Arzimovich - Elementary Plasma Physics - 1965

Elementary Plasma Physics
L e v A . A rz im o v ic h
Elementary
Plasma Physics
A Blaisdell Book in the Pure and Applied Sciences
CONSULTING EDITOR
Bernard T. Feld, Massachusetts Institute o f Technology

Elementary
Plasma Physics
LEV A. ARZ I MOVI CH

Academy o f Sciences ( U.S.S.R.)
TRANSLATED BY
Script a Technica, Inc.
BLAISDELL PUBLISHING COMPANY

A Division o f Ginn and Company
NEW YORK . TORONTO • LONDON
A translation of the original volume
Elementarnaya Fizika Plazmy
published by Gosatomizdat
(State Committee for Application of Atomic Energy
of the U.S.S.R., State Press for Atomic Physics
and Nuclear Engineering) Moscow, 1963
First English Edition, 1965
Copyright © 1965, by Blaisdell Publishing Company,

A Division of Ginn and Company.
All rights reserved.
Library of Congress Catalog Card Number: 65-14569.
Printed in the United States o f America
ONULP
Plasma is a state of matter, normally found in extraterrestrial space.
It possesses exceedingly interesting properties that are being in
creasingly applied to the solution of important problems of modern
technology. Thus, the subject is of major interest to the general
reader. However, the reader who desires an understanding of the
basic properties of plasma encounters considerable difficulties.
Most present-day books employ extremely complex mathematical
tools and require a good preparation in modern theoretical physics.
There is no need, however, to use complex mathematics to explain
the basic phenomena occurring in plasma. Such mathematics is
often misused and is consequently of no more than decorative
significance.
We shall attempt to interpret the fundamentals of plasma physics
in a manner understandable to a reader with a high-school knowl
edge of mathematics and physics. We do not mean to imply, how
ever, that the material is presented in a predigested form — for
assimilation without thought. Such an attitude is neither suitable
nor possible in the study of plasma physics.
This book discusses not only the various plasma processes but
also the methods of analysis of such processes. By becoming
familiar with these methods, the reader will be able to attain a
genuine understanding of the physical significance of the phe
nomena involved. The absolute cgs (centimeter, gram, second)
system of units is used in all computations. In some cases, the
practical system of electrical units (ampere, volt, joule) is also
used.
Contents
1. Introduction 1
1.1 Production of Plasma 1

1.2 Plasma as a Quasi-neutral Medium 7
1.3 General Nature of the Motion of Charged Particles in
Plasma 10
2. Motion of Charged Particles in Electric and Magnetic Fields 12
2.1 General Laws of Motion for Charged Particles in

Electric Fields 12
2.2 The Motion of a Particle in the Field of a Point Charge 16
2.3 Optical Analogy for the Motion of a Charged Particle
in an Electric Field 19
2.4 Motion in an Alternating Electric Field 22
2.5 The Motion of Charged Particles in a Uniform Mag
netic Field 26
2.6 General Nature of the Motion of Charged Particles in a
Nonuniform Magnetic Field 31
2.7 Motion of Particles in the Presence of Electric and
Magnetic Fields 43
3. The Motion of Charged Particles in Plasma 52
3.1 A General Review of the Properties of Gases 52

3.2 Collisions between Charged Particles in Plasma 58
3.3 Interaction of Electrons and Ions with Neutral Particles 67
Contents
4 . Radiation 81
4.1 Bremsstrahlung 81
4.2 Recombination Radiation 83
4.3 Radiation Emitted by Excited Atoms and Ions 84
4.4 Betatron Emission from Plasma 85
4.5 Flux of Energy Emitted by Plasma 87
5. Electric Current, Diffusion, and Thermal Conductivity 90

5.1 Electric Current in Plasma 90
5.2 Plasma in a High-frequency Field 99
5.3 Motion under the Action of a Pressure Difference 103
6. Experimental Techniques 114

6.1 The Probe Method 114
6.2 Radio-wave Measurements 119
6.3 Plasma Spectrometry 123
7. Plasma in a Magnetic Field 126

7.1 Forces Acting on Plasma in a Magnetic Field 126
7.2 Current in a Magnetized Plasma 134
7.3 Diffusion in a Magnetic Field 140
7.4 The Pinch Effect 148
7.5 Oscillations and Waves in Plasma 153
8. Technological Applications 160

8.1 Controlled Thermonuclear Reactions 160
8.2 Magnetohydrodynamic Conversion of Energy 175
8.3 Plasma Engines 181
T a ble of F u n d a m en ta l C o nstants 183
In d e x 185
Elementary
Plasma Physics
I
Introduction
1.1 Production of Plasma

Consider a closed vessel whose walls have a very high melting
point and suppose that the vessel contains a small amount of a
solid material. When the temperature of the vessel is gradually
increased the solid will melt and eventually evaporate, filling the
vessel uniformly with the resulting vapor. When the temperature
is increased still further all the molecules of the gas will dissociate,
that is, divide into individual atoms (if the gas was originally in
molecular form such as hydrogen, nitrogen, or oxygen). The vessel
will thus eventually contain a gaseous mixture of the elements
corresponding to the original composition of the solid material.
The atoms of these elements will be in a rapid and random motion
as a result of which they will collide with each other. The average
velocity of this random thermal motion is proportional to the
square root of the temperature of the gas and inversely proportional
to the square root of its atomic weight. The average thermal veloc
ity v can be calculated from the following formula
( 1. 1)
where T is the absolute temperature and A is the atomic weight of

the gas. In this formula the velocity is given in centimeters per
second. It follows from (1.1) that, for example, when T = 1000°K,
the mean velocity of hydrogen atoms is about 4 X 105 cm per sec,
while the mean velocity of mercury atoms is only 3 X 104 cm per
sec.
1
2 Introduction
By increasing the temperature from the lowest value (a fraction

of a degree above absolute zero, which is possible to achieve by
modern methods) to a few thousand degrees', it is possible to
take practically any material through the three possible states,
namely, solid, liquid, and gaseous. It is then natural to ask: How
will the properties of the material change if the heating is continued
still further and the temperature is increased beyond a few thousand
degrees Kelvin? At very high temperatures it is difficult to imagine
that the vessel containing the material under investigation will
remain solid, since most melting points lie below 3000 - 4000°K.
However, this is a practical difficulty which can, as we shall see, be
overcome. We shall therefore assume that we have a perfect
container whose walls have a higher melting point than any tem
perature which we shall consider. Under these conditions we shall
find that at temperatures as low as 3000 - 5000°K there are already
signs that certain new processes take place which are associated
with a change in the state of the atoms themselves, rather than of
the gas as a whole.
It is well known that each atom consists of a positively charged
nucleus, which carries practically all the mass of the atom, and a
number of electrons which surround the nucleus, forming the
electron shell of the atom. This shell, and in particular its outer
components, which are relatively weakly bound to the nucleus,
has a relatively brittle structure. Thus, when an atom collides with
a rapidly moving particle, one of the outer electrons may be
removed leaving behind a positively charged ion. This ionization
is the most characteristic process for this range of temperatures.
We thus arrive at the important result that, when the temperature
is high enough, the gas is no longer neutral but contains both
positive ions and free electrons. The number of ions and electrons
in the gas increases very rapidly with increasing temperature T.
When the gas is in thermal equilibrium with the surrounding
medium (the walls of the perfect container), most of the atoms will
become ionized by the time temperatures of the order of 10,000 —
100,000°K have been reached, and there will be practically no
neutral atoms left.
•All temperatures in this book are given in the absolute units (degrees Kelvin).
Production o f Plasma 3
F igure 1. Degree o f ionization o f hydrogen as a function o f temperature.
The relative number of ionized atoms in hydrogen is shown in

Fig. 1 as a function of the temperature. The degree of ionization
depends not only on the temperature but also on the density of the
gas (though less sensitively). For example, in the case of Fig. 1,
the total number of ions and neutral atoms per unit volume was
assumed to be 7 X 1016. At room temperature the pressure corre
sponding to this density is about 1mm Hg. It is evident from Fig. 1
that when T — 10,000°K, the number of ionized atoms is less than
10% of the total number of hydrogen atoms, while at 30,000°K
there is only one neutral atom for every 2 X 104 positive ions
(protons).
The electron shell of the hydrogen atom contains only one
electron, and therefore the ionization process is completed as soon
as this electron is removed. In other atoms the electron shell ex
hibits a much more complicated structure and the energy which is
necessary to remove an electron is very different for the different
electrons. The outermost electrons can be fairly easily removed,
and, as has already been pointed out, at a temperature of the order
of 20,000 - 30,000°K there are practically no neutral atoms left.
This means that the gas as a whole is completely ionized. However,
this does not mean that the ionization process has been concluded,
4 Introduction
since there are still a number of electrons left in each of the atoms
even after one or more of the outer electrons have been removed.
Thus, the larger the atomic number (that is, the number indicating
the position of the particular element in the periodic table) the
larger the number of electrons in the atom, and the higher the
energy necessary to remove the inner electrons. Therefore, the
final ionization of heavy atoms, in which even the inner electrons
are removed, will occur only at exceedingly high temperatures
(of the order of 106 - 107 degrees Kelvin). We know that the com
plete ionization of a heavy gas leads to a situation in which to each
positive ion there will be Z free electrons, where Z is the atomic
number, that is, the original number of electrons in the neutral
atom. However, the gas as a whole remains neutral, since the
ionization process as such does not produce an excess of charge of
either sign.
At high temperatures the ionization of a gas occurs as a result
of the various interactions between the individual atoms on the
one hand, and electrons, ions, and radiation, on the other. These
interactions are varied and very complicated. We shall, therefore,
postpone their description to a later chapter, and for the moment
confine our attention to the behavior of the ionized gas as a
whole.
A gas in which an appreciable number of atoms or molecules
are ionized is referred to as plasma. The term plasma was first
introduced by the American physicists Langmuir and Tonks in
1923. Plasma is the normal state of matter at temperatures of the
order of 10,000°K or more. It is the most common state of matter
in nature. For example, the sun and all the stars are nothing else
but gigantic condensations of high-temperature plasma. The outer
layers of the earth’s atmosphere, that is, the ionosphere, are also
known to consist of plasma.
In the above discussion we introduced the concept of plasma
in terms of the simple process of heating of a gas in a perfect
container. In practice, this is neither the best nor the easiest
method of producing plasma either in laboratory experiments or
in industrial processes. The normal conditions under which
plasma is produced are those which prevail in the various forms
Production o f Plasma 5
of gas discharges. When an electric discharge is produced in a

gas, a current flows; the current carriers are the electrons and
ions which are produced as a result of the ionization of the gas.
The ionization process itself is an integral part of the current flow
and cannot be separated from it. While the current is flowing, new
ions and electrons are being produced all the time, and the degree
of ionization is maintained at a definite level. The lightning
discharge, the spark, the glow of a neon tube in an advertisement —
these are all examples of phenomena which occur in highly ionized
plasma. However, there is one important difference between the
plasma obtained by heating a gas in a container, and that in the
gas discharge. The plasma in the gas discharge is not in thermal
equilibrium, but is heated from the outside, since it receives the
energy which is released during the passage of the current. Heat
is extracted from it, either by direct contact with the cold walls
of the discharge tube or by heat transfer to a surrounding ordinary
gas. The plasma which is produced in a high-intensity gas dis
charge may have a much higher temperature than the metal, glass,
or ordinary gas surrounding it. Moreover, plasma of this kind
exhibits a further nonequilibrium property: It consists of a number
of components which are not equally heated. These components
are electrons, positive ions, and neutral atoms. They are thoroughly
mixed just as, for example, the oxygen and hydrogen are in the
atmosphere. However, in contrast to an ordinary gas mixture in
which all the particles have the same mean kinetic energy of
thermal motion, the electrons, ions, and neutral atoms in the
plasma of a gas discharge have different mean kinetic energies.
As a rule, the electrons have a much higher energy than the ions,
and the kinetic energy of the ions may be greater than the energy
of the neutral atoms and molecules. It may therefore be said that
plasma consists of a mixture of components at different tempera
tures. It is well known that the average kinetic energy Wr which
is associated with the random thermal motion of atoms and
molecules is related to the temperature T by the following simple
expression:
WT = \ k T , ( 1. 2)
6 Introduction
where k is the Boltzmann constant, which is equal to 1.38 X 10~16

erg per deg. We recall that k is equal to the ratio of the universal
gas constant R to Avogadro’s number, that is, the number of
atoms in a gram-atom.
Because the mean kinetic energies of the electrons, ions, and
neutral particles in the plasma are different, the plasma as a whole
must be regarded in a sense as having three different temperatures,
that is, the electron temperature Te, the ion temperature 7",, and
the atomic temperature To. Usually, Te ^>Ti > To, where the
sign » means “much greater than.” The very large difference
between Te and Tt, which is characteristic of most forms of gas
discharge, is due to the very great difference between the mass of
the electrons and the mass of the ions. The external sources of
electrical energy, which are used to produce and maintain the gas
discharge, communicate energy directly to the plasma electrons,
since it is the light electrons which act as the current carriers.
The ions, on the other hand, obtain their energy by colliding with
the rapidly moving electrons. Since the mass of the ions is much
greater than the electron mass, the kinetic energy lost by an electron
to an ion in an electron-ion collision is quite small. This becomes
clear if one considers the collision between a light plastic ball and a
heavy steel sphere. Simple calculations, based on the conservation
of energy and the conservation of linear momentum, show that if
a body of mass collides with another body having a much greater
mass m2, then the fraction of kinetic energy communicated to the
heavier body must be less than 4w ,/ra2. Therefore, since the
ratio of the mass of the electron to the mass of an ion is equal to
1/1840^4, where A is the atomic weight of the atoms from which the
ions are produced, the maximum energy which can be received by
the ion is only about 2 X \0~3/A . It follows that an electron
must undergo a very large number of collisions with the ions before
it will lose most of its energy. Since the processes in which an
exchange of energy takes place between the electrons and ions
occur in parallel with the processes in which the electrons receive
energy from the external sources of current, there is usually a
large temperature difference between the electrons and ions in the
gas discharge. Thus, for example, in discharge devices such as
neon tubes, mercury rectifiers, and so on, the electron temperature
Plasma as a Quasi-neutral Medium 1
Tc is usually of the order of 10,000°K, while Ti and To are usually

less than one or two thousand degrees. In the arc discharge, which
is used in electrical welding, the rate of electron-ion collisions is
higher, and this tends to reduce the temperature difference
between the electrons and the ions. However, Tc is still greater
than Ti, the former being of the order of a few tens of thousands of
degrees while Ti and T0 are of the order of 6000°K. Under certain
special conditions the ion temperature of a highly ionized plasma
may be much greater than the electron temperature. Such con
ditions prevail, for example, in the high-power short-lived electrical
discharges which are used in studies of the controlled thermo
nuclear reactions.
1.2 Plasma as a Quasi-neutral Medium
We shall now discuss the definition of plasma and its main

properties in somewhat greater detail. In general, a plasma, that
is, an ionized gas, may consist of a number of components.
Even in the relatively simple case where plasma is produced by
ionization of a chemically simple gas, for example, nitrogen,
oxygen, or mercury vapor, the ion component will consist of a
number of different ions, that is, ions with one, two, three, or
more elementary charges. Moreover, in addition to the atomic
ions, the plasma may contain molecular ions, and also neutral
atoms and molecules. Each of these components can be character
ized by a corresponding concentration n and temperature T. In
the simplest case, when all the ions are singly charged atomic
ions, and the neutral component is fully dissociated and consists
of atoms only, the plasma will contain only three components,
namely, electrons, ions, and neutral atoms. These conditions will
occur only in high-intensity discharges in hydrogen, deuterium,
or tritium. In this particular case, the ion concentration /t; is
equal to the electron concentration ne. in the more general case,
when the plasma consists of singly charged ions with concentration
/?,, doubly charged ions with concentration m, and so on, we have
the following approximate equation
ne — «i "T 2n2 + 3n3 + • • •.

8 Introduction
This shows that the plasma as a whole is quasi-neutral; that is,

there is no appreciable surplus of charge of either sign. We must
discuss this particular property of plasma in greater detail since
it is particularly important and in the final analysis it forms the
foundation upon which the definition of plasma is based.
It is natural to ask: To what extent is an ionized gas quasi
neutral? Whatever the method which is used to produce the ioni
zation, it is quite clear a priori that the number of positive and
negative charges must be the same. Since the velocity of the elec
trons and of the ions are very different, the former are more likely
to leave the region in which they are produced. Therefore, although
the number of charges of either sign is originally the same, never
theless, it would appear that, because the electrons are faster than
the ions, they are lost much more quickly to the walls of the
discharge tube, and the ions are left behind. On the other hand,
it must be remembered that a net loss of electrons from the ionized
gas will immediately give rise to an excess charge of the opposite
sign, and this will tend to equalize the current of electrons and ions,
and to reduce the difference between the concentrations of particles
of opposite signs. The conditions under which this effect will be
sufficient to maintain the plasma in the quasi-neutral state may be
deduced as follows.
Suppose for the sake of simplicity that the ionized gas contains
electrons and singly charged ions only. If the plasma is to be
quasi-neutral, ne must be nearly equal to The question is:
How will a difference between ne and «,• effect the behavoir of the
individual particles? It is clear that this will depend on the magni
tude of the electric field which is produced as a result of the dif
ference between the two concentrations. There are two extreme
cases which will help us to understand this problem. Thus, if the
number of charged particles in a given volume is small, then the
electric field produced by them is too small to have any effect on
their motion and the individual electrons and ions move quite
independently of each other. The plasma need not then be quasi
neutral. The opposite case occurs when the charged-particle
concentration in a sufficiently large volume is very high. Here the
difference between ne and /;,• gives rise to the appearance of an
Plasma as a Quasi-neutral Medium 9
electric field which may be sufficient to equalize the currents and

re-establish the quasi-neutral property of the plasma as a whole.
In the final analysis everything depends on the ratio, of the potential
energy of an individual ion or electron in the electric field pro
duced as a result of the departure from quasi-neutrality to the
mean kinetic energy associated with its thermal motion. If the
potential energy Wp, corresponding to an appreciable departure of
ne from m, is much greater than kTe, which is a measure of the
energy of thermal motion of an electron, then the plasma will be
quasi-neutral to a high degree of accuracy. A more detailed
analysis of the relationship between Wp and kTe shows that the
plasma as a whole will be quasi-neutral provided
(1.3)
where n is the concentration of charged particles (number of elec

trons per unit volume) and r is a characteristic linear dimension
of the region occupied by the ionized gas, for example, the radius
of a spherical container. It is quite easy to see why the parameter r
should appear in this expression. Thus, at a given charged-particle
concentration, the potential due to these particles, and therefore
the potential energy of an individual particle, will depend on the
size of the region occupied by the particles. The characteristic
linear dimensions of the region must therefore necessarily enter
into the condition describing the quasi-neutral property of plasma.
The quantity 5 ^T e/n is called the Debye radius after the German
physicist who first introduced this quantity in his theory of elec
trolysis, where the situation is analogous to that in an ionized gas.
Henceforth, the Debye radius will be denoted by ro
ll follows from the condition given by (1.3) that, if the linear
dimensions of the region occupied by an ionized gas of given elec
tron concentration ne and electron temperature Te is much greater
than the Debye radius r D, then inside this region ne ~ m, where the
sign = means “approximately equal to.” Under these conditions,
a considerable departure of ne from n, will give rise to the appear
ance of an electric field which will eject particles of one sign (the
surplus particles) and will prevent the escape of particles of the
10 Introduction
other sign. This mechanism automatically maintains the equality

of ne and m, and will cease to be effective only when r « rD.
When the linear dimensions of the region occupied by the gas
are much smaller than rD, the electric fields produced as a result
of the difference between ne and m are too small to have any
appreciable effect on the motion of the individual particles.
We are now in a position to give a more precise definition of
plasma. As long as the relative number of charged particles giving
rise to a field is too small to affect their individual motions, there
is no point in introducing a new state of matter. The new form of
matter which we have called plasma corresponds to the state in
which the number of electrons and ions is so large that even a
small displacement of the electron component relative to the ion
component is impossible, because such a displacement gives rise
to strong electric fields tending to prevent any departure from the
equality of ne and Therefore, an ionized gas can be referred to
as plasma if, and only if, the condition given by (1.3) is easily
satisfied. ?■’
It must be emphasized that the quasi-neutral property of plasma
is only observed provided the volume occupied by the plasma is
large enough. If we consider a cube with sides of length x inside
the plasma, where x is much smaller than rD, then inside this cube
the number of ions may in fact be very different from the number
of electrons. However, as the ratio x/ro increases, the ratio
ne/m tends to unity.
1.3 General Nature of the Motion of Charged Particles in Plasma

Although plasma may be regarded as a special form of a gaseous
mixture (in the simplest case, a mixture of two components, namely,
electron and ion gases), there are many important physical prop
erties in which it differs from an ordinary gas containing only
neutral particles. These differences are especially evident in the
behavior of plasma in electric and magnetic fields. In distinction
to an ordinary neutral gas, which is practically unaffected by
electric and magnetic fields, the properties of plasma may be
altered very considerably by such fields. An electric field — even
Nature o f Motion o f Charged Particles in Plasma 11
a very weak one — gives rise to an electric current in the plasma.

In a magnetic field, plasma behaves as a diamagnetic material.
It also exhibits a very strong interaction with electromagnetic
waves, which is shown, for example, by the fact that radio waves are
reflected by plasma as if it were a mirror. This book will be largely
concerned with the specific properties, which are exhibited by
plasma in its interaction with electric and magnetic fields and
serve as the basis for many scientific and technological applications
of plasma. In order to understand the nature of these processes, we
must first consider the behavior of the individual electrons and ions
which make up the plasma. The analysis of the motion of these
particles must be based on the following:
1. The laws of motion of electrons and ions in electric and
magnetic fields produced by external sources.
2. The elementary interactions between the particles during
collisions. The change in the direction of motion of the colliding
particles, which occurs during such interactions, may also be
accompanied by the production of new charged particles and the
appearance of various forms of radiation which then escape from
the plasma.
The behavior of the individual particles can be used to interpret
the macroscopic properties of plasma as a whole.
Let us now try to describe, in very general terms, the motion of a
charged particle in plasma. The path of each electron or ion can
at first be regarded very approximately as consisting of segments
over which the particle executes free motion without interacting
with its neighbors. The free motion of the particle is interrupted
by collisions which result in changes in its direction of motion.
In the intervals between successive collisions, the particle moves
under the action of the general electric or magnetic field produced
in the plasma by the external sources. This is a very simplified
description of the motion of a plasma particle. It must be aug
mented by taking into consideration the intrinsic electric field
which exists in plasma even in the absence of external sources.
Thus, each charged particle produces a radial electric field in its
neighborhood. Since the number of particles of both signs is the
same, the average internal field must be zero. However, this does
12 Introduction
F igure 2. Spatial variation o f the internal electric field along a given

direction in plasma at a given instant o f time.
not mean that at every instant of time the electric field at every
point in space is exactly zero. In fact, the field at an internal point
in the plasma exhibits very rapid fluctuations, both in its magnitude
and in its direction. These random fluctuations yield zero average
field, provided that the interval of time over which the average is
taken is long enough.
However, if the electric field is measured at different points in
the plasma at a given instant of time, the resulting distribution is
found to be approximately of the form shown in Fig. 2. The
random space and time variations in the internal field of the
plasma are superimposed on the macroscopic fields due to external
sources. Therefore, the motion of an electron or an ion in plasma
can never be regarded as strictly consisting of free-motion intervals
during which the particle moves under the influence of the average
(that is, the external) field only. The random internal field cannot
be ignored altogether; it is implicit even in our simplified model.
When we say that the effect of all other charged particles in the
plasma on the trajectory of the chosen ion or electron is exhibited
by the collisions between them, we are, in fact, replacing the true
effect, which is due to the interaction of the particle with the
internal microfield, by the device of a collision.
After these preliminary remarks, we can proceed to an analysis
of the free motion of individual charged particles in given electric
and magnetic fields.
2
Motion o j Charged Particles

in Electric and Magnetic Fields
2.1 General Laws of Motion for Charged Particles

in Electric Fields
The laws governing the motion of a charged particle in an
electric field resemble the laws of motion in a gravitational field.
Consider Fig. 3 which depicts the trajectories of charged particles
in an electric field parallel to the y axis. The arrows indicate the
initial velocities of the particles at particular instants of time.
The force acting on a charged particle is equal to qE, where q is
the charge and E the electric field strength. For singly-charged
particles, q = ± e , where e is the elementary electric charge;2
for multiply-charged ions, q is an integral multiple of e. Under
the action of this force, a singly-charged positive ion of mass nu
will undergo an acceleration equal to eE/rru in the direction of the
positive y axis (Fig. 3). The acceleration of an electron would be
in the direction of the negative y axis and equal to eE/me, where
me is the mass of the electron. Since an electron is much lighter
than an ion, it undergoes a much greater acceleration.
The trajectory of a charged particle in a uniform electric field,
that is, a field of constant magnitude and direction, is always a
parabola. The specific form of this parabola depends on the
properties of the particle, on the initial conditions and on the
magnitude of the electric field E. Suppose, for example, that the
2The elementary charge e is equal to 4.8 X 10~10 esu = 1.6 X 10-19 coulomb.
13
14 Motion o f Charged Particles in Electric and Magnetic Fields
F igure 3. Trajectories of charged particles in a uniform electric field.
electric field is parallel to the y axis, and that the initial velocity
of the particle v0 is parallel to the x axis (path / in Fig. 3). The
motion of the particle in the direction of the x axis is then uniform,
whilst in the direction of the y axis it is uniformly accelerated.
The coordinates of the particle are given by the formulas
x = v0t, y = — /2, (2.1)
which are well known from elementary mechanics and represent

uniform motion and uniformly-accelerated motion respectively.
The time t can be eliminated between these two equations, to give
a relation between y and x, that is, to give the equation of the
trajectory of the particle. This equation is
qE
*2 ( 2 . 2)
y = 2^
It is well known that this represents a parabola. The increase in
the velocity of a charged particle, which it assumes in a time t in
the direction of the force acting upon it, is {qE/m)t. The kinetic
energy of the particle, which is equal to (l/2)m v2, does not remain
constant during the motion because the electric field does work on
the particle. This work is equal to the product of the force acting
Laws o f Motion for Charged Particles in Electric Fields 15
on the particle and the distance traversed by it in the direction of

the field; that is, it is given by W = qE(y2 ~ Jh) (see Fig- 3).
Therefore
mvj mv\
qE(y2 - yi), (2-3)
T ~ T
where Vi and Vi are the velocity of the particle at any two points
M i and M 2 on the trajectory. We note that the quantity E(y2 — y {)
is the potential difference between the points Mi and M 2 . There
fore, the right-hand side of Equation (2.3) is the product of the
charge of the particle and the potential difference between the end
points of the trajectory traversed by the particle. For example,
if the particle begins its motion from a state of rest, its kinetic
energy at any point on its trajectory will be equal to q U, where U
is the potential difference, in electrostatic units, between the initial
and final points of the trajectory. If the potential at the initial
point is conventionally taken to be equal to zero, then at any
other point on the trajectory U will be equal to minus the value
of the potential.3 If U is expressed in practical units, that is,
volts (V), then the relation between the kinetic energy and the
potential difference traversed by the particle can be written in the
following form:
— = Ze— = Z [ / - 1.6-IO-12, (2.4)
where Z = q/e is the number of elementary electric charges

carried by the particle. The unit of kinetic energy in Equations
(2.3) and (2.4) is the erg.
It follows from (2.4) that a singly-charged particle which has
traversed the potential difference of 1 volt receives an energy
equal to 1.6 X 10~12 erg. However, it is more convenient to
express the energy of particles such as electrons and ions directly
in terms of the potential difference traversed by them. In this
system of units, a convenient unit of energy is the energy attained
3A positively-charged particle can only be accelerated along a falling potential.

Therefore, if the initial potential is equal to zero, the potential at any other point
on the trajectory must be negative.
by a singly-charged particle, for example an electron, on passing

through a potential difference of 1 V. This new unit of energy is
widely used in electron, atomic, nuclear, and plasma physics, and
is referred to as the electron-volt (eV). In view of the above
definition, 1 eV = 1.6 X 10~12 erg. A particle of charge q = Z e
reaches an energy equal to Z U eV on falling through a potential
difference U.
We note that an electron with the energy of 1 eV has a velocity
of 5 X 107 cm per sec, while a hydrogen ion (proton) of the same
energy has a velocity of 1.4 X 106 cm per sec. This velocity is
greater than the mean velocity of ordinary hydrogen molecules at
room temperature by a factor of approximately 10. The energies
of atoms and molecules in neutral gases can also be expressed in
terms of electron-volts. All that is required is to divide the kinetic
energy, given in ergs, by the conversion factor 1.6 X 10-12. When
this is done, it is found that the molecules or atoms of a gas at room
temperature have a mean kinetic energy of about 0.04 eV. The
mean kinetic energy of a gas particle reaches 1 eV at about 7600°K.
In stellar interiors, where the temperatures are believed to be of the
order of ten million degrees, the mean energy of the constituent
particles is of the order of some thousands of electron-volts.
2.2 The Motion of a Particle in the Field of a Point Charge

So far, we have only considered the simple case of a charged
particle in a (uniform) electric field whose intensity is the same at
all points of space. In a more complicated field, the trajectory of
the particle will no longer be a parabola. However, Equation
(2.4), which gives the relation between the change in the kinetic
energy and the potential difference traversed by the particle, is
still valid in this general case. The increase (or decrease) in the
kinetic energy between two points in space is always given by the
potential difference between the two points and — this must be
especially emphasized — is independent of the particular path
traversed by the particle between the two points. Among the many
special cases of the motion of electrons and ions in an electric field,
a particularly interesting one is the motion in a radial field, that is,
Motion o f Particle in Field o f Point Charge 17
the field produced by a point charge. This case corresponds to the

interaction between the individual electrons and ions in plasma.
Figure 4 illustrates a coulomb collision between an electron and a
plasma ion in which the two colliding particles interact through
their electric fields. The precise form of the trajectory of a charged
particle in the field of a point charge can only be determined by
mathematical methods which are outside the scope of the present
book. However, the most important property of the trajectory,
that is, the magnitude of the angle t? through which the particle is
deflected as a result of the interaction with the center of force, may
be approximately estimated in a very simple way. Suppose that the
center of force is located at the point 0 (Fig. 4) and that the incident
particle passes it at a large distance, so that the deflection from its
original direction of motion is small. The minimum distance
between the incident particle and the charge at the point 0 will then
be approximately the same as the impact parameter b, which is
equal to the length of the perpendicular dropped from the point 0
on to the extension of the original direction of motion of the particle
(the line AB in Fig. 4). Since the force acting on the particle is in
versely proportional to the square of the distance from the center of
force 0 , it follows that most of the interaction between the incident
particle and the center of force takes place over a small part of its
trajectory, in the immediate neighborhood of the point of closest
approach to 0. Suppose that this part of the trajectory occupies
the region A 1A 2 in Fig. 4. For very approximate calculations it may
be assumed that the length of this part of the trajectory is equal to
A A\ B
F igure 4. Trajectory o f a charged particle in the field o f a

point charge o f opposite sign.
twice the impact parameter (at the ends of this segment, the force
acting on the particle is reduced by one-half as compared with its
value at the point of closest approach). A particle having a velocity
v will traverse the segment A 1 A 2 in a time equal to 2b/u. During
this time, the force acting upon it will be approximately perpendicu
lar to its path. This force may be approximated to be equal to
qiqi/b2, where q\ and qi are the two interacting charges. As a
matter of fact, the mean force will be smaller, but the true time of
interaction will be greater than that assumed here. During the
time taken by the particle to traverse the segment A 1 A 2 , this force
will transmit to the particle a velocity equal to
1 q\q2 2b
m b2 v
in the direction perpendicular to the initial direction of motion,
where m is the mass of the particle. The ratio of this velocity to the
initial velocity v must be equal to the tangent of the angle of deflec
tion t? (see Fig. 4), and since in the case of small deflections tan d is
approximately equal to we have
2qiq2
(2.5)
mv2b
where the symbol ~ indicates that the expression is only very
roughly true; that is, it gives the order of magnitude of 1? and the
true result may be different by an appreciable factor. Such approxi
mate expressions are widely used in modern physics and are fre
quently just as useful as the precise formulas. Approximate calcu
lations are useful whenever we have to analyze complicated phe
nomena involving a large number of different factors. Under such
conditions, it is first necessary to ascertain the relative importance
of the various factors. A typical example is the expression for the
Debye radius given by (1.3). There is very little point in trying to
establish the exact value of this quantity since the concept of the
Debye radius is introduced in order to distinguish between two
extreme cases in the behavior of a system of charged particles,
namely, ( 1 ) when the particles are independent of each other, and
(2 ) when they form a plasma.
Motion o f Charged Particle in Electric Field 19
Approximate formulas are a very powerful tool in the elucidation

of the general physical properties of a phenomenon because they
are free from the detail which has to be introduced in a more
rigorous mathematical calculation. Equation (2 .5 ) is valid — in
the sense indicated above — only for those values of b which cor
respond to a deflection much smaller than 1 radian. The precise
relation between t? and b, which is valid for all values of the impact
parameter, may be derived by means of a rigorous theoretical
analysis of the motion of a charged particle in the field of a point
charge. The exact result is
( 2. 6)
* 4 - B
For large b, that is, when the particles pass each other at a large
distance, this formula is identical with (2.5), and this justifies the
very approximate estimate previously derived. The range of ap
plicability of the formula may be illustrated by the following
special case: The deflection experienced by a 1 eV electron when it
interacts with a singly-charged ion at an impact parameter b = 1 0 ~ 6
cm is about 0.14 rad, that is, about 8.5°. The magnitude of the de
flection decreases with increasing energy.
2.3 Optical Analogy for the Motion of a Charged Particle

in an Electric Field
The above two examples, namely, motion in a uniform and
in a radial electric field exhaust all the possibilities as far as the
problems encountered in plasma physics are concerned. However,
it will be useful to consider one further special case which is interest
ing because it will be found to exhibit certain very important
general properties.
Consider a charged particle (in order to be specific, suppose that
it is an electron) passing through the boundary separating two
regions of space at different potentials (Fig. 5). This means that,
in a very thin boundary layer between the two regions, there is a
very strong electric field giving rise to a sudden drop in the potential
For the sake of simplicity, we shall suppose that the boundary
F igure 5. T r a je c to r y o f a p a r tic le p a s s in g th ro u g h th e b o u n d a ry b e tw e e n
tw o re g io n s a t d iffe re n t p o te n tia ls .
layer is infinitely. In thin practice, this can never be realized exactly

but it is possible to approach this situation approximately under
certain special conditions. If the potential is measured relative to
the point at which the velocity of the electron was zero, Equation
(2.4) will remain valid over the entire trajectory of the electron.
Suppose that at the boundary the potential changes from U\ to Ui.
Since the potentials on either side of the boundary are constant, an
electron will experience no forces, either before or after passing
through the boundary layer. It will therefore execute straight-line
motion both in region A and in region B. However, in the very
short interval of time during which it is within the boundary layer,
it experiences a very strong force at right-angles to the boundary of
separation between A and B. This force either increases or de
creases the component of the velocity of the electron perpendicular
to the boundary. The component which is parallel to the boundary
remains unaltered. This means that the trajectory must show a dis
continuity at the boundary, as shown in Fig. 5. The case illustrated
in this figure corresponds to an acceleration of the electron in the
boundary layer. Suppose that u, and v2 are the velocities of the
electron in the two regions A and B respectively, and let the angles
between these velocities and the normal to the boundary be de-
Motion o f Charged Particle in Electric Field 21
noted by a and /3 . Since the velocity component parallel to the

surface remains unaltered, it follows that
Vi sin a = v2 sin(i.
When Z = 1, it follows from Equation (2.4) that \m v\ = eU x,
\m v\ = eU2. When these equations are combined, it is found that
sin« = n2 = n -v
sin/3 Vl y jU i‘ { ’
The mathematical form of this result shows a striking similarity to
the law of refraction of light rays. It is well known that when a
light ray passes from a medium A to a medium B, its direction of
propagation changes in accordance with the relation
sina _ n2
(2.7a)
sin/3 n{
where a and (3 are the angles of incidence and refraction, and n{ and
n2 are the refractive indices of the two media. Comparison of
Equations (2.7) and (2 .7 a) will show that an electron beam passing
through the boundary between two regions at different potentials
behaves in a way which is analogous to the behavior of light rays
in optics. In the present context, the square root of the potential
plays the role of the refractive index. It may be shown that the
analogy between the mechanics of charged particles in an electro
static field and the laws governing the propagation of light rays is
very much wider than would be supposed on the basis of the above
example. In point of fact, any electric field may be imagined as
consisting of a very large number of thin layers, each of which cor
responds to an approximately constant potential which changes
discontinuously by a small amount on passing from one layer to the
next (Fig. 6 ). In such a field, the trajectory of a particle will consist
of a large number of very short, straight segments with breaks, at
the points of transition from one layer to another, at which the
trajectories are refracted in accordance with (2.7). The optical
analogue of this situation is the propagation of a light ray through a
transparent medium with a varying refractive index. Therefore,
the motion of charged particles in electrostatic fields is fully analo-
F igure 6. Motion o f a particle in an electric field o f a slowly varying

potential. U\, Ui, U3, . . . represent the potentials o f the regions
into which the space may be subdivided.
gous to the propagation of light rays. This fact is used in the design
of electronic instruments which are analogous to optical apparatus,
for example, the electron-optical converter, the electrostatic elec
tron microscope, and so on. Analysis of the general similarity
between ordinary mechanics and geometrical optics has led to the
development of modern quantum mechanics, which provides a
very satisfactory description of all atomic phenomena. However,
all this is outside the scope of the present book, which is exclusively
concerned with the physics of plasma.
2.4 Motion in an Alternating Electric Field

All of our previous discussion was concerned with time-independ
ent electric fields. However, in order to understand certain phenom
ena which take place in plasma, it will be useful to consider the be
havior of electrons and ions in rapidly varying electric fields. All the
main features of the interaction of an alternating field with a charged
particle can be explained by analyzing the following simple ex
ample: Suppose that the electric field-strength is a periodic function
of time at all points in space. The dependence of E on time is then
Motion in Alternating Electric Field 23
F igure 7. Motion o f a charged particle in an alternating electric field:

a. variation in the field strength, b. variation in the velocity o f the particle,
c. displacement o f the particle in the direction o f the field.
of the form shown in Fig. 7a (a cosine curve). The distance between

successive crests of the curve is defined as the period T of the oscil
lations, while the frequency v, that is, the number of oscillations per
second, is related to T by the obvious formula v = \/T . A quantity
often used in physics is the angular frequency, which is usually
denoted by co and is equal to 2tv, so that w = 2-ir/T. The equation
of the cosine curve shown in Fig. 7a may then be written in the
following form:
E = Eo cos2-wt/T = Eo coseot.
The field strength reaches its maximum value at times t = 0, T,
2T , . . . and is equal to zero at t = T /4, 3774, 5T/4, . . ., and so on.
Suppose now that a charged particle with zero velocity is placed in
the electric field at time t = 0. The particle will immediately be ac
celerated in the direction of the field (if its charge is positive) so
that its velocity will increase during the first quarter of the period
and will reach its maximum value when the field is zero (point A in
Fig. 7b). The magnitude of the maximum velocity is proportional
to the force acting on the particle during the acceleration, and
inversely proportional to the mass of the particle. If the field
strength were constant during the first quarter of the period and
equal to its maximum value, the velocity of the particle at time
t = T /4 would be
qEo T _ x qEp
m 4 2 mu
Since, however, the field gradually decreases, the final velocity will,
in fact, be smaller than this value. Exact calculations show that the
velocity of the particle at time t = 774 is given by
qEo
( 2 . 8)
mu
During the next quarter of the period, the field changes sign, the
particle is decelerated, and its velocity becomes zero at the end of
the second quarter-period (point B in Fig. 7b). As soon as this
point is passed the particle, accelerated in the opposite direction,
reaches a maximum negative value equal to —vmax at time / = 3J/4,
and then again returns to zero at the point D, that is, after one
complete period of the electric field. The whole process is then
repeated periodically. The graph showing the variation in the
velocity of the particle is shifted relative to the graph of the field E
by one-quarter of the period. When £ is at a maximum, the velocity
is zero, while zero values of the field correspond to maximum
velocities ( ± i w ) .
Consider now the variation in the length of the path traversed by
the particle. In the time interval between 0 and T /4, the particle
begins to move in the direction of the field and eventually acquires
its maximum velocity. The motion in this direction continues
throughout this time interval, as long as the velocity is positive.
The displacement reaches the maximum positive value at time
t = T/2, which corresponds to the maximum negative value of E.
During the next half-period, the particle moves in the opposite
direction and eventually returns to the original position at the end
Motion in Alternating Electric Field 25
of the period. The process is then repeated periodically. A plot of

the position of the particle as a function of time is shown in Fig. 7c.
This curve is exactly out of phase with that showing the variation in
the field E with time, that is, the minima of one correspond to the
maxima of the other.
The equation of the curve representing the position of the particle
as a function of time can be written in the following form:
5 = ^ - 5(1 — cos w/). (2.9)

mo2
In practice, the electric field cannot be established instantaneously

but its amplitude gradually increases until it reaches the value Eo
after a number of complete periods. This is illustrated in Fig. 8 , in
which the electric field is plotted as a function of time t. Calcula
tions show that in this most important case, after the amplitude Eo
has been reached, the oscillations of the particle in space can be
described by
( 2. 10)
which differs from the preceding expression by a constant term

which is independent of time. In the present case, the curve
representing the displacement S will be the mirror image of the
curve representing the electric field E. This concludes our brief
description of the motion of a charged particle in an alternating
electric field. The results we have obtained will be used later when
F igure 8. Growth o f the amplitude o f an alternating electric field.

we come to analyze the interaction of plasma with electromagnetic

waves.
2.5 The Motion of Charged Particles in a Uniform Magnetic Field

Consider now the motion of electrons and ions in a magnetic
field. The force F which acts on a particle of charge q and velocity v
in a magnetic field H is equal to (qvH/c)sin0 , where 6 is the angle
between the directions of v and H and c is the velocity of light
(c = 3 X 1010 cm per sec). This force is perpendicular to both v
and H and the three variables v, H, and F form a right-handed
system. This means that the directions of v, H, and F are given
respectively by the thumb, index finger, and the middle finger of the
right hand, when the three fingers are at right-angles to each other.
In the case of negative q, the force acts in the opposite direction. In
the abbreviated language of vector analysis, this is represented by
the simple formula
.F = - v X H, (2.11)
c
where the symbol v X H is called the vector product. By definition,

the magnitude of the vector product of two vectors A and B is equal
to AB sin0 , where 0 is the angle between them, while its direction
forms a right-handed system with A and B (Fig. 9). All this leads
to the following two consequences:
F igu re 9. V e c to r p r o d u c t o f tw o v e c to rs A a n d B.
Motion o f Charged Particles in Uniform Magnetic Field 27
1. If a charged particle moves in the direction of the magnetic

field, then F = 0, since 6 = 0. If, on the other hand, the particle
moves at right-angles to the direction of the field, then F = qvH/c.
2. The force F is always perpendicular to the velocity of the
particle. It therefore does no work and can have no effect on the
magnitude of the velocity of the particle.
The simplest case is that of the motion of a charged particle in a
uniform magnetic field, that is, a field which can be represented by
parallel lines of force whose magnitude 77 is the same at all points of
the space occupied by the field. A field of this kind can be pro
duced, for example, between the plane pole pieces of a large electro
magnet or inside a uniformly wound solenoid. Suppose now that
the velocity of a charged particle is perpendicular to the magnetic
field. The force acting on the particle is then constant at all points
along its path and is equal to qvH/c. Since the force is perpendicu
lar to the velocity, it gives rise to a centripetal acceleration with the
result that the direction of the velocity undergoes a continuous
variation. In this particular case, the acceleration is also constant
in magnitude (because v and H are constant). Therefore, the
particle moves in a plane perpendicular to the direction of H on a
curvilinear trajectory with constant centripetal acceleration (Fig.
10). It is known from elementary mechanics that under these
conditions the trajectory must take the form of a circle. According
to the second law of Newton, the force which maintains a particle
on a circular trajectory is equal to the product of its mass and the
F igure 10. Motion o f a charged particle in a transverse magnetic field,

that is, in the case when v jl H.
centripetal acceleration, v2/r , where r is the radius of the circle.

Therefore
m v2
r
and hence
m vc
( 2 . 12)
r~ w
The quantity r is called the Larmor radius after the British
physicist Larmor, who investigated the laws of motion of charged
particles in a magnetic field at the end of the nineteenth century.
In practice, the most convenient form of this expression is that in
which the radius r is expressed in terms of the energy of the particle.
This formula can be obtained from (2.12) and is found to be
r = 1. 12-1014— (2.13)
where Z is the number of elementary charges carried by the particle

and W is its energy in electron-volts. For an electron, m = 9 X
10~ 28 g and Z = 1. Therefore, the Larmor radius of an electron is
given by
3.4^W
rc (2.14)
H
The radius of curvature of the trajectory of a proton for the same

values of H and W is greater by a factor of 42, that is,
143 yJ W
fP (2.15)
H
In order to find the Larmor radius for a singly-charged ion of

atomic weight A , this last expression must be multiplied by ^ A .
Equations (2.14) and (2.15) are illustrated in Figs. 11a and lib .
The various straight lines in these figures correspond to magnetic
field-strengths of 100, 1,000, and 10,000 Oersteds. In order to cover
as large a range of energies as possible, W and r are plotted on
logarithmic scales.
Equation (2.12) can easily be used to determine two other
Motion o f Charged Particles in Uniform Magnetic Field 29
4 V . _____ l_______________l______________ I | (J d \ _______ i______ ________|______ _______ l
10-1 1 10 102 103 !F (eV ) 10-2 1 0 -1 1 10 1Q2 !F (eV )
F igure 11. Dependence o f the Larinor radius on the energy:

a . electrons, b . protons.
quantities characterizing the motion of a particle, namely, the

period TH and the angular velocity u>H. It is clear that
vT h = 27jt, (2.16)
and therefore substituting for r from equation (2 . 1 2 ), we have
_ lirmc
(2.17)
H ~ qH '
The angular velocity is related to TH by
2tt
0)ff —
1H
and therefore
CO IT
H
_qJL
me
. (2.18)
The quantity is also called the angular frequency or the Larmor

frequency of the particle.
It follows from (2.17) and (2.18) that the rotational period and
the Larmor frequency are independent of the velocity of the particle,
since v does not enter into the expressions for TH and u H. The
Larmor frequency of electrons is very high and therefore the period
is very small, even in very weak magnetic fields. For example,

when H = 1 Oe, the Larmor frequency is equal to 1 . 8 X 10~ 7 cps,
which is in the radio frequency range. The directions of motion of
electrons and ions in a magnetic field are opposite. A positive ion
rotates in a clockwise direction if its path is viewed against the
magnetic field. An electron moves in the opposite direction and its
trajectory forms a right-handed system with the magnetic field.
(The trajectory can be generated by clockwise rotation about the
positive direction of the magnetic field.)
We note that a charged particle moving on a circle gives rise to an
electric current. The direction of the current due to the motion of
electrons and ions in a given magnetic field is the same because
particles of different signs move in opposite directions. Each
elementary current ring associated with the circular motion of a
charged particle is itself a source of a magnetic field (Fig. 1 2 ). It is
quite easy to show that the field due to a Larmor ring current is
always opposite to the direction of the external field producing the
circular motion. The circulating particle behaves as if it were trying
to reduce the field forcing it into the circular path. The magnitude
of the magnetic field due to the circular motion of an electron or ion
is very small. However, if the density of the charged particles is
high enough, the resultant magnetic field may appreciably reduce
the external field in the region occupied by the particles. This effect
will be discussed in detail later.
Consider now a more general case of the motion of a charged
F igure 12. Magnetic field Hi produced by a charged particle during its

motion in an external magnetic field H.
Motion o f Charged Particles in Nonuniform Magnetic Field 31
F igure 13. Motion o f a charged particle in a uniform magnetic field.
particle in a uniform magnetic field; that is, suppose that the

velocity of the particle v makes an angle a with the direction of the
lines of force, where a is less than 90°. The velocity can be resolved
into two components which are respectively parallel and perpen
dicular to the magnetic field, namely, U| | = u co saan d i;± = usina.
The magnetic field will only affect the perpendicular component
and will force the particle into a circular motion in the plane per
pendicular to the vector H. The radius of the resulting circle is
given, as before, by
mu±c _ mvc
r sin a.
qH ~ ~qH
At the same time, the particle will move with a constant velocity v 11
in the direction of the lines of force. The resultant motion is there
fore made up of a uniform rotation and a uniform translation, and
can be represented by a screw thread (Fig. 13.) The pitch of the
screw, that is, the distance traversed by the particle in the direction
of the field during a complete rotation about the lines of force, is
given by
lirmvc
v\\T ------ z~z COS a . (2.19)
qH
For small angles a the trajectory is very shallow. For values of a
approaching 90° it resembles a compressed spring.
2.6 General Nature of the Motion of Charged Particles

in a Nonuniform Magnetic Field
Both in nature and in the apparatus employed for scientific and
technological purposes, the motion of charged particles usually
takes place in nonuniform magnetic fields. In such fields the direc-
tion of the vector H varies in space from point to point. A non-

uniform magnetic field may have a very complicated geometric
form and therefore the trajectories of the charged particles may also
be very complicated and tangled. There is little hope of finding a
general law which would apply to all such trajectories. However,
the situation is considerably simplified if one considers only those
particles for which the Larmor radius is relatively small. This
means that the magnitude of r is small in comparison with the
distance over which the field changes appreciably or, in the lan
guage of physics, in comparison with the characteristic linear
dimension of the field irregularities. A particle with a small Larmor
radius does not easily “see” the field irregularities and must com
plete a large number of rotations about the lines of force before it
passes through a region in which there is an appreciable change in
F igure 14a. T r a je c to r y o f a c h a r g e d p a r tic le m o v in g in th e d ir e c tio n o f

in c rea sin g f i e l d stre n g th .
the field. The effect of field irregularities on the path of a particle of

this kind is only important if we consider a sufficiently long segment
of its trajectory.
In order to investigate such effects, consider a case where the
field strength varies in the direction of the lines of force. This is
illustrated in Fig. 14a. The field strength increases from left to
right and therefore the lines of force form a converging system (the
stronger the field the greater the number of lines per unit area).
What will be the behavior of a particle entering a region of this
kind? At first sight, it would appear that, provided r is small, the
radius of curvature of the trajectory will be inversely proportional
to the field H and therefore the pitch of the helical path will de
crease, that is, there will be a contraction in all the dimensions by
the same factor. However, this implicitly assumes that the angle a
between the velocity of the particle and the lines of force is constant.
In point of fact, this assumption is not valid because in a non-
uniform field the particle experiences a force which is always in the
direction of decreasing H. The force retards the particle as it
moves from left to right and therefore reduces the longitudinal
component v cos a. Since the magnitude of the resultant velocity in
a magnetic field remains constant, this can only result in a change
in the angle a, which increases, and therefore the transverse com
ponent v sin a also increases. It follows that the ratio of the two
velocity components is no longer constant, and the trajectory of a
particle entering the increasing field becomes more and more
curved, as shown in Fig. 14a.
In order to establish the origin of the retarding force, consider
Fig. 14b, which shows a complete turn of the trajectory in an in
creasing field (it is assumed for the sake of simplicity that the
longitudinal component of the velocity is zero). The arrow MA
represents the magnetic field at the point M on the trajectory of the
particle. This vector can be resolved into two components — one
in the direction perpendicular to the plane of the trajectory and the
other along the radius of the trajectory. In a uniform magnetic
field the lines of force are parallel and the vector H has only the
single component MB. This component gives rise to the centripetal
force F\, which maintains the circular motion of the particle. The
F igure 14b. Forces acting on a charged particle in a magnetic field with

converging lines o f force. The force F\ maintains the Larmor circulation,
whilst the force Fi acts in the direction o f decreasing field.
component MC, which is due to the divergence of the lines of force,

that is, the nonuniformity of the field, gives rise to a force whose
direction is quite easy to find with the aid of the general rule given
at the beginning of the present section. This force must be per
pendicular to MC and to the velocity, that is, to the arc of the circle
at M. This results in the force F2 which tends to accelerate the
particle towards the region of the weaker field.
The same result can also be obtained in a different way. It has
been pointed out earlier that the Larmor circle described by a
particle in a magnetic field is equivalent to a current and that this
current gives rise to a magnetic field which acts in a direction op
posite to the external magnetic field in which the particle executes a
circular motion. This means that the Larmor circle exhibits proper
ties which enable us to regard it as the analogue of a particle of
diamagnetic material, since a diamagnet is defined as a substance in
which an external field induces a field of opposite sign. This means
that diamagnets are repelled by strong magnetic fields. A “dia
magnetic” Larmor circle which is produced by a circulating charged
particle will therefore also experience a force tending to accelerate
it into the region of weaker field.
It has already been noted that, since the total velocity of a particle
in a magnetic field is constant, the deceleration, during the motion
along a line of force in the direction of a stronger field, must neces
sarily be accompanied by an increase in the transverse velocity
component. Therefore, the magnetic field H and the magnitude of
the transverse velocity component v± should be related to each

other. Detailed calculations show that this is so: The square of the
transverse component is very nearly proportional to H, and
therefore the ratio vj_/H remains almost constant. Since is Pr0‘
portional to the kinetic energy of the Larmor rotation W±, it follows
that the condition v]_/H = const is equivalent to Wx /H = const.
This approximate theoretical result has one very important con
sequence. The word “approximate” must be understood to mean
that the quantity v \ / H may undergo a large number of small
changes about a mean value. These variations lie between relatively
narrow limits, within which the quantities vj_ and H themselves
may vary very considerably. The degree to which v ± /H remains
constant is greater for small Larmor radii, that is, small v and large
H.) This ratio is often called an adiabatic invariant, which is taken
to represent the approximate constancy (in the above sense) of this
quantity. Since the ratio v]_/H is constant, the angular velocity of
the particle is proportional to the square root of the magnetic field,
and therefore the Larmor radius r = mv±c/qH in a field of varying
strength is proportional to 1 /yfH.
Since the magnitude of the resultant velocity v is constant, the
quantity sin2a /H is also an adiabatic invariant for a given particle.
Suppose that at some initial point on the trajectory the magnetic
field is H0, while the angle between the velocity and the direction of
the lines of force is ar0. The angle at some other point on the tra
jectory can be found from the relation sin2a /H = sin2a0/H 0, and
hence
H
sin a: = sinao^/T r (2 .20 )
\ o
When the particle moves toward the region of stronger field, the
angle a continues to increase until it reaches 90°. This occurs at the
point at which the right-hand side of (2.20) is equal to unity. The
magnitude of the magnetic field H at this point can be found from
1 = sinof0
which yields
( 2. 21)
sin2a 0
F ig u re 15. M a g n e tic m irro rs.
This is the maximum magnetic field which can be reached by a

particle with given a 0 since sin a can never exceed unity. Therefore,
having reached the point at which H is given by (2.21), the particle
will change its direction of longitudinal motion and will return to
the region of lower field. Therefore, high-field regions may, under
certain conditions, act as magnetic mirrors. In particular, suppose
that the field is of the form shown in Fig. 15. In this field configura
tion, the magnitude of H increases on either side of the central
region and the density of the magnetic lines of force reaches a maxi
mum on the left and on the right of this region. (The maximum and
the minimum field strengths are denoted by Hmttx and Ho, respec
tively.) Consider the motion of an ion or electron which was
originally in the central region. The eventual fate of this particle
will depend on the magnitude of a0. If sina 0 < yjH0/ H max, the
particle will be retarded as it approaches the region of maximum
field but will, nevertheless, be able to pass through it. If, on the
other hand, sin a > -\]Ho/Hmnx, the particle will be reflected by the
magnetic mirrors. The particle will be trapped in the space between
the two mirrors. This is a simple case of a magnetic trap which is
capable of retaining charged particles. A trap of this kind can be
produced, for example, with the aid of two coils which are parallel
to each other and carry electric currents in the same direction.
The intensity of the magnetic field can vary not only along the
lines of force but also in the direction at right-angles to them. Thus,
for example, the field due to a current flowing in a straight wire de
creases with increasing distance from the wire, although the neigh
boring lines of force are parallel because they are in the form of
concentric circles. The trajectory of a charged particle in a non-
F igure 16. Magnetic field due to a straight-line current.
uniform field of this kind will, in general, be of a complicated form.

However, we can try to understand the situation by considering a
number of special cases. If the velocity of a particle is perpendicu
lar to the lines of force, the motion will take the form shown in
Fig. 17. (The plane of this figure is parallel to the current of
Fig. 16.) The intensity of the magnetic field is assumed to decrease
from left to right. The plane trajectory of the particle will then no
longer be a circle since the radius of curvature given by (2 . 1 2 ) will
be greater on the right than on the left and the particle will describe
a spiral. There will therefore be a resultant motion in the direction
of the y axis which is parallel to the current of Fig. 16. After a
large number of revolutions, the situation will appear as if the
particle were moving in an expanding groove whose width depends
on the velocity of the particle and the intensity of the magnetic field.
The direction of the vertical displacement of the particle depends on
its sign. In Fig. 17, the positive ion moves from bottom to top and
covers the vertical distance ab in a complete revolution, while an
electron moves from top to bottom and the corresponding vertical
displacement is cd.
This type of motion is referred to as magnetic drift or drift in a
nonuniform magnetic field. The word “drift,” which is taken from
marine terminology, is meant to indicate that the velocity of
displacement of the particle along the groove mentioned above
is small in comparison with its Larmor velocity. We note that,
during the drift, the particle does not escape into a region of
stronger or weaker field. On the contrary, during its motion along
F igure 17. The drift o f charged particles moving in a nonuniform magnetic

field in the plane perpendicular to H.
the relatively narrow groove it behaves as if it were tending to

retain the same magnetic field within the limits of its trajectory.
Suppose now that the particle moves along a line of force in a
nonuniform field whose intensity increases at right angles to the
vector H. An example of this is shown in Fig. 18, where a positive
ion is shown moving in the direction of the field due to a straight-
line current. It is clear that the particle cannot continue along
the line of force. If its velocity is parallel to the field, the force is
zero, and therefore there should be no centripetal acceleration.
Hence an ion whose velocity at the point Mi is parallel to the
direction of the field will leave the circular line of force because
F igure 18. Motion o f a positive ion along a magnetic line o f force due to a
straight-line current.
of its inertia. During its subsequent motion, the ion will intersect
the neighboring lines of force at small angles. For example, at
the point M% the angle between the velocity of the ion and the field
vector H may be, say, a. Since the velocity of the ion is now no
longer strictly parallel to the lines of force, the particle will ex
perience a force of -\qvH sin a. In the present case, this force is
parallel to the current producing the magnetic field. As a result,
the ion is given a velocity component in the direction of the positive
y axis and this is, in fact, the drift velocity of the particle in the
nonuniform field. The situation is the same in the case of electrons,
except that they drift in the opposite direction. It is easy to show
that the drift of a particle, due to the presence of a longitudinal
component to its velocity, occurs in the same direction as the
drift of a particle whose velocity is perpendicular to H.
Calculations show that in the general case, when the particle
has both a transverse velocity and a longitudinal velocity V\\,
the drift velocity is given by
( 2. 22)
where R is the radius of curvature of the line of force passing

through a given point on the trajectory and wh = qH/mc is the
Larmor frequency. The direction of the drift velocity ud is per
pendicular both to H and to the radius of curvature of the line of
force. In the language of vector analysis, this means that when
q > 0 , the drift-velocity vector u d is parallel to the vector product
H X R.
It follows from the above discussion that the direction of the drift
velocity is always such as to oppose any change in H. Equation
(2 .2 2 ) and all the other formulas discussed in this section are
strictly valid only when the field is not highly nonuniform, that is,
when the change in the magnetic field within the limits of a Larmor
circle is small. Under these conditions, ud <3Cu; that is, the drift
motion contributes a relatively small correction to the velocity
of the particle. It should be noted that there is one further con
dition which limits the range of applicability of Equation (2.22):
A particle moving in the space between the conductors giving
rise to the magnetic field must not enter the conductors, which is
equivalent to saying that the particle must not enter the region in
which a current is flowing. When this condition is not satisfied,
the equation for the drift velocity becomes more complicated,
but all the qualitative effects described above remain. When the
drift takes place across the lines of force, both velocity components,
Un and Ujl, remain constant in magnitude. In view of the above
analysis, the magnitude of H along the drift groove is also constant.
It follows that v \ / H = const and this means that the drift motion
does not affect the adiabatic invariance of v]_/H.
In order to obtain a clearer picture of the motion of a charged
particle in a slowly varying (in space) magnetic field, we note that
this motion may be resolved into two simpler component motions,
namely, the Larmor motion with a velocity v± = v sin a and a
displacement of the center of the Larmor circle. The latter, in
its turn, can be resolved into motion along the lines of force with
the velocity Un = ucosa and a slow drift at right-angles to the
lines of force with the velocity ud.
The general situation may be illustrated by a number of simple
examples. One of these is the motion of a charged particle in the
field of a straight conductor carrying a current. The motion of a
charged particle in the Earth’s magnetic field may be taken as
another example. The drift motion in the field of a straight-
line current may be described with the aid of Equation (2.22).
We shall confine our attention to the case when a = 90° and,
therefore, Uj_ = v. Under this assumption, Equations (2.22) and
(2 . 1 2 ) yield
Ud= vr
where r is the Larmor radius of the trajectory of the particle and R
is the radius of curvature of the line of force. In the present case R
is equal to the distance from the current. Suppose, for example,
that the current is equal to 1 0 5 amperes and a 1 0 0 electron-volt
proton is at a distance of 10 cm from the conductor. We thus
have r/R = 7 X 10~ 2 and therefore ud is quite small in comparison
with v, and is equal to 1 X 106 cm per sec. Thus, the particle
F igure 19. Schematic representation o f the earth's radiation belts:

A — inner belt, B — outer belt.
drifts slowly, parallel to the straight current, at a constant distance

from it.
The other case is represented in nature by the radiation belts
surrounding the Earth. The existence of these belts was first
established during the flights of the first satellites and space
rockets. They consist of electrons and ions of high energy (electron
energies up to several million electron-volts, protons, and light
nuclei with energies higher by two or three orders of magnitude).
These particles are trapped in the Earth’s magnetic field. The
disposition of the radiation belts in the Earth’s neighborhood is
illustrated schematically in Fig. 19. As can be seen, two belts
can be distinguished, although, in fact, there is no clear dividing
boundary between them. The inner belt is at an average distance
of a few thousand kilometers from the Earth, even though its
inner boundary approaches the Earth to a distance of about 500
km. The outer radiation belt begins at a distance of 10,000 to
15,000 km from the Earth. The precise origin of the particles
forming these belts is still not certain. Processes connected with
cosmic rays reaching the Earth from outer space, and the streams
of plasma ejected by the Sun during the solar flares, are probably
responsible for these particles. The charged particles in the
radiation belts execute a motion in the Earth’s magnetic field
which is of the form illustrated in Fig. 20. 4 For the sake of clarity,
4The trajectory shown in this figure represents the path of a high-energy ion.
F igure 20. Schematic representation o f the trajectory o f a charged particle

in the earth's magnetic field.
the Larmor rotation is indicated by the broken line, while the

solid line represents the “axis” of the trajectory, that is, the
motion of the center of the Larmor circle. An electron or ion
moving in the Earth’s magnetic field is therefore expected to be
reflected at the magnetic poles, where the field strength is highest,
and will therefore execute an oscillatory motion between the
magnetic poles. In addition, there is a drift due to the nonuniform
nature of the field. In view of the above analysis, this drift should
give rise to a displacement of the particles in a direction per
pendicular to the magnetic meridians, that is, from East to West
in the case of electrons and from West to East in the case of ions.
Each particle will travel around the globe while at the same time
executing a rotational motion about the magnetic axis of the Earth.
The lifetime of the individual electron or proton in this gigantic
magnetic trap is not infinite. In the inner radiation belt, the only
significant particle losses are due to collisions with atoms in the
outer layers of the Earth’s atmosphere. In the outer radiation
Motion o f Particles in Electric and Magnetic Fields 43
belt, charged particles may be lost as a result of changes in the

geometric structure of the Earth’s magnetic field, that is, in the
form of the lines of force. These changes occur whenever a large
mass of ionized gas, ejected by the Sun during a short-period
flare, reaches the Earth’s neighborhood. In the case of the radia
tion belts, we are dealing with a rather special form of plasma
which occupies an enormous volume. In point of fact, it follows
from Equation (1.3) that for linear dimensions of the order of
10,000 km and particle energies of the order of 106 eV, an ensemble
of electrons and ions will assume typical plasma properties at
concentrations as low as one fast particle per cubic meter. The
true density of fast electrons or protons in the radiation belts is
higher than this figure by at least several orders of magnitude and
this means that the Debye radius should not exceed 1,000 km.
2.7 Motion of Particles in the Presence of Electric,

and Magnetic Fields
Consider now some special cases of the motion of charged
particles in combined electric and magnetic fields. To begin with,
let us suppose that both the magnetic and the electric fields are
uniform. If the electric field is parallel to the magnetic field, it
will accelerate or decelerate the particle independently of the
magnetic field. At the same time, the motion of the particle in
the plane perpendicular to H will be unaffected by the presence
of the longitudinal electric field. It follows that under these
conditions the particle trajectory will be either a gradually extending
or gradually contracting helix (depending on the relative directions
of the electric field and the initial longitudinal velocity component).
This is illustrated in Fig. 21.
F igure 21. Motion o f a charged particle in parallel electric

and magnetic fields.
F igure 22. Motion o f a charged particle in crossed fields. The magnetic

field is at right angles to the plane o f the drawing, the electric field is in the
upward direction. The trajectory is a cycloid.
The situation is quite different when H is perpendicular to E.

Suppose that at some initial instant of time the particle is at the
point 0 (Fig. 22) and its velocity is equal to zero. Under the action
of the electric field it will be accelerated in the direction of the y
axis. (In order to be specific, we are assuming that the particle is
positively charged.) The force acting on the particle due to the
magnetic field will increase with increasing velocity, since the
force is proportional to the velocity. This force will deflect the
particle toward the * axis. The gradual curving of the trajectory
should eventually give rise to a reverse motion in which the particle
will move in the direction of the negative y axis. This occurs
at the point A in Fig. 22. Between A and B the particle will be
retarded by the electric field and the velocity will become equal
to zero when the particle returns to the .v axis. The acceleration
process will then begin again and the whole cycle will be repeated
indefinitely.
Therefore, the trajectory of the particle will consist of periodically
repeating, identical arcs. Each arc is identical in form with the
well-known curve called the cycloid. Calculations show that the
height of the cycloid is h = 2mc2E/qH 2, while the time taken by
the particle to complete one cycloid loop is 2wmc/qH; that is, it is
equal to the Larmor period. A negatively charged particle will
also follow a path consisting of repeated cycloids (see the broken
curve in Fig. 22). We note that the direction of displacement
along the axis is the same for particles of either charge.
Cycloidal trajectories of this type have long been known in
F ig u re 23. T ro c h o id a l m o tio n o f p a r tic le s in c r o s s e d fie ld s .
mechanics. In fact, the cycloid is the path described by a point

on the wheel of a railway wagon as it moves along the track
without slipping. This analogy suggests that the motion of a
charged particle in perpendicular electric and magnetic fields can
also be resolved into rotational and translational motion. In point
of fact, it is easy to show that the motion of an ion or an electron
along a cycloid can be resolved into a rotation on a circle of
radius \h and a translation along the x axis with velocity cE/H.
The velocity v of this translational motion is the same for all
particles whatever their mass and charge. It is perpendicular
both to E and H and is therefore parallel to E X H. This result
may be generalized by considering the motion of a particle which
starts off with a certain initial velocity v0. Detailed calculations
show that, in the more general case, the particle will travel on a
trajectory which is one of the family of curves illustrated in Fig. 23
and known as trochoids. The trochoidal motion can also be
resolved into a rotation and translation along the x: axis, and
it turns out that the velocity of the translational motion is
always equal to cE/H. Therefore, in crossed, that is, mutually
perpendicular, electric and magnetic fields, a charged particle
undergoes both a helical motion under the action of H and an
electric drift at right-angles to the magnetic lines of force.
Usually, the electric field both in laboratory experiments and
in nature is so small that the rotational Larmor velocity of the
particles is much greater than the drift velocity. The effect of the
transverse electric field on the energy of the particle can then be
neglected, and therefore the adiabatic invarience of v]_/H remains
valid even when the particles move in crossed electric and magnetic
fields. If the magnetic field is not strictly uniform, then in ad

dition to the electric drift there should also be a drift due to the
variation in H . The resultant drift velocity will, of course, be equal
to the sum of the two drift velocities. It should be noted that the
drift which is found to take place in crossed fields is a special case
of a more general law, according to which any constant force
acting on a charged particle in the direction perpendicular to H
gives rise to a drift at right angles to the magnetic lines of force.
The direction of this motion is perpendicular to the force acting
on the particle. The magnitude of the drift velocity is equal to
cF/qH, where F is the force and q is the charge of the particle.5
If the force F has the same direction both for ions and electrons,
then their drift velocities will be opposite to each other. Fig. 24
shows the drift of charged particles in the gravitational field of
the earth in the case where the magnetic field is parallel to the
earth’s surface. Although the ion drift velocity is much greater
than the electron drift velocity, the absolute magnitude of the
former is still quite small, even for heavy ions. In a magnetic
field of 1 Oe, the drift velocity of a singly charged molecular ion of
oxygen is about 3 cm per sec.
A further possible situation occurs when the magnetic field
varies both in space and in time. We shall consider two simple
examples of the motion of ions and electrons in variable magnetic
fields. These examples will illustrate certain general regularities.
F igure 24. Drift o f charged particles in a gravitational

field perpendicular to H.
5This expression for the drift velocity can be easily derived if it is recalled that
the effect of a force F on the given particle is identical with the effect of an equiva
lent electric field E, where F = qE.
F igure 25a. Motion o f a charged particle in a uniform magnetic field

which increases with time. The lines o f force are perpendicular to the plane
o f the drawing, b. Variation in the velocity and Larmor radius due to an
induced electric field.
In the first example, the charged particle is located in a long,

cylindrical chamber inside a solenoid producing a uniform mag
netic field (Fig. 25a). Suppose that this field increases with time,
and the problem is to determine the motion of the particle. The
first point to note is that, in accordance with the law of induction,
an electric field is necessarily associated with a time dependent
magnetic field. The lines of this induced electric field will in the
present example take the form of concentric circles, whose centers
lie on the axis of the chamber. It is possible to show from the law
of induction that the electric field strength E is proportional to the
distance from the axis and the rate of change of H, that is, the
increase in H per unit time. The electric field gives rise to a drift
motion, and the drift velocity is perpendicular to both E and H,
and therefore takes place in a radial direction. We can now use
the various rules which we established above, in order to show that
as H increases the drift occurs towards the axis, while a decreasing
H is associated with an outward drift of the particles. Calculations
show that as a result of the drift, the distance of the particle, or
more precisely the distance of the center of its Larmor circle from
the axis, is proportional to y]H. However, this is not the only
I//// /////////////////////A
F ig u r e 2 6 . A charged particle in the space between
moving magnetic mirrors.
effect of the induced electric field. There is also a change in v± and

therefore in the kinetic energy of the Larmor motion. The reason
for this effect can easily be established by considering the special
case of a particle for which the center of the Larmor circle lies on
the axis of the chamber (Fig. 25b). In this case, the induced
electric field is always parallel to the rotational velocity. Therefore,
it will either tend to increase the perpendicular component of the
velocity or to retard the particle. The former occurs in the case
of increasing H and the latter in the case of decreasing H. A
quantitative analysis shows that the kinetic energy associated
with the rotation of the particle is proportional to H; that is, the
quantity W J H is again constant. This result is valid for any
particle in a variable magnetic field and can be extended to all
cases involving time dependent H , provided the variation in the
magnetic field is sufficiently slow.
The second example involves a charged particle with a given
initial value of i»n trapped between two magnetic mirrors produced
by the short coils A and B shown in Fig. 26. It is assumed that
the field due to these coils is strong only in their immediate
neighborhood and that in the region between the coils the particle
moves in a relatively weak constant magnetic field produced by
the long coil extending from A to B. Suppose now that the two
short coils are moved towards each other. At first sight this should
Motion o f Particles in the Electric and Magnetic Fields 49
/
/
/
/ Uh + « Q
/.
/.
/
/ vu+ «
/
/A
u X
F ig u r e 27. Reflection o f a mass point from a moving wall.
merely give rise to a reduction in the region in which the longitu

dinal oscillations of the particle take place. However, in reality,
as the coils approach each other, there is a gradual increase in the
longitudinal velocity of the particles.
This increase in un may be explained in two quite different
ways. First, the motion of the two short coils towards each other
produces a time dependent magnetic field. Thus, if at an initial
time the field at the point M was H0, then as the coils approach
each other, the field will rapidly increase to a much greater value.
However, the increase in H gives rise to an induced electric field
which will exist near the two short coils. As the charged particle
approaches the coil during its motion along the x axis it is retarded
and its longitudinal component is transformed into the transverse
component. The latter component also increases as a result of the
effect of the electric field. This means that the total kinetic energy
must increase. When the particle is reflected from the mirror, the
increase in the kinetic energy associated with the transverse
motion is transformed into an increase in the energy associated
with the longitudinal motion, and this results in an increase in U||.
A similar result may be obtained by noting that the process involved
in this particular case is analogous to the elastic reflection of a
mass point from a moving wall. This will be understood from
consideration of Fig. 27. On the left of this figure, the collision of a
mass point Q with the wall is shown relative to the laboratory
system of coordinates, whilst on the right it is shown relative to a
system of coordinates which moves together with the wall. In

the system of coordinates which is attached to the wall, everything
occurs in accordance with the usual well-known rules: the particle
is reflected and its velocity before and after the reflection is the
same. Therefore, in this reference frame the particle will have a
velocity V\\ + u after the collision with the wall, where u is the
velocity of the wall. When the event is considered relative to the
laboratory system of coordinates, it is found that the velocity of the
particle moving to the left after reflection is equal to i>n + 2u;
that is, it is increased by twice the velocity of the wall as compared
with the initial value. One would therefore expect that the velocity
of the charged particle trapped between the two magnetic mirrors
discussed above will gradually increase. It may be shown that in
fact U|| will be inversely proportional to the distance between the
mirrors, whatever the magnitude of H.
We thus see that under certain conditions, a time dependent
magnetic field may affect both the transverse and the longitudinal
velocity components of charged particles. The true physical
reason for the change in the velocity is always the same, namely,
the acceleration of the particles in electric fields which are always
associated with time dependent magnetic fields.
The above mechanisms which are available for the acceleration
of charged particles are being currently used in various experimental
studies whose aim is to produce high-temperature plasma. By
radially contracting the plasma with the aid of an increasing
magnetic field and then compressing it with movable magnetic
mirrors, we are merely copying the phenomenon which occurs
on a gigantic scale in nature. It is well known that streams of
particles of very high energy, known as cosmic rays, reach the
earth from outer space. Even now, there is no direct experimental
evidence concerning the origin of these particles. However, the
fact that they have such high energies is no longer puzzling because,
as we have seen, it is possible that they are accelerated during
interactions with the magnetic fields of stars and nebulae as they
wander through space for a very considerable time. The dis
tinguished Italian physicist Enrico Fermi, who contributed so
much to modern physics (in particular, he built the first nuclear
Motion o f Particles in the Electric and Magnetic Field 51
reactor), pointed out in 1949 that cosmic ray particles may be

accelerated by colliding with moving magnetic “clouds,” that is,
regions of space in which there are nonzero magnetic fields. Astro
nomical data suggest that such clouds wander randomly through
the galaxy, giving rise to tangled up clusters of magnetic lines of
force. When a charged particle collides with a magnetic cloud
moving towards it, its energy increases. It is quite easy to show
that a collision with a receding cloud gives rise to a reduction in the
energy. At first sight it would appear that the two effects would
cancel out. However, Fermi was able to show that a complete
compensation does not, in fact, take place, and that the velocity
received by a charged particle moving towards a magnetic cloud
is greater than the velocity received during a recessive collision.
Therefore, during its very long random motion in the interstellar
space, a charged particle will gradually attain greater and greater
velocity, and its energy may reach many billions of electron volts.
This mechanism for the acceleration of cosmic rays, which was
first put forward by Fermi, may be augmented by the induction
mechanism which leads to an increase in the final velocity of
particles in longitudinal fields which are increasing with time.
3
The Motion of Charged

Particles in Plasma
3.1 A General Review of the Properties of Gases

In this and in the subsequent chapters, we shall make maximum
use of the analogy between a plasma and an ordinary gas. To
begin with, let us recall briefly the main properties of gases.
The most important parameters which characterize the physical
state of a gas are the density n (the number of molecules or atoms
per unit volume) and the temperature T. The mean kinetic
energy of the particles depends only on the temperature [see
Equation (1.2)], while the gas pressure is proportional both to the
density and the temperature. This is an expression of the laws of
Boyle and Gay-Lussac.
The formula relating the pressure p, the density n, and the
temperature T is
p = nkT. (3.1)
This equation is valid for any gas at a sufficiently high temperature,
provided the density n is not too great. When the gas consists of a
mixture of components with densities /q, nj, and so on, each of
the components gives rise to its own partial pressure, and thus,
for a mixture of gases, Equation (3.1) may be written in the form
P = Pi + P2 + • • • = (m + n2 + • • •)kT, (3.2)
where pi, p i, . . . are given by
pi = riikT, pz = nzkT, . . .
52
A General Review o f Properties o f Gases 53
It is, of course, assumed that the temperatures of the components

of the gas mixture are the same. In practice, this condition is
always satisfied in a neutral gas because there is a rapid energy
transfer between molecules belonging to the different components.
Consider the simplest case when the gas molecules are all of the
same kind. Even when the temperature of the gas is maintained
at a constant value, this does not mean that all the molecules
have the same velocities and therefore the same energies. On the
contrary, both theoretical calculations and direct measurements
show that the molecules may have very different velocities, lying
between zero and very large valeus. In the middle of the nineteenth
century, Maxwell showed that the number of gas molecules which
have velocities lying between v and v + Av is given by
N = Av2e~mvy2kTAv, (3.3)
where e is the base of natural logarithms (e = 2.718). The
coefficient A can be determined from the condition that the total
number of molecules per cubic centimetre must be equal to the
density n. It is found that A is proportional to n and is a function of
both the mass m of the molecules and the temperature T.6
The velocity distribution is illustrated in Fig. 28, in which the
relative number of molecules per unit velocity interval is plotted
F ig u r e 2 8 . Maxwellian velocity distribution.
6T h e coefficient A is given by the form ula A =

54 Motion o f Charged Particles in Plasma
as a function of the velocity v. A characteristic feature of this

distribution is the maximum at v = yjlkT /m (this is the most
probable velocity), and the very rapid decrease in the number of
molecules at low velocities. Calculations show that for the
Maxwellian velocity distribution the relative number of molecules
having velocities in excess of five times the most probable velocity
is only about 3 X 10" n %. The kinetic energy distribution can
easily be deduced from the velocity distribution. It can be shown
that the number of particles with energies between W and W + A W
is proportional to ■fWe~wlkTAW, provided A IT is small enough.
We note that the individual properties of the molecules do not
enter into the expression for the energy distribution, and therefore
the energy distribution will be the same for all the components of a
mixture.
If we confine our attention to one of the molecules of the gas
and follow it in its random thermal motion, we shall find that
its path will take the zig-zag form illustrated in Fig. 29. Each
break in this curve is the result of an elastic collision of the par
ticular molecule with some other molecule of the gas. The dur-
F igure 29. B row n ian {ra n d o m ) m o tio n o f a g a s m o lecu le.

ation of each collision is very small in comparison with the time

between successive collisions. For the sake of simplicity we shall
assume that aft£r a collision the particle can, with equal probability,
continue in any direction, irrespective of its initial direction of
motion. The elementary kinetic theory of gases, which provides
an explanation of most properties of gases, is based on this simple
assumption. We shall now examine the various concepts used
in this theory which, in a somewhat modified form, are widely
used in plasma physics.
The mean length of the straight-line sections making up the
zig-zag path of the molecules in the gas is known as the m ean fr e e
p a th of the molecule, and is usually denoted by X. The average
time taken by a molecule to traverse these straight-line sections is
known as the mean-time between collisions and is usually denoted
by r . The mean collision frequency, that is, the average number of
collisions experienced by the particle per second, is then 1 / r and
is denoted by v. The parameters X, r , and v can be related to the
characteristics which determine the collision process between the
molecules by introducing the concept of the effective collision
cross section. The effective collision cross section has a simple
geometric interpretation. A collision will occur only if the centers
of the colliding molecules are at a certain minimum distance
from each other. For example, if the molecules behave as if they
were rigid spheres of radii a, collisions will occur when the distance
between the centers is less than 2a. This quantity is called the
effective collision radius and the quantity 4wa2 is called the effective
collision cross section. We shall denote it by a. The idea of an
effective collision cross section is useful because it leads to the
following interpretation: A collision may be regarded as the
impact of a molecule on a target of definite dimensions. The
larger the target, that is, the larger the magnitude of 4wa2, the
more frequent the collisions between the molecules, and the smaller
the mean free path (other things being equal). It is evident that the
mean free path X will also depend on the density of the molecules.
Thus, the larger the density n, that is, the larger the number of
molecules per unit volume, the smaller the magnitude of X. The
algebraic relation between the effective collision cross section and
• •
• •
#• •
F igure 30. Passage o f a molecule through a thin layer o f gas: a. lateral

view, b. front view (molecule moving into the plane o f the paper).
the mean free path can be derived by considering Fig. 30a. This
figure shows the passage of a molecule A through a very thin
layer of gas which we can imagine as existing in space. The path of
the molecule is indicated by the dashed line. It is perpendicular
to the surfaces of the layer. In order to simplify the calculations,
let us suppose that all the molecules in the layer, except for the
molecule A , whose fate we are investigating, are at rest. If the
thickness of the layer d is sufficiently small, the probability that
the collision will occur is d/X . Thus, for example, when d / \ = 0.01,
only one out of a hundred molecules passing through the layer
will, on the average, experience a collision within this layer.
On the other hand, the probability of a collision depends on the
number of molecules in the layer and on the magnitude of the
effective collision cross section a. Fig. 30b shows the gas layer
as seen by the incident molecule A . Circles of diameter 2a are
drawn around each of the target molecules. Whenever an incident
molecule enters one of these circles we say that a collision has
taken place. In order to determine the probability of a collision
of this kind, we must find the relative area occupied by these targets
Suppose that the area of the layer shown in Fig. 30b is S so that
the volume occupied by the layer is Sd. The total number of

molecules in the layer is nSd and the total area occupied by the
target circles is nSda. The ratio of this area to the total area of the
layer S is, in fact, the required collision probability, which is
therefore equal to nda. However, it was shown above that this
probability is also d /\. Therefore d / \ = nda and hence
This expression was derived on the assumption that all the mole
cules but one are at rest. This assumption is clearly invalid and
the above formula must be corrected for the fact that the target
molecules are also in motion. If it is assumed that all the molecules
behave as if they were perfectly rigid spheres, then the expression
relating X, n, and a is found to be
X = _ L .J_ (3.5)
■ y j2 n < r
Thus, when the motion of the molecules is taken into account,

the mean free path is reduced by a factor of about 1.4. The mean
time between successive collisions can then be found from
where v is the mean velocity of the molecules which is very nearly

equal to the most probable velocity.7
The mean number of collisions which a molecule experiences
per second is equal to v /\. If all other molecules are assumed to
be at rest, then
v = nv<r. (3.7)
When the motion of the target molecules is taken into account,

this formula will include an additional factor of V2* For a 8as
7This is not quite true since the mean of a quotient is not equal to the quotient
of the means. However, this mathematical distinction is not of particular
importance in the present context.
consisting of neutral atoms and molecules, the cross section u can,

to a first approximation, be taken as constant, that is, independent
of velocity. It follows that the mean free path X is also independent
of velocity and therefore of temperature, and varies only with the
density n. Table 1 gives some typical values for the parameters
<j and X. In the case of nonatomic gases, for example, helium or
argon, the collision frequency and the mean free path refer to
collisions between atoms. In the case of molecular gases (hydrogen,
nitrogen, oxygen) these parameters refer to collisions between
molecules. The mean free paths are given for n = 3.5 X 1016
particles per cm3 which corresponds to a pressure of 1mm Hg at
room temperature (T = 300°K).
T able 1
Gas a (cm2) X( X 10 3cm)
hydrogen 2.4-10"' 5 8.4

helium 1.510-15 13.5
oxygen 4.1 10-15 5.0
nitrogen 4.4-10-16 4.6
argon 4.2-10-16 4.8
3.2 Collisions between Charged Particles in Plasma

After this brief review of the behavior of atoms and molecules
in a neutral gas, let us proceed to the laws governing the motion
of ions and electrons in plasma which, by definition, is a gas
consisting (if only partly) of charged particles. It seems quite
natural to discuss the effects of the interactions between the
electrons and ions on the nature of their motion in terms of the
concepts which have been so successful in the conventional kinetic
theory of gases. In particular, this means that we can use the
concepts of the effective collision cross section, mean free path,
and so on. However, in the case of plasma the use of these param-
Collisions between Charged Particles in Plasma 59
eters leads to certain difficulties which are directly connected

with the particular nature of the interaction between the plasma
particles. In contrast to the forces acting between neutral mole
cules, which have a very short range of action and become appreci
able only when the particles approach each other to within
10“8 to 10-7cm, the Coulomb forces between the ions and electrons
have a very long range and are appreciable even at large distances.
(They are inversely proportional to the square of the distance.)
Therefore, the path of an ion or an electron in plasma will be
quite different from the zig-zag curve shown in Fig. 29. The path
of a charged particle in a highly ionized plasma, in which the
electrostatic interaction between the particles is the most important
interaction, cannot be resolved into straight line sections which
begin and end at the points at which the collisions take place.
In plasma, each individual particle is always in the field due to the
remaining electrons and ions, which is subject to continuous
fluctuations both in magnitude and direction (see Fig. 2). This
plasma microfield gives rise to a continuous variation in the
magnitude and direction of the velocity of a charged particle in
the plasma; and this variation takes place continuously since the
intensity of the microfield is on the average quite small.
The order of magnitude of the microfield E can easily be
estimated as follows:
If it is assumed that the ions are singly charged, then the total
number of charged particles of both signs in a cubic centimetre is
2ne, where ne is the number of electrons per cm3. Therefore, the
volume occupied by a single particle is 1/2ne, and the mean
distance between neighboring particles is equal to \ / \ / 2 n e. The
field at a point halfway between the particles, that is, at a distance
1/2 -\f2 n e from one of the electrons or ions in the plasma, can be
taken as a measure of the intensity of the microfield. From this,
it can be shown that the mean intensity E of the internal electric
field is of the order of e fr2 = 6en^13. More rigorous calculations
show that the mean field is of the order of 20en*/3.
Because the changes in the direction of the velocities are statistical
in nature, the trajectory of the particle will resemble the random
F igure 31. Trajectory o f a charged particle in plasma.
walk of a person lost in a desert at night. A trajectory of this

kind is illustrated in Fig. 31.
We shall now try to analyze the motion of particles in a plasma
in terms of the known two-body interactions between the particles.
We shall try to show that the gradual change in the direction of
the velocity is the result of a large number of very weak interactions.
These interactions occur through the electrostatic field acting
between the particles. The angular deflection i? experienced by an
electron or an ion in an elementary collision is given by Equation
(2.5) provided & is not too large. It follows from this equation
that the scattering angle is inversely proportional to the impact
parameter b. In order that the particle should be deflected through
an angle greater than d it must strike the scattering center within
an area of radius b. The area of this circle is r b 2 and is inversely
proportional to i?2. It follows that collisions giving rise to large
scattering angles have a very low probability, while small angles
of deflection occur very frequently. A detailed theoretical analysis
shows that it is these very frequent small deflections which combine
in accordance with statistical laws to give rise to the gradual
variation in the direction of the moving particle as illustrated in
F *g - 31 •
One further point is worth noting here. Since the plasma
contains charged particles of two kinds, namely, electrons and
ions, there are, in fact, three kinds of electrostatic interactions:
electron-electron, electron-ion, and ion-ion interactions. In this
chapter we shall be interested mainly in the behavior of the plasma

electrons. The motion of each electron is affected both by the
Coulomb field due to the ions and the Coulomb field due to the
other electrons. Both interactions play an approximately equal
part in the continuous variation in the direction of the velocity of
an electron during its motion. It is not possible to determine,
from a consideration of the trajectory of an electron, which of the
two interactions is responsible for any particular small deflection.
However, it will be seen later that, in many plasma processes, the
electron-ion and the electron-electron interactions play quite
different roles. It will therefore be useful to discuss the two inter
actions separately. The electron-ion interaction will be character
ized by the quantity rei which can be formally defined as the mean
interval of time between successive collisions between electrons
and ions. This definition may be justified as follows: Suppose, for
the moment, that the plasma consists of ions only, except for a
test electron executing a random motion in the space occupied
by the positive ions. The test electron will start its motion in a
certain definite direction. After a small number of successive
deflections in the electric fields of the ions, the test electron will
“forget” its original direction of motion. Roughly speaking, this
means that the direction of its velocity will be deflected through
an angle of the order of 90°. The average time which is necessary
for this to happen is the quantity rei. The mean time between
electron-electron collisions can be defined in a similar way by
ignoring the collisions with the ions. We shall denote this quantity
by ree. It must also be remembered that, when the plasma is not
fully ionized, the electrons will also experience collisions with the
neutral atoms. These can be characterized by r e0 which can be
defined in analogy with rei. The quantity re, which represents all
interactions and is therefore the mean interval of time between
successive collisions experienced by an electron in the plasma,
can be expressed in terms of rei, ree, and Te0 and is given by
- = —+ —+ — (3.8)
In point of fact, \ / tc = ve is the total number of “collisions”

undergone by the electron per second, and must clearly be equal

to the sum of the electron-ion, (yei = 1/ r ei), electron-electron
{vee = 1/ r ee) and electron-atom (t>e0 = l / r e0) collisions per second.
The concept of the mean free path can also be used to describe
the motion of electrons in plasma. It is defined as the distance
of motion necessary for memory of the intial direction of motion of
the electron to be lost. The mean free path Xe and the mean time t c
are related by Xe = vere, where vc is the mean thermal velocity of
the electrons. In addition, it is possible to introduce the quantities
Xei, Xee, and Xrf which characterize the mean free paths of the
electrons with respect to the various forms of interaction.
The next step in our analogy between plasma processes and the
kinetic theory of gases should be the introduction of the concept
of an effective Coulomb collision cross section. As before, we
must distinguish between electron-ion, electron-electron, and elec
tron neutral atom interactions, that is, we must distinguish between
the cross sections aeh <ree, and <re0. It must be recalled that each
of these is used to replace the effect of a large number of small
deflections by a single “collision” which is analogous to a collision
between two spheres. The derivation of the expressions for the
effective cross sections is beyond the scope of this book, but it
will be useful to discuss certain approximate ideas which will pro
vide us with an insight into the exact expressions.
According to Equation (2.8) the impact parameter corresponding
to a collision between two charged particles, in which the velocity
vector of one of the particles is turned through 90°, is given by
b0 — gigi (3.9)
m ,v f
where the subscript i refers to the particle which has been scattered
through 90°. For the sake of simplicity we shall suppose that the
other charged particle (represented by the subscript 2 ) has a much
greater mass and simply acts as a center of the electrostatic force.
Let us draw a circle of radius b0 about this center. Its area will be
•S' = {q\q-i!m^)1. If we set qx = e and q2 = Ze, then
Z 2e4
5 = 7T (3.10)
m\v\
The area given by this expression can be interpreted as the effective

electron-ion interaction cross section if mi and V\ represent the
mass and the velocity of the plasma electron respectively, and Z
represents the number of elementary charges carried by the ion.
This expression was derived on the assumption that distant en
counters corresponding to large impact parameters, in which the
electron is deflected through very small angles, can be ignored.
However, detailed analysis shows that in reality the overall effect
of these small deflections is much greater than the effect of the
rare close encounters in which there is a sudden change in the
direction of motion of the electron. The reason for this is that the
weak interactions are much more frequent than the strong inter
actions. It follows that Equation (3.10) is, in fact, an underestimate
of the effective cross section. Exact calculations show that the
true cross section is greater by an order of magnitude as compared
with Equation (3.10).8 In approximate calculations using <rei one
can use the expression
4 - 10~5 Z 2
- t2 '> (3-H)
which gives the mean value of <re; taking into account the fact
that the electron velocities are not equal, but are distributed in
accordance with Maxwell’s law. This is the reason why the
denominator involves the square of the electron temperature.
The expression for <jee is analogous in form:
, m « 6 - 1 0 - 5-JL. (3.12)
The mean free path for electron-ion collisions is related to aei by

an expression which is analogous to (3.4):
2 5 - 104T 2
\ ej Z,D U ' (3.13)
Z 2n
Moreover,
4-10 ~2T l12
T„i « 25 niz
JV1
i (3.14)
Z 2d h ’ e
^However, the energy dependence of the effective cross section is given correctly
by Equation (3.10).
The symbol = in these formulas indicates their approximate

nature.
The interaction between charged particles in plasma can therefore
be described within the framework of the ordinary kinetic theory
of gases by replacing the continuously varying trajectories of the
electrons by certain conventional broken lines, and by combining
the statistical effect of a large number of weak interactions into
one strong interaction. The advantage of this not very rigorous
analysis is that Equations (3.11)-(3.14) provide a convenient
conceptual picture of what happens in a plasma in which the
electrons and ions are ascribed properties analogous to those of
ordinary billiard balls.
Moreover, the formulas given above exhibit the most important
properties which characterize the behavior of the constituent
particles in plasma. In particular, it is evident that the cross
sections aei and aee decrease rapidly with increasing electron
temperature. A few specific examples will show the relative
importance of the Coulomb interactions in cold and hot plasma.
Suppose that «,• = 1014 particles per cm3 and Z = 1. These
numbers correspond to a hydrogen plasma of very high density.
At Te = 10,000°K, which corresponds to a mean electron energy of
about 1.5eV, Equations (3.11), (3.13), and (3.14) yield
<rel ~ 4 - 10-13 cm2
\ ci ~ 0.03 cm
rei ~ 4 - 10-10 sec
vet ~ 2.5-109 sec"1.
When Te = 108°K, which is the goal of physicists concerned with
controlled thermonuclear reactions, the corresponding values are
crei — 4 - 10“21 cm2
\ ei ~ 3 • 106 cm
r ei ~ 4 - 10-4 sec
vel ~ 2.5 • 103 sec- 1.
Comparison of these special cases will show the very considerable
effect which a change in the temperature of the plasma has on the
collisions between the charged particles. Thus, in a low-tempera
ture plasma (we shall not refer to matter at a temperature of
10,000°K as cold plasma), the electron experiences about ten collis-
ions per centimetre, while in a plasma at a temperature of 108°K, and

with the same density, the electron will traverse tens of kilometers
without experiencing any appreciable interaction with its numerous
neighbors. This means that the Coulomb interactions can be
practically ignored in the case of a high-temperature plasma.
The interaction between the charged particles leads not only to a
change in the direction of their velocities, but also to energy
transfer between them. Suppose, for example, that a fast particle
having a large kinetic energy passes by another parricle which
may be regarded as the center of force (Fig. 32). As a result of
the encounter between the two bodies, the incident particle ex
periences a small deflection. This means that there is a change in
the angular momentum (moment of momentum) of the incident
particle. According to the basic laws of mechanics, the total
angular momentum of a system of two interacting charges must
remain the same before and after the collision. Therefore, the
second particle, which acts as the center of force, will receive an
impulse in the direction indicated by the arrow in the figure, and
will be given a velocity u3; that is, a fraction of the kinetic energy
of the incident fast particle will be transferred to the target particle.
This fraction will increase with decreasing distance between the
particles and decreasing mass of the particle to which the energy is
transmitted, so that for a given impulse, the energy given to the
second particle is inversely proportional to its mass.9 Therefore,
for the same initial energy, a fast ion will transfer to an electron
F igure 32. Collision o f a fast particle with a target particle at rest:

y, and v2 are the velocities o f the fast particle before and after the collision;
i>3 is the velocity o f the target particle after the collision.
9Since the kinetic energy is given by W = %mv2 and the momentum by p = mv,
it follows that W = p2l1m.
at rest much more energy than a fast electron will transfer to an

ion at rest. An electron with kinetic energy considerably greater
than the mean kinetic energy of the ions surrounding it, will
traverse a very long path and will change its direction of motion a
great many times before it gives up its energy to the ions. The
amount of energy which is transferred from a fast electron to the
ions per unit time decreases with increasing velocity of the electron.
This is explained by the fact that, in accordance with Equation
(2.5), the angle of deflection decreases rapidly with increasing
velocity of the incident particle, and therefore there is a rapid
decrease in the amount of transferred momentum. Calculations
show that the energy lost by a fast electron of energy W per unit
time to the plasma ions is given by
e “ ' ' 3 ' 10" 25i f r (315)
where «, is the ion concentration, Z is the number of elementary

charges per ion, and A is their atomic weight. It is assumed that
the ions may be regarded as being at rest; that is, the thermal
energy of the ions is very small in comparison with the electron
energy. In plasma, the various physical processes usually proceed
so that the energy taken from external sources is given to the
electrons, and some of it is then transferred from the electrons to
the ions. Therefore, the electron temperature Te will be much
higher than the ion temperature 7). In order to estimate the rate
of thermal energy transfer from the electrons to the ions, we can
use Equation (3.15) and replace W by the mean thermal energy of
the electrons. The quantity neQ is then equal to the amount of
energy transferred in one cubic centimetre per second from the
electrons to the ions. If Te and 7) are comparable, then the energy
transfer is smaller in the ratio of Te — 7’, to Tc.
Consider now some numerical examples characterizing the
transfer of thermal energy from the electrons to the ions in a
hydrogen plasma. Suppose that n = 1013 particles per cm3,
Te = 105°K, and T, = 0. The amount of energy received per second
Interaction o f Electrons and Ions with Neutral Particles 67
per cubic centimetre by the ions from the electrons is then equal to
4 X 106 erg, which corresponds to a power transfer of 0.4 kilowatts
per litre. Suppose now that the external source of energy is
switched off, so that after a short interval of time a virtually
complete thermal equilibrium is established between the electrons
and the ions, and T t is practically equal to Te. In point of fact,
with the above rate of heat transfer, each electron will lose on the
average about 4 X 10~12ergin 10 microseconds, which corresponds
to a reduction in its temperature by 30,000°K and a corresponding
increase in the temperature of the ions. Therefore T, and Te will
become equal in only about 20 microseconds after the external
source of energy is switched off. However, if the electron tempera
ture is increased to 108°K, without a change in their concentration,
the rate of heat transfer will be reduced by a factor of 30. The
time necessary to establish thermal equilibrium between the
electrons and the ions will then increase by a factor of 3 X 104,
and will therefore be of the order of a second.
The above examples are purely illustrative. The ratio of Te to T {
may, in general, vary within very wide limits, depending on the
conditions prevailing in the plasma. In a plasma with a low
density of charged particles, the heat transfer between the electrons
and the ions may be very slight at relatively low electron tempera
tures, and therefore a hot electron gas may be mixed with a cold
ion component. However, there are other possible cases when
Te and T,- may be approximately equal. For example, studies of
the radiation emitted by the plasma forming the outer layer of the
sun suggest that the electron and ion temperatures in this plasma
are not very different.
3.3 Interaction of Electrons and Ions with Neutral Particles

A two-particle interaction, in which the total kinetic energy of
the interacting particles remains the same and in which there is
only a redistribution of energy between them, is called an elastic
interaction. The collisions between the charged particles in plasma
which we have discussed so far belong to this category. However,
there are also inelastic collisions in which a fraction of the available

kinetic energy is transformed into other forms of energy, for
example, radiation, or the internal energy of one of the interacting
particles. Inelastic collisions leading to the emission of radiation
will be discussed in Chapter 4. For the moment we shall merely
note that there are two main types of such processes, namely,
deceleration of electrons in the electric field of the ions, which is
accompanied by the emission of light or x rays, and capture of
electrons by ions, which is also accompanied by the emission of
radiation. Inelastic collisions of the second kind, in which the
kinetic energy is transformed into internal energy of one of the
particles, are analogous to the corresponding processes which
occur with the participation of neutral atoms and must therefore
be considered together with these processes.
A fully ionized plasma consisting of bare atomic nuclei and free
electrons is a theoretical abstraction; in a real situation one has to
deal with plasma containing both positive ions and neutral atoms
and molecules. The number of charged particles in the plasma
produced in a gas discharge tube or in the ionosphere is very
small compared to the number of neutral atoms. It is only recently
that almost fully ionized plasma has been produced under labora
tory conditions in experimental apparatus for the production of
high-temperature plasma. Therefore, the interaction between
charged and neutral particles may play an important part in
many physical phenomena which are encountered in plasma.
In particular, when the density of charged particles is relatively
low, the motion of these particles will be affected to a greater
extent by collisions with neutral atoms or molecules than by
collisions with other charged particles. Among the various possible
interactions between the neutral particles on the one hand and the
electrons and ions on the other, the most important from the
point of view of plasma physics are the interactions with electrons.
An electron colliding with a neutral atom may undergo elastic
scattering, which is accompanied by a change in its direction of
motion and by a small loss of energy. However, there are also
collisions in which the atom is ionized or excited; that is, it is
raised to a higher energy level. To each of these processes can be
ascribed an effective cross section. As before, the effective cross

section is a measure of the strength of the interaction, and can be
used to replace a complicated set of true collision processes by a
simple model involving an incident particle and randomly dis
tributed spherical targets.
The effective cross section for the collision of an electron with
an atom, whether it be the elastic cross section or the cross section
for ionization or excitation, is very dependent both on the energy of
the electron and on the chemical properties of the target atom.
Figure 33 shows the effective cross section us for elastic scattering
of electrons by hydrogen molecules, and helium molecules, argon
and krypton atoms as a function of the electron energy. In all
cases there is a decrease in < j s with increasing electron energy W
provided W is high enough. At energies below 10 eV the de
pendence of < j s on W may be quite complicated. For example, the
noble gases are characterized by a minimum in as at low values of
W. These minima are responsible for a very considerable reduction
0 5 10 15 20 W(eV)
F igure 33. Elastic scattering cross section o f electrons by different gases

as a function o f the electron energy in electron-volts.
in the interactions between electrons and atoms in a very narrow

range of energies. This phenomenon is known as the Ramsauer
effect. It is due to the fact that the electrons have associated wave
properties. Under certain conditions, when the ratio of the wave
length of the moving electron to the linear dimensions of the atom
is a favorable value, the wave can pass directly through the atom.
(This is analogous to the interference effects which are observed
during the passage of light through thin films.) We shall not go
into the details of these wave effects, but will simply regard them
as an experimental fact.
The reduction in the magnitude of <rs at high energies can easily
be explained. It is due to the fact that an increase in the velocity
leads to a reduction in the time of interaction, which always tends
to reduce the effect due to the collision. This is a general property of
elastic collisions. At very high electron energies, the collision of
an electron with a neutral atom is not very different from a collision
with an ion, since in order to undergo a large deflection, the
electron must penetrate into the atom and closely approach the
atomic nucleus. The electric field near the nucleus is practically
due only to the nucleus itself; the electric field due to the extraneous
electrons is quite negligible. Therefore, a very fast electron, which
is appreciably deflected by the atom, interacts only with the
electric field of the nucleus.
For relatively slow electrons the atom behaves quite differently
and the electron must be regarded as interacting with a field of
force which becomes so strong, as the electron penetrates into the
atom to prohibit such penetration for slow electrons. Outside
the atom these forces are negligible, since the electric field of the
nucleus is screened by the outer electrons. In the cases illustrated
in Fig. 33, the maximum effective cross section for relatively slow
electrons lies between 0.5 X 10“ 15 and 2.5 X 10-15 cm2. This is
of the same order of magnitude as the dimensions of the atoms
themselves. The atoms of certain other chemical elements have
much larger electron shells, and therefore much greater <rs than
the noble gases. This is particularly so in the case of the alkali
metals. Thus, for example, the maximum value of as for cesium
is 5 X 10~14 cm2.
F ig u re 34. C a s c a d e p r o c e s s d u rin g th e io n iza tio n o f a g a s.
The scattering of an electron by an atom gives rise to a very

small energy transfer, just as in the case of the Coulomb, or
electrostatic interaction between an electron and an ion. The
mean energy transfer is of the order of (2me/m a) W, where ma is
the mass of the atom. Therefore, when an electron of energy
equal to 1 eV collides with an oxygen atom whose mass is greater
than me by a factor of about 30000, the energy transferred to the
atom is less than 10~4 eV. This is smaller by a factor of about 300
than the thermal energy of the oxygen atom at room temperature.
Among the various forms of inelastic interaction between
electrons and atoms, the most important for most plasma phe
nomena are those which give rise to the ionization of the atoms.
Each ionizing event gives rise to the appearance of a new pair of
charged particles (Fig. 34). It is precisely as a result of such
processes, leading to the cascade development of a gas discharge,
that plasma is usually formed. A definite amount of energy W,
must be spent in removing an electron from an atom. This energy
is known as the ionization energy and is usually expressed in
electron volts. It is also frequently referred to as the ionizaton

potential, which is numerically equal to the ionization energy Wh
but is expressed simply in volts. Thus, for example, the ionization
energy of the hydrogen atom is 13.6 eV and therefore the ioniza
tion potential of hydrogen is said to be 13.6 V. This terminology is
convenient because it indicates immediately the minimum potential
difference through which an electron must fall before it is capable
of ionizing a given atom. An electron which has fallen through a
potential difference smaller than 13.6 V will not succeed in ionizing
a hydrogen atom. If the energy of the electron incident on a
neutral atom is greater than the ionization energy, then the surplus
energy W — Wt- is shared between the two electrons, that is, the
primary electron and the electron leaving the atom.
In particular, when W is much greater than Wh the most
probable processes are those in which the secondary electron is
given relatively little energy (of the order of W,). This can be
easily explained. A fast electron flying past one of the atomic
electrons transmits to it a very small momentum, and therefore a
small amount of energy. If the transmitted energy in this almost
elastic collision is greater than JV,-, then ionization will take place,
and a low energy free electron is produced. If, however, W is only
slightly greater than Wh then the two electrons are indistinguishable,
and their energies are roughly the same.
Since, in general, an atom other than hydrogen and its isotopes,
deuterium and tritium, has more than one electron, and each of
them is bound in a different way to the nucleus, a number of
different ionization potentials must be ascribed to a given atom.
The minimum ionization potential gives the energy necessary to
remove the outermost electrons. These are the valence electrons,
which determine the chemical properties of the atom and many of
its important physical properties, for example, the geometric
dimensions, some' of the magnetic properties, the spectral com
position of the emitted radiation, and so on.
Table 2 lists the minimum ionization potentials of a number of
elements. As can be seen, the smallest ionization potential occurs
in cesium — an alkali m etal— while in the noble gases, whose
valence electrons form closed shells, the ionization potential is
quite high.
F igure 35. Typical variation o f the ionization cross section

with electron energy.
T able 2
Element W t (eV) Element W i (eV)
hydrogen 13.54 argon 15.68

helium 24.48 iron 7.83
nitrogen 14.51 cesium . 3.86
oxygen 13.57
The ionization potentials corresponding to the removal of

inner electrons are much greater than the minimum ionization
potentials given in Table 2. In the case of heavy atoms, they may
reach the order of 104 V or more (such electrons are located
nearer to the atomic nucleus and are therefore more strongly bound
to it).
Figure 35 shows a typical variation of the effective ionization
cross section of an atom with the energy of the incident electron.
The cross section is zero at W = Wh where Wt is the ionization
potential for the most loosely bound valence electron. The cross
section then increases rapidly and reaches a maximum at W = W\,
which lies between 40 and 80 eV, the position depending on the
particular element. As the energy is increased still further, it is
found that a,■ decreases at first slowly and then more rapidly.
This form of the ionization curve may be explained in terms of the
ideas which we have previously discussed. As long as the energy
of the incident electron is low, it must approach a bound electron
very closely before it can give up to the bound electron an energy
which is sufficient for ionization to occur. Therefore, with in-
creasing W, the value of the impact parameter b for which ioniza

tion is possible should also increase.
For very fast electrons there is a further important factor,
namely, the interaction is effectively reduced because it takes
place in a very short interval of time. This is the same factor
which is responsible for the reduction in the heat transfer between
the particles (see above). Therefore, it is not surprising that very
fast electrons have a low ionizing efficiency. The maximum value
of <r, is different for different elements: It varies between 0.7 X 10“ 16
cm2 (hydrogen) and 6 X 10“ 16 cm2 (heavy elements, for example,
mercury). We note that the curve shown in Fig. 35 summarizes
all the properties of ionization processes; that is, it includes both
the removal of the valence electrons from outer surface of the
atom and the removal of electrons belonging to the inner shells.
In fact, however, the contribution due to the internal electrons
to the effective ionization cross section is small provided W is not
too high. The probability that an electron colliding with an atom
will eject one of the inner electrons is small in comparison with
the probability of removing a valence electron. It is difficult
therefore to say what part of the total ionization curve is due to
strong interactions giving rise to the ejection of inner electrons.
The incident ionizing electron will not only lose energy, but
will also undergo a change in its direction of motion, and this
change increases with increasing proportion of transferred energy.
In this respect, the ionization process is analogous to the scattering
of electrons by neutral atoms. However, the role of ionizing
events in the energy balance of a fast electron moving in matter
is very much greater than the role of elastic scattering by neutral
atoms, since for ionizing events the electron loses a large pro
portion of its energy instantaneously, while in elastic collisions
the energy loss is quite negligible.
So far, we have been concerned with the individual ionizing
events involving the collision between an electron of given initial
energy and an atom. In order to evaluate the ionization processes
in plasma, it must be remembered that the plasma electrons have
different energies. An individual electron will produce n0a-j ions
per centimetre of its path, where n0 is the neutral particle density.
Since the path traversed by an electron per second is numerically

equal to its velocity v it follows that the number of ionizing events
per unit time is equal to n^v^,. This quantity is a function of the
energy of the electron, since both <rf and ve are functions of W.
In order to determine the number of ionizing events per second per
cubic centimetre of plasma, one must consider all the electrons,
whatever their energy. The final result will, of course, depend on
the electron temperature Te for a given electron and neutral-
particle concentration. As long as the electron temperature is
not too high, and kTe is small in comparison with the ionization
energy, the total number of ionizing events will also be small, so
that they can only be due to fast electrons which lie on the tail of
the energy distribution, that is, that part of the curve in Fig. 28
which lies well to the right of the maximum. The number of such
electrons is very small, and therefore the total rate of ionization
in plasma is also quite small. In this temperature range, the rate of
ionization is a very rapidly increasing function of Te. Conversely,
at very high temperatures, when kT e is much greater than the
ionization energy of the atoms, the number of ions produced per
second in the plasma is a much more slowly varying function of Te
and at very high Te the ionization rate tends to decrease.
Collisions between electrons and atoms can also give rise to
the excitation of the atoms. Under these conditions the electrons
transfer to the atom a fraction of their energy, and this lifts the
atom from the ground state to one of the possible higher energy
states.
It is well known that the energy of an atom cannot vary con
tinuously, but must be equal to one of a series of discrete values,
say, Wj, fVji, and so on. This can be represented by an energy-
level diagram which is illustrated schematically in Fig. 36. The
ground state of the atom is represented by the lowest energy level
A v The impact of an electron may raise the atom to one of the
higher energy levels. Transitions of this kind are indicated in the
figure by the arrows. The energy lost by the electron, and absorbed
by the atom during the transition to its excited state, is equal to
the difference between the energies corresponding to the excited
and the ground states. For the hydrogen atom, the minimum
A iv
>
> «
II II
.^ »—
i
* £
1 i
a >
^ £
F ig u r e 36. Schematic energy-level diagram fo r hydrogen.
energy which is necessary to raise it to the first excited state is

10.15 eV, while for helium and mercury atoms this figure is 20.55
and 4.9 eV, respectively. The minimum excitation energies for
other atoms lie between a few electron volts and 15-20 eV. Elec
trons are also capable of exciting ions provided the latter still
retain some of their original electrons. Thus, for example, when a
sufficiently fast electron collides with a singly charged helium ion,
which carries a single electron, the electron may be raised to the
first excited state. The minimum energy necessary for this to
happen is 40.5 eV.
The effective excitation cross section acx is a function of the
energy of the incident electron. A characteristic form of this
dependence is shown in Fig. 37. As can be seen, after the incident
electron has reached the threshold value, equal to W\\ — W\,
the effective cross section increases very rapidly, reaching a
maximum at an energy which is not very different from
W \ \ — W\. This is followed by a decrease in the cross section.
In this part of the curve the energy dependence of aex is similar to
the corresponding curve for the ionization cross section. The
maximum value of the excitation cross section for most neutral
atoms is of the order of 10"17 to 10“ 16 cm2. The excited atom
usually retains its extra energy for a very short interval of time
(less than 10“8 sec). It then returns to the ground state and gives
F ig u r e 3 7 . Characteristic form o f the dependence o f the excitation cross

section on the energy o f the incident electron.
up the surplus energy in the form of radiation by emitting a

quantum of light, that is, a photon.
It is well known that the photon is a discrete bundle of radiant
energy. The energy of a photon is equal to hv, where h is a universal
constant known as Planck’s constant and v is the frequency of the
corresponding oscillations. Planck’s constant h is equal to
6.6 X 10“27 erg sec.
The frequency of the light emitted by an excited atom when it
undergoes a transition from a state of energy WB to a state of
energy WA is given by
h v = W B - W A. (3.16)
The wavelength of the emitted radiation is given by
X = -. (3.17)
Hence, using (3.16) and (3.17) we find that

he
X= (3.18)
WB - Wa
In this expression the energy difference WB — WA, which is
converted into radiation, is expressed in ergs. When it is expressed
in electron volts, we have
X = 1.23-10' 4 ______!-------- , (3.19)
WB - WA
Aw
Am
An
A i ------ 1------ 1---------

F igure 38. Example o f a stepwise transition o f an excited atom from the
level A iv to the ground state.
where X is in centimetres. It follows from this expression that

when the emitted energy amounts to 1 eV, the corresponding
wavelength is 1.23 g which lies in the infrared region, adjacent to
the visible region. In the case of the hydrogen atom, the first
excited state lies 10.15 eV above the ground state, and therefore
when the atom returns to the ground state the wavelength of the
emitted radiation is 0.12 g which lies in the far ultraviolet. Under
the conditions which are usually realized in the laboratory or in
technological installations, the energy of most of the photons
emitted by excited plasma atoms is not less than 4-5 eV, and there
fore most of the radiation is emitted in the ultraviolet. However,
an appreciable portion of the radiation may be emitted in the
form of low energy photons. This is due to the fact that an excited
atom may return to the ground state by a number of successive
transitions rather than by a single transition to the ground state
(Fig. 38). The radiation emitted by excited atoms or ions usually
plays a leading role in the general energy balance of plasma
radiation.
The excited atoms in plasma can also lose their surplus energy
by collisions of the second kind. Thus, the excited atom may
transfer its surplus energy to one of the plasma electrons which
happens to collide with it. The process is, in fact, the exact inverse
of the excitation of an atom by electron impact. Collisions of the
second kind are usually unimportant in plasma with a low con

centration o f charged particles, since an excited atom will usually
succeed in getting rid of its extra energy by emitting a photon
before it undergoes a collision of the second kind. However, in
some cases the lifetime of the excited atom may turn out to be
relatively long (on the atomic scale), and therefore the probability
that a collision of the second kind will take place will be greater.
This occurs whenever the excited atom is in a metastable state.
For example, a helium atom raised to the energy level which lies
at 19.75 eV above the ground state may remain in this level for a
time of the order of 10“4 sec, since the laws of atomic physics
lead to the conclusion that the transition to the ground state with
the emission of a photon is not very probable (such transitions are
referred to as forbidden). Therefore, a metastable helium atom
has a distinct chance of disposing of its surplus energy either by
giving it up to one of the free plasma electrons or by colliding with
the solid wall of the container. The atoms of other noble gases
and many other elements exhibit analogous properties in some of
their states.
So far, we have been concerned with collisions in which one of
the colliding particles was a free electron. Under certain con
ditions, collisions between ions and neutral atoms, in which a
charge transfer takes place, may be important from the point of
view of the properties of plasma. In this process, an ion colliding
with an atom receives an electron from it, leaving the atom in an
ionized state while itself becoming neutral. This may be described
by the following formula
A+ + B A + B+.
Charge-transfer processes have a high probability when the ion
A+ and the atom B belong to the same element, for example, in
the case of a collision between a proton and a hydrogen atom, or
between a helium ion and a helium atom.
In the case of resonant charge transfer, the effective cross
section for the process may be quite high, particularly at low
energies when the two particles slowly pass each other. For
example, the effective charge-transfer cross section for protons
on hydrogen atoms is about 0.5 X 10~14 cm2 when the energy of

the protons is of the order of a few electron volts. It should be
noted that this is greater than the geometrical cross section of the
hydrogen atom by a factor of the order of 100. Nonresonant
charge transfer is characterized by a smaller effective cross section
whose maximum value is of the order of 10~16cm2. Charge
transfer processes are important since, under certain conditions,
they are an effective mechanism for cooling the plasma ions.
4
Radiation
Hot plasma containing a high concentration of charged particles

may act as a very powerful source of radiant energy. The radia
tion is produced largely as a result of various types of collisions
between the plasma particles. Let us consider them in turn.
4.1 Bremsstrahlung
When a free electron passes through the electric field of an ion or
atom, there is a change in the direction and magnitude of its veloci
ty. A large instantaneous change in the velocity of a charged
particle leads to the appearance of electromagnetic radiation whose
energy is supplied at the expense of the kinetic energy of the particle.
This means that, during its collision with an atom, an electron may
lose a part of its energy by emitting a photon. The energy hv of
the photon may amount to any fraction of the initial kinetic energy
We of the electron, but its maximum value is We. The appearance
of a photon with this maximum energy signifies that all the kinetic
energy of a fast electron has been converted into electromagnetic
radiation. Since hv can also assume any other value between 0 and
We, the emitted radiation, or the bremsstrahlung as it is commonly
called, has a continuous spectrum of frequencies between 0 and
We/h, in contrast to the radiation emitted by excited atoms which
consists of discrete spectral lines of definite wavelengths. Plasma
is not the only source of bremsstrahlung. A more common example
is the radiation emitted by x-ray tubes, used for diagnostic or
therapeutic purposes in medicine. In an x-ray tube a beam of fast
81
82 R a d ia tio n
electrons, which are emitted by a hot tungsten wire (cathode), is

accelerated by a potential difference of some tens of thousands of
volts and is allowed to bombard a second electrode, the anticathode
or anode. On penetrating into the anode, the fast electrons interact
with the electric field of the atoms and emit bremsstrahlung as they
come to rest. Owing to the high energy of these electrons, the fre
quency of the emitted electromagnetic waves is very high and there
fore the wavelength is very small (of the order of 1 0 - 9 to 1 0 - 8 cm,
characteristic of x rays.)
The frequencies of the electromagnetic waves emitted by plasma
at an electron temperature Te lie within very broad limits, but most
of the emitted radiation corresponds to photons with frequencies of
the order of kTe/h and wavelengths of the order of ch/kTe. There
fore, the bremsstrahlung emitted by plasma with Te = 10,000°K lies
mainly in the infrared and in the visible parts of the spectrum, while
at Te = 1 0 8°K (high-temperature plasma in experimental thermo
nuclear reactors) most of the radiation lies in the x-ray region. The
total intensity of the bremsstrahlung emitted by plasma per unit
volume is proportional to the number of electron-ion collisions per
second, which in turn is proportional to ne«,. Moreover, the in
tensity of the radiation is very dependent on the charge carried by
the ions since the probability that a photon will be emitted during
an electron-ion collision increases with the electric field which acts
on the electron and changes its velocity. Theoretical calculations
show that the energy emitted per second per unit volume by a fully
ionized plasma at the temperature Te is given by
Q = 1.5- lO -^neiiiZ ^T e, (4.1)
where Q is expressed in ergs per unit volume per second andZ is the
atomic number of the element whose ions are the constituents of the
plasma. As an example, let us suppose that we have succeeded in
producing a hydrogen plasma with electron temperature Te = 108°K
and an electron concentration equal to 1016 cm-3. According to
(4.1), each liter of this plasma will generate about 150 kW in the
form of x rays, which is equivalent to the total intensity generated
by several thousand simultaneously operating x-ray tubes. The
intensity is even higher in the case of heavier ions.
Recombination Radiation 83
I
13.6
1 I I
F igure 39. Possible transitions during the recombination o f an electron
and a proton (schematic).
4.2 Recombination Radiation

The phenomenon of recombination involves the capture of a free
electron by an ion when the electron passes close to it and, there
fore, results in a bound state. The energy which is liberated in this
process is equal to the sum of the kinetic energy of the free electron
and its binding energy. For example, when an electron with energy
We is captured by a proton to form a hydrogen atom in the ground
state, the total energy which becomes available is W e + 13.6 eV
(Fig. 39).
The shaded area in the energy diagram corresponds to free elec
trons. The kinetic energy of such electrons is measured by the
vertical distance from the zero line. The ground state of the elec
tron in the hydrogen atom corresponds to a negative energy 10 of
13.6 eV. The available energy can be emitted in the form of a
photon with energy W e -f 13.6 eV. Another process by which the
system may dispose of its energy is that in which the atom first
10T he zero level o f energy is conventionally assum ed to correspond to the state

in which the bond betw een the proton and the electron is just reduced to zero
and the tw o particles are taken to a very large distance from each other with
zero kinetic energy.
84 Radiation
undergoes a transition to one of the possible excited states from

which it then decays to the ground state. The number of photons
which will be emitted under these conditions may be two or more,
as shown on the right-hand side of the energy-level diagram in
Fig. 39. The emission of bremsstrahlung can also be indicated on
this diagram. It corresponds to a change in the energy state of the
electron in the shaded region and can be represented by a transition
such as a —►b.
Since free electrons have continuous energy spectrum, the
photons which are emitted in the recombination process form a
continuous spectrum upon which is superimposed the line spectrum
of the excited atoms, produced as a result of the stepwise transitions
mentioned above. The total intensity of the recombination radia
tion is, clearly, also proportional to ncni. In contrast to bremsstrah
lung, the intensity of recombination radiation decreases with in
creasing electron energy and is therefore enhanced by a reduction
in the electron temperature Te. In hydrogen plasma with «,• = ne,
the intensity of recombination radiation per unit volume can be
estimated from the expression
f) _
n2
in-22 n e
^'recom b in ation 1^ (4.2)
Vt;
Up to temperatures of the order of 107 °K,the intensity of recombina
tion radiation exceeds that of bremsstrahlung. At higher tempera
tures bremsstrahlung predominates. Moreover, the intensity of re
combination radiation increases rapidly with the number of ele
mentary charges carried by the ions. (It is approximately propor
tional to the fourth power of the ionic charge.)
4.3 Radiation Emitted by Excited Atoms and Ions

The origin of this radiation has already been discussed. In the
case of plasma, its intensity is strongly dependent on the electron
temperature and the chemical composition, and increases rapidly
with the number of heavy ions in the plasma. The spectrum of this
radiation consists of discrete lines corresponding to the various
transitions between the excited energy states of the atoms and ions.
B e ta tr o n E m iss io n f r o m P la s m a 85
With increasing electron temperature, there is a gradual increase in

the charge on the heavy ions and the line emission spectrum is
modified. Thus an atom in a plasma at a very low electron tempera
ture will either remain neutral or will lose one of its most weakly
bound electrons. With increasing electron temperature, the inner,
more tightly bound electrons will also be removed, and therefore
the mean charge per ion will increase. Moreover, this will be as
sociated with an increase in the excitation energy of the ions, with
the result that the line spectrum will be displaced from the visible
region into the ultraviolet and x-ray regions. The situation may
become clearer if we consider some numerical examples. Hydrogen
plasma can be practically fully ionized at an electron temperature
of 10 eV (100,000°K). If the concentration is 1013 cm-3, the total
intensity of radiation which is emitted at this temperature is
5 X 10" 24 erg per cm3. Most of this radiation is due to electron-
proton recombinations. Suppose now that we add one atom of
oxygen to every hundred atoms of hydrogen. The introduction of
this small amount of oxygen impurity will increase radiation losses
very considerably. In fact, the radiation loss will now be mainly
due to the line spectrum of the excited atoms and ions of oxygen,
and if the electron temperature is the same (100,000°K), the total
intensity emitted by the plasma will be higher by a factor of about
10, 000.
The line spectrum emitted by excited atoms and ions in plasma
at a not too high electron temperature is one of the main means
by which the plasma loses its energy. The energy lost in this way by
hydrogen plasma at an electron temperature of the order of
100,000°K with a 10% oxygen impurity is greater by three orders of
magnitude than the energy lost by bremsstrahlung from hydrogen
plasma with Te = 108°K and the same electron concentration.
4.4 Betatron Emission from Plasma

A new source of radiation comes into play when plasma is placed
in a magnetic field. The electrons now execute circular motion in
the plane perpendicular to the magnetic field, and since the motion
is accelerated (the acceleration is directed towards the center of the
86 Radiation
circle) it should, in accordance with the laws of electrodynamics,

give rise to the appearance of radiation. For a single electron the
intensity of this radiation is proportional to the square of the
magnetic field and is also a function of the velocity of the electron
executing the Larmor rotation. The amount of energy radiated by
the electron per second is 6.4 X 10~2lH2fV sin2a, where H is the
magnetic field, W is the energy of the electron in electron volts, and
a is the angle between its velocity and the direction of the magnetic
field. This expression is valid provided the velocity of the electron is
small in comparison with the velocity of light, which is practically
always the case in plasma. Radiation of this kind is usually referred
to as betatron radiation and is of major importance in accelerator
technology. When electrons are accelerated to high energies in a
modern circular accelerator, a stage is eventually reached when the
energy spent in accelerating the electrons is used mainly to com
pensate the betatron radiation losses.
Betatron radiation has a line spectrum in which the fundamental
frequency corresponds to the Larmor frequency of the electrons.
In addition to the fundamental frequency there are also harmonics,
that is, frequencies which are multiples of the fundamental fre
quency. The relative importance of these harmonics in the total
intensity of the radiation increases with increasing W. Betatron
radiation lies mainly in the radio frequency range. The wavelength
corresponding to the fundamental frequency is [in view of (2.17)]
given by
2irmec2 ,
x = cT« = ~ tr' (4J)
The wavelengths of the harmonics are A/2, X/3, and so on. In a

magnetic field of the order of 1000 Oe, the radiation consists largely
of centimeter waves, while at H ~ 10,000 Oe it lies in the millimeter
range.
The radiation emitted by plasma is made up of the contributions
from the individual electrons. The total intensity of betatron radia
tion produced per unit volume is therefore proportional to ne. At
high electron concentrations and magnetic fields, a very consider
able amount of energy could be lost by betatron radiation if it were
Flux o f Energy Emitted by Plasma 87
allowed freely to leave the plasma. However, in reality, the long

wave radiation corresponding to the fundamental frequency and its
lowest harmonics are strongly absorbed by the plasma itself, and
therefore the total flux of radiation escaping through the surface of
the plasma is only a small fraction of the energy generated at inter
nal points. B. A. Trubnikov, working at the Kurchatov Institute of
Atomic Energy, has shown by direct calculation that, at very high
electron temperatures ( 1 0 8°K or more), the escape of betatron radia
tion from plasma increases rapidly because there is an increase in
the fraction of the energy associated with the higher harmonics
which correspond to short wavelengths. Betatron radiation is
therefore a limiting factor in the production of high-temperature
plasmas.
4.5 Flux of Energy Emitted by Plasma

In order to obtain an insight into the nature of the radiation
emitted by plasma, it will be useful to compare plasma with a perfect
blackbody. A perfect blackbody is defined as a body which com
pletely absorbs all radiation falling upon it, whatever the wave
length. Theoretical calculations, which have been confirmed by
experiment, predict that the total intensity of the radiation emitted
by a blackbody is proportional to the fourth power of its absolute
temperature, and that the energy emitted by a blackbody is equal to
5.7 X 10~ 5 T4 erg per cm2, sec. The spectrum of blackbody radia
tion is illustrated in Fig. 40, in which the energy emitted per unit
area per unit frequency interval from the surface of a blackbody
is plotted as a function of the frequency of the radiation v.
The most important features of the blackbody spectrum can be
summarized as follows:
1. At low frequencies, when the photon energies hv are much
smaller than k T (the energy which characterizes the thermal energy
of the blackbody particles) the intensity is proportional to ?3. If
we compare the intensity per unit frequency range at a given fre
quency in different temperatures, we find that the intensity is pro
portional to the cube of the temperature.
2. The intensity of blackbody radiation has a maximum at
88 Radiation
0 10 i 2 ioi3 2-1013 p (cps)
F igure 40. Spectra o f blackbody radiation at different temperatures.
hv = 2.75 kT. With increasing temperature, this maximum is dis

placed towards higher frequencies. At temperatures between
7,500° and 15,000°K, the maximum lies in the visible range. At
temperatures of the order of some tens of thousands of degrees, the
maximum lies in the ultraviolet and corresponds to wavelengths of
the order of 0.1 p. As the temperature is increased still further, the
blackbody will eventually become an extremely powerful source
of x rays.
In distinction to a blackbody, plasma at a relatively low electron
temperature is almost totally transparent to visible and ultraviolet
radiation. It will strongly absorb only radiowaves, that is long
wave radiation. In view of the above discussion, this radiation can
only be produced in a strong magnetic field. Therefore, if the radio
frequency region of the spectrum is ignored, it may be considered
that the total amount of energy emitted by plasma of constant
density and temperature is proportional to its volume. Suppose, for
example, that this volume is spherical in form. The total energy
which is radiated by the plasma is then proportional to a3, where a
is the radius of the ball of plasma, and the intensity of radiation
emitted per unit area of its surface is proportional to the ratio of the
volume to the surface area; that is, it is a linear function of a.
Flux o f Energy Emitted by Plasma 89
However, this increase cannot continue indefinitely since, accord

ing to the general laws of radiation, the energy which is emitted by
the surface of a heated body cannot exceed the energy emitted by
the surface of a blackbody. This refers not only to the total amount
of energy, but also to the intensity in each individual spectral
region. The physical reason which leads to the reduction in the
rate of increase of the intensity of the plasma radiation is that,
when the volume occupied by the plasma is large, absorption be
comes more and more important. Absorption first becomes ap
preciable in the long-wave region of the spectrum; it is only when a
becomes very large that it affects ultraviolet and x-ray radiation.
We note, by the way, that at a temperature of 108 °K, which is
considered the minimum temperature for future thermonuclear
reactors, the radiation emitted per unit area of a blackbody is ap
proximately equal to 5 X 1017 kW. At this rate of emission a
sphere of unit volume placed at a distance of one million kilometers
from the earth would give rise to a flux of radiation at the earth’s
surface which would be greater than the flux of solar radiation, al
though the radiation would consist mainly of x rays. Further
calculations show that, in fact, a plasma with a concentration of
1016 cm - 3 and a temperature of 108°K will emit an energy flux
corresponding to a blackbody at a temperature of only about
2000°K. Therefore, a small volume of high-temperature plasma
will, in point of fact, be a relatively modest source of radiation.
5
Electric Current, Diffusion,

and Thermal Conductivity
5.1 Electric Current in Plasma
Directed motion of electrons and ions in plasma may be due to

two causes: an electric field producing a current, or a difference in
the concentration between different regions in the plasma. In a
nonuniformly heated plasma, the transfer of particles between
regions at different temperature gives rise to plasma heat conduc
tion, that is, the flow of thermal energy. Phenomena of this kind
are collectively referred to as transfer phenomena. The appearance
of an electric current, which we shall consider first, is of particular
importance in plasma physics. A current flows because the electric
field acting on charged particles of different sign makes them move
in opposite directions. In a time /, an electron will acquire the ad
ditional velocity (eE/me)t, while a singly charged ion will receive
the additional velocity (eE/M i)i. If there were no collisions be
tween the electrons and the ions, and also between the electrons and
ions and the neutral particles, their velocities would increase in
definitely and the plasma would soon be converted into two streams
of fast particles travelling in opposite directions. However, in
reality, collisions do take place and the particles maintain a directed
motion for only a limited time.
The acquired velocities are lost as a result of collisions, after
which the particles are again accelerated by the field. We note that
the term “collision” is used in the sense defined in Chapter 2 .
90
Electric Current in Plasma 91
Usually, the total velocity acquired by an electron or ion in one free

path is quite small in comparison with the thermal velocity, and
therefore an external electric field has only a slight effect on the
motion of the particles. The small extra velocity in the direction of
the electric field is superimposed on the fast random motion which
takes place at thermal velocities. We can therefore speak of a
relatively slow drift of charged particles in the direction of the field,
rather than the formation of a current in which the velocities of all
the particles have the same direction. A current in plasma is due to
the directed motion of charged particles of both signs. However,
the contribution due to the positive ions is negligible in compari
son with the contribution due to the electrons, since the ions have a
much larger mass and therefore smaller acquired velocity. The
contribution of the ions to the electric current will therefore be
neglected henceforth and we shall confine our attention to the mo
tion of the electrons.
A constant current in plasma consists of a steady stream of elec
trons in which the force on each particle due to the electric field is
balanced by the frictional force due to collisions between the elec
trons on the one hand and ions and neutral atoms on the other.
The frictional force is equal to the momentum (mass multiplied by
velocity) which is lost per second by the moving electron to the ions
and atoms in the plasma . 11 Let the velocity of the beam of electrons
in the plasma be denoted by u, so that the average momentum of
electrons is meu. Since an electron undergoes v collisions per
second, in each of which it loses its velocity and transmits a momen
tum meu to an ion or atom, it follows that the frictional force on an
electron is meuv. The condition, that in equilibrium the forces
acting on the electron must be equal and opposite, is then
meuv = eE, (5.1)
and hence the average drift velocity is given by

eE 1 eE
u --------- — r, (5.2)
me v me
“ W e recall that, in accordance w ith N ew to n ’s secon d law, force is equal to the

rate o f change o f m om entum .
92 Electric Current, Diffusion, and Thermal Conductivity
where r = \ / v is the mean time between successive collisions ex

perienced by electrons. The current passing through a unit area at
right-angles to the direction of the electric field in the plasma
(Fig. 41) is defined as the current density and is equal to the net
electric charge passing through the unit area per second. If the
average drift velocity of the charges — in the present case, the
electrons — is u, all the charges which lie in a parallelepiped of unit
base area and length u will pass through the unit area in one second.
If the density of the electrons is ne, the total number of electrons in
the parallelepiped is neu and the total charge is neeu. Therefore, the
current density j is given by
(5.3)
This can conveniently be rewritten in the form
j = vE, (5.4)
where 77 is the conductivity and is given by
nee2r
(5.5)
F igure 4 1 . Flux o f electrons passing through a unit area per second.

Equation (5.4) is another way of writing Ohm’s law.12 In Equation

(5.5), r represents the mean time between collisions which reduce
the velocity of the electron current. Therefore, r is related to the
quantities r ei and r e0, which were defined earlier, by
1
(5.6)
T Tei TeQ
In contrast to Equation (3.7), the latter expression does not include

the term representing electron-electron collisions since these should
not lead to a retardation of the electron current, that is, to a reduc
tion in the drift velocity.
In order to obtain an insight into the properties of plasma as a
conductor of electricity, it will be useful to consider two extreme
cases. The first of these cases corresponds to fully ionized plasma
and the second to plasma with a very small relative concentration of
charged particles. To begin with, consider the first of these two
cases. Since there are no neutral atoms, there are ho collisions be
tween electrons and neutral atoms, and therefore r = T e i . Substi
tuting for r ei from (3.14) into (5.5), and remembering that in a fully
ionized plasma ne = n{Z where Z is the charge of the ions, we have13
7" ’3/2
t, = 0.9-102— - (5.7)
Equation (5.7) has certain important consequences. First, it shows

that the conductivity of fully ionized plasma is independent of the
concentration of electrons. This is understandable because as ne in
creases, there are more particles acting as current carriers and more
collisions; that is, there is a reduction in the mean free time.
12F or readers w h o are accustom ed to the m ore general way o f w riting O hm ’s
law , nam ely, I = U/R, w here I is the current, U the potential difference, and R
the resistance, w e m ust point out that E quation (5.4) is equivalent to this expres
sion . In point o f fact, I = jS, w here S is the cross section o f the conductor,
U = El w here / is the length o f the conductor, and R = kI/S, w here k is the
resistivity. Substituting these expressions in to the form ula relating /, U, and R,
w e find that j = E / k, w here 1 /* is by definition the electrical conductivity o f the
plasm a. W e thus arrive again at E quation (5.4).
l3This expression is given in cgs units. In order to convert it into practical
units, the electrical conductivity should be divided by 9 X 10u .
Second, for a given electron temperature, the conductivity is in

versely proportional to the mean charge per ion. Finally, the con
ductivity of a fully ionized plasma is proportional to T \n and will
therefore become very great at high temperatures. For example, the
conductivity of hydrogen plasma at different temperatures is given
in the following table:
Te (°K) V (cgs)
104 10*3
105 3 • 1014
106 10*6
107 310*7
108 10*9
At an electron temperature of about 15 X 106, hydrogen plasma

has the same electrical conductivity as ordinary metallic copper at
room temperature.
Under certain conditions, an interesting phenomenon, known as
electron runaway, occurs in fully ionized plasma. We have assumed
that an electron is accelerated in a time r and its drift velocity is lost
on collision with an ion. This means that the electron component
of the plasma moves as a whole with a constant velocity. However,
a simple argument will show that all the electrons cannot have the
same mean velocity. Thus consider an individual electron whose
velocity is much greater than the mean thermal velocity of the elec
tron component. An electron of this kind corresponds to the tail
end of the energy distribution and its mean free time is much
greater than the mean value r . Moreover it has been shown (see
Equation (3.10)) that the effective scattering cross section is in
versely proportional to the fourth power of the velocity, and
therefore the mean free path is directly proportional to v4 and the
mean free time is proportional to i>3. It follows that, in the interval
between successive collisions, our particular fast electron is ac
celerated by the field to an additional velocity which is much greater
than the mean drift velocity w. For example, if the kinetic energy of
the electron is higher than the mean thermal energy by a factor of
10, the drift velocity to which it is accelerated in the electric field is

found to be higher than the mean drift velocity by a factor of 30.
Thus, if the initial velocity of the electron is high enough, the mean
free path may be so large that the additional velocity acquired ex
ceeds the thermal velocity. However, this means that we can no
longer use the simple model in which the electron loses its extra
velocity as a result of collisions, since this model is only valid when
the extra velocity is small in comparison with the thermal velocity.
Moreover, we should recall the discussion of the physical meaning
of quantities such as the mean free path and the mean free time
between collisions in plasma, which was given in Chapter 2 where
it was noted that these quantities cannot be easily fitted into a
simple scheme involving gas-kinetic concepts only.
The extreme case we have just considered is an illustration of a
situation in which the simple gas-kinetic model must be abandoned
in favor of the true physical description. In point of fact, accelera
tion and deceleration of an electron occur continuously rather than
discretely. While the electron is receiving its additional velocity,
collisions with ions gradually alter its direction of motion. If the
initial velocity of the electron is very high, its interaction with the
ions will be considerably reduced, and therefore the increase in the
drift velocity produced by the electric field will not be compensated
by a collisional loss of the longitudinal velocity component. This
means that the accelerating and frictional forces will no longer be in
equilibrium and the electron will be continuously accelerated by the
electric field. This process will occur in the case of those electrons
in the Maxwell tail which succeed in reaching in one free path an
additional velocity greater than the initial velocity v . Mathe
matically, this condition may be written in the following form:
— r > v. (5.8)
me
It has already been pointed out that r is proportional to v3 and
inversely proportional to the ion concentration, so that r = (a///,)u3,
where a is a numerical coefficient. Substituting this expression into
(5.8), we see that the continuous acceleration of the electron begins
when E v 2 / n i exceeds a certain maximum value. As the electric field
increases and the concentration of the plasma decreases, the magni

tude of the initial velocity v which is necessary to satisfy (5.8) de
creases, and therefore there is a corresponding increase in the
fraction of the electron component of the plasma which is con
tinuously accelerated. This means that in addition to the current
which obeys Ohm’s law, there will also be a current of accelerated
electrons whose velocity, and therefore energy, will increase with
time.
When E/m is high enough, the condition given by (5.8) will be
satisfied not only for electrons in the tail of the Maxwell distribution
but also for those with velocities of the order of the mean thermal
velocity. Most of the plasma electrons will then be continuously
accelerated and the usual ohmic current will be absent altogether.
Theoretical calculations show that electron runaway becomes ap
preciable when the ratio of the mean drift velocity of the plasma as a
whole to the mean thermal velocity of the electrons is greater than
about 0.1. The ratio of the drift velocity to the thermal velocity is
proportional to the kinetic energy of the electrons since this ratio
involves the factor v2. Therefore, if u/v = 0.1 for electrons with the
mean energy, then for electrons with kinetic energy of the order of
10 kTe, the ratio of u to v will be of the order of unity and such
electrons will take part in the runaway process.
The necessary condition for continuous acceleration of electrons
is that the plasma conductor must be long enough; that is, the
electrons must have a long distance available for the acceleration
process. This condition can easily be satisfied in a toroidal chamber
in which the electrons are accelerated by an electric field induced,
for example, with the aid of the transformer arrangement illustrated
in Fig. 42. Continuous acceleration of a relatively small number of
electrons can, in fact, be observed experimentally. Energies of the
order of a few million electron-volts can be achieved in a field of
only a few tens of volts per orbit.
Theoretical analysis of the behavior of a beam of continuously
accelerated electrons in plasma leads to the conclusion that, under
certain conditions, such currents are capable of exciting various
plasma oscillations and waves which take over the energy acquired
by the electrons during their acceleration in the electric field.
Processes of this kind may give rise to an additional deceleration of

charged particles, so that the acceleration process will terminate
when the electrons reach a certain energy.
Consider now the passage of an electric current through an
ionized gas with a very low degree of ionization, that is, the case
where the deceleration of the electrons is due mainly to collisions
with netural atoms and molecules in the gas. Under these condi
tions t = \/n 0Vaa, where no is the concentration of neutral particles
and <ra is the effective collision cross section between electrons and
neutral particles. It has already been pointed out in Section 3.2 that
<ia is, in general, a relatively complicated function of electron energy.
In cgs units, the conductivity of a weakly ionized gas is given by
n0 nte ve<ja
where the bar over the product veoa represents the mean of this
quantity (with allowance for the electron-velocity distribution and
the dependence of <ra on ve). In very approximate calculations, we
may assume that ve is equal to the mean thermal velocity of the
F igure 42. Toroidal discharge chamber surrounding

the core o f a transformer.
electrons and <ja can be taken at this velocity. The product of these
two parameters should not then be very different from the true
mean of the product ve<ra.
Equation (5.9) shows that the conductivity of a weakly ionized

plasma is proportional to the relative ionization ne/n 0. Since this
ratio enters as a factor into (5.9), the conductivity of a weakly
ionized plasma should be small (simply because of the very small
number of current carriers). Low-density plasma is encountered in
low-current electric discharges, where the temperature of the
electrons depends on the ratio of the electric field E to the gas
pressure p0. This may be written in the following symbolic form
Te = f(E /p 0). This dependence of Ta on the discharge parameters
has a simple qualitative explanation. The thermal energy of an
electron is determined by the difference between the work done by
the electric field and the energy losses due to collisions between
electrons and atoms. It is clear that the higher the field strength, the
higher will be the energy which an electron can acquire between
successive collisions. On the other hand, with increasing pressure
there is an increase in the energy losses due to collisions, and there
fore the temperature of the electrons should decrease with increas
ing pressure p0. Hence, Te should increase with increasing E and
decrease with increasing p0. The quantity ueaa is, of course, also a
function of E/po.
It follows from the above discussion that the conductivity of
plasma with a low-charged particle concentration may vary over
very wide limits and is a function of the electric field strength, the
pressure, and the chemical composition of the gas. It is therefore
quite difficult to find a simple relationship which could be used to
predict the conductivity of a weakly ionized plasma in each specific
case. All that one can do is to note some general points which may
be useful in very approximate calculations. First, one would expect
that for given E/p0, the ratio ne/n 0, that is, the relative ionization,
should increase rapidly when an impurity with a low ionization
potential is added to the gas. The vapor of an alkali metal is suit
able for this purpose. On the other hand, the addition of halides
(chlorine or fluorine) should reduce the conductivity very considera
bly at low E/p0, since chlorine and fluorine have high electron
affinities and can easily capture free electrons. The negative ions
produced as a result of this capture have large mass and therefore
small mobility in the electric field. The conductivity of plasma with
Plasma in High-frequency Field 99
a low relative ionization cannot, in general, be high. A high con

ductivity 7] at a low E /pQcan only be achieved if the effective col
lision cross section is small, but this requires a high electron temper
ature which is inconsistent with the assumed low relative ionization.
The practical limit for the electrical conductivity of a weakly ionized
gas is of the order of 1012 cgs. This means that the resistivity of a
weakly ionized gas will be higher than the resistivity of copper by a
factor of at least a few tens of thousands. Continuous acceleration
of electrons is, in principle, also possible at very low pressures and
very high electric field-strengths in neutral or weakly ionized gases.
However, in this case, the initial conditions necessary for this
phenomenon to occur are much more rigid than for fully ionized
plasma, since the electrons must now be accelerated through the
energy range in which the collision cross section has a maximum.
5.2 Plasma in a High-frequency Field

Suppose now that a high-frequency alternating current flows
through the plasma. We shall assume that the frequency of the
electric field, which gives rise to this current, is so high that a large
number of oscillations take place during the time corresponding to
a mean free path. The effect of the collisions on the behavior of the
electrons is then very small, and can be neglected; that is, it may be
supposed that the only forces acting on the electrons are those due
to the high-frequency field. In order to determine the current due to
the alternating field, we can use the results of Section 2.4, which are
represented graphically in Fig. 7. One of the more important con
clusions of that analysis is that the phase of the velocity lags behind
the phase of the field by 90°. Since the electric current is propor
tional to the electron velocity in the direction of the current, it must
also lag behind the field. Therefore, the phenomenon is analogous
to that encountered in AC theory when one considers the properties
of an inductance, for example, an ordinary coil.
However, the origin of the phase lag is quite different in the two
cases. The alternating current flowing through a coil gives rise to an
alternating magnetic field; this induces an additional electromotive
force in the wire which must be compensated by the external
voltage applied to the coil. It is the induced field which is responsi

ble for the phase difference between the voltage and the current. In
the case of a high-frequency current in plasma, the phase difference
between / and U is not associated with a magnetic field. It is
simply due to the fact that the current carriers, that is, the electrons,
have a finite mass and oscillate as mass points in the periodically
varying field of force. The density j of the high-frequency current
at any given time is neeu, where u is the velocity of the electrons in
the direction of the field and is given by (2.8). Therefore, the ex
pression for the current density flowing through plasma is
j = £ 0 sinco/, (5.10)
mew
where o>is the angular frequency of the electric field. All the quanti
ties in this expression are in cgs units. The amplitude j 0 of the
current, that is, its maximum value, is related to the amplitude of
the electric field £ 0 by j 0 = nec2E0/m eo>. Equation (5.10) is valid
when the time between collisions is large in comparison with the
period of the high-frequency oscillations. When this conditions is
not satisfied, the plasma exhibits not only a kind of inertial induct
ance but also a finite resistivity.
The lower part of Fig. 7 shows the variation in the position of an
electron under the influence of a high-frequency field. Just as in the
case of mass point executing oscillations under the influence of a
periodic force, the displacement of the electron under the action
of the alternating electric field lags in phase behind the velocity by
90°. Hence, the displacement S lags 180° behind the force; that is,
the displacement is in a direction opposite to the direction of the
force. It follows that the high-frequency properties of plasma are
the reverse of those of ordinary dielectrics.
In order to obtain a better understanding of these special proper
ties of plasma, consider the situation illustrated in Figs. 43a and b.
These figures represent the displacement of charges in an ordinary
dielectric and in plasma. For convenience, it is assumed that in
both cases only the electrons are displaced. Inspection of the left-
hand drawing, which represents an ordinary dielectric, shows that
the electrons move in the direction of the forces acting upon them,
Plasma in High-frequency Field 101
E E
F igure 4 3 . Displacement o f charges in an electric field:

a. ordinary dielectric, b. plasma.
that is, in the direction opposite to the direction of E. The displace

ment of the electrons results in the appearance of charges on the
surface of the dielectric. The electrons appear on one surface and
the positive ions on the other. (In the absence of the field, the posi
tive charges neutralize the electrons.) The surface charges produce
a field AE whose direction is opposite to that of the external field;
that is, the resultant field in the dielectric, E ', is less than the applied
field. The ratio of the external field to the reduced field inside the
dielectric is known as the dielectric constant or permittivity and is
usually denoted by e. In general, the permittivity « is a function of
the frequency of the electric field and is always greater than unity.
For some substances, for example, barium titanate, the value of e at
moderate frequencies is greater than 104.
Consider now Fig. 43b, which illustrates what happens in plasma.
The electrons are now displaced in the direction of the field, and
thus the charges appearing on the surface tend to increase the. field
strength inside the plasma. Therefore, the permittivity of plasma
is less than unity. The lower the frequency of the oscillations, co, the
larger the amplitude of the electron oscillations. Therefore, with
increasing a>, the permittivity of plasma decreases; detailed calcula
tions show that 6 is zero when
which is the “critical frequency.” We shall encounter it again later.

It turns out that the angular frequency of the natural electron
oscillations in the plasma is equal to cok and the permittivity e is
given by
(5.12)
It is evident that when oj < <jk, the permittivity is negative.

This effect is related to an important property of plasma. About
one hundred years ago, Maxwell showed that the relation between
the permittivity t and the refractive index y for electromagnetic
waves in matter is
T= (5.13)
which shows that a negative t leads to an imaginary refractive index

and therefore an imaginary velocity of propagation of electromag
netic waves in matter (since this velocity is proportional to the ratio
of the velocity of light in vacuum to the refractive index). In other
words, when e is negative, electromagnetic waves cannot propagate
through matter and are reflected from it. Therefore, at frequencies
smaller than the critical frequency, plasma is a perfect reflector of
electromagnetic waves. This is the reason why radio waves can be
sent around the Earth. It is known that a layer of rarefied plasma,
called the ionosphere, is located between 40 and 200 km above the
Earth’s surface. Reflection of radio waves from the ionosphere is
responsible for the long-range propagation of radio waves, that is,
propagation to points lying beyond the visible horizon. The mini
mum wavelength \ k of radio waves which can pass through plasma
is given by
c _ lire
(5.14)
Tk wk
Substituting for the critical frequency wk from (5.11), we have
(5.15)
Motion under Action o f Pressure Difference 103
When X < X*, the refractive index of plasma can be obtained from
(5.16)
which is a consequence of (5.12), (5.13), and (5.14).
5.3 Motion under the Action of a Pressure Difference

Directed currents of electrons and ions can be produced in
plasma, not only by an electric field but also by a gradient in the
concentration of the particles. Consider the behavior of a non-
uniform plasma with a high degree of ionization. Figure 44 depicts
a region occupied by a nonuniform plasma whose concentration
decreases from left to right, in the direction of increasing x. We
shall assume for simplicity that the electron temperature Te is equal
to the ion temperature T, throughout the region. A change in the
concentration will then be proportional to the change in the pres
sure of the plasma since p = nk{Te + Tf). The decrease in the
o o o o
o
o o
o
o o
o
o ° o
O O o
F ig u r e 44. S ch em a tic represen tation o f a g as with a nonuniform distribu tion

o f con centration along the x axis.
concentration in the direction of the x axis means that there is a

corresponding decrease in the pressure.
Plasma, just as an ordinary gas, cannot exist in equilibrium in
the presence of a pressure gradient and will redistribute itself until
there are no pressure differences. In the present example, the
plasma will move from left to right. This process is analogous to
the expansion of a gas in an apparatus in which, by opening a
valve, we release the gas from a high pressure region into a low
pressure region.
The simplest example of this is the expansion of a gas into a
vacuum. This expansion occurs with a velocity of the order of the
mean thermal velocity of the molecules. A similar situation should
obtain in a stream of expanding plasma. However, plasma con
sists of two components and the problem arises as to which of the
two thermal velocities — the electron or the ion velocity — should
play the leading role in the expansion process. This question can
easily be answered. Since the plasma as a whole is quasi-neutral,
the electrons cannot leave the slowly moving ions far behind, and
therefore the plasma as a whole will expand with the thermal
velocity of the ions.
There are, however, some exceptions to this general rule. If a
concentration of plasma with a nonuniform pressure distribution is
formed in a very short interval of time in a region bordering on a
high vacuum, then at the initial instant of time a relatively small
number of electrons will leave the surface of the plasma with
thermal velocities. The electric field which is produced as a result
will, under certain and still somewhat obscure conditions, lead to
the appearance of a very brief and weak stream of very fast ions
moving with velocities approaching the thermal velocity of the
electrons. A very large number of electrons is necessary to produce
a single fast ion, since the velocity of the ion is taken at the expense
of the velocity of the electrons. This situation is occasionally ob
served in electric discharges of short duration in a low-density gas.
However, it may be regarded as an exception which does not in
validate the general rule governing the motion of expanding
plasma as a whole.
So far, we have only been concerned with the limiting case of
nonuniform plasma with a very large relative variation of pressure.

Small pressure changes are much more frequently encountered.
The time necessary for the pressures and concentrations to become
equal is then much longer. In the general case, analysis of this
situation can be based on the basic laws of mechanics.
A pressure difference gives rise to a force which accelerates the
gas in accordance with the second law of Newton. Let us try to
determine the effect of this force by first estimating its magnitude.
Consider a unit cube in a gas, which is oriented so that one of its
edges is parallel to the direction of the pressure drop. The force
acting on a side surface of the unit cube is numerically equal to the
pressure (since pressure is equal to the force acting per unit surface
area). On the left, the pressure is equal to p0, and on the right it is
equal to p 1, which is smaller than p0. It follows that the force acting
on the unit cube is equal to p0 — p x. This pressure difference will,
in fact, accelerate the gas in the unit cube. Since the mass of the
gas in the unit cube is equal to its density p, it follows that the ac
celeration is equal to (p0 — pf)/p. This is, of course, only a very
approximate calculation, since the acceleration will vary from point
to point within the cube, owing to the differences in the density and
the relative pressure difference.
Let us therefore replace the unit cube by a very small cube of side
£. The area of a side surface of this cube is then £2, its volume is £3,
and the mass of the enclosed gas is p£3. The resultant force giving
rise to the motion of the gas is (p0 — pf)i-2, where po is the pressure
on the left and p x the pressure on the right. The acceleration is
given by
= Po ~ P i, 1 (5.17)
£ p
The quantity (p0 — /?,)/£ is the pressure drop per unit length.
However, this is not a simple quantity, since it is characterized both
by a magnitude and a direction. In point of fact, it is a special type
of vector. When preceded by a minus sign, it is called the pressure
gradient, which in vector analysis is represented by the symbol
grad p. The negative sign is necessary to allow for the fact that the
direction of the gradient is in the direction of increasing pressure,
whereas the force acting on the gas is in the direction of decreasing

p. The concept of the gradient will be frequently encountered in the
following chapters, because in the physics of plasma we are always
concerned with the gradients of pressure, concentration, and
temperature. It will therefore be useful to discuss this in somewhat
greater detail.
In the above simple case there was no doubt as to the direction of
the pressure gradient, since it was assumed that the pressure varied
along the x axis only; that is, it was a function of a single co
ordinate only. In general, the pressure inside the plasma may vary
in all directions and the form of the variation will not, in general,
be very simple. Suppose, for example, that the plasma is in the
form of a cylinder, not necessarily of circular cross section. A
cylinder of this kind can be produced in an electric discharge in a
long glass or ceramic tube. When the cylinder of plasma is long
enough, the pressure in its middle region will be nearly constant in
the longitudinal direction, but will decrease in the transverse
direction. The pressure distribution over a typical cross section of
the cylinder might be of the form shown in Fig. 45. The closed
lines in this figure represent lines of equal pressure. In physical
geography they are known as isobars. The larger the number of
F igure 4 5 . Definition o f gradient. The direction o f the gradient is

indicated by an arrow.
isobars which are drawn on a particular map, the more accurately

they represent the distribution of pressure in the plasma. The
point M lies pn an isobar on which the pressure is, say, p {. The
pressure varies in all directions except for the direction parallel to
the isobar (the line DD in Fig. 45). However, the form of the varia
tion in p is different along different lines drawn through M. The
next isobar, which represents the somewhat higher pressure p2, can
be reached in various ways, for example, along MA, MB, or MC.
In all such cases the total change in the pressure is equal to p2 —Pu
although it occurs over different lengths. The maximum rate of
change in the pressure occurs in the direction of MA, which cor
responds to the shortest distance between the two isobars. The ray
MA is approximately perpendicular to both isobars. The direction
of the most rapid change in the pressure is, in fact, the direction of
the pressure gradient which is numerically equal to {p2 — p f)/M A ,
provided that the neighboring isobars do not differ in pressure by
too much, that is, provided that the network of isobars is sufficient
ly dense. This definition of the gradient may be generalized to the
case in which the pressure varies not only in a plane, but in all di
rections in space. In this most general situation we must consider
surfaces of constant pressure, and the gradient is in the direction of
the shortest distance between two neighboring surfaces of constant
pressure, that is to the perpendicular between them (Fig. 46). This
F igure 46. Surfaces o f constant pressure. The line M A is parallel to the

direction o f the gradient.
analysis of the concept of the gradient is quite general. It can be

used to define gradients in the concentration, temperature, electric
potential, and so on. The gradient of any quantity varying in space
is measured by the increase in that quantity per unit length in the
direction of its most rapid variation.
Under certain conditions, a nonuniform distribution of charged
particles in plasma may give rise to another kind of motion which
is known as diffusion. In a completely ionized plasma, the diffusion
process may be isolated from dynamic processes due to pressure
differences, only if the plasma contains a number of different kinds
of ions with different spatial distributions. However, a much more
interesting case is the phenomenon of diffusion of charged particles
in weakly ionized plasma when the pressure due to the electron and
ion components is small in comparison with the pressure of the
neutral gas.
Diffusion is similar to the flow of a current in that it involves the
mixing of the components of the medium rather than the displace
ment of the medium as a whole. A typical example of diffusion in
everyday life is the spread of the scent due to a drop of aromatic
substance in a room, or the gradual disappearance of a puff of
cigarette smoke. A clear picture of the diffusion process can be ob
tained by observing the behavior of a microscopic particle in a
liquid under a microscope. (This is the Brownian motion; see
Fig. 29.)
Diffusion in a gas, and in particular in plasma, is the result
of the random motion of the particles (atoms, ions, or elec
trons). As a result of collisions, a particle frequently changes its
direction of motion, and its path takes the form of a tortuous zig
zag line. If at an initial time the particle is at the point 0 (see Fig.
29), then as a result of its random motion it will depart from this
point in the course of time. However, since the motion of the parti
cle takes place in a randomly varying direction, its distance from the
point 0 will not be proportional to time. Calculations based on the
theory of probability show that in point of fact the distance r is
proportional to fft, that is r = sjAt, where A is a constant which
characterizes the rate of diffusion under different conditions.
Consider now the macroscopic effect of such random motions in
o "
rC t *
Oa
cU rdn cx
c<D
O
c 3
o Ea
U
n • cwo M ^
•
• x
. 0*On« o • •
• ^ • u
. 0.0 • • • . o , • • •
n*
o’. °.O o^ 0o 0-0 . .*
• • • o
•
o
° - o °
o . ° • o • . •
o
oo.°*o
• ° * o • n : • : o
o o n
F ig u r e 4 7 . Schematicrepresentation o f a nonuniform distribution o f

impurity particles (large circles) in the volume occupied
by a gas (small circles).
the case of a gas containing a small amount of impurity, which is

distributed nonuniformly as shown in Fig. 47. On the left of the
plane M N the concentration of the impurity particles is greater than
on the right. Therefore, during their random motion, more parti
cles will travel from left to right than in the reverse direction.
This means that there is a net flux of impurity particles in the
positive direction of the x axis. The intensity of this current across
the plane M N is equal to the difference between the number of
particles passing per second per unit area in the direction of the
arrow, and the number of particles passing through unit area per
second in the opposite direction. It is clear that the larger the differ
ence in the concentration of the particles from left to right, the
larger the magnitude of the current. The variation in the concentra
tion is indicated by the curve in Fig. 47. The slope of this curve
gives the rate of drop in the concentration n of the impurity. The
relative change in the concentration over a small part of the x axis is
a measure of this drop. It is precisely the concentration gradient
(taken with a negative sign). According to the basic assumptions of
the theory of diffusion, which have been verified experimentally, the
intensity of the current of particles should be proportional to
—grad n. If we denote the diffusion current by Q, we have
Q = —D grad n, (5.18)
where D is the diffusion coefficient. This coefficient is related to the
various quantities which govern the motion of the particles in the
gas. It is clear that the larger the mean velocity and the mean free
path X of the particles, the faster they will diffuse. Therefore, for a
given variation of the concentration, the diffusion current should
increase with increasing velocity and mean free path. This means
that the diffusion coefficient should be a function of v and X. Cal
culations show that
D = -v \. (5.19)
For readers who are interested not only in the theoretical results
expressed by Equations (5.18) and (5.19), but also in the way in
which these formulas may be obtained, we give a derivation based
on the simplest ideas (whose main disadvantage is its somewhat
formal nature). This derivation is unimportant to our subsequent
discussion and may be omitted altogether by readers who are not
interested in mathematical proofs.
A change in the concentration of particles in space means that
there is also a change in the pressure of the component of the gas to
which these particles belong. If the temperature is assumed to be
the same at all points, then Equation (3.1), which relates p, n, and T,
shows that the pressure gradient is simply proportional to the con
centration gradient, and their ratio is equal to kT. However, as has
Motion under Action o f Pressure Difference
already been pointed out, the quantity —grad p is simply the force
acting on all the particles enclosed in a unit volume. The force
acting on a single particle is therefore given by
kT
F = — grad n. (5.20)
In diffusive motion this force is balanced by the frictional force

due to collisions between impurity particles and atoms in the main
mass of the gas. The latter force is proportional to the mean veloc
ity of directed motion u of the impurity atoms. To determine u we
can use an argument analogous to that employed in Section 5.1 to
calculate the current in a plasma. The frictional force is equal to the
momentum which is transferred per unit time by a moving particle
to the atoms in the main component of the gas. The total loss of
momentum per second is equal to muv, where m is the mass of the
particle and v is the number of collisions per second. In equilibrium
this quantity must be equal to F, and therefore •
kT
muv = ------grad n. (5.21)
n
Hence
kT
nu = ------grad n. (5.22)
mv
The quantity on the left-hand side of this equation is the diffusion
current Q. We have thus proved Equation (5.18). The coefficient in
front of the concentration gradient is none other but the diffusion
coefficient D, and therefore
kT _ kT \
(5.23)
mv mv
The mean kinetic energy of the particles is related to the tempera
ture by Equation (1.2), which we shall now write in the form
W = -U 2 = jjkT.
Substituting the expression for k T into Equation (5.23), we arrive at

Equation (5.19).
The above derivation (which does not pretend to be rigorous) is

particularly instructive in one respect. It shows that diffusion is a
process which is in some ways similar to the electric current. It is
due to a force which brings the particles into motion. This force is
the pressure difference. Owing to the presence of the gaseous phase
in which diffusion takes place, the pressure difference is balanced
by a frictional force. Moreover, there is a clear distinction between
diffusion and all processes involving the motion of a gas as a whole,
which occur whenever there is an uncompensated pressure gradient
in the main component.
So far, we have been concerned with diffusion in a neutral gas.
Let us now consider how it can take place in plasma. It should be
clear from the foregoing discussion that if we are dealing with fully
ionized plasma, we cannot speak of diffusion in the presence of a
concentration gradient because the plasma is then set in motion as a
whole. When the degree of ionization is small, the situation is
quite different. The electrons and ions then form an impurity com
ponent whose density and pressure are small in comparison with the
corresponding quantities for the main neutral component of the
partially ionized gas. If there are irregularities in the distribution of
the charged particles, they will diffuse through the neutral com
ponent. A characteristic feature of this process is that, since the
plasma is quasi-neutral, the rate of diffusion should be the same for
both electrons and ions. Since the electrons have a greater mobility,
they leave the ions behind giving rise to an electric field which de
celerates them very rapidly while slightly accelerating the heavy
ions. As a result, the velocities tend to become equal, and the diffu
sion process occurs at a rate which is nearly the same as the rate of
diffusion of the ions in the absence of the electric field. The
diffusion coefficient for plasma can be determined from (5.19) by
substituting into it the values of v and X calculated for ions. The
process which involves the simultaneous motion of ions and elec
trons through the gas is known as ambipolar diffusion.
The thermal conductivity of plasma is also due to the motion of
particles. The main role in the transport of heat from the hotter
parts of plasma to the cooler parts is played by the electrons (be
cause of the higher thermal velocity which they have). If there is a
temperature drop in a given direction, electrons with higher energies

will travel toward one side, while electrons with lower energies will
travel in the opposite direction. As a result, there is a net flow of
thermal energy toward the colder layers of plasma. This flow is
proportional to the relative temperature drop, that is, the tempera
ture gradient. In fact, the expression for the flow of heat through
unit area is
Qt = ~ Cr grad T, (5.24)
where CT is the thermal conductivity. The higher the temperature
of the plasma (we are assuming for the sake of simplicity that the
temperature is the same for both electrons and ions), the higher the
thermal conductivity. The magnitude of CT in fully ionized plasma,
containing singly charged ions only, can be estimated from the
following very approximate formula
CT = 1.24 X 10" 6 T5/2 erg per cm deg. (5.25)
The thermal conductivity given by this formula is expressed in c.g.s.
units. If the heat flow is expressed in calories, then the thermal con
ductivity is given in calories per degree per centimeter. This will
necessitate the replacement of the numerical factor in (5.25) by the
new factor 3 X 10~ 14 cal per deg. It follows from Equation (5.25)
that the thermal conductivity of a fully ionized plasma increases
very rapidly with temperature. At T ~ 105 °K, the thermal conduc
tivity of hydrogen plasma is higher than the thermal conductivity
of silver at room temperature.
6
Experimental Techniques
In order not to give the impression that the physics of plasma is a

purely theoretical subject, consisting solely of formulas and abstract
ideas, we must give, even if only briefly, an outline of the more im
portant experimental methods which are used to investigate the
properties of plasma. The main practical problem is to determine
the basic parameters which characterize the state of plasma, that is
to say, the concentration of charged particles, the mean kinetic
energy, the energy distribution of electrons, and the mean energy of
the ions. Moreover, it may be necessary to measure the plasma
drift velocity, the flow of energy to the walls, the current of acceler
ated electrons, and so on. A very large number of different experi
mental techniques is available for these purposes. The most
valuable information has been obtained from (a) Langmuir probes,
(b) investigations of the passage of radio waves through plasma,
and (c) spectroscopic studies of plasma. These three methods are
considered in turn in the following sections.
6.1 The Probe Method

The probe method, which was first introduced by Langmuir
about forty years ago, can be used to determine the electron con
centration, temperature, and energy spectrum, and to estimate the
ion temperature. In this method, a measurement is made of the
current to a small metal electrode (probe) when different voltages
are applied to the probe. This yields a curve called the probe
characteristic of plasma. Figure 48 illustrates schematically the
114
Probe Method 115
F igure 48. The Langmuir probe: 1 — insulating tube, 2 — wall o f discharge

chamber, 3 — plasma, 4 — probe.
experimental arrangement, whilst Fig. 49 shows a typical probe

characteristic (in somewhat idealized form). The probe may be in
the form of a plane disk, cylinder, or sphere. The wire which is
used to conduct the current from the probe is surrounded by an
insulating cover. It is important to ensure that the dimensions of
the probe are small, since otherwise the presence of the probe
affects the state of the plasma.
Consider now a probe characteristic in which the voltage applied
to the probe is plotted along the * axis, and the current flowing
F igure 49. Tpyical probe characteristic.

116 Experimental Techniques
into the probe from the plasma along the y axis. Positive values of
the current represent an electron current; negative values represent
an ion current. On gradually increasing the voltage applied to the
probe, we successively pass through three different parts of the
probe characteristic. In the region marked A in Fig. 49, the current
flowing to the probe is due to positive ions. In region B there is a
rapid increase in the electron current which reaches saturation (con
stant value) in the region C .'
This form of the curve can be quite simply explained. When the
voltage applied to the probe relative to the plasma is large and
negative, the electrons cannot reach the probe and the current is
due to positive ions only. The current due to these ions is rela
tively small since, for equal concentrations of electrons and ions,
the latter have a much smaller velocity, whilst the magnitude of the
current is proportional to the product of the concentration of the
particles and their velocity. The electron current is cut off so long
as the negative potential on the probe is high enough to prevent the
fastest electrons, in the tail of the Maxwellian distribution, from
reaching the probe. In order to reach the probe, the electrons must
ascend the retarding potential difference and lose kinetic energy by
doing work against the electric field.
Suppose that the absolute magnitude of the retarding potential
difference is U, which means that the potential of the probe relative
to the plasma is — U. Only those electrons can reach the probe
whose kinetic energy is greater than e U. Since the kinetic energy of
the electrons is of the order of kTe, it follows that the electron cur
rent to the probe will remain cut off as long as kTe is much less than
e U. A current will appear when kTc and e U are of the same order of
magnitude. Calculations show that if the electron energy distribu
tion follows Maxwell’s law, then the fraction of the total electron
current which reaches a probe at a potential U is e~eUlkT«. Therefore,
when eU = kTc, only 37% of the electron current will reach the
probe. When eU = 2kTe this figure becomes 12%, and when
eU = 3kTe, only 4% of the electrons reach the probe. Thus the
general expression for the probe current in this region of the char
acteristic is
/ = /, + Ioe-'VtkTe, (6 . 1)
Probe Method 117
where e is the base of the natural logarithms (e = 2.718), To is the

electron current in the absence of a retarding field, and /,• is the ion
component of the current. The latter component is opposite in
sign to the electron current, but its magnitude is very small in com
parison with the electron current. The ion current /, may be deter
mined from the part of the characteristic which lies in region A, and
we may assume that it remains practically the same in region B.
By subtracting the ion current, we obtain the electron component
for any retarding potential U. The potential U must, of course, be
measured from the point on the * axis at which the knee of the curve
begins, that is from the point at which the electron current ceases
to increase, having reached its limiting value.
In practice, it is convenient to interpret probe measurements by
plotting the logarithm of the electron current as a function of the
probe potential U. In view of Equation (6.1) this graph (Fig. 50)
represents the equation
( 6 . 2)
The slope of the graph is e/kT e from which the absolute electron
temperature Te can be calculated. We note that this method of
determining Te is based on the assumption that the electron energy
distribution is Maxwellian, so that the electron current is propor
tional to e~eU,kT«. The probe characteristic itself may serve as a
Inle a
—U
F igure 50. Probe characteristic on a semilogarithmic scale.

confirmation of the fact that the distribution is Maxwellian. This

is best shown by the logarithmic graph which we have just men
tioned. The dependence of the logarithm of the electron current on
U will be linear, provided Maxwell’s law is valid. Experiment
shows that this is practically always the case, and therefore the
above method for the determination of Te can be used in most cases.
Knowing Te and the total electron current / 0, we can determine
the concentration of electrons in plasma. In order to simplify our
discussion, let us suppose that the probe is in the form of a flat disk.
The electron current per square centimeter of surface of the probe is
proportional to neVe, where ve is the mean thermal velocity of the
electrons in the plasma. If the velocity of each of the plasma elec
trons were perpendicular to the plane of the probe, and all the elec
trons were moving towards the probe, then the magnitude of the
current which we have just mentioned would not be merely propor
tional, but actually equal to neve. In reality, only one half of the
electrons move towards the probe, and the other half move in the
opposite direction. This reduces the current by a factor of 2. More
over, it is important to take into account the fact that the velocity
of the electrons is not necessarily perpendicular to the plane of the
probe. Calculations show that when this effect is taken into ac
count the current is reduced by an additional factor of 2. There
fore, the final expression for the maximum electron current to a
probe of area 5 is given by
/. = (6.3)
This formula may be used to determine the electron concentration

in plasma from probe measurements. In principle, it is possible to
determine both the electron and the ion temperature from the probe
characteristic. However, the ion temperature is usually much
smaller than the electron temperature and is exceptionally difficult
to determine, since there are many factors which mask the effect of
the ion temperature on the form of the probe characteristic.
Probe characteristics provide unambiguous information about
properties of plasma only under certain special conditions. For
example, when the density of the plasma is high, so that the mean
free paths of the electrons and ions are very small, the above simple
Radio-wave Measurements 119
method of measuring Te and ne cannot be used. The interpretation

of probe measurements is also very difficult, or even quite impossi
ble, when a strong magnetic field is present in the plasma. However,
in spite of these limitations, the probe method is in practice one of
the most widely used methods because of its relative simplicity.
6.2 Radio-wave Measurements

The electron concentration can be measured, or at least estimated
by studying the transmission of radio waves through plasma.
According to Equation (5.15), the maximum wavelength of radio
waves which will pass through plasma with electron concentration
ne is 3.3 X 106/yjne. Therefore, when ne = 106 particles per cm3,
the critical wavelength is equal to 33 m, while when ne — 1012
particles per cm 3 the limiting wavelength is reduced to 3.3 cm. The
basic arrangement for this kind of experiment is illustrated sche
matically in Fig. 51.
F igure 51. Basic arrangement fo r measuring the absorption o f electro

magnetic waves in plasma: 1 — high-frequency oscillator, 2 — receiver.
Radio waves from a high-frequency oscillator are transmitted

through a hollow metal waveguide A leaving it through the horn B
in the form of a collimated beam. This beam is then allowed to
pass through the chamber containing the plasma and is received by
another horn C. The wavelength X is determined by the properties
of the oscillator and cannot be varied within wide limits in a given
measuring channel. If the receiver records that the radiation has
passed through the chamber, this means that the wavelength X is
smaller than the critical value. This immediately provides a value
for the upper limit of the electron concentration. Thus, if the
plasma transmits the radiation we have ne < 1.1 X 1013 /X2, while
if it does not, then ne > 1.1 X 1013/X2.
These upper and lower limits are not, of course, sufficient.
More complete information may be obtained by using two or more
wavelengths at the same time. Suppose, for example, that a beam
containing waves with X = 4 mm and X = 8 mm is employed, and it
is found that the 4 mm waves are cut off. This means that ne must
lie between 2 X 1013 and 8 X 1013. In order to determine the
electron concentration rather than find its upper and lower limits,
it is necessary to modify the experimental method. For example,
the determination of ne may be based on the fact that the refractive
index of plasma is a function of the electron concentration, and
therefore the velocity of propagation of electromagnetic waves in
plasma is also a function of ne. Let us suppose that electromagnetic
oscillations of frequency v pass through a layer of plasma of thick
ness d. The number of wavelengths which may be fitted into a
segment of length d is d /\, where X is the wavelength in the plasma.
Since X = u/v, where u is the velocity of propagation of the waves
in the plasma, and u = c/y where y is the refractive index, we find
that the number of wavelengths which can be fitted into the seg
ment of length d is, equal to vyd/c. This may be rewritten in the
form yd /k0 where X0 is the wavelength of waves of the same fre
quency but propagating in a vacuum.
Figure 52 illustrates schematically an apparatus which may be
used to determine y by measuring the number of wavelengths which
can be fitted into a layer of plasma. The high-frequency oscillator
emits electromagnetic waves which pass along a metal waveguide.
Radio-wave Measurements 121
F igure 52. Basic arrangement for measuring the refractive index o f plasma
by measuring the number o f wavelengths which can be fitte d into a layer o f
plasma: 1 — high-frequency oscillator, 2 — receiver.
At the point A the radiation is divided into two parts, one of which
travels via B i along the bent continuous waveguide, while the other
travels along B 2 and passes through a chamber containing the
plasma. The two waves are reunited at C. The intensity of the
radiation recorded by the receiver depends on the result of the re
combination at C. If the two waves reach the point C in phase,
that is if the crests of one coincide with the crests of the other, as in
Fig. 53a, then the two waves interfere constructively and the signal
recorded by the receiver reaches a maximum value. If, on the other
hand, the waves differ in phase by 180° and the crests of one of the
waves correspond to the troughs of the other, then there is destruc
tive interference, and the strength of the signal is reduced (Fig. 53b).
Suppose that, to begin with, there is no plasma in the container.
F ig u r e 53. Graphs illustrating the interference o f signals propagating along

the two channels B t and B i in Fig. 52: a. maximum resultant amplitude,
b. minimum resultant amplitude.
The geometry of the waveguides can be adjusted so that the waves

reaching C have zero phase difference, and therefore add construc
tively. If the chamber is now filled with plasma whose concentra
tion is gradually increased, the phase difference between the waves
reaching C will gradually increase, since the number of wavelengths
occupying the region filled by plasma will vary. If the concentra
tion of plasma is small, the refractive index y will be very nearly
equal to unity, and the number of wavelengths fitting into the plas
ma layer will differ from the corresponding number in the absence
of the plasma by a small fraction of one oscillation. This means
that the crests of the waves at C will be separated by a very small
amount. As ne is increased, the phase difference will increase and
the amplitude of the resultant wave reaching the detector will de
crease. For a certain value of ne the phase difference will be 180°
and the signal recorded by the receiver will be a minimum. Further
Plasma Spectrometry 123
increase of ne will lead to the eventual coincidence of the crests,

which is associated with a new maximum in the strength of the
signal, and so on.
All this is shown graphically in Fig. 53. It follows that the ampli
tude of the received signal is a periodic function of ne. By measur
ing the strength of the signal with gradually increasing plasma
density, it is possible to determine the refractive index 7 at any
particular density and then, by using Equations (5.15) and (5.16),
it is possible to find the concentration. This is called the radio
interferometer method, since it is based on the interference between
two waves. A necessary condition for the successful operation of
the interferometer is that the frequency of the radio waves should
be greater than the critical frequency for the particular state of the
plasma. In addition, the thickness of the layer of plasma, through
which the radio waves are transmitted, must be known.
6.3 Plasma Spectrometry

One of the more important sources of information about the
properties of plasma are studies of the spectral composition of the
radiation emitted by plasma. Such studies may be carried out with
various spectographs, that is, normal prism instruments with glass
or quartz optics, which are suitable for the analysis of the visible
and near ultraviolet radiation, or vacuum spectographs with diffrac
tion gratings, which are suitable for short wave radiation (far ultra
violet or even x rays). The radiation enters the slit of the spectro
graph, either through a window in the vessel containing the plasma
or through a connecting evacuated pipe (if one is interested in
radiation which does not pass through glass or quartz). The in
tensity of the spectral lines emitted by plasma depends on the con
centration of electrons and ions, and on the electron temperature.
With increasing electron temperature, the spectrum changes: New
lines, corresponding to a higher degree of excitation of the atoms,
appear. At high electron concentrations, lines due to excited
neutral atoms have an appreciable intensity only if the electron
temperature is low enough, and disappear very rapidly as this
temperature increases (owing to ionization of the neutral compo-
nent). They are replaced by lines due to singly- or multiply-

charged ions.
A very approximate estimate of the electron temperature can be
obtained by measuring the relative intensity of the spectral lines
emitted by a given atom or ion. For example, let us suppose that an
excitation energy W\ must be communicated to an atom or ion
before it can emit a particular spectral line, while the energy neces
sary for the appearance of another line is fVn > W\. The intensity
of each of these lines will depend on the number of electrons which
have an energy sufficient to produce the excitation process. If the
excitation energy is much greater than kTe, the relative number of
such electrons, and therefore the intensity of the line, will be largely
determined by the factor e~wlkT«. Therefore, the ratio of intensities
of the two lines is primarily governed by the ratio
Q -W pkTe
e - W u lk T e
eW u-w pkT e' (6.4)
This does not mean that the ratio of intensities will be precisely
equal to this quantity, since the intensity is also a function of the
internal properties of the atom, which have an effect on the transi
tion probability. However, these additional factors are tempera
ture-independent and can be approximately taken into account.
Therefore, when W\\ — W\ is much greater than kTe, the factor
e (wi-wu)ikT, is v e r y sensitive to the ratio of the line intensities, and
its magnitude may be used to estimate the electron temperature.
In some cases, it is possible to use other spectroscopic measure
ments to determine Te. For example, if the plasma has a very high
concentration, then in the far infrared region it emits as a perfect
blackbody at the temperature Te. Therefore, by measuring the
absolute flux of radiant energy emitted by plasma in the form of
infrared radiation, it is possible to estimate the electron tempera
ture.
Spectroscopic determinations of plasma concentration are very
difficult because they involve the accurate measurement of the
absolute intensities of spectral lines. The temperature of the atoms
and ions in plasma is usually deduced from various indirect data.
However, in some cases, spectroscopic measurements can also be
Plasma Spectrometry 125
used for this purpose. The point is that the temperature of the
atoms is related to the width of the spectral lines emitted by them.
This is due to the Doppler effect, which relates the frequency of a
wave to the velocity of the source. A simple example of the
Doppler effect is the increase in the pitch of the whistle from an ap
proaching locomotive, and the reduction in this pitch which is ob
served as soon as the locomotive moves away. These phenomena
are encountered both in acoustics and in optics. For example, if a
source of light moves towards the receiver with a velocity u, the
frequency of the oscillations measured by the receiver is given by
where vQis the frequency emitted by a stationary source, u is the

velocity of the source, and c the velocity of light. The wavelength
is related to frequency, and must therefore also depend on u. As
t. e source of light approaches the recording instrument (spectro
graph), the wavelength will decrease, and vice versa. In plasma, the
atoms and ions move at random with a mean velocity proportional
to ffT. Therefore, different particles emit slightly different wave
lengths, with the result that a narrow spectral line is broadened
(Fig. 54). Measurements of the width of spectral lines may be
used to determine the temperature of the particles which are re
sponsible for the emission of the radiation.
F ig u re 54. D o p p le r b ro a d e n in g o f a s p e c tr a l lin e : a. lin e p r o file a t lo w

ion te m p e r a tu r e , b. b ro a d e n in g o f th e lin e f r o m p la s m a
w ith a h igh ion te m p e ra tu re .
7
Plasma in a Magnetic Field
7.1 Forces Acting on Plasma in a Magnetic Field

The properties of plasma are radically altered when it is placed in
a strong magnetic field. The reason for this is that the motion of the
charged plasma particles is affected by the magnetic field. In a
strong magnetic field, the electrons and ions cannot move freely
in the direction perpendicular to the lines of force. The trajectory
of each particle takes the form of a helix with axis parallel to the
magnetic field, and the motion is therefore highly anisotropic
(directed). Figure 55 illustrates the situation for a fully ionized
plasma both in the absence and in the presence of a magnetic field.
The displacement of electrons and ions across lines of force is
a b
55. Schematic illustration o f the motion o f particles in fully ionized

F ig u r e
plasma: a. in the absence o f a magnetic field, b. in a strong magnetic field.
126
Forces Acting on Plasma in Magnetic Field 127
only possible as a result of collisions between the particles. In each

such collision, the particles are displaced only through a distance of
the order of the Larmor radius. In a fully ionized plasma of given
concentration, the probability of a collision decreases rapidly with
increasing temperature since the average interval of time between
collisions is proportional to the cube of the mean velocity, that is,
to T 3/2. Therefore, although at high plasma temperatures a charged
particle, having a larger Larmor radius, is displaced through a
larger distance after each collision than at low temperatures, the
mean distance which it succeeds in traversing per second in the
direction perpendicular to the field is, nevertheless, found to de
crease with increasing temperature because of the rapid decrease in
the collision probability. This distance will also decrease with in
creasing field strength.
Because a magnetic field restricts the motion of charged particles,
it may be used to prevent the plasma from coming into contact
with the walls of the container (Fig. 56). In Figure 56, the fully
ionized high-temperature plasma occupies a cylindrical volume
inside a tube whose axis is parallel to the direction of the external
magnetic field. In the space between the surface of the plasma and
the walls of the tube there is only vacuum and magnetic lines of force.
— 1
—2
— 2
— 1
F igure 56. Insulation o f the walls o f a discharge chamber from plasma by

means o f a magnetic field: 1 — walls o f the chamber,
2 — vacuum, 3 — plasma.
128 Plasma in a Magnetic Field
The situation illustrated in Fig. 56 is not surprising and can easily

be explained by considering the motion of electrons and ions in the
magnetic field. However, there are considerable difficulties if one
tries to interpret the phenomenon from the macroscopic point of
view. Plasma retained by a magnetic field exhibits an internal pres
sure. In the absence of the magnetic field, this pressure would im
mediately lead to the expansion of the cylinder of plasma, which
would thus come into contact with the walls. The problem now
arises as to what is the force which balances this pressure.
It is clear that this force must, in some way, be related to the
magnetic field. It may be calculated as follows. A force due to a
magnetic field can only arise when an electric current flows. For
example, consider a conductor carrying a current at right-angles to
the magnetic field (Fig. 57). According to Ampere’s law, the force
acting on the conductor is I H / c per unit length, where / is the cur
rent in electrostatic units, H is the magnetic field, and c is the
velocity of light. This force acts at right-angles both to the lines of
force and to the current. Its direction is indicated by the arrow and
is given by the right-hand rule. If the current is distributed con
tinuously through the medium, we can determine the force per unit
volume. Thus consider a unit cube in a medium through which a
current flows and suppose that the cube is oriented so that the
current is perpendicular to one of its faces. The magnitude of the
current flowing through the unit area is equal to the current density
j in the medium. Suppose that the magnetic field is perpendicular to
the current. The unit cube may be looked upon as a conductor of
l 1H
c
F igure 57. Interaction o f a current-carrying conductor with an

external magnetic field.
unit length which carries a current j. The force acting on this con
ductor must be equal to jH /c.
This conclusion can be generalized somewhat by considering the
case when j and H are at an angle different from 90°. Under these
conditions, the field strength may be resolved into two components,
one of which is parallel (H n) and the other perpendicular (H±) to
the current. The component H\\ has no effect whatsoever on the
current and therefore the force is equal to jH ± /c. It can also be
written in the form jH sin d/c, where 0 is the angle between the direc
tions of j and H. In the language of vector algebra, with which we
are already familiar, this means that the force acting on the unit
cube may be written in the form j X H/c.
This is the force which balances the pressure in the plasma when
it is placed in a magnetic field. For equilibrium, it is necessary that
the electrodynamic force be equal and opposite to the forces acting
on the boundary layer due to the pressure drop across it. It has al
ready been explained in Chapter 3 that this force is equal to —grad
p, and therefore the condition for equilibrium of plasma in a mag
netic field is
i j X H = grad p. (7.1)
This formula is quite general. It holds both for the boundary layer
and for any volume element within the plasma.
In order to elucidate the physical significance of Equation (7.1),
we must first establish the reason for the appearance of a current.
This can be done with the aid of Fig. 58a, which shows the cross
section of a cylinder of plasma. The trajectory of each particle in
this cross section may be represented by a Larmor circle. For the
sake of clarity, only the electron Larmor orbits are shown. Let us
suppose that the pressure is constant over the cross section except
for a very narrow boundary layer where it suddenly drops to zero.
Under these conditions, the Larmor circles are uniformly dis
tributed over the area occupied by the plasma. The motion of the
particles inside the plasma does not result in a net current flow
since any given point is traversed with equal probability by particles
with opposite directions of motion. However, near the boundary,
F ig u r e 58. a. Cross section through a column o f plasma placed in a

longitudinal magnetic field , b. the distribution o f magnetic lines
o f force over the cross section.
the circular Larmor motion leads to the appearance of a current

which flows in the form of a thin ring. This current is due to the
fact that while at internal points the various Larmor currents cancel
out, at the surface they add up. The interaction of the macroscopic
surface current with the magnetic field is the reason for the ap
pearance of electrodynamic forces which prevent the plasma from
expanding.
One further important consequence of Equation (7.1) must be
noted. The force j X H /c is perpendicular to the directions of
both the current and the magnetic field. Therefore, grad p is also
perpendicular to j and H. This means that the pressure of the
plasma is constant along the lines of force. It should also be
constant along the lines of current.
The behavior of plasma in a magnetic field can also be described
in a different way by recalling that, as a result of the Larmor

rotation of the electrons and ions, the plasma as a whole exhibits
diamagnetic properties, so that the field inside the plasma is
smaller than the field outside, The conventional concept of
magnetic pressure, first introduced by Faraday, may be used in
this connection. The pressure due to the field is equal to H 2/8-ir
and its direction is perpendicular to the lines of force which in
Faraday’s theory are regarded as if they were elastic filaments
stretched in the direction of the field and compressed at right-
angles to it. In the case illustrated in Fig. 58b, all the lines of
force are parallel, but their density, that is, the number of lines
per unit area, which is a measure of the field strength, undergoes a
change at the surface. The density is higher outside the region
occupied by the plasma. Therefore, the magnetic pressure acts
in the inward direction and the difference between the magnetic
pressures, that is, the difference between H^/Zir and must
be equal and opposite to the intrinsic pressure of the plasma which
we are assuming to be constant throughout the cylinder. Hence,
_ H2 (7.2)
^ 87T 8t
In particular, if the diamagnetic Larmor currents reduce the

magnetic field strength inside the plasma to zero, then
m (7.3)
Sir
The higher the concentration and temperature of the plasma, the
greater the field strength required to balance the pressure.
The above three ways of describing the behavior of plasma in a
magnetic field are equivalent, since they describe the same physical
phenomenon in different ways. In particular. Equation (7.1) may
be applied to the current flowing in the boundary layer of the
plasma, and this will again yield Equation (7.2). Without repro
ducing this proof, we note that, qualitatively, the equivalence of
the two methods of determining the electrodynamic forces is a
consequence of the relationship between the current density j and
the change in H. If the current flows at right-angles to the uniform
4/
F igure 59. Origin o f the forces which compress a cylinder o f plasma when a
current flows through it (Hi — magnetic field due to the current).
magnetic field H, it gives rise to a field whose direction is the same

as that of H. Therefore, the field strength varies from point to
point, giving rise to a field gradient and therefore a difference in
the magnetic pressures.
The above formulas, which relate the pressure of the plasma
and the electrodynamic forces, are only valid if the plasma is in
equilibrium. It is possible, however, to imagine a situation when
there is no equilibrium. As an example, consider a cylinder of
cold, low-pressure plasma along the axis of which a rapidly in
creasing current is allowed to flow. This current gives rise to a
magnetic field whose lines of force surround the plasma as indicated
in Fig. 59. Interaction of the current with its own magnetic field
tends to compress the plasma in the radial direction. This effect
may be discussed in terms of the well-known elementary rule
which states that parallel currents attract; therefore, any current-
carrying conductor tends to contract in the radial direction. If
the pressure in the plasma is low, the force tending to compress it
is not completely balanced and the plasma will contract at an
increasing rate. We shall not discuss this particular case in detail
(it will be dealt with later) but will merely note that such non-
equilibrium situations are very frequent. They may be described

by replacing Equation (7.1) by the more general equation
pa = I j X H - grad p, (7.4)
where p is the density, that is, mass per unit volume, and a is the
acceleration of the plasma. All the quantities entering into this
equation are given per unit volume. Therefore, the left-hand side
consists of the density p multiplied by the acceleration, while the
right-hand side consists of the vector sum of forces acting on a
unit volume of the plasma.
Inspection of Equations (7.1) and (7.4) will show an important
property: These equations have nothing in them to indicate that
they refer to plasma, since the microscopic structure of the medium
enters into them under the guise of the macroscopic parameters p,
a, and p. The equations can equally well be used for the analysis
of phenomena occurring in a strong magnetic field in any con
ducting medium capable of changing its form under the action of
external forces. This means, in particular, that Equations (7.1)
and (7.4) are valid not only for plasma but also for a conducting
liquid, provided its conductivity is high enough and gravitational
and capillary forces can be neglected. Therefore, Equation (7.4)
is usually referred to as the basic equation of magnetohydrody
namics.
In our analysis of the behavior of plasma in a magnetic field,
we saw that Equation (7 .1) could be derived on the basis of macro
scopic considerations without explicitly taking into account the laws
governing the motion of the ions and electrons, and without
discussing the mechanism responsible for the appearance of the
electric field in plasma. As long as we are using the conducting
liquid model to describe the various processes occurring in plasma,
we may consider that the current can flow freely both along and
at right-angles to the lines of force. However, we cannot confine
our attention to this simplified theory because sooner or later we
shall encounter the following question: What effect has the
magnetic field on the current flowing through the plasma, and in
particular, under what conditions will the current flow at right-
angles to the magnetic field? The importance of this question

will become apparent if we recall that electrons and ions cannot,
in fact, move freely at right-angles to magnetic lines of force.
7.2 Current in a Magnetized Plasma

To begin with, let us consider the simple case when the electric
and magnetic fields are parallel. A given electron is then ac
celerated along the magnetic lines of force, and therefore the
magnetic field has no effect on the final velocity communicated
to the electron. Since collisions between charged particles occur
when they approach each other at very small distances, it follows
that the electric forces acting during the collisions are much
stronger than the interaction of the charged particles with the
magnetic field. (We are concerned here with magnetic fields of
the order of a few thousand oersteds, which can easily be produced
in the laboratory.) Therefore, when E is parallel to H, the magnetic
field should have very little effect on the flow of current.
Let us suppose now that an electric field E is introduced into a
uniform plasma in which there is a uniform magnetic field H
perpendicular to E. When the electric field is switched on, the elec
trons and ions begin to move in cycloidal or trochoidal orbits in
the direction parallel to the vector E X H. At the initial instant of
time, this will give rise to the appearance of a current since the elec
trons and ions will be displaced through a certain distance along the
electric field (see Fig. 60 where it is assumed for the sake of sim-
F igure 60. Motion o f particles in perpendicular electric and magnetic fields.

A t the initial instant o f time all the particles move so that a current is
produced in the direction o f the electric field.
Current in Magnetized Plasma 135
plicity that the trajectories are cycloidal). However, this current

will disappear as soon as the cyloidal motion settles down and the
entire plasma moves at right-angles to both E and H with the drift
velocity cE/H. Since the relative drift velocity of the electrons and
ions is zero, there will be no momentum transfer from the electrons
to the ions.
If in addition to the electrons and ions there is also a neutral
component, the charged particles will undergo collisions with the
neutral atoms and molecules, and this will tend to retard the drift
of the charged particles. This frictional force will have the same
direction but, in general, different magnitude for the electrons and
ions. It was pointed out in Chapter 2 that any force which is per
pendicular to the magnetic field gives rise to a drift. Therefore, the
frictional forces acting on the electrons and ions as they move
through the neutral component will also give rise to drift. The
velocity of the drift motion due to frictional forces will differ both in
magnitude and direction for the electron and ion components, and
this will give rise to an electric current in the plasma. A detailed
analysis shows that this current will have a component in the direc
tion of E and a further component parallel to E X H. It follows
that the resultant current will be at an angle both to the electric and
magnetic fields. Its magnitude decreases with increasing H and E
and with decreasing frequency of collision between the charged
particles and the neutral atoms.
This situation persists if the neutral component of the plasma
remains at rest as a whole. This is possible under certain conditions.
However, if the concentration of the charged particles in the
plasma is high enough and the volume of the plasma is large enough
for it to be considered practically infinite, then soon after the
electric field is switched on, the neutral component is brought into
motion with the drift velocity cE /H as a result of collisions with
electrons and ions. The relative velocity of the charged and neutral
particles is then zero and there is no transfer of drift momentum.
Therefore, the retarding force and the current due to it are also zero.
The volume of the plasma need not be infinite in all directions in
order to ensure that there is no electric current. It is only neces
sary that the drift motion with velocity cE /H should take place
V / / / / / / / / / Z ___
F igure 61. M otion o f plasma in perpendicular electric and magnetic fields:

a. in the electric field o f a plane parallel capacitor,
b. in a cyclindrical capacitor.
freely. This will occur, for example, when a layer of plasma is

placed inside a plane-parallel condenser of infinite length (Fig.
61a). A similar situation obtains in the case of plasma rotating
inside a cylindrical condenser (Fig. 61b). The current which flows
immediately after the electric field is switched on gives rise to
charges on the surface of the plasma, just as in an ordinary di
electric. The dielectric properties of plasma in a magnetic field
will be discussed later. For the moment, let us assume that there is
only an initial pulse of current which rapidly drops to zero. How
ever, a situation of this kind, where the drift motion occurs freely
in crossed fields, is the exception rather than the rule. In practice,
it is much more probable that, having begun its motion with
velocity cE/H, the plasma will come to rest after a short interval
of time, either because it encounters a wall or as a result of the ap
pearance of a large pressure drop which prevents any further
motion.
Any drift motion perpendicular to the magnetic field in plasma
is always due to a force which is perpendicular both to H and to
the direction of the drift itself. (The cycloidal motion of charged
particles in crossed fields is a special case of this general rule.)
Therefore, if the drift velocity is zero, this must mean that the force
which is the immediate cause of this drift is balanced by some other
force. It follows that in the case under consideration, the drift due
to the electric field E will terminate only if the force on each

particle due to this field is balanced by an equal and opposite force.
In a fully ionized plasma, this can only be the force of friction be
tween the electrons and ions. (In the case of free drift of particles
of either sign with the same velocity cE/H, this frictional force is
zero.) The compensation of forces leads to the reappearance of
normal plasma conductivity in the direction of the electric field,
since the conductivity is expressed by the equation eE = muv,
where u is the velocity of an electron in the direction of the electric
field.
Although it is perfectly correct, this discussion may nevertheless
appear to be unsatisfactory to the critical reader. Thus termination
of drift in the direction of E X H leads to the reappearance of
drift in the direction of E. Since this motion is also perpendicular
to the magnetic field, there should be a force perpendicular to E
which gives rise to drift of the electrons and ions in opposite direc
tions. This force is, in fact, the true cause of the electric current.
The problem now is: How can we derive this force?
In order to answer this question, we must start with the fact that
the required force should be directly connected with the physical
processes which appear as a result of drift in the direction of
E X H and which tend to prevent this drift. One such process is
the appearance of a variation of concentration and pressure in the
plasma. It is precisely this pressure variation which, in the presence
of a transverse magnetic field, will give rise to a force tending to
maintain the current even though the direction of the current and
the direction of the pressure gradient are at 90° to each other.
From the formal point of view, the situation is quite clear.
The force equal to —grad p gives rise to electron and ion drift
across the lines of force and at right-angles to the direction of the
force. The ions and electrons move in opposite directions, that is,
an electric current is produced. If grad p is perpendicular to E,
the current will be parallel to the direction of the electric field.
However, this purely formal argument must be amplified by a
discussion of the microscopic situation. It is useful to be able to
have a picture of how a nonuniform pressure distribution gives
rise to the appearance of currents of charged particles.
F igure 62. Appearance o f a current in a nonuniform plasma in a magnetic

field. Only the electron trajectories are shown.
In the special case when the pressure drop is localized in the

boundary layer, the mechanism responsible for the appearance of
the current has already been discussed above. A more general
case is illustrated in Fig. 62 where it is assumed, for the sake of
simplicity, that all the electrons have the same velocities, and
therefore the trajectories can be represented by circles of equal
radius. The concentration of particles increases from left to right.
Consider a narrow band which is parallel to the vertical axis (in
the figure it is defined by the two dashed lines). Electrons whose
orbits lie to the right of the band AB move in a downward direction
inside the band, while those on the left of the band move in an
upward direction. Since it is assumed that the electron concentra
tion increases from left to right, it follows that the number of elec
trons moving in the downward direction within the band must be
greater than the number of electrons moving in the opposite direc
tion. There is, therefore, a net current flowing within the band.
We thus arrive at an almost paradoxical situation: Each of the
electrons moves in a circular orbit without undergoing a net dis-
placement, but there is nevertheless a constant electron current in

the direction perpendicular to the concentration gradient, that is,
perpendicular to the pressure gradient.
The above discussion may be summarized as follows: If an
electric field E is set up in some way in plasma at right-angles to a
magnetic field H, the plasma as a whole will move with velocity
E X H. If there is nothing to retard this motion, the electric current
will flow for a very short time only. If, on the other hand, an
equilibrium is set up in which the drift motion reaches a steady
state, the current is reestablished, and its density is given by the
usual formula j = ? The real reason for the reappearance of the
current in this case is a pressure gradient in the electron component,
which is produced as a result of the drift motion.
The following remarks will complete our discussion of currents
in plasma:
1. It is clear that an electron pressure gradient may be present
when the plasma reaches its equilibrium state. The current-carriers
are the electrons, and therefore the electrodynamic forces due to the
interaction between the current and the magnetic field act on the
electrons. The fate of the ions can be discussed separately. Since
the plasma as a whole is quasi-neutral, it follows that the change in
concentration of the electrons should lead to a similar change in the
concentration of the ions. The problem then arises as to what is
the force which produces and maintains the ion pressure gradient.
It is evident that this force is due to the electric field which appears
as a result of the fact that electrons moving under the action of
electrodynamic forces tend to leave the ions behind. In equilib
rium, the ion pressure gradient is compensated by the forces on the
ions, which are due to the forward-moving electrons. This is
expressed by
eE' = - grad p i} (7.5)
where E' is the electric field due to the initial separation of the
electrons and ions. The factor 1/n on the right-hand side of
Equation (7.5) is introduced because we are dealing with the force
on a single ion rather than all the ions in a unit volume. Therefore,
when one is concerned with the validity of Ohm’s law in the case
of plasma in a magnetic field, one must remember that the relation
between the current and the electric field strength in the plasma does
not include the component of the electric field which is balanced by
the ion pressure gradient. This means that the field E' in Equation
(5.4) need not be taken into account.
2. When the plasma as a whole moves with velocity v, the rela
tion between the current and the field is
(7.6)
This equation shows that, during its motion across the magnetic
field, the plasma experiences an induced field represented by the
second term in the brackets. 14 This field must always be added to
the electric field produced by external sources. It follows from
Equation (7.6) that, in the case of plasma moving with a drift
velocity cE/H, the quantity in brackets must be equal to zero.
This gives a new and independent interpretation of the disappear
ance of the current in plasma moving freely in crossed fields.
The current in a drifting plasma is equal to zero simply because the
resultant electric field is zero.
3. Although the relation between j and E in the presence of a
transverse magnetic field in plasma in equilibrium is the same as in
the absence of the field, this does not mean that the physical picture
is identical in the two cases. In particular, it is obvious that, in the
presence of a strong transverse magnetic field, the electrons cannot
be continuously accelerated.
7.3 Diffusion in a Magnetic Field

The diffusion process, which is due to a nonuniform distribution
of the concentration of plasma particles, is completely altered in
the presence of a transverse magnetic field. Thus, in the absence
of a magnetic field the collisions between particles tend to retard
the process causing their concentrations to become equal. In
contrast, if a strong magnetic field is present in a nonuniform
14This field appears in any conductor moving through magnetic lines of force.
Diffusion in Magnetic Field 141
plasma, collisions are the only mechanism tending to equalize the

concentration in the direction at right-angles to H. If the particles
did not collide with each other, each of them could move freely
only in the direction of the lines of force, that is, parallel to the
direction of the magnetic field. The higher the temperature of the
plasma, the lower the collision frequency, and therefore the lower
the rate of diffusion. The diffusion coefficient for a fully ionized
plasma in a transverse magnetic field is given by
D = (7.7)
where the constant A is different for different substances. For

hydrogen it is equal to about 0.014, while for helium it is 0.0035.
Figure 63 illustrates schematically the change in the concentra
tion of plasma as a result of diffusion across the boundary between
two regions with different values of ne. It must be noted that, when
we are concerned with the diffusion of plasma in a magnetic field,
it is understood that the plasma as a whole is in motion rather than
n2
-►
F ig u r e 63. Variation in the concentration o f plasma as the result o f dif

fusion at the boundary between two regions with different ne.
Three successive situations are illustrated .
some small impurity within it. (This was assumed in the analysis
of diffusion processes in the absence of the field.) This aspect
represents a further fundamental difference between plasma
processes occurring in the absence and in the presence of a field.
When H = 0, a nonuniform distribution in the plasma pressure
leads to a current rather than to diffusion. If the magnetic field
strength is not zero, the difference in the plasma pressure is balanced
by a corresponding difference in the electrodynamic pressure in
such a way that the sum p + H 2/%-k is constant over the cross
section of the plasma (provided that the lines of force are straight).
Therefore, under these conditions, one should observe a gradual
disappearance of the boundary between the regions of different
values of p and a tendency towards a uniform spacial distribution
of plasma.
The outward diffusion current from the region occupied by
plasma leads to dispersion of the plasma which originally occupied
a clearly defined region in a magnetic field, with the result that
plasma eventually reaches the walls of the vessel. This phenome
non exhibits another aspect, namely, a decrease in the magnitude of
the current flowing in the boundary layer of the plasma. We
have already pointed out, in the preceding section, that this current
is a direct consequence of the very existence of plasma confined
by a magnetic field. The length of persistence of this current depends
on the duration of the diffusion process. If there were no collisions
between the particles, there would be no change in the configura
tion of the plasma in the plane perpendicular to H (since under these
conditions D = 0). It follows that the current would remain con
stant. This result can obviously be related to the fact that, in the
absence of collisions, the plasma behaves as a superconductor
(since the electrical conductivity is proportional to the average time
between collisions). By definition, the flow of current through a
superconductor does not require an electromotive force to maintain
the current.
The situation is quite different when collisions do take place.
The diffusive dispersion of the plasma gives rise to a change in its
concentration, and this, in its turn, leads to a change in the magnetic
field inside the plasma (since with decreasing concentration there
is a reduction in the diamagnetic effect, and therefore an increase

in H). According to the law of induction, a change in the magnetic
field passing through a plasma gives rise to an electromotive force
which is responsible for maintaining the current in the plasma.
It is easy to estimate the order of magnitude of the time necessary
for the plasma to disperse in space as a result of diffusion. If the
linear dimension, (for example, the diameter) of the region oc
cupied by the plasma in the direction perpendicular to the field is
L, the time necessary to retain the plasma in the magnetic field is
given by yj2Dt, and hence L 2/D.
Let us use this result to estimate the time during which a high-
temperature plasma can be retained in the magnetic field of a
thermonuclear reactor. Suppose that L = 50 cm, T e = T i = 1 0 8 °K,
ne = 1015 particles per cm3, and H = 10,000 Oe. Under these
conditions, the time interval in question turns out to be 180 sec.
In our previous analysis of the interaction of plasma with a
magnetic field, we have been concerned mainly with those processes
in which the plasma is a passive participant. However, in general,
the plasma may also behave as an active factor affecting the field.
This affects not only the diamagnetic properties of plasma, deter
mined by the resultant electron and ion pressure but also, in the
limiting case, leads to the complete exclusion of the field from the
region occupied by the plasma. The effect of the plasma on the
field may also be quite appreciable when the plasma as a whole
moves rapidly across the lines of force, even if the gas-kinetic
pressure nk(Te -f Ti) is small in comparison with the magnetic
pressure H 2/S t . The interaction is then due to the fact that the
plasma behaves as a conductor in which an electromotive force,
giving rise to a current, is produced as it moves through the
magnetic field. The induced current gives rise to its own magnetic
field and hence the initial magnetic field distribution is modified.
There is one further interesting property which characterizes
the interaction of plasma with a magnetic field. Suppose that in a
certain region of space the plasma and the magnetic field are
completely intermingled, so that the magnetic field strength H
and the plasma concentration n can, in principle, vary in an
arbitrary manner from point to point. We shall confine our at-
tention to the simple case where H and n vary only in the plane
perpendicular to the lines of force, but remain constant along the
lines of force. We can then imagine a filament of plasma in space,
which is parallel to H and is thin enough so that H and n may be
regarded as essentially constant. We shall suppose further that the
plasma as a whole is moving across the magnetic lines of force.
In all other respects, the motion will be assumed to be arbitrary.
Our imaginary filament of plasma may have a variable cross section
and may undergo compression and expansion, in which its cross-
sectional area will decrease or increase. Accordingly, there may
be a change in the concentration of the particles in a particular
volume element. Suppose that, at the initial instant of time, the
cross-sectional area of the filament of plasma is Si and the plasma
concentration is n\ so that the total number of particles per unit
length is n\S\. After a certain interval of time, the cross-sectional
area will change and become equal to, say, S 2 while the concentra
tion will be «2 - Since the total number of particles per unit length
must remain constant, it follows that «i*Si = H2S2 and, therefore
rti = S 2
(7.8)
112 Si
It must be noted that this result will only be valid if the motion
of the plasma particles is collective in character, that is, if the plasma
moves as a whole and the diffusion processes, tending to equalize
the concentration at different points, may be ignored. This condi
tion holds only for sufficiently rapid processes for which the time
intervals under consideration are much smaller than the time
necessary to retain the plasma in the magnetic field. For fast
motion, the magnetic flux in each filament of the plasma is also
conserved. The magnetic flux is equal to the product of the mag
netic field H and the cross-sectional area S. The conservation of
the magnetic flux in fast processes is a consequence of the law of
induction, according to which a change in the magnetic flux
through a conductor results in the appearance of an induced
electromotive force. The latter gives rise to the current which pro
duces a magnetic field tending to oppose the original change in the
magnetic flux. The electromotive force is proportional to the rate
of change of the flux while the current is, in addition, proportional

to the conductivity of the material. Therefore, in sufficiently fast
processes, that is, processes occurring over short intervals of time,
even a small change in the magnetic field in a highly conducting
plasma leads to a large instantaneous current, whose field com
pensates the increase or decrease in the flux. Therefore, in the case
of fast plasma motion, Equation (7.8) is augmented by the further
relation
Si Hi = S2H2, (7.9)
showing that the magnetic flux remains constant, Hence, it
follows that
//. = ^2
(7.10)
H2 S i
and comparison of Equations (7.8) and (7.10) yields
Hi = m
(7.11)
H2 U2
Therefore, the magnetic field strength is proportional to the con

centration, that is, the plasma density.
Moreover, it is clear that the ratio H /n remains constant for a
given volume element. This characteristic feature of the interaction
of plasma with a magnetic field is often referred to as the “freezing-
in” of magnetic lines of force. This term is consistent with Fara
day’s model of a magnetic field, in which the lines of force are
imagined to be real rather than conceptual. On this interpretation,
the magnetic flux is measured by the number of lines of force passing
through a given cross section of the plasma, and the fact that the
flux is conserved means that these lines move together with the plas
ma just as if they were frozen into it. This idea has been very con
venient and useful in the analysis of many phenomena involving
conducting plasma in a magnetic field. However, its usefulness
should not be overestimated. It is valid only if the deformation of
the plasma occurs rapidly, and only in the direction perpendicular
to H. Frozen-in magnetic lines of force are not peculiar to plasma.
A similar effect occurs in the interaction between the magnetic
field and any other good conductor.
F igure 64. Appearance o f a plasma electric dipole moment in crossed

electric and magnetic fields.
A magnetic field has a very considerable effect on the dielectric

properties of plasma. This may be understood with the aid of
Fig. 64, which shows a layer of plasma in a uniform magnetic field.
Suppose that an electric field parallel to the y axis is switched on
with the result that the electrons and ions execute a cycloidal mo
tion. The mean displacement along the y axis, from the initial
position of each particle, is equal to one-half of the height of the
cycloid. Displacement of the ions and electrons in opposite direc
tions results in the formation of an excess positive charge on one
side and excess negative charge on the other. The height of the
cycloid is proportional to the mass of the particle, and therefore in
calculating the amount of charge on the two surfaces, the displace
ment of the electrons may be ignored (in contrast to our earlier
calculations of the permittivity of plasma in a high-frequency
electric field with H = 0). The charges appearing on the bounda
ries of the plasma layer produce an additional electric field which is
opposite to the external field E. This means that, when the electric
and magnetic fields are perpendicular to each other, the plasma be-
haves as a dielectric with permittivity greater than unity. The

permittivity o.f plasma in crossed electric and magnetic fields is
given by
4Tpc2
€ m ag 1 + ~ lP ~ ’ (7.12)
where pis the density of the plasma; that is, p = nrm. Provided the
density is not too low, the permittivity €mag is practically always high
Thus, for example, in hydrogen plasma with n = 1013 particles per
cm3 and H = 10,000 Oe, the permittivity is found to be about 2,000.
The large magnitude of €raag gives rise to some very interesting prop
erties of the propagation of electromagnetic waves in magnetized
plasma (see Section 7.5).
Equation (7.12) is not universally valid. It holds only when the
electric field varies slowly. If the frequency of the field is com
parable with the Larmor frequency of the ions, Equation (7.12)
can no longer be used.
We shall use the above expression for emag to analyze a fairly
simple paradox concerned with the behavior of magnetized plasma
in a gravitational field. Consider a plasma concentration in the
form of a parallelepiped placed in a uniform magnetic field which is
at right-angles to the gravitational field. At first sight, it would
appear that the plasma will not fall since each charged particle will
execute a drift motion at right-angles to both the gravitational and
magnetic fields. However, it is evident from Fig. 64 that the drift
motion leads to the appearance of equal and opposite charges on
the surfaces of the condensation which, in turn, produce an electric
field. This field also gives rise to a drift in the direction of the
gravitational force. The drift of the particles under the action of the
force mg takes place with the velocity u = (c/eH )m g. If the plasma
concentration is n, the extra charge q formed on each square centi
meter of the lateral surface of the plasma in a time t will be neut
(only the drift of the ions is taken into account), so that
nmicgt
(7.13)
H
According to the laws of electrostatics, the electric field inside the
plasma due to these charges is equal to 4irq/€mag. If we now suppose
that «mae is much greater than unity, we need only take the second
term in Equation (7.12) and assume that €maR = A-wnmic2/ H 2. The
electric field in the plasma is then given by
4wq H
E =
and the drift velocity due to this field is

cE
(7.14)
v = H = *'
in agreement with the usual law of free fall.
7.4 The Pinch Effect

We shall now consider some specific examples of the interaction
of plasma with magnetic fields which are of interest for the various
possible applications of plasma physics. One such special case is
the “pinch effect” which is observed in high-current gas discharges.
The passage of current through a gas discharge is associated with
the appearance of electrodynamic forces acting on the ionized gas.
The electrodynamic force j X H /c acting on a unit volume is always
perpendicular to the direction of the current; that is, it tends to
compress the plasma (see Fig. 59). One would therefore expect that
in high-current electrical discharges the plasma will leave the walls
and form a very narrow filament along the axis of the discharge
tube. Let us suppose that there is equilibrium between the electro
dynamic forces which compress the plasma and the opposing
gaseous pressure in the filament. We should then have a relatively
simple relation between the temperature of the plasma and the
current in the filament. We shall derive this relation under the
following simplifying assumptions:
1. The current flows only over the surface of the filament of
plasma. This should occur in discharges of short duration if the
plasma exhibits high electrical conductivity, since the laws of elec
trodynamics predict that a high-frequency current does not pene
trate into a conductor but flows only within a thin surface layer.
This phenomenon is known as the “skin effect.”
Pinch Effect 149
2. The temperature of the electrons and ions is the same and re
mains constant over the cross section of the filament.
3. The plasma filament is in the form of a cylinder of circular
cross section.
It follows from the first assumption that the electrodynamic
forces act on the surface layer of the filament and, since they must
be balanced by the pressure, we have
where Hi is the magnetic field due to the current at the outer surface
of the filament. Since Te = 7), the pressure in the plasma is 2nkT.
The magnetic field strength due to the current on the surface of the
filament is directly proportional to the current / and inversely pro
portional to the radius of the filament ro. As is shown in textbooks
on electricity and magnetism,
H, = — , (7.16)
cr0
where c is the velocity of light and / is measured in electrostatic
units. Substituting these expressions for p and H into (7.15), we
have after some rearrangement
/ 2 = \c \r \n k T . (7.17)
The quantity ■nr^i is the number of particles of given sign per unit
length of the plasma filament. We shall represent it by N. Equa
tion (7.17) may then be rewritten in the form
p = 4c2NkT. (7.18)
If we now substitute the numerical value for k and express I in
amperes (1 ampere = 3 X 109 esu), Equation (7.18) becomes
I 2 = 5.5-10-14AT, (7.19)
which is more convenient for practical calculations.
The relation between / and T was derived above for a current
flowing on the surface of the plasma filament. A more detailed
theoretical analysis shows that this relationship remains valid for
any distribution of current over the cross section of the plasma
filament, provided only that the temperature is the same through

out.
Let us consider a concrete example which approaches the condi
tions obtaining in a real physical experiment concerned with the
pinch effect. Suppose that the discharge tube is filled with hydrogen
at an initial pressure of 0.1 mm Hg. The radius of the tube will be
assumed to be equal to 10 cm and the discharge current will be
taken as 5 X 105 amp. Under these conditions N = irr\n = 2 X
1018. The temperature of the plasma, as given by (7.19), will then
be equal to two million degrees. This numerical example is interest
ing because it shows that it is possible, at least in principle, to
achieve ultrahigh temperatures under laboratory conditions since it
is quite easy to produce with modern apparatus a short current
pulse of the order of 5 X 105 amp or even greater. The results of
experimental studies of such high-current discharges will be dealt
with in a later section.
The pinch effect, that is, the lateral compression of plasma by
current passing through it, can also be achieved inductively by
using a doughnut-shaped discharge tube and winding a coil on it.
When the induced current is great enough, a circular filament of
plasma is produced as before. However, as the cross-sectional area
of the filament is reduced as a result of the pinch effect, there is also
an increase in the radius of the current ring. The expansion of the
F igure 65. Expansion o f the current flowing in a ring.

Pinch Effect 151
ring is due to the fact that each element of length of the ring ex
periences a force due to the magnetic fields of all the other elements.
For example, the segment ab experiences the field due to the seg
ment cd. Since the currents are equal and opposite, the two ele
ments repel as indicated in Fig. 65. The current ring will therefore
tend to expand, and its radius R will increase.
This property of the current ring is an example of a general
result in electrodynamics: Namely, a conductor which carries a
current always tends to increase its self-inductance. For a linear
conductor, the increase in the self-inductance occurs when its length
increases and the cross-sectional area decreases.
If there were no forces preventing the expansion of the ring, the
F igure 66. “Image currents'' produced in the conducting wall

o f a discharge chamber.
plasma filament would eventually reach the outer wall of the

chamber. However, under certain conditions, the forces which
tend to expand the filament may be balanced. This occurs when the
chamber is constructed from a metal with a high electrical con
ductivity. As the plasma ring expands, the magnetic lines of force
due to it cut the walls of the chamber as shown in Fig. 66. This
gives rise to an induced electromotive force which produces a cur
rent in the wall. The direction of this current is such as to prevent
the penetration of the magnetic field into the wall. It follows that
the direction of this current must be opposite to the direction of the
current in the plasma, and therefore the two currents must repel
each other. The force of repulsion between the current in the
plasma and the induced current in the wall increases rapidly as the
plasma filament approaches the wall. Calculations show that a
small displacement of the plasma ring is sufficient to ensure that the
induced current in the wall will balance the force tending to extend
the ring, and stop the motion of the ring.
In fact, the situation is somewhat more complicated since the
current induced in the wall is only maintained as long as the mag
netic lines of force cut the wall, and therefore as soon as the filament
stops moving, the currents in the wall disappear altogether. In
reality, after equilibrium has been reached, the plasma filament
continues to expand but the expansion takes place very slowly.
Therefore, in discharges of relatively short duration (much smaller
than 1 sec), the balancing effect of the metal wall is sufficient to
prevent the approach of the plasma ring toward the wall.
In addition to steady-state processes, in which electrodynamic
forces are balanced by the pressure in the plasma, there are also
dynamic effects which give rise to acceleration of the plasma. The
most direct application of this effect is the plasma injector. A
plasma injector is an apparatus which injects into a vacuum either
a continuous jet of high-speed plasma, or individual plasma con
densations, that is, plasmoids. Figure 67 shows a simple pulsed
plasma injector intended for the generation of fast plasmoids. The
injector consists of two coaxial metal cylinders. The cylinders are
fed by a capacitor bank which serves as a reservoir of electrical
energy. A definite amount of gas is let into the space between the
Oscillations and Waves in Plasma 153
G as leak Plasm a
4E J
t/=E
V / / '/ 7 7 V / / T
/ / 7
F igure 67. Basic arrangement o f a coaxial plasma injector.
cylinders through a fast-acting inlet valve at a predetermined in

stant of time. Before the gas succeeds in expanding into the avail
able space, the high voltage from the capacitor bank is applied to
the electrodes and plasma is produced as a result of the electrical
discharge. A current flows in the radial direction, and the inter
action of the current with the magnetic field due to it gives rise to an
electrodynamic pressure which accelerates the plasma along the
injector. As a rough approximation, we can use Equation (7.4) and
neglect the gas-kinetic pressure, that is, the term containing grad p.
The larger the current passing through the plasma and the smaller
the mass of the gas, the higher is the velocity reached by the plas-
moid at the exit of the injector.
Devices of this type may be used under laboratory conditions to
produce plasmoids accelerated to velocities of the order of a few
hundred km per sec and containing 1018 to 1019 particles. Another
example of a situation in which the interaction of plasma with a
magnetic field is dynamic in character, is the deceleration of
ionized gas in a magnetohydrodynamic converter. This new type of
generator of electrical energy will be discussed in Section 8.2.
7.5 Oscillations and Waves in Plasma
Various forms of oscillations and waves may appear and propa

gate in plasma. They manifest themselves as a periodic variation in
the charged-particle concentration and the electric or magnetic
field strengths. One of the simplest processes of this kind, peculiar
to plasma, is the Langmuir oscillations, illustrated in Fig. 68.
Si S%
1----- 11
©
© © 00 ®
© ©
© ® © ©F ©
A <
0 © © ©
0 © ©
©
0 © ©
© © ___
©
__Il
u
x
F ig u r e 68. Drawing illustrating the origin o f Langmuir

oscillations in plasma.
Suppose that, in the volume A, the electrons have been displaced

through a small distance £ in the direction of the x axis. The reason
for this displacement is unimportant for our purposes and may in
fact be quite accidental. As a result of the displacement of the elec
trons to the right, electrical charges will appear on the surfaces S\
and S 2 which bound the region A ; these charges produce an electric
field which tends to return the electrons to their original position.
The resulting force may be compared with the force acting on a
pendulum and tending to return it to its position of equilibrium.
In both cases, the restoring force is proportional to the deflection
and gives rise to periodic oscillations. The dependence of the de
flection on time can then be represented by a sine curve, and
the frequency of the oscillations, that is, their number per second,
is given by
* ( 7 ' 2 0 )
where / is the elastic constant, that is, the ratio of the force to the
deflection, and m is the mass of the body executing the oscillations.
This formula is valid for any process in which the restoring force is
proportional to the displacement. In particular, in the case of

electron oscillations in plasma,15 the force is given by
F = eE = 4irne 2£. (7.21)
Therefore
/ = 4wne2
and
J'o = — J—
2lT = \ 7Wle = 9 -!03V^.
(7.22)
We recall that we have already encountered this quantity in a some

what different context. According to Equation (5.11), it represents
the critical frequency of electromagnetic waves, that is, the mini
mum value of the frequency at which electromagnetic waves can
penetrate into plasma.
The oscillations of plasma ions are somewhat more complicated.
Owing to the much greater mass of the ions, these oscillations are
slower and therefore the electrons, which follow the ions but have a
much greater mobility, compensate almost completely the electric
fields produced as a result of such oscillations. We shall not con
sider in detail the mechanism of ion oscillations and will merely
note that their propagation through plasma is analogous to the
propagation of sound through a neutral gas. If there is thermal
equilibrium between the electrons and ions in the plasma, that is, if
Te = Ti, this “ion sound” propagates with a velocity of the order of
the thermal velocity of the ions. When T e is much greater than T i ,
the velocity of propagation of the sound is found to be of the order
of -\JkTe/mt; that is, the sound wave travels with the velocity which
the ions would have if their temperature were equal to the electron
temperature. Theoretical analysis shows that plasma sound waves
can propagate freely in the latter case, that is, when the ions are
much colder than the electrons. If, on the other hand, the velocity
of sound is of the order of the thermal velocity of the ions, which
should occur when 7) = T e , the sound waves will be rapidly at-
15The electric field inside the region A is equal to 4 irq, where q is the charge per
unit area on Si or Si. Electrons displaced through a distance £ produce an extra
charge o f q = ne£.
156 Plamsa in a Magnetic Field
F igure 69. Electric and magnetic fields in a

transverse electromagnetic wave.
tenuated in the plasma; that is, their energy will be absorbed and
converted into heat. Under these conditions, the sound can only
propagate through plasma over very short distances.
The Langmuir electron oscillations and ion sound are forms of
periodic motion of matter in the longitudinal direction. This
motion gives rise to longitudinal electric fields. However, ordinary
electromagnetic waves can also propagate through plasma. Let us
recall some of the main properties of such waves: In their simplest
form, electromagnetic waves are a periodic process in which the
electric and magnetic fields are strictly related to each other. In
vacuum, the velocity of propagation of electromagnetic waves is
always the same and equal to c. Both the electric and magnetic
fields in the wave are perpendicular to the direction of propagation;
that is, the waves are transverse oscillations of the two components
of the electromagnetic field as illustrated in Fig. 69. In a material
medium, the velocity of propagation of electromagnetic waves is
c/'yjt, where e is the permittivity. In the absence of a magnetic field
the permittivity of plasma is less than unity, and therefore the
velocity of propagation of electromagnetic waves in plasma is
greater than the velocity of light in a vacuum. Electromagnetic
waves cannot penetrate into plasma and propagate in it if their
frequency is less than p0, since the permittivity is then negative
and the velocity of the wave complex.
The propagation of electromagnetic waves in magnetized plasma
is of particular interest. The phenomenon is very complicated and
a detailed analysis would be neither easy nor useful within the

framework of this book. We shall, therefore, confine our attention
to a general description of a few important special cases. To begin
with, it must be noted that the magnetic field should have no ap
preciable effect on the passage of an electromagnetic field through
plasma if the frequency of the waves is much greater than the
Larmor frequency. The oscillations of the electrons are then so fast
that the magnetic field has no chance to affect them, and therefore
the permittivity of plasma is the same as when H = 0. The op
posite case occurs when the frequency is very low. If it is much
lower than the Larmor frequency of the plasma ions, the propaga
tion of electromagnetic waves through plasma exhibits some very
special features. The simplest result is obtained when the direction
of propagation is parallel to the direction of the constant magnetic
field Ho. The alternating electric and magnetic fields in the wave
are then perpendicular to Ho (Fig. 70) and the field strength at each
point in space is the vector sum of Ho and the alternating field in the
wave. As a result, the lines of force assume a wave-like form, and
since they are frozen into the medium, their deformation brings the
plasma into motion, so that it takes part in the oscillation. A large
proportion of the electromagnetic energy associated with the wave
is thus converted into the kinetic energy of oscillation of the
medium. Therefore, the velocity of propagation of electromagnetic
waves in a dense plasma should be much smaller than the velocity
of electromagnetic waves in vacuum. The situation is somewhat
similar to the propagation of elastic waves along a wire. If the
density, that is, the mass per unit length, is higher at some point
along the wire than elsewhere, the velocity of the wave is reduced
F igure 7 0 . Distortion o f a magnetic line o f force (the curved line) due to

the propagation o f an electromagnetic wave, carrying an alternating magnetic
field H_, in the direction o f a constant magnetic field Ho.
v?
F igure 71. Propagation o f a sound wave in magnetized plasma. The wave
is associated with the bending o f the lines o f force.
accordingly. In the propagation of electromagnetic waves in the

direction of a constant magnetic field in plasma, the lines of force
act as elastic threads along which oscillations are transmitted.
These threads are “loaded” by the plasma, and therefore the veloc
ity of the wave is reduced.
The change in the velocity of electromagnetic waves in plasma
can also be predicted in a more formal way by considering the per
mittivity of magnetized plasma (Section 7.2). In a dense plasma lo
cated in a magnetic field, the permittivity is equal to 4xpc2/ / / 2 (see
Equation (7.12)), and therefore the velocity of propagation in the
direction of the magnetic field is given by
= <7-23>
Such transverse electromagnetic oscillations, which propagate
along the lines of force as if these were elastic threads, are called
Alfven waves, after the Swedish astrophysicist who predicted them.
The properties of sound waves are also modified in the case of
magnetized plasma. In a sound wave, the oscillations of the medi
um occur in the direction of propagation, and therefore a magnetic
field frozen into the plasma may have an important effect on the
propagation of the waves when they travel at right-angles to the

lines of force. This is different from the case of electromagnetic
waves which are, of course, transverse in character. The propaga
tion of a sound wave through magnetized plasma is illustrated
schematically in Fig. 71. It is connected with the periodically
varying bending of the lines of force. If the temperature of the
plasma is not too high, the velocity of propagation of such waves is
also given by (7.23); that is, it is equal to the velocity of Alfven
waves. A sound wave propagating at right-angles to the magnetic
lines of force in plasma can be excited, for example, by applying a
rapid periodic variation to the magnetic field near the boundary of
the plasma.
In addition to the simple special cases which we have discussed
above, there are also other forms of oscillatory and wave processes
which can develop in plasma. Further discussion of these problems
is beyond the scope of this book.
8
Technological Applications
8.1 Controlled Thermonuclear Reactions

The current trend in plasma research has been to concentrate on
those topics which are likely to lead to useful practical applications.
The most important of these is the utilization of the enormous
quantities of energy which are stored in the nuclei of the very
abundant light elements. Although the problem is still largely
unsolved, its importance is so great, and it is so intimately con
nected with the development of plasma physics, that the tech
nological applications of plasma are virtually inseparable from it.
In contrast to the nuclei of such heavy elements as uranium or
thorium, whose energy can be released in the fission process, that is,
division into lighter fragments under neutron bombardment, the
nuclei of light elements can give up their energy only in synthetic
processes, in which two colliding nuclei fuse and form a heavier
nucleus. One of the simplest examples of such a synthetic reaction
is the fusion of deuterium nuclei which results in the formation of
helium or tritium.16 This nuclear reaction may be written in the
form
H2 + H2
cHe2 + «'
H 3 + / / ',
( 8. 1)
when nl is a neutron, H l, H 2, H 3 represent the three isotopes of hy

drogen, and He3 represents the helium isotope of atomic weight 3.
16W e recall that deuterium is the hydrogen iso to p e with atom ic w eight 2, w hile
tritium is the hydrogen isotop e with atom ic w eight 3.
160
Controlled Thermonuclear Reactions 161
The two arrows show that a collision between two deuterons may
result in the -appearance of either a helium nucleus and a neutron,
or a tritium nucleus and a proton. In the first case, the energy re
leased in the fusion reaction is about 3.3 MeV (MeV = million
electron-volts), while in the second case, the energy is about 4 MeV.
When the masses involved in the reactions are the same, the energy
released in the fusion process is found to be approximately the same
as the fission energy released in uranium reactors. However, the con
ditions which must be obtained before the reactions can take place
are quite different in the two cases. Fission can occur in a medium
whose atoms are at rest, since atomic nuclei play a passive role and
are simply targets for the bombarding neutrons. In contrast, the
fusion reaction can only occur in a medium whose nuclei are in very
rapid motion. The basic fusion reaction can only occur when the
two interacting particles approach each other to within a distance of
the order of 10“ 13 cm. This can only happen if they overcome their
mutual electrostatic repulsion, that is, if they have a very high
relative velocity. This in its turn means that the temperature of the
medium must be very high. It follows that the necessary condition
for the fusion reaction to proceed at a high rate is that the medium
must be very hot. Hence the term thermonuclear reaction.
Calculations show that fusion reactions of the type described by
Equation (8.1) occur at an appreciable rate only above temperatures
of the order of a few million degrees but in order that the liberated
energy be of practical interest, the temperature of the deuterium
must be increased to a few hundred million degrees. At such
temperatures, the deuterium can, of course, no longer exist as a
neutral medium and is converted into highly ionized plasma con
sisting of fast deuterons and electrons. It is evident that the main
difficulty lies in the insulation of the high-temperature plasma from
the walls of the chamber in which it is contained. Since plasma has
an enormous thermal conductivity, this insulation is essential, be
cause otherwise all the energy would immediately escape to the
walls and the required high temperature would not be reached.
This means that the plasma must, in effect, be contained in a high
vacuum, and this can only be achieved with the aid of a magnetic
field whose lines of force surround the plasma. In point of fact, the
162 Technological Applications
idea of magnetic thermal insulation of plasma is contained in

Equation (7.2).
In the Soviet Union, this idea was first used in connection with
controlled thermonuclear fusion by A. D. Sakharov and I. Ye.
Tamm in 1950. Similar work was performed independently and at
the same time in Great Britain and the U.S.A. However, most of
the work was classified at the time, and therefore there was no ex
change of information.
The idea of magnetic thermal insulation led to a very rapid de
velopment of experimental studies in the Soviet Union. Initially,
the experimental work was based on use of the pinch effect. As has
already been pointed out in the preceding chapter, the passage of a
high current through plasma gives rise to a compression of the
medium by the resulting electrodynamic force. This removes the
plasma from the vicinity of the walls of the discharge tube and
confines it to the central region. The electric current performs
three functions: (1) in the initial stage it produces the plasma by
ionization; (2) the electrodynamic forces due to it maintain the
plasma ih a compressed state, and (3) it heats the plasma to a very
high temperature as a result of the liberation of Joule heat. The
simple example illustrated by Equation (7.3) shows that, with rela
tively modest and easily realizable initial conditions, it is quite
possible to heat deuterium plasma to a temperature of many
million degrees, and hence produce thermonuclear reactions at an
appreciable rate. However, it was soon shown experimentally that
the rather optimistic original estimates were not entirely justified.
It has been found that the plasma filament, produced as a result of
the pinch effect, is very unstable. The geometry of the filament is
radically altered within a few millionths of a second as a result of
this instability. The result is that the heated plasma is splashed onto
the walls of the discharge tube, and the very high temperature
reached during the initial stages of the process, as the plasma is
pinched by the electrodynamic forces, exists only for an exceedingly
short interval of time.
In the early experiments performed in the Soviet Union, the
maximum current in the plasma during the discharge pulse was I05to
IO6 amperes and the rise time of the current from zero to its maxi-
mum value was 5 to 10 microseconds. The initial pressure of the

deuterium, which was the main substance used in these early ex
periments, was usually between a few hundredths and a few tenths
of a millimeter of mercury. Under these initial conditions, the
process developed as follows. At first, the plasma was found to
pinch very rapidly. In the final stage of this process, the tempera
ture of the plasma filament was of the order of a million degrees, or
even a few million degrees. As soon as the pinch process is over, the
plasma cord executes very rapid radial oscillations: The ring alter
nately expands and contracts. During this stage, the instability of
the filament sets in and gives rise to an interaction of the plasma
with the walls, leading to the contamination of the deuterium by
impurities and to rapid cooling. All this occurs within a few
microseconds. When the current reaches its maximum value, the
temperature of the plasma is already much lower than that during
the initial stages of the pinch effect. Successive stages in the de
velopment of the plasma filament are illustrated in Fig. 72. Each
of the photographs corresponds to a 0.5 microsecond exposure and
the separation between the photographs is roughly of the same
order.
The thermonuclear reaction yield produced under these condi
tions is, of course, rather modest. The observed short burst of
neutrons, which is of the order of a few tenths of a microsecond, is
in fact due to the acceleration of a small group of very fast deuter-
ons by the electric fields developed in the plasma as a result of its
strong instability, rather than the thermonuclear reactions. Al
though this phenomenon is interesting in itself, and was certainly
unexpected, it does not resolve the main practical problem, since it
involves only a small proportion of the available number of
particles and gives rise to a very brief train of nuclear reactions.
The results of these experiments soon led to the following crucial
question: Under what conditions will magnetically supported hot
plasma remain stable? Clearly, the answer to this question will
govern the successful utilization of controlled thermonuclear re
action. In spite of a very large number of theoretical calculations,
the question is still an open one. We are not yet absolutely certain
that there are states in which a dense hot plasma, completely
F igure 7 2 . Photograph o f seven suc

cessive stages in the development o f a
4
plasma filament in a pinching dis
charge. The frames are separated by
5 X 1 0 ~ 7 seconds
7
separated from the walls and supported by magnetic forces in a

vacuum, will remain in equilibrium for a sufficiently long interval
of time. The phrase “sufficiently long” is used in the sense that
during the existence of the hot plasma, each deuteron should have
an appreciable chance of entering into a nuclear reaction. At
present, the development of methods for producing and maintain
ing high-temperature plasma is intimately connected with investiga
tions of the stability of various plasma configurations. We must
therefore consider the problem of stability before we proceed to
the more general problem of thermonuclear fusion.
Let us return, therefore, to the consideration of a plasma filament
through which a current is flowing. In order to exhibit the main
physical features of the phenomenon, we shall simplify the problem
by assuming that the plasma is perfectly conducting. As a result of
the skin effect, the current will then be confined to a thin surface
layer. Figs. 73 and 74 show two simple kinds of deformation of a
cylindrical column of plasma which are produced by random fluctu
ations. In the first case, there is a local reduction in the radius of the
cylinder. Since for a given current the magnetic field at the surface
of the plasma is inversely proportional to the radius of the cylinder,
it follows that the field in the plane ab must be greater than the field
at the undeformed surface. Therefore, the magnetic pressure H 2/ 8x
a b
F igure 73. Local contraction o f a plasma filament.

which compresses the plasma in the deformed region, must also be

greater in this region. Since, however, the pressure p in the plasma
must be the same throughout (since the plasma can flow freely
along the cylinder), it follows that in the region of contraction, the
equilibrium between electrodynamic and plasma pressures is upset
and a small initial deformation will increase rapidly. This means
that the plasma cylinder is unstable with respect to local changes in
its diameter.
In the case of deformations of the form illustrated in Fig. 74, the
magnetic field strength is different on opposite sides of the bent
segment. On the outer (convex) side, it is reduced, while on the
inner (concave), it is increased. There is, therefore, a difference in
the magnetic pressures in the direction ab and this leads to a rapid
increase in the initial deformation. It follows that the plasma
cylinder is also unstable with respect to bending. This form of in
stability is particularly dangerous from the point of view of thermal
insulation of plasma since it eventually leads to contact between
the plasma and the wall. These two forms of instability are not
exclusive to plasma — they are equally characteristic of any linear
conductor carrying a current, or any medium which is not rigid,
that is, which does not oppose changes in its form.
Having discovered the possible instabilities of a plasma con
ductor, we must now consider how we can stabilize it. This can be
done through the aid of additional external magnetic fields which
are independent of the field due to the plasma current itself. Sup
pose, for example, that a coil has been wound on the outer surface
of the discharge tube and is used to produce a strong magnetic field
whose lines of force are parallel to the current in the plasma
F igure 74. Bending o f a plasma filament.

4H
F igure 75. Compression o f the magnetic lines o f force o f a longitudinal

magnetic field during a local contraction o f a
current-carrying column o f plasma.
(Fig. 56). As explained in the preceding chapter, properties char

acteristic of elastic rubber bands may be ascribed to the magnetic
lines of force. They are compressed in the transverse direction and
stretched in the longitudinal direction, and therefore tend to con
tract longitudinally and expand laterally. These two properties
may be used as a stabilizing factor opposing the deformation of a
cylinder of plasma. A local contraction of the form illustrated in
Fig. 75 gives rise to a compression of the lines of force within the
plasma conductor, and therefore the magnetic pressure in the
plasma increases. In contrast, the density of the lines of force out
side the cylinder is reduced, and therefore the magnetic pressure due
to the longitudinal field is also reduced. This gives rise to a mag
netic pressure gradient which leads to the return of the plasma con
ductor to its original form (if the longitudinal field is high enough).
In the case of bending (Fig. 76), the lines of force associated with
the longitudinal field are stretched and this leads to the appearance
of a force F tending to reduce the bending. A conducting metal
envelope can also be used as a means of enhancing the longitudinal
field. As has already been pointed out earlier, when the plasma
F igure 76. Stretching o f magnetic lines o f force during the bending o f a

current-carrying column o f plasma.
conductor approaches the metal wall, it induces a current within it,

and the two currents repel. This phenomenon reduces the tendency
of the plasma filament to depart from its axial position in the dis
charge chamber, when the chamber is made out of metal or is sur
rounded by a metal envelope.
These methods of reducing the instabilities are being used at the
present time in installations designed for the study of high-tempera-
ture plasma. An installation of this kind is illustrated in Fig. 77.
F igure 77. Toroidal discharge chamber with a longitudinal magnetic field:

1 — inner chamber or liner, 2 — outer copper chamber, 3 — coil producing
the magnetic field 4 — primary o f transformer (the plasma filament acts
as the secondary), 5 — iron core.
The discharge chamber is toroidal in form and surrounds the iron

core of a transformer. The current in the gas is excited inductively
and the circular plasma filament, which is produced during the dis
charge, acts as the secondary of the transformer. The plasma fila
ment is stabilized by means of the longitudinal magnetic field due to
the coil wound on the outer surface of the chamber.
The chamber itself is usually of the two-layer kind. The plasma
is produced inside a thin, corrugated stainless-steel shell, 0.2 to
0.3 mm thick. This thin inner chamber is known as the liner and is
inserted into a thick copper envelope. The space between the liner
and the outer envelope is evacuated to a very low pressure. This
design ensures that the experiments can be carried out under very
clean vacuum conditions since the liner can be baked, by passing a
current through it, and thus thoroughly degassed. Experimental
installations of this kind have been developed in the Soviet Union
and are collectively referred to as “Tokamak” installations. The
biggest of these is designed to produce currents Up to 2.5-3 X 105
amperes with a stabilizing longitudinal field of up to 4,000 oersteds.
In the Tokamak installations, the plasma filament exists for
0.005-0.02 seconds. So far, maximum temperatures of the order of
two million degrees have been achieved.
It is still difficult to prophesy the final maximum temperatures
which it will be possible to produce by heating the plasma with a
current. The stability of the plasma filament is also an unresolved
problem. Although all the measurements indicate that the Toka
mak installations produce relatively stable rings of plasma, if the
longitudinal magnetic field is high enough, absolute stability has not
been achieved. There is also a further form of instability which
leads to the gradual spread of the plasma filament as a result of a
process which may be referred to as anomalous diffusion. The latter
is due to the fact that plasma is capable of diffusing at right-angles
to the lines of force much more rapidly than predicted by the
classical laws of diffusion in a magnetic field.
Recent theoretical studies have unearthed a number of processes
which may be responsible for anomalous diffusion. They all in
volve the same physical mechanism, that is, the excitation of various
forms of plasma oscillations due to the transfer of energy from di-
rected beams of plasma particles. The directed motion of one of the

components of plasma relative to another, that is, the motion of
electrons relative to the ions, is nothing else than an electric current.
We have seen in the preceding chapter that such currents can exist
in plasma in a magnetic field (even in the absence of external
electromotive forces). The excitation of oscillations in plasma by a
directed beam of particles is, in a sense, analogous to the excitation
of sea waves by the wind, except for the important distinction that
the wind produces surface waves, while the electrical wind in the
plasma gives rise to three-dimensional waves.
It should be noted that a directed beam of particles passing
through plasma can only excite plasma waves which propagate
with a velocity of the order of the velocity of the beam itself. The
oscillations and waves produced in plasma give rise to electric
fields which may have components at right-angles to H. This, in
its turn, is responsible for a drift motion at right-angles to the mag
netic lines of force, that is, the anomalous diffusion mentioned
above. It is still not clear whether the anomalously high rate of
diffusion is an unavoidable feature of the circular rings of plasma
which are produced in installations of the kind we have discussed
so far. It may be that when the longitudinal magnetic field is very
much greater than the field due to the current in the plasma, all the
minor instabilities, including anomalous diffusion, will be sup
pressed. If this optimistic assumption turns out to be right, it
might be possible to heat the plasma to temperatures of the order of
ten million degrees. A plasma ring with a concentration of the
order of 10'3 to 1014 particles per cm3 would be a relatively powerful
source of thermonuclear reactions.
The pinch effect is not the only way of insulating a high-tempera
ture plasma. The plasma may be prevented from coming into con
tact with the walls of the chamber by means of an externally pro
duced magnetic field. Systems of this kind, in which the thermal
insulation is achieved with the aid of external magnetic fields, are
known as magnetic traps. Simple forms of magnetic traps are il
lustrated in Figs. 15 and 78. The first of these two figures shows a
system with two magnetic mirrors and has been described in
Chapter 2. In Fig. 15, the magnetic field increases in the axial
F igure 78. Magnetic trap with opposing fields (the coils producing the
field are represented by the rectangles).
direction on either side of the central region. Plasma particles

travelling at a large angle to the lines of force will be trapped and
will not succeed in leaving the system. A plasma captured in a trap
of this kind will occupy the spindle-shaped central region.
We note that, even if all the stability conditions are satisfied, the
plasma will be retained by the trap for a long interval of time only if
its temperature is very high and its density very low. A single
collision is sufficient to ensure that the direction of a particle is
changed, so that it may succeed in escaping from the trap. There
fore, it is necessary that the collision frequency be small, and this
can only occur at high temperatures or very low concentrations.
The second magnetic system (Fig. 78) differs from the first in that
the magnetic fields due to the two coils are in opposite directions.
As a result, the lines of force in the space between the two coils are
deflected away from the axis, rather like the bristles of two shaving
brushes pressed against each other. The field strength at the central
point 0 is zero. In the neighborhood of this point, the magnitude
of //increases, both in the axial and radial directions, in proportion
to the distance from 0. In this system, the plasma occupies a
volume which has the form of a spinning top. This magnetic trap
will retain particles for which the angle between the direction of
motion and the lines of force is not too small.
Since the external magnetic field in these magnetic traps serves
only as the thermal insulator, the problem of heating the plasma is
quite independent of the trapping problem. It can be resolved in
various ways; for example, it is possible to inject into the trap
plasmoids, accelerated to high velocities with the aid of electro
dynamic injectors of the kind described in the preceding chapter.
However, the most important problem is not the development of
methods for filling up magnetic traps with plasma, but the elucida
tion of the conditions under which stable retention of hot plasma in
the trap is possible.
The first of the above mentioned two traps presents the most
serious difficulties as far as stability is concerned. This system has
one obvious disadvantage: The magnetic field decreases in the
radial direction, that is, at right-angles to the axis. Since the plasma
exhibits diamagnetic properties, and therefore tends to move in the
direction of decreasing magnetic field, it follows that if a small
pimple or hump appears on the surface of the plasma, it will tend to
grow in the radial direction. As a result, random surface deforma
tion will lead to the appearance of streaks of plasma whose length
will continuously increase. This will eventually result in the expan
sion of the plasma, bringing it into contact with the walls.
A similar conclusion can be reached by considering the variation
of the ratio of plasma pressure to the magnetic pressure during a
deformation of the surface of the plasma. Let us suppose, for the
sake of simplicity, that the magnetic field in the plasma is zero.
Under these conditions, we have p = H 2/ 8 t on the boundary be
tween the plasma and the field. If a small section of the surface is
deformed so that it enters a region of lower field strength, the
equilibrium of the forces is upset, and the pressure in the plasma
will no longer be balanced by the electrodynamic force. Therefore,
the deformation will tend to increase. These qualitative conclusions
are supported by detailed theoretical calculations and are in agree
ment with the experimental results. In spite of the large number of
attempts at achieving stable retention of plasma within mirror traps,
and a number of premature reports that this has been achieved, in

reality nobody has succeeded in maintaining pure hydrogen plasma
at high ion temperature in a magnetic trap for more than a few
tenths of a microsecond.
Traps of the kind illustrated in Fig. 78 do not suffer from the
above disadvantage, since they are characterized by an increase in
the magnetic field strength in all directions away from the central
region which is filled with plasma. Therefore, as far as surface de
formations are concerned, this configuration should be stable.
However, theory shows that systems of this kind suffer from an
other kind of defect. It is found that the plasma flows out through
the sharp edge of the region occupied by it, that is, through the
boundary of the plane of symmetry of the plasma confined in the
trap. This difficulty is directly related to the fact that the trap con
tains a region in which the field strength passes through zero and
therefore W ± /H does not remain constant for particles entering
this region. In the neighborhood of the region where the field is
zero, there is no unique relation between the direction of motion of
the particles and the direction of the lines of force. It follows that
the condition for the retention of the particles in the trap, which
requires that the angle between the velocity and the magnetic field
must be large, breaks down. A particle passing through this region
has an appreciable probability of changing its direction of motion
relative to the lines of force and leaving the plasma at a small angle
to the direction of H (if it moves near the plane of symmetry of the
magnetic system). This theoretical conclusion has not yet been
confirmed experimentally but there is little doubt that it is correct.
It would appear therefore that the most suitable configurations
ensuring efficient thermal insulation of high-temperature plasma are
traps with composite magnetic fields, combining the properties of
the systems illustrated in Figs. 15 and 78 but free of their basic
defects. In such magnetic systems the magnetic field should not go
to zero anywhere in the region occupied by the plasma, whilst
the magnitude of H should increase in the outward direction. A
magnetic system satisfying these requirements is illustrated in Fig.
79. The field is produced by two coils with six linear, current-
carrying conductors disposed symmetrically relative to the line
F igure 79. Magnetic trap with composite fields: 1 — additional current-

carrying conductors, 2 — main coils.
joining the centers of the two coils. In the absence of the con
ductors, the system would form a simple magnetic trap of the form
previously illustrated in Fig. 15. However, the additional con
ductors carry currents which flow in directions normal in the plane
perpendicular to the axis and produce a field whose magnitude in
creases with the distance r from the axis. In particular, for a system
consisting of six such conductors, the field H near the axis increases
in proportion to r2. This magnetic trap was used in the first success
ful experiment in which high-temperature plasma was maintained
in a stable configuration for time intervals of the order of a tenth of
a second.
Although an interval of time of the order of one-tenth of a second
seems rather short in comparison to the time intervals encountered
in every-day life, it is nevertheless very long in comparison to the
time intervals which characterize the development of micro-
processes in plasma. For example, a fast electron in hot plasma
may succeed in traversing the plasma several hundred times in one-
tenth of a second, and the corresponding path length in the plasma
will exceed 100 km. This means that one-tenth of a second repre
sents a satisfactory retension of the particles in the plasma. This
result was achieved by M. S. Ioffe in the Department of Plasma
Studies of the Institute of Atomic Energy of the USSR, and is the
major achievement in the development of controlled thermonuclear
reactions in recent years.
M agnetohydrodynamic Conversion o f Energy 175
It is, of course, still too early to say that the way is now open to
the successful utilization of thermonuclear generators of energy.
So far, it has only been possible to retain the plasma particles in a
magnetic field at relatively low concentrations and it is still not
known what will be the behavior of dense plasma under these
conditions. It must also be remembered that there are still no effec
tive methods of producing high-temperature plasma of high con
centration. However, one gains the impression that one of the main
barriers to the successful development of thermonuclear generators
has been removed and that technological developments will finally
lead to a successful solution of this problem. Thermonuclear fusion
will then become an important new source of energy available for
the requirements of mankind.
8.2 Magnetohydrodynamic Conversion of Energy

Let us now briefly consider other possible applications of plasma
processes in the near future. We shall consider only those which
are promising enough to justify extensive research. The first of
these applications is concerned with direct transformation of ther
mal energy into electrical energy. The basic idea is quite simple. A
jet of ionized gas, in which a large proportion of the initial store of
energy is transformed into the kinetic energy of directed motion, is
injected into a magnetic field at right-angles to the lines of force
(Fig. 80). When the plasma jet intersects the lines of force, an
F ig u re 80. B a s ic a rra n g e m e n t o f a m a g n e to h y d r o d y n a m ic c o n v e r te r
{th e n o z z le is on th e le ft).
electromotive force is induced in the plasma. The magnitude of

the induced electric field E is equal to vH/c, where v is the velocity
of the plasma jet. The induced electric field gives rise to an electric
current which flows at right-angles to both v and H and is allowed
to pass through the external load R i. The load is connected across
the electrodes A\ and Ai, which are in contact with the plasma jet.
An installation based on this idea is known as a magnetohydro-
dynamic generator. The electrical energy liberated in the external
circuit of the generator is taken at the expense of the kinetic energy
of the plasma jet, so that the jet is decelerated as a result of the
interaction between the current flowing in the plasma and the
magnetic field. The force per unit volume of the moving plasma is
equal to jH /c , and therefore the work done in displacing the unit
volume through a distance / is equal to jH l/c. It follows that the
change in the kinetic energy per unit volume must also be equal
to this quantity. The work done by the electrodynamic force is, in
fact, the source of energy for the entire electric circuit. Let the re
sistance in the external circuit be R i and the internal resistance of
the region between the electrodes be Ri. The current flowing
through the magnetodydrodynamic generator is then given by
Ohm’s law
V
/ = ( 8. 2)
Ri + Ri
If all the quantities in this formula are expressed in practical units,
then U = 10-8 vHa, where a is the gap between the electrodes, that
is, the thickness of the plasma jet, and v is the mean velocity of the
jet. (We are assuming that the velocity of the plasma is only slightly
reduced on passing through the gap defined by the electrodes.)
The power dissipated in the external circuit is
U2R i
W = P-Rx (8.3)
(Ri + R2)2
The internal resistance of the generator depends on the con
ductivity of the plasma t; and the geometrical dimensions of the
system, and is given by
(8.4)
Magnetohydrodynamic Conversion o f Energy 177
where b is the width of the plasma jet and / the length of the elec
trodes, that is, the length of the region in which the plasma is
decelerated. It is assumed that the magnetic field is uniform over
the entire area bl. For a given set of parameters defining the geo
metric dimensions of the generator and properties of the plasma
jet, the maximum transfer of energy (although not the maximum
efficiency) corresponds to Ri = R 2 . Under these conditions
W = 1 0 -1 6 (8.5)
where fi = dbl represents the working volume in which the kinetic

energy of the jet is converted into electrical energy. The efficiency
with which the energy is converted is proportional to y\vzH 2.
Therefore, in order to increase the efficiency, both the conductivity
and the velocity of the plasma jet must be increased. It must be
noted that Equation (8.5) was derived under highly simplifying
assumptions and can only be used for very approximate estimates
of the electrical power. It cannot, of course, be used as a practical
formula in the theoretical design of magnetohydrodynamic
generators.
Some idea about the orders of magnitude involved in the
operation of magnetohydrodynamic generators may be obtained
as follows: Suppose that v = 105 cm per sec, H = 2,000 Oe and
d = 100 cm. Under these conditions, the electromotive force
U is equal to 2,000 V. If the resistivity of the plasma is of the
order of 1 ohm cm, b = 10 cm, and I = 100 cm, Equation (8.5)
shows that, when the external and internal resistances are equal,
the power released in the external load is about 10,000 kVa.
So far, we have confined our attention to the motion of a
plasma jet in a magnetic field because the principle of the magneto
hydrodynamic converter is based on this process. However, it is
evident that the converter must also incorporate certain other
elements. The original form of the energy is the thermal energy
of the gas, given to it either by a combustion process or by heating
the material by nuclear radiation, for example, in a nuclear reactor.
The first requirement is therefore to transform the energy of
random thermal motion into the energy of a directed beam. This
can be done by allowing the heated gas to expand into a vacuum

through a specially designed gas-dynamic nozzle (see Fig. 80).
In an expansion of this kind, no work is done against external
forces and the thermal energy of the gas is simply transformed into
the kinetic energy of the directed beam. Therefore, if the tempera
ture in the thermal reservoir (inside the nuclear reactor or in
ordinary space) is equal to 7j, it will be reduced to Ti at the mouth
of the nozzle. Consider an element of unit mass in the jet. The
initial store of heat in this element is C t T \ , where C T is the specific
heat. After expansion, the thermal energy is reduced to CjTi.
The difference between these two quantities is equal to the kinetic
energy of directed motion which for our element of unit mass is
equal to jV2. Therefore, the velocity of the beam leaving the nozzle
is given by
v = V2Cr(T, - T i ) .
However, it is not enough to produce a directed beam of rapidly

moving gas. It is also necessary to ensure that the jet should have a
sufficiently high degree of ionization, since otherwise it will not
be an adequate conductor of electricity.
The production of highly conducting ionized jets is one of the
main problems at the present stage of development of magneto-
hydrodynamic converters. In principle, this can easily be done if
there were no practical limits to the initial temperature of the gas.
In fact, at a sufficiently high temperature 7j, it is possible to
convert a considerable proportion of the original thermal energy
into electrical energy while still ensuring that the final temperature
72 is high enough for the gas to be thermally ionized and therefore
highly conducting. However, so far, such favorable conditions
have not been achieved in practice. A magnetohydrodynamic
generator with an initial gas temperature well in excess of 3,000-
5,000°K is beyond the reach of present technology, and therefore
the temperature of the gas as it enters the magnetic converter
cannot be much greater than 3,000°K. At temperatures of this
order, the air mixed with the combustion products of the mineral
fuel is only weakly ionized and its electrical conductivity is not
high enough.
Magnetohydrodynamic Conversion o f Energy 179
The conductivity could possibly be increased by introducing

into the gas jet a small proportion of impurities with low ionization
potential, for example, the vapors of alkali metals. The situation is
apparently 'more favorable if a nuclear reactor is employed as the
source of thermal energy, since any mixture of gases may then be
used as the carrier of thermal energy, provided only that it is not
made too radioactive by the nuclear radiation. In particular,
one could use helium with small amounts of cesium to increase the
electrical conductivity. (The ionization potential of cesium is
about 3.8 eV,, while the ionization potential of helium is 24.5 eV.)
However, even the use of impurities does not lead to plasma jets
with resistivities in excess of a few ohm cm, owing to the low
electron temperature of the plasma.
A jet of plasma entering the working region, where the magnetic
transformation of energy takes place, must necessarily be very
dense since it carries a high flux of energy. For example, in the
above numerical calculation, the energy flux entering the magnetic
converter must exceed 10,000 kVa, and therefore the energy flux
density must be greater than lOkVa per cm2. Since the energy
flux density is equal to the product of the amount of energy stored
per unit volume and the velocity of the plasma; that is, it is equal
to }pv2 X v, it follows that |pu3 > 1011 erg/cm2 sec. where p is the
density of the plasma and all the quantities are expressed in cgs units.
When v = 105 cm per sec, the density must exceed 2 X 10~4 g per
cm2. In air, this corresponds to an atomic (and ionic) concentra
tion of about 4 X 1018 particles per cm3. For comparison, we
note that at normal temperatures this would correspond to a
pressure of about 100 mm Hg. At such densities, the transfer of
thermal energy between the electrons, ions, and neutral atoms in the
plasma of the magnetohydrodynamic converter should proceed at
a high rate, and this means that the temperature of the electrons
cannot be very high.
The electron temperature Te is determined by the balance betweep
the energy communicated to the electrons by the induced electric
field and the energy lost by collisions, mainly with the neutral
atoms. The properties of the gas itself play a very important role
in this balance. In an atomic gas with a high ionization potential,
elastic electron-atom collisions are the most important. In each

such collision the electron transmits to the atom only a very small
fraction of its kinetic energy, and therefore the total heat transfer
from the electron component to the ions and atoms is relatively
small. Excitation of atoms should have only a minor effect on the
energy balance, since most of the energy lost by the electrons in
inelastic collisions is eventually returned as a result of the strong
absorption of the radiation by the gas itself. It may therefore be
supposed that the electron temperature in a jet of ionized atomic
gas is much higher than 7/ and To. Under these conditions, the
process may become practicable with jet resistivities of the order
of 1 ohm cm. In a jet of molecular gas, for example, a mixture of
air and combustion products, the electron temperature can hardly
be different from the temperature of the neutral component,
owing to the very strong inelastic interaction between the electrons
and the molecules. Therefore, the conductivity of the jet cannot be
very high. It follows that at velocities smaller than 105 cm per sec,
the plasma jet can travel relatively freely across the magnetic
lines of force without dragging them along (frequent collisions lead
to displacement of the particles relative to the lines of force).
In other words, in this case there should be no appreciable
freezing-in of the lines of force into the plasma; that is, the passage
of the plasma beam through the magnetic field should not lead to a
strong distortion of the field.
The development of magnetohydrodynamic generators is still
at a very early stage. There is a lack of important fundamental
data upon which the basic design calculation may be founded.
In particular, very little is known about the conductivity of plasma
under working conditions and the nature of the interaction between
plasma beams and the working electrodes. Nevertheless, it seems
that the magnetohydrodynamic converter will find its place in
technology and will eventually displace conventional machinery.
This expectation is based on the fact that the efficiency of con
version of thermal into electrical energy in magnetohydrodynamic
generators should, at least in principle, be much higher than in
steam and gas turbogenerators.
There is an important difference between the physical problems
Plasma Engines 181
arising in connection with magnetohydrodynamic conversion

of energy and the physical aspects of thermonuclear fusion. In
the former case, one is concerned with plasmas at enormous
temperatures, while in the other with ordinary low-temperature
plasma.
In principle, the magnetohydrodynamic converter is a reversible
device. It can convert the kinetic energy of plasma into electrical
energy, or it may be used to accelerate plasma with the aid of
electrodynamic forces by drawing electrical energy from external
sources. In the second case, it becomes an accelerator or injector of
plasma and can be used as the basis for a plasma jet engine.17
8.3 Plasma Engines

Plasma engines are also among the possible future applications
of the physics of plasma. Injectors of this kind have already been
discussed in one of the preceding chapters. It is convenient, at
this juncture, to return to this topic, which has been so widely
discussed in the popular literature.
The more striking applications of plasma engines are, of course,
those concerned with long-range cosmic flights. In order to ensure
that a spaceship will have the necessary maneuverability and
autonomous means of transport, rather than be a ballistic object
moving as a result of its inertia under the laws of gravitation, the
spaceship must be supplied with adequate sources of energy and
with jet engines which could be used to perform complicated
maneuvres in space. If the trajectory is such that the ability to
maneuvre in space is important, then fuel storage becomes an
acute problem. The thrust produced by the jet engine is equal to
w,y, where m v is the mass ejected per second and v the velocity of
the jet. For a given thrust, the consumption of fuel is inversely
proportional to the velocity v. In conventional jet engines, using
chemical fuels, the velocity of the jet is of the order of 105 cm per
sec. For a thrust of only 10 kg, a velocity of 105 cm per sec will
17These remarks are concerned only with physical principles and are not meant
to indicate that the practical design problems of magnetohydrodynamic generators
and plasma engines are identical. The two devices are, in fact, quite different.
require the consumption of 100 kg of fuel per day. It is apparent

that under these conditions the spaceship will only have limited
maneuverability.
The amount of fuel which must be carried on board the spaceship
can be reduced by a very large factor if conventional jets are
replaced by plasma jets with velocities of the order of 107 cm per
sec. However, this leads to a considerable increase in the rate of
energy consumption, since the power necessary to accelerate the
jet is iw iu2, and therefore for a given thrust the fuel consumption
is proportional to v. It follows that the application of plasma jets
for the control of spaceships will require the development of
powerful sources of energy, capable of being carried on board the
spaceship. The construction and operation of plasma jet engines is
still a matter for conjecture. The simplest kind of jet engine is the
coaxial plasma injector illustrated in Fig. 67.
This concludes our brief account of some of the possible applica
tions of plasma processes in the near and more distant future.
It is quite clear that other, less fanciful, applications will develop
in the course of time and will enable the very varied and interesting
properties exhibited by plasma to be used for practical purposes.
Fundamental Constants
E le c tr o n c h a r g e e = 4 .8 - 1 0 - * ° e s u
E le c tr o n m a ss me = 9 . 10- 28 g
P r o to n m a ss trip = 1 .6 7 - 1 0 - 24 g
S p e e d o f lig h t c — 3 * 1 0 10 c m /s e c
P la n c k ’s c o n s ta n t h = 6.62* 10“27 erg sec
B a ltz m a n n c o n s ta n t k = 1 .3 8 * 10—1 6 e r g /d e g r e e
183
186 Index
E la stic in te r a c tio n , 67 H e lic a l m o tio n , 4 5

E la stic sc a tte r in g , 68 H ig h -fr e q u e n c y field , 100
E lectrical c o n d u c tiv ity , 9 3 , 99 H ig h -te m p e r a tu r e p la sm a , 5 0 , 65 ,
E lectric d rift, 4 5 68 , 8 2 , 8 7 , 8 9 , 1 2 7 , 143, 1 6 1 , 165,
E le c tr o d y n a m ic fo r c e , 1 2 9 ,1 3 1 ,1 3 9 , 1 6 8 ,1 7 3 - 1 7 5
148, 149, 152, 1 6 2 ,1 7 6 H y d r o g e n p la sm a , 6 4 , 8 2 , 8 5 , 113,
E le c tr o d y n a m ic in je c to r s, 172 1 47 , 173
E le c tr o d y n a m ic p ressu re, 1 42, 153
E le c tr o m a g n e tic w a v es in m agne Im p a c t p a ra m eter, 17, 6 2 , 63
tized p la sm a , 156 In d u ced ele c tr ic field , 4 9
E le c tr o m o tiv e fo r c e , 143, 144, 176, In d u c e d e le c tr o m o tiv e fo r c e , 152
177 I n d u c ta n c e , 9 9
E le c tr o n -e le ctr o n in te r a c tio n , 6 1 , 62 In d u c tio n , 4 7
E le c tr o n -io n in te r a c tio n , 61, 62 In e la stic c o llis io n s , 68
E le c tr o n -io n in te r a c tio n c r o ss In e la stic in te r a c tio n , 71
se c tio n , 63 Ioffe, M . S ., 174
E le c tr o n -n e u tr a l a to m in te r a c tio n s, I o n iz a tio n , 3
62 I o n iz a tio n c r o ss s e c tio n , 73
E lectro n ru n a w a y , 9 4 I o n iz a tio n e n e r g y , 71
E le c tr o n tem p era tu re, 6, 66, 117, I o n iz a tio n p o te n tia l, 7 2 , 7 3 , 179
1 1 8 ,1 2 4 I o n o s p h e r e , 102
E le c tr o n -v o lt, 16 Io n s o u n d , 155
E le c tr o sta tic in te r a c tio n , 5 9 , 71 Io n te m p e r a tu r e , 7, 66 , 118
E n e r g y -le v e l d ia g r a m , 7 5, 84 Iso b a r , 106
E x c ita tio n , 75
E x c ita tio n c r o ss se c tio n , 76 J o u le h e a t, 162
K in e tic th e o r y o f g a se s, 55
F araday, 131, 145
L a n g m u ir e le c tr o n o s c illa tio n s , 156
F ermi, E nrico , 50
L a n g m u ir o s c illa tio n s , 153
F is s io n , 160
L a n g m u ir p r o b e s, 114
F r e q u e n c y , 23
L a r m o r c ir c le , 3 4 , 129
F u lly io n iz e d p la sm a , 68, 82
L a rm o r fr e q u e n c y , 2 9 , 3 9 , 147, 157
F u s io n , 160
L a rm o r m o tio n , 4 0
F u s io n r e a c tio n , 161
L a r m o r p e r io d , 4 4
L a r m o r r a d iu s, 2 8 , 3 2 , 3 5 , 127
G a s d isc h a r g e , 148 L a rm o r r o ta tio n , 4 2 , 86, 131
G a s-k in e tic p ressu re, 143 L a w o f in d u c tio n , 144
G ay-L ussac, 52 L a w s o f B o y le , 52
G e o m e tr ic a l c r o ss s e c tio n , 80 L o n g itu d in a l e le c tr ic field , 43
G r a v ita tio n a l field , 147 L o w -te m p e r a tu r e p la sm a , 64
Index 187
M a g n e tic d rift, 37 P la sm a je t, 1 7 6 , 177, 179, 1 8 0 , 182

M a g n e tic m ir r o r s, 3 6 , 4 8 , 5 0 , 170 P la sm a j e t e n g in e , 181, 182
M a g n e tic p r e ssu r e , 131, 132, 143, P la sm a m ic r o fie ld , 59
167 P la sm a o s c illa tio n , 96
M a g n e tic tra p , 3 6 , 4 2 , 170, 1 7 2 - 1 7 4 P la sm a r a d ia tio n , 78
M a g n e tiz e d p la sm a , 134 P la sm a w a v e s, 170
M a g n e t o h y d r o d y n a m ic c o n v e r te r , P la sm o id s, 1 5 2 , 172
1 5 3 , 1 7 7 -1 8 0 P ressu re g r a d ie n t, 105, 107, 139
M a g n e to h y d r o d y n a m ic g e n e r a to r , P r o b e c h a r a c te r istic, 114, 117, 118
1 7 6 - 1 7 8 , 180
M a g n e to h y d r o d y n a m ic s , 133 Q u a si-n e u tr a l, 8, 10
M axwell , 102
R a d ia l field , 16
M a x w e ll d is tr ib u tio n , 9 6 , 117
R a d ia tio n b e lts, 41
M a x w e llia n v e lo c ity d istr ib u tio n , 5 4
R a d io -in te r fe r o m e te r m e th o d , 123
M a x w e ll’s la w , 5 3 , 6 3 , 1 16, 118
R a m sa u e r effe c t, 70
M a x w e ll ta il, 95
R e c o m b in a tio n r a d ia tio n , 8 3 , 84
M e a n c o llis io n fr e q u e n c y , 55
R e fr a c tio n , 21
M e a n free p a th , 5 5 - 5 8 , 6 2 , 6 3 , 95
R e fr a c tiv e in d e x , 2 1 , 102, 122, 123
M e a n free tim e , 95
R e fr a c tiv e in d e x o f p la sm a , 120
M e a n tim e , 62
R e sis tiv ity , 100
M e ta s ta b le s ta te , 7 9
R ig h t-h a n d r u le, 128
N ewton , 105
Sakharov, A . D ., 162
N o n u n if o r m m a g n e tic field , 31
S k in effe c t, 1 4 8 , 165
N u c le a r r e a c to r , 17 9
S p e c tr o m e tr y , 123
O h m ’s la w , 9 3 , 1 40, 176 S u p e r c o n d u c to r , 142
O p tic a l a n a lo g y , 19
T amm, I. Y e., 162
O s c illa tio n s in p la sm a , 155
T h e r m a l c o n d u c tiv ity , 1 1 2 ,1 1 3 , 161
P e r io d , 23 T h e r m o n u c le a r fu s io n , 162, 165,
P e r m ittiv ity , 101, 156 175,181
P e r m ittiv ity o f m a g n e tiz e d p la sm a , T h e r m o n u c le a r g e n e r a to r , 175
158 T h e r m o n u c le a r r e a c tio n , 6 4 , 1 6 0 -
P e r m ittiv ity o f p la sm a , 1 46, 157 163, 1 7 0 , 1 7 4
P in c h e ffe c t, 1 48, 1 5 0 , 1 62, 170 T h e r m o n u c le a r r e a c to r , 8 2 , 8 9 , 143
P la n c k ’s c o n s ta n t, 7 7 T o k a m a k in s ta lla tio n s , 169
P la s m a , 4 , 7, 10 T r a n sv e r se e le c tr ic field , 4 5
P la sm a c o n d u c tiv ity , 1 37 T r o c h o id a l m o tio n , 4 5
P la sm a e n g in e , 181 T r o c h o id a l o r b it, 134
P la sm a fila m e n t, 1 6 3 , 165, 169 T r o c h o id s , 4 5
P la s m a in je c to r , 152 T rubnikov , B . A ., 87
188 Index
Uniform field, 12 Valence electrons, 72, 73

Uniform magnetic field, 26, 31 Variable magnetic fields, 46
Uranium reactors, 161 Vector product, 26
ABOUT THE AU TH OR
Lev A. Arzimovich is the Director of the

Plasma Physics Division of the Institute
for Atomic Energy in Moscow. In this
job he has assumed the position of the late
Dr. Kurchatov who was, until his death a
few years ago, considered one of the
world’s leading authorities on thermo
nuclear problems.
Academician Arzimovich is a full mem
ber of the Academy of Sciences (U.S.S.R.)
and a member of its Presidium. He is the
foremost Soviet plasma physicist and one
of the acknowledged world authorities in
this field.
THIS BOOK WAS SET IN
TIMES ROMAN AND PERPETUA TYPES
BY TRADE COMPOSITION, INC.
IT WAS DESIGNED BY THE STAFF OF
BLAISDELL PUBLISHING COMPANY.

Elementary Plasma Physics
Lev A. Arzimovich
Plasma physics, the study of highly ionized gases, posse sses

exceedingly interesting properties th a t a re being increasingly
applied to the solution of im portant problem s of m odern te c h
nology and applied sciences. Thus the basic properties of
plasm a are of major interest to th e general reader. However,
m ost present-day books employ extremely com plex m a th e
matical tools an d require a good preparation in m odern th e o
retical physics.
This book a tte m p ts to interpret the fu n d a m e n ta ls of plasm a

physics in a m a n n er u n d erstan d a b le to a reader with a high-
school knowledge of m a th e m a tic s and physics. This d o es not
imply, however, th a t the material is presen te d in a predigested
form —for assimilation without thought. The book d iscu sses not
only the various plasm a processes but also th e m e th o d s of
analysis of such processes. By becom ing familiar with th e se
m ethods, the reader will be able to attain a genuine u n d e r s ta n d
ing of the physical significance of the p h en o m en a involved.
Blaisdell Publishing Company

A Division of Ginn and Company
New York • Toronto • London

Arzimovich - Elementary Plasma Physics - 1965

Uploaded by

Copyright:

Available Formats

Arzimovich - Elementary Plasma Physics - 1965

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Arzimovich - Elementary Plasma Physics - 1965

Uploaded by

Copyright:

Available Formats

Elementary Plasma Physics

Bernard T. Feld, Massachusetts Institute o f Technology

LEV A. ARZ I MOVI CH

BLAISDELL PUBLISHING COMPANY

First English Edition, 1965

Copyright © 1965, by Blaisdell Publishing Company,

1.1 Production of Plasma 1

2. Motion of Charged Particles in Electric and Magnetic Fields 12

2.1 General Laws of Motion for Charged Particles in

3. The Motion of Charged Particles in Plasma 52

3.1 A General Review of the Properties of Gases 52

5. Electric Current, Diffusion, and Thermal Conductivity 90

6. Experimental Techniques 114

7. Plasma in a Magnetic Field 126

8. Technological Applications 160

T a ble of F u n d a m en ta l C o nstants 183

1.1 Production of Plasma

where T is the absolute temperature and A is the atomic weight of

By increasing the temperature from the lowest value (a fraction

F igure 1. Degree o f ionization o f hydrogen as a function o f temperature.

The relative number of ionized atoms in hydrogen is shown in

of gas discharges. When an electric discharge is produced in a

where k is the Boltzmann constant, which is equal to 1.38 X 10~16

Tc is usually of the order of 10,000°K, while Ti and To are usually

1.2 Plasma as a Quasi-neutral Medium

We shall now discuss the definition of plasma and its main

ne — «i "T 2n2 + 3n3 + • • •.

This shows that the plasma as a whole is quasi-neutral; that is,

electric field which may be sufficient to equalize the currents and

where n is the concentration of charged particles (number of elec­

other sign. This mechanism automatically maintains the equality

1.3 General Nature of the Motion of Charged Particles in Plasma

a very weak one — gives rise to an electric current in the plasma.

F igure 2. Spatial variation o f the internal electric field along a given

Motion o j Charged Particles

2.1 General Laws of Motion for Charged Particles

F igure 3. Trajectories of charged particles in a uniform electric field.

x = v0t, y = — /2, (2.1)

which are well known from elementary mechanics and represent

on the particle and the distance traversed by it in the direction of

— = Ze— = Z [ / - 1.6-IO-12, (2.4)

where Z = q/e is the number of elementary electric charges

3A positively-charged particle can only be accelerated along a falling potential.

by a singly-charged particle, for example an electron, on passing

2.2 The Motion of a Particle in the Field of a Point Charge

the field produced by a point charge. This case corresponds to the

F igure 4. Trajectory o f a charged particle in the field o f a

Approximate formulas are a very powerful tool in the elucidation

2.3 Optical Analogy for the Motion of a Charged Particle

layer is infinitely. In thin practice, this can never be realized exactly

noted by a and /3 . Since the velocity component parallel to the

F igure 6. Motion o f a particle in an electric field o f a slowly varying

2.4 Motion in an Alternating Electric Field

F igure 7. Motion o f a charged particle in an alternating electric field:

of the form shown in Fig. 7a (a cosine curve). The distance between

of the period. The process is then repeated periodically. A plot of

5 = ^ - 5(1 — cos w/). (2.9)

In practice, the electric field cannot be established instantaneously

which differs from the preceding expression by a constant term

F igure 8. Growth o f the amplitude o f an alternating electric field.

where n is the concentration of charged particles (number of elec

4 V . _ l_l I | (J d \ _ i | _____ l

fields. If the magnetic field is not strictly uniform, then in ad

merely give rise to a reduction in the region in which the longitu