Caltech Li

il|t r illlll

llllll

llll llll

;0q? 0l,t

illlilllll

APPLIED PHYSICS LIBRARY

CALIFORNIA INSTITUTE OF TECHNOLOGY

.-{ ;1;. "r

- --.t77
sl.lN\

UNITED STATES

'{'r"s* DEPARTMENT oF
',,--.,oi-

ENERGY

r,/l

AB0ur rHE rEcHNrcAL TNF0RMATT0N cENTER

The Technical Inforrrration Center in Oak Ridge, Tennessee.

has been the national centel fcrr scientific and technical
infornration lor the Department cif Energy (DOE) and its

James B. Edwards
S et:

reta ry

William S. Heffelfinger
Ass

s 1.r

Sec rt-'ta ry

Manirrler rrenl anci Ad

rn

nistratlr-rn

Joseph G. Coyne
Manager

Tectrnrcal lnf orrnatron Center

lrt developing and managing DOE's technical information program, the Center places
under biblitlgraphic control not only DOE-originated intbrpredecessor agencies since 194(:.

mation but also worldrvide literature on scientific ancl

in the energy tleld and announces the

tecl-rnical advances

source und availabilitv

of this rnforr"nation. Whereas the

rded to
i.tr::

ii

QC

,,=, =
TITLE

k"g"'"i

11"

Otf""n uave.

ac
718.5

%lBo*leY
ligsagawa.
The Alfven wave,

lvrse=
-

Applied physics Library

ABOUT BOOKS PUBLISHED BY THE
TECHNI CAL I NFORMATION CENTER
Books published by the Center include
undergraduate, graduate, and workshop
textbooks; harrdbooks and manuals; reference books; monographs; specialized
reviews; and DOE-sponsored symposium,

workshop, and short-course proceedinqs.

nrental,

7L8"5

cAilr0nt{tA il{Silrurr

0r ruclilol0fiy

DOE/T|C-11 197

DOE Critical Review Series-Advances in Fusion Science and Engineering

DE82001 702)

(DO E/ER/53093-1

11's

t.XL-_
"'7 t 11

ti6 ":

--

i4'j o

The
a

Fl

!,*l

ATFVENWE
.ir.',

d'',!

Akira Hasegawa
Bell Laboratories
and

Chanchal Uberoi
lndian Institute of Science

1982
Prepared for the

Office of Fusion Energy

Office of Energy Research

U. S. Department of Energy

Jean S. Smith, Publicatlon Editor

Editing, composition, proofreading, illustrations. book design, and page

for this publication were performed by staff members of the

makeup

Technical lnformation

Center.

i":niifOfnia
r:nOlOgS
hli,r,, ,
t"l;":,.,., ;. "

il i*r,,-;fi*gfifig

P"bftiftw
TECHNICAL INFORMATION CENTER
U. S. DEPARTMENT OF ENERGY

Library of Congress Cataloging in Publication Data

Hasegawa,

Akira, 1934-

The Alfvdn wave.

(DOE critical review series)
lncludes bibliographical references and index.

"DOE/TlC-11197."
Supt. of Docs. no.: E 1 .28: DOE/TlC'1 1 197
Uberoi, Chanchal.
1. Magnetohydrodynamic waves. l.
United States. Dept. of Energy. Office of Fusion Energy.

ll.
lll. Title.
lV. Series. QC71S.5.M36H37
81-507894 r5BN0-87079-125-7 AACR2

530.4'4

Available as DE82001702 tDOE/TIC-11197l' for $11.50 f rom

National Technical I nformation Service
U. S. Department of Commerce
Springfield, Virginia 22161
DOE Distribution Category UC-209

Printed in the United States of America

1

982

To
Professor Hannes Alfven

the discoverer of the wave that

stimulated so much importan t

search in plasma physics

has
re_

FOREWORD

This publication is the second in the Advances in Fusion Science and Engineering series, a part of the DOE Critical Review Series. The purpose of the
Advances in Fusion Science and Engineering series is to provide the fusion
research community with high-quality reviews in specialized areas of fusion science and engineering.

Although Alfv6n waves have been the subject of study in space plasmas for a
long time, only recently have plasma dimensions and betas reached sufficiently
high values to make it possible to study Alfvdn waves in the laboratory. Auxiliary heating by means of Alfv6n waves has been tried both in the United States
and abroad. Preliminary indications are that heating may be possible with the
very high absorption efficiency predicted by theory. Because of the localized
nature of wave conversion, Alfv6n waves have the potential of providing plasma
profile control. Alfv6n waves have recently been proposed for space-resolved
measurement of magnetic fields in tokamaks and may become important as a
diagnostic tool in the future.
This monograph deals with the properties of Alfvdn waves and with their
application to fusion. The book is divided into seven chapters dealing with linear
properties in homogeneous and inhomogeneous plasmas. Absorption is treated
by means of kinetic theory. Instabilities and nonlinear processes are treated in
Chapters I to 6, and the closing chapter is devoted to r,heory and experiments in
plasma heating by Alfv6n waves.

FOREWORD

ln view of the growing importanc to the tokamak program of auxiliary

heating by means of electromagnetic waves, I believe this monograph will fill a
need Ior an exposition of the essential elements of Alfvdn waves, with emphasis
on geometries relevant to tokamaks. The extensive reference section at the end
ol'each chapter will assist the reader in expanding his knowledge of the subject.
W. L. Sadowski
Etlitor, Advances in Fusion Science and Engineering
DOE Critical Review Series
Division oJ' Applied Plasma Physics
OJJ'ice of Energy Research
U. S. Department of Energy

PREFACE

Of the many types of

waves that exist in plasmas, the Alfvdn wave is probably the best
studied. However, some of its unique properties, such as the absence of a discrete
eigenmode in a nonuniforrn plasma and the effect of a finite ion gyroradius, are not
widely known. Although the importance of the Alfven wave in astrophysical and space
plasmas was realized soon after its discovery in 1942, plasma physicists paid relatively

little attention to it until recently because of difficulties in exciting it in the

laboratory. However, the recent progress in thermonuciear fusion research has
increased the possible physical size and magnitude of plasma beta (the ratio of the
plasma to magnetic-field pressure) so that the Alfven mode is now accessible in
laboratory plasmas. Consequently a large part of this book explores areas of relevance
of Alfven waves to fusion. This book also serves as a good review for scientists in the
areas of astrophysics, space physics, and solid-state physics. Recent references in these
areas are listed at the end of Chap. l.
Chapters I and 2 introduce the Alfvdn wave and describe its linear properties in a
homogeneous medium. Chapters 3 and 4 cover the effects of inhomogeneities on these
linear properties. Particular emphasis is placed on the appearance of a continuum
spectrum and the associated absorption of the Alfven wave which arise as a result of
the inhomogeneity. Kinetic theory is used to explain the physical origin of absorption.
Chapter 5 is devoted to the associated plasma instabilities.

In recent publications considerable interest has been shown in the study of

nonlinear phenomena associated with the Alfv6n wave. Since it is not possible to
vil

vilt

PREFACE

describe all this work, we restricted our study to nonlinear processes in laboratoryscale plasmas. Nonlinear effects discussed in Chap. 6 include quasiJinear diffusion,
decay, a solitary wave, and a modulational instability. To compensate for our
omissions, we have included a large number of references on recent work at the end of
the chapter.
The Alfven wave is a potential candidate for radio-frequency heating of a fusion
plasma because low-cost power sources in the appropriate frequency range are readily
available and its absorption rate is high. The principles of Alfven-wave heating. a design
example, and present-day experimental results are described in Chap. 7.
The idea for this review originated when the authors met at Bangalore, India, in
December l976.It has been diffioult to co-author a book while the authors are sitting
on opposite sides of the earth, but thanks to our very efficient typists, Hildegard
Franks (United States) and K. S. Rajagopal (India), it has gone rather smoottrly.
After the manuscript was finished, one of the authors (A. Hasegawa) used it in a

plasma seminar course at Columbia University graduate school to teach plasma

physics, using the Alfven wave as an example. The richness of phenomena related with
the Alfven wave, which involves the magnetohydrodynamic theory, the kinetic theory,
plasma instabilities, and nonlinear effects, was very useful in showing how the
plasma-physics theory can be applied to a real problem.
We both are grateful to our spouses, Miyoko Hasegawa and Professor S. N.
Balasubrahmanyam, for their encouragement and endurance. One of the authors
(A. Hasegawa) is particularly appreciative of his favorite music colleague, Sachiko
Nishida, who has continuously helped to calm his impatient spirit. The other
(C. tlberoi) is thankful that her children, Rajju and Soonu, "understood" when their
mother had to go to the "office."
Akira

Hasegawa

Bell Laboratories
Chanchal Uberoi
Indian Institute of Science

CONTENTS

Foreword

vii

Preface

Alfv0n

lflf

6vg-"ld eal"

Dispersion Relation
Properties
Reflection and Transmission

3
4

Finite Conductivity

Compressibi

ity Effects

Coriolis Force Effects

Cylindrical Geometry: Torsional Alfv6n Wave
Experimental Studies

8
10

References

11

Alfvdn Waves in Plasmas

14

Dispersion Relation
Compressional Alfv6n Wave

15

Hall Current Effects: Finite Frequency Alfvdn Wave

lon-Neutral Collision Effects
Finite Larmor Radius Effects: Kinetic Alfv6n Wave
Alfv6n Waves in Cylindrical Plasmas
Experimental Studies of Alfvdn Wave Properties

18

17

ix

19

19

22
25

CONTENTS

29
30

Concluding Remarks
References

Spectral Analysis of ldeal Alfvdn Wave in

lnhomogeneous Magnetic Fields
The Wave Equation
Solution of the Wave Equation

Concluding Remarks

41

References

42

Resonant Absorption and Mode Conversion of

Alfvdn Surface Waves in Nonuniform Plasmas
Alfven Surface Wave
Resonant Absorption of Surface Wave
Resonant Mode Conversion

44
46
49
54
60

References

Plasma lnstabilities Asociated with the Alfvdn Wave

Fire-Hose lnstability

Two-Stream lnstability
Kelvin-Hel mholtz I nstabil ity

Drift Alfven-Wave

nstabi

lity

References

62
62
63
64
65

66

Nonlinear Processes
Ouasi- Linear D iffusion

68
69

Parametric Decay

73
80
86
89

Solitary Wave of the Kinetic Alfv6n Wave

Modulational lnstability and Envelope Soliton
References

33
34
36

Plasma Heating by Alfvdn Wave

Resonant-Absorption and Energy-Deposition Rate

Heating Rate
Design Examples

92
92

Experimental Results

97
100
103

References

110

Appendix A: List of Symbols

112

Appendix B: Equations Used in the Book

115

Appendix C: Studies of Nonidealizing Parameters

(Diagrammatized)

119

Author lndex

120

Subject lndex

123

AI,FYEN WAVE

-"'DEAL"

The well-known basic principle of magnetohydrodynamics is that, when a conducting

fluid moves in the presence of a magnetic field, the motion of the fluid gives rise to
electromagnetic force, which produces electric currents. These electric currents modify
the initial magnetic field, and, in the presence of the magnetic field, the mechanical
force due to the electric currents modifies the motion of the fluid. Thus there is an
interaction between the magnetic field and the motion of the fluid. The mechanical
effects of a magnetic field are equivalent to a hydrostatic pressure,B2 f 2pr, combined
with a tension , Bt lpo, along the lines of force. In an incompressible fluid, the
hydrostatic pressure can be balanced by the pressure of the fluid so that only the
tension, B'l2po, remains effective. Analogy with the theory of stretched strings
suggests that this tension may lead to the possibility of transverse waves along the lines
of force, with a velocity v4 given by
vA.

(l.l)

= 82 (pop)- t

where p is the mass density of the fluid.

The possibility of such waves was first established by H. Alfvdn

in 1942. The

name

given by Alfven was "electromagnetic-hydrodynamic" waves.

DISPERSION RELATION

To study small-amplitude wave propagation in an ideal incompressible fluid, we

linearize the ideal magnetohydrodynamic (MHD) equations (see the appendix), taking

ALFVEN WAVE-'(IDEAL''

p to be constant and

o -+ oo, an

infinite conductivity assurnption. The fluid is initially

stationary, and the equilibrium magnetic field is Bo = (0,0,8s).W. consider a linearly
polarized wave with the wave magnetic and velocity fields in the y direction;

B"=Br=vx=vz=0
of wave propagation in the x-z plane. This is equivalent to
taking an incompressible perturbation, p ' v = 0. With the use of (v 'p)v =
vr (Ovy/DV) = 0, the equation of motion of fluid becomes

We take the direction

dv., Dv.,
pt=pH=-J*Bo
where the

x component of the cunent density J*

relates

(r.2)

to B, through the Ampdre

law,
l..oJx =

Br and vy

- lalt

(1.3)

are related through the electric field by Ohm's law

E*+vrBq=0

Q.+7

& --E

(l.s)

and the Maxwell equation

0E*

0t

lf

we substitute Eqs. I .2, 1.4, and 1.5 into Eq. 1.3, we have

otB, _

B'?o

otB,

(1.6)

aT- tttyE

and

vy=_d*u, __vehBy

(1.7)

Equation 1.6 represents a wave propagating along z direction with Alfven velocity

va

given by tB6(pto

p)-'h.

The sinusoidal solution of Eq. 1.6 is

Bv=A

sin c.r

(#-

,)

(1.8)

PROPERTIES

where A is the amplitude;v, is then given by

vv =

The lines

1r;,)qtin

of force now lie in the y-z

."(*-,

(l.e)

plane, and their shape is given by the

equation

dv B'
...:=dz Bo

( l .10)

Solving this, we find that (OV/d| = vy, i.e., the velocity of the lines of force is the
same as the fluid velocity. This leads to the well-known concept of "frozen-in" fields.
The magnetic lines of force are "frozen in" to the conducting fluid and are thereby
constrained to move with the fluid. It is this concept which gives rise to the analogy of
Alfven waves to those waves along an elastic string. The nugnetic lines of force have a
lateral pressure and a tension equal to 82 /trrs. Since the magnetic lines are frozen in the
fluid, the density can be taken as p, the fluid density. Hence the wave velocity along
the magnetic lines of force is given as Be(p6p)- y'. Wr also note that the wave equation
1.6 is exact in that we did not have to linearize to obtain it [(v . V )v term in Eq. 1.2 is
identically zerol. In this respect the Alfven wave is considered as an exact solution.

PROPERTIES

The most striking property of the Alfven wave is that it propagates in one
direction only (i.e., along the magnetic lines of force that pass through it) and
therefore suffers no geometric attenuation with distance (i.e., it does not spread out

three-dimensionally around a source of disturbance which leads to "spherical

attenuation"). There are cases of astrophysical importance in which densities are so
low that attenuation due to collisional effects becomes negligible over relevant length
scales.

The Alfven wave is transverse in nature since the particle velocities are
perpendicular to the direction of wave propagation.
It is a low-frequency wave; we have already set an assumption of neglection of
the displacement current. The discovery of this wave in an incompressible medium
became a very important and highly interesting phenomenon since the energy could be
transmitted without large-scale exchanges of the fluid elements.
Alfvdn velocities are several orders of magnitude smaller than the velocity of light
(Table l.l). Because of this velocity, the wave can be easily controlled. This fact is
used in the coupling of the Alfven wave to other waves as seen in the excitation of
sound waves in solids.

ALFVEN WAVE-"IDEAL"
Table

l.l

ORDER OF MAGNITUDE OF ALFVEN WAVE

VELOCITIES IN FOUR DIFFERENT MEDIA

Magnetic

Media

field,

Solar corona
Galactic arms
Hydrogen gas (at
100 pm Hg pressure)

Bismuth

Alfvdn wave
velocity,

Density,
El"mt

cm/sec

10-2
10-5

10-t 6

3 x lOs

10-24

3 x 106

10o
103

10-E
l0-1o

3 x 107
3 x l0?

In an Alfven wave the energy is transmitted partly as magnetic and partly as

kinetic. The electric energy is small and is insignificant for an Alfvdn wave. The
kinetic-energy density

l1/z\82

ior'=frA2

sin2." (*

-) =;G

equals the magnetic-energy density. Therefore, for an MHD Alfven wave, the
kinetic-energy density is equal to the magnetic-energy density. This equi-partition of
energy is, of course, a consequence of the frozen-in lines of force, and this will
therefore hold only for a medium of infinite conductivity.
However, as indicated in Chap. 2, the Alfven wave, which has a perpendicular
wavelength comparable to the ion gyroradius, carries electric energy that is comparable
to the magnetic-field energy.
REF LECTION AN D TRANSMISSION

Here we consider reflection and transmission of the Alfven wave. We take

By=Asinc^rffi-)
Bi = A'
Bil =

A"

sin c,r

(* - )

sin <^r

E*=Av41

(-*

-)

sin<^r(*-)

(z < 0, incoming wave)

(z > 0, transmiued

wave)
(1.')

(z < o,reflected wave)

(r<

o)

F'INITE CONDUCTIVITY

El=A'v42sino'e-

EI = - A" uA1 sin c,.r (*

-)

(z>o)

(t.t2)

(z<o)

(1.

l3)

Boundary conditions are

Br+Bil =Bi,

(1.14)

E*+E!=B'

(1.r s)

which gives

2v,qt
o,- v41 * v42 o
( I .16)
ra
^rr-vAt
--rr

-VAz

VAT*VAz

Rigid boundary (conductinE, pz ="o): A =A";there is no phase shift in the induced

magnetic field. From the relation between v, and By, it can be easily shown that there
is a l80o phase shift in the velocity field.
Free surface (pz = 0): A.= A"; hence there is a 180o phase shift in the magnetic
field, and the velocity field is reflected with the same phase.

At a nonconducting

boundary, the current density, J*, associated witfi the wave

- - A". Hence the phase shifts in the magnetic and
velocity fields are the same as those for the free surface.
A general treatment of the reflection and refraction of Alfven waves, when the
magnetic field is inclined at an angle of the plane of separationz= 0, has been given by
Ferraro (1954). It was shown that the general laws of reflection and refraction no
longer remain simple, as in the case when the magnetic field is normal to the surface of
separation, since they depend on the orientation of the field with respect to the
surface of separation of the two liquids.

must be zero, which gives A

F!NITE CONDUCTIVITY
The "infinite conductivity" is an ideal situation. What are the conditions in which
the Alfven wave can still exist if the conductivity is large but finite? This can be
studied by introducing the effect of finite conductivity in ohm's law,

E"*vyBo=lt*

(l.aa)

ALTIVEN WAVE_"IDEAL''

The dispersion relation is

s2

-('i - #)k2

=o

k=+#('-*HC'

(r.r7)

(I

.18)

where o is the conductivity.

For small damping,

k=,(*

#)

(l.r e)

The damping distance is

21ts ov3s

zo=V-

(1.20)

For an Alfvdn wave with wavelength ). to exist,

) ).

(1.21a)

r=$>r

(1'21b)

z6

That is,

where L is the Lundquist numbor (Lundquist,1949a,1949b).

COMPRESSIBI LITY EF FECTS

When compressibility is taken into account, the dispersion relation assumes the
form

t*

- S *e + 4)

+ vlvN

.o,' o]

(# -vf; cos2 ,) = o

(1.22)

where vs = (7Po/po)% ip the sound speed and 0 is the angle between the equilibrium
magnetic-field direction and the wave propagation direction.

CORIOLIS FORCE EFFECTS

There are three modes of propagation. The first two are given by the upper and
lower signs of the equation

# = *rr", + vl) * j

tt"3 + ui.)'

au?ul ,os, o1k

(t.23)

and the third by

=vN

cos2 g

(1.24)

They are termed fast, slow (magnetosonic waves), and intermediate (Alfven waves)
modes, respectively.

When 0 = nf 2, the only mode that persists is the fast mode. The fast mode has
about the same velocity as the Alfven speed when the magnetic pressure is large
compared with the material pressure and represents the compressional Alfven mode.
The compressjonal Alfven mode arises owing to compression of the magnetic lines of
force.
When 0 = 0, we have the Alfven and sound waves.

cos

vl

If va ) ur, then the compressibility effects are not very large; the two modes, v4
0 and v, cos 0, are anisotropic in direction; and the third mode, va(l + ll2v?l

sin2 g), is not affected much by the direction.

However, if vr/v4 is not small, we no longer have a disturbance transmitted along
the lines of force as a coherent whole unless the motion is wholly normal to 86.

CORIOLIS FORCE EFFECTS

In the presence of the Coriolis force, 2pvx O, where Q is the uniform angular
velocity directed along the magnetic-field direction, the dispersion relation for Alfven
waves assumes the form

"fi

(t

-ffi

-2vf,v2s+vf, = s

(t.2s)

where vo = c.r/k is the phase velocity.

The two modes of propagation are

fof

"i (t *4)-'

known as the "ineftial," or "fast," mode and the "magnetic," or.,slow," mode.

(t.26)

ALFVEN WAVE-"IDEAL''

Thus, in the presence of the Coriolis force, a plane Alfven wave splits into two
<;ircularly polarized transverse waves with velocities vps < v4 ( vpf. It can be easily
shown that there is no equipartition between the kinetic and the magnetic energies
(Lehnert, 1954a).

CYLINDRICAL GEOMETRY:
TORSIONAL ALFVTN WNVE

Most

of the experiments

use cylindrical containers

in the laboratory. To

understand the theory of these experiments, we consider the Alfvdn wave propagation
in cylindrical geometry for incompressible and finitely conducting fluid.
Taking the magnetic field in the z direction and considering axisymmetric
perturbations so that 0ld0 = 0, we have

vg=v

vr=0

Vz=0
(1.27)

Br=0 Bo=B Br=0

The equation for B is given by

#.+#.(u,-i)

B=o

(r.28)

where

ktul)po
k3=-[u'* (.' - iun
o- t; is the resistivity.
The solution of B, which is finite at r = 0, is given

(t.2e)

and n (=

B=

as

AJr(k"r).i(kz-ot)

(1.30)

where J1(k.r) is the Bessel function of the first.order. The field variables are

v-J, = '

J,'

B"k
"
ltoPd)

j\ g

(1.31)

Ilo

=&'='!5"']
J1(k"r)
lto

CYLINDRICAL GEOMETRY: TORSIONAL ALFVEN WAVE

E,=*(l-H'
(1.32)

u,=#u"j#ilu
In the absence ot resistivity, the electric field is entirely radial. The Alfven wave
becomes a torsional, or shear, wave. Adjacent magnetic surfaces are able to shear past
each other without mutual coupling in their motion. If we take k = k, * i//, where k, is
the propagation constant and / is the attenuation distance, / is given by
(1.33)

The quantity k" will be determined by the boundary conditions.

For example, if we take a liquid in the cylinder with radius R and height H and at
the bottom z=0 excite a motion, v(r,t)= (vor/R)e-t@t, the boundary conditions are

r=

Jr=0

z=0

v=vo

z=H

Jr=

(condition

l)

R' -ic^rt (condition 2)

(t.34)

(condition 3)

Condition I gives the wave numbers for different modes of propagation as

Jr(kcnR)= 0; i.e., k"n is given by the zeros of the Bessel function. The corresponding
kn is then given by Eq. 1.17.
Conditions 2 and 3 can be satisfied as follows: Consider the repeated reflections of
the waves. A wave traveling up to the surface is reflected without phase difference in v,
but H is in the opposite phase in the upward and the downward waves. At the bottom
the conditions for H and v are reversed. At a point z, if we add all the waves traveling
up and down, i.e.,

v(z) = v(O) [eik'

eik(2H -z) +r-ik(2H+z) + e-ik(4H-z) + . .

.]

(1.35)

the solution becomes

v(r,z,t)=

AnJr(k"nr)%ffil).-i<,-,t

(1.36)

ALFVEN WAVE-('IDEAL''

10

Forz=0andr(

R,wehave

vor-$

AnJl (k"nr)
R-L
n=l

(1.37)

frorn which An can be determined. Finally, the expression for v(r,z), valid for

0(v< Rand0(z(H,is

v(r,z)-n

(,pt-

p_

2J1(k"nr)
kgnrJ2

cos k1(H

(k"nR)

cos knH

z)

(1.38)

The preceding analysis was given by Lundquist (1949a) while studying the Alfven
waves experimentally.

EXPERIMENTAL STUDIES
The concept of hydromagnetic waves was introduced by Alfvdn in 1942 (Alfven,
1942), but the first experimental observation in the laboratory was made in 1949,7
years later. This was due mostly to the difficulty in obtaining conditions under which
the wave attenuation is sufficiently low to allow observation. The early experimental
studies of Alfven waves were conducted by using liquid metals. Liquid mercury was
used by Lundquist (1949b) and liquid sodium by Lehnert (1954b). Since more-recent
experiments in plasmas (discussed in Chap. 2) are very similar to these early
experiments, we shall give a brief account of these in this chapter.
The mercury, in a cylindrical vessel, was placed between the pole pieces of an
electromagnet with the axis of the cylinder aligned parallel to the direction of the
field. The cylinder diameter was 15 cm, the height 15 cm, and the magnetic field
about 104 G. These parameters satisfy the condition L > 1. A circular disk was placed
at the base of the vessel so that it would be free to oscillate in its plane about the axis
of the cylinder.
Because of the inertia of the mercury, the oscillations of the disk do not produce
any motion of the mercury in the absence of the magnetic field. However, as soon as
the magnetic field is switched on, the motion of the disk interacts with the mercury
electromagnetically, and its oscillations are transmitted along the rnagnetic field. This
is, in fact, the propagation of hydromagnetic waves, which, in this case, are torsional
waves along a cylinder. As these waves reach the surface, they produce a rotational
motion. This could be detected by a mirror floated suitably on the surface. The
deflections of the light reflected from the mirror would indicate the rotational motion
at the surface. Very low frequencies, less than 1 cycle/sec, were used to keep the
damping at a low level. Hence it was not possible to obtain any standing wave effects.
The waves here will, obviously, be standing for L = zl4, and the amplitude will be quite

REFERENCES

11

large. However, owing to the damping effect, which originates from the flniteness of
the conductivity of mercury, the amplitude on the surface is much decreased.
Lehnert repeated the experiment with sodium, because its higher electrical
conductivity and lower density are more suitable for MHD experiments (the Lundquist
number is 35 times that in the mercury), but the experiment is more difficult to

perform because

of the nature of

sodium. Since

it

is more easily oxidized,

it

is

enclosed in a vessel in an atmosphere of some inert gas. So that the motion of the
surface in hydromagnetic oscillation can be observed, use is made of the probes, which
are held at various points on the surface. The motion of the material carrying frozen-in
lines of force with it induces in the probe an electric field given by

f,=-vX B

(1.39)

The relative magnitudes of these electric fields induced in different probes give at
any instant the relative motions of the surface elements and hence the oscillation of
the surface.
The experimental results agree qualitatively with the theory.

HEFERENCES

Text Citations
Alfvdn, H., 1942, Existence of Electromagnetic-Hydrodynamic Waves, Nature ,150: 405.
V. C. A., 1954, On the Reflection and Refraction of Alfvdn Waves, Astrophys. J.,
119:393.
Lehnert, 8., 1954a, Magnetohydrodynamic Waves Under the Action of the Coriolis Force,Astrophys.

Ferraro,

Il9: 647.
1954b, Magnetohydrodynamic Waves in Liquid Sodium, Phys. Rev., 94: 815.
Lundquist, 5., 7949a, Experimental Investigations of Magnetohydrodynamic Waves, Phys. Rev.,
-,
J.,

76:1805.

-,

I949b, Experimental Demonstration of Magnetohydrodynamic Waves, Nature , 164: 145.

General References
Alfvdn, H., 1950, cosmical Electrodynamics, oxford University Press, New york.
Amte, P. G., I9'14,Dual Alfvdn Waves in Superfluids with High ElectricalConductivity, IndianJ.
Phys. ,48: 843-845.
Cowling, T. G., 19 57, Magnetohydrodynamics, Interscience Publishers Inc., New York.
Cramer, N. F., J. A. Lehane, R. M. May, and F. J. Paoloni, 1970, Compressional Alfvdn Wave
Reflection from a Magnetic "Cliff," Phys. Lett., A,32:307-308.
Herlofson, N., 1950, Magnetohydrodynamic Waves in a Compressible trluid Conductor, Nature,

165:1020.
Lundquist, S., 1952, Studies in Magnetohydrodynamics, Ark. Fys.,5: 297 ,
Narasimha, M. S., 1973, The Influence of Coriolis Force on MHD Waves, hoc. IndianAcad.

A,78:202-207.
Novikov, I. I., and A. E. Kutnik,1974, Analogue

Sci.,

Sect.

of the Alfvdn Velocity

Associated

with

the

ALFVEN WAVE-"IDEAL''

12

Translation and Rotation of an Ideal Infinitely Conducting Liquid, Fiz. Khim. Obrab. Mater.,

4:32-34.
Roberts, P. H., 1955,

On the Reflection and Refraction of Hydrodynamic

Waves, Astrophys. J.,

I2l:720.

Shah, S. K. D., and K. M. Srivastava, 1969, Hydro.-nagnetic Waves in an Infinitely Conducting

Compressible Fluid with an Inhomogeneous Magnetic Field of Constant Direction, Nuovo

Am.,8,63:

105-114.

Stein, R.F., 1971, Reflection, Refraction and Coupling of Magnetohydrodynamic Waves

Density Step, Astrophys. J., Supplement Ser., 22: 479.

at

Alfvdn Waves in Astrophysics

Beckers, J. M., 1976, The Flux of Alfvdn Waves in Sunspots, Astrophys. J.,203:739-752.
Belcher, J. W., and R. Burchsted,1974, Energy Densities of Alfvdn Waves Between 0.7 and 1.6 AU,
J. Geophys. Res., '79: 4'765-4'168.
L.Davis, Jr., and E. J. Smith, 1969, Large-Amplitude Alfveln Waves in the Interplanetary
-, Medium, J. Geophys. Res., 74: 2303-2308.
and S. Olbert, I9'15, Stellar Winds Driven by Alfvdn Waves, Astrophys. J.,200:369-382.
-, and C. V. Solodyna,l975, AlfvdnWaves and Directional Discontinuities in the Interplanetary
-, Medium, J. Geophys. Res.,80: 181-186.
Burlaga, L. F., and J. M. Turner, 1976, Microscale "Alfvdn Waves" in the Solar Wind at I AU,,I.
Geophys. Res., 8l : 73-77.
Cummings, W. D., C. Countee, D. Lyons, and W. Wiley, 1975, The Dominant Mode of Standing
Alfvdn Waves at the Synchronous Orbit, J. Geophys. Res., 80: 3705-3708.
Davydov, V. M., 1971, Diffraction of Low-Frequency Electromagnetic Waves on a Stratified Earth
Surrounded by a Thin Gyrotropic lonosphere, Geomagn. Aeron., 11 : 845-851.
Dwarkin, M. L., A. J. Zmuda, and W. E. Radford, 1971, Hydromagnetic Waves at 6.25 Earth Radii
with Periods Between 3 and 240 Seconds, ,1. Geophys. Res., 76: 3668-3674.
Hollweg, J. V., I9'12, Supergranulation-Driven Alfvdn Waves in the Solar Chromosphere and
Related Phenomena, Cosmic Electrodyn., 2: 423444.
Alfvejn Waves in a Two-Fluid Model of the Solar Wind, Astrophys.,r., 181: 54'l-566.
-,1973, Improved Limit on Photon Rest Mass, Phys. Rev. Lett.,32:961-962.
I974b, Transverse Alfvdn Waves in the Solar Wind Arbitrary K, J. Geophys. Res.,79:
-,1974a,

-t

1539-1

s4l.
A., 1975, The Confinement of

Galactic Cosmic Rays by Alfvdn Waves, R. Aston. Soc.,

170:251-260.
loffe, Z. M., 1969, On the Mechanism of Alfvdn Wave Generation in Comets, Astron. 2h.,46:

Holmes, J.

1328-1329.

Lanzerotti, L. J., I974, Magnetohydrodynamic Waves in the Magnetosphere and the Photon Rest
Mass, Geophys. Res.

Lett.,l:

229-230.

Mullan,D.J.,l9T4,SunspotModelswithAlfvdnWaveEmission,Astrophys.J., 187:621-63I.
Owens, A. J., 1975, A New Test for Alfvejn Waves in Interplanerary Space,Astrophys. Space

Sci.,

38:469473.
Piddington, J.H.,1974, The Alfvdn-Wave Theory of Solar Flares, Sol. Phys., 38: 465-481.
Radoski, H.R., 1968, The Polarization of AlfveinWaves, J. Geornagnet. Geoelect.,285-287.
Richter, A. K., 1974, Wave-Trains in the Solar Wind. II: Comments on the Propagation of Alfvdn
Waves in the Quiet Interplanetary Medium, Astrophys. Space Sct.,26:95-105.
Skilling, J., 1975, Cosmic Ray Streaming. I. Effect of Alfvdn Waves on Particles,R. Astron. Soc.,

172:557-566.
Toptygin, I. N., 1973, Magneto-Hydrodynamic Waves

in Plasma of Solar Wind, Izv. Radioftz.

(USSR),16: 971-980.
Uchida, Y., and T. Sakurai, 1975, Oscillations in Sunspot Umbras Due to Trapped AlfvdnWaves
Excited by Overstab ility , Astron. Soc. Jpn., Publ. ,27 : 259-27 4 .

REFERENCES

13

Vinogradov, P. A., and V. A. Parkhomov,I975, MHD Waves in the Solar Wind. A Possible Source
of Geomagnetic PC 3 Pulsations, Geomagn. Aeron. (USSR) (Engl. TTansl.), 15: I34-l3:.;
Geomagn. Aeron., 15 : 109-1 I 1 .
Volk, H. J., and W. Aplers, 1973, The Propagation of Alfvdn Waves and Their Directional
Anisotropy in the Solar Wnd, Astrophys. Space Sci. ,20: 267 -295.
Whang, Y.C., 1973, AlfvdnWaves in Spiral Interplanetary Field, J. Geophys. Res.,'18: :-221-7228.
Wilson, P. R., 1975, Alfvdn Waves and Meyer's Sunspot Model, Astron. Soc.,172:535-543.

Alfvdn Waves in Solids

Ashkinadze, B. M., and P. D. Aitukov,I975, Magnetoplasma Effects in Optically Excited GE and
Sl, Fiz. Tverd. Tela, 17: 1004-101 1 ; Soy. Phys.-Solid State (Engl. Tiansl.) , l'l : 634-637 .
Baynham, A. C., and A. D. Boardm'an, I97I, Plasma Effects in Semiconductors; Helicon and
Alfvdn llaves, Taylor & Francis, London, England.
Beletskii, N. N., A. P. Tetetvov, and V. M. Yakovenko, 1973, Nonpotential Surface Waves in
Magnetically Active Semiconductor Plasma, Sov. Phys.-Semicond. (Engl. Transl.), 6:
1

807-1

10.

Caille, A., and P. R. Waliace, 1969, Solid State Commun.,

j:

1283-1286.

R.

Wallace, 1971, Interaction of Magnon and Alfvdn Waves in Itinerant Anti-, ferromagnetic Chromium lI, Solid Stote Commun., 9 : 1437 -I44I.
Chen, F. C., J. Kirschbaum, and Y. H. Kao, 1972, Temperature Dependence of Alfvdn-Wave
Amplitudes and Carrier Relation Times in Bismuth, Phys. Rev.,8,6:327-333.
Fahidy, T. 2., I976, Wave Phenomena in Magnetoelectrolytic Systems, Electrochim. Acla,2l:
2r-24.
Guthmann, C., J. P. D'Haenens, and A. Libchaber, 1973, Effect of Anisotropy and Landau

and

P.

DampinginaGeometrywhereQ IsNotPerpendiculartoB,Phys.Rev.,8,8:561-570.

Hetmann, R., W. Baune, and G.. Kuka, 1974, Alfvdn Waves Cyclotron R.esonance in Semimetallic
Bi-SB, Phys. Status Solidi (b),6I: K77-K80.
Hess, H. D., and H. Hinsch, 1973, Propagation of Alfvdn in Bismuth, Appl. phys.,z:27-30.
Hu, P., V. Narayanamurti, and R. C. Dynes, 1975, Metals Using Nanosecond Pulses of Laser Light
Case of Alfvdn Waves in Bismuth, Phys. Rev. Lett",34: i0l5-1018.
lsaacson, R. T., and G. A. Williams, 1969a, Oscillations of the Fermi Surface ofBismuth, Phys.
Rev. , I85 : 682-688.
and G. A. Williams, 1969b, Alfvdn-Wave Propagation in Solid-State Plasmas. II. Bismuth,
-, Phys. Rev., 177 : 738-746.
Kaner, E. A., and V. G. Skobov, I97I, Plasmn Effects in Metals: Helicon and Atfvdn llaves,
Taylor & Francis, London, England.
Krauss, A. R., and J. K. Furdyna,l9'1.3, Altvdn-Wave Propagation and Damping in Pyrolytic and
Single-Crystal Graphite , Phys. Rev., B,7 : 2520-2539.

Marston, E. H., and Y. H. Kao, 1969, Damped Affvdn Waves in Bismuth. A Determination of
Chzuge-Carrier Relaxation Times, Phys. Rev. , I82: 504-510.
Nakahara, J., H. Kawamura, and Y. Sawada, lg70,Interaction of Alfvdn Waves with Bernstein
Modes in Bismuth, Phys. Lett.,31A: 27I-272.
Platzman, P.M., and P. A. Wolff, I973, liloves and Interactions in Solid State Plasmas,Academic
Press, New York,
Schwarz, 8., and H. Hinsch, 1976, Magnetoplasma Waves in Anisotropic Semiconductors and
Semimetals (Alfvdn Waves in Bismuth) , Appl. Phys. ,10: 325-330.
Suematsu, H., and S. Tanuma, 1976, Alfvd,nWave Propagation in Antimony,J. Phys. Soc. Jpn.,4l:
496-501.

V. G., M. V. Glushkov, and L. V. Lyn'ko,1971, Propagation of Magnetoplasma Waves in

Bi Alloyed with Te, JETP Lett. (USSR) (Engl. Transl.) ,13: 24j-249.
phys.,42:333-340.
wilson, P. R., 1975, Alfvdn waves and the sunspot Phenomenon,
Vaselago,

^Sol.

AI,FYEN WAYES

PLASMAS

'N

It is now well known that the MHD, or ideal, Alfven wave is a fundamental rnode of
wave propagation in plasmas involving low-frequency ion oscillations in the presence
of a magnetic field.
At first, it

appears surprising that a wave propagating in an incompressible and

medium can exist in an ionized gas in which equations V' v = 0
conducting
infinitely

and E+v x B=0 hold. As pointed out by Herlofson (1950), these difficulties are
more formal than real. Alfven wave exists in plasmas for very low frequencies,
<^r

ar"i. Hence the equation of motion for plasmas,

*=:(E+vx
dt m'

B)

(2.1)

at low frequencies becomes equivalent to the infinite conductivity assumption in a

liquid. Actually, this characteristic condition means that the magnetic flux through a
surface moving with the liquid is independent of time. In both cases the magnetic lines
of force will therefore move with the plasma. Concerning the incompressibility
condition, the Alfven-wave perturbations are such that the velocity of the medium is
always perpendicular to the wave normal and thus do not compress the plasma.
Therefore V ' v = 0 is satisfied without suggesting that the plasma is incomplessible.

14

DISPEi(SION RELATION

15

DISPERSION RELATION

We shall use the two fluid descriptions for the plasma to understand the electric
field and current which are set up in the plasma with the propagation of the Alfven
wave.

Since the perturbations do not compress the plasma, the plasma can be treated as
"cold." The equivalent dielectric tensor derived from the two-fluid model for a cold
plasma is

.fi ;l)

(2.2)

with

s=*(R+L)
(2.3)
D

=+(R

- L)

R= I

-'p"

r-, '

atp"

rD-1
-r----

.3i
a'pi

a(a: + c^lce)

<^.r(c,r

c..:.1)

-'p" .3i

a'

(2.4)

(2.s)

(2.6)

u|

The notations are standard and are listed in the appendix.

The dispersion relation for the waves propagating at an angle g with respect to the
magnetic-field, Bs = (0,0,86), direction is given by

Ana-Bn2+C=0

(2.7)

where

A=Ssin29+Pcos2g
B = RL sin2 g +PS

C=PRL

(l

+ cos2 0)

(2.8)

ALFViN WAVES IN

16

where n

ckl as is the

PLASMAS

index of refraction. When 0 = 0, the two transverse modes

(2.e)

n2=L

n2=R
These are circularly polarized modes. In the

limit

^l
n'=R=L=

co

are

(."i,

we tlnd
(2.10)

r+olJocz
B3

waves' are degenerate

where p = 1ot1i. The two modes, shear and compressional Alfven

al0 = O.
From Eq.2.10, we tind

4k2 = l+(c2
-i^,

=,
lv26)

*urA

(2.1 1)

wave,
c2. Since p,q,.210 gives the dielectric constant for a low'frequency
the
by
modified
is
the Alfven wave can be treated as an electrornagnetic wave, which
high dielectric constant of the medium'
then given as
The electric field is in the x direction. The velocity components are

with vf,

ui*=,"fr('

.-,i 1,, _ 4\-'

u'r=G
V-,;/
The corresponding solution

(2.r2)

*)-'u
u

(2.13)

of the electron equation of motion gives, in the limit

at 4 -'.",
(2.t4)

vex-0
rr=--

'ey

The fluctuating
VxE=-DB/DI,as

magnetic

(2.1s)

Bo

field, By, is

u=?u,

given

by the Maxwell equation,

(2.t6)

E x Be drift
Thus we have E in the x direction and Bs in the z direction, which gives
Thus
electrons.
and
(.3i,
ions
is the same for both
in the y direction, which, for os2

COMPF.ESSIONAL

ALFVEN

WAVE

17

the motion of the fluid in the y direction has magnitude of the velocity lE/Bol. The
in the magnetic-field lines of force move with the velocity (<,r/k)lBvt'Bs i, which
is equal to fluid velocity. The plasma particles thus bound to the magnetic lines of
force provide an added inertial effect, which lowers the frequency of an electromagnetic oscillation of the given wavelength. The principal mode induces a shear of
the magnetic lines but no compression (i.e., magnitude of the field does not change in
the first order). The restoring force is thus derived solely from the tension (i.e.,
curvature) of the magnetic lines. The string analogy, as in the conducting liquid case,
holds for ionized media also. It should be noted, however, that the concept of plasma
frozen to lines of force and moving with them is an accurate one so long as theie is no
ripples

electric field along Bs.

What sustains the electric field that was presupposed to be there? As E fluctuates,
the ion inertia causes ions to lag behind the electrons, which induces the polarization
drift, vo, in the direction of E. This drift, vi", given by Eq. 2.I2,causes a current J to
flow in the x direction. The resulting J x Be force on the fluid is in the y direction and
is 90o out of phase with the velocity v. This force perpetuates the oscillation in the
same way as that in any oscillator in which the force is out of phase with the velocity.
It is, of course, always the ion inertia that causes an overshoot and sustains the
oscillation, but in a plasma the momentum is transferred in a complicated way via the
electrom agnetic forces.

COMPRESSI ONAL ALFVEN WAVE

In Eq. 2.7,when 0 = nf 2, the dispersion equation gives

Pltoc2
" RL
n'=T=l+Til

(o(<,.:.1)

(2.r7)

In this case the wave propagation is perpendicular to the magnetic lines of force,
and the compression of the magnetic lines contributes to the restoring force along with
the curvature.

The (shear or torsional) Alfven wave has no magnetic field induced

in

the

longitudinal direction, whereas the compressional Alfven mode does have a component
B parallel to Bs in general.
Considering 0, not in the vicinity of nf2, we get

.
.
no-l*---

PFoc2
B6

(2.t8)

and

n2cos2

o.-l*gttolz
B3

(2.re)

ALFVEN WAVES IN PLASMAS

18

The first isotropic mode is the compressional mode, and the second is a torsional
or shear mode, This can be determined by calculating E, which shows that E, = 0 for
both the modes, whereas E* =0 for Eq.2.l8 and Ey =0 for 8q.2.19. From the
relation

E+vx B=0

(2.20)

v-;cos(k*x +krz-@t)

(2.2r)

we get

for the mode of Eq. 2.18 and

v

-i

cos

(k*x + k"z -

@t)

(2.22)

for the mode of Eq. 2.19, which confirms the compressional and torsional nature of
these modes.

HALL CURRENT EFFECTS:

FII{ITE FREOUENCY AtFVdN WAVE
The Hall current effect becomes more important than the electrical resistivity
whenever the magnetic field, B, is so large that the gyrofrequency of electrons greatly
exceeds their collision frequency. This can happen in high-temperature low-density
plasma for reasonable strength of the magnetic field.
When c^r ( @ce but <^r < <*.1"i, then

oci
vt1^ @ci i c,,'t
^2

R=l

and Lx +
v"4

(^)ci_

(nci

0)

(2.23)

These two modes give

'i="i( .*)
where

* and -

(2.24)

are for compressional and shear Alfv6n waves, respectively.

of finite ion gyrofrequency disperses the Alfven wave, and the

Hence the effect

unidirectional property, which allows the "ideal" Alfv6n wave to propagate without
attenuation, is not met in the case of a "finite frequency" Alfv6n wave.
When the Coriolis force is also taken into account, an interesting effect appears
(Uberoi, 1976). The presence of rotation and Hall effects gives rise to a critical
frequency , G)cr,lying between the rotational and the ion-cyclotron frequencies,

ION_NEUTRAL COLLISION EFFECTS

at.r

19

=rr (+ - ,n)

(2.2s)

where L = B6/(4rne), at which the slow and fast modes have the same phase velocity.

Across this frequency, both modes undergo polarization reversal. At this critical
frequency, conditions may exist for strong coupling between the two modes of
propagation.

ION-NEUTRAL COLLISION EFFECTS

In a partially ionized

plasma, the ion-neutral collisions can be.come very

important. We use subscript "i" for ions and "n" for neutrals.
The effect of neutral atoms is introduced through the collision term Pin in the
momentum equation for the plasma ion. Since ffii - lrrn,

Pit

flnilnn

(-

vi + vn),rni

(2.26)

where rni is the ion-neutral collision frequency. The velocity, vn, is determined by
the equation

*=r"'

(2.27)

pni = _ pin

(2.28)

with

The Alfven velocity is then given by

Bo

(2.2e)

'A itop
with

p"(l

Pr=Pi* Ur,
where

r=

ir)
(2.30)

asfvni.

FINITE LARMOR RADIUS EFFECTS:

KINETIC ALFVEN WAVE
When the perpendicular wavelength,

2nfky

becomes comparable

to the

ion

gyroradius, p|, ions can no longer follow the magnetic line of force, whereas electrons

20

ALFVEN WAVES IN PLASMAS

are still attached to the field line because of their small Larmor radius. This produces
charge separation and coupling to the electrostatic mode (Stepanov, 1958; Stefant,

1970). To treat this effect, we must use Vlasov equations for ions. When kpi ry 1, the
frequency of the fast (compressional) mode becomes the order of the ion-cyclotron

frequency; hence we can consider that the fast mode is decoupled. Then the
of the magnetic-field perturbation, Br, can be assumed to
be much smaller than the transverse components. This allows us to use a scalar
potential, @, to represent the transverse component of the electric field, E1,
compressional component

Er=-Vrd

(2.3t)

because then B, becomes zero. To represent the z component of the electric field, we
must use a different potential, ry',

E,=-y
so that the transverse components
equations are Poisson's equation,

vio

of

(2.32)

x E are not zero. The appropriate field

e,(ni
*{4
=:-n")=o
dz'
e6

(2.33)

and the z component of the Ampere law,

*o"(Q -

{t)=

tto

at

(Ji, + Jsr)

(2.34)

The quantities ni, fie, Jiz, and J"" are obtained from the linearized Vlasov equation
(appendix) of each species by assuming the unperturbed distribution function to be
Maxwellian (Rukhadze and Silin, 1969).If the electron thermal speed is larger than the

Alfv6n speed,
eni

-=_
s

.7i
_T
vTi

[1

rs(r1)

e-ri] r

-ryro(rJ r-^,(1 - i6i)0

+=
co *u+id")o
vte
u7; k.

rtoJiz=;;

to(xJ e-^i

(l - i6i)f

(2.3s)

(2.36)

(2.37)

FINITE LARMOR RADIUS EFFECTS: KINETIC ALFVEN WAVE

a7^

/.\

FoJez=-;";'#tt+i6e)*
c'vte Lz

21

(2.38)

where tri = k? p? ,lo(\i) is the modified Bessel function of the first kind, and 6; and 6,
are the fractional Landau damping rates,

6i= Z(n)h gi'/' ,*P (-gi- t )

(2.3e)

6"=(n),hBir(+J(H'

(2.40)

pr=z*

(2.4t\

and

vA

where vTe and

vTi are the electron and ion thermal speeds, respectively. The
by substituting Eqs. 2.35 to 2.38 into 2.33 and

dispersion relation can be obtained

2.34. lf we ignore the damping,

(.--^,

-#,) tt **,,

-roe-^,,]

#,(r -roe-riy

e42)

where c, is the ion sound speed with electron temperature, cl = Ts/mi. The preceding
dispersion relation shows the coupling of the Alfven wave and the ion acoustic wave.
In a low-beta plasma, since v, ) v4 ,they are decoupled, and the dispersion relation of
the Alfvdn wave becomes

T"
,"s2
* .r.
k|u'n I - Ise-tri Ti "t
Here we cail the wave represented

by this dispersion relation the "kinetic

wave" because of this kinetic property.

Alfv6n wave reduces to
(.,2

If

).1

(2.43)
Alfven

1, the dispersion relation of the kinetic

= kluTl, . utr,

(i- +)]

(2.44)

The kinetic Alfven wave can propagate across the magnetic field and faces both
electron and ion Landau damping because of its coupling to the electrostatic mode. It
is also important to note that the wave accompanies the electric field in the direction
of the ambient magnetic field.
If the plasma is relatively cold such that vTe ( v4, the electron inertia becomes
important and the dispersion relation is modified to

,,)2

=ulv26(t-

ulri)

(1

.5r)-'

(2.44')

ALFVEN WAVES

22

IN

PLASMAS

ALFVTN WAVES IN CYLINORICAL PLASMAS

The theory of Alfv6n wave in cylindrical geometries in Chap. 1 was developed to
understand experiments in which only the torsional waves were excited. However, in a
plasma-wave guide (a long electrically conducting, cylindrical tube containing a plasma
confined by a uniform magnetic field directed along the axis), there can be many
different modes of propagation. It therefore becomes necessary to study the properties

of Alfven wave relative to other modes of propagation'

Under laboratory conditions it is usually assumed that the plasma pressure

is

negligible compared with the magnetic pressure.

The ideal MHD equations for axisymmetric perturbations (m = 0) of the type
exp (- i@t + ikz) give the following three equations for the components of the

magnetic-field vector:

(kl-k2)Bu=g

(2.4s)

(ki - k')B, = tO *

(2.46)

+*(,+) +t248,='u++

(2.47)

The equations thus decouple to give the pure torsional and compressional Alfven
waves.

For the compressional mode, B" is governed by the equation

E -i*-(or -k2-i)t'=o

(2.48)

Bz = AJt (k3r)

(2.4e)

k?=k2t_

(2.s0)

d'8,

which gives

where
k2

The compressional modes therefore have a lower cutoff frequency, which

is

obtained by setting k equal to 0 in Eq. 2.50,

a)o c = kcvA

(2.s 1)

ALFVEN WAVES
where

k.

IN CYLINDRICAL

PLASMAS

23

is determined by the eigenvalue condition

Jt(k"R)=O

(2.52)

which arises owing to boundary conditions at r = R, the radius of the cylinder.

For torsional or shear modes, the frequency is determined by
fr: = kV4

(2.s3)

The quantity 89 is an arbitrary function of r. These modes are therefore charact erized,
by a constant phase velocity v4 and by the absence of a cutoff. Since R is not

contained

in

Eq. 2.53, the walls have no effect. The Alfven rnode thus has a very

special nature.

The compressional and shear Alfvdn modes correspond to transverse-electric (TE)

modes and principal modes, respectively, in the hydromagnetic wave guide theory,
which was first discussed by Newcomb (1957).

The cutoff frequency, a)sct is the lowest frequency below which the compressional mode does not exist. In laboratory plasmas the frequency, a)oc, is often
comparable to or is greater than the ion-cyclotron frequency. This makes the study of
Alfv6n wave, excited well below the ion-cyclotron frequency, simpler. For example, in
the experiment of Wilcox, Boley, and DeSilva (1960) described in the next section, the
Alfven velocity, V4, wrls 2.8 x 107 cm/sec for a 86 of 10,000 G. Therefore

eci=108/sec and .0.=#

Hence c*ro. is comparable to

(withvA

- 4x l0landR=5

cm)

c^.r"1.

If

we move away from the ideal situation and take the Hall effect, finite
conductivity, and ion-neutral collisions into accouirt, the general dispersion relation,
when m # 0, becomes (Woods, 1962)

[rk'-kl (l +i6rk2 + i6'k3)] {s(k, +k3)-ki [t +i6r (k, +k3)l ]

= k2 (k2 + k3)st2
wneres=l+
-

.a)
v4

,f)=

a)ci

Pnl+ir
Pil+12

Q.54)

I-

ALFVEN WAVES IN PLASMAS

24
and

D=

uoaLn

The two modes in Eq. 2.54, which are now coupled, represent "slow" and "fast"
modes corresponding to torsional and compressional modes at low frequencies.

Irt k = k, * (i//); then at low frequencies

Eq. 2'54 gives

(slow)
\-^_...,
_ .-A _ I _r.2
o,-g-{r^
s1 I u: (fast)

(2.55)

and

+=*(".Erkl-u:o')

:il'

(2.s6)

whereS=Sr +is2.
When Sr =

and s2 = 0, we get

1=

2tto-kru'A_

a:n(k2e, +

2.57)

kZ)

for the slow wave, which is the same as Eq. 1.33 to lowest order in 4 in Chap. I .
When 6 = 0,
So V.t
l= ' n=

(2.58)

2l.i,s!

wave. The damping distance, which is due to both neutral particle

collision and conductivity, is proportional to Bo for the torsional Alfven mode.
When m ;& 0, the components of the magnetic-field vector associated with the slow

for the slow

and fast waves to

first order in ,f),

B, =

ic[#

are given, respectively, as

r*,u"r)

,oftui

r;(k.,)]

25

EXPERIMENTAL STUDIES OF ALFVEN WAVE PROPERTIES

and

QJ-(k"r)

krJh(U.r)]

k,m
I
,u=olffir[1r.r; -oJ.(k"r)l

(2.60)

B, = Ak"J* (k.r)
When m

0, the boundary conditions give the following eigenvalue conditions:

J',n1qn)=

J*(k.R)=0
When

ill = 0,

(fast wave)

(2.61)

(slowwave)

these become

f'o1t.R) = O (for both

(2.62)

waves)

The boundary conditions are for conducting walls. When ffi = 0, however, the
torsional wave has the boundary condition of F,q.2.62, even in case of insulating walls.
We note from Eq. 2.59 that, for low frequencies, <,r ( <,.r"1 and m = 0, the slow
wave

is

almost entirely

torsional wave. Hence

at low

frequencies and for

axisymmetrical perturbations, the torsional wave decouples from the compressional

mode, and, if eoc- <,.r"1, the Alfven torsional mode is the only mode propagating in
laboratory plasmas at very low frequencies.

EXPERIMENTAT STUDIES
OF ALFVEU WNVE PROPERTIES

The experimental generation and detection of Alfven waves in a magnetized

first reported at Berkeley, Calif., by Allen, Baker, Pyle, and Wilcox (1959)

plasma was

and in England by Jephcott (1959).

The experiments reported in plasma, though conceptually very similar to the experiment conducted by khnert (1954b), allowed a detailed study of the properties of
Alfven waves (see Chap. 1). The major difficulty in these experiments was that the
damping distance, which was due both to ion-neutral collisions and finite conductivity, was proportional to Be. Very high magnetic fields of the order of l0 kG were
needed to overcome the damping. If B0 is large, v4,aod hence the wavelength,
becomes uncomfortably large unless the density is high.
In the experiment of Wilcox, Boley, and DeSilva (1960) and Wilcox, DeSilva, and
Cooper (1961), with Be equal to 10kG, a density of 6 x l01s/cm3 was used to

ALFVEN WAVES IN PLASMAS

achieve a low Alfven speed of 2.8 x l0? cm/sec. The work was done in a highly
ionized hydrogen plasma created in a cylindrical copper tube immersed in a uniform
a,rial magnetic field that could be varied in strength up to 16 kG. The plasma was
produced by an ionizing wave front that in many respects was similar to a

"switch-ion" shock wave. The plasma thus produced was initially 80 to 100% ionized,
with an ion density greater than 5 x lOs/cm3 and an estimated temperature of around
104 "K. Torsional hydromagnetic waves were induced in the plasma by discharging a
small capacitor between the copper cylinder and a coaxial electrode mounted in an
insulator at the end of the tube. The coaxial driving-electrode system generates
azirnuthally symmetrical waves. The wave magnetic fields were observed with small
magnetic probes, and the integrated wave electric field was observed as a voltage on
coaxial electrodes. Since the theoretical analysis postulates an azimuthally symmetrical
wave propagation, care was taken to measure the azimuthal symmetry. It was
measured experimentally with four magnetic probes disposed 90o apart on the same
base circle. The symmetry was found to be present within a few percent.

o
()
0)

C)

b+
tb

I
o

o
E

()

o
J

ul^
>J

o
o

Llj

812
AXIAL MAGNETIC FIELD,

2.1

KG

of Alfv6n wave as a function of axial magnetic-field strength.

10-E glcmt. [From A. W. DeSilva, J. M. Wilcox, W. S. Cooper lII, and F. I.
Boley, Experiments on Alfv6n-Wave Propagation, in Magnetohydrodynomics, Proceedings of the
Fourth Biennial Gas Dynamics Symposium, Evanston, Ill,, Aug.22-24, 1961, p. 149, A.B.
Cambel, T. P. Andelson, and M. M. Slawsky (Eds.), Northwestern University Press, Evanston, Ill.,
Fig.

Initial

1962.1

Measured phase velocity

mass density,

EXPERIMENTAL STUDIES OF ALFVEN WAVE PROPERTIES

The experimental results are shown in Figs.2.l to 2.6. Figure 2.1 shows
of wave phase velocity vs. magnetic field, depicting the linear

measurements

dependence as given by Eq. 1.1 in Chap. 1.

Figure

2.2

shows

the attenuation against the magnetic-field strength. The

attenuation is mainly due to finite conductivity since the plasma is fully ionized. The
agreement between theory and experimental values is good. The attenuation of Alfven
waves due to neutral damping was measured by Jephcott and Stocker (1962) and by
Brown (1965).
The most important check on the theoretical predictions for the wave field was
provided by the observation of wave reflections. The reflections were observed from
an insulating boundary, represented by a pyrex end plate, from a conducting boundary
(a copper plate), and from a plasma-neutral interface. The observed change of phase
of the wave fields on reflection in Figs. 2.3 to 2.5 is in agreement with the theory.
The magnetic-probe measurement of the azimuthal magnetic field, Be, associated
with the wave indicated that the strength of 86 was about 100 G. Since the static field
is 10,000 G, the wave field is about l%, and therefore a smali-amplitude theoretical
treatment is valid.
Figure 2.6 gles the radial distribution of Be, which is described by a first-order
Bessel function as predicted by the theory.

llJ

o
)_
F5
o
=-

(,

=4
E
o
o

UJ

f
F
rrt
JJ

CL

ul

LrJ )

ot.t J

16

12

AXIAL MAGNETIC FIELD,

Fig.2.2 Attenuation

'measured

plot. [From A. W. DeSilva,

as a function

of

KG

axial field strength. Solid curve is a theoretical

M. Wilcox, W. S. Cooper III, and F,I. Boley, Experiments on

Alfvdn-Wave Propagation, in Magnetohydrodynamics, Proceedings of the Fourth Biennial Gas
J.

Dynamics Symposium, Evanston, IlL, Aug. 22-24, 1961, p. 150, A. B. Cambel, T. P. Anderson,
and M. M. Slawsky (Eds.), Northwestern University Press, Evanston,

Ill.,

1962.1

ALFVEN WAVES IN PLASMAS

28

Et'oo
<,..

l!

'oo

t! ll

le
g:cE

s5
L
'-J
a

TIJ

Ll)

(J

lo

34

3"1

=E
r!
uJ

=
TIME, psec

Fig,2.3

of Alfv6n wave from a pyrex end plate. The voltage (E.)

phase and the magnetic field (86) out of phase, in accord with theory for a
nonconducting boundary. IFrom A. W. DeSilva, J. M. Wlcox, W. S. Cooper III, and F.I. Boley,
Experiments on Alfvdn-Wave Propagation, in Magnetohydrodynamics, Proceedings of the Fourth
Biennial Gas Dynamics Symposium, Evanston, Ill., Aug.22-24,196t, p. 151, A. B. Cambel, T. P.
Anderson, and M. M. Slawsky (Eds.), Northwestern University Press, Evanston, 111,,1962,1
reflects

Oscillogram showing reflection

in

t! _

500

5E
ur 250
^

u, tl

kg

trr

FE
-J

trJ

()
L

lo
(,^'

s;

UJl!

68

3
TIME, psec

Fig.2.4

Oscillogram showing reflection from a copper plate with ionizing current still flowing to
the plate. The electric field has reflected out of phase and the magnetic field in phase, in agreement
with theory for a conducting boundary. IFrom A. W. DeSilva, J. M. Wilcox, W, S. Cooper III, and
F. L Boley, Experiments on Alfv6n-Wave Propagation, in Magnetohydrodynamics, Proceedings of
the Fourth Biennial Gas Dynamics Symposium, Evanston, Ill., Nre.22-24, 196I, p. 152, A. B,
Cambel, T. P. Anderson, and M. M. Slawsky (Eds.), Northwestern University Press, Evanston, Ill.,
t962.1

CONCLUDING REMARKS

3>
gE

29

750

U^ uJ 500
lu lt

kg

5E zso
u llJ
'- -s
aur
O
9
F

!<t

70

3ol

;-*

10

TIME. psec

Fig.2.5 Oscillogram showing reflection from an interface

phases indicate a nonconducting boundary at

III,

between plasma and neutral gas. The

reflection. [Frorn A. W. DeSilva, J. M. Wilcox, W.

S.

and F. I. Boley, Experiments on Alfv6n-Wave Propagation,inMagnetohydrodynomics,

Proceedings of the Fourth Biennial Gas Dynamics Symposium, Evanston, Ill., Aug.22-24,1961,
p. 153, A. B. Cambel, T. P. Anderson, and M.M.Slawsky (Eds.), Northwestern University Press,
Cooper

Evanston, lll.,1962.1

CONCLUDING REMARKS

In low-beta ideal MHD plasmas, the axisymmetdc perturbations propagate as two

types of decoupled waves, the torsional, or shear, Alfven wave and the compressional
Alfven wave. When pressure perturbations are taken into account, the torsional modes
are not affected, because they do not involve any compression of the plasma. The
properties of the compressional wave, however, change considerably, and the wave
couples to the ion acoustic wave. In a nonuniform plasma, on the other hand, the
density gradient introduces a frequency-dependent coupling between the torsional and
compressional waves of uniform plasma. As a result, an axial component of the
magnetic-field perturbation, which is normally associated only with the compressional
wave, is now carried by the torsional wave as well, and was, in fact, detected for
low-beta plasmas at frequencies well below the lowest waveguide cutoff (McPherson

and Pridmore-Brown, 1966). The pressure perturbation is not important for the
torsional wave but affects the compressional modes in a uniform plasma. However, in a
nonuniform plasma, it plays a very important role also in the torsional mode. This
change is rather striking and important for the Alfven wave in nonuniform plasmas, as
discussed in Chap. 4.

ALFVEN WAVES IN PLASMAS

30

,/
0

/t- --i

40

c;
lrJ

t!-

\-Theory

C)

ul 30

z(,
llJ

= 20

q
fr'

RADIUS, cm

2.6 Radial distribution of the magnetic field, Bg, measured near the receiving end of the tube
after the wave made three transits of the tube. - - -, theoretical distribution for the lowest mode
J, (k"r). Higher modes are all negligible at the distance represented by this measurement. [From
A. W. DeSilva, J. M. Wilcox, W.S. Cooper III, and F. L Boley, Experiments on Alfvdn-Wave
Propagation, in Magnetohydrodynomics, Proceedings of the Fourth Biennial Gas Dynamics
Symposium, Evanston, Ill., Aug, 22-?4, 1961, p. 155, A. B. Cambel, T. P. Anderson, and M, M
Slawsky (Eds.), Northwestern University Press, Evanston, Ill., 1962.I

Fig.

REFEHENCES

Text Citations

Allen, T. K., W. R. Baker, R. V. Pyle, and J. M. Wilcox, 1959, Experimental Generation of Plasma
Alfvdn Waves, Phys. Rev, Lett,, 2: 383.
Brown, L G., 1965, Some Experimental Observations of the Attenuation of Alfvdn Waves in a
Laboratory Plasma, Aust. J. Phys., 18: 437.

REFEKENCES

Herlofson, N., 1950, Magnetohydrodynamic Waves in a Compressible Fluid Conductor, Nature,

168: 1020.
Jephcott, D. F., 1959, Alfv6n Waves in a Gas Discharge, Nature, 183: 1652.
and P. M. Stocker, 1962, Hydromagnetic Waves in a Cylindrical Plasma,J. Fluid Mech., l3:

-,

587.

Lehnert, B., 1954, Magnetohydrodynamic Waves in Liquid Sodium, Phys. Rev.,94: 815.
McPherson, D. A., and D. C. Pridmore-Brown, 1966, Density Gradient Effects on Alfv6n Wave
Propagation in a Cylindrical Plasma, Phys. Fluids,9: 2033,
Newcomb, W. A., 1957, The Hydromagnetic Wave Guide, in Magnetohydrodynamrcs, Rolf K. M.
Landshoff (Ed.), Stanford University Press.
Rukhadze, A. A,, and V. P. Silin, 1969, Kinetic Theory of Drift Dissipative Instabilities of a
Plasma, Sov. Phys.-Usp. (Engl. Transl.), 11: 659.
St6fant, R. J., 1970, Alfvdn Wave Damping from Fluids Gyroradius Coupling to the lon Acoustic
Mode, Phys. Fluids, 13: 440,
Stepanov, K, N., 1958, Kinetic Theory of Magnetohydrodynamic Waves, Sov. Phys.-JETP (Engl.
Transl.), 34:892.
Uberoi, C.,1976, Low Frequency Waves in a Rotating Magnetoplasma, Phys. Fluids, 19:921.
Wilccx, J. M., F. I. Boley, and A. W, DeSilva, 1960, Experimental Study of Alfv6n Wave Properties,
Phys. Fluids, 3: 15.
A.W. DeSilva, and W. S. Cooper III, 1961, Experiments on Alfvdn Wave Propagation,Phys.
-, Fluids, 4: 1506.
Woods, L. C., 1962, Hydromagnetic Waves in a Cylindrical Plasma, J. Fluid Mech., 13: 570.

General Beferences
Astrdm, E., 1950, Magnetohydrodynamic Waves in a Plasma,Nature, 165: 1019.
1961, On Waves in an Ionized Gas,Ark. Fys., 2: 443.
Bellyustin, N, S., and V. V. Tamoikin,I973, Diffraction of Alfvdn in aMagnetoactive Plasma,Izv.
-,
Vuz Rodiofiz., 16: 1467-1474.
Cross, R. C., and J. A. Lehane, 1967, Propagation of Compressional and Torsional Alfvdn Waves
Under ldentical Plasma Conditions, Nucl. Fusion, 7: 219.
Chen, F. F., l9'l4,Introduction to Plasma Physics, Plenum Press, New York.
Daily, W.D., 1973, Alfv6n Wave Refraction by Interplanetary Inhomogeneities, J. Geophys. Res.,

78:2043-2053.

III, and F. I. Boley, L962, Experiments on Alfvdn Wave

Propagation, in Magnetohydrodynamics, Proceedings of the Fourth Biennial Gas Dynamics
Symposium, Evanston, Ill., Aug. 22-24, 1961, p. 141, A. B. Cambel, T. P. Anderson, and
M. M. Slawsky (Eds.), Northwestern University Press, Evanston, Ill.
Fejer, J. A., and J. R. Kan, 1969, A Guiding Centre Vlasov Equation and Its Application to Alfv6n
Waves, J. Plasmo Phys., 3: 331-351,
Hollweg, J. V., 1975, Alfvdn Wave Refraction in High-Speed Solar Winds Streams, J. Geophys,
Res., 80: 908-916.
Knox, S. O., F. J. Paolini, and M. Kristiansen, 1975, Helical Antenna for Exciting Azimuthally
Asymmetric Alfv6n Waves, J. Appl, Phys., 46:2516-2520,
Kotik, D. A., 1970, Refraction and Reflection of an Alfv6n Wave at a Plane Boundary with
Allowance for Its Transformation into a Magnetoacoustic Wave, Geomagn. Aeron., 10:
779-782.
Lighthill, M. J., 1960, Studies on Magnetohydrodynamic Waves and Other Anisotropic Wave
Motions, Philos, Trans, R. Soc, London, Ser, A, 252: 397 ,
Muller, G., I974, Experimental Study of Torsional Alfvdn Waves in a Cylindrical Partially Ionized
Magnetoplasma, Plasma Phy s,, 76 : 8 13-822,
DeSilva, W., J. M. Wilcox, W. S. Cooper

31

32

ALFVEN WAVES IN PLASMAS

Murphy, G. L., 1975, Relativistic Effects in Magnetohydrodynamic Wave Propagation, Pltys.
Fluids, 18: ll2-114.
Paoloni, F. J., 1973, Non-Axisymmetric Alfvdn Wave Propagation in an Argon Plasma, Plasma
Phys., 15: 475481.
1975 , Boundary Effects on M=0, +OR- I Alfven Waves in a Cylindrical, Collisionless Plasma,
-, Phys. Fluids, 18: 640-644.
Rauchle, E., and P. G. Schuller, 1975, Reflection Phenomena of Torsional Alfven Waves in
Inhomogeneous Magnetic Fields, Z. Naturforsch., o, 30: I594-1599.
Sobolev, S. V., 1975, Influence of Pressure Anisotropy of the Near-Solar Plasma on the
Propagation of Alfvn Waves, Astron. Zh., 52: 346-350.
Spitzer, L., 1956, Physics of Fully lonized Goses,Interscience Publishers, Inc,, New York.
Stix, T. H., 1962, The Theory of Plasma Waves, McGraw-Hill Book Company, New York.
Uberoi, C., and C. Devanathan, 1964, Small Amplitude Hydromagnetic Waves in Compressible
Medium with Hall Current, Proc. Ind. Acad. Sci., LX: 42.
Venkataraman, N, S., and W. A. Gustafson, 1973, Alfv6n Waves Associated with Long Cylindrical
Satellites, J. Astronout. Sci., 2l: 99-ll7.

SPECIRAI. ANAIYS'S OF IDEAL

A[FYEN WAVE
'N 'NHOMOGENEOUS
MAGNEI'C FIEIDS

The propagation and properties of the ideal Alfven wave in the direction of a constant
uniform magnetic field are discussed in Chaps. 1 and 2.In many cases, however" the
magnetic field, though constant in time, is not uniform, and the magnetic lines of
force are, in general, not straight lines. The question then naturally arises as to what
type of solution the hydromagnetic-wave equation will admit. Will it have well-defined
discrete modes with a dispersion relation that gives rise to wave propagation analogous
to the Alfven wave? These questions are discussed in this chapter for a class of
magnetic fields for which the strength of the magnetic field varies along the direction
perpendicular to the plane in which the field lines lie. The choice of this class of
magnetic fields derives from the fact that a theoretical study of propagation of Alfven
waves in such fields has led to very striking and important results. The local variation
of Alfven-wave velocity gives rise to a singularity in the wave equation at the point
where the wave phase velocity equals the local Alfven velocity, which, in turn, leads to
a continuum spectrum of Alfvdn-wave frequencies.

Such a study was initiated by Gajewski and Winterberg (1961). Actually, the
equation arrived at by Gajewski and Winterberg did not show any singular behavior,
because these investigators neglected the effect of pressure perturbations. The basic
characteristic of the ideal Alfven waves, as noted in Chap. l, is that the total pressure
in the fluid, defined as the sum of the fluid and magnetic pressures, remains constant
during the passage of the wave as a consequence of the incompressibility condition.
33

34

IDEAL ALFVdN WAVE IN INHOMOGENEOUS MAGNETIC FIELDS

For inhomogeneous medium, however, the total pressure, in general, couples with the

of the motion, and the assumption of

neglect of pressure perturbations

(1964)
problem,
the
including the pressure
considered
becomes invalid. Uberoi
perturbations, and found that a singularity arises in the wave equation. She showed
that there was no possibility of any dispersion relation. Pridmore-Brown (1966)
discussed various features of the singular behavior of the wave equation but did not
arrive at the continuum spectrum of Alfven-wave frequencies. It is interesting that
Grad (1969), in his popular review, mentioned the possibility of the existence of a
continuum spectrum of Alfven waves. Uberoi (1972) noted that the Alfven-wave
equation exhibits rather complete similarity to that governing high-frequency
electrostatic oscillations of a cold plasma of nonuniform density. The latter equation
had been rigorously analyzed by Barston Q96$. This similarity made it possible to
exploit and to extend the generality, rigor, and inclusiveness of Barston's treatment to
dynamics

the case of Alfvdn waves.

Interestingly, as mentioned by Uberoi (1974), the wave equations obtained in the
study of the following four problems associated with inhomogeneous flows of liquids
and plasmas, i.e., the two-dimensional shear parallel flow of inviscid incompressible
fluid (the Rayleigh problem), the two-dimensional motions of low-density crossedfield electron beams, the electrostatic oscillations of inhomogeneous plasma, and
Alfven-wave propagation in inhomogeneous magnetic fields, show complete similarity
to each other. The similarity between the first two problems was pointed out by
Briggs, Daugherty, ffid Levy (1970). We note, however, that this singular behavior of

the Alfven wave originates from the assumption that exists behind the ideal
magnetohydrodynamics, e.g., the zero I-armor radius and/or zero dissipation approximations. It will be shown later that the singularity can be eliminated by relaxing these
approximations.

THE UUAVE EOUATION

The variation

in the static magnetic {ield is assumed to be perpendicular to the

plane in which the field lines lie; therefore

Bo(x) = Bov(x)

i + nor(x) i

This is consistent with the initial equilibrium state of the system:

Yo=0
Po = constant
lBn l2

Po+-=constant
ztrto

(3.1)

35

THE WAVE EQUATION

We have considered the density to be uniform so that the effect of varying

magnetic-field strength on the frequency spectrum associated with the Alfven mode of
propagation can be more easily understood. As shown in Chap. 4, for incompressible

the wave equation is similar either when only the magnetic-field

is considered or when both the density and magnetic-field variations in the
direction perpendicular to the magnetic field are considered.
For the initial magnetic-field configuration given by Eq.3.1, the ideal MHD
perturbations,

variation

equations for incompressible fluid, after linearization, combine to give

*=
dt'

o
o

d-

d
y
y

of

al

I (Bo.v)rn
-1v

FoQo

Bo

(r- -;)

'B\

(32)

As shown in this equation, the total pressure p = [p + (Bo ' B/po)J is,'in general,
coupied with the dynamics of the motion unless the medium is homogeneous.
If the medium is homogeneous, we can take the divergence of Eq.3.2and obtain
V2(ap/at) = 0, whose only solution vanishing outside a bounded region is Dp/Dt =
constant, which makes fl = constant. This then glves the ordinary Alfven wave
satisfying the equation

gv=O
where the

local Alfven-wave operator , A, is given by

i-

g -- ttopo

se

he

-l)

In the present case,

-(Bo

'v)'

(3.3)

is not constant. We take the x component of the curl of Eq.3.2,

A@xv)*=0
where

(V x v)* = (dvrldyl

- (avrlAz).

from F,q. 3.2 and if V. v = 0, the final equation,

v*, the velocity component along the direction in which the magnetic field

If vy, vr,
satisfied by

(3.4)

varies, can be

and

are eliminated

put in the form

"{ffi
This equation is

(Bo
Ito

9o

'vl']

v".)

=o

(3.s)

to be solved subject to the boundary condition that v* -+ 0 as

lxl + o".

Equations 3.4 and 3.5 govern the two modes that can propagate in the
inhomogeneous fluid. In general, the individual velocity components are coupled
together by the incompressibility condition, V ' v = 0. In one particular case, there is

IDEAL ALFVy'N WAVE IN INHOMOGENEOUS MAGNETIC FIELDS

36

no coupling, namely, when there is no dependence on z. Then v, satisfies CIvz = 0, i.e.,

it propagates at local Alfven speed and is uncoupled from v*. An even simpler case is
one in which the coupling has two dimensions, i.e., the motion is confined to the x-y
plane. Then F,q.3.4 drops out altogether, and v" and vy are conveniently computed

from a single-stream function that satisfies Eq. 3.5. In general, however, in any
arbitrary velocity disturbance, there will be a part, namely, the component of the
vorticity, which will propagate in one dimension along the field lines at the local
Alfven speed. Other components of the disturbance are coupled to v*, which satisfies
the more-complicated three-dimensional equation (Eq. 3.5). Thus we can expect that
at sufficiently large distances the latter components will die out and only (V x v)* will
survive.
We now turn to a more-detailed analysis of Eq. 3.5. For the Fourier analysis of Eq.
3.5, we take the time and space deperrdence as exp [i(-c.rt+ kyy + krz)]. The
equation satisfied by U1-(x ,ky ,kz,c.r), the Fourier transform of v*, is found to be

{r-'
a*'1

-<^,ltx)J

qF} - k'uk- l.-'

-a,f,{x)l = o
ox )

(3.6)

where k2 = k? + kZ

[Bo(*)' k]' lQtopo)

:uf;1*; t' ,or' 0, with va(x) = lBo llltrpo)'h,the local Alfven speed
0(x) = the angle between the direction of magnetic field and wave normal

@1(x) =

direction
Equation 3.6 can be written
d

a;
where

(*) = o|(x) -

as

[ougts] - *',

o)2,

--

r(x) up-(x) = o

< x < , and g(x) = k2, the boundary condition

Up-(x) +

lxl

--

(3.7)

being

(3.8)

SOLUTION OF THE WAVE EOUATION

We consider the magnetic field, Bo(x), to be everywhere continuous and constant

for all sufficiently large lxl so that the contributions to the discrete spectrum of the
regions of constant Bs are taken into account. Therefore the functions <.rl(x) in
F,q.3.7 can be considered as real valued, continuous, nowhere constant in some finite
interval, I, and constant outside this interval. For 8q.3.7 to have nontrivial solutions
in this interval satisfying boundary conditions 3.8, a2 must be real and satisfy the
inequality [c'l|(*)]min (<.r2 ^( [r1(*)lmax. This then defines the continuous
spectrum S = { col [c^r|(x)] min ( c^r2 ( trtr (")] *"* l] of the problem.

SOLUTION OF THE WAVE EQUATION

We assume that f(x) does not vanish anywhere. Multiplying Eq.3.7 with
integrating from -- to *- with the boundary conditions, we get

,[-0.,(l+l'
If

c.r

is complex so that

nt

he

in

te

NS

he

US

+ k2

tuu- t) dx=0

Uf

and

(3.e)

: (rft. + ic^rf , the real and imaginary parts of Eq. 3.9 give
a

<,.r2

rn(x)(l+f '*u, rur- r) dx=0

(3.10)

and

r r,(")(l*l'*u,

rur- ,)

dx=0

(3.r

l)

From Eqs.3.10 and 3.11, we can say that, if f1(x)*0 and fn(x)*0, then
d(x) = 0 is the only solution. Therefore, for a nontrivial 0(*), f1(x) equals 0 (i.e., <^r2
should be real), and fp(x) equals 0 (i.e., c.r2 should lie in the spectrum S). This implies
that a point x6 existS in the interval I so that a)2 = c^rtr(xo).
In Eq. 3.7,the points X = Xo for which f(xo) = 0 [i.e., lva(x) cos 0(x)l = lor/kl] 4p
the singular points of the equation, and thus the eigenfunctions Ut.(x) may not be
well behaved there. We shall now investigate the behavior of the solutions of this
equation near such a singular point.
lrt f(x) and g(x) be analytic functions in a neighborhood of xs so thaf

37

f(x)

fn(x

n=I

x6)n

fn(x)*0 for n 2 l. Then, fle&r X6, which is a regular singularity, the general
solution of Eq. 3.7 is of the form

where
series

Uru,(x) = A'1 W(x) + 81 [W(x) ln lx

= 42 W(x) + B, [W(x)

where

Ar, Az, Br,

ln lx -

xo I + (x

xo) V(x)J

(x > xo)

(x

xs) V(x)]

(x < xo)

xo | +

(3.r2)

and B2 are constants, w(x) and V(x) are analytical in

neighborhood of X6, &fld W(xo) = l. Thus the general solution of Eq.3.7 diverges at
the point X = Xo, the divergence being logarithmic.
It is easily shown that these singularities will appear in solutions that obey the
boundary conditions 3.8. Hence, in general, the mode solutions of Eq. 3.7 arc singular.
However, a well-behaved solution of the original boundary-value problem can be
constructed in the realm of generalized functions by making an integral superposition

IDEAL ALFV/N WAVE IN INHOMOGENEOUS MAGNETIC F'IELDS

38

of

over the whole spectrum S

At(<,r) so that

these eigenfunctions with some expansion coefficient

Us(x,t) =

l, ln(.)

Ur-(x) r-ic';t6-

(3. l 3)

We shall now try to determine the arbitrary constants Ar , Az, Br , 4fld Bz by

applying certain matching relations that must be satisfied across the singular point
X = X0. Constants A1 and A2 are arbitrary, whereas 81 equals Bz . To prove this, we
define

Ur = w(x) H(x -

6)

uz = B1 [w(x) ln lx Ua = Bz [W(x) ln lx
U+ = U2 H(x

xo I +

(x

- xo) v(x)]

- xo I + (x - xo) V(x)]

xo) + UrH(xs

- x)

where H(x) is the Heaviside function. Then, from Eq. 3.12, we have

Ur-(x) = A2 W(x) + (At -

A2)

Ut + Uo

(3.14)

It is then sufficient to show that

* ('#) -e(x)r(x)u1 =o

(3. 1s )

and
d

d"

(t*)

-e(x)r(x)ua

= Br

- u, ry6(x -

xe)

Note that

f(x)

x fn(x-xe)n

n=l

fr+0

f(xe)=0

w(xe)=l ry=D(x-xo)
and,

if y(x)

is continuous at x6, that

y(x) 6(x

xo) = y(xo) 6(x

xs)

(3.16)

SOLUTION OF THE WAVE EQUATION

39

We shall first provg that A1 and A,2 are arbitrary. We have

*=#","-xo)+D(x-xo)
and thus

r# r# H(x - xs)
=

Therefore

* ('*)

= H(x

- xo)*

('#). r$ a1* - xo)

=H(x_xo)*('#)
and so we obtain

* t*)

-Brur

=H(x-xo)[*

(#)

-u*]=o

it is demonstrated that A,1 and A,2 are arbitrary. Let us now show that
Br = Bz. For this we define a function G(x) as follows:

Therefore

G(x)=(.)*
(*) du"
d*

(x>xo)

(*.

*o)

For very small values of e , e > 0, we have

ry= [ry,n* H9* *9+

ry= [ryrn -ry- dv$'.
.

B'

u,

Since

f(x)= X f,,(x-xs)n
n=l

u)

v(xe

.,)]

v(xe

_.)]

IDEAL ALFVI]N WAVE IN INHOMOGENEOUS MAGNETIC ITIELDS

40
and

f,' =df!*o)*o
clx
we find

= Brr, w(xo) = B,f,

Y
-o%3=Bzr,
G(xo -) =
l,s r(*o

G(xs +) = j,3x r(*o

and thus

G(xo +)

G(*o

-) = (B, - Bz)

ry

whichshowsthatBr=Bz.
The fact that the coefficients A1 and Az are independent across a singularity
precludes any functional dependence of c^r on k; i.e., given any fixed value of k, c*.r is
free to range over the continuous spectrum S, and thus our medium does not give rise
to dispersion relations. Well-behaved solutions to Eq. 3.5 can be obtained by
integrating their singular modes over the spectrum S. The dominant terms of the
asymptotic expansion for u1(x,t), as estimated by Barston (1964),arc
U1(x,t) =

t.'o6(x)

oi

Thus each infinitely thin fluid layer oscillated with the Alfven-wave frequency c^,:a(x)
and the oscillations are damped as the inverse of time.
Hence, if c^14(x) is continuous and nowhere constant in an interval I, the MHD
plasma admits of only singular modes and an entirely continuous spectrum. No
dispersion relation exists. The spectrum consists of values of c',r that satisfy the
equation e)2 = o'A(x) for some xI. We nclte that the anisotropy induced by

themagneticrie^ta,whichcauSesthebandwidth{I,i(*)]mincos20(X)<

c.,'( [.L(*)]rnu" cos2 g(x)] of the continuous spectrum to continuously change

with g, cannot be considered as a dispersion in the classical sense (see Fig. 3.1).
What about the discrete spectrum? Suppose c.,.r2a(x) assumes constaut values <-,-rf,
anda2s2 oneithersideof theopenintervall.Inthiscase,tc^141 andtl'o;z willform
the entire discrete spectrum of the system. If, however, jump discontinuities are
introduced in the function oa(x) in the interval I, so that oa(x) jumps discontinuously from a value crr41 to a value c^lAi+t (><oal upon crossing a surface Si [i.e',
planes parallel to(y-z) plane] , and if (oa1,c.rAi+1) and the range of r.ra(x) have no
1

CONCLUDING REMARKS

41

-Y.

;o
f
E
F

LlJ
CL

U'

[vo(x)i.;n

0, radians

Fig.3.l Distribution of the spectrum of lo'/kl

sketched as a function of O, the anglebetween the

of magnetic field. IAdapted from C. Uberoi, Alfvdn Waves
in Inhomogeneous Magnetic-Fields, Phys. Fluids,15: 1674 (I972).1

wave normal direction and the direction

point in common, then we can get well-behaved modes, the number being equal to the
number of discontinuities, which will also form part of the discrete spectrum. The
well-behaved modes arising owing to step discontinuity in the function <rra(x) are the
well-known Alfvdn surface modes; the existence and properties of these are discussed
in Chap. 4.

CONCLUOING REMARKS
Knowledge of the existence of a continuum spectrum for Alfven waves plays an
important role in practical problems both in space plasmas, where inhomogeneity is
due both to the density variation and to structured magnetic fields, and in laboratory
the density is nonuniform.
space plasmas this knowledge has led

plasmas, where

In

to an understanding of the phenomenon

of magnetic pulsations in the ultra-low-frequency ranges (Lanzerotti et al., 1973 Chen
and Hasegawa,l974).
A practical method for heating plasmas in fusion experiments by creating
conditions for spatial resonance of the Alfvdn wave has been proposed. This aspect is
discussed in detail in Chap. 7.

42

IDEAL ALFVIIN wAVE IN INHOMOGENEOUS MAGNETIC FIELDS

The understanding of the spectral theory of magnetohydrodynamics has led to a

new approach in the study of MHD stability theory. Relevant early references are Grad
(1973),Appert, Gruber, and Vaclavik (1974), attd Goedbloed (1975).
The inhomogeneous nature of solar magnetic fields has been firmly established by
observations (e.g., Harvey, l97l). We believe that the theory of Alfven waves in
structured magnetic fields presented in this ohapter will play an irnportant role in
delimiting various effects in solar physics.

REFERENCES

Text Gitations
Appert, K., R. Gruber, and J. Vaclavik, 1974, Continuous

Spectra

of a Cylindrical Magneto-

hydrodynamic Equilibium, Phys. Fluids, 1'l: 1471.

Barston, E. M., 1964, Electrostatic Oscillations in Inhomogeneous Cold Plasmas, Ann. Phys.

(N. Y,),29:282.
Briggs, R. J., J. D. Daugherty, and R. H. Levy, 1970, Role of Landau Damping in Crossed-Field
Electron Beams and Inviscid Shear Flow, Phys. Fluids, 13: 421.
Chen, L., and A. Hasegawa, 1974, A Theory of Long-Period Magnetic Pulsations: Steady State
Excitation of Field Line Resonance, J. Geophys. Res.,79: 1024; Impulse Excitation of Surface
Eigenmode, J. Geophys. Res.,'19 : 1033.
Gajewski, R., and F. Winterberg, I96t, Alfven Waves in lnhomogeneous Magnetic Fields, Report
DI{2-0111, Boeing Scientific Research Laboratories;later published inAnn. Phys. (N. Y.),32
348 (196s).
Goedbloed, J. P., 1975, Spectrum of Ideal Magnetohydrodynamics of Axially Symmetric Toroidal
Systems, Phys. Fluids, 18: 1258.
Grad, H., t969, Frontiers oi Physics Today, Plasma Phys. Today,22(12):34.
1973, Magnetofluid-Dynamic Spectrum and Low Shear Stability, Proc. Natl. Acad. Sci.,

-,

U. S.

A.,70:3277.

., I97 | , Publ. Astron, Soc, Pac. , 83: 539 .

Lanzerotti, L. J., H. Fukunishi, A. Hasegawa, and L. Chen, 1973, Excitation of the Plasmapause at
Ultra Low F-requencies, Phys. Rev. Lett.,3l:- 624.
Pridmore-Brown, D.C., 1966, Alfvdn in a Stratified Incompressible Fluid, Phys. Fluids,9: 1220.
Uberoi, C., 1964, Hydromagnetic Waves in Ideally Conducting Uniform Medium with Inhomogeneous Magnetic Field of Constant Direction, Indian J. Pure Appl. Phys.,2: 133.
Alfvdn Waves in Inhomogeneous Magnetic Fields, Phys. Fluids, l5(9): 1673-1675.
-,1972, On the Similarities Detected in lUaves in Plasmas ond lUoves in Incompressible Fluids,
Applied Mathematics Seminars, Indian Institute of Science, Bangalore.
-,19'14,
Halvey,

General References
Deutsch, R. V., and F. Fornwald, 1958, Propagation of Alfvdn Waves in Non-Homogeneous Plasma
in the Presence of the Homogeneous Grayitational Field, Rev. Roumoine Phys., l3: 895-900.
and L. Ralan, 1968, The Propagation of Alfvdn Waves in Nonhomogeneous Systems in
-, Relativistic Magnetohydrodynamics, Bull. Inst. Politeh. IASI, 14 : 17 5-17 8.

Goedbloed, J. P., 1975, Spectrum of Ideal Magnetohydrodynamics of Axisymmetric Toroidal

Systems, Phys. Fluids, 18: 1258-t268.
and P. H. Sakanaka, 1974, New Approach to Magnetohydrodynamic Stability: A Practical
-, Stability Concept, Phys. Fluids, 16: 908.

REFERENCES

A., and Liu Chen, 1974, Theory of Magnetic Pulsations, Space Sci. Rev.,16:347-359.
Infeld, 8., 1969, Solutions of the Linearized Equations of Magnetohydrodynamics in Nonhomogeneous Magnetic Fields, Phys. Fluids,12: 7845-1848.
Motrow, R., and M. H. Brennan, 1974, The Propagation of Compressional Alfvdn Waves in
Nonuniform Plasmas, Aust. J. Phys.,27: 181-194.
Muller, G., R. Rauchle, and P.G. Schuller, TgT3,Investigation of Plasma-Magnetic Waves in an

Hasegawa,

Inhomogeneous Cylindrical Plasma, Plosma Phys., 15: 925-928.

D. G., and Y. C. Ngan, 1975, Warm-Plasma Effects on Fast-Alfvdn-Wave Cavity
Resonances, Phys, Rev. Lett., 35: 51 7 -520.
Tataronis, J. A., 1975, Energy Absorption in the Continuous Spectrum of Ideal MHD, J. Plasma
Phys., l3:87-105.

Swanson,

43

tI

RESONANT ABSORPI'ON AND I{'ODE

coNvERs,oN oF awviN suRFAcE

wAyEs,N NoNUNtFoR/lt Pll.st{tAs

The study of Alfven'wave propagation in nonuniform cylindrical plasmas considered in

Chap.2 was limited to low-beta plasmas with the assumption that the pressure
perturbations of the plasma could be neglected. This assumption does not hold good
for high-beta plasmas. Moreover, as noted in Chap. 3, the coupling of pressure
perturbations to the dynamics of the motion can completely change the properties of
Alfudn-wave propagation in inhomogeneous media. The Alfudn-wave equation shows a
singular behavior at the point where the externally applied frequency, o), matches the
local Alfven (shear Alfvdn) frequency, krr(*) ua(x), where k1(x) is the local wave
number parallel to the ambient magnetic field. Owing to the presence of this
singularity, a nonuniform plasma with continuously varying density is characterized by
a continuum spectrum in the hydromagnetic frequency regime. On the other hand, if
jump discontinuities are introduced in the density profile, the system admits of
discrete spectrum. The incompressible perturbations in the hydromagnetic regime give
rise to Alfven surface waves propagating along the discontinuous plane with definite
frequencies. Most of the instabilities that originate from the nonuniformity of a
plasma are, in fact, the instabilities of a surface wave (Hasegawa, 1975). (The
instabilities associated with Alfuen surface waves are discussed in Chap. 5.)
The earlier part of this chapter is concerned with the question of how to unite the
two dissimilar pictures presented above; i.e., we ask what happens to the surface wave
when the discontinuous density profile is smoothed and whether there are some

INTRODUCTION

collective modes of propagation even in the case of continuous density. We shall,

therefore, treat the Alfven-wave equation (which is similar to Eq. 3.5 in Chap. 3) afresh
by using the Laplace transform technique to solve the corresponding initial-value
problem. Once again, using the analog between the Alfven-wave equation and the
plasma-wave equation, we follow closely the analysis given by Sedkidek (1971) for the
study of electrostatic oscillations of cold inhomogeneous plasma. This approach,
though equivalent to the normal-mode analysis, is known to provide results in a form
suitable for finding the whole asymptotic time expansion (not just the dominant
terms) of the solution simply by studying the singularities of its Laplace transform in

the complex frequency plane. By constructing the Green function of the wave
equation, we find that there are branch point singularities on the real axis of the
complex frequency plane which correspond to the continuous spectrum and which,
asymptotically, give rise to noncollective oscillations with position-dependent frequency and damping proportional to negative powers of time. In addition, there are an
infinite number of new singularities (simple poles) of the analytic continuation of the
Green function into the lower half of the complex frequency plane with positionindependent frequency so that they represent exponentialiy damped, collective
(surface eigenmodes) modes of wave propagation. Thus the effect of a continuous
density profile is to introduce damping to the originally undamped surface modes of a
discontinuous plasma. This damping arises owing to the presence of the continuous
frequency spectrum. The normal modes that make up the continuum can phase mix in
time, and this mixing leads to decay of macroscopic variables. The damping of Alfven
surface waves is thus similar to the well-known I.andau damping of plasma oscillations
in warm plasma. The damping of surface waves has been observed experimentally
(Grossmann, Kaufmann, and Neuhauser, 1973) and in a magnetohydrodynamic
computer calculation (Pritchett and Dawson, 1978; Balet et a1., 1979; Appert et al.,
1e80).

Another aspect of a continuum, related to the absorption of surface waves, is that

the system can irreversibly absorb energy when it is externally driven. The energy
absorption rate in the presence of an external driving source, which is represented by a
sheet current located in the vacuum, is discussed later in this chapter. The absorption
is a consequence of the local Alfvdn resonance which occurs when the externally
applied frequency matches the frequency in the continuum. A characteristic of this
absorption is that it occurs locally in the space in which the continuum is defined. The
relationship between a continuous spectrum and energy absorption has been noted and
discussed extensively for zero-temperature plasma oscillation in a nonuniform plasma
(Baldwin and lgnat, 1969).
Extensive studles on linear transformation and absorption of electromagnetic
waves in nonuniform plasma have revealed the fact that collisionless resonant
absorption of waves in a plasma is connected with wave transformation (Golant and
Piliya, 1972).In the section Resonant Mode Conversion, we study the resonant mode
conversion of the Alfven surface waves (Hasegawa and Chen, 1976), which will enable
us to understand the mechanism by which the surface-wave energy is absorbed and
propagated in the plasma, away from the resonant region.

NI-PVEN SURFACE WAVES

46

IN NONUNIFORM PLASMAS

Fig.4.lDiscontinuityindensity(.-_)andsmoothvariation(---).

ALFVEN SUBFACE WAVE

I.etusconsideraplasmawithasharpdensitygradient(Fig.a.1).

n(x)=ns1 (x>0)
n(x) =

ne2

We shall also consider discontinuity

(x < 0)

in the magnetic field taken along

the

z direction:
Bo

=Bor2 (*>0)

Bo=Boz2 (x<0)
for the plasma' we -consider incompressible
+k"z)l' We seek for
perturbati,ons in the y-z pf.tt given by f(x)exp [i(-3t+kvy
decay away from the
Taking the ideal MHD model

amplitudes
the zurface-wave solution, ,u.h thut the *uut
interface x = 0'

FollowingtheprocedureinChap.3,wecanwritethebasicMHDequations,after
linearization, in the form

t*

'v )'.l,
ItoPo I'

(Bo

S'Y=Q

1a

(4.1)

^a,vP
(4.2)

47

ALFVb,N SURFACE WAVE

[p + (Bo ' B/po)] and pq = rilioo, with m1 the ion mass and
unperturbed plasma density. Eliminating p' from Eqs.4.1 and 4.2 gives
where F'=

V2v* = g

ns

the

(4.3)

which is to be satisfied in both region 1 and rcgSon2.

Therefore the equation governing the surface perturbations

is the

Laplace

equation, with the boundary conditions v* and p'continuous. From Eq.4.1, pcan be

written

as

F=#

(."0,-#J*

(4.4)

=kr, kr =ky, and k2 -kfr'; *k]. fhe boundary conditions, therefore,

continuity of v,. and e(bv*/0x), with = Q lk2)l-' po - (B3kfr/tro)l .

where ktt

The solution of Eq. 4.3 for surface waves is given

vx

lklx

are

as

Ae-

Be+tktx (x < 0)

(x > o)
(4.s)

Applying the boundary conditions, we get the following dispersion relation for the
surface waves:

,,,(# -vu,)*oo,(fr "i,) =

@6)

If we consider a plasma-vacuum interface, 8q.4.6, with p62 =0, gives the

following dispersion relation for the Alfven surface waves (Kruskal and Schwarzschild,
re54):

G)As=

In the particular

case

with

Bo

/sl, * nt, \%

\ff/

krr

r = 862 , this frequency

becomes (Chen and Hasegawa,

te74)
&)As =

(2)h v6k,,

We note that the surface-wave frequency in this case is higher than the Alfven-wave

frequency (the bulk resonant frequency), which is in contrast to the electrostatic

surface wave where the surface-wave frequency is given by <,r. = upl(2)'h

ALFVEN SURFACE WAVES IN NONUNIFORM PLASMAS

48

3
o
J
UJ

u-

cs
F

2.O

t.rJ

z
(t

4'o

o.orro oJ*or,.r

,n,

12'o

Fig 4.2 Ratio of magnetic fields and plasna densities on either side of the plane of surface-wave
propagation plotted against each other for various values of the ratio of surface-wave velocity, vp5,
to the Alfvdn-wave velocity, v4r.

In

general, Eq. 4.6

will

have real roots only when the surface phase velocity

satisfies the inequality

vllcos2 U.#{Y2Azcos20
where

0 is the

angle between

and Bo. This condition, which was discussed in

Chap. 3, was obtained from Barston's theorem.

If we define \ = pozl ps1 and "y = BozlBo r , Eq. 4.6 can be written

" = l+^f
I +n

v6r

as

(4.7\

where vp, 3 c^rl(k1;va r ). Figure 4.2 gives an estimate of the surface parameters 4 and 7
for given ratios of surface-wave velocity, vps , to Alfven-wave velocity, vAr.

RESONANT ABSORPTION OF SURFACE WAVE

B

49

ESONANT ABSO BPTION

F SU

FAC E WAVE

Instead of having a sharp density discontinuity, we now consider the situation in

which the plasma density is smoothly varying in x. For this we consider a small
transition layer, -a < x < a, in which the density varies smoothly. The density profile,
as compared with that in the preceding section, then looks like the broken line in
Fig. 4.1.

To study the surface waves in region 0, we proceed as in Chap. 3 to obtain the

in which Bs = Bs2 is uniform and density, go = po(x), is continuously

wave equation,

varying with x,

,,d' _(Bo.v),1I =o
-v ' Jl
'i-lvvx/
a(tlpo(x)

(4.8)

Equation 4.8 shows the coupling of bulk Alfvdn waves with the surface Alfven waves.
the perturbations to be independent of the coordinate y, the Fourier

If we consider
analysis

with

respe

ct to z,
vx = ut(x

,i = I

U(x,z,t) ,+ikz 6t

yields the equation

*rh.o,

#-#l%*

- r.,[,.t.)S -#]ut=o

(4.e)

This equation is now to be solved for U1(x,t), subject to the boundary condition that
U1(x,t) vanishes at infinity and to the initial conditions U1(x,0) and its time derivative
Up(x,O). The initial conditions for the velocity themselves must, of course, satisfy the
appropriate boundary conditions and must be physically acceptable, i.e., must be
sufficiently smooth and square integrable in the infinite interval.
The Laplace transform of up(x,t) is defined as usual by the formula

Ur-(x)
where

I; Uk(x,t) ,-ic^it 61

(4.10)

Imo > 0. The application of this transformation to Eq. 4.9

*{[r.o ),"

-#]*]

where S(c^lf) corresponds to the

We shall now solve Eq. 4.1

the homogeneous
Green's

- r'

[o.r*

),,

-#]

uk-

gives

= s(<o,x)

(4.1l)

initial conditions.

I with the aid of Green's functions. As will be noted,

equation 4.ll is the same as Eq.3.7 discussed in Chap.3. The

function G(x,sf,c^r) can be expressed in terms of two linearly independent

ALFVEN SURFACE WAVES

50

IN NONUNIFORM PLASMAS

solutions, Ut-(rl(x) and U1-(2)(x), of the homogeneous equation 4.11, which

satisfies the boundary conditions at x = -6 and at x = +"". The following formula
holds:
G(x,s;k,<,.r) =

J-t [Ur-(r)(x)

U1.,121(x) H(s

x)

* Uk-rel(x) Ut-(1y(s) H(x -

s)l

(4'r2)

where H(x) is the unit step function and J(k,c^r) is the so-called conjunct of the two
solutions Ut.,(r) and Uko,(z),

J(k,c^,) =

where e(x) = po(x) a)2

e(x){H#tur-(,,] - |uuHt" u*-,,,]]

-

(4.13)

(BAk' ltto).

A theorem exists [Theorem (3.2) in Friedman, 1956] which shows that J(k,c^r) is
independent of x and s.
To make the homogeneous equation 4.1 I solvable, we choose a monotonic density
profile composed of linear sections (Fig. a.1) so that
e(x) = er = <,s2ltogr
= r

- k'BA

o(i- t)

tto12)

(p,

( -

a, region 1)

(- u(x(

= r 1 = u2 ltopz

with A = (u2

(*

a, region

- k'BA (x2 a,region

0)

(4.14)

2)

- p).

With this density profile, Eq,, 4.11 assumes constant coefficients in regions I and 2
and becomes a modified Bessel equation in region 0 (the transition region).
The solutions U1.,(r;, bounded at x = -@, and U1-12y, bounded at x = , in the
three regions can be written as follows:

ut-("

lL';,,].,
=

Cr

81 Ks(zx)

(4.1s)

e-kx 1 Pr akx

and

ur-(2,:::;;::T;r*,
(4.16)
=

s-k(x+a)

RESONANT ABSORTTfION OF SURFACE WAVE

where zx = -tr(c,r'pop*
Bessel functions.

51

k2BA), ).=ka/A, and Iq(z) and Ks(z) are the modified

With the use of the matching conditions as given in the preceding section, at the
boundaries x = tn, the constants ,{1,2,8t,2,Ct,2, aLrtdD1,2 Citn be decided.
Regardless of whether the solutions in region l, 0, or 2 arc used, by substituting
the two solutions into Eq. 4.l3,we get
J(k,<^r) =

D(k,c,r) = flo(zz) + It (zz

(4.t7)

*trz2 D(k,a:)

)l [Ko(zr) + K, (zr )]

[Io

@')

Ir (zr )] [Ko@r)

K'

(zr)l

(4.18)

We shall now discuss the properties of D(k,c^r) as a function of c.r in great detail. As

will be evident, the study of its analyticity, the structure of its Riemann surface, and
its zeros and singularities will give the details of the continuum spectrum and the
discrete eigenvalues of the differential equation 4.11. Hence it is appropriate to call
the function D(k,<^r) a dispersion function.
The modified Bessel functions Is(z) and

11

(z) arc entire functions, and Ks(z) and

(z) can be decomposed into a logarithmic term plus an entire analytic function,
wo(z) and w1(z), respectively (Erdelyi and Bateman, 1953). Hence D(k,ol) can be
Kt

written as (SedleiU ek,

197

D(k,c^r)

l)

[Io(zt )

+ fls(zr)

on writing

h#

= llo(zr ) - I, (21)] [Is(zr) + Ir tdl

@A | ,2 =

Ir (zr )l

rr@)l

[*o,rr) -

[*0,r,

W,

d - *]

) + W, (', ) *

(4.re)

*l

ku.l t ,2 t we determine the multivaluednes of D(k,<,1) by the

logarithmic term

. 7.n

At</JAt

(v.J-|v.JLz

+ln
ln"" =-ln
21
U*v.)Az
@-aSA,

(4.20)

which shows that D(k,c.r) possesses four logarithmic branch points on the real axis:
-aAz,-@At,o)A1,afld QA2,, each of which coincides with a simple pole. If the
o-plane is cut along the intervals connecting the first pair and the second pair of these
points (Fig.4.3), the function ln (zrlzt) becomes a single-valued functionof co.These
cuts (or spectral cuts) are identical with the regions of the continuous spectrum of the
differential equation 4.1

l.

ALFVEN SURFACE WAVES IN NONUNIFORM PLASMAS

52

Fig 4.3 Branch cuts and leastdamped roots of the dispersion function D(k,<^:) on the n = -l
sheet

of its Riemann surface.

The function D(k,c^.r) has the same Riemann surface as the function In(z2lz1). All
the branches of D(k,o) may then be constructed simply by replacing ln(z2lz1) in
Eq. 4.19 with ln (z2lz1) + ZnnL The result is
Dn(k,<^r) = D6(k,c^.r) + 2nnrillo(zt )

Ir (21)] Us(zr) + It

(zr)l

@.21)

We shall now investigate the zeros of D(k,<,1). The zeros lying on the principal
sheet of the Riemann surface areidenticaltothediscreteeigenvaluesof Eq.4.l1.The
zeros lying on the other sheets are usually referred to as "virtual" eigenvalues (because
no independent eigenfunctions correspond to them) and are associated with the
phenomenon of spectral concentration, whose physical interpretation is quite similar
to the theory of Landau damping in warm plasma oscillations.
With the aid of the Nyquist criterion, it can be shown that there are no zeros of
D(k,<.r) on the principal sheet. There are, however, zeros on the other sheets of the
Riemann surface. We shall calculate those which, at least for small values of the
parameter ka (i.e., the least-damped modes), lie nearest the real axis, which are the
ones associated with the phenomenon of spectral concentration. Considering first the
sheet 11 = - l, we expand D_ 1(k,c^l) in terms of the small parameter ka, retaining only
the terms of the zeroth and first order.

D-1(k,c.r)=-*{

h opr(az - c,rlt)

.ffi(r+ka)
pz(a:2

-r['' prks2 --

a'a,r)

rta.r)

r'']l

s22)

RESONANT ABSORPTION OF SURFACE WAVE

53

If ka = 0, i.e., if the density profile is discontinuous, the dispersion function D(k,o)

gives the dispersion relation representing surface modes, as derived in the preceding
section,
pzk

f -

r'Ar)

+ p,

(r' -.1r)

(4.23)

=0

The roots of this equation are purely real; therefore we can assume that the imaginary
part of the solution of Eq. 4.22 will be of the order ka. We take c^r = (rR * ic^rl, and
then, retaining only the terms of the zercth and first order in ka, we split F;q.4.22 into
real (R) and imaginary (I) parts, which gives the following two equations to determine
<^lp and ar1:
t-

Li

<^ltp

t'a'r)P'

(cR

- of^)e,l

,^,ffi) =o

(4.24)

and

We see from Eq. 4.25 that the effect of a continuous density profile (ka *0) is to
introduce damping to the originally undamped surface-wave modes of a discontinuous
(ka = 0) MHD plasma.
We denote the roots on n = -1 sheet as c^r(-). Since the dispersion relation is a
function of only a2,the point -@(-) is also a root. The roots on n= I sheet, <,r(+)
and -c,r(+), can be found in the same way.
Considering that the oscillation frequency ,[-) is given approximately by

8q.4.23, we have

,[-) o(2)Y'o^t#r;;
Substituting into Eq. 4.26 this value of c^.rf- ), we get the damping rate,

-9fl- o -n -,y^Pr - Pz
4 grtPz
t[-)
lf p2

4pr

(4.26)

,r(- ), ut
(4.27)

we have

,k) -(2)'hku.^, and + =-$t^

(.^/R

(4.28)

ALFVEN SURFACE WAVES IN NONUNIFORM PLASMAS

54

o [d ln ptdx]*io is the scale length of the

density variation. We note,

(i.e.,
the smaller ka),the weaker
therefore, that, the sharper the density discontinuity

Here a

the damping.

The complete solution

transform, which then gives

Eq. 4.9 will be given by performing inverse l-aplace

U(x,t) in terms of the Bromwich integral,

of

u1(x,t)

--*.,f

uu-t*; e-itrt

dc,r

(4.2e)

where the integration path C runs parallel to the real axis of the plane above all
singularities of Ur (*).
Without going into the complete details of finding the asymptotic behavior, we
give the results as obtained by Sedkilek (1971):
Ug(x,t) = A1(x;k)

t- I e-i<^'r a(x)t + A2 (x;k) 1- I .io4(x)t

+

Br(x*) e-i-(-)t 1Bz(x*).-i<^:(+)t

(4.30)

'4.25 correspond to the noncollective modes {the

The first two terms in Eq.
continuous spectrum S = [<^rl(c^r1)*in ( c,r2 ( (.'a)*"*l]] with asymptotically posi
tion-dependent frequenc/ <.,r4(x) and damping proportional to inverse power of time.

The second two terms, though negligible in the asymptotic expansion (Eq.4.30)'
represent the discrete spectrum corresponding to the only collective modes of
oscillations (the surface eigenmodes) with position-independent frequency and
exponential damping (resonant absorption). The damping can be rather weak if the
plasma parameters have sharp variations. The damped surface eigenmode is used by
Chen and Hasegawa (1974) to explain the impulseexcited damped magnetic
pulsations.

mechanism of the resonant

absorption, using kinetic theory. This absorption is a manifestation of the resonant
mode conversion of the surface wave to the kinetic Alfvenwave introduced inChap.2.

In the next section we elucidate the physical

RESONANT MOD E CONVERSIO N

understand the energy absorption mechanism, we shall study the linear

transformation of the Alfven surface wave to the kinetic Alfven wave in the

To

neighborhood of the resonance region (Hasegawa and Chen. 197 6).

The logarithmic singularity in the solution of the Alfvdn-wave equation arises
because the ideal MHD equations do not have any dissipative mechanism and/or
assume zero l-armor radius. This singularity is removed when the finite resistivity
effect or the finite Larmor radius correction is introduced into the ideal MHD
equations. The effect of resistivity on the Alfven surface wave has been considered by
Kappraff and Tataronis (1977) and Uberoi (1977). This treatment, however, is valid

RESONANT MODE CONVERSION

55

only when the plasma skin depth is shorter than the ion gyroradius. We note that finite
resistivity effects are smaller than kinetic effects by 0(z"1v2a lrou|), where z"g is the
electron-ion collision frequeoct, v1" is the electron thermal speed, and cr6 is either
the applied frequency or the eigenfrequency of the surface wave. For a typical hot
plasma, vl"lvz;=(pm1/m)V!, v"rSl0s/sec, and e)ox 106 rads/sec. Therefore the
skin depth is generally much larger than the ion gyroradius. Then the kinetic Alfven
wave excited at the singular surface can propagate for a considerable distance
measured by the size of the ion gyroradius, and the energy is dissipated more
uniformly in the plasma. This phenomenon cannot be seen by the one-fluid MHD
theory but can be studied only by the kinetic theory. We consider the unperturbed
distribution function for the ions to be given by

{or

= s (,.

#) do)(")

(4.3r)

where f(o)(v) is the Maxwell distribution with the ion itemperature Ti and g is
function representing the plasma nonuniformity in tJre x direction.
The linearized Vlasov equation for the perturbed distribution function of ions is

aq afi r
af,(o) e
df.
d*nax+"\(a*vx Bo).#**0x Bo).#

=o

(4.32)

The perturbed distribution function of electrons obeys the drift kinetic equation

af

Dt

'z

*i)

*.(*.",

#-du'#=o

The field equations are the same as those derived

i.e., the quasi-neutrality condition

and Ampere's

(4.33)

inChap.2 (Eqs.2.33 and2.34),

fli=tre

(4.34)

$vira - ,lt)= po$t1,, + J"r)

(4.3s)

law in the z direction

In these expressions the number density, n, and the current density, J,are given by the
distribution function, f,
nj = no-f

t du

(i = i,.)

(4.36)

F
ALFVEN SURFACE WAVES IN NONUNIFORM PLASMAS

56
and

J;

=qino/vfdv 0=i,r)

(4.37)

The set of equations from 4.31 to 4.37 describes the electromagnetic waves in an MHD
frequency range without restrictions on the size of the perpendicular wavelength (but
with a restriction of no compressional magnetic-field perturbation).
To study the resonant mode conversion of the Alfven wave, we can assume that
k' the wave number perpendicular to the magnetic field as well as to the density
giadient, is much smaller than p1 I and that the scale size of the density gradient is
much larger than the ion grroradius, and thus

*u

kv*p?

rc-t is the inhomogeneity scale. This assumption allows us to ignore the term
vrla"i in g in Eq. 4.31, which is equivalent to neglecting a drift wave.
In addition, we assume that the wavelength in the x direction near the mode

where

conversion point is small but larger than the ion gyroradius; thus we can expand the
wave equation in the power of p1 d/dx. We also assume vTe ) vA.
With these assumptions the Fourier amplitude of the number density and the
current-density perturbations are given by

,n,

6=&
,oli

( .;,?

#6#)

ski,

tre a'o"

T=
,
co vi"

4.*k"ee

(4.38)

(4.3e)

gV

ttoriz=+Ztr

(4.40)

and
(t)2no U)

FoJez=-c-vTe
;T Kz
,

In

Erlt

(4.4t)

these expressions the ion and the electron plasma frequencies, c,rpi and clpe,
respectively, are constants and refer t<l the values at the maximum plasma density,
where g(x) is normalized to unity.
To study the resonant mode conversion to the kinetic Alfven wave, we consider
only theregionneartheresonantpointX=X0,where a)2 =Vrt"[u'nlg(x=xo)] .Wecan
then further simplify the expressions by using the assumption of a low-beta plasma so

RESONANT MODE CONVERSION

that vf;

).3

coupling

57

T"/mi. This assumption eliminates the possibility of a simultaneous

acoustic wave. The wave equation can then be derived by

to the ion

eliminating r/ from Eqs. 4.34, 4.35, and 4.38 to 4.41,

ffii,,#.#ifi,'*)
.

(':g)

ffig

- 1) *-

.?

(*,-')] o
d=

(4

42)

where the Alfven speed, vf;, is that of the maximum density and g is normalized to
unity.
We can immediately notice that, if we put pi + 0 in this expression, the wave
equation reduces to

frlno#J-k]e(x)

Q=o

(4.43)

(.' lV3r"t) g - l. Equation 4.43 has frequency structure identical to the

MHD wave equation 4.ll.It is also noted that in a uniform plasma, g= 1,8q.4.42
gives two decoupled wave equations,
where e(x) =

(4.44)

Vtt"O=o
and

lo#.(&
where D2 =

lpl4)

(Tr/Ti)l

p?

.fn"

)]

quantity kv

p=o

(4.4s)

d/dx is used in Eq.4.45.

Equation 4.44 rcpresents a quasi-static electromagnetic perturbation (cutoff mode)

associated with an external source. In the absence of a source, this equation represents
a surface wave. Equation 4.45 is the wave equation for the bulk kinetic Alfven wave
(Eq.2.dA).
We can thus identify that Eq. 4.42 represents a coupling between a surface MHD
mode or an externally applied electromagnetic perturbation and the kinetic Alfven
wave. From Eq. 4.45 we can see that the kinetic Alfvdn wave propagates, after the
mode conversion, to the higher density side, where?,3rv2a(x).Prv|(x = *o).
To study the mode conversion, we must specify the actual density profile. Since
the wave after the mode conversion is expected to propagate on the higher density
side, the solution depends on whether the converted wave can propagate all the way
across the plasma column or whether it dissipates significantly before it reaches the
other side. We assume here that the plasma is semi-infinite in the x direction"

Ia

ALITVEN SURFACE WAVES IN NONUNIFORM PLASMAS

58

As a simple example, we take a linear profile for the plasma density so that

g(x)=rcx*a
where x is a normalized distance whose origin is located at the resonant point where

g(x=

0).'l$lu't)= l, or aollQ<lv2o)=1, with 0<a< l.

Equation4.42 is then

reduced near x rY 0 to a simple form

t #f,

rcxEx = Eo

(4.46)

o'=C.ryT)"

(4.47)

where

and
a6'

tr =ux

Ex

(4.48)

Eo is an integration constant representing a nominal value of E* at a large negative x

(E* associated with the external source field or surface vzave).
In the derivation we also assumed that l(dg/dx)/gl < l@Qldx)lQl;i.e.,the variation
wave amplitude is much faster than the variation of the density (W'K.B.

of

the

approximation). The wave equation of the type of Eq.4.46 has been studied
exlensively in relation to the mode conversion of the electromagnetic wave to the
Langmuir wave. It is well known that the solution can be expressed in terms of the
Airy functions.

If we introduce

a scale

length,

^=(+)-

(4.4e)

and use a normalized electric-field intensity,

E*=-

E*

(rcP)%

Eo

(4.s0)

the general solution is given by

E1 = gr

Ai(-xlA)

+ c2

Bi(-x/A) + G1(-xlA)

(4.s r)

RESONANT MODE COIWERSION

59

KrNETrc

alrv6ru

wAVE

a--*l

Fig 4.4 Schematic diagram of E* field of the kinetic Alfvdn wave excited at the AlfvCn tesonance
surface x = 0. Eo sltows the x component of the external field or the zurface Alfvdn wave. The
Alfvdn speed is assumed to decrease toward the right.

where

Ai and Bi are Airy functions and Gi is a function involving integrals of Ai and

81.

The integration constants c1 and c2 can be decided by applying appropriate

boundary conditions. For a semi-infinite plasma as considered here, they are decided
whether or not the energy source exists internally or externally to the plasma.
For the resonant absorption or the surface-wave damping as discussed in Chap.3
and earlier in this chapter, the suitable boundary condition for the kinetic Alfven wave
is to accept only the right-going wave (no reflection at x - -) and that which has no
divergence at x + -"". We can then find c2 = 0 and cr = i. The asymptotic solution for
lx/Al > I can then be written as
Ex =

- ffi(i)',.e{'F(;)* .

;]}.*

(4.s2)

F
ALFVEN SURFACE WAVES IN NONUNIFORM PLASMAS

60

forx>0,and
Eo

b-^Kx
=-

(4.s3)

for x < 0. The first term in Eq. 4.52 represents the kinetic Alfven wave and the second
term, aswell as the expression in Eq. 4.53, the source or surface-wave field.
As is expected, the kinetic Alfven wave propagates on the higher density side
the resonant point, as shown in Eq.4.52.In Eq.452the first few peak
amplitudgs of the kinetic Alfven wave after the mode conversion are given by

(x>0) of

Eo(*p)-%, with an effective wave number of (.rc1p2)'h, whereas away from the
resonant point, say at x N K- I , the amplitude and the wave number of the kinetic
Alfven wave become Eo(rcp)-'h and p-r, respectively. The qualitative feature of the
mode-converted kinetic Alfvdn wave is shown in Fig.4.4.
We can see that the solution for x < 0 is identical to that obtained under the ideal
MFID approximation. Hence the plasma impedance, and consequently the absorption
rate, seen from the external circuit remains unchanged from the previous MHD
calculation.
If the energy source exists inside the plasma, we must take the left-going wave also.
Such a problem is important in the analysis of the drift Alfven instability (Chap. 5).
If v1s < v4 at the resonance surfaco, Xs, the appropriate dispersion relation of the
kinetic Alfvdn wave is approximately given by
(,)2

=krrro( -

#-t

(2.44')

Hence, after the mode conversion, the kinetic wave propagates toward the lower den-

sity side.
REFERENCES

Text Gitations
Appert, K., et al., 1980, Resonant Absorption

of Alfvdn Waves in Fat Tori with Circular

Cross-Section, in hoceedings of the 2nd Joint Grenoble-Vsrenna International Symposium on
Heoting in Toroidal Plosmas, Como, Italy, Sept. 3-12,1980, International School of Plasma
Physics/Comissariat a I'Energie Atomique.
Baldwin, D. E., and D. W. Ignat, 1969, Resonant Absorption in Zero-Temperature Nonuniform
Plasma, Phys. Fluids, 12: 697.
Balet, B., et a1., 1979, Studies of Resonant Absorption of Alfvdn Waves by Means of an Evolution
Code, in Proceedings of the 9th European Conference on Controlled Fusion and Plasma
Physics, Oxford, England, Sept. 17 -21, 1979, p. 170, European Physical Society.
Chen, L., and A. Hasegawa, L974, A Theory of long Period Magnetic Pulsations. 2. Impulse
Excitation of Surface Eigenmode,J. Geophys. Res., 79: 1033.
Erdelyi, A., and H. Bateman, 1953, Higher Transcendental Functions, Vol.II, McGraw-Hill Book
Co., New York.

61

REFERENCES

Friedman, B., 1956, Principles of Techniques of Applied Mathematics, John Wiley & Sons, Inc.,
New York.
Golant, V. E., and A. D. Piliya, 1972,Linear Transformation and Absorption of Waves in a Plasma,
Sov. Phys.-Usp. (Engl. Transl.), 14:413.
Grossmann, W., M. Kaufmann, and J. Neuhauser, t973, Damping of Alfvdn and Magneto-Acoustic
Waves at lligh Beta, Nucl. Fusion, 13: 462.
Hasegawa, A., 197 5, Plasma Instability and Nonlinear Eflbcts, Springer, New York.
and L. Chen, 1976, Kinetic Processes in Plasma Heating by Resonant Mode Conversion of
-, Alfvdn Wave, Phys. Fluids, 19: 1924.
Kappraff, J. M., and J. A. Tataronis, 19'77, Resistive Effects on Alfvdn Wave Heating,J. Plasma

Phys.,18: 209.
Kruskal, M., and M. Schwarzschild, 1954, Some Instabilities of a Completely lonized Plasma,P)"oc.
R. .Soc. (London), A223:348.
Pritchett, P. L., and J. M. Dawson,1978, Phase Mixing in the Continuous Spectrum of Alfvdn
Waves, Phys. Fluids, 2l: 516.
Sedldiek, 2., 1971, Electrostatic Oscillations in Cold Inhomogeneous Plasma. 1. Differential
Equation Approach, J. Plasma Phys., 5: 239.
Uberoi, C., 1977, A Note on Resonant Absorption of Alfvdn Waves in Nonuniform Plasma, paper
presented at Kiev Conference on Plasma Physics, Trieste, April1977.

General Beferences

A. Hasegawa, l9'14, A Theory of Long-Period Magnetic Pulsations. I. Steady State

Excitation of Field Line Resonance,J. Geophys. Res., 79:. 1024-1032.
Cummings, W.D., R. J. O'Sullivan, and P. J. Coleman, Jr.,1969, Standing Alfvdn Waves in the
Magnetosphete, J. Geophys. Res, 74: 778-793.
Davies, 8., L969, Alfvdn Waves irf a Nonuniform Resistive Plasma,Phys. Rev. Lett.,22: 1246-1247.
Chen, Liu, and

Deutsh, R. V., 1968, hopagation of Alfvdn

Waves

in

Nonhomogeneous Plasma, Rev. Roum.

Phys.,13:673-681.
Gerwin, R., 1967, Hydromagnetic Surface Waves in a Conducting Liquid Surrounded by a
Compressible Gas, Phys. Fluids, l0:.2164.
Grossmanno W., M. Kaufmann, and J. Neuhauser, 1973,Damping of Alfvdn and Magneto-Acoustic
Waves at High Beta, Nucl. Fusion, 13: 462464.
and J. Tataronis, 1973, Decay of MHD Waves by Phase Mixing. II. The Theta-Pinch in
-, Cylindrical Geometry, Z. Pltys., 261: 2l'l-236.
Morrow, R., and M. H. Brennan, 1971, Compressional Alfvdn Waves in a Non-Uniform Plasma,
Plasma Phys., 13: 75-79.
Swanson, D.G., I975, Mode Conversion of Toroidal Alfvdn Waves, Phys. Fluids, 18:1269-1276.
1976, Mode Conversion and Tunneling at the Two-Ion Hybrid Resonance, Phys. Rev. Lett.,
-. 36: 316-319.

E
r,

PASMA tNSrABll,'ilEs ASSOCTATED

WITH THE AVVilN WAVE

In this chapter plasma instabilities associated with the excitation of Alfven waves are
discussed. As in many other cases of plasma instabilities, the Alfven wave becomes
unstable owing to either the velocity space nonequilibrium or the coordinate space
nonequilibrium. These instabilities appear by a conversion of free energy stored in the
plasma to the Alfven pdrturbation. The free energy exists when the velocity
distribution of the plasma particles is not an isotropic Maxwellian or when the plasma
is not uniform in space. The former case, a case of velocity space nonequilibrium,
produces an instability called "velocity space instability," whereas the other case, the
case of coordinate space nonequilibrium, produces an instability called "coordinate
space instability." The instabilities work to equilibrate the distribution. The instability
is found by solving the dispersion relation for the eigenfrequency, <,r, for a given value
of the wave number k. For an assumed Fourier amplitude expression of
exp [i(k ' x - @t)l , the instability exists when a; has a positive imaginary part. We shall
present some of the typical A1fv6n-wave instabilities.

FIRE.HOSE INSTABILITY
The fire-hose instability occurs when the plasma has an anisotropic pressure. If we
take the direction of k vector in the x-z plane, where z is the direction of the ambient
62

63

TWO-STREAM INSTABILITY

magnetic field, the Alfv6n-wave dispersion relation is written by

c2k?

(s.l)

p-xx=0

where xx is the xx component of the plasma dielectric constant (listed in the

appendix). In the magnetohydrodynamic (MHD) frequency (<^r/c,:.1 ( 1) and the
wave-number (kv1i/@ci ( l) range, e** is given by

xx

T*t(f *ll - r,tl

k?(r1il

(s.2)

where the summation j is over different species (electrons and ions), vs is the velocity
in the direction perpendicular to the ambient magnetic field, and angle brackets (( ))
indicate average. Hence the dispersion relation of the Alfven wave becomes

-\$n - 0r)

(s.3)

0g1 and 0r are the parallel and perpendicular components, respectively, of the
ratio of the plasma pressure to the magnetic-field pressure,

where

o.=\J1m:n(v?)j
.,,
pr
2
Brol2po

P,,

=X
j

rqin(v?):

BZl2po

As shown in Eq. 5.3, when 0lr -0r >2, a becomes purely imaginary; hence the
instability appears. Since the physical mechanism of this instability is similar to that
which generates oscillations in the water hose when the water pressure exceeds a
critical value, this is called "fire-hose instability" (Rosenbluth, 1956; Parker, 1958).

TWO.STR EAM I NSTABI

When a stream of charged particles exists in the plasma, the Alfven wave becomes

To analyze this instability, we cannot use the dispersion relation of the

Alfv6n wave (Eq.5.l),because the existence of a streaming particle produces a
unstable.

ITY

WAVE
PLASMA INSTABILITIES ASSOCIATED WITH THE ALFVEN

64

component. For simplicity, we consider the wave propagating in the

written
direction of the ambient magnetic field. Then the wave equation catr be

nonvanishing

e*,

u,+
I o' $ tl e**) t#.',1

=o

(s.4)

wave which is
Since the Alfven wave is the lower frequency limit of the ion cyclotron
vector
left-hand polarized, we have introduced a left-hand polarized electric-field
defined by El = E* * iErthe
For a low-beta plasma, the dispersion relation can be written, by using e"u in
appendix, as

k' -4.*(<^r

kvs) = o

(s.s)

=+( *)

(s 6)

or, alternatively,

to the plasma density and v6 is the

velocity of the electron stream. lt is easily seen that the instability occurs when
k=ov6<,)"il2r'a, and the threshold condition is ve > v4. The interesting point of this
instability is that the frequency at which the instability occurs is given by uurif 2;i'e',
it is proportional to the density ratio of the stream to the plasma. Another type of
two-stream instability exists in the Alfven wave when the finite Larmor radius
correction is included. In this case the kinetic Alfven wave is excited by electrostatic
coupling, which requires a density perturbation. The instability can be obtained by
using the dispersion of the kinetic Allven wave and by including an additional term
that originates from the streaming particles. The instability can be produced by
parstreaming ions in the tokamak (Rosenbluth and Rutherford, 197 5) or by alpha
where

a is a ratio of

streaming electron density

ticles (Sigma, 1980).

KE

LVIN-HE LMHO LTZ I NSTABI LITY

When a velocity shear exists, the Kelvin-Flelmholtz instability appears in a plasmaIn the Ml{D-frequency lange, the instability excites the Alfven wave'
For simplicity, we assume two types of plasmal one is stationary and exists at
x < 0 (medium 1) and the other exists at x > 0 (medium 2) and moves at the velocity
ve with respect to the former. The boundary surfaoe between these two plasmas is
+
x = 0. We consider a wave that propagates along this surface as exp li(kvy krz -

DRIFT ALFVfN.WAVE INSTABILITY

65

@t)] , where z is the direction of the ambient field. The analysis is similar to that in
Chap. 4 except that one c,.r is Doppler shifted with respect to another by k . vo. The
dispersion relation is then modified from Eq. 4.6,Chap.4.

porIa2

-(k. var)rf" * por[(.-k.

vo)2

-(k.

vnz)3]

=0

(5.7)

where the vector Alfven velocity, v4, is taken in the direction of the magnetic field.
The condition of instability can then be obtained from the condition of a complex
root of c^r as (Chandrasekhar, 1961)

(k'no)' r

(* .#)

[por(k.va)? t poz(k .nn)?]

(s.8)

We see that the instability is most easily excited when the plasma flow, vs, iS
perpendicular to the ambient magnetic field and fbr the wave that propagates in the
direction of the flow. The excited wave is elliptically polarized.

DBI FT AL FVEN.WAVE INSTABI LITY

In

an inhomogeneous plfsma the dispersion relation of the Alfven wave is modified

by the diamagnetic drift of

ions.

If

we ignore the finite Larmor radius effect, the

dispersion relation becomes

as2

where

c^riF

is the ion diamagnetic

+ c.lc,.lf

(kr

rc

(s.e)

drift frequency
<^lf =

where

- klv2s=g

x x)'

2v2^11

@ci

(s.10)

is the measure of the density gradient

K=

V ln

no

(s.11)

The finite Larmor radius correction produces the coupling to the electrostatic electron
drift wave. The local dispersion relation for <^l ) krvli then becomes (Mikhailovskii,
1967)

(.'*

(r<^lf

-k|r'n) (

*) ['.'(,{

#=ftr,r.7ut('-9

(s.12)

PLASMA INSTABILITIES ASSOCIATED WITH THE ALFVEN WAVE

66

where ).1 = (k1v1ilu"i)2. The Alfven wave becomes unstable iI' 2krv a <
frequency, cd, and the growth rate,y, respectively, given by

-=

k?.u'n

c^.tf

with the

(s.1 3)

uls

and

u2rl

7=triW 4"

k'vr"

(s.14)

Note that the Alfven frequency is reduced in this case. The dispersion relation is
obtained by assuming that k* equals 0, where x is the direction of the density
gradient. To include the local effect, we must solve the eigenvalue prclblem for a wave
equation in the x direction. Such an analysis gives results that distinguish whether the
instability is convective (spatial growth in x) or absolute (temporal growth). This is a
problem of recent controversy. An analysis by Tsang et al. (1978) shows that the

instability is convective.

If it is convective, the wave energ)' may escape from the plasma in a manner that is
exactly opposite from the resonant mode conversion, as discussed in Chap. 4. This
may contribute to anomalous heat loss from a plasma.
REFERENCES

Text Gitations
Chandrasekhar, S., 1961, Hydrodynamic and Hydromngnetic Stabiliry, Chap. 13, Clarendon Ptess,
Oxford, England.
Mikhailovskii, A. B., 196'7 , Review of Plasma Physics, M. A. Leontovich (Ed.), p. 172, Consultants
Bureau, New York.

Parker, G. N., 1958, Dynamic Instability of an Anisotropic Ionized Gas of [,ow Density,Phys.

Rev.,l09:

1874.

Rosenbluth, M. N., 1956, Stobility of the Pinch, USAEC Report LA-2030, Los Alamos Scientiltc
Laboratory, NTIS.
ild P. H. Ruthelfold, 7975, Excitation of Alfvdn Waves by Hieh Energy Ions in a Tokamak,
-, Phys. Rev. Lett.,34: 1428.
Sigma, D. J., 1980, Velocity Space Instabilities of Alpha Particles in Tokamak Reactors, proceedings of course held in Vatenna, Italy, Aug. 27-Sept. 8, 1980'
Tsang, K. T., J. C. Whitson, J. D. Callen, and J. Smith, 1978, Drift Alfven Waves in Tokamaks,
Phys. Rev. Lctt.,41: 557.

General References
Benezin, Yu. A., and

V. A. Vshivkov, 1976, On the

Computer Phys., 20 : 8196.

Firehose Instability of Alfv6n Waves,,I.

REFERENCES

Bizli, K. O., D. G. Lominadze, and A. G. Mikhailovskii, 1976, Stimulation of Shortwave Alfv6n

Waves by High-Energy lons in Tokamak , Fiz. Plazmy, 2(l7O):95-96.
Chao-Kung, Yang, and B. U. O. Sonnerup, 1976, Compressive Magnetic Field Reconnection a

Slow Wave Model, Astrophys. J.,206: 570-582.

DobrowolnY, M., 1972, Kelvin-Helmholtz Instability in a High-Beta Collisionless Plasma,phys.
Fluids, 15 : 2263-22'1 0.
Egorenkov, V. D., 1973, Low-Frequency Instabilities oi Cold Plasma with a Group of Hot Ions
with Anisotropic Velocity Distribution Function,Ilv. Fiz. Zh., IB:1472-1475.
ild K. N. Stepanov, 1973, Electromagnetic Instabilties in a Finite-Beta Plasma with Hot
-, Electrons, Zh. Tekh. Fiz., 43 : 27 0-27 5 ; Sov. Phys.-Tech. phys., 1 8 : 1 78-1 80.
Elsasser, K., and H. Schamel, 1972, Nonlinear Evolution of Firehose-Unstable Alfven Waves, ,I.
Plasma Phys., 7 : 47 5-502.
Hasegawa, A., 1975, Plasma Instabilities and Nonlinear Effects, Springer-Verlag, New York.
Hudson, M. K., and C. F. Kennel, 1975, The Electromagnetic Interchange Mode in a Partly-Ionized
Collisional Plasraa, J. Plasma Phys., 14: l2l-134.

T. I)., D. G. Lominadze, and K. N. Stepanov,l973, Excitation of Alfv6nWavesby Fast

Ions in a Finite-Beta Plasma, Zh. Tekh. Fiz., 43:2517-252O; Sov. Phys.-Tech. Phys., 18:

Kaladze,

1s83-1s84 (1974).
Kan, J. R', and R. R. I{eacock,l9T6,Generationof Irregular(TypePIC)PulsationsinthePlasma
Sheet During Substorms,J. Geophys. Res., 8l:2371-2371..
Kats, M. E., V. N. Makarenko, and A. K. Yukhimuk, 1976,E. N{. Wave Instability in Plasma with a
"Reverse Loss cone" (Application to Magnetosphere), zh. Tekh. Fiz.,2: 133-137.
Landau, R.W., and S. Cuperman, 197I, Stability of Anisotropic Plasmas to Almost-Perpendicular
Magnetosonic Waves, J. Plasma Phys., 6:495-512.

Mazur, V. A., and A. B. Mikhailovskii, 1976, Excitation of Alfv6n Waves by Fast Ions in Tokamak
with Non-Circular Cross-Section, Fiz. Plazmy, 2: l't2-17 S.
Mikhailovskii, A. B., l973, "Drift" Instabilities Distorting the Magnetic Surfaces of Tokamak-Type
Toroidal Systems, Nucl. Fusion, 13 : 259-269.
l975, "Drift" Stimulation of Alfv6n Waves by Trapped Electrons in Tokamak at Beta/Sub
-, Jl#1, Fiz. Phzmy, l: 378-392.
and O. A. Pokhote'ov, 197 5, A New Mechanism of Generation of Geomagnetic Pulsations by
-, Fast Particles, Fiz. Plazmy, 1: 786-792.
Montgomery, M. D., S. O. Gary, W. C. Fieldman, and D.W. Forslund, 1976, Electromagnetic
Instabilities Driven by Unequal Proton Beams in the Solar Wind, J. Geophys. Res., 8l:
2',143-2749.

Nishida, Y., and K. Ishii, 1974, Observation of the Coupled Mode of aCollisionalDriftWaveand
an Alfv6n Wave, Phys. Rev. Lett., 33:352-3SS.
Ohsawa, Y., K. Nosaki, and A. Hasegawa, 7976, Kinetic Theory of Magnetohydrodynamic
Kelvin-Helmholtz Instability, Phys. Fluids. l9: 1139-1143.
Roberts, 8., 797 6, overstability and cooling in Sunsp ots, Astrophys. J., 204: 269-2g0.
Tang, J. T., N. G. Luhmann, Jr., Y. Nishida, and K. Ishii, 1975, Destabilization of Hydromagnetic
Drift-Alfv6n Waves in a Finite-Beta Collisional Plasma, Phys. Rev. Lett., 34: j0-j3.

67

Irrr-

NoNT.TNEAR PRocEssEs

In the preceding

chapters we considered small-amplitude Alfven waves. When the

perturbed magnetic field is no longer smail compared with the ambient magnetic field,
the study of nonlinear magnetohydrodynamic (MHD) equations becomes necessary. It
is now well-known that a circularly polarized Alfven wave described by B = (Bo ,By,Bz)
and v = (O,vy,v") is an exact solution of the nonlinear MHD equations (Ferraro, 1956).
Ferraro also pointed out the equi-partition of kinetic and mechanical energy, which is
characteristic of the MHD waves. Can a large-amplitude Alfven wave persist in a rcal
plasma? The decay instability of this wave into a small-amplitude Alfven wave and a
sound wave has been discussed in detail by Sagdeev and Galeev (1969).
The large-amplitude linearly polarized Alfven wave, on the other hand, has the
local phase velocity

" BA + B(0'
,p---F
which can be seen to approach the Alfven velocity as B(t)/no - 6. The rapid
development of a plane polarized Alfven wave into hydromagnetic shock has been
discussed by Montgomery (1958) and has been observed experimentally by Boley and
Forman (1964).
The possibility of the propagation of solitary hydromagnetic waves in the direction
parallel to a uniform magnetic field in a cold collision-free plasma was shown by
Shaf'fman (1961). It was noted that solitary Alfven waves move much faster than
linearized Alfven waves in the plasma.
68

-l'QUASI-LINEAR DIFFUSION

69

A large number of papers on the nonlinear effects of the MHD Alfven waves have
been published in the comparatively recent literature. However, as we have seen in the
linear theory of the Alfven waves, in most laboratory-scale plasmas, its inhomogeneity
prohiLrits the existence of the ideal MHD Alfven wave. Hence the nonlinear theory of
the NIHD Alfven wave is not applicable in most laboratory plasmas. ln this chapter,
therefbre, we shall consider nonlinear processes associated with the Alfven-wave
perturbation with particular emphasis on its kinetic effects.
We discuss quasi-linear diffusion in the presence of the kinetic Alfven wave
turbulence, parametric decay and nonlinear Landau damping of the kinetic Alfven
wave, sonte solitary wave solutions, and the modulational instability of the
finite-frequency Alfven wave.

OUASI.LINEAH

IFFUSION

In the collisionless plasma with a stationary ensemble of waves, the quasi-linear

diffusion occurs only in the presence of wave-particle resonant interactions. The MHD
Alfven wave does not have the wave-particle resonant interactions;hence there should
be no quasi-linear diffusion. One should introduce either a finite-frequency or a
nite-Larmor-radius e ffe ct.
Consequently we consider here the quasilinear diffusion in the presence of the
kinetic Alfven-wave turbulence (Hasegawa and Mima, 1978). This is relevant in fusion
plasma because the kinetic. Alfven wave is excited, for example, by the drift-wave

fi

instability.
Electron Diffusion
We first consider the elecnon diffttsion. We use the drift kinetic equation (see the
appendix) for the electron guiding center distribution function fs(v11,x,t).

t?- ,,#+vr

. (vor")

- #tu" +(vaxBr).21 #=

(6.1)

Here we used a local Cartesian coordinate in which the z axis is taken to be in the
direction of the ambient magnetic field Bo i E and B are the wave electric and
magnetic fields, respectively; subscripts z and 1 indicate the components in the
direction parallel and perpendicular to the ambient magnetic field, respectively; 2 is
the unit vector in the z direction; and v6 is the drift velocity perpendicular to the
magnetic field, which, for electrons, is given by

nd =

E1XBs.81
*
u,
-l?E;

(6.2)

Irr>
NONLINEAR PROCESSES

70

p is small (compared with unity) to warrant

neglecting the nragnetic compression Br.
We assume that the density gradient exists in the x direction. Then the quasi-linear

We assumed also that the plasma

equation for the average distribution function (f") can be written

#- efie,

v,B*){')

-fr

F,

as

#)=

(6.3)

where fjt) i, the linear response and O shows the ensemble average.
Fourier amplitude expression for the linear terms, such as

If we use the

fjl) = *.1 f[t )1x; .i(k'x-t';t)

f{t)

the linear response of the distribution function

ff'1)

=''

ffiP-

E,k

fr #)
+ n6(kzvz

<o)

.un be expressed

as

##
"n]
(*;#-ffi H

u,u

6.4)

where P indicates the principal value, 6 is the Dirac delta function, and the Maxwell
equation V x E = - 0B/0t is used to express 81 in terms of E, and E7.

n
_kyErt -k2Eyk
DxK---T_

(6.s)

lf

we substitute Eq. 6.4 into Eq. 6.3, the term multiplied by fhe delta function, which
represents the dissipation due to the Landau damping, contributes to the quasi-linear
diffusion coefficient. The first term contributes to the diffusion in velocity space and
the second term to that in coordinate space. Taking the second-term contribution

here, we have

Y = * I lffitE'r't2 n6(k7v2-

c'r)

#]

(6.6)

As is expected, only the z component of the wave electric field contributes to the
diffusion. The preceding result is valid for any kind of low-frequency electromagnetic
waves not restricted to the kinetic Alfven wave.

The diffusion equation for electron density, oe = -f (f") dvr, can be constructed by
integrating Eq. 6.6 over v". If we assume a Gaussian distribution for (fs), i.e.,

(f") = n.(x)

r_r

-v2' l

(6.7)

Tvl
"

QUASI.LINEAR DIITFUSION
we have, for v1"

71

uA,

(6.8)

*=*(""*)
where

o.=

(f)nX

tfi lErrl2
-w-tm
1

(6.e)

As stated previously, a unique feature of the kinetic Alfven wave is that it

accompanies the electric field E, in the direction of the ambient magnetic field.

lon Diffusion
We now discuss the ion diffusion. The diffusion coefficient can be formally
obtained by using fi in place of f" in Eq. 6.6. However, because fi(u = ulk, = vA)
^,
exp (-vf,lZv'?-firy eXp eP- t), if the plasma beta is small, the diffusion coefficient
obtained this way gives a value negligibly small compared with that of electrons.
However, if a local energy balance is maintained between electron and ion
dynamics, the ambipolarity results, namely, the ion-diffusion coefficient, become
identical to that of the electrons. To present an example of such a case, we consider a
situation in which the fluctuation is generated by the drift-wave instability. When the
local energy balance exists, the fluctuation excited by the inverse Landau damping of
electrons is absorbed by ion damping and the stationary state is maintained. If we take
the viscous damping as the damping mechanism by the ions and ignore the finite ion
Larmor radius effect for simplicity, the ion-density fluctuation, ni1, is obtained from
the equation of motion and the equation of continuity,
nik =

no I t Eno E_y.r kul"rr(l + iuk?la:) Evr.l

rL* Jlgo *
oci
B;J

(6.10)

where z is the viscosity coefficient and the parallel ion inertia is ignored for the Alfven
branch considered here. We note that the diffusion coefficient, Di, is obtained by
calculating the diffusive ion flux,

JDi = hipv11)
1

iRe(nsEyr)

=-o,P
dx

(6.11)

Irr'72

NONLINEAR PROCESSES

If we substitute Eq. 6.10 for this expression, we can formally obtain the ion flux
in terms of the viscosity coefficient r,

r -llEvkl2klun6
rDi-z
BA -"r.t

(6.12)

To compare this with the electron diffusive flux (Eq.6.8), we need to express y ancl
terms of 5. and Er1. For this, we derive the dispersion relation by using

Eyt in

quasi-neutrality and the AmpAre law, including the effect of the density gradient. By
assuming stationary state (real co), we have, from the real part of the dispersion
relation (Eq. 5.12),
(c,-r

- a*)(u" - k?u'd)= ).rklvf,<,.'

(6.13)

and, from the imaginary part,

6 eG)

= uk2"
v\.

a*l

<'t

(6.t4)

where c^r* is the drift-wave frequency given by

0)a

kvui. I
@ce

0n6

116 0X

The relation between E21 and Ey1 can be obtained from the Ampire law:

Eyk=llr,*t'

{ffi#t]

(6.1s)

Using the dispersion relation (Eq. 6.13) and substituting Eqs.6.14 and 6.15 into
Eq.6.12,wehave

rDi=tc'wff )**
If we compare this expression with Eq. 6.1 l, we see

o,=

(t)nTtr#,,,fu=o"

(6.16)

and hence the ambipolarity is proved.

However, we point out that, if the local energy balance between electron and ion
dynamics does not exist, the ambipolarity disappears. For example, if the stationary

PARAMETRIC

DECAY

73

state is achieved whereby the fluctuation produced by the electrons is convected away

by the wave or is reabsorbed by

electrons, the ion diffusion coefficient becomes

practically zeto.

PARAMETRIC DECAY

The MHD Alfvdn wave is known to decay into another A1fv6n wave and the ion
acoustic wave (Sagdeev and Galeev, 1969). Hence we discuss the decay of the kinetic

Alfvdn wave, which has a much larger decay rate than the MHD Alfvdn wave
(Hasegawa and Chen,1976). We consider a self-consistent pump wave, @s(x,t) (the
kinetic Alfvdn wave), of the following form:
do(x,t)

=){O"exp [-i(coot-ko . x)l +c. c]

(6.17)

where (oo, ko) satisfies the linear dispersion relation for the kinetic Alfven wave. The
@0, is assumed to be sufficiently weak so that only interactions up to
O(ldo12) need to be kept. Furthermore, since decay instabilities are considered here,
we ignore the upper sideband as being off-resonant and discuss only the couplings
among the pump wave (t<,.ro, tko), the lower sideband (-_, k_)= (.,
-.0,
kr - ko), and the low-frequency wave (c.l., kr). Note again that (t.ro, tko) and (<.,r- ,
k-) are kinetic Alfven wavs and (or, kr) is the ion acoustic mode. For low-beta
plasmas, l<,:el=lkrvel> l<^rrl ez lkrcsl. To further simplify the analyses, we make the
additional assumptions that lkrpil < I and T./Ti > l; thus we ignore the finite ion
Larmor radius effect. The dynamics of both species are then described by the
following drift kinetic equation:

pump field,

$. " #. vr . (vroj . (fi),

l},+("r x BJ . 2Jj

fr=

(6.18)

where

vrj

=(vB+vp+vg)i

0 =e,i)

(6.1e)

with vB = E1 x Bs lB,,, vpj = (m/qB3)i (dEt/dt), and vs = v2B1/Be, and where

!(x,v",t) is the drift distribution function and other notations are standard. Note that
here d/dt contains a convective term; i.e., d/dt = dldt+ (v . V).
Letting t=4o'*4tl*fiNL and using the.two potentials @ and r/ defined in
Eqs.2.7 and 2.8, we obtain tlie linear response,

{t

fj') = fjo) U

r, ut

(6.20)

NONLINEAR PROCESSES

74
and
ri(, ) =

- rr.(or o + c?

(6.2t)

#r;*t

where V = eQlT",6 = eQlT", and L = k1(c3 l-'"). Equations 6.20 and 6.21 are valid
for both kinetic Alfven and ion acousticr waves. Note, however, since (<^rr, kr) is an
electrostatic mode'; @* is equal to gs;whereas, for the kinetic Alfven modes, (,^r-' k-)
(c^r6 , ko ), 0- ,o is not equal to {
- ,o .
In dealing with the nonlinear analyses, let us note that, because the kinetic Alfven
waves have I-o O( l), our results are therefore valid in the regime tr ) .o la"i, whereas
the classic MHD results of Sagdeev and Galeev (1969) are valid in the opposite limit.

and

The details of nonlinear interactions are different for the different modes as well

as

for

species and will be treated separately.

Let us first consider the ion acoustic (<^rs, kr) mode. Since we are interested in
either the resonant decay to the ion acoustic mode or the induced scattering decay
when this mode is heavily damped, only nonlinear perturbations up to O(l0ol) need to
be kept here; i.e.,we must calculate fi(') onty. For the electrons lc^rsl,l<^rol (c^r"., and
lkrpel ( lkrpil 41, vpe, the electroh polarization drift can therefore be neglected.
Referring to Eq.6.8, we then find the dominant nonlinear contributions to fj"' to
originate from the V1 . [vs(Oo) fjt)(O-)+ (O+ -)] term as well as from the

(q/*).{[vB(Oo)
o-- = (;- , k-),

x Br(O-)l'2 * (0 + -)]{af;otl}vr) tgtp,where Oo = (,^ro, ko),

and Q, = (rr, kr). With lc^rsl ( ik"svr"l, rj') (Q*) is given by
rj'z)(o,)

where

[g

- tjo, [Ue(Qr) + fs(o,)]

(6.22)

corresponds to the nBfr(t) term;

1

7s(o,)=;s*(k,xko)'2
and where

[B corresponds

try

k1-),0(1 + tr-)l

,-J

- -Qo
Q

(6.23)

to the vB X 81 term;

(1 + \o)
Us(or)=#(k,x ko)'tFU (ng -- -*]t-l

Q-Qo (624)

In deriving Eqs. 6.23 and 6.24, we have made use of the fact that Qe and O- are
kinetic Alfven waves and hence ,,l, (or E2) and 81 are related to @ by Eq. 6.5.
In treating the ions, we note first that, because lc^r.-1, lc^r6l ) lkrrovTil, lk11-v1il, v3
has negligible contribution, and, from Eq.6.2l,fr(t) of the kinetic Alfv6n wave is
reduced to

PARAMETRIC DECAY

75

f,(')(o) = - Ir,,o, O(rl) = ftlt) *

f#)

Ytf., + 0(o)

(for

o = oo, o-)

(6.2s)

Second, we note that vpi contains a nonlinear term from its convective part; i.e.,

"['r)the
(m/qB3)i (" 'V)i E1. However, it can be shown that the contribution to
cancels to the order. aolasci with that from rEfilt).
V1 ' [vlfli term from

"[1)f[t) comes from the

Thus the only net contributi6n
t.r-, which corresponds to't-he
""t#)
usual (nE . V )v, convective nonlinear (ponderomotive) force term in the parallel (to
Be) equation of motion. Another important nonlinear contribution, which is similar to
that of the electrons, comes from the (vB x Br) . 2 term. Combining these nonlinear
terms, we obtain
ri(2)(o,)

=.3

J4o/au' [p,(n,) + 7p(or)]

where ,/r. corresponcls to the Vr

u.(oJ = -

. ["Bf#)]

,;:,k^(k,

(6.26)

nonlinearity antl isgiven by

x ko)

.r(*

e=L"J 6-60

and 7r(Or) is given in Eq. 6.24. Substituting Eqs. 6.20,6.21,6.22, and 6.26 into the
quasi-neutrality equation

1nj')(or) + nj'){e.)l = [ni(')(or) + nf')(or)l

(6.27)

we have

er(Or) 6, =

ArtF(I)

+ er(Or)l

6o6-

(6.28)

where

s=l+trr+1i

(6.2e)

xi--"?1ffi0",

(6.30)

n,=#e- kr-\
a- I

N-

A
e)s

(6.31)

-f

(6.32)

NONLINEAR PROCESSES

76
and

F(I)

= Lo +

l-

+ tre ).-

(6.33)

).t

Equation 6.28 describes the coupling of the ion acoustic mode to the lower-sideband
kinetic Alfven wave via the pump.
pt us now consider the kinetic Alfven (a-,k-) mode. For this mode we have to
calculate both the charge and the parallel current-density perturbations to O(l6ol2),
i.e., {t)(O-), to take into account the induced scattering process.
fhu, nonlinear contributions due to both i(')(Or) and {2)(Or) must be included.
Let us first consider the electrons. Again vo" hai negligible contribution, and dominant
nonlinear contributions come from the
[f"(t)(Or) * f[t)(Or)] term aswell as
"fi(Oo)
out, howevJr, that nJ') it negligible
turns
It
term.
2
x
Bf(Oo)]
'
from the [vB(Qr)
nonlinear terms, whereas n|'/
dominant
of
the
canceling
to
owing
compared with nf2)
is given by

nj.)tn-)
Perturbations
equation,

'-i"?$ffi(k,

x ko) .zdfi ni'?)@J

(6.2+s

in parallel current density can then be obtained from the continuity

r"(3)(n-) = ekr-

[vE(os) n*(oo) + vg(oo) n(l)(Os)]e

(6.3s)

kz_

and

ljl)1o-;

. v$(oo) n(2)(os)

*.

nta)1O_)l

kz_

(6.36)

As to the ions, because l<^r-lllkz-vTil, the dynamic is mainly in the

perpendicular (to Bo) plane. Thus we can neglect J7i, and, from the continuity
equation, we find the ion-density perturbations to be

nfr)1o*y ^, kr_

[nov[2) + vE(os) n*(oo) + vfi(oo)

n(1)(os)]i

6.37)

and

n{3)1o-y
where
term.

u$l)tO-l is the

- kr- .

tvf,(oo)

nl)(oJli

(6.38)

nonlinear ion polarization drift due to the convective (vr

'Vr)

PARAMETRIC

DECAY

77

The two field equations, Eqs. 2.9 and 2.10, including nonlinear perturbations,

becomeatQ=Onr(t)

#rl

n"(r)

* n.(r) = nr(t) *

n-(2)

1n(a)

(6.3e)

-,r)= uo X #u"!' 1 * t$)+ rjj)t

(6.40)

j=e,i

which can be combined into a single equation,

ea(a-) 6-

Noc,r-).-(t

L)
I

+ 43ll
nf))- (r + X_)k, -tl[']
e

)
lc,r- (n9 +
L

(6.41)

where
I

en(O-)

K"-Va

=, -ff(l

n!2)'(3) = [ni(e_)

I-)

+-

(6.42)

(2),(3)
- n"(e_)]

(6.43)

and

J[z)'1t1

(6.44)

= 1Q)'G)(O-)

In deriving Eq. 6"4I, we have noted that O- is a resonant kinetic Alfven mode, and
( 1. If we substitute Eqs. 6.34 to 6.38 for the nonlinear charge density and

hence leal

parallel current density perturbations into Eq. 6.41, it reduces to

6a(o-) 6-

= AA

F(I) dffds

(6.4s)

where

6A=A-.f;)
with ef;) due to third-order perturbations,

ef;)

= tdot'

rt# [' - "tl:: tl k

(6.46)

(6.47)

78

NONLINEAR PROCESSES
and where

Aa

co-f-1t

I-)

,^roL-(t

f,-)

(6.48)

Combining the two coupled equations, Eqs.6.38 and 6.45, we then derive the
following dispersion relation for the parametric decay instabilities:

lea

- ef;)l es = A4A, pr(l) ldol,

6.49)

where

ef;) = .f;) * AaA, p(r) lo-ol'

(6.s0)

With Te ) Ti, the acoustic wave is weakly damped, and we have the resonant
In this cut., uf ) can be neglected. kt or, = G)sr + i7 and e)- =
*
-@A iT, whero G)sr and c^r4 = G)0 - c^rr. satisfy, respectively, the dispersion relations
for the ion acoustic and kinetic Alfven waves. Equation 6.49 then reduces to
decay instability.

P*(z+rn)(r+rJ=i*l'ffi
o(rsr o@A

0rr,

2(l + Is)
=
.*-, A-*
De.q,,

(6.s2)

Ocol=QA
and

14

and

Ir

(6.s 1)

(6.s3)

are the corresponding linear damping rates. From Eq.6.51 we can

deduce the threshold pump field by letting 7 = 0. Well above the threshold field, the
growth rate is given by
0l
lF(),)
^,D="(S l#l u-* [(r+\o)(r+r_1r+t)]b
sin

(6.s4)

ln deriving Eq. 6.54, we have let (k, x ko) . 2 = (k- x ko) . 2 = kt-k1e sin g and
have used the relation between 816 and @6(=e@o/T") expressed in Eq.6.5. The
growth rate obtained here is larger than that of the ideal MHD results by a factor of
).c,.r.1/c,la.

This enhancement is expected because, owing to the finite f's, nonlinear effects
by the E x Bq drifts of electrons and ions do not cancel each other to the
order of L in the case considered here, whereas only ion polarization drift [which is
induced

PARAMETRIC DECAY

smaller than the E x Bs drift by a factor (c^rn/<.r"Jl contributes in the ideal MHD
limit. Furthermore, unlike the MHD case in which only the backscattering is allowed,
three different types of decay are possible here as illustrated in the c,r vs. k, diagrams

in Fig.6.1. Note also, since

F(I)+

as ).q

-+0, this decay processispertinenttothe

pump wave's being a kinetic Alfven wave.

Let us now consider the case in which T" { 5T1 so that the ion acoustic wave is
heavily Landau damped by ions. In this case the decay instability is made through
nonlinear ion Landau damping; i.e., it is an ion-induced scattering process or a

\o'\a

Xo

t).o

tro?trA
Fig. 6.1 Dispersion diagram showing the decay process of the kinetic Alfv6n wave at uro to
another kinetic Alfv6n wave at ur4 and the ion acoustic wave at cds.

NONLINEAR PROCESSES

80
quasi-mode decay instability (which
growth rate, Tp, is obtained to be

it is sometirnes

also called). From Eq.6.49, the

7N=-+t-W.rf,]

-ra

(6.ss)

As indicated in Eq. 6.50, 6f;) aoes not contribute to the growth rate. and 7p further
reduces to

'[=)
which has its maximum value at lcosl

F'(I) sin2 g lmxi _

rA
(1 + IoXl + tr--) le*l'

= lkrrvlll

and

sin2 o
(rN)."*o?p-'lHnl'(S (t F'(r)
+ I-)
+

IoXl

(6.s6)

1t +

T"/Ti
lr)2 + (T"/T)2

fa

(6.56')

Main conclusions are: (l) the growth rate obtained here is larger than the classic MHD
value by a factor (Lc-r"1/c^,r 1J';Q) the results are pertinent to the pump wave's being a
kinetic Alfven wave; and (3) there is a similarity to the resonant decay instability in
that three types of decay are possible here (see Fig.6.1). Note that the threshold
pump field depends on f4, which, in the collisionless regime, is due mainly to electron
Landau damping and, typically, fa/c^rci - O(10-'). Then, for a reasonable choice of
other parameters, such as g- lO-2, osciluol= 10, and T" =Ti, the threshold
amplitude of B1.s, (Bro)tn/B6 becomes approximately 10-2.

SOLITARY WAVE OF THE

KINETIC ALFVEN WAVE
The solitary wave is a nonlinear wave that propagates in a stationary form. Usually
such a stationary waveform is generated by a balance between a nonlinear steepening
effect and a dispersion effect.
Since the MHD Alfven wave does not have dispersion. to construct a solitary wave,
one should introduce either the finite frequency effect (Kakutani,1974) or the finiteLarmor-radius effect. We consider the latter example and construct the scllitary wave of
the kinetic Alfven wave (Hasegawa and Mima, 1976).
The kinetic Alfven wave has a dispersion for an oblique propagation. The
nonlinearity originates from its compressible nature.
We rnake two assumptions. The first is that the plasma Bf 2p6nT/83) is small but
much larger than the electron/ion mass ratio. The effect of a high-beta correction was
studied recently by Yu and Shukla (1978). The second is that the variation exists in
the x-z plane, where z is the direction of an ambient magnetic field, Bs, and x is a
perpendicular direction.

SOLITARY WAVE OF THE KINETIC ALFVEN

WAVE

81

The low-beta assumption allows one to use the two potential fields 0 and 0 to

describe the electric field, Ex = - Ailldx and E, - - a9lAz. They produce only shear
perturbations in the magnetic field; B, = Bo (constant), Bx = 0, and

aB,,
--'=*a2 (0-P)
dt

dx

(6.s7)

dz'

The relevant field equations are the quasi-neutrality relation for ion and electron
densities n1 and ns,

(6.s8)

fli = l1e
and the Ampere law for

Jr,the current density in the z direction,

a4

ffi@-'r)=uo

d'Jt

(6.se)

ar;.

In a low-beta plasma, the ion density is given by

(2

Dni- !

0t

a"o

n) =o

(6.60)

ax \eBfr 0x 0t'^/

The electron density is obtained from the drift kinetic equation,

u,*+
dz

$1u6*t")'

dx

*9PP=
m dz |vr.

(6.61)

Since the drift velocity v6[= v2(B/Bo) + nx (2lgr)] has only the y component'the
second term does not contribute. Hence the electron density can be assumed to be
given by the Boltzmann distribution,

Ile = oo

et!

.xPE

(6.62)

where Ts is the electron temperature. In Ec1. 6.60 the contribution of the ion parallel
inertia term is ignored by the low-beta assumption.
Similarly, the contribution of ions to the current density, Jr, is negligible;hence J,
is given by the electron densitY.
AJz

at
dz=O(en.)

(6.63)

to 6.63 become dimensionless if we introduce the new variables

E=xlpr, l=zasplc, r=<rrsit, il= nlno,6= e{lTe, and V= ellTe, where pr=
Equations 6.59

ItlrNONLINEAR PROCESSES

lTslm)+Lle)ci, e)ci= eB6/mi, @pi= (e2n6lesmi)h, and urhere ne is the unperturbed

of light. We then consider a one-dimensional simple
wave solution propagating obliquely with respect to the ambient magnetic field. The
simple wave is chosen to be stationary with respect to a coordinate 4, which is defined
plasma density and c is the speed

by

e=KxE+Krf-r

(6.64)

For the boundary conditions, we require that the number density perturbation,
n(n), as well as its derivatives, }nldt and 0n/0f, vanish as | -++ * and f *+ "". In
terms of 6 and [, this means ,t,=art,rcE= arttldl= 02610E2 = 036/(02 af)= 0 as

t-t-andf++oo.

With the use of these boundary conditions and the variable ?, Eqs. 6.59 and 6.60
can be integrated. If we then eliminate @, ry', and J, from Eqs. 6.59 to 6.63, we have
the following nonlinear wave equation:

(K:

--

D (1 -

n"1

rlrjn #(+*uJ=o

(6.6s)

The linear dispersion relation is obtained by substituting fr = I + 6n exp (iKn) = exp

[iK(KxE+Krl-")1 :exp [i(k*x +k z-

<^rt)]

t=Kl(1+k'zK?)

(6.66)

In terms of the original coordinates,

(,)2

=k|u'n^1t + t<lo3)

(6.67)

of the kinetic Alfven wave (8q.2.19). The related

solitary wave solution is expected to have a structure scaled by the ion gyroradius p1 in
the perpendicular direction and to propagate almost in the parallel direction.
Equation 6.65 can be readily integated once by changing the independent variable
from 4 to n. Using the condition dfr/dn = 0 at fr = 1, we obtain

This is the dispersion relation

ffi)'

=&(r
_ F(n,K?)

KiKl

Expanding F around
solitary wave.

fr=

- f,(K+D+(r+c)nhnl
(6.68)

I identifies d= I asadouble root,aconditionrequired

fora

SOLITARY WAVE OF THE KINETIC ALFVEN WAVE

The potential function F(nKZ) is plotted in Fig. 6.2 for various values of Kl. The
nonlinear wave(s) that we are seeking exists only for the range of density that makes
F>0, including n= 1. As shown in this figure, no range exists if K? < 1; only one
range,0<n( l, exists if Kl = 1; and tworanges,0(n< l and l<n<frr"*,exist
it K? > 1. K, = I corresponds to a wave that propagates at the Alfven speed in the z
direction; hence KZ > | 67 < 1) corresponds to a sub-Alfvenic (super-Alfvenic) wave.

\-

-?1

Fig. 6.2 Nonlinear potential function, F, vs. the normalized number density. The solitary wave
solution exists when F ) 0.

For the sub-Alfoenic range, there are two types of solitary wave solutions, one
accompanying a density hump (n > 1) and the other accompanying a density dip

< 1). Qualitatively this result agrees with the small-amplitude result. However, the
wave corresponding to n( I has a minimum density of zero, which is beyond the
scope of our theoretical framework (since the Alfven speed becomes infinity), and
other effects, such as the finite frequency effect, need to be introduced to obtain a
reasonable solution in this range of il-.

(n

FDD

NONLINEAR PROCESSES

Consequently we consider only the wave corresponding to i> l. lf K? deviates

say M =KZ - 1 < I (M > 0), the maximum amplitude
d."* of the solitary wave is correspondingly small. For such a case, we can integrate
Eq. 6.68 analytically for a small density perturbation, A - n - 1, to give

from unity by a small amount,

a=f,.,n'[#-l

(6.6e)

The corresponding magnetic-field perturbation, By (normalized by Bo), is obtained

from Eq. 6.57.

#=Ey=-*I,'- t)#.

- 3(M)Y' sgn (K*Kr)rL

f

The wave accompanies a finite

given by E*(n

+t

-) = uAK,

x component of the electric field as n -> t

Bv(tt * t ""). In general, E* is given by
E-

Ex:t^Br-rtn-

tanh

cs

f n- t orl

"^J n R;

"o, which is

(6.7

r)

The present solution is exact to an arbitrary amplitude so long as the result is

consistent with the low-beta assumption. This means that the effective phase velocity
of the wave should be larger than the ion sound speed, cr. Hence the madmum
allowable value of K" is v4/cr, or Y-7. S P- t . For a large value of K2, the peak density
fr*"* is related to K? through KZ o fl*"*/ln frrnu*. Hence the present result should
t
be valid if Rmax S p- .
Figure 6.3 shows the density profiles for F*u* = l0 and nrnul = 5, which are
obtained by numerically integrating Eq. 6.65. The broken lines show the sech2
solutions that are fitted to the maximum density point. The exact solitary wave has a
wider structure than the ideal sech2 solution. The'corresponding values By and E* are
shown in Fig. 6.4.The integration constant is chosen so that these values vanish at the
origin of the 4 coordinate. As expected, they have kink structures.
The structure of the total magnetic line of force is also interesting. Because
Br(n * t "") -+ + constant, the line of force must be straight as ? + t -. Hence the
projection of the field line into the q - y plane is basically a triangular strape with a
round top as shown in Fig.6.5. The actual line of force exists in the y-z plane.
Because the phase stationary line at t = 0 exists on ? = 0, such a line can be expressed
by apizlc = - [(K* lKr)xl pt], namely, z = - (K*/Kr) (va/cr)x. Hence the z axis has
an angle Q = -- tan-1 (K2c./Kr1va) with respect to r7= 0line. Therefore the actual
line of force follows the edge of an oblique slice of the triangular roof-like surface by

85

SOLIT.ARY WAVE OF THE KINETIC ALFVEN WAVE

nkl/Kx,K?l

hmqx=fO; X!

ilmox = 5;

K2, =

2.g5

ot25

\-2-'l

rl/K^
Fig.

6.3

Normalized density as a function

of 4/K*.

B, vs/c,

t',

vs/c,

6, v4lc., E,

va7c.

(-ntot = tO)

1/Kx

-?-lo
-4

Fig.

6.4

Normalized electric

field, E*, and magnetic field, By, pbtted

as a

function of 4/K*.

F!
NONLINEAR PROCESSES

86

Fig. 6.5 Kink structure of magnetic line of force of the kinetic Alfv6n solitary wave (shown by
the curve with arrowheads).

y-z plane as shown by the arrowed solid curve in the figure. For a linear wave,
approximately equals @cila ) I , as can be seen from Eq. 6.67. Hence the angle 0
is usually very small.
the

K*

MODU LATIO NAL INSTABI LITY

AND ENVELOPE SOLITON

When a large-amplitude wave propagates in a nonlinear dispersive medium, the
amplitude modulation is known to grow under a certain condition. Such phenomena
are called modulational instability. When the nonlinear dispersion relation for c.r is
given as co(k,a), where k is the wave number and a is the amplitude of the wave,the
nonlinear equation that describes the dynamics of a wave amplitude @(x,t) with a
monochromatic carrier frequency {d6 and wave number k6 is given by (Karpman and
Krushkal, 1969)

i(H.",9-i#

#-f

rrr'' -roor')o=o

(6.t2)

where

ug

6a:l

= aElt=ko,

Oue

a=o

E-

d".l
lplk=ko, a=o

Doo
-

Dc^rl

Da =_lDa lk=t o , o=0

and l0ol is the magnitude of the stationary amplitude. The envelope function @(x,t) is
shown to be unstable if the product (0vr/0k . Dc^ro/Oa) has a negative sign. To prove
this, we put

Q:

Lp(x,t)ll4 eio(x't)

(6.73)

MODULATIONAL INSTABILITY AND ENVELOPE SOLITON

87

and obtain the dispersion relation O = Q(K) for small perturbations on p and o so that

(*) = e'9.*'(s)

exp {i[K(x

vrt)

- At] ]

(6.14)

By substituting Eqs. 6.73 and 6.74 into 6.72, we derive the dispersion relation,
dL2

=+ -

(6.7s)

apoK2

where

u- - diao16a
Ovr/0r
From Eq. 6.75 we

see

(6.76)

easily that an instability arises when o > 0. This means that

in the wave amplitude

(envelope) grows in time. Such an

instability is called "modulational instability" and is originally discovered in nonlinear

small perturbation

optics"

The physical picture of the instability can be easily understood if we realize that
F,q.6.72 represents the Scfuodinger equation of a quasi-particle in a potential field
given by (0<,ro/0a) (Ol2 * ldol2); if a> 0, the potential is attractive and the
quasiparticles are trapped in the potential whose size is proportional to the
quasi-particle density lfil2 itself. Hence the more the quasiparticles are trapped, the
deeper the trapping potential becomes; thus the perturbation grows.

This self-trapping process leads to a solitary wave solution for ldl2. If we put
i0ol2 = 0 and obtain an exact stationary solution of Eq. 6.72 by imposing the
condition that ldl2 -+ 0 at x + t -, wo have, for a > 0,
l4lz = po sech2 [(opo)% (x

vrt)]

(6.77)

and

aPo

=a-,

(6.78)

If o < 0, the solitary wave solution appears only under the condition that l|l2 pr(* 0) at x -+ * *, The solitary wave appears in the form of an absence of wave
ener$/ (Hasegawa and Tappert,
l0l2 =

197

pr U -

3),

ps sech'

[(ol

prpo)'h (x

- vrt)] ]

(6.79)

and

o= fiolnr(r

po)l%

(*

- vet) r tan- t {f-"i(\t - Pol

x(x -

tanh [(lalp
opr)'h
I

'r')1) -+pr(3 -

po)t

(6.80)

NONLINEAR PROCESSES

88

Zakhalov and Shabat (1972) have shown that the nonlinear Schrodinger equation
(Eq. 6.72) can be solved exactly by using the inverse scattering rnethod and that these
solitary wave solutions are, in fact, solitons: i.e., the time asymptotic solution of
Eq. 6.72 for an arbitrary initial condition can be expressed in terns of solitary wave
solutions of Eq. 6.77.
In contrast to the solitary wave solutions of the wave itself, these solitons are
often called envelope solitons.
Now, let us see whether the Alfven wave is modulationally stable or not. As we
have seen, the modulational instability can be found by deriving the nonlinear
dispersion relation <,:(k,a) and finding the sign of (Ov*/Dk) ' (0c.r/0a).
First, we note that the Alfven wave propagating in the direction of the ambient
magnetic field has no dispersion; therefore 0vr/0k = 0. Hence, to find the rnodulational instability, we must consider the finite frequency correction. The dispersion
relation as derived in Chap. 2 (F,q.2.24) then becomes

-F-;m;re
"'k' =

atpi

(6.8 r)

02 a:10k2 < 0. Now, to find Dl..if da, we must apply a

nonlinear perturbation theory. The perturbations should include amplitude to the
third order. The most elegant method for this is the reductive perturbation theory
developed by Taniuti and Yajima (1969). The reductive perturbation leads directly to
the nonlinear Schrcidinger equation of the form of Eq. 6.72.
Leaving the detail of the perturbation processes to a reference (Hasegawa,1972),
we show here the nonlinear Schrcidinger equation which describes the modulation
amplitude of the Alfven wave,

Hence we see that Dvr/Ok =

'(#.",9.+*#

gi
- ++Vo (w2 _rdor')

=o

(6.82)

We see here that 0co/0a1u=g = kvzp.l(4ue) > 0. Since Ovr/Dk < 0, we find that the
Alfven wave is modulationally unstable.
The nonlinear frequency shift 0c,.r/0a is closely related to the ponderomotive
force. The modification of the background plasma density due to the electromagnetic
wave pressure (ponderomotive force) leads to the local shift of the wave frequency.
For the Alfven wave, the wave energy is primarily magnetic, and the corresponding
ponderomotive force becomes negative; this means that ttre plasma is attracted to the
region of larger intensity of wave magnetic field (Kaufman, Cary, and Pereira, t978).
This leads to the modulational instability in the region of negative Dvr/Ok, which is
opposite the case of the well-known Langmuir wave, where the ponderomotive force
is positive.

89

REFERENCES

REFERENCES

Text Citations
Boley, F. I., and P. R. Irorman, 1964, Steepening of Large Amplitude Alfvdn Waves, Phys. Rev.

Lett.,12:

384.

Ferraro, V. C. A., 1956, Hydromagnetic Waves in a Rare lonized Gas and Galactic Magnetic Fields,
hoc. R. Soc., A233: 310.
Hasegawa, A., 19'72, Theory and Computer Experiment on Selt'-Trapping Instability of Plasma
Cyclotron Waves, Phys. Fluids, 15: 870.
and L. Chen, 1976, Kinetic Processes in Plasma Heating by Resonant Mode Conversion of
-, Alfv6n Wave, Phys. Fluids, 19: 1924.
and K. Mima, 1976, Exact Solitary Alfv6nWave, Phys. Rev. Lett., 3'l: 690.
-, and K. Mima, 1978, Anomalous Transparent Produced by Kinetic Alfvdn Wave Turbulence,./.

-,

Geophys. Res., 83: 1l 17.

-,

Fibers.

and F. Tappert, t9'13, Transmission of Stationary Optical Pulses in Dispersive Dielectric

II. Normal Dispersion, Appl. Phys. Lett.,23: 17l,

Kakutani, T., 1974, Plasma Waves in Long Wave Approximation, hog. Theor. Phys. (Kyoto),
Suppl,,55:97.
Karpman, V. I., and E. M, Krushkal, 1969, Modulated Wave in Nonlinear Dispersive Media,,Sou.
Phy* JETP (Engl. Transl.), 28:277.
Kaufman, A. N., J. R. Cary, and N. R. Pereira, 1978, Universal Formula for the Quasi.Static
Second Order Density Perturbation by a Cold Magnetoplasma Wave, inProceedingsof the 3rd
Topical Conference on R. F. Plasra Heating, Pasadena, Calif., Jan. 11-13, 1978, American
Physical Society.

of Hydromagnetic Shocks from Large Amplitude Alfv6n

36.
Sagdeev, R. 2., and A. A. Galeev, 1969, Nonlinear Plasma Theory, T. M. O'Neil and D. L. Book
(Eds.), W. A. Benjamin,Inc., New York.
Shaftman, P. G., 1961, Propagation of a Solitary Wave Along a Magnetic Field in a Cold
Collision-Free Plasma, J. Fluid Mech., l1: 16.
Taniuti, T., and N. Yajima, 1969, Perturbation Method for Nonlinear Wave Modulation,I,J. Math.
Phys., 10: 1369.
Yu, M. Y., and P. K. Shukla, 1978, Finite Amplitude Solitary Alfv6n Waves, Phys. Fluids,2I:
1457.
Zakhalov, V. E., and A. B. Shabat, l9'72, Exact Theory of Two-Dimensional Self-Focusing and
One-Dimensional Self-Modulation of Waves in Nonlinear Media, ,Sou. Phys. JETP (EnSl.
Montgomery, D., 1959, Development
Waves, Phys. Rev.

Lett.,2:

Transl.), 34: 62.

General References
Aleksin, V. F., and V. D. Khodusov,l973,On MagnetohydrodynamicWave Kinetics, lll., Ukr. Ftz.
ZIL , t8: l7 07 -17 16.
Barnes, A., 1976, On the Nonexistence of Plane-Polarized Large Amplitude Alfv6n Waves,,I.
Geophys. Res., 81 : 281-282.
*-, and G. C. J. Suffolk, 1971, Relativistic Kinetic Theory of the Large-Amplitude Transverse
Alfv6n Wave, Plosma Ph1ts., 5: 3t5-329.
Bate, R. T., and W. R. Wisseman, 1969, Second-Harmonic Generation by Damped AlfvdnWaves
and Helicons in Pure and Doped Bismuth, Phys. Reu., t8l:'163'7'13.
Belcher, J.W., and L. Davis, h., 197t, Large-Amplitude Alfv6n Waves in the Interplanetary
Medium. ll.,J. Geophys. Res.,76: 3534-3563.
Berezin, Yu. A., 1971, Nonlinear Motions in an Anisotropic Plasma, Zh. Eksp. Teor. Fiz.,6I:.
1877-1881.

T
90

NONLINEAR PROCESSES
Cohen, R.

I{.,

1975, Mode Decay and Evolution

of thc

Solar Wind Alfvtln Wave Spectrum,,/.

80: 3678-3680.
Dewar, 1974, On the Backscatter Instability

GeoPh1t* Res.,

and R. L.

of

Solar Wind Alfvdn Waves,

../.

-,

Geophys. Res., 79: 4l'1 4-4I'1 8.

Cramer, N. l.-., and R. G. L. Hewitt, 1975, Parametric Excitation of Compressional Alfv6n Waves,
Phys. Lett. A, 54A: 493-494.
Dokuchaev, V. P., 1970, The Theory of Magnetic Bremsstrahlung fiom Altv6n Waves in an
lncompressible Mediunl Magn Gidrodin., 2: 23-30.
Elachi, C., L9'13, Parametric Interactions Between Alfv6n Waves and Sonic Waves, IEEE Trans.
Antennas Propag., AP-21 : 9A7-909.
Elfinrov, A.G., and F. M. Nekrasov,l972, Contribution to the Theory of HydrodynamicPlasma
Stabitity in the Field of a Finite-Amplitude Alfv6n Wave, Nucl. Fusion, 12: 201-206.
and F. M. Nekrasov, 1974, Excitationof Alfv6nWaves in Decay Instabilities,Sov. Phys.-Tech.
Phys. (Engl. Transl.), 19: 9-1 I.
and F. M. Nekrasov, 1975, Nonlinear Saturation in Parametric Excitation of Alfv6n Waves,
Sot,. Phys.-Tech. Phys. (Engl. Transl.),20: 1300-1302.
F'u, K. Y., l9'13, Nonlinear Mode-Mode Coupling of Altvdn Waves in the Interstellar Medium,
Plnsma Phys., l5: 57-76.
Grebinskii, A. S., 1968, The Quasilinear Theory of the Growth of Alrven Waves ina Nonuniform
Medium, Sot,. Phys-Tech. Phys. (Engl. Transl.),38: 1190-1193.
Hollweg, J. V., 19'11a, Nonlinear Landau Daniping of Allv6n Waves, Phys. Rev. Lett.,27:

-,
-,

r349-r352.

-,
-,
-,

1971b, Density Fluctuations Driven by Allvdn Waves, J. Geophys. Res., 76:5155-5161.

1973, Wave Coupling Between Solar Wind Protons and Helium, Ark.Geophys. Union,54:

I 190.

and G. Skadron, 1975, Fokker-Planck Theory for Cosmic Ray Diffusion in the Presence of
Alfvdn Waves. I. Theory, J. Geophys. Res., 80: 2701-2707.
Hung, N. T., 1974, Parametric Excitation of Alfv6n and Acoustic Waves, Plasma Phys., 12:
445-453.

Inutake, M., and K. Kuriki, 1974, Lxperiments on Affvdn Wave and Shock Wave in a Plasma Wind
Tunnel, Bull. Inst. Spoce Aeronaut. Sci, Univ- Tokyo B, l0:799-817.
Ionson, J. A., and R. S. B. Ong, 1976, The Long Time Behavior of a Finite Amplitude Shear Al1v6n
Wave in a Warm Plasma, Plasnu Phys., 18: 809-819.
Kakutani, T,, l9'14, Plasma Waves in the Long Wave Approximation, Prog. Theor. Phys. (Kyoto),

Suppl.,55: 97-119.
Kaw, P. K., 1976, Parametric Excitation of Electromagnetic Waves in Magnetized Plasmas, .4dv.
Plasma Phys., 6: 207-236.
Kennel, C. F., and R. Pellat, 1976, Relativistic Nonlinear Plasma Waves in a Magnetic Field, Plasmo

Phys.,15: 335-355.
Lashmore-Davies, C. N., 1976, Modulated Instability of a Finite Amplitude AlfvdnWave,Phys.
Fluids, 19: 587-589.
and R. S. B. Ong, 19?4, Parametric Excitation of Alfv6n and lon AcousticWaves, Phys. Rev.

-,

Lett., 32: ll72-1t75.

Lee, M. A., and H. J. Volk, 1973, Damping and Nonlinear Wave-Particle Interactions of

--,

Alfvrr-Waves in the Solar Wind, Astrophys. Space Sci., 24: 3l-49.

and H. J. Volk, 1975, Hydromagnetic Waves and Cosmic-Ray Diffusion Theory, Astrophys.

"I., 198: 485-492.

Lehane, J.A., and F. J. Paolini, 1970, Observation of Parametric Amplification of Alfv6nWaves,
Phys. Lett. A, 32A:409-410.
Leonard, B. P., t972a, Magnetic Structure of lonizing Shock Waves. IIL Normal Shocks, Plasma
Phys., 7: 1'l'l-185,
1972b, Magnetic Structure of Ionizing Shock Waves. II. Oblique Shocks, Plasma Phys.,7:
| 57-r',l6.

-,

REFERENCES

91

1972c, Magnetic Structure of Ionizing Shock Waves. I. Skew Shocks, Plosma Phys., 7:
I 33-15s.
McKenzie, J. F., and M. Bornatici, I974, Effect of Sound Waves, Alfv6n Waves, and Heat liiow on
Interplanetary Shock Waves, J. GeophSts. Res.,79: 4589-4594.
Mikhailovskii, A.8., V. I. Pietviashvili, and A. M. Fridman,I976,The Alfvdn Soliton, JETP Lett.

-,

(USSR) (EngL Transl.),24: 53-56.

__, and O. A. Pokhotelov, 1975, Intluence of Whistlers and Ion-Cyclotron

Oscillations on
ExCitation of Alfvdn-Waves in Magntospheric Plasma, Fiz. Plazmy, l: 1004-1011.
Mio, K., T. Ogino, K. Minami, and S. Takeda, 19'16, Modified Nonlinear Schrodinger Equation for
Altv6n Waves Propagating Along the Magnetic Field in Cold Plasmas, J. Phys. Soc. Jpn., 4I
265-271.

Nguyen, H., 1975, Nonlinear Decay of a Large-Amplitude lon-Acoustic Wave, Plnsma Phys., 14
347

-35r.

K. Nozaki, Propagation of Solitary Pulses in Interactions of Plasma Waves. II.,l

Phys. Soc. Jpn., 36:591-595.
Patterson, ts., lg7l, Exact Nonlinear Evolution of Alfvdn-Mode in the Guiding CenterModel, Plzl's.
Fluids, 14: 1127.
Petrukhin, N. S., 19'74, Parametric Amplification of Altvdn Waves in the Solar Atmosphere,
Astron. Zh., 51: 571-576.
1975. On the Mechanism of Sound Pulses Generation in the Solar Atmosphere, Pis'mn Astron.
Ohsawa, Y., and

Zh., I: 32-36.
Pozwolski, A.,1973, Parametric Amplification of Altv6n Waves, Onde Electr., 53: 4L2-414.
Rogister, A., I971, Propagation of Alfv6n Waves in lon-Sound Turbulent Plasma, Plasmn Phys., 6:
309-323.
Saka, O., and T. Kitamura, 1975, Turbulent Spectra of the Transverse Alfv6n Waves in the
Corotating Solar Wind Structure, Rep. Ionosphere Space Res. Jpn., 29: I27-132.
Scholer, M., and J. W. Belcher, I97I, The Elfect of Alfvdn Waves on MHD Fast Shocks, Solar
Phys., 16: 472-483.

-,

Selim, A. A., and S. Krishan, 1971, Amplitude Dispersion in the Electromagnetic Modes, Phys.
Fluids, l4: I 166-1 170.
Sharma, S. K., and V. K. Tripathi,1972. Alfv6n Wave Second-Harmonic Generation in Bismuth'
Kinetic Theory, J. Appl. Phys. (Jpn.), 453:2155-2160.
Tamoikin, V. V., and S. M. Irainshtein, 1975, Stimulated Mandelstam-Brillouin Scattering on
Propagation of Alfv6n Waves in a Plasma with Random Inhomogeneities, Sov. Ph1ts. JETP
(Engl. Transl.), 68: 948-955.
--, and S. M. Fainshtein, 1976, Kinetic Equations for Sound and Allvdn Waves in a Plasma with
Random Inhomogeneities, Sov. Phys. JETP (Engl. Transl.),71: 531-536.
Tsytovich, V. N., and S. A. Kaplan, l9?2, Relativistic Turbulent Plasma in Pulsars, Astrophysics

(Engl. Transl.), 8: 44I-460.

Vahala, G., and D. Montgomery, 1971, Parametric Amplification of Alfvdn Waves, Phys. Fluids,

14:113'l-1140.
Valley, G. C., 1973, Scattering of Alfv6n Waves by Random Density Fluctuations (in the Solar
Wind), Am. Geophys. Union, 54: 1792.
Scattering of Alfvn Waves by Random Density Fluctuations, Astrophys. J. (Engl.

Transl.),
-,1974,

188: 181-189.
Williams, D. J., 1975, Magnetohydrodynamic Accretion and the lnstability of Smooth TransAlfvenic Flow, Mon. Not. R. Astron. Soc., I7I 53'l -549.
Yamagishi, T., 1979, Diffusion Coefficient of Low Frequency Electromagnetic Turbulent Plasma,
Phys. Fluids, 22: 2431,-2433.
Yuang-Chung, Chin, and D. G. Wentzel,I9'12, Nonlinear Dissipation of Alfv6nWaves, Astrophys.
Space ,Scl.,

16: 465-47 7 .

P.r.rMA HEATNc BY arrvdru wAvE

As we have seen in a homogeneous plasma, an externally applied electromagnetic field
resonance absorption at the Alfven resonant surface,
r = f o, given by

with frequency, c,le, faces spatial

G)o = t<,,(ro)

va(ro)

If

a proper choice of parameters is made, the absorption rate approaches 100%;i.e.,

the wave enerry is deposited into the plasma at the same rate as the frequency of the
wave. Plasma heating with this absorption has been proposed independently by
Grossmann and Tataronis (1973) and Hasegawa and Chen (1974). The absorbed
energy is shown to be converted to the kinetic Alfvdn wave. The plasma is heated
when the kinetic Alfvdn wave is dissipated by the particles by collision damping,
Landau damping, and/or nonlinear effects (Hasegawa and Chen, 1975).We derive the
rate of absorption, the heating rate of each species, and some design examples in the
following sections.

RESO NANT.ABSO

RPTION AND

ENERG Y.D EPOSITIO N RATE

As we discussed in Chap. 4, in an inhomogeneous plasma the Alfven wave has

continuous spectrum. If a monochromatic wave or an oscillating magnetic field is
applied externally to an inhomogeneous plasma, the wave energy is absorbed at the

RESONANT-ABSORPTION AND ENERGY.DEPOSITION RATE

93

resonant surface. The absorption rate can be obtained by using the ideal magnetohydrodynamic (MHD) equations. Ihis rate can be shown to be the same, even if the
kinetic theory is used, if the kinetic Alfven wave, which is excited by the mode
conversion, propagates into the plasma and is eventually absorbed by wave particle
interactions. Here we calculate the energy absorption rate, using a model boundary

condition for a conceivable plasma. We take planar geometry as shown inFig.7.l,in

which the inhornogeneous plasma is located in x ) 0. The current in the antenna coil is
represented by a surface current, J' at x = h, where X = Xo represents the Alfven
resonant surface. The plasma is assumed to be uniform in x2 a. The equilibrium
properties of the plasma [mass density (po), pressure (Po), and the confining magnetic
field (Be)] vary only in the x direction. Bs(x) has a shear component; i.e.,
Bo(x) = Bor(x), + Ber(x),

(7.1)

(In a toroidal plasma, x, y, and z correspond to radial, poloidal, and toroidal direqtion,
respectively.) Ps(x) and Bs(x) satisfy the equilibrium condition
d

dx

F'.$=o

Linearning the ideal MHD equat ions, including the compressibility effect, from
the equation of motion, we have

ttopE=-peV(p+86

Fig'

7.1 Variation of

B)+(B.V)Bo +(Bo .V)B

(7.2)

E as defined in Eq. 7.8 and the location of the surface current J, to excite

the kinetic Alfvdn wave for plasma heating.

PLASMA HEATING BY ALFVEN WAVE

94

where I is the fluid displacement vector and B, the perturbed magnetic field, is related
to t through the Maxwell equations
B = (Bo'V )E - (E 'V )Bo

Bo

(v ' )

(7 .3)

Combining the adiabatic equation of state and the continuity equation, we can express
the perturbed plasma pressure, P, in terms of f , as
p

--

(7'4)

E.* rPo(v't)

If we assume perturbations of the form

t(x) exp [i(krz + kry

@t)l

= a)o+ i6(6 = 0*),and adopt the local rectangular coordinates with ett = Bo/
lBe I and r = ll X 1, Eq.7 .2 then becomes, for each component,

with

c^r

ef' = ik't/o$

+ iB2ok'

" +

(it

it., s,.

"

**)

(7.s)

elr = ikrpoF

Q'6)

egx=Pr#

(7.7)

and

Here

e(x) = u2 tto.

p=p*

kfr Be

B' Bo
Iro

(7.g)

k,,(x) Bo(x) =krBo, + krBov

kr(*) Bo(x) = kyBo, - krBoy

Using Eqs. 7.3 and 7.4,we can express

tr(x) in terms of f*(x)

iakrBB dg"

tr(x)=Gfrr*

as

RESONANT-ABSORMION

AND ENERGY.DEPOSITION RATE

95

Here

a(x)=1+

a2

v?

fi(.' -

where v, is the sound speed. Substituting Eqs.

at the wave equation for f*:

(7.e)

kiu3l

.6 and 7.8 into Eq.7

#ffii$-ex=o

.7 ,

we then irrive

(7.10)

Equation 7.10 contains the shear Alfven, magnetosonic, and ion-aoodstic waves.
Owing to the nonuniformities, these three waves are coupled. Note that Eq.7.l0 has a
singular solution at the resonant point X = X0 where e = 0. This singularity then causes
the wave phase mixing, and hence the energy of the excited wave is dissipdted.
Equation 7.10 is a generalized form of Eq. 3.5 in that the effect of cornpressibility is
included. When vs -+ oo and k, = 0, this equation reduces to the ideal Alfvdn-wave
equation (Eq. 3.5). We also note here that, near the spatial resonance where e ry 0, Eq.
7.10 is reduced to
ek2rt* = 0

(7.rt)

and hence it recovers the structure of Eq. 3.6. This indicates that near the Alfvdn
resonance the effect of compressibility becomes irrelevant. In the subsequent analysis

we shall assume that the low-beta plasma is vr' (v2a; thus <ofr >vft<fr and
a(x)=1+(v3 luT)-o(1).
Lel us now obtain the energy absorption rate, dWdt. From the power
conservation law, dW/dt should be equal to the difference in power flow toward the
plasma before and after it reaches the resonance surface Xo. The total power flow, P,
in a cross-sectional area, S = LrL* is given by

, =\:,

Re(E x H* + vp*)

= \rL' R'

i@og*F

L,,L,e)o
6 0l*
! L--"_ rrrr.*
I_ l*
qlro A*
2
If we note that f*

has a logarithmic solution near x = Xs So thot

(7.12)

PLASMA HEATING BY ALFVf,N WAVE

96

(7.13)

*=Cln(x-xe+i6)

where C is a constant. Equation T.l2indicates that the power flow is discontinuous (but
not divergent) across the plane X = X0. Hence the absorption rate is obtained as

LrL'as

dw

dt

-%l*lX=Xo

un (x

Xo

* i6)ll'l
(7.14)

=+nL,L"#lHl.=..
where e, is the real part of e. The preceding expression shows that the absorption rate
is proportional to the gradient of e given in Eq. 7.8 and to lCl2 ; C is proportional to
the amplitude of the externally applied field intensity. For a design purpose, C should
be expressed in terms of driving field. To do this, we must solve the boundary-value
problem of Fig.7.1 and express C in terms of the current density, Jr, of the exciter
coil. This process is straightforward but tedious. If we use the process by Chen and
Hasegawa (1974), the result is

LrL, I

dw
o,=2ras

181(o)

Tili-r*tmi-

(7.1s)

with

= [Io(Xo)

I, (Xo)1

Ko(&)

+ K, (Xe)

lK,(xo) - Ko(xo)l
lr,(xo) + ro(xo)l

D1

\' - lkrlel
^o--T
lkrlell
a,
1\^ -oK

e(x)=ffxr6t

=eu

(0(x(a)
(x)

a)

Bessel functions of the first and second kind,

respectively, and B1(0), the magnetic field induced by Jr in vacuum, is

where

lrr(*) and Kn(x) are modified

VAVE

HEATING RATE

7.13)

Bi(o)

=+exp (- lkyl h)

(7.16)

(but

lt

can be shown that,

if the plasma nonuniformity

lXo l = lkrxe

lX"l= lkr(a

is weak so that rc/k,

l)

( l, then

- xs)l)

7.14)

y o 12r)'h(lkr l xe)-'^

rate
al to

ould

,*n(lkr l xs)

Then the absorption rate can be written

'alue

dWo

;iter

dt

and

rd,

LuL, B*(*o)'
tt"

to lirl

(7

.17)

This expression shows that, at each cycle of the wave, the magnetic-field energy given

by lB*(xe)l'lpo in the volume

efficient rate of absorption.

of IrLr lkyl-r is absorbed,

an indication

of avery

HEATING RAT

To obtain the heating rate of each species in the plasma, we must use the kinetic
theory. As was shown in Chap. 4, the absorbed energy due to the spatial resonant is
converted to the kinetic Alfvdn wave. Hence the heating rute of each species can be
obtained by the damping rate of the kinetic Alfven wave by the species (Hasegawa and
Chen, 1975,1976).
First, let us consider the collisional regime. Here the ion heating is dominated by
the viscous damping of the transverse component of the wave field. The heating rate is
given by

no

dT,

ai=ine(l ' E*)i

=

0J vi{k*o)2

= 0.9

uo

lE*l'

.fu

,?i

eo lEx l' .'o

zii(k*pi)-t "o f*' :+
2 rt2-

c1.

(k*pi <

1)

(k*pi > l)

(7.1s)

PLASMA HEATING BY ALFVEN WAVE

where zii is the ion-ion collision rate. Since k*, lE*12, and o:f,ilc':2"iarlallfunctions
of x, the heating rate varies as a function of the distance away from the mode
conversion point. However, the variations tend to cancel among each other, and the
heating rate remains roughly constant and is given approximately by

dTi lB,,(K- t )12

no
6 =uii 2F

(7.te)

where Br(x-t) is the value of the wave magnetic field at x = K-t-.8, is related to the
amplitude of the source field, Br' through the Airy function, as shown in Chap. 4,
rBr l^,

(*p)'-'a ( - o?-r)
i

tBsyt

(7.20)

lB.vl

(7.2r)

near the mode conversion region,

rBrl ^, (*pi)-h

(-iolri)

t
and away from the mode conversion region (at x = K - ). If we take a tokamak plasma,

K^or-l e lo-t pit

Hence the y component of the wave magnetic field can be enhanced by a factor of 100
near the mode conversion point.
The electron heating in a collisional regime is governed by the ohmic dissipation of
the field aligned current,

".

+= |*4r,r;y
=,.,+*(r - ree-N ir,

eo t*-*t'

*#

(7.22)

If we compare this heating rate with that of the ions in Eq. 7.18, we have

ffi-,ry-(S'(i,I;

(7.23)

For most tokamak parameters, this ratio remains of the order of unity. Hence, in a
collisional regime, electrons and ions are heated approximately at an equal rate.
However, electrons are heated in the parallel direction, whereas ions are heated in the
perpendicular direction.

HEATING RATE

In a collisionless regime, the linear heating occurs as a result of the l,andau

damping. Hence particles are heated in the parallel direction, no (dT/dt) o Re(J"Ef/
2). The Landau damping rate for trapped electrons by a local mirror is reduced by the
bounce motion. For simplicity, we assume the case of a straight magnetic field. The
heating rate for ions in a collisionless regime is then given by
", #

($'

e)olu,

to'r,

^t(r

rse-r.;;z

eollxl2

(7.24)

where
6

2(n)k 91Y'

i=

,*p (- pi t )

(7.2s)

and p1 is defined for the ion pressurc, Ai--2vlilv2g, at the resonant point. Other
quantities except <^r are a function of position. Maximum heating is achieved at tri * 1.
The heating rate for electrons in a collisionless regime is given by

no
where

dr" c)ooe^ Telrt

dt

-Iee-riyz
\ n,

eo lE*12

oii

o:?i

(7.26)

6, is given by
6" = 1n1'h pyk

In Eqs. 7.24 and7.26,

eo lEx

t2

lG:\ila!1)

($-(ff)-

(7.27)

can be approximately identified as the wave

magnetic-field enerry Bl I 2uo.

The ratio

of the

heating rate for ions and electrons in the collisionless regime

becomes

dTildt
dT"/dt

(l I 2)9il o e- r'i 1t"/t1)

6e

s'H- e-Fit

lfor),1

(l)

(7.28)

The factor 0i on the ion heating rate appears because the ion l,andau damping is
possible only through the coupling to the ion acoustic wave. This ratio is negligibly
small for a low-beta plasma; i.e., only electrons can be heated in a low-beta plasma.
However, when 0i approaches unity, the ratio becomes of the order of unity. For
example, if 0i = A.2, TelTi = 2 makes this ratio unity, and the corresponding value of
6e becomes 0.13. Because of the trapped-particle effect, this value of 6" is an
overestimate. The true damping rate will probably be an order of magnitude less.
Let us now consider a nonlinear regime. Because the kinetic Alfvdn wave has an
enhanced amplitude as a result of the resonant mode conversion, even with a

PLASMA HEATING BY ALFVEN WAVE

lOO
reasonably small amplitude

of the external source field, nonlinear

expected to take Place.

As was discussed in Chap. 6, if

Te

processes are

} Ti, the kinetic Alfven wave decays

acoustic wave' If T" = 1i,

parametrically into another kinetic Alfven wave and an ion
of
it decays into another kinetic Alfvdn wave through the nonlinear Landau damping
is roughly
the ions. In either case, the ions arc heated, and the heating rate
6'57
or
6'54
Eq'
in
given
proportional to the decay rates
'

dTi

(7.2e)

2po
dt -tBl'

where T = ?o(Te

> TJ or 7y(T, 3 TJ, with

^,D*?ili:l
and
@ci tr2

}N

lgrlt aci

=-tr p lB"l

QA

where

I > tr(= tip3)

= max(tr0,trA,trr)

DESIGN EXAMPLES

kinetic
In this section we present some design examples of the application of the
ohmic
the
Since
plasmatqkamak-type
Alfven wave for supplementary heating of a
to
is
needed
heating
supplementary
a
keV,
heating saturates at a temprrutu* n.ui 1
have
methods
Several
keV.
10
of
increase the temperature to a fusion temperature
conclusively to be the best
been proposed for this purpose, but none has been proved
dimensions: Density of
following
the
has
choice. A reactor tokamak typically
a
(a), 3 m; and Saior
radius
minor
kG;
(Bo),
70
electrons (n"),2 x l0r ; toroidai nttO
2 x 103 m3 and
approximately
of
volume
plasma
radius (R), 12 m. This gives the
energy
expected
the
within
plasma
a
such
output power of 3 GW. To heat
heating of
supplementary
the
of
po*er
input
the
need
we
containment time of l0 sec,
that we need a power
200 MW. If we subtract heating efficiency, this number indicates
requirement for a
severe
is
a
rather
S}%.This
of
source with an efficiency of the order
power source with 200 MW.

DESIGN

EXAMPLES

101

The Alfven wave, whose frequency range is of the order of I MHz, is attractive in
this respect because a power source of this magnitude and efficiency is currently
available.
Plasma is heated when the kinetic Alfvdn wave is absorbed by the plasma particles.
Supplementary heating has meaning only in the collisionless regime where the ohmic
heating can no longer operate. In this regime Landau damping, or nonlinear
wave-particle interaction, is the important absorption process.

Since the linear Landau damping, which effectively operates only on electrons, is
expected to saturate quickly owing to the plateau formation of the distribution
function, the nonlinear heating is probably the most important process. There ions can
be heated by the nonlinear ion Landau damping. The threshold value of the
normalized wave amplitude, BriBo, is given by

*,
r1i(0.tts;
us2.,

Here we have assumed that the electron Landau damping has been saturated by the
plateau formation; it.nce only the ion viscous damping, ziitri, is a dissipative process.
The ion-ion collision frequency, uii, for kiloelectron volt plasma with a density of
5 x l01t cm-3 is approximately 103 sec-r. Hence the threshold is given by
(

tsr(G)

If

we use

crs])'h

=;-

MHz, the threshold amplitude

of the kinetic Alfven wave becomes

approximately 50 G. This is quite a reasonable value if we recall that the amplitude of

the kinetic Alfven wave is enhanced, owing to the resonant mode conversion, by a
factor of (pra)-% 30 compared with the externally applied field.
^,

If the kinetic Alfvdn wave dissipates completely as it propagates through the

plasma column, the heating rate is given by the absorption rate of the externally
applied oscillating field by the resonant absorption. This means that the damping rate
per cycle should be larger than the wavelength (-p*) over the minor radius of the
column, which, for the example shown here, is l0-3. The absorption rate is given by
dwov
dt

Bg(x^)2
jftu

V is the plasma volume and Bf(xs) is the value of the magnetic field at the
location corresponding to the resonant surface in the absence of the plasma. This
where

relation shows that, to provide 200 MW to the plasma, we need only 4 G of oscillating
magnetic field. Inside the plasma this field will be enhanced to about 100 G and will
be well beyond the threshold value for the nonlinear Landau damping.

102

PLASMA HEATING BY ALFVfN WAVE

The oscillating radio-frequency (r-f) field with a typical frequency

of I M[Iz is

launched by a coil that is placed at the wall and is helically wound with respect to the
toroidal magnetic field. The ohmic (skin) resistance, R, of the coil is given by

*_nrd.
where d and / are the width and length, respectively, of the coil and R, is the skin
resistance of -10-a ohm for copper and-10*3 ohm for tungsten.
On the other hand, the radiation resistancg, Rr, is given by

Rr-#ur,N(ioL
where

L is the coil inductance.

Hence the heating efficiency of the coil, 4, is given by

n=-

"

1+

(R/Rr)

where

R1
Rr 2n x l0' d(m)
even

for a tungsten coil.

If d = 0.1 m is used, R/&

is much smaller than

and 4 is approximately equal to

of the coil itself is given by <..roL/R ev 2r x 102 d. The Q value of the

radiation impedance, is of the order of unity because of the high
including
the
coil,
1. The Q value

efficiency of the resonant absorptio4.

The induced terminal voltage, V, of the coil is given by

V = c^lo LI

To produce a lO-G amplitude of E1(Xo) at I MHz with a coil of l-m length, the induced voltage becomes V

100 V.

The excitation by a coil has a merit in that it can be designed so that the
magnetosonic wave is cut off by choosing kyvl(=kova)) a, whereby one can
suppress the unnecesspry compressional motion of the plasma. However, it has
demerits that originate from the installation of a large coil inside the reactor chamber.

An alternative method is to use a waveguide or a smaller antenna coupling to

excite a magnetosonic wave that is resonantly absorbed at the Alfven layer as shown

EXPERIMENTAL

RESULTS

103

by Eq. 7.l0.lf a plasma can tolerate the compressional motion of the magnetosonic
wave, this method can eliminate the problem of the large antenna (Ott, Wersinger, and
Bonoli, 1978; Kearney and Perkins, 1978).
EXPERIMENTAL RESULTS
We introduce four major experiments of Alfven-wave heating: (1) Theta-pinch
plasma
plasma

in Lausanne, Switzerland; (2) stellarator plasma in Wisconsin; (3) stellarator

in Sukhurni, Russia; and (4) heliotron plasma in Kyoto, Japan. In all these

experiments, efficient absorption and heating have been observed;however, there seem
to be discrepancies in the existence of enhanced loss of plasma.

Theta Pinch
Alfven-wave heating on linear theta-pinch plasma has been studied by Keller and
Pochelon (1978). The pinch configuration and the helical launching structure are
shown in Fig. 7 .2. The theta coil is 142 cm long and has an inner diameter of 9 cm.
The main field reaches 16kG in 3.8psec, at which time it is crowbarred. Thequartz
discharge tube has an inner diameter of 5.2 cm. Two electrodes,l.2 cm in diameter,
are 142 cm apart. The typical plasma parameters are: percentage of ionization, l5%;
mean plasma radius, A.76 cm; mean electron density, 2x l0l6 cm-3;and mpximum
plasma ternperature (before heating), 40 eV. The helical launching structure has an
m= I configuration. The energy source is a 45-kV,0.1-trrF, low-inductor condenser.

Fig. 7.2 Alfvdn-wave heating experiments on the lausanne linear theta pinch.
[From R.
and A. Pochelon, Alfvdn Heating of a Theta Pinch, Nucl. Fusio,n, l8(8): 1052 (197g).1

m=lcoil
Part of the electrostatic shielding

3
4

(insulation not indicated)

Diamagnetic coil
Theta coil

Viewing port

5
6

Quartz tube

Electode

feller

104

PLASMA HEATING BY ALFVEN WAVE

Coupled with the coil system, the ringing frequency is 1 MHz. The temperature and
heating power are measured by diamagnetic probe. Figure 7.3 shows the increase of
temperature (T) and plasma beta (0) as functions of time. The heating efficiency
deduced from the experimental data is approximately 50%. Evidence of resonant
absorption is observed.

>40
o
ui

tso

0.6

E.

t.Ll

LU

o-

t!

r-

0.4 >

20

tU
co

0.2 e
Ll,

TIM E, psec

Fig.

7.3

absence

Plasma temperature and mean beta value as functions of time in the presence (o) and
(o) of the Alfvdn wave. IFrom R. Keller and A. Pochelon, Alfvdn Wave Heating of a Theta

Pinch, Nucl. Fusion, 18(8): 1055 (1978).1

Proto-Cleo

Alfven-wave heating on Proto-Cleo plasma at Wisconsin has been reported by

Golovato, Shohet, and Tataronis (1976). The Proto-Cleo is operated as an / = 3,7 field
period stellarator with a major radius of 0.4 m and an average plasma minor radius of
5 cm. The plasma density is approximately 1012 cm-3, and the electron temperature
is of the order of 10 to 20 eV. The poloidal field is 3 kG, and the energy confinement
time is I msec. The experimental setup is shown in Fig. 7.4. The rf coil is the helical
winding that goes around the major axis of the torus three times in going around the
minor axis once. It is inside the higher pitched confining field windings. A generator
operating up to 200 kW at a pulse length of I msec was used to launch the rf. Figure
7.5 is the plot of the radial electron-temperature profile immediately after rf was cut
off, normalized to the temperature profile at the same time without the presence of rf.

105

E)GERIMENTAL RESULTS

ALUMINA
H E LICAL
MAGNETIC-

SUPPORT
RINGS

FIELD COILS

LICAL
LAUNCHING

HE

STRUCTURE

RF

FEED

Fig. 7.4 Experimental setup of the Wisconsin Proto{leo stellarator. [From S. N. Golovato, J. L.
Shohet, and J. A. Tataronis, Alfudn Wave Heating in the Proto{leo Stellarator, Phys. Rev. Lett.,

37(19): 1273 (1976).1

LL

E.

F
l
o
I
F
ulB
EuJ
)g6
Fr

e.i
o--

e. o,

Fr>
F

g-1

zlO7

tro

5r
u-lb
di
LU

LL

E.

L
=

RADtUS, cm

Fig. 7.5 Change

heating.

of

temperature as a function

of radial position in

the

Proto{leo Alfvdn-wave

fF106

PLASMA HEATING BY ALFVEN WAVE

Notice that several peaks are observed. These peaks roughly correspond to the spatial
resonant surfaces if we assume one-dimensional theory. Both the electron and ion
temperatures are reported to be doubled. One unique feature of this experiment is the
observation of enhanced transport owing to the applied rf field. The anomalous
transport varies approximately linearly with respect to the amplitude of rf and is
believed to be caused by magnetic island formation.
R.02

The Alfven-wave heating experiment on R-02 at Sukhumi is interesting because of

the high-intensity rf field (Demirkhanov et a1.,1977). The schematic diagram of R-02
and its coupling coil is shown in Fig. 7.6.
Efficient heating has bee4 found to take place when the plasma density is made
larger than 5 x 10lt cm-3, at which density the Alfven wave can be coupled into the
plasma. Figure 7.7 shows the clear nonlinear aspect of the heating as a function of rf
magnetic-field intensity. It is also interesting that the temperature reached was almost
20 times as great as the initial temperature. In actual numbers the ion temperature
increases from 20 to 250 eV. The nonlinear threshold of the heating takes place at the
rf magnetic-field amplitude of about 50 G. One remarkable aspect of this experiment
is that the fluctuating field observed by loop voltage has disappeared, and better

TRANSFORMER

CT3-1
PHOTOMU LTIPLIE

ROGOWSKY COIL

BOLOMETER

Ho

FTELD

COILS

HE LICAL
WINDING

l=2
DIAMAGNETIC GAGE
Fig.

7,6

Experimental setup of Sukhumi R-02 sbllarator. (From Demirkhanov et al., 1977.)

107

EXPERIMENTAL RESULTS
18

'16

14

";I

;12
0)

lr)

io
X

10

rlij

u)o
F
uJ

20

40

60

100

80

RF INTENSITY,

120

140

Fig.7.7 Diamagnetism which indicates the plasma temperature as a function of Alfvdn-wave

intensity. A clear nonlinear heating is shown. q l l kG. q 5 kG. (From Demirkhanov et al., 1977 .)
confinement is achieved when the Alfven wave is coupled in the higher density phase
of the discharge. The better confinement in the presence of the large-amplitude Alfven
wave may be understood in terms of the negative ponderomotive effect, as discussed in
Chap. 6. Unlike electrostatic ponderomotive force, magnetic portderomotive force
tends to attract plasma into the high-intensity region of the rf field. If this is proved to
be the case, the Alfvdn-wave heating has a very attractive feature in that the applied rf
field works to heat as well as to confine the plasma.
Heliotron-D
Alfvdn-wave heating of the Heliotron-D has been reported by Obiki et al. (1977).
The machine has a major radius of 1 .l m, a minor radius of 10 cm, a toroidal field of 2
to 3 kG, and an electron density rangingbetween 3 x 1012 and 3 x 1013 cm-3. The
plasma is Ohmically heated before the Alfven-wave heating is applied. A schematic
diagram of the Heliotron-D is shown in Fig. 7.8. The initial experimental data tend to

PLASMA HEATING BY ALFVEN WAVE

108

MONOCHROMATOR
HIGH-SPEED CAMERA

TOROIDAL COIL

ELECTRON TEMPERATURE
(LASER SCATTERING)

MAGNETIC PROBE

RF COIL
ION TEMPERATURE
(

DOPPLE

BROADENI NG)

IIELICAL
COIL

PLASMA DENSITY
(4-mm MICROWAVE)
MONOCHROMATOR

0
Fig.

7.8

1m

DIAMAGNETIC LOOP

Experimental setup

Wave Heating Experiment

of

the Kyoto stellarator Heliotron-D. [From T. Obiki et al., Alfvdn

in the Heliotron-D,Phys.

Rev.

Lett.,39(13): 812(1977).1

show successful heating but poor confinement of electron heat. However, most recent
results seem to indicate good heating and good confinement, as shown in Fig. 7.9.The
improvement in confinement is achieved by better shielding of the rf coil. One
remarkable result of this experiment is the clear threshold of efficient heating as the
function of initial electron temperature, as shown in Fig. 7 .lO. As was shown earlier in
this chapter, the heating occurs owing to the dissipation of the kinetic Alfven wave
which is excited at the resonant surface. However, the kinetic Alfven wave propagates

toward the plasma center only when the electron thermal speed is larger than the
Alfven speed at the resonant surface. Otherwise, the kinetic Alfven wave propagates
toward the plasma edge and heats only the plasma surface (see Eq. 2.44').In addition,
when the electron thermal speed is larger than the Alfvdn speed, electron Iandau
damping and the consequent heating of electrons can take place. This threshold is

in Fig. 7.10; the temperature increases only when this condition is satisfied. If
the kinetic Alfv6n wave cannot be excited, the heating takes place only locally at the

shown

resonant surface. Therefore, having an initially large-enough electron temperature so

that its thermal speed is larger than the Alfven speed is essential for a bulk heating.
Among three toroidal experiments introduced here, the Wisconsin results fall into the
case of surface heating. The enhanced plasma transport, which seems to exist only in
the Wisconsin experiment, may be related to this point.

109

EXPERIMENTAL RESULTS

O WITH RF
O WITHOUT

RF

-uJ
ia

Y:

100

qL

hk

Ecc
tlJ
__.1

tl.l3
LU

Fo
10

RADIUS, cm

Fig.7.9 Electron tcmperature profile with and without Alfveh wave as a function of radial
position and time (Obiki et al., 1977).
150

100

0)
L

He

k50

o
o
Plasma density,

6 to 10

100

2.9 kG
2.6 kc

"

1618

r-3

200

ELECTRON TEMPERATURE, EV

Fig.7.l0 Heating threshold as a function of the initial electon temperature in the Kyoto
Heliotron-D. [From T. Obiki et al., Alfvdn Wave Heating Experiment in the Heliotron'D, Phys.
Rev. Lett.,39(18): 815 (1977).1

r
I

PLASMA HEATING BY ALFVEN WAVE

110
REFERENCES

Text Citations
Chen, Liu, and A. Hasegawa, 1974, Plasma Heating by Spatial Resonance of Alfvdn Wave, Phys.
Fluids, l7: 1399-1403.
Demirkhanov, R. A., et al., 1977, Plasma Heatingin a Toroidal Systern by a Helical Quadrupole RIr
Field with c,.r ( <rBi, in Plasma Physics and Contolled Nuclear Fusion Research, 1976,
Symposium Proceedings, Berchtesgaden, Oct. 6-13, 1976, pp. 3l-37, STI/PUB/439, Vol. 3,
International Atomic Energy Agency, Vienna.
Golovato, S. N., J. L. Shohet, and J. A. Tataronis, 1976, Alfvdn Wave Heating in the Proto-Cleo
Stellarator, Phys. Rev. Lett.,37: 1272-1274.
Grossmann, W., and J. Tataronis, 1973, Decay of MHD Wave by Phase Mixing II, Theta Pinch in
Cylindtical Geometry, Z. Phys., 261: 2L7 .
Hasegawa, A., and L. Chen, 1974, Plasma Heating by Alfvdn Wave Phase Mixing, Phys. Rev. Lett.,

32:454.
and L. Chen, 1975, Kinetic Process of Plasma Heating Due to Alfvdn Wave Excitation,Phys.

-, Rev. Lett.,35: 370.

and L. Chen, 1976, Kinetic Processes in Plasma Heating
--, Alfvdn
Wave, Phys. Ftuids, 19: 1924.

by Resonant Mode Conversion of

Kearney, C. F. F., and F. W. Perkins, 1978, Alfvdn Resonance Heating via Magnetosonic Modes in
Large Tokamaks, Report PPPL-1471, Plasma Physics Laboratory, Kyoto University, Japan.
Keller, R., and A. Pochelon, 1978, Alfvdn Wave Heating of a Theta Pinch, Nucl. F'usiozr, 18(8):
1 05 1-l 057.
Obiki, T., T. Mutoh, S. Adachi, A. Sasaki, A. tiyoshi, and K. lJo, 1977, Alfvdn-Wave Heating
Experiment in the Heliotron-D,Phys. Rev. Lett.,39:812 (197'l):in.Plasma Physics and Controlled Nuclear Fusion Research, L976, Symposium Proceedings, Berchtesgaden, Oct.6-13,
I976, STI/PUts/439, International Atomic Energy Agency, Vienna.
Ott, E., J.-M. Wersinger, and P. T. Bonoli,l978, Theory of Plasma Heating by Magnetosonic Cavity
Mode Absorption, Report LPS-236, Laboratory of Plasma Studies, Cornell University.
Pochelon, A., and R. Keller, 1977, Alfven Wave Heating Experiment of a High Ileta Plasma,Helv.
Phys. Acta, 50: 172.

General References
Alazraki, G., and P. Couturier, 1971, Solar Wind Acceleration Caused by the Gradient of Alfvdn
Wave Pressure, Astron. Astophy*, 13: 380-389.
Appert, K., B. Balet, R. Gruber, F. Troyon, and J. Vaclavik, 1980, Optimization of Resonant
Absorption of Alfvdn Waves in Low-p Plasmas, in hoceedings of the International Conference
on Plasma Physics and Controlled Nuclear Fusion Research, Brussels, Belgium, July 1-10,
1980, International Atomic Energy Agency.
et al., 1980, ResonantAbsorptionof AlfvdnWavesinFatToriwithCircularCross-Section,in
-, Proceedings of the 2nd Joint Grenoble-Varenha International Symposium on Heating in
Toroidal Plasmas, Corno, Italy, Sept. 3*12, 1980, International School of Plasma Physics/
Comissariat a I'Energie Atomique.
Balet, 8., et al., 1979, Studies of Resonant Absorption of Altvdn Waves by Means of An Evoiution
Code, in hoceedings of the 9th European Conference on Controlled Fusion and Plasma
Physics, Oxford, England, Sept. l7-21,I9?9, p. 170, European Physical Society.
Hasegawa, A., and L. Chen, 7976, Parametric Decay of "Kinetic Alfvdn Wave" and lts Application
to Plasma Heating, Phys. Rev. Lett., 36:1263-1265.

REFERENCES

Hollweg,

J' v.,

fi1
1974, Alfvdnic Acceleration

Theory, J. Geophys. Res., 79: 1357-1363.

Lacombe, C., 1976, Contribution

of

Solar Wind Helium and Related phenot'nena. l.

of the Non-Linear Landau Damping of AlfvCn Waves to tlte

Heating of the Solar Wind protons, Aston. Astrophys.,4g: ll_17.
Piddington, J - H., L973, Solar Atmosphere Heating, Sot. Phys. (Netherlands),
33: 363-373.
Tataronis' J' A., and W. Grossmann, 1976, On Alfvdn Wave Heating and
Transit Time Magnetic
Pumping in the Guiding-centre Model of a plasma, Nucl. F-usion, 16:
667-67g.
Uchida, Y', and o' Kaburaki, L974, Excess Heating of Corona and Chromosphere
Above Magnetic
Regions by Non-Linear Alfvtin Waves, Sol. ph1.ts. (Netherlands/, 35:
4SId6G.
Wentzel, D. G', 1974,Coronal Heating by AlfvdnWaves, Sot. Phys. (Netherlands),39:129.140.

T-

APPENDIX:
[,sr oF sYMBor.s
B
c
cs
D
X-,k)
E
E1
e;e
F
f(x,v,t)
Im
In
i
J
Jn
Kn
k
kp
L
I
rns
mi
n

Magnetic-field vector
Speed of light in vacuum ;2.998
Ion sound speed, v1"(m"/mi) h

x 108 m/sec

Diffusion coefficient
Dispersion relation;D(<^,r,k) = 0
Electric-field vector
Fourier amplitude of an electric-field vector with the wave vector k
Electron charge, 1,602 x l0-re cclulomb; base of natural logarithm
Force

Distribution function in

v-x

phase space

Imaginary part

Modified Bessel function of the first kind

(-

1)

Current density
Bessel function of the first kind
Modified Bessel function of the second kind
Wave vector

Debye wave number;

Lundquist number
Attenuation distance

uplvr

Electron mass; 9.107 x 10-' t kg

Ion mass
Number density of particles;harmonic number
112

LIST OF SYMBOLS

o(*)

113

Order of x in magnitude
Principal value

Pressure

F'

r
R
Re

t
V
v
Vo
b

vp

Total pressure [p + (Bo . B/po)]

Charge of a particle (e for proton,
Radial coordinate

-e for electron)

Radius
Real part
Temperature (energy unit)
Time
Plasma volume

Velocity
Group velocity, 0l,ol0k
Phase

velocity, c,r/k

vA

Parallel speed; in cylindrical coordinates v,

Alfvdn speed; cra"if api

vD

Drift

vd

Diamagnetic drift speed

MHD sound speed

v11

vs

vT
V1

w
x
x
v
Z
z
p
^tk

V
6ij

6(x)
q

e;e(co,k)

speed

Thermal speed;

(Tln)h

Perpendicular speed in cylindrical coordinates with z 1186 ; (vl +

Kinetic energy; m(vl + villz
Position vector
Coordinate
Coordinate
Plasma dispersion function
Coordinate
Pressure ratio of plasma to magnetic field; n6T/(B2o/2po)
Growth rate of a wave with the wave vector k
e* (0/0x) + eu (010v) + @lAz)
",
Kronecker symbol; = I (i = j), = 0 (i +j)

Delta function
Permittivity of free space; 8.854

rl

Equivalent dielectric constant

Resistivity

Angle between B and k

Measure of density gradient;

),

Wavelength

Lp
p
l/o
Uo$

!ei

k2

x l0- t' Fl^

-V(ln ns)

vi rlaZ

Debye waveiength; v7 I u:o

Adiabatic invariant ; mv] I pnol
Permeability of free space;4n x 10-7 H/m

Collision frequency of particle c to particle p

Collision frequency of electron to ion

v2rS'h

7
I

'|-14

LIST OF SYMBOLS
Reduced spatial coordinate vector

E
g

Plasma displacement;0f/0t = v
Mass density

L,armor radius

Conductivity
Equivalent conductivity; -icoe6

o(<o,k)
T

Mean free tinre

Reduced time

Amplitude of

e(<^r,k)

Q6

a modulated wave
Electrostatic potential
Parallel potential
Solution for <^r for a given value of k of a linear dispersion relation
Plasma (angular) frequency ; (e2 ns I esm)k
Cyclotron (angular) frequency i eBs /p
Drift-wave (angular) frequency; krlr V (ln no )/c^.rc
Alfven wave (angular) frequency

Ensemble average

Superscript *

Complex conjugate

a
tl)

@k

Qp

(rc
Q)a

Subscript e
Subscript i
Subscript i
Subscript p
Subscript r

Subscript

Subscript 0
Subscript 1

Subscript

ll

Electron

Ion (proton)
Imaginary part
Plasma

Real part
Surface

Unperturbed (d-c) quantities

Perpendicular to the magnetic field
Parallel to the magnetic field

APPENDIX:

EQUAT'ONs USED

'N

IH'S BOOK
MAG NETO HYD RODYNAMIC EOUATIONS

#*o.(nv)=6
*,n#=Jx B-Vp
E+vx B=?J

dp=Ip
dn 3n

vxE=-#
V x B=1,roJ

V. B=0

115

EQUATIONS USED

116
V

LASOV-MAXWEL

IN THIS

t EOUATI O NS

*P*

. ut(i*n,t) 1-9- 1n1x,t)

+v

x B(x,t)l .

$=

vxE=-#
vx

B=

p0.",

(l:vri

dv

- I:_ *" o) -+

V. B=0

v.E=# Il_,,u"_-,[:

r.dv

DRI FT KINETIG EOUATIONS

*-v.(vprp)-*(*r)
fp

=o

= fp(v1g,It,X,t)

VD=VDr*vntt
no,, =

vDl=na+

FxB
Ofl,

nu=#
FL-

-!vrB -

mv?
R-i!

^ mfdve
R-

F1=-lrV1B+qE,,

1 mB

}|

BOOK

PLASMA DIELECTRIC

TENSOR

117

PLASMA DIELECTRTC TENSOR

k(k. E,)-krE,

?=-,,e,SF."[

-5(l+ ?).

Er =o

[o"

T]

xy;
/n'"
e

S=

\'

vlJnJ'n

\ k, '"/

-i?v1JnJl,

(urJl,),

-iv1vxJnJl.,

ff u',t1

ivlvxJnJl,

(urJn)'

The argument of the Bessel functions, Jn, is k1v1/oq,

ff

c.rc

u,,t;

In MHD limit:

.**=r*[('-+)')-\91
J

xy=-6yx=i)*( -ry)

xz=zx=T**t#.(*f *))]

T*(#/.EP')

= (qBo/m), k = (k1,0,kr),

and Bs = (0,0,Bo).

vv=xx-

USE D

EQUATIONS

118

y, =

- zy= - i I

*,(ry-

fr

IN THIS

BOOK

*t, ,t)

Ezz --

where

,i=.[_**o',,I]

2m,dv,**#i

(rmc^,

> o)

and

q, = ,f
-

u,

11

.f

2rv,

dv

,vn

f6i(vr,v

n)

PLASMA DISPERSt0N FUNCTt0N

7K)=#[:-fr.,.

rorlmr>o

= analytic continuat ion of the above integral for Inrf < 0

-+) <r;
-r' - -+) cn>r;
lt

Z(l) =i(zr)% e- ,"

z(9 *

i(n)h oe

- 2f (

o=0

Imf > 0

=l

Imf=g
Imf<0

_,)

(rrr

U
-

APPENDIX:

SIUDTES OF NON,DEALIZ,NG PARATI,IEIERS

@TAGRAMMAT/iZED)
INHOMOGENEOUS MAGNETIC
FIELDS CONTINUUM
SPECTRUM 11972!.
PLASMAS @ <<

ect

YIELDS THE ALFVEN MODES

SHEAR AND COMPRESSIONAL (1950}

DISSIPATIVE EFFECTS
DAMPING (1949)

HALL CURRENT EFFECTS

,/, {FINITE

THE "IDEAL" AL rveru

wnve g

// a*vtN

FREOUENCY

wAVE)

DlsPERsloN (1960)

(1942)

FINITE LARMOR
RADIUS EFFECT.S
(KINETIC ALFVEN WAVE)
OF CURRENT INTEREST
THE CORIOLIS FORCE SPLITTING
OF THE NORMAL MODE INTO LEFT
AND RIGHT CIRCULARLY
POLARIZED WAVES (1954)

119

AUTHOR

'NDEX*
Adachi, S., (107), (108),

(109),110
Alazraki, G., 110
Aleksin, V. F., 89
Alfv6n, H., (1), (10), 11
Allen, T.K., (25), 30
Altukov, P. D., 13
Amte, P. C., 11
Aplers, W., l3
Appert, K., (42), 42, (45),

60,110
Ashkinadze, B. M., 13
Astrcim, E., 3l
Baker, W. R., 30

Baldwin, D. 8., (45),60

Balet, B., (45), 60, 110
Barnes,

A., 89

Barston, E. M., (34),42

Bate, R. T.,89
Bateman, H., (51),60
Baune, W., 13

Baynham, A. C., 13

Beckers, J. M., 72
Belcher, J. W., 12, 90, 9L
Beletskii, N. N., 13

Bellyustin, N. S., 31
Benezin, Yu. A.,66, 90
Bizli, K. O.,67
Boatdman, A. D., l3
Boley, F. t., (23), 31, (68),
89

Bonoli, P. T., (103), 110

Bornatici, M., 91
Brennan, M. H.,61
Briggs, R. J., (34),42

Brown, I. G., (27),30

Burchsted, R., 12

Chen, F. F.,31
Chen, L., (41),42, (45\, (47),

(54), 60-61 , (73),89, (92),

(96), (97), I 10
Cohen, R. H.,90
Coleman, P. J.,61
Cooper, W. S., III, (25), 31
Countee, C.,12
Couturier, P., 110
Cowling, T. G., 11
Cramel, N. F., 11
Cross, R. C., 31
Cummings, W.D.,
Cuperman, S.,67

Burlaga, L. F., 12
Caille, A., 13
Callen, J. D.,66, (66)
Cary, J. R., (88), 89
Chandrasekhar, S., (65), 66

Chao-Kung,Yang,67
Chen, F. C., 13

*,Numbers in parentheses indicate the references cited in the text.

124

12,6t

AUTHOR INDEX
DeSilva, A.W., (23), (25),
Deutsch, R. V., 42,61
Devanathan, C.,32
Dewar, R. L.,90
D'Haenens, J. P., 13
Dobrowolny, M., 67
Dokuchaev, V. P., 90

121

3t

Hasegawa, A., (41), 42, (44),

(45), (47), (54),60-67,67 ,

(69), (73), (80), (87), (88),
89, (92), (96), (97), 110
Heacock, R. R.,67
Herlofson, N., 11, (14), 31
Hermann, R., 13

Krishnan, S., 91
Kristiansen, M.,31
Krushkal, E. M., (86), 89
Kruskal, M., (47), 60

Kuka, G., 13

Kuriki, K.,90
Kutnik, A. E.,

11

Dwarkin, M.L.,12
Dynes, R. C., 13

Hess, H. D., 13

Egorenkov, V. D., 67
Elachi, C., 90

Hollweg, J. V.,

Elfimov, A. G.,90

Holmes, J. A., tz
Hu, P., 13
Hung, N" T., 90

Irhnert, B., (8), (10),

Ignat, D.W., (45),60

Leonard, B. P., 91

Elsasser,

K.,67

Erdelyi, A., (51),60

Fahidy, T.2.,13
Fainshtein, S. M., 91
Fejer, J. A., 31
Ferraro, V. C. A., (5). 11,

(68), 89
Fieldman, W. C., 67
Forman, P. R., (68), 89

Fornwald, F.,42
Forslund, D. W., 67
Fridman, A. M.,91
Friedman,8., (50), 60
Fu, K. V.,90
Fukunishi, H., (41),42
Furdyna, J. K., 13
Gajewski, R., (33),42
Galeev, A. A., (68), (73),

(74),89
Gary, S. O.,67

Gerwin, R.,60
Glushkov, M. V., 13
Goedbloed, J.P.,42
Golant, V. E., (45), 60

Golovato, S. N., (104), (105),

110
Grad, H.,34,42,(42)
Grebinskii, A. S., 90
Grossmann, W.,45, 60-6 1,

(92),

tl0,

111

Gruber, R., (42), 42, (45),

60,110
Gustafson, W. A.,32
Guthmann, C., l3
Hadson, M. K.,67
Harvey, J., (41), 42, (42)

Hewitt, R. G. L.,90
Hinsch, H., 13

t2, 31, 90,

111

Lacombe, C., 1l I
Landau, R. W.,67

Lanzerotti, L. J., 12, (4I),42

Lashmore-Davies, C. N., 90
Lee, M. A.,90
Lehane, J. A., 11, 31,90

Iiyoshi, A., (107), (108),

l.evy, R. H., (34),42

(109),110
Infeld, 8.,43
Inutake, M.,90

Libchaber, A., 13
Lighthill, M. J., 3l
Lominadze, D. G.,67
Luhmann, N. G.,67

loffe,Z.M.,12
Ionson, J. A., 90
Isaacson, R.

T.,

13

Ishii, K.,67
Jephcott, D. F., (25),(27),
31

Kaburaki, O., 1l I
Kakutani, T., (80), 89-90

Kaladze,T.D.,67
Kan, J, R., 31,67
Kaner, E. A., 13
Kao, Y. H., 13
Kapalan, S. A.,91
Kappraff, J. M., (54), 60
Karpman, V.I., (86),89
Kats, M. 8.,67
Kaufman, A.N., (88), 89
Kaufmann, M.,45,5G61
Kaw, P. K.,90
Kawamula, H., 13
Kearney, C. F. F., (103), 110
Keller, R., (45),60, (103),

(104),110
Kennel, C. F., 67, 90
Khodusov, V. D., 89
Kirschbaum, J., 13

Kitamura, T.,91
Knox, S. O., 31

Kotik, D. A.,31
Krauss, A. R., 13

11,

(11), (25),31

Lundquist, S., (6), (10), 11,

(11)
Lyn'ko, L. V., l3
Lyons, D.,72
McKenzie, J. F., 91
McPherson, D. A., (29), 3L
Makarenko, V. N.,67
Malston, E. H., 13
May, R. M., 11
Mazur, V. A.,67

Mikhailovskii, A. 8., (65),

66-67,91
Mikhailovskii, A. G., 67
Mima, K., (69), (80), 89
Minami, K.,91
Mio, K.,91
Montgomery, D., (68), 89
Montgomery,M. D.,67
Morrow, R.,43, 61
Mullan, D, J.,12
Muller, G.,3I,43
Murphy, G.L.,32
Mutoh, T., (107), (108),

(109),110
Nakahara, J., 13
Narasimha, M. S., 11
Narayanamurti, V., 13
Nekrasov, F. M.,90

AUTHOR INDEX

122
Neuhauser, J., (45), 60-61
Newcomb, W.A., (23), 31
Ngan, Y. C.,43
Nguyen, H., 91
Nishida, Y.,67
Nosaki, K.,67

Novikov, I. I., 1l

obiki, T., (107), (108),

(109),110

Y.,67,91

Olbert, S., l2
Ong, R. S. B.,90
O'Sullivan, R. J., 61
ott, E., (103), 110
Owens, A. 1., 12

Sakurai, T., 12
Sasaki, A., (107), (108),

(109),110
Y., 13
Schamel, H.,67
Sawada,

2., (45), (51), (54),

l'etervov, A. P., 13
Toptygin, I. N., l2
Tripathi, V. K.,91
Troyon, F., (45), 60, 110
Tsang, K.T., 66, (66)
Tsunematsu, T., (45), 60
Tsytovich, V. N., 91
Turner, J. M,, 12
Uberoi, C., (18), 3l-32, (34),
(41\,42, (54), 61
Uchida, Y., 12, III
Uo, K., (107), (108), (109),
110

61

lll

Pietviashvili, V. I., 91

Piliya, A.D., (45),60

Platzman, P. M., 13
Pochelon, A., (103), (104),
110

Pokhotelov, O. A.,91
Pokhote'ov,O. A.,67
Pozwolski, A.,91

Pridmore-Brown, D. C., (29),

3r, (34), 42
Pritchett, P. L., (45), 60
Pyle, R. V., 30

Radford, W. E., 12
Radoski, H. R., 12
Ralan,

Saka, O., 91
Sakanaka, P.H.,42

SedldUek,

Paoloni, F. J., I l, 3l-32,9A

Parker, E. N., (63), 66
Parkhomov, V. A., 13
Patterson, 8., 91
Pellat, R., 90
Pereira, N.R., (88), 89
Perkins, F.W., (103), 110

Petrukhin, N. S., 91
Piddington, J. H., L2,

R.2., (68), (73),

(74),89

Scholer, M., 91
Schuller, P. G.,32,43
Schwarz, B., L3
Schwarzschild, M., (47), 60

Ogino, T., 91
Ohsawa,

Sagdeev,

L.,42

Selim, A. A., 9l
Shabat, A. 8., (88), 89
Shaffman, P. G., (68), 89
Shah, S. K.D., L2
Sharma, S. K., 91

Shohet, J. L., (104), (105),

110
Shukla, P. K., (80),89
Sigma, D. J., (64), 66
Silin, V. P., (20), 31
Skadron, G., 90

Skilling, J., 12
Skobov, V. G., 13
Smith, E. J.,12
Smith, J.,66, (66)
Sobolev, S. V., 32
Solodyna, C.V.,12
Sonnerup, B. U. O., 67
Spitzer, L,,32
Srivastava, K. M.,

tz

Stefant, R. J., (20), 31

Stein, R. F., 12
Stepanov, K. N., (20),3L,67
Stix, T. H.,32
Stocker, P. M., (27),3L
Suematsu, H., 13
Suffolk, G. C. J., 89
Swanson, D. G., 43,61

Rauchle,8.,32
Rauchle, R.,43

Takoda, S., 91

Richter, A,K.,12

Tamoikin, V. V.,91

Roberts, 8.,67
Roberts, P. H., 12
Rogister, A.,91
Rosenbluth, M. N., (63),

(64),66
Rukhadze, A. A., (20), 31

Rutherford, P. H., (64),66

Tang, J. T.,67

Taniuti, T., (88), 89

Tanuma, S., 13

Tappert, F., (87), 89

Tataronis, J.

A.,43, (54),
(t04), (105),

60-6 1, (92),
110, 111

Vaclavik, 1., (42\, 42, (45),

60,110
Vahala, G., 91
Valley, G. C., 91
Vaselago, V. G.,

l3

Venkataraman, N.5.,32
Vinogradov, P. A., 13
Volk, H. J., 13, 90

Vshivkov, V. A., 66
Wallace, P. R., 13
Wentzel, D.G.,91,111
Wersinger, J. M., (103), 110
Whang, Y. C., 13
Whitson, J. C., 66, (66)
Wilcox, J. M., (23), (25),
30-31

Wiley, W., 12
Williams, D. J., 91
Williams, G. A., l3
Wilson, P. R., 13
Winterberg , F ., 42
Wisseman, W. R., 89

Wolff, P. A., 13
Woods, L. C., (23),31
Yajima, N., (88), 89
Yakovenko, V. M., l3
Yamagishi, T., 9l
Yu, M. Y., (80), 89
Yuang-Chung, C.,91
Yukhimuk, A. K.,67
Zakhalov, V. E., (88), 89
Zmtrda, A. J.,12

SUBJECT

'NDEX
Alfv6n-wave frequencies, 33, 66
spectrum of, continuum, 33-34, 36

Absorption nte (see Energy absorption rate)

Alfv6n resonance (see Spatial resonance of
the Alfv6n wave)
Alfv6n resonant sutface, 92'93,101, l'06-107
Alfvdn speed, 26, 83, 108
Alfv6n surface waves, 4L, 44,46
coupling of bulk Alfv6n waves with, 49
damping

of,45,

discrete,40
Alfv6n-wave heating (see Plasma heating)
Alfv6n wave-"ideal," l, 10, 14, 18, 33,69,
101

energy transmission of, 4

magnitude of velocities in different media

53

of,4

computer calculation of, 45

experimental observation of, 54
dispersion relation of, 47
effect of resistivitY on, 54-55
frequency of,47,53
instabilities associated with, 44

properties of, 3
reflection and transmission of,4-5

refraction of, 5
velocity of , I-2,7
wave equation for, 3
Alfv6n-wave instabiliti es, 62
See also Instability

See also Instability

linear transformation of, 54

phase velocity of,48
resonant absorption of , 49-54,92
physical mechanism of, 54

Ambipolarity, T l-72
Amplitude modulation, 86
Anomalous transport (see Enhanced transport)

resonant mode conversion of,45, 54-60, 66

Alfv6n-wave equation, 4445, 54,95
See also Inhomogeneous magnetic fields and

Boltzmann distribution, 81
Boundary conditions, 36-37, 49, 82, 93
for Alfv6n surface waves, 47
for cylindrical geometry, 9
for kinetic Alfv6n wave, 59
for propagation in cylindrical plasma, 23,25
for reflection and transmission, 5

Nonuniform plasmas
asymptotic behavior of the solution of' 54
when compressibility effects are taken, 95
discrete eigenvalues of, 52
Fourier-Laplace transform of, 49
Green's functions of, 45, 49

123

124

SUBJECT INDEX
Circularly polarized Alfv6n wave, 68
Coil inductance, 102

with Hall effect, finite conductivity, and

ion neutral collisions taken into account,

Collective modes, 45, 54

Collision damping, 92

23

in partially ionized plasma, 19

for compressional Alfv6n wave, 17
for ideal or magnetohydrodynamic (MHD)
Alfvdn wave (see Alfvdn wave-"ideal")
in an inhomogeneous plasma, 65-66,72

Compressibility effect, 6, 93, 95

See also Dispersion relation
Compressional Alfv6n wave, 7, 16-18, 20,

22-23,25,29
Confinement time, 104
Continuum spectrum, 36, 40, 4445,

5 1,

54,

92

bandwidth of,40 (Fie. 3.1)

Coriolis force, 7-8, l8
Critical frequency, 18
Cutoff frcquency,22
Cylindrical geometry, Alfv6n-wave propaga-

tion in,

86-88

for parametric decay instabilities, 78

for plasma with an anisotropic pressure, 63
in the presence of velocity shear, 64
with stream of charged particles existing
in the plasma, 64
for surface waves, 47 , 53
for waves in a cold magnetized plasma, 15

Cylindrical plasmas
Alfv6n waYes in,22-25
general dispersion relation for, 23
lower cutoff frequency for compressional

Distribution function

drift,73
electron guiding center, 69
Gaussian, 70

mode in,22

nonuniform,44

Maxwellian,20, 62
quasi-linear equation for the average,20
Drift kinetic equation, 55, 69, 73, 81
Drift velocity, 69, 81
Drift wave frequency, 72

Damping distance
due to both neutral particle collision and
conductivity, 24-25
due to finite conductivity, 6
in cylindrical geometry (or for torsional
Alfvdn wave), 9
Damping rate, 101
Design examples (see Plasma heating)
Diamagnetic drift frequency, 65
Diamagnetic drift of ions, 65

Drift

plasma, 15
Discrete spectrum, 36, 40,44, 54
Dispersion effect, 80
Dispersion function, 51, 53
expansion of,52
logarithmic branch points of, 51 (Fig. 4.3)
zeros of, 52
Dispersion relation

sidered, 6

with finite-frequency effects considered

(see

Finite-frequency Alfv6n wave)

wave

See also

Dielectric constant, 16
for a low-frequency wave, 16
Dielectric tensor for a cold magnetized

of Alfvdn wave-"ideal," L
with compressibility efTects considered,
with Coriolis effects considered, 7
with finite-conductivity effects con-

for ion-acoustic wave, 78

for kinetic Alfv6n wave,2I,78,82
nonlinear, for modulational instability,

instability, 69,7I
Instability

Electromagnetic-hydrodynamic waves, I
See olso Alfv6n wave-"ideal"
Electron drift wave, 65
Electron heating (see Plasma heating)
Electrostatic surface wave, 47
frequency of,47
Energy absorption rate, 45, 92-93,95-97 , 701
Enhanced loss of plasma, 103
Enhanced transport, 106, 108
Envelope solitons, 86, 88
Experimental studies
in liquid metals, 10-11

in plasma heating, 103-109

in plasmas, 25-29
measurements of attenuation,2T (Fie. 2.2)
measurements of wave phase velocity,2T

(Fie. 2.1)
Fast mode,7,20,24-25
compressional Alfv 6n, 7, 24

inertial,

SUBJECT INDEX
Finite-froquency Alfvdn wave, 1 8-19
dispersion relation for, 18, 88
modulational instability of, 69
See also

Instability

Finite-frequency effects, 69, 80, 83, 88

Finite-Larmor-radius effects, 79 , 54, 64-65,

71,73
Kinetic Alfv6n wave
Finite-resistivity effects, 6, 55
Frozen-in fields, concept of, 3
Fusion temperature, 100
See also

Hall effect, t8,23

See also Finite-frequency Alfvdn wave
Heating efficiency, 102, 104
See also Plasma heating
Heating rate (see Plasma heating)
Heliotron-D plasma, 1 07-1 09
Hydromagnetic shock, 68
Hydromagnetic wave equation (see Wave
equation)
Hydrotnagnetic wave guide, 23
Hydromagnetic waves (see Alfv6n wave-

"ideal")
Induced scattering decay, 74
Induced scattering process, 76

Inertial wave,7
See also Fast mode
Inhomogeneous magnetic fields, 33
Alfvdn-wave propagation in, 34
local variation of Alfv6n-wave velocity in, 33
spectral analysis of ideal Alfv6n wave in, 33
the wave equation in, 33, 35
asymptotic time expansion of the solution

of,40,45
eigenfunctions of, 37
Fourier analysis of, 36
singular behavior of ,33-34,44
singular points of , 34, 37

solution

of,364l

Inhomogeneous plasma (see Nonuniform

plasmas)

Instability
absolute, 66
convective, 66
coordinate space,62

decay,79
drift Alfvdn wave,65
anomalous heat loss due to, 66
growth rate of, 66

125
fire hose, 62

Kelvin-Helmholtz,64
modulational, 69, 86-89
parametric decay, 69 , 73
growth rate of, 78
threshold pump field for, 78
plasma, associated with the Alfv6n wave,62
quasi-mode decay, 80
growth rate of, 80
resonant decay, 78, 80
two stream, 63
due to finite Larmor radius effects, 64
See also Kinetic Alfv6n wave
velocity space,62
Intermediate mode, 7
Inverse scattering method, 88

Ion acoustic (sound) wave, 2I, 29,

79,95, 100

73-7

4, 76,

Ion heating (see Plasma heating)

Ion-induced scattering process, 79

Ion-ion collision frequency,

Ion-ion collision rate, 98

101

Ion neutral collision frequency, 1,9,23

lon polarization drift, 78
nonlinear, 76
Kinetic Alfv6n wave, 19,55, 7I,74,76,
79-80, 92-93, 97, gg, 10G101, 109
coupling between a surface MHD mode
and,57
damping rate of,97
electron and ion Landau damping of,21
excitation by the drift wave instability of,
69
linear dispersion relation of, 19-21 ,73,82
mode converted, 60
nonlinear Landau damping of,69
parametric decay of ,69,73, 100
solitary wave of, 80, 82 (Fig. 6.5)
threshold amplitude of, 1.01
turbulence, 69
two-stream instability associated with, 64
wave equation for the bulk, 57
Kyoto, Japan, 103
See also Heliotron-D plasma
Landau damping, 21, 52, 7 0, 92, 101, 108
of Alfv6n surface waves, 45

electron, 80, 101

fractional rates of, 2L
inverse,

Tl

SUBJECT INDEX

126

Nonlinear dispersive medium, 86

ion,99

Nonlinear effects, 92
Nonlinear equation describing the dynamics

of ion acoustic wave,79

of kinetic Alfvdn wave, 21
nonlinear ion, 69,79, 100-101
threshold value for the nonlinear, 101
Langmuir wave,58, 89
Laplace equation,4T
Laplace transform technique, 45
Large-amplitude linearly polarized Alfv6n
wave, local phase velocity of, 68
Larmor radius, 19-20, 34, 54
See also Finite-Larmor-radius effects
Lausanne, Switzerland, 103

of a wave amplitude, 86
time asymptotic solution of, 88
Nonlinear frequency shift, 88
Nonlinear magnetohydrodynamic equations,
68

Nonlinear optics, 87
Nonlinear perturbation theory, 88
Nonlinear processes, 68, 100
Nonlinear steepening effect, 80
Nonlinear wave, 80, 83

See olso Theta-pinch plasma

equation for, 82

Local Alfv6n frequency, 44

Local Alfv6n resonanoe, 45
Local Alfv6n speed, 33,36
Local Alfvn-wave operator, 35
Lundquist number, 6, 1l

Nonlinear wave-particle interactions, 101

Nonuniform plasmas, 29, 44, 92-93
coupling between the torsional and compressional Alfvdn waves in,29

Magnetic mode, 7
See olso Slow mode

Magnetohydrodynamic (MHD) Alfv6n wave,

4,74,68-69, 73, 80
See also Alfv6n wave-"ideal"

electrostatic oscillation of cold, 34,45

linear transformation and absorption of
electromagnetic waves in, 45
resonant absorption and mode conversion
of Alfvdn surface waves in,44
Normal-mode analysis, 45

Nyquist criterion, 52

decay instability of ,68,73

nonlinear theory of,69

Magnetohydrodynamic (MHD) equations,

l,

22, 35,46, 54,93

nonlinear,6S
Magnetohydrodynamic (MHD) experiments
(see Experimental studies)

Magnetohydrodynamic (MHD) frequency

range, 63-64
Magnetohydrodynamic (MHD) model for the
plasma, 46,53
See also Magnetohydrodynamic (MHD)
equations
Magnetohydrodynamics

principle of, I
specfial theory of,42
'l

Magnetosonic wave, ,95, 102, 103

See also Slow mode
Matching relations, 38
Mode conversion (see Resonant mode
conversion)
Modulation amplitude of the Alfvdn wave, 88
Monochromatic carrier frequency, 86

Monochromatic wave,92

Ohmic dissipation, 98
Ohmic heating, 100-101
Ohm's law,2,5
One-fl uid magnetohydrodynamic (MHD)

theory, 55
See also Magnetohydrodynamic

Partially ionized plasmas, 19

Alfvbn velocity in,19
Phase mixing, 45,95
Plasma

cylindrical (see Cylindrical plasmas)

enhanced loss

of, 103

inhomogeneous (see Nonuniform plasmas)

nonuniform, 29, 44, 92-93
partially ionized (see Dispersion relation)
stellarator, 103-104
Proto-Cleo, 104-105
R-02, 106-107
theta-pinch, 103-104

tokamak,98, 100
toroidal, 93

two-fluid description for,

Noncollective modes, 45, 54
See also Continuum spectrum

(MHD)

equations

15

Plasma heating, 92, 103-t04, 106-107

design examples for, 100

SUBJECT INDEX
127
experimental studies in, 103-109
See also Experimerltal studies
heating late in, 92,97-gg,l00-l0l
heating rate in a collisional regime in

for electrons, 98
linear heating rate in a collisionless regime

in
for electrons, 99
for ions, 99
nonlinear heating rate in collisionless
regime in, for ions, 100
Plasma impedance, 60
Plasma instability associated with the Alfv6n
waver 62
Plasma-wave equation, 4S
Plasma-wave guide,22
Poisson's equation, 20

Schrddinger equation, 87

nonlinear, 88
Self-trapping process, 87
Shear Alfv6n waye (see Torsional Alfvdn wave)

torsional, 24

7.j)

for electrons, T0-71

for ions, 71, 73
of electrons, 69-71,
equations for electrons, 70
of ions, 7 L-7 3
Quasi-neutrLljty, 7Z
equation for, 25
Quasi-particle, 87
Radiation impedance, 1 02
Reactor tokamak, 100
Reflection and transmission (see Alfvdn

ohmic, 102
radiation, 102
skin, 102
Resonant absorption, l0l-102, 104
See also Alfv6n surface waves

R-02, 106 (Fie. ?.6)

See also Stellarator plasma

Solar magnetic fields, 42

Solitary Alfv6n wave, 68
Solitary wavej 80, 82, 94,87
Solitary wave of the kinetic Alfvdn wave, g0,
82 (Fig.6.5)

Quasilinear diffusion, 69
coefficient of

wave-"ideal")

Alfv6n surface waves

Resorrant surface (see Alfvdn resonant strrface)

sheet curent, 45
Simple wave, 82
Slow mode,7,19,24-25
magnetic, 7
magnetosonic, 7

Polarization ddft, 1?
electton, T4
nonlinear ion,76

Resistance

108
See also

for ions, 97-98

Polarization reversal, l9
Ponderomotive force, 75, gg-g9
electrostati c, 107
magnetic, 102
Power flow,95-96
Power source, 100-101
hoto-Cleo, 104, 105 (Figs. 7.4 and
See also Stellarator plasma
Pump w av e, 7 3,'l 6, Z 9-80

Resonant decayrT4
Resonant kinetic Alfv6n mode, Z?
Regonant mode conversion, 66, 93.,9g-99,

nonlineat wave equation for, g2

Solitary wave solution, 69, B2-g3,g7-gg
Solitons, onvelopel 86, Bg
Sound waVe (MHD), 7
Space.plasmas, lhenomenon of magnetic

pulsatiohs of, 41, 54

Spatial resonance absorptiort, 92
Spatial resonance of the Alfv6n wave, 41, 95,
97

Spatial resonant surface (see Alfv6n resonant

surface)
Spectral analysis of idCal Alfvdn wave, 33
Spectral concenttation, phenomenon of, 52
Spectral cuts,51
Spectral theory of magnetohfdrodynamics,
42
Stellarator plasma, 103"104
Proto-Cleo, f 04-105
R-02, 106-107
Structured magnetic fields, 4142
Sub-AlfV6nic wave, 83
Sukhumi, Russia, 103
See also Stellaretor plasma, R-02
Super.Alfv6nic wave, 83
Supplementary heating, 100,
See also Plasma heatlng
Sutface current, 93
Surface pardrheters, 48

l0l

SUBJECT INDEX

128
Theta-pinch plasma, I 03'l 04
Tokamak parameters,98
Tokamak plasma, 98, 100
Toroidal plasma,93
Torsional Alfv6n wave, 8-10, 16'18, 22,

Wave equation, 33-35

with compressibility effects considered

(see Alfv6n-wave equation)

for electromagnetic plane waves, 64,66

for ideal Alfv6n wave,2'3
in inhomogeneous magnetic fields (see

24-26,29,95
See also

Cylindrical geometrY

Torsional mode, 24
Trapped electrons, 99

nonlinear, 101

Two-fluid description for plasma, l5

Viscosity coefficient,

Inhomogeneous magnetic fields)

Wave-particle interactions, 9 3

l'72

Viscous damping, 97, LOl

Vlasov equation, 20

resonant, 69
Wisconsin, 103, 108
See also Stellarator plasma, Proto-Cleo

WKB approximation,58

DISCLAIMER
This book was prepared as an account of work sponsored by an agency of the
United States Government. Neither the United States Government nor any agency
thereof, nor any of their employees, makes any warranty, express or implied, or
assumes any legal liability or responsibility for the accuracy, completeness, or
usefulness of any infdrmation, apparatus, product, or process disclosed, or
represents that its use would not infringe privately owned rights. Ref erence herein
to any specif ic commercial product, process, or service by trade name, trademark,
manufacturer, or otherwise, does not necessarily constitute or imply its
endorsement, recommendation, or favoring by the United States Government or
any agency thereof. The views and opinions of authors expressed herein do not
necessarily state or reflect those of the United States Government or any agency
t

hereof.

U. S. DEPARTMENT OF ENERGY CRITICAL REVIEW SERIES

.'\ i
' '' , r i ,, r ,.t rrt ,l,t1r ,1 tlt,, ul -,llrl {', [,t;ir1t',1,t'rl l,\ 1lt,' 1191 T''t irrrr];t it)it)r-ltlill (,!'l
i'r'
I ', .t l' '. ,,.,,
rrr ,i,'1,r,]rt,'i1 i, , ,,tiit,tlt' iltr' ,'!. I 1rl il :.l,lii' |1 ktlrrvi,lr'tltl{i trt ,l
i ,
,,t ltl'rri..1 l,,rilrtlr!,,
,i,, i
,/tlri(,111 (1{,,,"1'rl)ltlll,, iitllt ftL;irlr:,iti:rj ,trr,J
r'r1 ', l,iiltl
Iitrrr,.
IL:l
.lI''i'l,l
,t.t
,.,r rt r lr r, N,ri,,r .t lr,r i11L,i, lttl,ri;1,i11i,1 Slt',tf ',. tI !r [)i,l);]rlptr,f tl i11 (-Lrf nr|i,f (f r.
frl ,i l llr rt \''1,,,t
lirl I)r rrrlri,ri, tJtil'.i.ll ll r, []rtir,il Sl,r1,,:,. (.,rtr,tri,r ,rni1 Nlrrt.o l:,lll)iti(1
.,.,ri, l!T'l l'l',
Sources
(TrD

of

Tritium

24635). 1968,

and

lts

Behavior

upon

Release

to

the

Environment

$ 10.25.

Reactor-Noise Analysis in the Time Domain (TlD ?4512\, 1969, $10"50.

Plume Rise (TlD 250151, 1969, $10

OO.

Atmospheric Transport Processes

Part 1: Energy Transfers and Transformations (TlD 24868), 1969, $14.25.
Part 2: Chemical Tracers (TlD 25314\, 1971, $17 50.
Part 3: Hydrodynamic Tracers (TlD-257311, 1912, $13.25.
Part 4: Radioactive Tracers 1JID-211141, 1978, $23.5O.
The Analysis of Elemental Boron (TlD-25190), 1970, $11.25.
Aerodynamic Characteristics

of Atmospheric Boundary Layers

(TlD

25465\,

191

1,

$ 1 2.75.

Nuclear-Explosion Seismology \fD-25512), 1971, $12 00.

Boiling Crisis and Critical Heat Flux (TlD 25881 , 1912, $10.25
Neptunium-237 Production and Recovery (TlD-259551, 1972 $1O.50.

The Kinetics of the Oxidation-Reduction Reactions of Uranium, Neptunium,

Plu-

toniunr, and Americium in Aqueous Solutions (TlD-26506), 1975, $11.50.

Particle-Transport Simulation with the Monte Carlo Method (TlD-266071 1975,

$11 00

The Chernistry of Americium (TlD 26971), 1976, $15.50.

lchiban: Radiation Dosimetry for the Survivors of the Bombings of Hiroshima and
Nagasaki (TlD 27OBO), 1911, $11.25

An lntroduction to Trapped-Particle lnstability in Tokamaks

(TlD 21 1571, 1911

$10 50

Fuels from Sugar Crops, Systems Study for Sugarcane, Sweet Sorghum, and
Sugar Beets 1JID 22181) 1978 $1 1 75
Radiostrontium Movenrent in Soils and Uptake in Plants \flD-21564), 1978, $11.50.
The Alfvtjn Wave IDOEiTIC
(Lrstrncl crrrrent .rs

'11

197 (Ordt-'r No. DEB2001]O2jl ,1982, $1 1.50'

of Janttary 1982)