Introduction to Plasma Physics

Univ.-Prof. Mag. Dr. Helmut O. Rucker

Institute of Physics
Karl-Franzens-University Graz, Austria

The lecture Introduction to Plasma Physics is intended as a first approach to the

physics of particles and fields in space with fundamental phenomena and corresponding
laws. The content is a selection out of the books of Chen [1], Kippenhahn and Mllenhoff
[2], and Lyons and Williams [3] with specific typical emphasis to space physics as a base
for further lectures like Planetary magnetospheres and Radio and plasma waves.
It will be highly appreciated to send any comments, corrections, and suggestions for improvements to [email protected].

Helmut O. Rucker

Contents
1 Basics of Plasmaphysics
6
1.1 The Three Plasma-Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 Different types of Plasmas . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2 Influence of a Magnetic Field on the State of Plasma
2.1 Gyration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 Special Magnetospheric Processes . . . . . . . . . . . . . . . . . . .
2.2.1 Motion of Charged Particles in Magnetic and Electric Fields
2.2.2 Derivation of the dipole moment . . . . . . . . . . . . . . .
2.2.3 Derivation of the magnetic moment . . . . . . . . . . . . . .
2.2.4 Oscillation (bounce motion) . . . . . . . . . . . . . . . . .
2.2.5 E B-Drift . . . . . . . . . . . . . . . . . . . . . . . . . . .
3 Plasmas as Fluids
3.1 Plasmaphysics . . . . . . . . . . . . . .
3.2 The Convective Derivative . . . . . . .
3.3 Consideration of Temperature . . . . .
3.4 Consideration of collisions . . . . . . .
3.5 Hydrodynamics . . . . . . . . . . . . .
3.5.1 Equation of continuity . . . . .
3.5.2 Equation of state . . . . . . . .
3.5.3 The full set of MHD equations
3.5.4 MHD-Drift perpendicular to B:

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

17
17
19
19
22
24
25
33

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

46
46
47
48
49
49
50
51
52
55

4 Plasma Oscillations and MHD Waves

4.1 Plasma oscillations . . . . . . . . . . . . . . . . . . . .
4.2 Magnetohydrodynamic Waves . . . . . . . . . . . . . .
4.2.1 Linear perturbation theory: . . . . . . . . . . .
4.3 MHD waves with arbitrary angle to the magnetic field
4.3.1 Development of a hodograph: . . . . . . . . . .
4.4 CMA-diagram . . . . . . . . . . . . . . . . . . . . . . .
4.4.1 Structure of the CMA-diagram . . . . . . . . .
4.4.2 Wave-modes and transitions . . . . . . . . . . .

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

59
61
61
63
71
82
85
85
86

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

5 Appendix
90
5.1 Physical Quantities in Plasma Physics . . . . . . . . . . . . . . . . . . . . 90
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

Contents
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

1 Basics of Plasmaphysics
1.1 The Three Plasma-Criteria
Matter can appear in four different states. The particles setting up matter have the
more possibilities to move the higher the temperature is. In solid-state matter atoms
and molecules have the least freedom to move. In liquid state the freedom to move
is a little more extensive. In gas, atoms and molecules can move independently but
the electrons are obeying the laws of nuclear physics bound to their atoms. In the
plasmatic state the electrons are completely separate from the atoms and therefore have
entire freedom of movement. If atoms or molecules have lost one or more electrons they
carry positive charge outwardly, in this case they become positive ions. Plasma is
therefore considered as gas showing collective behavior and consisting of particles which
carry positive and negative charges in the extent that the overall charge comes to zero.
Definition: Plasma is electrically neutral to the outside, if the number of
positive and negative charges equals in a sufficiently large volume and for
a sufficiently long interval of time. This balance is referred to as quasineutrality.
ne
or

Zni ne
Zni

Zni

 1

(1.1)
(1.2)

ne . . . electron-density (number of electrons per volume unit)

ni . . . ion-density (number of electrons per volume unit)
Z . . . Z-times charge
Space plasmas like the solar wind are quasi-neutral over given areas. In a given volume
(in a defined span of time) there are as many positive as negative particles. The physical
explanation is the good conductivity of plasma.
Positive charges lead to electric fields that attract negative charges, the latter are much
more mobile and thus cause compensation of charges. Electric charge-effects have to
be dominant (effects must be at least of the same intensity) over effects of thermal
motion so that typical plasma properties occur or can be held up. Only in this case the
far-reaching electric forces can affect neighbouring particles. This results in collective
reactions which are determined by electric forces, this is how plasma and ordinary gas
can be distinguished. One of the Maxwellian equations is used to determine the first

1.1 The Three Plasma-Criteria

plasma-criterion:
E = div E =

(1.3)

0 . . . dielectric constant (= 8.85 1012 A s/V m)

. . . charge-density (mass density and electric charge)
E . . . electric field
The electrical potential is directly connected to the electric field:
E =

(1.4)

Equivalent notation:
= grad =

i+
j+
k
x
y
z

(1.5)

Relation between charge density and potential:

div E = div grad = (

2 2 2
+
+
)
x2
y 2
z 2

(1.6)

Forming the divergence of the gradient is expressed by the symbol 2 (Nabla squared)
or by the Laplacian operator :
2 = =

2
2
2
+ 2+ 2
2
x
y
z

The Laplacian operator is the scalar product of the Nabla operator with itself.

 


=
i+
j+
k
i+
j+
k = 2 = ,
x
y
z
x
y
z

(1.7)

(1.8)

where i j = ij . The relation div E = /0 thus leads to Poissons equation:

=

In spherical coordinates this relation can be written as:





 2 
1
1

2
2
r
+ 2
sin
+ 2 2
=
= 2
2
r r
r
r sin

0
r sin

(1.9)

(1.10)

The charge density symbolizes the overall charge density of electrons and ions:
e = ne (e)
ne
ni
Z
e

...
...
...
...

and i = ni (Ze)

(1.11)

electron density (number of electrons per volume)

ion density (number of ions per volume)
charge (Z-times)(to simplify matters Z = 1)
elementary charge

1 Basics of Plasmaphysics
According to classical statistics the number of particles of a given overall number n0 staying at the position x, y, z in a potential field (x, y, z) is given by Boltzmanns equation:


e(x, y, z)
n(x, y, z) = n0 exp
kT
e
kT
k

...
...
...

(1.12)

potential electric energy

thermal energy
Boltzmanns constant (k = 1.38 1023 W s/K)

Boltzmanns equation for electrons is given by:



e(x, y, z)
ne (x, y, z) = ne,0 exp +
kTe


(1.13)

Physically this relation expresses that electrons are very mobile due to their low mass
and that they can be accelerated to high energies quickly by an appropriate force. Electrostatic forces emerge because electrons cannot leave a plasma region as a whole without
generating a significant positive (ion-) space-charge, which balances electrons and ions.
Because of the condition of quasi-neutrality the following relation is applied to the quantities with indices zero:
ne,0 = Zni,0 = n0
ne,0 = ni,0 = n0

(1.14)
for Z = 1

Applying ne , ni and n0 to Poissons equation (1.9) we get

1
e
en0
e
[en0 exp(
) + en0 ] =
[exp(
) 1]
0
kTe
0
kTe

(1.15)

e
The term [en0 exp( kT
)] is applied to electrons, the second term (en0 ) is applied to
e
e
positive charges. In areas where ( kT
)  1, where thermal energy surpasses the potential
e
electric energy substantially the exponential expression can be developed into a Taylor
series:

exp (

e
e
1 e 2
)=1+(
)+ (
) + ...
kTe
kTe
2 kTe

The series is aborted after the second term. Applying these terms to eqn. 1.15, now
using spherical coordinates, see eqn. 1.10, and considering the radial component solely,

1.1 The Three Plasma-Criteria

we get
=


2
r
=
r



2
2
1
2
r
+
r

=
r2
r
r
r2


2
1

2
2r
+
r
=
r2
r
r2
1
r2 r


en0
1+
0

en0
1+
0


e
1
kTe

e
1
kTe

e 2 n0

kTe 0
e 2 n0

kTe 0

2 2
e 2 n0
+
= 0

r2
r r
kTe 0


2
n0 e 2
2
D + D
= 0
r
0 kT

r .

using the differential operator D :=

2 +

(1.16)

This leads to the characteristic equation

n0 e 2
2

= 0,
r
0 kT

(1.17)

which has the two following solutions

1,2

1
=
r

1
n0 e 2
+
,
r2 0 kT

(1.18)

whereupon only the negative part provides a solution that is physically meaningful, as
will be shown below. The potential is given by:
(r) = C1 e1 r + C2 e2 r
= C1 e

( r1

n e2
1
+ 0kT
r2
0

= C1 e1 e
| {z }
f1
C

1+

(1.19)
)r

n0 e 2 r 2
0 kT

+ C2 e

( r1 +

n e2
1
+ 0kT
r2
0

+ C2 e1 e
| {z }

1+

n0 e 2 r 2
0 kT

)r

f2
C

This differential equation has to satisfy the following initial conditions:

() = 0
(0) = 0

1 Basics of Plasmaphysics
Calculating the term

n0 e2 r2
0 kT

at terrestrial orbit one obtains

107 m3 (1.602 1019 )2 A2 s2 (1.5 1011 )2 m2

1021  1.
8.85 1012 A s /V m 1.38 1023 A V s/K 4 104 K
As a consequence the following approximation is evident:
s
s
r
n0 e 2 r 2
n0 e 2 r 2
n0
1+
'
=e
r
0 kT
0 kT
0 kT

(1.20)

f2 = 0
() = 0 C
f1 = 0
(0) = 0 C
q n
e k0 T r

(r) = 0 e

= 0 e D
r
0 kTe
D =
n0 e 2

(1.21)

Within the Debye shielding length D the potential (r) of an individual charge decreases
by a factor of 1/e due to space charge effects of the neighboring charges. D decreases
if plasma density increases and D increases with increasing temperature Te . Electron
temperature is very important in this context because shielding mainly occurs on account
of electrons and their high mobility.
The Debye shielding length, eqn. 1.21 is of great importance: in areas smaller than
D electric fields are too weak to take influence on the motion of particles. Thus quasineutrality is only given beyond a given volume. The lower limit in space for the occurrence
of quasi-neutrality is given by the Debye sphere, with Debye length D as radius. Quasineutrality is only given in areas greater than this sphere, thermal motion of particles is
dominant within it. Quasi-neutrality therefore is a good assumption for studying plasma

Figure 1.1: Description of the Debye length

phenomena with a scale length L, being substantially longer than the Debye length.
D  L

10

1st plasma criterion

(1.22)

1.1 The Three Plasma-Criteria

Example Is quasi-neutrality of the solar wind given in some area of the solar system
(orbit of Neptune: r 30 AU)?
To answer this question one has to investigate if the first plasma criterion can be
applied everywhere in interplanetary space. The variables in the relation for D (eqn.
1.22) are temperature and density. Assuming approximately constant temperature of
solar wind plasma the density of solar wind remains the only variable quantity. Over a
distance of r 30 AU it decreases to 1/r2 of the density at earth orbit.
How large is the Debye length D :
(a) at orbit of Earth (r = 1 AU)?

(b) at orbit of Neptune (r = 30 AU)?

Putting in the parameters in eqn. (1.21):

0
ne
np
k
e
Te
Tp

...
...
...
...
...
...
...

dielectric constant 8.85 1012 A s/V m

electron density 107 m3 (at 1 AU)
proton density 107 m3 (at 1 AU)
Boltzmanns constant 1.38 1023 W s/K
elementary charge 1.602 1019 C
electron temperature 1.5 105 K
proton temperature 4 104 K

ad (a) At earth orbit, calculation for protons:


D =

8.85 1012 A s/V m 1.38 1023 A V s/K 4 104 K

107 m3 (1.602 1019 )2 A2 s2

= [8.8 1.38

 12

1
1
4
1035+47+38 m2 ] 2 = [18.975 100 m2 ] 2
2.56

D,proton = 4.4 m

D,electron = 8.4 m

With a scale length of Ldensity 108 m  D = 4.4 m for protons and 8.4 m for
electrons, respectively. Thus the first plasma criterion is met.
ad (b) At Neptune-orbit n decreases by 1/r2 within the solar wind, that is
n(r = 30 AU) n(r = 1 AU)/302

If n decreases by about a factor of 900, D increases by 900 = 30. Ldensity

108 m  D 131 m for protons and 252 m for electrons, respectively. Here the
first plasma criterion is also met.

11

1 Basics of Plasmaphysics

Figure 1.2: Scales of plasma, [source: M. Heyn: Vorlesungsunterlagen zur Plasmaphysik]

Although this is only a coarse estimation it can be said that in the entire known space
of the solar wind plasma (with the exception of the direct vicinity of the Sun) it is in
quasi neutral condition. Using magnetohydrodynamics therefore is fully guaranteed for
the description of large and small scaled ( 103 m) structures.
The Debye sphere is defined by the volume VDebye :
VDebye =

4 3

3 D

(1.23)

In a Debye sphere there are VDebye ne electrons NDebye charged particles in general. For
a magnetohydrodynamic approach the following relation has to be applied:
NDebye =

12

ne 4 3
1
3 D

(1.24)

1.1 The Three Plasma-Criteria

NDebye is the number of particles in the Debye sphere.
= ne 3D  1

2nd plasma criterion

(1.25)

If , which is called plasma parameter, is much greater than 1, the term plasma is
used. In other words: collective behavior of particles in a plasma requests NDebye  1.
A charged particle can leave its neutral position up to a defined length, the Debye length
D . Then the particle oscillates around its former neutral position, attenuation can be
neglected. The mathematical approach in neglect of attenuation is written as:
m
2

m t2x
eE

2x
+ eE = 0
t2

(1.26)

deviating force
counterforce due to electric field

To simplify matters a one-dimensional approach is taken: only the x-direction is taken

into account. From the Maxwellian equation (eqn. 1.3), mentioned before, we get:
E=

ne
x
0

(1.27)

applying (1.27) to (1.26):

m

2 x ne2
+
x=0
t2
0

With the approach x = x0 eit , the solution of this differential equation can be determined
immediately:
x = ix0 eit
x
= 2 x0 eit
We obtain:
ne2
0
s
ne2
=
m0

m 2 =

(1.28)

This leads to the plasma frequency or Langmuir frequency:

p
1
fp =
=
2
2

ne2
m0

(1.29)

The equation of motion leads to the fact that the deviated particle carries out a harmonic
oscillation with the frequency p . This frequency is called Langmuir or plasma frequency.

13

1 Basics of Plasmaphysics
It is a characteristic frequency for a system of charged particles and it depends on the
particles mass and density. One can clearly see that plasma consisting of two components
has a characteristic frequency for electrons as well as for ions. Inserting corresponding
values in eqn. 1.29 results in plasma frequency for, e.g. heavy particles (ions), and for
lighter particles (electrons), leading to a lower plasma frequency for ions with respect
to electrons. The Langmuir frequency fp is the characteristic oscillation frequency for
electrostatic disturbances in a plasma.
Example Estimating the plasma frequency in the solar wind
0 . . .
ne . . .
np . . .
me . . .
mp . . .
e ...

dielectric constant (= 8.85 1012 A s/V m)

electron density 107 m3 (at 1 AU)
proton density 107 m3 (at 1 AU)
electron mass 9.1 1031 kg
proton mass 1.67 1027 kg
elementary charge 1.602 1019 C

Langmuir frequency for electrons: fp,e 28.4 kHz

Langmuir frequency for ions: fp,p 663 Hz
p
The ratio fp,e : fp,p = mp /me 43.
In order to define a plasma, a further condition, the 3rd plasma criterion must be met:
> 1

3rd plasma criterion

(1.30)

is the average time between collisions of charged and neutral particles, is the circular
frequency of a typical plasma oscillation (see e.g. eqn 1.29). If this condition is not given,
in other words: if there are too many collisions with neutral particles, particle motion
can rather be described by hydrodynamic equations than by electromagnetic equations.
Plasma physics in the classical sense can not be applied.
Summary of the plasma criteria:
1. D  L
2. = ne 3D  1
3. > 1

14

1.2 Different types of Plasmas

1.2 Different types of Plasmas

Space plasmas vary drastically regarding density and temperature, as we have already
seen before (fig 1.1). The Debye-length, i.e. the screening distance of the ions, depends
on these parameters and is a characteristic quantity of any plasma configuration.
The interplanetary gas is characterized as a low-density plasma, evincing densities of
about 102 cm3 and temperatures of 105 K. These values result in a huge screening distance of about 102 cm. For the higher density-plasma of the ionosphere, n 106 cm3 ,
T 103 K, D has already declined to 101 cm.
For the solar corona that is a characteristic hot (106 K), dilute plasma ((106 . . . 108 ) cm3 ),
showing gas discharges, D is in the order of about 10 cm, while for the extremely high
energetic plasma in the solar center ((1022 . . . 1025 ) cm3 , 108 K) the ions are screened
within a length scale of roughly 106 cm.
By exploring the sun in a distance of 0.3 AU with Helios 1 & 2, at present there are
known four different kinds of solar wind, each having unique properties:
1. Fast solar wind: The source of the fast solar wind is located at the coronal holes.
It is stable for long periods and is therefore accredited to the quiet sun. v
(400 . . . 800) km/s; the density in a distance of 1 AU is very low: 3 106 m3 .
The fraction of helium is about (3 . . . 4) %.
2. Slow solar wind in times of solar minimum: This type is characterized by a low
velocity of (250 . . . 400) km/s and a very high density of 11106 m3 at a distance of
r = 1 AU from the sun. For the most part it occurs at the borders of interplanetary
magnetic field sectors. Fraction of helium: < 2 % and variable.
3. Slow solar wind in times of solar maximum: Similar to type 2, though this type provides a higher fraction of helium (about 4 %). It is extremely turbulent (shockwaves
are likely to appear).
4. Coronal Mass Ejections: CMEs have a velocity of about (400 . . . 2000) km/s and
therefore propagate as shock waves. The fraction of helium (He++ ) may increase
up to 30 %.

15

1 Basics of Plasmaphysics

Figure 1.3: Speed of the solar wind depending on direction (Geophys. Res. Lett., 25, 1-4,
1998)

16

2 Influence of a Magnetic Field on the

State of Plasma
Space plasma is in general a magnetoplasma. Electric charges describe circular orbits
around homogeneous (invariant in space) and static (invariant in time) magnetic fields.

2.1 Gyration

dv
dt

= e(E + v B)

(2.1)

Assumption: E = 0
m

dv
dt

= e(v B)

(2.2)

Inner product with v

m

dv
v = ev (v B)
dt

v (v B) = 0
m

dv
v =
dt

d mv 2
(
)=0
dt 2

Kinetic energy of the particles and thus |v| remain constant. The magnetic field is given
by B = (0, 0, B)


 i
j k 

v B =  vx vy vz  = i(vy B) j(vx B) + k(0)
 0 0 B 

mv x = eBvy

(2.3)

mv y = eBvx

(2.4)

mv z = 0

(2.5)

17

2 Influence of a Magnetic Field on the State of Plasma

Second time derivative:
m
vx = eB v y
m
vy = eB v x ,

inserting (2.3): v x =

eB
vy
m

eB
vy
m
 2
eB
=
vy
m

m
vy = eB
vy

vy = c2 vy
vx = c2 vx ,
the cyclotron frequency being
c =
With rL =

v
c

eB
m

(2.6)

the Larmor radius becomes

rL =

mv
eB

rL =

mv sin
.
eB

(2.7)

rL . . . Larmor radius (also gyration radius rg , or cyclotron radius rc ); [r] = m

m . . . particle mass; [m] = kg
v . . . component of particle velocity perpendicular to B; [v] = m/s
B . . . magnetic field; [B] = T
e . . . elementary charge 1.602 1019 C
m
According to (2.7) electrons with equal v have a by the factor mpe = 1836.1 shorter

Figure 2.1: Components of the velocity vector regarding B

18

2.2 Special Magnetospheric Processes

cyclotron radius than protons.
me v
eB
mp v
rL,p =
eB
rL,p
rL,e =
1836.1
rL,e =

(2.8)
(2.9)

Therefore the cyclotron radius can be written as:

rc =

v
c

(2.10)

2.2 Special Magnetospheric Processes

The magnetosphere provides a variety of different plasma populations. Particles move
along complicated paths which leads to specific current systems.

2.2.1 Motion of Charged Particles in Magnetic and Electric Fields

The motion of a charged particle can basically be regarded as a helical path around a
field line. The helical path can be separated into
gyration around the field line and
translation along the field line.

Figure 2.2: Motion of the electron around a magnetic field line

19

2 Influence of a Magnetic Field on the State of Plasma

If there is no or only an insignificant temporal variation within the magnetic field the
particle motion can be regarded as circular motion around a so-called guiding center ,
which moves along the field line.
r = R + rc
(2.11)
r . . . position vector of the path of the charged particle
R . . . position vector of the guiding center
rc . . . radius vector of the cyclotron path
The cyclotron radius of a charged particle in a magnetic field B can be obtained by

Figure 2.3: Gyration of an electron around a magnetic field line where v and B are not
parallel
setting equal Lorentz-force and the centrifugal force:
F = e[v B] = m

2
v
r

(2.12)

2
v
r
mv
eB

ev B = m
rc =

mv sin
eB
v . . . component of velocity perpendicular to magnetic field
. . . (v, B), i.e. angle between velocity vector of the particle and magnetic field;
(sin = v /v)
rc =

In order to take into account relativistic effects it is necessary to introduce the Lorentzfactor (c . . . velocity of light):

20

R =

1
q
1

rc =

mR v
eB

v2
c2

(2.13)

(2.14)

2.2 Special Magnetospheric Processes

In the following, we do not assume relativistic conditions, thus we will not use this factor
any more, R 1.
The gyration period c is given by the time a particle with velocity v needs to pass the
circular orbit 2rc .
rc
c = 2
(2.15)
v
rc
m
=
v
eB
Thus c is given by:
m
c = 2
(2.16)
eB
the cyclotron frequency is therefore:
fc =

1 eB
1
=

c
2 m

(2.17)

A field can be considered as not or slightly variable in time or space if the following
conditions are met:
1
| B|

small spatial changes in B
(2.18)
|B|
rc


 dB 
B


small temporal changes in B
(2.19)
 dt   c
If these requirements are met the field can be seen as static, the particle is not supplied
with energy. In this case the magnetic flux through the circular area described by the
cyclotron radius is constant:

Figure 2.4: The magnetic flux, MF

d
=0
dt

(2.20)

can be written as:

MF = rc2 B
MF =

2
m2 v
p2
B
=

= const
e2 B 2
e2 B

(2.21)
(2.22)

21

2 Influence of a Magnetic Field on the State of Plasma

p . . . momentum perpendicular to B.
If MF is multiplied by a constant value of e2 /2m we get:
MF

p2
e2
p2
e2
= 2
=
2m
e
B 2m
2mB

= const

(2.23)

p2
2mB

= const

(2.24)

=
is the so-called first adiabatic invariant.

2.2.2 Derivation of the dipole moment

In the following sections we will need a relation between the magnetic moment and the
magnetizing force of a dipole; in particular we will also require the vector components
of the magnetic induction in spherical coordinates. For this reason, these physical quantities shall be derived right here. A fictitious magnetic monopole +p at P (x, 0, z) has a

Figure 2.5: Magnetic dipole with vector r to test point P, under the angel to the z-axis
potential of
1 p
.
4 r+
Therefore the entire potential of a dipole becomes


1 +p p
+ + =
+
.
4 r+
r
+ =

22

(2.25)

(2.26)

2.2 Special Magnetospheric Processes

The poles are placed at +p 0, 0, 2l and p 0, 0, + 2l and have a distance of r+ =
h
h

2 i1/2

2 i1/2
and r = x2 + z 2l
respectively from the test point P . Since
x2 + z + 2l
y = 0 we get r2 = x2 + z 2 . The distances may be written as
r+


1/2 
l2
2
2
= x +z +zl+
= r2 + z l +
4

1/2 
l2
2
2
= x +z zl+
= r2 z l +
4

l2
4

1/2

l2
4

1/2
.

Is l  r, quadratic terms of l may be omitted and the entire potential becomes



+p
1
p
+
4 (r2 + z l)1/2 (r2 z l)1/2

1/2

1/2 i
1 h 2
p r +zl
r2 z l
.
4

=
=

(2.27)

By performing the following development we may simplify these terms:

2

1/2

(r + z l)





z l 1/2 1
z l 1/2
r 1+ 2
=
1+ 2
r
r
r


1
1 zl
1
+ ...
r
2 r2


zl
1
1 2
r
2r


=
=
'

So we get for :
=
=





zl
1
zl
1 2
1+ 2
2r
r
2r


p 1
zl
1
zl
plz
1
cos
3 3 =
= pl 2 ,
3
4 r 2 r
r 2r
4r
4
r
p
4

1
r

(2.28)

with cos = z/r. The dipole moment M being p l, we now may equate the potential of
the dipole
1 M cos
1 Mr
=
=
.
(2.29)
2
4
r
4 r3
Out of this we get the vector components of the magnetic field.
B = 0 grad =

0 M r
4r3

(2.30)

23

2 Influence of a Magnetic Field on the State of Plasma

with (M r)/r3 = M r cos /r3 . Since the potential does not depend on the azimuth
angle , the azimuth component of B, i.e. B equals zero.
Br



1 M cos

= 0
= 0
r
r
4
r2
 


0 M
1
0 M

2
=
=
cos
cos 3
4
r r2
4
r
=

0 M 2 cos
4
r3

B = 0
=

(2.31)



1 M cos
1
1

= 0
r
r
4
r2

0 M 1
0 M 1
() sin
cos =
3
4 r
4 r3

0 M sin
4 r3

B = 0

(2.32)
(2.33)

The dipole is axially symmetric; every meridian layer offers the same structure of field.
For p and +p are placed at (0, 0, +l/2) and (0, 0, l/2) respectively, the dipole moment
is directed downwards (as is presently the situation at the earth).
Using the magnetic field components we may equate the absolute value of magnetic
induction:
"
2 
2 #1/2
q
2
cos

M
sin

M
0
0
+
B =
Br2 + B2 + B2 =

4
r3
4 r3
=
B =
B

0 M p
4 cos2 + sin2
4r3
0 M p
1 + 3 cos2
4r3
1
r3

(2.34)
(2.35)

2.2.3 Derivation of the magnetic moment

Under the conditions for a static field a gyrating particle can be considered as a circular
current.
ev
i = efc =
,
(2.36)
2rc

24

2.2 Special Magnetospheric Processes

fc being the gyration frequency already derived in equations (2.15) and (2.17). This
circular current flows round the area A = rc2 and generates a magnetic moment B :
B = iA =

ev 2
1
rc = ev rc
2rc
2

(2.37)

B =

2m
2
mv
1 ev
=
2 eB
2B

(2.38)

B =

Ekin,
B

(2.39)

This is the 1st adiabatic invariant:

The ratio of kinetic energy of the particle and the magnetic induction is constant everywhere along the field line.

2.2.4 Oscillation (bounce motion)

Non-relativistic representation
The relation for the magnetic moment of a particle directly leads to the analysis of particle
motion in magnetic fields with converging field lines. In this case B is not constant in
space:
Ekin,
Ekin sin2
=
=
= const
(2.40)
B
B
Ekin, . . . kinetic energy of the particle to B
=

2
mv
mv 2 sin2
=
= const
2B
2B

(2.41)

For a particle moving in a converging B field, the angle increases until it becomes /2
(sin /2 = 1). In this point the guiding center of the particle inverts its direction and
moves back towards the direction it has come from. In a static field:
mv 2 sin2 1
mv 2 sin2 2
=
= const
2B1
2B2
sin2 2
sin2 1
=
= const
B1
B2

(2.42)

Principle of the magnetic mirror: The invariance of the dipole moment results in the
reflection of gyrating particles by the so- called magnetic mirror. The effect of reflection
is also characterized by the direction of the force (v B) which is always directed away
from the region with higher magnetic induction.

25

2 Influence of a Magnetic Field on the State of Plasma

Figure 2.6: The principle of the magnetic mirror

Figure 2.7: The magnetic bottle

26

2.2 Special Magnetospheric Processes

Magnetic bottle in the terrestrial magnetic field
B0 . . . magnetic induction at the equator
0 . . . pitch angle between velocity vector and direction of the magnetic field (see fig.
2.7)
sin2
B

sin2 0
B0

sin2 = sin2 0

B
B0

(2.43)

Equation for the magnetic induction of the dipole:

B =
B =

0 MD p
1 + 3 cos2
4r3
0 MD p
1 + 3 sin2
4r3

r = r0 cos2
B=

0 M D p
1 + 3 sin2 =
4r3

p
1 + 3 sin2
0 MD

4
r03 cos6

(2.44)

(2.45)
(2.46)

Applying sin2 = 0 and cos2 = 1 for B0 at = 0:

0 MD 1
3
4
r0
p
p
1 + 3 sin2
1 + 3 sin2
B
0 MD
4
r03
=

=
B0
4
0 MD 1
cos6
r03 cos6
p
1 + 3 sin2
sin2 = sin2 0
cos6
B0 =

(2.47)

(2.48)

= 90 in the mirror point. With sin2 = 1 the transformed equation leads to:
1 = sin2 0
BM =

B
B0

B0
sin2 0

(2.49)

This equation gives the magnetic induction BM in the magnetic mirror point, with a
given B0 and 0 , which is determined by
cos6 M
sin2 0 = p
1 + 3 sin2 M

(2.50)

27

2 Influence of a Magnetic Field on the State of Plasma

There is only gyration for a pitch angle of 0 = 90 . The mirror points of a gyrating
particle on the equator ( = 0 ) are located there too (M = 0 ). Therefore B/B0 = 1.
If the pitch angle = 0 , the particle moves exactly parallel to the field lines, it can
theoretically reach a latitude of = 90 , but here B/B0 .
The relation between pitch angle 0 , equation (2.49), and the magnetic latitude M
follows the curve as shown in fig. 2.8. Intermediate values:
M = 55

BM
B0
M = 30

0 = 8.2

sin 0 = 0.1432
= 48.7

0 = 34.4

sin 0 = 0.5647
BM
B0

= 3.1

Perpendicular and parallel component of the velocity of a particle during oscillating

motion can be given (in addition to the magnetic induction BM at the mirror point and
the relation between pitch angle and magnetic latitude M of the mirror point):
2
v
= v 2 sin2 = v 2 sin2 0

B
B0

vk2 = v 2 cos2 = v 2 (1 sin2 0

B
)
B0

If the mirror point BM is located high enough above the atmosphere, a magnetic reflection will occur without a problem. If the pitch angle is too small if the particle can
spiral along a field line longer and thus gets into deeper atmospheric layers then collision
and absorption by atmospheric particles is more probable than reflection. h = 100 km
can be considered as effective height of the dense terrestrial atmosphere. The following
relation can be determined:
sin2 M
BM

sin2 0 =

sin2 0
B0
B0
B100 km
r

0,100 km = arcsin

(2.51)

B0
B100 km

(2.52)

All particles with pitch angle between

0 0,100 km
enter the atmosphere and are considered lost. The velocity vector of the particles lies in
the atmospheric loss cone with the aperture angle
0,100 km = loss

28

(2.53)

2.2 Special Magnetospheric Processes

Figure 2.8: Relation between the pitch angle 0 and magnetic latitude M

29

2 Influence of a Magnetic Field on the State of Plasma

Figure 2.9: The atmospheric loss cone with loss cone angle Loss
If reflection comes to pass at location BM above the atmospheric loss zone, continuous
oscillation occurs between the mirror points in the northern and southern hemisphere of
the dipole: Within a full period of oscillation the distance between = 0 and = M

Figure 2.10: Particle oscillation between the two mirror points

is passed four times. The oscillation period b (bounce ) is given by:
Z
b = 4
0

dl
=4
vk

Z
0

dl d
d vk

(2.54)

The infinitesimal path element along a field line is given by (curve length dl):
(dl)2 = (dr)2 + r2 (d)2

30

(2.55)

2.2 Special Magnetospheric Processes

Figure 2.11: Path element of a magnetic field line



dl
d

2

dl
d


dr 2
=
+ r2
d
s 
dr 2
=
+ r2
d


field line equation (2.45): r = r0 cos2

dr = r0 2 cos ( sin )d
2
dr
= 4r02 cos2 sin2
d
q
dl
=
4r02 cos2 sin2 + r02 cos4
d

cos4 = cos2 (1 sin2 ) = cos2 cos2 sin2

q
dl
=
3r02 cos2 sin2 + r02 cos2
d
p
dl
= r0 cos 1 + 3 sin2
d
If this term is inserted into the equation for the oscillation period b instead of
vk , we get:
Z

B = 4
0

r0
= 4
v

Z
0

dl 1
d = 4
d vk
M

dl
d

and

p
r0 cos 1 + 3 sin2
p
d
v [1 sin2 0 B/B0 ]

p
cos 1 + 3 sin2
1/2 d


(1+3 sin2 )
2
1 sin 0
cos6
{z
}

(2.56)

S(0 )

31

2 Influence of a Magnetic Field on the State of Plasma

For this expression approximate solutions can be given:
Approximation 1: Hamlin, Karplus, Vik, Watson, Jgr, 1961
S(0 ) ' 1.30 0.56 sin 0

(2.57)

Approximation 2: doctoral dissertation by Wentworth, 1960

S(0 ) ' 1.38 0.32(sin 0 + sin1/2 0 )

(2.58)

The more accurate approximation is approximation 2.

0
0.1
10
20
30
..
.

Hamlin et al.
1.299
1.202
1.108
1.02
..
.

Wentworth
1.366
1.191
1.083
0.994
..
.

60
70
80
89.9

0.815
0.774
0.749
0.740

0.805
0.769
0.747
0.740

From these non-dimensional numbers one can see that the oscillation period is widely
independent from the equatorial pitch angle 0 . Within the possible extent 0 0 /2
the oscillation period varies less than about factor 2. If the magnetic field does not change
significantly during one oscillation period b the following relation is given:
b

1 B
1
B t

(2.59)

In this case the second adiabatic invariant can be derived from the integral that represents
the oscillation:
I
J =
pk dl
Z
J

= 2

lM2

lM1

lM1 . . . position of the mirror point M1

lM2 . . . position of the mirror point M2

32

mvk dl

(2.60)

2.2 Special Magnetospheric Processes

The second invariant is determined by integration along the field line between two conjugate mirror points. The so-called integral invariant I is defined by:
Z lM
2
J
1
I=
=
2
mvk dl =
2mv
2mv
l M1
q
Z lM v 1 sin2 0 B
Z lM
2
2 vk
B0
dl =
dl
v
v
lM
lM
1

The magnetic induction in the mirror point is known already:

Z lM r
2
B
dl
1
I=
BM
lM

(2.61)

The advantage in using the integral invariant is that in a static magnetic field I only
depends on the field configuration. If B/t 6= 0 equation (2.60) has to be applied.

2.2.5 E B-Drift
Starting from the equation of motion
dv
m
= e(E + v B)
dt
the parallel component is examined first:

(2.62)

mv k = eEk

(2.63)

v B has no component parallel to B. The electric field Ek cannot be maintained

because electrons parallel to B are extremely mobile. In the following the perpendicular
component will be investigated. In order to do so we assume: E = Ex i
mv x = eEx + eBvy
v x =
v x =
v y =

eB
e
Ex +
vy
m
m
e
Ex + c vy
m
0
|{z}

(2.64)

c vx

(2.65)

there is no Ey

Second time derivative:

vx = c v y = c2 vx
vy = c v x = c (
vy = c2 (vy +
Substitution: vy0 = vy +

Ex
B

Ex
)
B

(2.66)
e
Ex + c vy )
m
(2.67)
(2.68)

33

2 Influence of a Magnetic Field on the State of Plasma

After this substitution the second derivative leads to the following expressions:
vx = c2 vx

(2.69)

vy0 = c2 vy0

(2.70)

This is a cyclotron motion. It is superimposed by a drift of the guiding center in ydirection. This drift is the so-called E B-drift:
vD =

EB
B2

(2.71)

The physical fundamentals of the EB-drift are founded in the Lorentz transformation.

Figure 2.12: Scheme of the E B-drift

In a moving system the following relation is given:
E0 = E + v B

(2.72)

For an independent particle the following condition has to be met: E0 = 0. E = v B.

The Lorentz transformation is independent from charge, therefore the E B-drift is
charge-independent too.
If an external force F acts on a charged particle; it carries out a motion that can be
described by the following equation:
dp
= F + e(v B)
dt

(2.73)

In a moving system, which translates with the velocity of the guiding center, the particles
conduct a circular motion around the field line. With these assumptions the force F
must be compensated by an induced electric field or by a force, which can be deduced
from the induced electrical field: eE0 . In a moving system the fields are given by:

34

B0 B

(2.74)

E0 = E + (vD B)

(2.75)

2.2 Special Magnetospheric Processes

The B field is almost equal in both systems. The induced E field however consists of
an E field that may be already existing and a part that occurs in the dashed system:
(vD B). vD is the transversal velocity of the moving system, and as well the velocity
of the guiding center.
The sum of the appearing forces is zero as well in a static system as in a moving system:
eE0 + F0 = e(vD B) + F = 0

(2.76)

Cross product with B leads to:

(2.76) e [vD B] B + F B = 0
| {z }

(2.77)

E0

F B = eB [vD B]
= e{(B B)vD (B vD )B}
= eB 2 vD eB(vD B)
F B
FB
=
(2.78)
2
eB
eB 2
For the cross product only the perpendicular component of F is used. Due to this fact
the cross product is made with F instead of F . The drift velocity is perpendicular to
the external force F and to the magnetic induction. The external force F can be replaced
e.g. by an electrical field with the so-called electric force eE:
vD =

F = eE

(2.79)

eE B
EB
=
,
(2.80)
eB 2
B2
which, in addition, shows v = E/B. The following assumptions are applied: The drift
does not depend on charge of a particle, mass or energy. It is only dependent on the
configuration of the E and B fields. To give further details about particle motion in the
magnetosphere, those external forces that stem from the dipole are especially relevant:
vDE =

(a) the gradient of the magnetic field

(b) the curvature of the field lines
ad (a) Drift generated by a gradient in the magnetic field The force component parallel
to the gradient of the B field is given by:
FL cos = ev cos (B + | B|rc cos )

(2.81)

The Lorentz force is always of inverse direction of the gradient

= 0 :
= 90 :
= 180 :

B + | B|rc = B2
B+0=B
B | B|rc = B1

35

2 Influence of a Magnetic Field on the State of Plasma

Figure 2.13: Gradient of magnetic field strength

Figure 2.14: Drift caused by the magnetic field (GC . . . guiding center)

36

2.2 Special Magnetospheric Processes

Force averaged over full gyration:

|F| =

1
2

ev (B + | B|rc cos ) cos d

0

Z 2
Z 2

1
2

(|
B|r
cos
)
d
+
ev
(B
cos
)d
c

2
0
0
{z
}
|
0

Z
0

1
1
1
(cos2 ) d = ( + sin(2)) |2
sin(4) (0 + 0)
0 =+
2
4
|4 {z }
0

(2.82)
Fav =

1
ev Brc
2

1
Fav = ev Brc
(2.83)
2
This force is directed oppositely to the Lorentz force. It gives rise to a drift, the
so-called gradient drift. With
mv
rc =
eB
one obtains:
mv 2
1
mv
Fav = ev B
= B = B B
(2.84)
2
eB
2B
From equation (2.78) follows:
vDG =

FB
B B B
=
2
eB
eB 2

(2.85)

B gradient drift is dependent on the energy of particles (the kinetic energy can be
found in B ) and on the charge e. Again it is demanded that the variation of the
external field B must be small compared to the cyclotron radius:
Postulation :

rc

| B|
1
|B|

(2.86)

ad (b) The curvature drift occurs due to a force FC which results from the motion of a
mass along a curved field line, thus centrifugal forces emerge.
centrifugal force: FC =
vDC =

mvk2
RC eB 2

nB =

mvk2
RC

mvk2 n B
eRC B 2

(2.87)

(2.88)

37

2 Influence of a Magnetic Field on the State of Plasma

Figure 2.15: Parameters describing the curvature drift

Important: Neither the drift resulting from the gradient of a B field nor the
curvature drift appear isolated because curvature of the field line always appears in a natural non-homogeneous B field. Thus both kinds of drift always
appear combined. If currents are negligible, the rotation can be applied for
the B field: rot B = 0.
The perpendicular gradient B can be juxtaposed with the curvature radius of the
field lines RC :
B
B =
n
(2.89)
RC
The exact derivation can be conducted within the scope of MHD. The normal vector n
is given by:
RC B
n=
(2.90)
B
Insertion into (2.88) yields:
vDC =

mvk2 (RC )( B) B
B3

eRC

vDC =

mvk2 ( B) B
eB 3

(2.91)

From eq.(2.85) follows:

vDG =

2 BB
mv

2BeB 2

(2.92)

Addition of both drifts leads to:

2 BB
mvk2 B B
mv

2eB 3
eB 3
m
(v 2 + 2vk2 ) B B
2eB 3

vD =
vD =

38

(2.93)

2.2 Special Magnetospheric Processes

If the relations for v and vk are put in, the following can be obtained:
2
(v
+ 2vk2 ) = v 2 sin2 + 2v 2 cos2 = v 2 (sin2 + 2 cos2 )

v 2 (1 + cos2 ) = v 2 (2 sin2 )
| {z }
1sin2



B
2
2
2
2
2 sin 0
v (2 sin ) = v
= v2
B0

!
p
2
1
+
3
sin

2 sin2 0
cos6

Thus a drift velocity vD for a particle with a defined equatorial pitch angle 0 along a
given field line in geomagnetic latitude can be obtained.
!
p
2
mv 2
1
+
3
sin

2 sin2 0
B B
(2.94)
vD =
2eB 3
cos6
Including the following approximation: B = (B/RC )n and with the restriction that
oscillation only takes place near the equator in magnetic latitudes about 20 the
dipole field line can be approximated by the osculating circle with the curvature radius
RC =

r0
,
3

(2.95)

r0 is the apex distance of the field line.

B = B

vD =

|vD | =
B =
B =

3
n
r0

(2.96)

!
p
3
1 + 3 sin2
2 sin 0
B( B)b n
6
cos
r0
!
p
2

3 mv 2
1
+
3
sin
2 sin2 0
2 eBr0
cos6
mv 2
2eB 3

0 MD p
1 + 3 sin2
4r3
p
0 MD 1 + 3 sin2
cos6
4r03

One finally obtains:

|vD | =

|vD | =

!
p
4r03 cos6
1 + 3 sin2
3 mv 2
2
p
2 sin 0
2 er0 0 MD 1 + 3 sin2
cos6
"
!#
p
2
6mv 2 r02
cos6
1
+
3
sin

p
2 sin2 0
2
0 MD e
cos6
1 + 3 sin

(2.97)

39

2 Influence of a Magnetic Field on the State of Plasma

This is the approximation for oscillation along the osculating circle ( 20 ). vD gives
the drift velocity of charged particles in a dipole field. To simplify matters the structure
of the dipole field line is replaced by the osculating radius. Therefore equation (2.97) can
only be applied to a restricted range of latitude of about 20 +20 .
Another estimation for the value of the combined drift velocity is made, supposing that
oscillation occurs only narrowly over and under the magnetic equator: pitch angle 0
90 , 0 . So the expression in parentheses in equation (2.97) is reduced to:

}|
{
z
z }| {
p

6
2

cos

1
+
3
sin

p
= 1
2 sin2 0

6
2
cos

1 + 3 sin

| {z }
|
{z
}
1

With the definition of the shell parameter

L=

r0
rp

(2.98)

r0 . . . apex distance of the field line

rp . . . radius of the planet
In case of minimum oscillation around the magnetic equator:
vD =

6 mv 2 rp2 2

L
0 M D e

The relation vD = r0

(2.99)
(2.100)

(
gives the angular velocity of a particle moving away of the guiding field line.
t = ),
This motion occurs transverse to the field structure.
The angular velocity can be estimated:
=

vD
r0

6mv 2 rp2 L2

0 MD e rp L

vD
r0

6 mv 2 rp

L
0 MD e

(2.101)

Protons hardly ever reach velocities high enough for relativistic effects to occur, for
electrons the Lorentz factor is relevant. Another interesting effect is to be mentioned
here: L, this means that particles with greater distance in the magnetosphere, with
a given amount of energy move faster around the planet! These relations for the drift
velocity are only valid if the pitch angle 0 90 .

40

2.2 Special Magnetospheric Processes

Figure 2.16: Motion of the guiding field line

Summary of particle drift Oscillation of a particle between the two mirror points is
juxtaposed with a drift resulting from the gradient of the B field and the curvature of
the field line. gradient curvature drift: This drift is directed westward for protons and
eastward for electrons for the case of the Earth dipole. For Jupiter and Saturn the drift
is of opposite direction due to the upright orientation of M: westward for electrons and
eastward for protons. Characteristic times for particles performing these three kinds of
motion in the terrestrial magnetic field: r0 = 4rp L = 4; 0 90 .
e - , 1MeV
p + ,1keV
p + ,1MeV

Gyr.period c
220 s
0.14 s
0.14 s

Oscill.period b
0.27 s
172 s
5.45 s

Driftperiod d
0.061hr 3.66 min
184 hr
0.18 hr 10.8 min

(Source: Lyons and Willimas, Quantitative Aspects of Magnetospheric Physics, p. 26)

Motion of charged particles in a magnetic dipole field is composed of three separate
movements:
Gyration - around the field line
Oscillation - along the field line
Drift - across the magnetic field lines (around the planet)
From the appearing periods one can see characteristic time scales:

41

2 Influence of a Magnetic Field on the State of Plasma

Electrons
Protons

Gyration
s
ms

Oscillation
tenth s
s

Drift
min
hrs

If charged particles get to their starting point after a drift of 2, in other words after a
full orbit around the planet, this motion starts from the beginning: Particles are trapped
in a dipole field due to their complicated paths, they are referred to as trapped particles.
The motion around the field line, the so-called gyration is crossed out by averaging, so
the gyration is not relevant for the motion between the mirror points. Motion between
the mirror points, the so-called oscillation is crossed out if drift motion is considered,
oscillation is not relevant for motion around the Earth. oscillation and drift are important
if the surface on which the guiding center moves is viewed. Complete azimuthal motion

Figure 2.17: Part of a drift shell

yields a surface which is in three dimensional view arched: the so-called drift shell . The
drift shell is the sum of all guiding field lines.
The 3rd adiabatic invariant In case temporal variations occur very slowly compared to
the drift period (if the following relation is given):
|
| B
drift t  1
|B|

(2.102)

The third theorem of conservation can be expressed: The magnetic flux , which is
encircled by the drift shell of the particle is constant.
I
= A0 dx = const
(2.103)

42

2.2 Special Magnetospheric Processes

Figure 2.18: Magnetic flux through the drift shell

A0 . . . magnetic vector potential
The integral is performed along a curve located in the drift shell of the particle. The

Figure 2.19: Drift shell above the magnetic equator

drift shell of a particle is defined by the shell parameter L:
L=

r0
rE

(2.104)

In case of a dipole field the magnetic flux through the drift shell (through the northern
or southern opening) is given by:
=

2MD
r

43

2 Influence of a Magnetic Field on the State of Plasma

r0 . . . zenith distance of a field line
We get the magnetic flux by using equation (2.103) and applying the integral theorem
of Stokes:
I
ZZ
ZZ
( A0 ) dS =
= A0 dx =
B dS
(2.105)
{z }
|

It is identical with the value derived from integration over the part of the equatorial plane
which lies outside the intersection line of the shell with . All particles reflecting on
the same magnetic field line (dipole field line) have the same value for the third adiabatic
invariant.

Figure 2.20: Illustration of the determination of the magnetic flux

/2 Z 2

Br dS

Z
=

(2.106)

/2

0 M 2 sin r d

3
r
| 4 {z
}

2r cos d
0

Br

For the radial component of B we take the result already obtained above, (eqn. 2.34)
Br =

0 M 2
cos
4 r3 | {z }
sin

44

2.2 Special Magnetospheric Processes

Z /2
Z 2
0 M 2
2
=

2r
sin cos d
d
4 r3

0
Z
Z 2
0 M /2
sin cos d
d
=
r

0
Z

/2

sin cos d =

1
sin2 /2 1
| =
1 sin2 = cos2
2
2
2

0 M
0 M
cos2 2 =
cos2
2r
r

cos2

(2.107)
(2.108)

The magnetic flux therefore reaches a maximum at = 0 (defining the equator) and a
minimum at = 90 (defining the pole, here = 0).

45

3 Plasmas as Fluids
3.1 Plasmaphysics
In plasma the situation is substantially more complicated than the one-particle consideration that has been dealt with up to now. Electric and magnetic fields cannot been stated
fixed, they are determined by positions and movements of the particles themselves.
One has to solve a self consistent problem: The globally dominant fields have to be
supplied with the fields that are generated or attenuated. If one tried to consider every
single particle even with a super computer and to determine resulting paths and
fields, this would turn out to be impossible. One thing to consider: change in position of
one single particle causes a deviation of position of all the other (n 1) particles (e.g. in
a laboratory plasma there are n 1012 protons and electrons within 1 cm3 ), in addition
the particle causes a change of all prevailing three-dimensional fields.
In the MHD theory (magnetohydrodynamics) the individual particle is negligible, only
the motion of a fluid element, this is an ensemble of particles is considered. Of course
there are problems that can be poorly or not at all be dealt with. In such cases one
has to go back to the one-particle model. In this context the Monte Carlo method is
worth mentioning. Position, velocity and resulting fields of up to 104 or 105 particles can
be defined by means of this method. The main problem concerning this method is the
requirement for a huge storage capacity within the computer. Another possibility if the
MHD theory fails is the kinetic theory, but it is connected with a substantially higher
level of difficulty.
In plasma-physics one uses the Maxwell equations for vacuum. They are written as:
div E = E =

curl E = E =

(3.1)
B
t

div B = B = 0
curl H = H = j +
(B = 0 H)

46

B = 0 j +

1 E
c2 t

(3.2)
(3.3)

D
t

(3.4)
(3.5)

3.2 The Convective Derivative

In the so-called vacuum equations E and B are used (instead of D and H) as well as
the relations = 1 and = 1.
Maxwells equations determine the conditions of E and B for a given state of plasma. In
order to solve this self consistent problem it is necessary to know the response of plasma
to the E and B field, in other words the response has to be defined by an equation. For
a single particle the already known equation of motion can be applied:
m

dv
= e(E + v B)
dt

(3.6)

Supposing that collisions and thermal motion are not considered a general velocity u can
be defined for all particles in a fluid element:
mn

du
= en(E + u B)
dt

(3.7)

n . . . particle number density (particles per volume unit)

At this point it is necessary to define the so-called convective derivative:

3.2 The Convective Derivative

In the single particle model (equation with v) the derivative with respect to time is used
in a coordinate system moving with the particle. An equation for a liquid element must
be defined in a fixed coordinate system. The following relation can be applied for any
quantity if the corresponding transformation is conducted:
G = G(x, t)
G is one-dimensional at the moment because it depends on x. In a coordinate system
moving with the fluid temporal variation of G is given by the sum of two terms:
dG(x, t)
G G x
G
G
=
+
=
+ ux
dt
t
x t
t
x

(3.8)

G/t represents the variation of G in a fixed reference point, ux G/x represents the
variation of G that way that the observer moves with the fluid into a region with a
different value in G.
The following relation can be given generally in three dimensions:
dG
G
=
+ (u )G
dt
t

(3.9)

This equation is the convective derivative. Sometimes it is written as DG/Dt. The

expression (u ) (means u in Nabla) is a scalar differential operator. The following

47

3 Plasmas as Fluids
operator can be obtained with u = u1 i + u2 j + u3 k.
u = (u1 i + u2 j + u3 k) (

i+
j+
k)
x
y
z

The in-product gives: i i = 1

ij = 0
..
.
u = u1

+ u2
+ u3
x
y
z

(3.10)

(From: Schaum, Vektoranalysis, p. 74)

(u applied to the vector quantity G = G1 i + G2 j + G3 k)




G

G
G
G = u1
(u )G =
u1
+ u2
+ u3
+ u2
+ u3
x
y
z
x
y
z




G1
G1
G2
G2
G1
G2
+ u2
+ u3
+ u2
+ u3
=
u1
i + u1
j
x
y
z
x
y
z


G3
G3
G3
+ u2
+ u3
k
+
u1
x
y
z
If plasma is concerned u can be used instead of G. The following relation is given:


u
mn
+ (u )u = en(E + u B)
(3.11)
t
u/t . . . derivation with respect to time within a coordinate system fixed in space
The initial assumption that there are no collisions and no thermal motion in a plasma
cannot be held up. These terms can be added to the equation of motion.

3.3 Consideration of Temperature

If thermal motion is considered this leads to a force of pressure. This force results from
the motion of particles into a liquid element and out of it. Such a force does not occur
in the equation for a single particle. Pressure is connected to temperature:
p = nkT

(3.12)

n . . . particle number density

k . . . Boltzmann constant
T . . . absolute temperature
When a magnetic field is present two different temperatures are given T and Tk . Two

48

3.4 Consideration of collisions

different pressures emerge from this: p = nkT
pressure tensor which is symbolized by P , e.g.:

p 0

0 p
P =
0
0

p 0 0
P = 0 p 0
0 0 p

and pk = nkTk . Pressure becomes a

0
0
pk

Isotropic temperature distribution leads to the second of the tensors mentioned above.
According to this the motion equation has to be generalized too:


u
+ (u )u = en(E + u B) P
(3.13)
mn
t
Remark: In this case an isotropic liquid element was assumed,this means that the characteristic features of the liquid element are constant in all directions. This is not justified
for in presence of a magnetic field B.

3.4 Consideration of collisions

If collisions between plasma particles are taken into account this has of course an effect
on the equation of motion. To simplify matters only collisions with neutral particles
are considered. (Collisions among charged particles require a more detailed discussion of
this subject, which is not necessary at this point). Any collision of plasmatic and neutral
particles leads to a change of momentum. This is a change in mnu where the change of
momentum is proportional to the relative velocity u u0 . Here, u0 is the velocity of the
neutral fluid. If the time between two successive collisions can be considered constant
with as mean free path time the resulting force resulting from such collisions is
given by:
mn(u u0 )
. . . Momentum per time

Thus the generalized equation of motion (it includes non-isotropic pressure and collisions
with neutral particles) is given by:


u
mn(u u0 )
mn
+ (u )u = en(E + u B) P
(3.14)
t

3.5 Hydrodynamics
Ordinary liquids obey the Navier-Stokes equation:


u
+ (u )u = p + 2 u

(3.15)

49

3 Plasmas as Fluids
...
...

mass density
kinematic coefficient of viscosity

Because the two equations are equal apart from the consideration of the E and B field
a certain question arises: Does the derived relation, the so-called generalized equation
of motion for a fluid element really describe the motion of a plasma component? The
answer to this is: yes! The reasons for this show the limits of the fluid theory at the
same time: as temperature was taken into account in the equation of motion, plasma
(or the plasma components) was implicitly considered to correspond to a Maxwellian
distribution. If this derivation is deliberately conducted with another distribution, the
result does not differ significantly from the relation mentioned above. This means that
the fluid theory is not very sensitive to deviations from the Maxwellian distribution. If
there are significant deviations from this distribution the kinetic theory has to be applied.
If a magnetic field is present, plasma which is collision free in general behaves like
a fluid in which collisions are apparent. The magnetic field forces particles that flow
independently if they have a velocity component perpendicular to the B field to
conduct gyration motions. Fluid theory is an especially good approximation for motions
perpendicular to B.

3.5.1 Equation of continuity

The mass conservation law requests that the overall number of particles N in a given
volume V can only change if there is a net flux of particles normal to the surface of the
included volume V
Z
N
n
=
dV
(3.16)
t
V t
I
N
= n u dS
(3.17)
t
S
Equation (3.16) is equal to the change in particle number density in the volume element.
Equation (3.17) is the particle flux density through the surface (negative sign!). Applying
the Gaussian integral theorem:
ZZZ
ZZ
A n dS =
A dV
(3.18)
S

S . . . surface
A . . . arbitrary vector function
n . . . outwardly directed boundary normal
V . . . volume
Therefore the following equation is given:
Z
I
Z
N
n
=
dV = n u dS =
(n u) dV
t
V t
S
V

50

(3.19)

3.5 Hydrodynamics
The integrands have to be equal because this relation can be applied to any surface:
n
= (n u)
t
n
+ (n u) = 0
t

(3.20)

This is the equation of continuity. On the right hand side any sources and sinks have to
be taken into account.

3.5.2 Equation of state

In order to complete the system of equations one more relation is necessary: the relation
between p and which is substantially determined by the behavior of temperature:
p = p(, T )
Supposing an isothermal state the following relation is applied:
d
p = const
dt

 
p
= 0

With p = nkT we can write:

p = (nkT ) = kT n

(3.21)

p
p

(3.22)

n
n

(From Chen, p. 58)

The adiabatic relation is applied to a gas (plasma) when it cannot release energy:
p = const
Adiabatic change of state
d
dt


=0

is the so-called adiabatic exponent, which is the ratio of the two specific heats: =
cp /cV . In an ideal gas can be determined by the number of degrees of freedom:
=

f +2
f

(3.23)

51

3 Plasmas as Fluids

3.5.3 The full set of MHD equations

Now the complete set of MHD equations is given, where the plasma is to consist of two
components electrons and ions:
charge density: = ni ei + ne ee (ei = ee )

(3.24)

current density: j = ni ei ui + ne ee ue = e(ni ui ne ue )

(3.25)

At this point velocities of the fluid element are used instead of velocities for single particles. ui is applied for ions and ue for electrons.
Maxwellian Equations: See eqns. (3.1) to (3.5) for the definitions of the four Maxwellian
equations
div E =

(3.26)

div E = 0 (without particular space charges)

curl E =

B
t

(3.27)
(3.28)

div B = 0

(3.29)

curl B = 0 j +

1 E
c2 t

(3.30)

Equation of motion: (without collisions and viscosity)



uk
m k nk
+ (uk )uk = ek nk (E + uk B) pk
t

(3.31)

k = i . . . ions
k = e . . . electrons
equation of continuity:
nk
+ (nk uk ) = 0 (k = i, e)
t

(3.32)

Equation of state:
pk = C(mk nk )k

(k = i, e)

(3.33)

Normally, Ohms law is added to these equations; Ohms Law:

jk = (E + uk B)
. . . conductivity
The following unknown quantities appear:

52

(3.34)

3.5 Hydrodynamics
- ni , ne
- pi , pe
- ui , ue (three directions in space)
- E (three directions in space)
- B (three directions in space)
16 scalar unknowns arise from these quantities. The number of scalar equations is sufficient to determine all unknowns: While each curl-equation of the Maxwellians counts
three for each spatial dimension the divergence operator contracts each vector it operates
upon to a scalar thus it contributes as a single equation to our set. As a consequence,
the Maxwellian equations provide 8 eqns., the equation of motion 3 eqns., the equation
of continuity and equation of state each 1 eqn., and Ohms law 3 eqns. which gives on
the whole 16 equations with 16 unknowns that represents a self-consistent set of fields
and motions in MHD theory within the scope of the mentioned approximations. With
these equations plasma can now be described as a continuum, as a conductive fluid. In
order to recognize some connections, some transformations will be done at this point: In
MHD approximation the current density is from Ohms law given by:
j = (E + [u B])
. . . conductivity
E . . . electric field strength
u . . . velocity of the fluid element
B . . . magnetic induction
The problem within the MHD approximation is the consideration of the conductivity
being a constant of matter, which is not true for a plasma in general. In fact, a lot of
plasmas (e.g. the solar wind) are connected with conductivity that high that it can be
approximated to . Effects on this will be dealt with later on.
Transformation of the equation of motion of the conducting medium leads to:



u
mn
+ (u )u = en(E + [u B]) p
t
mn . . .
enu = j
For reasons of simplicity, E = 0 is applied. Transformation:



u
1
1
+ (u )u = p + [j B]
t

53

3 Plasmas as Fluids
If the processes in MHD approximation are considered to be sufficiently slow, the displacement current on the right side (second expression on the right) of the corresponding
Maxwellian equation is negligible:
B = 0 j +

1 E
c2 t

Inserting into equation above:



u
1
1
+ (u )u = p +
[( B) B]
t

0
Vector Analysis provides the rules, the gradient operates on products of vector-fields. In
the following ( B) B is going to be calculated:
(a b) = (b )a + (a )b + b ( a) + a ( b)
(a b) (b )a (a )b b ( a) = a ( b)
( b) a = (b )a + (a )b + b ( a) (a b)
a = b:
( b) b = (b )b + (b )b + b ( b) (b b)
|
{z
}
(b)b

2( b) b = 2(b )b b2
1
( b) b = (b )b b2
2
Thus:
1
( B) B = (B) B B 2
2

(3.35)

The following is obtained:



1
1
1 1
u
+ (u)u = p +
(B)B
B 2
t

0
0 2
The expression (B )B/(0 ) becomes zero if the magnetic field does not change with
progression in direction of B. The following is obtained for the acceleration perpendicular
with respect to the B field:
du
dt
du
dt

1
1 B 2
= p

20


1
B2
= p+

20

(3.36)

Plasma moves perpendicular to the magnetic field as if the magnetic pressure B 2 /20
would work besides the pressure p.

54

3.5 Hydrodynamics


Thus the force B 2 /20 can be considered either as interaction force or as magnetic
force of pressure. The fact that a magnetic force of pressure acts on the plasma is a very
important result from the MHD approximation, this can not be derived from the single
particle model. A magnetic force of pressure occurs at equilibrium between plasma of
the solar wind and the magnetic field of any magnetized planet.

3.5.4 MHD-Drift perpendicular to B:

It seems obvious that if a fluid element within plasma consists of a great number of
particles, it shows the same drift as individual particles: drift perpendicular to an external
magnetic field B. Because there is an expression p (force of pressure) in the equation
of motion, there is a drift linked directly to this force of pressure. This drift appears
in MHD theory but not in single particle model. The equation of motion for a plasma
component (simplified representation without consideration of collisions):

mn
= en(E + (v B)) p
)v
t + |(v {z
| {z }
}
|{z}
2
3
1

Ratio: term 1 / term 3:

term 1
term 3

mn v
t
en(v B)


regarding the


 perpendicular component v


= i
and applying t
=








  mniv  =

 env B 
eB
c

m

Suppose the drift occuring slowly compared to the time scale of c , term 1 can be
neglected. Term 2 can be neglected too, this will be explained later. The situation
prevailing in an isolated column of plasma can be simulated if E and B are homogeneous
and if they have a gradient in n or in p: A drift not appearing in single particle theory
but connected with the p drift will be derived.
Equation of motion, cross product with B:
0 = en[E B + (v B) B] p B
Application of: (a b) c = (c a)b (c b)a
(v B) B = (B v )B (B B)v
= B(v B) v B 2
0 = en[E B + B(v B) v B 2 ] p B

55

3 Plasmas as Fluids
v B v B is connected with the cosine cos /2 = 0. Therefore:

0 = en[E B] env B 2 p B
en v B 2 = en[E B] [p B]
[E B]
1

[p B]
2
B
enB 2
= vE + vD

v =

vE =

EB
B2

vD =

normal E B-drift

p B
enB 2

(3.37)

vD is the so-called diamagnetic drift which only occurs if plasma is regarded as fluid, it is
described by the expression p. vD only appears perpendicular to p, this is why (v)v
can be neglected; this expression takes into account variations of v in direction of v, but
not variations perpendicular to v. The plasma- and field configuration is illustrated in
figure 3.5.4. The equation of state is given by:

Figure 3.1: Representation of B and the diamagnetic drifts of ions and electrons

p
p

n
n

p = p

56

n
n

3.5 Hydrodynamics
Inserting p into the relation for diamagnetic drift (+/ signs for ions and electrons,
respectively), we get:
vD =

p B
enB 2

n
1
p
B
enB 2
n

nkT n zB
enB 2 n

= +

kT z n
eBn

electron: vDe

kT
z n
eBn

(3.38)

ion: vDi

= +

kT
z n
eBn

(3.39)

What is the reason for this drift? The following field and density configuration should
be given: The gradient in density is symbolized by the number of gyration radii. In an

Figure 3.2: Schematics of the diamagnetic drift

arbitrary volume element a greater number of particles moves downward than upward
because the particles streaming downward come from a region of higher density. Thus a
fluid drift perpendicular to n and B exists where the guiding centers stay stationary.
Because of the opposite direction of gyration, the diamagnetic drift of electrons is directed

57

3 Plasmas as Fluids
differently from the drift of ions. A diamagnetic current occurs due to the different drift
directions for electrons and ions: = 1 leads to:


kTi
kTe
jdrift = ne(vDi vDe ) = ne
B n
B n
eB 2 n
eB 2 n
jdrift = (kTi + kTe )

B n
B2

(3.40)

If the particle model is applied, a current cannot be considered to stream if guiding

centers do not drift. In a liquid model a current, jdrift occurs as soon as a gradient in
pressure is apparent. Diamagnetic drift was proved experimentally in the so-called Q
machine.

58

4 Plasma Oscillations and MHD Waves

Any periodic motion of a medium can by Fourier analysis be divided into a superposition of sine or cosine waves with varying frequency and wavelength . If the oscillation
frequency is small, the waveform is sinusoidal in general. Only one component has to be
used.
Example: Oscillation in density
n=n
exp[i(k r t)]

(4.1)

n
. . . constant, defined by the amplitude of the oscillation in density
k . . . propagation vector
Written in cartesian coordinates:
k r = kx x + ky y + kz z
k only has an x-component for wave propagation in direction of x:
n=n
exp[i(kx t)] = n
ei(kxt)
It is intended by convention that if exponential notation is applied, the real part describes
the quantity to be measured. Applying De-Moivres theorem:
ei = cos + i sin
Re(n) = n
cos(kx t)
A wave of time-constant phase, d(phase)/dt = 0 shows that
d
(kx t) = 0
dt
kx t = const

x =
t+C
k

dx
= = vphase
dt
k

(4.2)

Within a wave the E field oscillates.

E = E0 ei(kxt)

(4.3)

real part: Re(E) = E0 cos(kx t)

59

4 Plasma Oscillations and MHD Waves

Because it does not carry any information the phase velocity vphase can be greater than
the velocity of light. An unmodulated wave of infinite length and constant amplitude
does not carry information. The modulation itself the information propagates by the
group velocity, which cannot exceed the velocity of light.
A modulation will be viewed in the following, to do so two waves with almost equal
frequency are added:
E1 = E0 cos[(k + k)x ( + )t]
E2 = E0 cos[(k k)x ( )t]
(k k)x or (k + k)x: Both waves must have the same phase velocity both of them
propagate in the same medium a difference of 2k has to be allowed in the spreading
factor k. ( )t or ( + )t: E1 and E2 vary about 2 in frequency.
substitution: a = kx t
b = (k)x ()t
E1 + E2 = E0 cos[kx t + (k)x ()t]
+E0 cos[kx t (k)x + ()t]
= E0 cos(a + b) + E0 cos(a b)
= E0 (cos a cos b sin a sin b + cos a cos b + sin a sin b)
= 2E0 cos a cos b
E1 + E2 = 2E0 cos[kx t] cos[(k)x ()t]

(4.4)

This is a sinusoidally modulated wave. Information is carried by the envelope of the

Figure 4.1: Modulation of the wave

wave, it is given by cos[(k)x ()t]. The corresponding velocity of information is
/k. Group velocity can be determined in the transition 0:
vg =

60

d
dk

(4.5)

4.1 Plasma oscillations

This group velocity cannot exceed the velocity of light.

4.1 Plasma oscillations

An important form of plasma oscillation is the plasma frequency p :

for electrons: pe =

for ions: pi =

ne2
me 0

1/2

ne2
mi 0

1/2

(4.6)

(4.7)

This plasma oscillation is only dependent on particle density n. It is important to

mention that it does not depend on k, i.e. the group velocity turns out to be zero:
d/dk = 0. The oscillation or disturbance does not expand! One process that can
lead to expansion of plasma oscillations is the thermal motion: Electrons getting out of
regions with such oscillations by thermal motion have impressed information about the
oscillation. This case can be understood as plasma wave, to be exact, as electron plasma
wave. A derivation which will not be conducted in detail here.
3 2 2
2
2
plasma
wave = p + k vth
2

(4.8)

2 = (2kT )/m, where k is Boltzmanns constant.)

(One dimensional consideration with vth
e
The frequency now depends on k; the group velocity can be derived by differentiation of
the equation above:

2d =

3 2
v 2kdk
2 th

vg =

d
3 2 k
= vth
dk
2

vg =

2
3 vth
2 vphase

(4.9)

vg is the group velocity of the electron plasma wave.

4.2 Magnetohydrodynamic Waves

Starting point are the MHD equations.

61

4 Plasma Oscillations and MHD Waves

Maxwellian equations:
B = 0 j
E =

B
t

B = 0

) will not be used here

( E =
0
Ohms Law:
j = (E + v B)
Equation of motion:

v
+ (v )v = j B p
t

Equation of continutity:

+ (v) = 0
t
Equation of state:
p = p(, T )
These fundamental equations will be used to describe simple wave solutions in the following section. A linearization of the equations will be conducted: The equilibrium solution
is indexed 0, the time dependent disturbance is indexed 1. It is also important to note
that the disturbance must be small compared to the not-disturbed quantity. Linearization is conducted for: E E = E0 + E1 , B, v, j, and p. The objective is to describe a
resting plasma, a set of equations for stationary state will be compound. The following

relation is given in stationary state: t

= 0. Thus the MHD zero-indexed equations are
given by:
Maxwellian equations:
B0 = 0 j0

(4.10)

E0 = 0

(4.11)

B0 = 0

(4.12)

( E0 =

) not neccessary here

0

(4.13)

Ohms law:
j0 = (E0 + v0 B0 )

62

(4.14)

4.2 Magnetohydrodynamic Waves

Equation of motion:
0 (v0 )v0 = j0 B0 p0

(4.15)

(0 v0 ) = 0

(4.16)

p0 = p0 (0 , T )

(4.17)

Equation of continuity:
Equation of state:

4.2.1 Linear perturbation theory:

Disturbance variables are now added to this simplified stationary case. Distortions
of quadratic order and higher are neglected. A set of MHD fundamental equations is
obtained for perturbations. The set of equations will be simplified by the following
assumptions:
B0 = (B0 , 0, 0), (B0 being constant), , v0 = 0
(4.18)
0 und p0 are constant in space.
B1 = 0 j1
E1 =

(4.19)

B1
t

B1 = 0
j

(4.20)
(4.21)

= E1 + v0 B1 + v1 B0

for and v0 = 0 we get:

0 = E1 + v1 B0
0

v1
t

= p1 + j1 B0

1
+ (0 v1 ) = 0
t
p1
1
=
p0
0

(4.22)
(4.23)
(4.24)
(4.25)

= 5/3 for a single-atom gas. This set of equations provides a homogeneous linear
system for spatial and temporal behavior of the disturbance quantities E1 , B1 , j1 , v1 , p1
and 1 . The disturbance quantities should show the same behavior in time as e.g. v1 :

t (1 ) = i1 . In order to solve the set of equations (4.19) (4.25) v1 can be given

and a differential equation for v1 can be searched for; approach:
All disturbance variables dependent on v1 are inserted into equation (4.23) subsequently.
From this a linear, homogeneous differential equation for v1 is derived. Any special case
can be investigated with this scheme, where a well defined assumption has to be made
at the beginning.

63

4 Plasma Oscillations and MHD Waves

1st step: v1 into equation (4.22) gives E1 (v1 )
2nd step: E1 into equation (4.20) gives B1 (v1 )
3rd step: B1 into equation (4.19) gives j1 (v1 )
4th step: v1 into equation (4.24) gives 1 (v1 )
5th step: 1 into equation (4.25) gives p1 (v1 )

1st special case: Alfvn waves

Assumptions:
(a) The disturbance variable is dependent only on x and t and is given in the y component: v1 = (0, v1 (x, t), 0). Approach: v1y (x, t) = C eikxit
(b) The B field is given by B0 = (B0 , 0, 0)

Figure 4.2: Directions of B0 and v1,y in space

(c) Supposition of an incompressible plasma oscillation p1 = 0

64

4.2 Magnetohydrodynamic Waves

Inserting into the solution scheme:
1st step:
E1



 x
y z 

= v1 B0 =  0 v1y 0  = +(v1y B0 )
z
 B0 0 0 

E1 = (0, 0, v1y B0 )
2nd step:
E1 =

E1

B1
t


 x
y
z

=  /x /y /z
 0
0
v1y B0


 





z
v1y B0 x
v1y B0 y + 0

y
x


v1y
B0 , 0)
x
v1y depends on x and t. For v1y exclusively, a spatial derivation may be found
because B0 is homogeneous and constant.
Z
B1 = ( E1 )dt + C1
E1 = (0,


Z 
v1y
=

B0 dt
x
C1 = const, is supposed to be 0, B0 is stationary, therefore temporally constant.
Z
Z

1
B0
(v1y )dt = B0
(C eikxit )dt =
(v1y )B0
x
x
i x
therefore: B1 = (0,

1 v1y
B0 , 0)
i x

3rd step:
B1 = 0 j1


 x


y
z  


1 v1y
/y
/z  =

B1 =  /x

B0 x
z
i x
1 v1y

 0
0
i x B0




1 vly
[0]
y+

B0 z
x
i x
j1 =

2 v1y
1
(0, 0,
B0 )
i0
x2

65

4 Plasma Oscillations and MHD Waves

Now all disturbance variables (dependent on v1 ) are inserted. An incompressible, cold
plasma is assumed Cold Plasma Approximation (CPA) (T = 0, therefore p1 = 0);
0

v1
t

j1 B0

v1
t

= j1 B0

 x
z
 y

2 v1y
1
=  0 0 i x2 B0
0

 B0 0
0
= p1 + j1 B0

0 (iv1y ) =
0 v1y =

0 v1y +

p1 = 0( because of CPA)

2 v1y
1
B02
i0
x2
B02 2 v1y
0 x2

B02 2 v1y
0 x2

= 0

2 v1y
0 2 0
+
v1y = 0
x2
B02
k 2 v1y +




1 2 v1y 2

B , 0)
 = (0,

i0 x2 0


DE of an oscillating string

0 0 2
y1y = 0
B02

By using the approach v1y = C eikxit the following relations are obtained:

v1y = ikv1y and

x
2
v1y = k 2 v1y
x2
The characteristic equation is written as:
k 2 v1y +

0 2 0
1y = 0
B02

k 2 +

0 2 0
B02
2

66

0 0
B02

= 0
= k2

4.2 Magnetohydrodynamic Waves

k2
2

0 0
B02

2
k2

B02
2
vA
0 0

vA =

B0
=
k
0 0

(4.26)

vA is the so-called Alfvn velocity

= vphase velocity of the distortion

k
Two modes are obtained by taking the root: one mode with +vA and one mode with
vA , both of them propagate along the field line B0 . k = /vA is derived by inserting
vA again:
i xit
i( vx t)
A
=Ce
v1y (x, t) = C eikxit = C e vA
The initial situation of this special case is to be repeated at this point:
incompressible plasma
conductance
B0 = (B0 , 0, 0) and
v1 = (0, v1 (x, t), 0)

Figure 4.3: Propagation of the Alfvn wave

The distortion B1 the excursion of the field line impressed onto B0 has only a ycomponent, it propagates along the field line with velocity vA . This so-called Alfvn-wave

67

4 Plasma Oscillations and MHD Waves

is the most simple kind of MHD waves.
Thus the assumption of initial conditions leads to a special analytic solution, the Alfvn
waves. The Alfvn velocity only depends on the magnetic induction B and the plasma
mass density; it is therefore a characteristic quantity of the given plasma configuration. Alfvn waves are transversal waves because the direction of propagation (here:
x-direction) is directed perpendicular to the changing amplitude (y-direction). If 0,
the Alfvn velocity becomes higher, within the scope of this approximation vA .
Detailed investigation shows that vA converges towards the velocity of light, in this connection Alfvn waves change into electromagnetic waves.
2nd special case: magnetohydrodynamic compression-waves
Assumptions:
B0 = (B0 , 0, 0)
v1 = (0, v1y (y, t), 0), the disturbance variable only depends on y and t
v1y (y, t) = C eikyit wave propagation is permitted in positive and negative
direction
The y-component of the distortion depends on y, thus the wave is a longitudinal wave
(the direction of propagation coincides with the direction of the distortion amplitude), a
so-called MHD compressional wave. For the plasma is compressed, the adiabatic relation

Figure 4.4: Propagation of MHD compressional waves

is needed:
p1
p0
p1

1
0
p0
= 1
0


vsonic =

p0

1/2

2
p1 = vsonic
1

68

(4.27)
(4.28)

4.2 Magnetohydrodynamic Waves

A differential equation for the disturbance variable v1 is to be found, the same scheme
like in the first special case is used. The only difference is the expression grad p1 , it is
now part of the motion equation:
0

v1y
= p1 + j1 B0
t

Approach:
1st step:
E1 = v1 B0 = (0, 0, v1y B0 )
2nd step:
E1 =

E1

B1
t


 x
y
z

=  /x /y /z
 0
0
v1y B0

E1 = (


 





 = (v1y B0 ) x

(v1y B0 ) y + [0] z

y
x


v1y
B0 , 0, 0)
y

B1 is derived from integration over t:

1 v1y
B1 = ( + i
y B0 , 0, 0 )

3rd step:
B1 = 0 j1



x

y
z 


/y /z  = [0] x
B1 =  /x

 1 v1y B0

0
0
i y


 

 

1 v1y
1 2 v1y

B0 y +
B0 z
i z
y
i
y 2

 2 

v1y
1
B
= 0 j1
B1 =
0, 0, i
0
y 2
1
j1 = ( 0, 0, i
0

Inserted into the equation of motion:


 x


j1 B0 =  0

 B0

j1 B0 =
0,

2 v1y
y 2

B0 )

y
1
0 i
0

0
1
i
0

 z2

v1y
y 2

0


2 v1y
y 2





B0 



B02 , 0

69

4 Plasma Oscillations and MHD Waves

4th step: This step is necessary because of the use of the adiabatic relation. Insertion into the equation of motion:
1
+ (0 v1 )
t

(The other derivations /x and /z are zero)

i1 + 0

v1y
y

= 0

1 =

1 v1y
0
i
y

5th step: Inserting 1 into the state equation p1 /p0 = 1 /0 :

p1
p0

p1 =

1 v1y
0
0 i
y
p0 1 v1y
0
0 i
y
|{z}
vs2

p1 = vs2

0 v1y
i y

Putting into the equation of motion:

0

v1
t

= p1 + j1 B0 =

i0 v1y = vs2

p1
+ j1 B0
y

2
0 2 v1y
1
2 v1y

B
i y 2
i0 0 y 2

This leads to an equation of motion in the y-component:

2 v1y = vs2

2 v1y
B02 2 v1y

y 2
0 0 y 2
}
| {z
2
vA

2
= (vs2 + vA
)

2
(vs2 + vA
)

70

2 v1y
+ 2 v1y = 0
y 2

2 v1y
y 2

4.3 MHD waves with arbitrary angle to the magnetic field

This is the required differential equation. Approach:
v1y = C eikyit
v1y
y
2 v1y
y 2

= ikv1y
= k 2 v1y

2
(vs2 + vA
)(k 2 v1y ) + 2 v1y = 0 characteristic equation

2
2
= vs2 + vA
u2MHDcompr.
k2

(4.29)

This is the velocity of phase of a magnetohydrodynamic compressional wave that propagates perpendicular to the field lines. The square of the phase velocity can be expressed
2 . Which situation is
by the sum of the squares of sonic speed vs2 and Alfvn velocity vA
given in the limit cases B0 0 and B0  1 ?
for B0 0 u becomes u vsonic
for B0  1 u becomes u vAlfvn
The two special cases of wave propagation in plasma lead to: Alfvn waves and MHD
compressional waves. In both modes, the velocity of phase does not depend on frequency
or wave number, therefore in both cases wave propagation is non-dispersive. (The statement that propagation of waves of various frequencies is equally fast is given only because
of the assumed approximation. Exact investigation shows frequency-dependent behavior
Kippenhahn, Mllenhoff, p. 242 ff.).

4.3 MHD waves with arbitrary angle to the magnetic field

The following situation should be given:
(a) The coordinate system is directed that way that the wave propagates in z-direction
(b) The angle lies in between z-axis and B0 field
(c) B0 disappears in x-direction, it is given by:
B0 = (0, B0 sin , B0 cos )
Further assumptions:
- v0 = 0 as before (Plasma is resting in undisturbed case)
- j0 = 0 as before (B0 is homogeneous and B0 = 0 j0 )

71

4 Plasma Oscillations and MHD Waves

Figure 4.5: Image of wave surfaces of MHD waves

- E0 = 0 as before (from Ohms law: from and v0 = 0 follows E0 = 0)
additional assumptions as before: , p0 = const, 0 = const
The following approach is given for the disturbance variables:

v1

..
exp[i(t kz)]
.

p1
Because of this the following is given: /x = 0, /y = 0, /z = ik, /t = i
Special cases:
= 90 . . . MHD compressional waves
= 0 . . . Alfvn waves
It is not possible for any arbitrary value of to reduce the disturbance equation to a
single differential equation. A real system consisting of seven equations must be solved.
The following unknowns are delivered by the subsequent equations:
distort. eq. (4.19), (4.23): 3 equations for v1x , v1y , v1z
distort. eq. (4.20), (4.21), (4.22): 2 equations for B1x , B1y
distort. eq.(4.24), (4.25): 2 equations for 1 , p1

72

4.3 MHD waves with arbitrary angle to the magnetic field

Solving the set of MHD-equations The velocity-vector v1 is oscillating with angular
velocity and propagating into the direction of the wave-vector k = k ez :
v1 = C exp [i(t kz)]
This implies the following substitutions for derivatives operating on velocity components,
we will need shortly after,

0,
0,
ik,
i
x
y
z
t
as well as
2
k 2 and ik ez .
z 2
We start with evaluating the components of the electric field vector E1 .


 ex

ey
ez



v1y
v1z
E1 = v1 B0 =  v1x

 0 B0 sin B0 cos 
= [ex (v1y B0 cos v1z B0 sin ) ey v1x B0 cos + ez v1x B0 sin ]
= (v1z B0 sin v1y B0 cos ) ex + v1x B0 cos ey v1x B0 sin ez
{z
} |
{z
}
|
{z
}
|
c

= a ex + b ey + c ez
We evaluate the magnetic field B1 via the Maxwellian equation
B1
E1 =
.
t
To start with we calculate curl E1 :


 ex ey ez 


E1 =  x y z 
 a b c 






b
c
a
b
a
c

ey

+ ez

= ex
y z
x z
x y


v1y
v1x
v1z
= ex ()
B0 cos + ey
B0 sin
B0 cos
z
z
z
Integration by components yields the magnetic induction.
Z
B1 = E1 dt + C1 with C1 = 0
Z

v1x
B0 cos dt ex +
z

v1y
B0 cos dt ey
z

v1z
B0 sin dt ey
z


1 v1x
1 v1y
v1z
B0 cos ex +
B0 cos
B0 sin ey
=
i
z
z
|i z {z
}
|
{z
}

73

4 Plasma Oscillations and MHD Waves

We evaluate the current density j1 via B1 = 0 j1 .

j1 =



 ex ey ez 


1 

x
y
z

0 
d
e
0 

1
0

() e ex + d ey +

z
z

e d
ez

x y
|
{z
}
0



1
1 2 v1y
1 2 v1z
=
()
B0 cos +
B0 sin ex
0
i z 2
i z 2


1
1 2 v1x
+
B0 cos ey
0 i z 2
=

1
()k 2 v1z B0 sin + k 2 v1y B0 cos ex
i0
+

1
()k 2 v1x B0 cos ey
i0

As we will see soon it is convenient to express j1 by the components of B1 which is why

we evaluate B1 a bit further:
B1 =

1
1
(i)kv1x B0 cos ex +
[(i)kv1y B0 cos + ikv1z B0 sin ] ey
i
i

k
1
= v1x B0 cos ex + (kv1z B0 sin kv1y B0 cos ) ey
| {z
}
|
{z
}
B1x

B1y

Using the components B1x and B1y we evaluate the current density:

j1 = B1y (1)
=

k
k
ex + B1x
ey
i0
i0

ik
ik
B1y ex B1x ey
0
0

The result we insert into the equation of motion,

74

v1
= p1 + j1 B0 ,
t

4.3 MHD waves with arbitrary angle to the magnetic field

Here we have to determine j1 B0 first:

j1 B0



ex
ey
ez

ik
ik

0
0 B1y 0 B1x

0
B0 sin B0 cos 






ik
ik
ik
= ex B1x B0 cos ey
B1y B0 cos + ez
B1y B0 sin
0
0
0



= 


ikB0
(B1x cos ex B1y cos ey + B1y sin ez )
0

We finally come to the equation of motion, being

i0 v1 = p1 + j1 B0

with p1 = vs2 1

or, equivalently by components

v1x
B1x cos
0
B0
.
B1y cos +
0
i0 v1y = ik
0
2
v1z
B1y sin
ikvs 1
Here, we used the adiabatic relation p1 = vs2 1 . An equation for 1 has to be found
subsequently. According to the approach two equations must be found for the disturbance
variables B1x and B1y . From eq. (4.20) and eq. (4.22):
E1 =

B1
t

0 = E1 + v 1 B 0 ,
the following relation is obtained:
E1 = v1 B0

B1x
B1
(v1 B0 ) =
= i B1y
t
0
Since all disturbance variables v1 , B1 , 1 , . . . are proportional to exp[i(t kz)] the last
evaluation is justified. Again, we first calculate the cross-product whereupon we let
operate the cartesian curl operator.


 ex

ey
ez




v1y
v1z
v1 B0 =  v1x

 0 B0 sin B0 cos 
= ex (v1y B0 cos + v1z B0 sin ) + ey v1x B0 cos ez v1x B0 sin

75

4 Plasma Oscillations and MHD Waves





ex
ey
ez



x
y
z
(v1 B0 ) = 

 v1y B0 cos + v1z B0 sin v1x B0 cos v1x B0 sin 
!

...
v1x B0 cos
= ex
y
z





ey
(v1z B0 sin v1y B0 cos )
. . .
x
z

!



+ez
...
. . .
x
y



v1y
v1z
v1x
B0 cos + ey
B0 sin
B0 cos
= ex ()
z
z
z
= ex ikv1x B0 cos + ey (ikv1y B0 cos ikv1z B0 sin )
We get an equation for B1 :

B1x
ikv1x B0 cos
i B1y = ikv1y B0 cos ikv1z B0 sin
0
0
Two relations for 1 and p1 are still missing here. By inserting v1 into the equation of
continuity 1 /t+(0 v1 ) = 0 and consideration of ik ez , we get ik ez 0 v1 =
ik0 v1z for the divergence-term. Thus the equation of continuity finally becomes
i1 = ik0 v1z .
Now we have received six linear equations for the six variables v1x , v1y , v1z , B1x , B1y , 1
and p1 .
i0 v1x =

ikB0
B1x cos
0

i0 v1y =

ikB0
B1y cos
0

i0 v1z = ikvs2 1 +

ikB0
B1y sin
0

iB1x = ikB0 v1x cos

iB1y = ikB0 (v1y cos v1z sin )
i1 = ik0 v1z
Since this is a linear set of equations of form Ax = c, A being the quadratic matrix of
coefficients A = (a1 , . . . , a6 ) = (aik )66 over field R and inhomogenous term c R6 there

76

4.3 MHD waves with arbitrary angle to the magnetic field

is a unique solution, given by Cramers law:
xi =

1
det(a1 , . . . , ai1 , c, ai+1 , . . . , a6 ) for i = 1, . . . , 6.
|A|

Solving a system of equations by determinants An inhomogeneous system of equations generally has the form:
a11 x1 + a12 x2 + . . . + a1n xn = c1
a21 x1 + a22 x2 + . . . + a2n xn = c2
a31 x1 + a32 x2 + . . . + a3n xn = c3
..
.

.. ..
. .

an1 x1 + an2 x2 + . . . + ann xn = cn

In the case of a homogeneous system of equations c1 = c2 = = cn = 0. The unknowns

xi are derived by use of the following relation:
xi =

Di
D

D is the determinant of coefficients in this relation. Di is derived from crossing out

the ith column of the determinant, which is replaced by the right side of the system of
equations. Zeros are inserted for the right side in a homogeneous system of equations,
all Di = 0. A homogeneous system of equations has non-trivial solutions only if the
determinant of coefficients D becomes zero.
A linear homogeneous system of equations with as many equations as variables
has a non trivial solution exactly if the determinant of coefficients becomes
zero.
The system of equations can be solved easier when it is written as matrix of coefficients,
(see Kippenhahn, Mllenhoff: Elementare Plasmaphysik ):
v1x
B1x
v1y
v1z
B1y
1
eq. for
kB0
0
0
0
0
0
v1x
0 cos
B1x
kB0 cos

0
0
0
0
kB0
0
0
0
0
0
v1y
0 cos
kB0
2
0
0
0
0
0 sin kvs v1z
0
0
kB0 cos kB0 sin

0
B1y
0
0
0
k0
0

1
This determinant has to equal zero. It is easy to see from this arrangement of equations
and unknowns that the system can be divided into two subsystems:

77

4 Plasma Oscillations and MHD Waves

(a) system for v1x and B1x
(b) system for v1y , v1z , B1y and 1
According to this the determinant falls apart into a product of two subdeterminants:
D = D1 D2
At least one of the two determinants D1 or D2 , or both of them must disappear if D
shall equal zero. To get the first particular solution D1 = 0:
D1 = 0
0 2

kB0
cos kB0 cos = 0
0

k 2 B02
cos2 = 0
0

phase velocity:

2
k2

B02
2
cos2 = vA
cos2
0 0

(4.30)

This particular solution is an Alfvn wave. For the propagation direction along the z-axis
only the component of the magnetic field B0 in z-direction is relevant.
For the solution D1 = 0, D2 must be D2 6= 0, in this case only trivial solutions are
obtained. The other particular solution can be found if D2 = 0:


kB0


0
0
cos

0


0


kB0
2
0

sin

kv

0
s 
0
D2 = 

 kB0 cos kB0 sin

0 


0
k0
0

The first and fourth line can be exchanged to obtain the solution, it is important to
mention that the sign changes.



2 
0
0
kB

0 sin kvs 



0 
D2 = { 0  kB0 sin


kB0

0
0 
0 cos


kB0

2 
sin

kv
0


s 
0



0 
(k0 )  kB0 cos


kB0

0
0 
0 cos


2 

0

kv
0
s


0 
+0  kB0 cos kB0 sin

0
0
0 


0
0
0
kB

0 sin


 kB0 cos kB0 sin


kB0

0
0
0 cos

78

4.3 MHD waves with arbitrary angle to the magnetic field

D2



kB0
2
2
cos (kvs )0
= k0 (kvs )kB0 cos
0




kB0
kB0
+ 0 0 0 kB0 cos
cos
sin (kB0 sin )0
0
0
= 0

D2 = k 4 0 vs2
2 0

D2 = vs2

B02
k 2 B02
cos2 k 2 vs2 2 20 + 4 20 2 0
cos2
0
0
k 2 B02
sin2 = 0
0

B02
2
2 B 2
4 2 B 2
cos2 2 vs2 + 4 2 0 cos2 2 0 sin2 = 0
0 0
k
k
k 0 0
k 0 0
4 2 2 2 2
2
2 vA 2 vs + vs2 vA
cos2 = 0
k4
k
k

(4.31)

2
2 2
u4 u2 (vA
+ vs2 ) + vA
vs cos2 = 0

(4.32)

Inserting for /k u:

For = 90 , i.e. when the wave surface is directed parallel to the magnetic field, the
following picture arises:

Figure 4.6: Surface of a wave parallel to the B field

2
u4 u2 (vA
+ vs2 ) = 0
2
u2MHD compr. wave = vA
+ vs2

(4.33)

This equation is the solution of the longitudinal MHD compressional wave, which propagates normal to the field lines. For = 0 it is given cos = 1, in this case the following

79

4 Plasma Oscillations and MHD Waves

is obtained:
2 2
2 2
u4 vA
u vs2 u2 + vA
vs = 0

Subst.: u2 = w
2
2 2
w2 (vA
+ vs2 )w + vA
vs = 0

w1,2 =

2 + v2
vA
s

1 2
2 v2
(v + vs2 )2 vA
s
4 A

w1 =

2
vA
v2 v2
v2
2
+ s + A s = vA
= u2
2
2
2
2

(4.34)

w2 =

2
vA
v2 v2
v2
+ s A + s = vs2 = u2
2
2
2
2

(4.35)

First solution: Alfvn wave which propagates transversally along the magnetic field
Second solution: sonic wave which propagates parallel to the magnetic field with
vs
The particular solution D2 = 0 of the determinant therefore contains both special cases
of wave propagation.
D1 = 0
D1 6= 0
D1 D2 = 0 :

D2 =
6 0
D2 = 0

2 cos2 eq. (4.30)

u2 = vA
2 + v 2 eq. (4.33)
for = 90 : u2 = vA
s

2
2
for = 0 : u = vA eq. (4.34) and u2 = vs2 eq. (4.35)
3 different types of waves into a given direction

If the coefficients of the system of equations are calculated exactly, amplitude, phase and
degree of polarization are obtained. Arbitrary angles can be investigated, starting point
is again the already known equation:
2
2
u4 u2 (vA
+ vs2 ) + vs2 vA
cos2 = 0
2
2
u4 u2 (vA
+ vs2 ) + vs2 vA
(1 sin2 ) = 0
2
2
2
u4 u2 (vA
+ vs2 ) + vs2 vA
= vs2 vA
sin2
2
2
2
= vs2 vA
sin2
u4 u2 vA
u2 vs2 + vs2 vA
2
2
(u2 vA
)(u2 vs2 ) = vs2 vA
sin2
2 )(u2 v 2 ) = v 2 v 2 sin2 has two square roots, u2 and u2 . The indices
The relation (u2 vA
s
s A
1
2
can be chosen that way that the following relation is given: u21 > u22 . The solution of
this relation without conducting the calculation in detail leads to:
2 and v 2
u21 maximum values of vA
s

80

4.3 MHD waves with arbitrary angle to the magnetic field

2 and v 2
u22 minimum values of vA
s
2 cos (Alfvn velocity projected onto the direction of the B field)
u22 vA

Figure 4.7: Propagation of the three kinds of MHD waves in a magnetoplasma

The velocities vA and vs are defined by:
vA =

vs

r
r
p
kT
=
=

(4.36)

(4.37)

(1) Fast MHD compressional wave: It is faster than vA and vs , at least vA = vs .

(2) Slow MHD compressional wave: It is slower, maximally in the range of the minimum values of vA or vs .
(3) Alfvn wave: It is given by vA cos and lies between slow and fast MHD compressional wave
In a given direction to the magnetic field B the velocities (1), (2) and (3) are grouped in
the same sequence but at positions according to the equations for the particular velocities.
A circulation of the vector of propagation over 2 leads to a complete polar diagram of
the velocities with the polar angle and the phase velocity u as distance to the coordinate
center. A polar diagram like that is also called Friedrich diagram or hodograph of wave
propagation.
Development of a hodograph for the three kinds of waves:
(a) fast MHD compressional wave
(b) Alfvn wave
(c) slow MHD compressional wave

81

4 Plasma Oscillations and MHD Waves

4.3.1 Development of a hodograph:

1. Drawing vA < vs into the coordinate system containing B0 as abscissa in both
directions +B0 and B0
2. The relation u = vA cos yields two circles with diameter vA
3. The slow MHD compressional wave u2 is always smaller than vA cos due to the
2 cos2 . Two ovals lying horizontally on the abscissa result from
relation u22 vA
this.
4. The vector of u1 leads to an oval standing upright:
(a) = 0 : u1 = vs
q
2 + v2
(b) = 90 : u1 = vA
s

Figure 4.8: Hodograph for the case of vA < vS

The various types of waves were derived from plane waves. It was not taken into account
that waves can be intensified or compensated because of different phase positions. The
interrelationship of hodograph and wave front leads to the realization that in special
positions wave front and tangent to the hodograph correspond to each other. The wave
front is perpendicular to the vector v(). The entity of all wave fronts can be determined
from assuming the coordinate origin as magnetohydrodynamic bang. From this bang
the MHD waves propagate into all directions. Neighboring wave fronts cut on the envelope, on which neighboring waves are in phase, here waves intensify. At other positions
waves are extinguished. The envelope moves with wave propagation, starting from the
center of the MHD bang, is becoming bigger, not changing its shape.
Wave fronts (in phase) for fast MHD waves: Waves propagate into all directions,
though with varying velocity.

82

4.3 MHD waves with arbitrary angle to the magnetic field

Figure 4.9: Interrelationship of hodograph and wave front

Figure 4.10: Wave fronts of fast MHD waves

83

4 Plasma Oscillations and MHD Waves

Figure 4.11: Image of wave fronts of Alfvn waves

Wave fronts for Alfvn waves: For geometric reasons all planes in phase intersect in
one point (principle of Thales) on the right and on the left, these two points mould
the envelope, which is the sum of all wave fronts. The points move away from the
center with vA . Two wave packs emerge from the MHD bang, they propagate along
the magnetic field with vA .
Slow MHD wave: A similar procedure is necessary to find the envelope: the envelope
consists of curves bounding an area with three spinodes. Because of the propagation
away from the center of the MHD bang the propagation area grows, again with
constant shape. Outside this area there is no information about the wave, here the
wave is extinct.

Figure 4.12: Image of the envelope of slow MHD waves

84

4.4 CMA-diagram

4.4 CMA-diagram
Drawing the hodograph into a specific diagram, the CMA-diagram, some characteristics
of the plasma are displayed in a very intuitive way. CMA derives from the inventors of
the diagram: C . . . Clemmov, M . . . Mullaly, and A . . . Allis. We may take up a quite
simplifying position and regard the diagram as a plasma-pond with different parts, each
forming other shapes of wave-propagation as it happens to get distorted. The CMAdiagram, further on abbrev. as CMA, displays the relation of dispersion in a plasma,
since we may qualitatively observe the change of the index of refraction, thus also of
the velocity of propagation with frequency. Using this diagram it is possible to classify
different waves in a cold plasma.
Definition of a cold plasma: A plasma may be regarded as cold, if the socalled plasma-beta is much smaller than 1.

thermal pressure of gas

1
magnetic pressure

(4.38)

nkT
B 2 /20

(4.39)

In a cold plasma, thermal movements may be wholly neglected, since Te = Ti = 0.

4.4.1 Structure of the CMA-diagram

For a specific MHD-wave we may first determine the frequency of gyrationpvia c = eB/m
and so, c B. The plasma frequency may be equated also from p = ne2 /m0 , thus
p2 n. Since a cold plasma is principally defined by its density n and by magnetic
induction B, any MHD-wave with specific frequency is assigned to a specific position
on the CMA.
The abscissa is given by the ratio of p2 and 2 or, equivalently, by the density of
the plasma n.
The ordinate is identical to the ratio of c and , which accords to the magnetic
induction of the plasma wave.
Thus the CMA provides information of n and B. As already mentioned before, a specific
frequency of a plasma wave has a specific position on the CMA, which is to be set as
the origin of the hodograph of the plasma wave. Since c / is vertically orientated,
the hodograph has to be rotated around the origin by an angle of +/2 (thus anticlockwise). Now we can read specific information of wave propagation directly from the
CMA. Additionally, there is quite a lot of information about wave propagation contained
in the CMA, not mentioned and discussed so far, e.g.:
resonances

85

4 Plasma Oscillations and MHD Waves

Figure 4.13: Hodograph, drawn into the CMA-diagram

cutoffs (barrier-frequencies)
polarization: right- and lefthanded
ordinary and extraordinary waves
Thus, the CMA-diagram is a graphical illustration of propagation characteristics of wave
phenomena in a magneto-plasma, which can be separated by resonance- and barrierfrequencies into different areas. Within a specific area the structure of the hodograph does
not change. In the course of propagation, however, a frontier crossing may come to pass;
here the wave experiences resonance and absorption respectively. Thus the hodograph
may: (a) remain the way it is, (b) change into another structure of hodograph, or (c)
vanish completely.

4.4.2 Wave-modes and transitions

The CMA-diagram displays which wave-modes may exist in specific regions on the one
side, and their change of velocity of phase with the angle to the magnetic induction B,
i.e. , on the other side. The distance from the origin to a point on the surface of an
ellipsoid or a lemniscate respectively indices the velocity of phase for this specific angle.
Dispersion relation for the ordinary mode: n2 = 1 p2 / 2 . If p , the index of
refraction and thus, the wave vector becomes zero. At the zero point of the function
n2 () it changes the sign and a wave experiences reflection. So a plasma wave with

86

4.4 CMA-diagram
plasma frequency has a cut-off here, whereas it can propagate for any frequency
> p . In this case, both, n2 and k 2 are positive and real numbers, which is a
necessary condition for propagation. For < p , k becomes an imaginary number:
k = i. By substituting k into the wave equation we get an exponential decrease of
the amplitude: exp(ikx) = exp(x) a wave propagating and thereby vanishing
exponentially like this is called an evanescent wave.
For n2 , the wave will be absorbed entirely by the plasma (resonance condition). This leads to a heating of the plasma.
Extraordinary, (X-) mode: the dispersion relation has a special dependency on the
magnetic field: n2 = n2 (p2 , c2 ).
righthanded polarized wave (R)
lefthanded polarized wave (L)
The CMA-diagram is divided into 13 regions, separated by boundaries. In each of these
regions different modes may be generated and may propagate too. E.g. the cut-off for
the extraordinary wave is given by 2 = c2 + p2 , thus is a quadratic relation between
c / and p2 / 2 . This parabola is displayed in figure 4.14 in the lower left corner, labeled
as upper hybrid resonance.
If, in the course of propagation, a frontier crossing may come to pass, the hodograph
will (a) remain constant concerning its structure (b) vanish entirely, or (c) change its
structure respectively. These transitions shall be discussed now in more detail (see also
figure 4.14).
1. Intact transition: No discontinuity is occurring in the course of transition (so neither resonance nor cut-off). Here, the topological form changes into another form
of same topology.
2. Destructive transition: Here one mode is vanishing exponentially (see evanescent
wave above). The velocity of propagation for this specific mode becomes an imaginary number for any angle . Two different forms may be distinguished:
ovals (0) and rad-lemniscates () experience a destructive transition at a
cutoff boundary, i.e. at P = 0; L = 0 or R = 0 respectively. The surface of
wave vanishes at this transition, see for expample 6 8.
The angle of resonance for a barbell-lemniscate (8), res becomes zero.
The angle of resonance, res for a rad-lemniscate becomes /2. See for example
4 5.
3. Reshaping transition: The real form of one topology changes into a form of another
topology (but real too). See for an instance 8 10. If the angle of resonance for
a barbell-lemniscate becomes res = /2, an oval sphere is formed. However, if the
angle of resonance for a rad-lemniscate runs to zero, here also this lemniscate will

87

4 Plasma Oscillations and MHD Waves

Figure 4.14: CMA-diagram for classification of plasma waves

88

4.4 CMA-diagram

Figure 4.15: Angle of resonance for the barbell-lemniscate

Figure 4.16: Angle of resonance for the rad-lemniscate

form an oval sphere. Although the topology is changed in these cases, the mode
will propagate into the opposite region.

89

5 Appendix
5.1 Physical Quantities in Plasma Physics
A0

0
loss
b
B
B0
BM
B
C
D
E
e
er , e , e
E
0
f
fp
F
FL

i, j, k
I
J
j
k
k
L
dl

90

magnetic vector potential

(v, B)
pitch angle at the equator
angle of the atmospheric loss cone
unity vector tangential to the magnetic field
magnetic induction
magnetic induction at the equator
magnetic mirror point induction
magnetic field gradient
constant, for the most part from integration
determinant
(kinetic) energy
elementary charge
vectors of unity in spherical coordinates
electric field
constant of dielectricity
number of degrees of freedom
plasma frequency
force, in general
Lorentz-force
azimuth angle
adiabatic exponent
vectors of unity in cartesian coordinates
invariant of integration
2nd adiabatic invariant
density of current
wave vector
Boltzmann-constant
characteristic scale-length
line-element of a field line
conductance, geographic latitude
Debye-length
parameter of plasma, scale length of density
mass of a particle

5.1 Physical Quantities in Plasma Physics

MD

n
n

p
P
p, P
r
r
R
rc
RC
rp

b
c
u
v
v
vD
vDC
vg
vS
vth

res
p
c

magnetic moment
1st adiabatic invariant
number of particles in a volume of unity, index of refraction
vector of unity perpendicular to the magnetic field
coefficient of viscosity
pressure, magnetic pole-force
tensor of pressure
force of pressure
radial coordinate
position vector of a particle
position vector of the guiding center
radius of gyration
radius of curvature
radius of a planet
density of charge, density of mass
temperature
mean time between collisions
bounce-period of oscillation
duration of gyration
velocity of an ensemble of particles
velocity of a particle
velocity of a particle perpendicular to the magnetic field
velocity of drift
curvature drift
group velocity
sonic speed
thermal velocity
electric/magnetic potential
magnetic flux
polar angle
angle of resonance
plasma frequency
gyration frequency

91

Index
adiabatic invariant
1st, 22, 25
2nd, 32
3rd, 42, 44
Alfvn
velocity, 67
wave, 64, 78
atmospheric loss cone, 28
Boltzmanns equation, 8
bounce motion, 25
CMA diagram, 85, 86
collisions, 49
convective derivative, 47
coronal mass ejection, 15
criteria of plasma, 6, 14
curvature, 35
radius, 38, 39
cyclotron
frequency, 21
motion, 34
radius, 20, 37
Debye
shielding length, 10
sphere, 12
diamagnetic current, 58
dipole, 22
moment, 22, 23
drift, 42
diamagnetic, 56
ExB, 33
fluid, 57
gradient, 37
gradient curvature, 41
MHD, 55

92

particle, 41
period, 42
shell, 42, 43
equation
Navier-Stokes, 49
of continuity, 50, 52, 62, 63
of motion, 47, 48, 52, 53, 62, 63
generalized, 49, 50
of state, 51, 52, 62, 63
field line, 19, 20, 28, 30, 31, 35, 37, 40,
44
fluid, 46, 49
element, 4648, 50
isotropic, 49
theory, 50, 53
Friedrich diagram, 81
gradient, 7
guiding center, 20
gyration, 19
helical path, 19
hodograph, 81, 82, 86
hydrodynamics, 49
integral invariant, 33
integral theorem
Gauss, 50
Stokes, 44
isothermal, 51
kinetic theory, 46, 50
Langmuir frequency, 13
Laplacian operator, 7
linearization, 62

Index
Lorentz
factor, 20, 40
force, 20, 35
transformation, 34
magnetic
bottle, 27
field, 17, 19, 20, 23, 32, 50
gradient, 35
flux, 21, 42, 45
force
of pressure, 55
mirror, 25
moment, 24
potential, 22, 24
vector potential, 43
magneto plasma, 17
Maxwellian
distribution, 50
equations, 6, 13, 46, 52, 62
MHD
approximation, 54
bang, 82
compressional wave, 68
equations, 52
theory, 46, 53
wave, 79
mirror points, 42
momentum, 49
monopole, 22
Nabla operator, 7, 47
Ohms law, 52, 53, 62
oscillation, 25, 32, 40, 42
period, 31, 32, 41
osculating circle, 39, 40
path element, 30
perturbation theory, 63
pitch angle, 27, 28
plasma
frequency, 13, 61
incompressible, 67
laboratory, 46

oscillations, 61
parameter, 13
wave
electron, 61
plasma criterion
1st, 11
2nd, 13
3rd, 14
Poissons equation, 7
quasi-neutrality, 6, 12
shell
parameter, 40, 43
shockwave, 15
solar wind, 15, 53
fast, 15
plasma, 11, 12
slow, 15
space plasma, 17
temperature, 48, 51
Thales
principle of, 84
thermal motion, 61
transition, 86
destructive, 87
intact, 87
reshaping, 87
translation, 19
trapped particles, 42
velocity
group, 60, 61
of phase, 59, 71
wave
compression, 68
longituinal, 68
MHD compressional, 68
modes, 86
transversal, 68
Wentworth, 32

93

Bibliography
[1] Chen, F. Francis: Introduction to plasma physics. Plenum Pr., 1974
[2] Kippenhahn, R., Mllenhoff C.: Elementare Plasmaphysik. Bibliogr. Inst., 1975
[3] Lyons, R. L., Williams, D. J.: Quantitative aspects of magnetospheric physics.
Reidel, 1984
[4] Spiegel, R. M.: Vektoranalysis. McGraw-Hill, 1977

For further reading

Literature, in general:
[5] Aleksandrov, A. F., Bogdankevic, L. S., Ruchadze, A. A.: Principles of plasma
electrodynamics. Springer, 1984
[6] Arcimovic, L. A., Sagdeev, R. Z.: Plasmaphysik fr Physiker. Teubner, 1983
[7] Diver, D. A.: A plasma formulary for physics, technology and astrophysics. WileyVCH, 2001
[8] Kegel, Wilhelm H.: Plasmaphysik. Springer, 1998
[9] Keudell, A.: Einfhrung in die Plasmaphysik. http://www.ruhr-uni-bochum.de/
reaktiveplasmen/german/lehre/vorlesungen/einfuehrungplasma/skriptPP.
pdf - Stand: 27. Juli 2006
[10] Reiter, D.: Einfhrung in die Plasmaphysik. http://www.eirene.de/lectures/
teil_1.pdf - Stand: 27. Juli 2006

94

Bibliography

Focus on Geophysics:
[11] Cross, R.: An Introduction to Alfvn Waves. Adam Hilger, 1988
[12] Glameier, Karl-Heinz: Plasmaphysik im Sonnensystem. BI-Wiss.-Verl., 1991
[13] Parker, Eugene N.: Cosmical magnetic fields. Clarendon Pr., 1979

Theoretical motivated texts:

[14] Davidson, Ronald C.: Methods in nonlinear plasma theory. Acad. Press, 1972
[15] Ecker, G.: Theory of fully ionized plasmas. Acad. Press, 1972
[16] Fussmann, G.: Einfhrung in die Plasmaphysik. http://www.ipp.mpg.de/BB/
Lehrstuhl/PlasmaphysikI2003.pdf - Stand: 27. Juli 2006
[17] Landau, L. D., Lifschitz, E. M.: Klassische Feldtheorie. Akademie-Verl., 1992

Computational Physics:
[18] Birdsall, C. K., Langdon, A. B.: Plasma Physics via Computer Simulation.
McGraw-Hill, New York, 1985
[19] Hockney, R. W., Eastwood, J. W.: Computer simulation using particles. Hilger,
1992
[20] Matsumoto, H., Sato, T.: Computer Simulation of Space Plasmas. Terra Scientific
Publishing, 1984
[21] Warsi, Z. U. A.: Fluid dynamics - Theoretical and Computational Approaches. CRC
Press, 1999

95