Chapter 1. Collective Plasma Phenomena

CHAPTER 1.
COLLECTIVE PLASMA PHENOMENA 1
Chapter 1
Collective Plasma
Phenomena
The properties of a medium are determined by the microscopic processes in it.

In a plasma the microscopic processes are dominated by collective, rather than
binary, charged particle interactions — at least for sufficiently long length and
time scales.
When two charged particles are very close together they interact through
their Coulomb electric fields as isolated, individual particles. However, as the
distance between the two particles increases beyond the mean particle separa-
tion distance (n−1/3 , in which n is the charged particle density), they interact
simultaneously with many nearby charged particles. This produces a collective
interaction. In this regime the Coulomb force from any given charged parti-
cle causes all the nearby charges to move, thereby electrically polarizing the
medium. In turn, the nearby charges move collectively to reduce or “shield
out” the electric field due to any one charged particle, which in the absence of
the shielding decreases as the inverse square of the distance from the particle.
In equilibrium the resultant “cloud” of polarization charge density around a
charged particle has a collectively determined scale length — the Debye shield-
ing length — beyond which the electric field due to any given charged particle is
collectively shielded out. That is, the “long range force” of the Coulomb electric
field is actually limited to a distance of order the Debye length in a plasma.
On length scales longer than the Debye length a plasma responds collectively
to a given charge, charge perturbation, or imposed electric field. The Debye
shielding distance is the maximum scale length over which a plasma can depart
significantly from charge neutrality. Thus, plasmas, which must be larger than a
Debye length in size, are often said to be quasineutral — on average electrically
neutral for scale lengths longer than a Debye length, but dominated by the
charge distribution of the discrete charged particles within a Debye length.
Most plasmas are larger than the Debye shielding distance and hence are not
dominated by boundary effects. However, boundary effects become important
DRAFT 10:26
August 12, 2003 °J.D
c Callen, Fundamentals of Plasma Physics
CHAPTER 1. COLLECTIVE PLASMA PHENOMENA 2
within a few Debye lengths of a material limiter or wall. This boundary region,
which is called the plasma sheath region, is not quasineutral. Material probes
inserted into plasmas, which are called Langmuir probes after their developer
(in the 1920s) Irving Langmuir, can be biased (relative to the plasma) and
draw currents through their surrounding plasma sheath region. Analysis of
the current-voltage characteristics of such probes can be used to determine the
plasma density and electron temperature.
If the charge density in a quasineutral plasma is perturbed, this induces a
change in the electric field and in the polarization of the plasma. The small
but finite inertia of the charged particles in the plasma cause it to respond
collectively — with Debye shielding, and oscillations or waves. When the char-
acteristic frequency of the perturbation is low enough, both the electrons and
the ions can move rapidly compared to the perturbation and their responses are
adiabatic. Then, we obtain the Debye shielding effect discussed in the preceding
paragraphs.
As the characteristic frequency of the perturbations increases, the inertia
of the charged particles becomes important. When the perturbation frequency
exceeds the relevant inertial frequency, we obtain an inertial rather than adi-
abatic response. Because the ions are much more massive than electrons (the
proton mass is 1836 times that of an electron — see Section A.8 in Appendix
A), the characteristic inertial frequency is usually much lower for ions than for
electrons in a plasma. For intermediate frequencies — between the characteris-
tic electron and ion inertial frequencies — electrons respond adiabatically but
ions have an inertial response, and the overall plasma responds to perturbations
via ion acoustic waves that are analogous to sound waves in a neutral fluid.
For high frequencies — above the electron and ion inertial frequencies — both
electrons and ions exhibit inertial responses. Then, the plasma responds by
oscillating at a collectively determined frequency called the plasma frequency.
Such “space charge” oscillations are sometimes called Langmuir oscillations after
Irving Langmuir who first investigated them in the 1920s.
In this chapter we derive the fundamental collective processes in a plasma:
Debye shielding, plasma sheath, plasma oscillations, and ion acoustic waves.
For simplicity, in this chapter we consider only unmagnetized plasmas — ones
in which there is no equilibrium magnetic field permeating the plasma. At the
end of the chapter the length and time scales associated with these fundamental
collective processes are used to precisely define the conditions required for being
in the plasma state. Discussions of applications of these fundamental concepts
to various basic plasma phenomena are interspersed throughout the chapter and
in the problems at the end of the chapter.
1.1 Adiabatic Response; Debye Shielding

To derive the Debye shielding length and illustrate its physical significance, we
consider the electrostatic potential φ around a single, “test” charged particle in
a plasma. The charged particles in the plasma will be considered to be “free”
DRAFT 10:26
charges in a vacuum. Thus, the electrostatic potential in the plasma can be

determined from
∇· E = −∇2 φ = ρq /²0 , Poisson equation, (1.1)
which results from writing the electric field E in terms of the electrostatic po-
tential, E = −∇φ, in Gauss’s law — see (??) and (??) Section A.2. The charge
density ρq is composed of two parts: that due to the test charge being consid-
ered and that due to the polarization of the plasma caused by the effect of the
test particle on the other charged particles in the plasma. Considering the test
particle of charge qt to be a point charge located at the spatial position xt and
hence representable1 by δ(x − xt ), the charge density can thus be written as
ρq (x) = qt δ(x − xt ) + ρpol (x) (1.2)
in which ρpol is the polarization charge density.

The polarization charge density results from the responses of the other
charged particles in the plasma to the Coulomb electric field of the test charge.
For slow processes (compared to the inertial time scales to be defined more pre-
cisely in Section 1.4 below), the responses are adiabatic. Then, the density of
charged particles (electrons or ions) with charge2 q and temperature T in the
presence of an electrostatic potential φ(x) is given by [see (??) in Section A.3]
n(x) = n0 e−qφ(x)/T , Boltzmann relation (adiabatic response), (1.3)
where n0 is the average or equilibrium density of these charged particles in the

absence of the potential. The potential energy qφ of our test particle will be
small compared to its thermal energy, except perhaps quite close to the test
particle. Thus, we expand (1.3) assuming qφ/T << 1:
qφ 1 q 2 φ2
n ' n0 (1 − + · · · ), perturbed adiabatic response. (1.4)
T 2 T2
The validity of this expansion will be checked a posteriori — at the end of this
section. To obtain the desired polarization charge density ρpol caused by the
effect of the potential φ on all the charged particles in the plasma, we multiply
(1.4) by the charge q for each species s (s = e, i for electrons, ions) of charged
particles and sum over the species to obtain
X X n0s q 2 · µ
qs φ
¶¸
ρpol ≡ ns q s = − s
φ 1+O (1.5)
s s
Ts Ts
in which the “big oh” O indicates the order of the next term in the series
expansion. In obtaining this result we have used the fact that on average a
1 See Section B.2 in Appendix B for a discussion of the Dirac delta function δ(x).
2 Throughout this book q will represent the signed charge of a given plasma particle and
e ' 1.602 × 10−19 coulomb will represent the magnitude of the elementary charge. Thus, for
electrons we have qe = −e, while for ions of charge Zi we have qi = Zi e.
DRAFT 10:26
plasma is electrically quasineutral:

X
n0s qs = 0, quasineutrality condition. (1.6)
s
Retaining only the lowest order, linear polarization charge density response
in (1.5), substituting it into (1.2), and using the resultant total charge density
in the Poisson equation (1.1), we obtain
µ ¶
1 qt
−∇2 + 2 φ = δ(x − xt ) (1.7)
λD ²0
in which the 1/λ2D term is caused by the polarizability of the plasma. Here, λD
is the Debye shielding length:
1 X 1 X n0s q 2 1 1 n0e e2 n0i Zi2 e2

2 ≡ 2 ≡ s
= 2 + 2 = + ,
λD s
λDs s
²0 Ts λDe λDi ²0 Te ²0 Ti
plasma Debye length. (1.8)
In the last expressions we have assumed a plasma composed of electrons with

density n0e and only one species of ions with charge Zi e and density n0i . Note
that for comparable electron and ion temperatures the electron and ion Debye
lengths are comparable. The overall plasma Debye length is obtained from the
sum of the inverse squares of the Debye lengths of the various species of charged
particles in the plasma. For a plasma composed of electrons and protons, which
we will call an electron-proton plasma, the lower temperature component will
give the dominant contribution to the overall plasma Debye length. Numerically,
the electron Debye length is given in SI (mks) units by
r s
²0 Te Te (eV)
λDe ≡ ' 7434 m, electron Debye length. (1.9)
ne e2 ne (m−3 )
The general solution of (1.7) in an infinite, homogeneous three-dimensional

plasma geometry3 is4
qt e−|x−xt |/λD qt e−r/λD

φt (x) = = , potential around a test particle.
{4π²0 } |x − xt | {4π²0 } r
(1.10)
Here, the subscript t indicates this is the particular solution for the potential
around a test charge qt in a plasma. That this is the solution can be verified by
substituting it into (1.7), noting that (−∇2 R+ 1/λ2D )φt = 0 everywhere
RR except
where r ≡ |x − xt | → 0 and there limr→0 d3x ∇2 φ = limr→0 ° dS · ∇φ =
3 For one- and two-dimensional geometries see Problems 1.4 and 1.5.
4 Here and throughout this book we write the mks factor {4π²0 } in braces; eliminating this
factor yields the corresponding cgs (Gaussian) forms for electrostatic response formulas.
DRAFT 10:26
φCoul
φt
Debye
ÃÃÃÃÃÃshielding
bmin n-1/3
Figure 1.1: Potential φt around a test particle of charge qt in a plasma and

Coulomb potential φCoul , both as a function of radial distance from the test
particle. The shaded region represents the Debye shielding effect. The charac-
−1/3
teristic distances are: λD , Debye shielding distance; ne , mean electron sep-
aration distance; bmin = q /({4π²0 }T ), classical distance of “closest approach”
cl 2
where the eφ/T << 1 approximation breaks down.
−qt /²0 . The solution given in (1.10) is also the Green function for the equation
(−∇2 + 1/λ2D )φ = ρfree /²0 — see Problem 1.6.
The potential around a test charge in a plasma, (1.10), is graphed in Fig. 1.1.
Close to the test particle (i.e., for r ≡ |x − xt | << λD ), the potential is sim-
ply the “bare” Coulomb potential φCoul = qt / ({4π²0 } |x − xt |) around the test
charge qt . For separation distances of order the Debye length λD , the expo-
nential factor in (1.10) becomes significant. For separations large compared to
the Debye length the potential φt becomes exponentially small and hence is
“shielded out” by the polarization
R cloud surrounding the test charge. Overall,
there is no net charge Q ≡ V d3x ρq from the combination of the test charge
and its polarization cloud — see Problem 1.7. The difference between φt and
the Coulomb potential is due to the collective Debye shielding effect.
We now use the result obtained in (1.10) to check that the expansion (1.4)
was valid. Considering for simplicity a plasma with Ti >> Te [so the electron
Debye length dominates in (1.8)], the ratio of the potential around an electron
test charge to the electron temperature at the mean electron separation distance
−1/3
of |x − xt | = ne can be written as
h ¡ ¢1/3 i
¯ −1/ 3
eφt ¯¯ exp n λ
e De 1
¯ = 3
' . (1.11)
Te |x−xt | = ne
−1/3 4π (ne λDe ) 2/3 4π (ne λ3De )2/3
DRAFT 10:26
For this to be small and validate our expansion in (1.4), we must require
ne λ3D >> 1, necessary condition for the plasma state. (1.12)
That is, we must have many charged particles (electrons) within a Debye cube
— a cube each side of which is the Debye shielding distance in length.5 Physi-
cally, (1.12) is a necessary condition for the plasma state because it represents
the requirement that, at the mean interparticle separation distance, collective
interactions of charged particles dominate over binary interactions. The number
of charged particles within a Debye cube (or more often its reciprocal 1/ne λ3D )
is called the plasma parameter since it must be large for the medium to be in
the plasma state.
As another check on the validity of the preceding expansion approach, we
next confirm that the electric field energy in the polarization cloud is small
compared to a typical thermal energy for the test particle — the temperature
of that species of particles. The polarization electric field is determined by the
difference between the potential φt around a test charge in the plasma and the
test charge’s Coulomb potential φCoul :
" ¡ ¢#
d q e−r/λD − 1
Epol = −∇ (φt − φCoul ) = − êr (1.13)
dr {4π²0 }r
in which r ≡ |x − xt | and êr ≡ ∇r = (x − xt )/|x − xt | is a unit vector in the

r ≡ x−xt direction. The variation of the polarization electric field as a function
of the distance r away from the test charge is shown in Fig. 1.2.
The energy density associated with this electric field is ²0 |Epol |2 /2. Using a
spherical coordinate system whose origin is at the position of the test charge,
we find that the total electric field energy obtained by integrating the energy
density, normalized to the electron temperature (again assuming Ti >> Te for
simplicity) can be written as
Z µ ¶2 Z ∞ · µ −r/λD ¶¸2
1 ²0 4π²0 q d e −1
d3x |Epol |2 = r2 dr
Te 2 2Te {4π²0 } 0 dr r
I
≡ . (1.14)
8πne λ3D
Here, the dimensionless integral I is simplified using x ≡ r/λD and is given by

Z · ∞ µ ¶¸2 Z ∞ · ¸2
d e−x − 1 1 − e−x
I ≡ dx x = dx e−x −
0 dx x 0 x
Z ∞ " # Z
∞
2 ¡ −x ¢ (1 − e )
−x 2
1
= dx e−2x − e − e−2x + 2
= dx e−2x =
0 x x 0 2
5 Since the intrinsic geometry of the polarization cloud around a test charge is spherical,
plasma physicists often use as the appropriate measure the number of charged particles within
a Debye sphere, (4π/3)ne λ3D , which by (1.12) must also be large compared to unity.
DRAFT 10:26
Er
Epol
Figure 1.2: Coulomb and polarization radial electric fields around a test par-
ticle of charge q in a plasma. Because the plasma polarization acts to shield
out the positive Coulomb electric field, the polarization electric field is negative.
The polarization electric field is finite at the origin, decays smoothly with dis-
tance away from the test charge, and shields out the Coulomb electric field for
separations larger than the Debye length λD .
in which in the first integral form on the second line we have integrated the last
term by parts and cancelled it with the second term, and the final integral is
evaluated using (??) in Appendix C. From (1.14) we again see that our expansion
approach is valid as long as there are many electrons within a Debye cube (or
sphere), since then the electric field energy in the polarization cloud around a
test charge is small compared to the typical, thermal energy of a charged particle
in a plasma.
We can also use the concepts developed in the preceding discussion to es-
timate the level of thermal fluctuations or noise in a plasma. The thermal
fluctuations are caused by the interactions between charged particles through
the electric field around one particle influencing the positions of other parti-
cles within approximately a Debye sphere around the original charged particle.
That is, they are caused by correlations between particles, or by electric field
correlations within the plasma. To calculate these properly requires a plasma
kinetic theory (see Chapter 13). However, the fluctuation level can be estimated
as follows.
A relevant measure of the magnitude of the thermal noise in a plasma is the
ratio of the electric field energy density in the fluctuations ²0 |Ẽ|2 /2 to the ther-
mal energy density nT . The polarization electric field given by (1.13) represents
the correlation electric field between a test particle at xt and an observation
point x. From Fig. 1.2 we see that the polarization electric field is localized to
within about a Debye length of any given charge, and its magnitude there is
not too different from its value at r ≡ |x − xt | = 0: Epol (0) = −q/(2{4π²0 }λ2D ).
Also, we note that all charged particles within about a Debye sphere [namely
DRAFT 10:26
∼ (4π/3)ne λ3D particles] will contribute to the electric field fluctuations at any
given point. Hence, omitting numerical factors, we deduce that the scaling of
the relative electric field fluctuation energy from “two-particle” correlations in
a plasma is given by
µ ¶h i
²0 4π ²0 2
|Ẽ| 2 n e λ 3
D |Epol (0)|
2 3 2 1
∼ ∼ << 1,
ne T e ne Te ne λ3D
thermal fluctuation level. (1.15)
We thus see that, as long as (1.12) is satisfied, the thermal fluctuation level is
small compared to the thermal energy density in the plasma and again our basic
expansion approach is valid. The thermal fluctuations occur predominantly at
scale-lengths of order the Debye length or smaller. The appropriate numerical
factor to be used in this formula, and the frequency and wave-number depen-
dence of the thermal fluctuations in a plasma, can be obtained from plasma
kinetic theory. They will be discussed and determined in Chapter 13.
1.2 Boundary Conditions; Plasma Sheath

A plasma should be larger than the Debye shielding distance in order not to
be dominated by boundary effects. However, at the edge of a plasma where
it comes into contact with a solid material (e.g., a wall, the earth), boundary
effects become important. The region where the transformation from the plasma
state to the solid state takes place is called the plasma sheath.
The role of a plasma sheath can be understood as follows. First, note that
for comparable electron and ion temperatures the typicalpelectron speed, which
will be taken to be the electron thermal speed vT e ≡ 2Te /me [see (??) in
Section
p A.3], is much larger than the typical (thermal) ion speed (vT e /vT i ∼
mi /me ∼ > 43 >> 1). Since the electrons typically move much faster than
6
the ions, electrons tend to leave a plasma much more rapidly than ions. This
causes the plasma to become positively charged and build up an equilibrium
electrostatic potential that is large enough [∼ a few Te /e, see (1.23) below]
to reduce the electron loss rate to the ion loss rate — so the plasma can be
quasineutral in steady state. The potential variation is mostly localized to the
plasma sheath region, which is of order a few Debye lengths in width because
that is the scale length on which significant departures from charge neutrality
are allowed in a plasma. Thus, a plasma in contact with a grounded wall will:
charge up positively, and be quasineutral throughout most of the plasma, but
have a positively charged plasma sheath region near the wall.
We now make these concepts more concrete and quantitative by estimating
the properties of a one-dimensional sheath next to a grounded wall using a simple
plasma model. Figure 1.3 shows the specific geometry to be considered along
6 Many people use an analogy to remember that electrons have much larger thermal veloc-
ities than ions: electrons are like fast moving, lightweight ping pong balls while ions are like
slow-moving, more massive billiard balls — for equal excitation or thermal energies.
DRAFT 10:26
φ plasma
presheath bulkÃplsama
φ∞ sheath
φ
s
0 xs x
n ionÃrich
transition quasi-neutralÃplasma
region
n
ni
ne
0 xs x
Figure 1.3: Behavior of the electrostatic potential and electron and ion densities
in the sheath, presheath (or transition) and bulk plasma regions of a plasma
2
in contact with a grounded wall. For the case shown Te /mi V∞ = 0.9. The
sheath parameters determined in the text are Φ∞ ' 3 Te /e and xS ' 2 λDe .
The long-dash line in the top figure indicates the approximation given in (1.26).
with the behavior of the potential, and electron and ion densities in the plasma
sheath and bulk plasma regions, as well as in the presheath (or transition) region
between them.
The electron density is determined from the Boltzmann relation (1.3):
½ ¾
e [Φ(x) − Φ∞ ]
ne (x) = n∞ exp (1.16)
Te
in which Φ(x) indicates the equilibrium potential profile in the plasma, and the
∞ subscript indicates evaluation of the quantities in the bulk plasma region
far from the wall (i.e., beyond the plasma sheath and presheath regions whose
properties we will determine). (In using this equation it is implicitly assumed
that the background electron velocity distribution is Maxwellian.)
For simplicity, we consider an electron-proton plasma with negligible ion
DRAFT 10:26
thermal motion effects (eΦ >> Ti ). The potential variation in and near the
sheath produces an electric field that increases the flow of ions toward the wall,
which will be assumed to be grounded. The ion flow speed Vi in the x direction
is governed by conservation of energy for the “cold” (eΦ >> Ti ) ions:
1 1
mi Vi2 (x) + eΦ(x) = constant = mi V∞ 2
+ eΦ∞ (1.17)
2 2
in which we have allowed for a flow of ions from the bulk of the plasma into the
presheath region so as to ultimately balance the electron flow to the wall. The
ion flow at any given point is given by
r s · ¸
2e 2e 2
mi V∞
Vi (x) = V∞ +2 [Φ∞ − Φ(x)] = Φ∞ + − Φ(x) .
mi mi 2e
The spatial change in the ion flow speed causes the ion density to change as
well — a high flow speed produces a low ion density. The ion density variation
is governed, in a steady equilibrium, by the continuity or density conservation
equation [see (??) in Appendix A] for the ion density: d(ni Vi )/dx = 0, or
ni (x)Vi (x) = constant. Thus, referencing the ion density to its value n∞ in the
bulk plasma (x → ∞), it can be written as
½ ¾−1/2
2e [Φ∞ − Φ(x)]
ni (x) = n∞ 1 + 2
. (1.18)
mi V∞
Substituting the electron and ion densities into Poisson’s equation (1.1), we
obtain the equation that governs the spatial variation of the potential in the
sheath, presheath and plasma regions:
d2 Φ e
= − (ni − ne )
dx2 ²0
"½ ¾−1/2 ½ ¾#
n∞ e 2e [Φ∞ − Φ(x)] e [Φ∞ − Φ(x)]
= − 1+ 2
− exp − .(1.19)
²0 mi V∞ Te
While numerical solutions of this equation can be obtained, no analytic solution

is available. However, limiting forms of the solution can be obtained near the
wall (x << xS ) and in the bulk plasma (x >> xS ). Even though a simple
solution is not available in the transition region, solutions outside this region
can be used to define the sheath position xS and the conditions needed for
proper sheath formation.
In the quasineutral plasma far from the plasma sheath region (x >> xS ) the
potential φ(x) is very close to its asymptotic value Φ∞ . In this region we approx-
imate the electron and ion densities in the limits 2e [Φ∞ − Φ(x)] /mi V∞ 2
<<1 and
e [Φ∞ − Φ(x)] /Te <<1, respectively:
½ ¾
e [Φ∞ − Φ(x)]
ne (x) ' n∞ 1 − + ··· ,
Te
½ ¾
e [Φ∞ − Φ(x)]
ni (x) ' n∞ 1 − 2
+ · · · .
mi V∞
DRAFT 10:26
Keeping only linear terms in Φ∞ − Φ(x), (1.19) can thus be simplified to

µ ¶
d2 [Φ∞ − Φ(x)] 1 Te
' 1 − [Φ∞ − Φ(x)] . (1.20)
dx2 λ2De 2
mi V ∞
Here, λDe is the electron Debye length evaluated at the bulk plasma density n∞ .
For mi V∞ 2
< Te the coefficient of Φ∞ − Φ(x) on the right would be negative;
this would imply a spatially oscillatory solution that is not physically realis-
tic for the present plasma model, which implicitly assumes that the potential
is a monotonic function of x. Thus, a necessary condition for proper sheath
formation in this model is
p
|V∞ | ≥ Te /mi , Bohm sheath criterion. (1.21)
Since this condition need only be satisfied marginally and the ion flow into
the sheath region typically assumes its minimum value, it is usually sufficient
to make this criterion an equality. The Bohm sheath criterion implies that ions
must enter the sheath region sufficiently rapidly to compensate for the electron
charge leakage through the sheath to the wall. In general, what is required for
proper sheath formation is that, as we move toward the wall, the local charge
density increases as the potential decreases:p ∂ρq /∂Φ < 0 for all x. Also, since
we will later find (see Section 1.4) that Te /mi is the speed of ion acoustic
waves in a plasma (for the plasma model being considered), the Bohm sheath
criterion implies that the ions must enter the presheath region at a supersonic
speed relative to the ion acoustic speed.
As long as the Bohm sheath criterion is satisfied, solutions of (1.19) will be
well-behaved, and exponentially damped in the presheath region: for x >> xS
2 −1/2
we have Φ∞ − Φ(x) ' C exp(−x/h) where h = λDe (1 − Te /mi V∞ ) and C is
a constant of order Φ∞ − ΦS . Thus, forpthis plasma model, in the typical case
where V∞ is equal to or slightly exceeds Te /mi , the presheath region where the
potential deviates from Φ∞ extends only a few Debye lengths into the plasma. In
more comprehensive models for the plasma, and in particular when ion thermal
effects are included, it is found that the presheath region can be larger and the
potential variation in this region is influenced by the effects of sheath geometry,
local plasma sources, collisions and a magnetic field (if present). However, the
Bohm sheath criterion given by (1.21) remains unchanged for most physically
relevant situations, as long as the quantity on the right side is interpreted to be
the ion acoustic speed in the plasma model being used.
We next calculate the plasma potential Φ∞ that the plasma will rise to in
order to hold back the electrons so that their loss rate will be equal to the ion loss
rate from the plasma. The flux of ions to the wall is given by −ni V∞ , which
whenpevaluated at the Bohm sheath criterion value given in (1.21) becomes
−n∞ Te /mi . (The flux is negative because it is in the negative x direction.)
A Maxwellian distribution of electrons will produce (see Section A.3) a random
flux of electrons to the p wall on the left side of the plasma of −(1/4)ne v̄e =
−(n∞ /4) exp(−eΦ∞ /Te ) 8Te /πme . Thus, the net electrical current density to
DRAFT 10:26
the wall will be given by
J = Ji − Je = − e(ni V∞ − ne v̄e )
hp p i
= − en∞ Te /mi − (1/4) 8Te /πme exp (−eΦ∞ /Te ) . (1.22)
Since in a quasineutral plasma equilibrium we must have J = 0, the plasma

potential Φ∞ is given in this plasma model by
r
Te mi Te Te
Φ∞ = ln ≥ 2.84 ' 3 , plasma potential, (1.23)
e 2πme e e
where after the inequality we have used the proton to electron mass ratio mi /me
= 1836. In the original work in this area in 1949, Bohm argued that a potential
drop of Te /2e extending over a long distance into the plasma (much further
than where
p we are calculating) is required to produce the incoming ion speed
V∞ ≥ Te /mi at the sheath edge. In Bohm’s model the density n∞ is e−1/2 =
0.61 times smaller than the bulk plasma ion density and thus the ion current
Ji is smaller by this same factor. For this model, the potential Φ∞ in (1.23)
increases by 0.5 Te /e to 3.34 Te /e. Since lots additional physics (see end of
preceding paragraph) needs to be included to precisely determine the plasma
potential for a particular situation, and the plasma potential does not change
too much with these effects, for simplicity we will take the plasma potential Φ∞
to be approximately 3 Te /e.
Finally, we investigate the form of Φ(x) in the sheath region near the wall
(x << xS ). In this region the potential is much less than Φ∞ and the electron
density becomes so small relative to the ion density that it can be neglected. The
equation governing the potential in this ion-rich region can thus be simplified
from (1.19) to
· ¸1/2
d2 Φ(x) en∞ mi V∞2
/2e
'− 2 /2e − Φ(x)
. (1.24)
dx2 ²0 Φ∞ + mi V∞
This equation can be de-dimensionalized by multiplying through by e/Te . Thus,

defining a dimensionless potential variable χ by
£ ¤
e Φ∞ + mi V∞ 2
/2e − Φ(x)
χ(x) ≡ , (1.25)
Te
the equation can be written as
d2 χ 1
' 2√ .
dx2 δ χ
in which δ ≡ λDe /(mi V∞2

/2Te )1/4 .
To integrate this equation we multiply by dχ/dx and integrate over x us-
√ √
ing (dχ/dx)(d2 χ/dx2 ) = (1/2)(d/dx)(dχ/dx)2 and dx(dχ/dx)/ χ = dχ/ χ =
√
2d χ to obtain
DRAFT 10:26
µ ¶2
1 dχ 2√
' χ + constant.
2 dx δ2
Since both χ and dχ/dx are small near xS , this equation is approximately valid
for the x < xS region if we take the constant in it to be zero. Solving the
resultant equation for dχ/dx, we obtain
dχ 2χ1/4 4 2dx
'− =⇒ d(χ3/4 ) ' − .
dx δ 3 δ
Integrating this equation from x = 0 where χ = χ0 ≡ (eΦ∞ + mi V∞
2
/2)/Te to
x where χ = χ(x), we obtain
3/4 3x
χ3/4 (x) − χ0 '− ,
2δ
or
Φ(x) ' (Φ∞ + mi V∞
2
/2e)[1 − (1 − x/xS )4/3 ]. (1.26)
Here, we have defined

3/4 µ ¶1/4 µ ¶3/4
2δχ0 25/4 Te 2
eΦ∞ + mi V∞ /2
xS ≡ = 2
λDe ,
3 3 mi V ∞ Te
sheath thickness. (1.27)
Equation (1.26) is valid in the sheath region near the wall (0 < x << xS ). We
have identified the scale length in (1.27) with the sheath width xS because this
is the distance from the wall at which the potential Φ(x) extrapolates to the
2
effective plasma potential in the bulk plasma, Φ∞ + mi V∞ p /2e.
Using the value for Φ∞ given in (1.23) and V∞ ' Te /mi , the sheath
thickness becomes xS ' 2 λDe . Thus, as shown in Fig. 1.3, for this model the
plasma charges to a positive potential of a few Te /e and is quasineutral up
to the non-neutral plasma sheath region, which extends a few Debye lengths
(∼ 2 xS ∼ 4 λDe in Fig. 1.3) from the grounded wall into the plasma region.
1.3 Langmuir Probe Characteristics*

To further illuminate the electrical properties of a static or equilibrium plasma,
we next determine the current that will be drawn out of a probe inserted into an
infinite plasma and biased to a voltage or potential ΦB . Such probes provided
some of the earliest means of diagnosing plasmas and are called Langmuir probes,
after Irving Langmuir who developed much of the original understanding of
their operation. The specific situation to be considered is sketched in Fig. 1.4.
For simplicity we assume that the probe is small compared to the size of the
plasma and does not significantly disturb it. The probe will be assumed to
have a metallic (e.g., molybdenum) tip and be electrically connected to the
outside world via an insulated tube through the plasma. Probes of this type are
DRAFT 10:26
I
ISe
plasma
φI φ φB
φB ISi p
Figure 1.4: Schematic of Langmuir probe inserted into a plasma and its idealized
current-voltage characteristics: current I drawn out of the probe as a function
of the bias voltage or potential ΦB . The labeled potentials and currents are:
Φf , floating potential; Φp , plasma potential; ISi , ion saturation current; ISe ,
electron saturation current.
often used in laboratory plasmas that have modest parameters (Te ∼ < 10 eV,
< 19 −3
ne ∼ 10 m — probes tend to get burned up at higher plasma parameters).
Since the bias potential ΦB on the probe will not affect the incoming ion
flow speed V∞ (for ΦB < Φ∞ ), following the discussion leading to (1.22) the ion
current out of the probe will be given by
p
Ii = AS Ji = −n∞ e Te /mi AS ≡ −ISi , ion saturation current (ISi ), (1.28)
where AS is the area of the probe plus sheath over which the ions are collected
by the probe. For the electrons we must take account of the bias potential ΦB
on the probe. The electron current into the probe is given by
 p
 n∞ e Te /2πme Ap ≡ ISe , ΦB ≥ Φp ,
Ie = Ap Je =
 n epT /2πm A exp [ − e(Φ − Φ )/T ] , Φ < Φ ,
∞ e e p p B e B p
(1.29)
in which Ap is the area of the probe and Φp is the plasma potential — the voltage
at which all electrons heading toward the probe are collected by it. (Whereas
the effective area for ions to be collected by the probe encompasses both the
probe and the sheath, for ΦB < Φp the relevant area Ap for electrons is just the
probe area since only those electrons surmounting the sheath potential make
it to the probe — see Fig. 1.3. However, when ΦB > Φp the relevant area,
and consequent electron current, grows slightly and roughly linearly with bias
DRAFT 10:26
voltage, which is then attracting electrons and modifying their trajectories in

the vicinity of the probe. In the idealized current-voltage curve in Fig. 1.4 we
have neglected this latter effect.)
The total current I = Ii + Ie drawn to the probe is shown in Fig. 1.4
as a function of the bias voltage or potential ΦB . For a large negative bias
the electron current becomes negligible and the current is totally given by the
ion current ISi , which is called the ion saturation current. The potential Φf
at which the current from the probe vanishes is called the floating potential ,
which is zero for our simple model. However, it is often slightly negative in real
plasmas, unless there is secondary electron emission from the probe, in which
case it can become positive. For potentials larger than the plasma potential
Φp all electrons on trajectories that intercept the probe are collected and the
current is given by the electron saturation current ISe . Except for differences
in the charged particle collection geometry (typically cylindrical or spherical
probes versus a planar wall), in the sheath thickness (relative to probe size)
effects and perhaps in secondary electron emission, the difference between the
plasma and floating potentials is just the naturally positive plasma potential
that we derived in (1.23). That is, Φp − Φf ' Φ∞ ∼ 3 Te /e.
In a real plasma the idealized current-voltage characteristic that is indicated
in Fig. 1.4 gets rounded off and distorted somewhat due to effects such as charged
particle orbit effects in the sheath, probe geometry, secondary electron emission
from the probe and other effects. Indeed, because of the practical importance of
Langmuir probes in measuring plasma parameters in many laboratory plasmas,
as we will discuss in the next paragraph, there is a large literature on the
current-voltage characteristics of various types of probes in real plasmas (see
references and suggested reading at the end of the chapter). Nonetheless, the
basic characteristics are as indicated in Fig. 1.4.
For bias potentials that lie between the floating and plasma potentials, the
current from the probe increases exponentially with bias potential ΦB . Thus,
the electron temperature can be deduced from the rate of exponential growth in
the current as the bias potential is increased: Te /e '(I − ISi )/(dI/dΦB ). Alter-
natively, one can use a “double probe” to determine the electron temperature —
see Problem 1.11. If the electron temperature is known, the plasma ion p density
can be estimated from the ion saturation current: n∞ ' ISi /(eAS Te /Mi ).
Langmuir probes are thus important diagnostics for measuring the plasma den-
sity and electron temperature in laboratory plasmas with modest parameters.
The thickness of the plasma sheath changes as the bias potential ΦB is
varied. The derivation of the sheath thickness xS given in (1.24) to (1.27) can
be modified to account for the present biased probe situation by replacing the
potential Φ∞ with Φp − ΦB . Thus, setting mi V∞ 2
/Te to unity to satisfy the
Bohm sheath criterion (1.21), the sheath thickness around a biased probe in a
plasma is given approximately by
µ ¶3/4
25/4 Φp + 0.5 − ΦB
xS ' λDe , sheath thickness. (1.30)
3 Te /e
DRAFT 10:26
This formula is valid for e(Φp +0.5−ΦB ) >> Te — small or negative bias voltages
ΦB − Φp . As the bias potential ΦB increases toward the plasma potential Φp ,
the plasma sheath becomes thinner; it disappears for e(Φp − ΦB ) ∼ < 0.5 Te .
For large negative bias potentials (|ΦB | >> Te /e), the electrical current
density flowing through the ion-rich sheath region is limited by “space charge”
effects and given by the Child-Langmuir law — see Problem 1.13. However, tran-
siently the current density can be larger that indicated by the Child-Langmuir
law — see Problem 1.14.
1.4 Inertial Response; Plasma Oscillations

In the preceding sections on Debye shielding and its effects we considered the
adiabatic or static response of charged particles and a plasma to the Coulomb
electric field around a charged particle in the plasma. Next, we discuss the
inertial (or dynamic) response of a plasma. To do this we consider the electric
polarization response of charged particles and a plasma to a small electric field
perturbation, which may be externally imposed or be collectively generated
within the plasma.
First, we calculate the motion of a charged particle in response to an electric
field. The velocity v of a charged particle of mass m and charge q subjected
to an electric field perturbation7 Ẽ(x, t) is governed by Newton’s second law
(F = ma) with force q Ẽ:
dv
m = q Ẽ(x, t). (1.31)
dt
The electric field perturbation will be assumed to be small enough and suffi-
ciently slowly varying in space so that nonlinear and translational motion effects
are negligible. Thus, it will be sufficient to evaluate the electric field at the ini-
tial position x0 and neglect the small variation in the electric field induced by
the motion x(t) of the charged particle. This approximation will be discussed
further after the next paragraph.
Integrating (1.31) over time, the velocity perturbation ṽ(t) induced by the
electric field perturbation for a particle with initial velocity v0 is given by
Z t Z t
q q
ṽ(t) ≡ v(t) − v0 = dt0 Ẽ[x0 (t0 ), t0 ] ' dt0 Ẽ(x0 , t0 ). (1.32)
m 0 m 0
Integrating once more over time, we find that the motion induced by the electric
field perturbation becomes
Z t Z t0
q
x̃(t) ≡ x(t) − (x0 + v0 t) ' dt0 dt00 Ẽ(x0 , t00 ), inertial response.
m 0 0
(1.33)
7 Perturbations to an equilibrium will be indicated throughout the book by a tilde over
the symbol for the perturbed quantity. Equilibrium quantities will be indicated by 0 (zero)
subscripts.
DRAFT 10:26
Because the response of the particle to the electric field force is limited by the
inertial force m a = m dv/dt, this is called an inertial response. Note that this
response is inversely proportional to the mass of the charged particle; thus,
the lighter electrons will give the primary inertial response to an electric field
perturbation in a plasma.
We check our approximation of evaluating the electric field at the initial
position x0 by expanding the electric field in a Taylor series expansion about
the charged particle trajectory given by
Ẽ [x(t), t] = Ẽ (x0 , t) + (x̃ + v0 t) · ∇Ẽ |x0 + · · · (1.34)
Our approximation is valid as long as the second (and higher order) terms in
this expansion are small compared to the first term:
(x̃ + v0 t) · ∇Ẽ << Ẽ. (1.35)
Thus, the electric field perturbation must vary sufficiently slowly in space (i.e.,
the gradient scale length |(1/|Ẽ|)∇Ẽ|−1 must be long compared to the distance
|x̃ + v0 t|), be small enough (so the nonlinear term x̃ · ∇Ẽ is small compared to
Ẽ) and the elapsed time must not be too long. These approximations will be
checked a posteriori — at the end of this section.
The inertial motion x̃ of a charged particle in response to the electric field
perturbation creates an electric dipole moment qx̃. A uniform density n0 of such
charged particles leads to an electric polarization density P̃ = n0 qx̃. Summing
over the species of charged particles in the plasma, the total plasma polarization
density becomes
X X Z t Z t0
P̃ = n0s qs x̃s = ²0 2
ωps dt0 dt00 Ẽ (t00 ) (1.36)
s s 0 0
in which for each charged species s
n0s qs2
2
ωps ≡ , square of species plasma frequency, (1.37)
ms ²0
is the inertial or plasma frequency for a species s, whose physical significance
will be discussed below.
Because the ions are so much more massive than the electrons (the ratio of
the proton to electron mass is 1836), they have much more inertia. Thus, their
plasma frequency is muchp smaller than that for the electrons — for example,
for protons ωpi /ωpe = me /mp ' 1/43 << 1. Since the electrons give the
dominant contribution to the plasma polarization and have the largest plasma
frequency, we have
X
2
ωps 2
= ωpe 2
+ ωpi ' ωpe
2
. (1.38)
s
DRAFT 10:26
Numerically, the electron plasma frequency is given by

s
ne e2 p
ωpe ≡ ' 56 ne (m−3 ) rad/sec, radian plasma frequency, (1.39)
me ²0
or
p
fpe ≡ ωpe /2π ' 9 ne (m−3 ) Hz, plasma frequency. (1.40)
The plasma polarization in (1.36) causes [see(??) and (??) Section A.2] a po-
larization charge density ρ̃pol given by the negative of the divergence of the
polarization P̃:
X Z t Z t0
0
ρ̃pol (Ẽ) = −∇· P̃ = − ²0 2
ωps dt dt00 ∇· Ẽ(t00 ). (1.41)
s 0 0
Now, to calculate the perturbed electric field Ẽ in a plasma we need to use

Gauss’s law, which is given in (1.1). For the charge density ρq we imagine that
there are polarization charge densities due to both the electric field perturbation
we have been considering, and an externally imposed electric field Eext which
satisfies the same conditions as Ẽ — namely condition (1.35). Thus, the relevant
form of Gauss’s law becomes
X Z t Z t0 h i
1 0
∇· Ẽ = ρ̃pol = − 2
ωps dt dt00 ∇ · Ẽ(t00 ) + Eext (t00 ) . (1.42)
²0 s 0 0
This differential and integral equation in space and time, respectively, can be
reduced to a simpler, completely differential form by taking its second partial
derivative with respect to time to yield
" #
∂ 2 Ẽ X 2 ³ ´
∇· + ωps Ẽ + Eext = 0. (1.43)
∂t2 s
Using the approximation in (1.38), we thus find that taking into account the
inertial effects of charged particles (mostly electrons), nontrivial (i.e., nonvan-
ishing) electric field perturbations satisfying condition (1.35) are governed by
the differential equation
∂ 2 Ẽ 2
+ ωpe Ẽ = − ωpe
2
Eext . (1.44)
∂t2
This is a linear, inhomogeneous differential equation of the harmonic oscilla-
tor type with frequency ωpe for the perturbed electric field Ẽ induced by the
externally applied electric field Eext .
The “complementary” (in the current langauge of mathematics) solutions of
the homogeneous part of this equation are of the form
Ẽh = Cc cos ωpe t + Cs sin ωpe t, (1.45)
DRAFT 10:26
plasma
V(t)
Figure 1.5: Schematic of circuit for imposing an oscillating potential Φ(t) =

Φ0 sin ω0 t across a plasma.
where Cc and Cs are arbitrary coefficient vectors to be fixed by the boundary

conditions. These plasma oscillation solutions show that the plasma responds
inertially to electric field perturbations by oscillating at the electron plasma
frequency ωpe . Externally imposed electric fields will induce perturbations in
the plasma that are combinations of the time dependence of the externally
imposed field and the electron plasma oscillations.
In the present simple model plasma oscillations are undamped. Collisions
(electron-neutral or Coulomb) damp them at rates proportional to the relevant
collision frequency ν — see Problem 1.18. Also, as we will discuss in Chapter 8,
kinetic effects will lead to evanescence of these oscillations due to wave-particle
resonance effects — Landau damping.
To illustrate the plasma responses more concretely, we consider the response
of a plasma to an externally imposed sinusoidal electric field. (An alternative
illustration for just plasma oscillations is developed in Problem 1.19 using a
one-dimensional plasma slab model.) As shown in Fig. 1.5, the electric field will
be induced by imposing an oscillating potential Φ(t) = Φ0 sin ω0 t at time t = 0
across plates on opposite sides of a plasma of thickness L (implicitly >> λD )
in the x direction. For simplicity the plasma will be assumed to be infinite in
extent (or at least >> L) in the other two directions so that their effects can
be neglected. Thus, the applied electric field will be given for t > 0 by
Φ0
Eext = êx sin ω0 t ≡ E0 sin ω0 t. (1.46)
L
The particular solution of (1.44) in response to this externally applied electric
DRAFT 10:26
field is
2
ωpe
Ẽp = E0 sin ω0 t. (1.47)
ω02 − ωpe
2
Adding together the homogeneous, particular and externally applied electric

field components (E = Ẽ + Eext = Ẽh + Ẽp + Eext ) of the solution of (1.44), and
subjecting them to the boundary conditions that E(t =£ 0) = 0 ¡and dE/dt ¢¤|t=0
= dEext /dt |t=0 = ω0 E0 , we find Cc = 0 and Cs = − ω0 ωpe / ω02 − ωpe 2
E0 .
Hence, the total electric field E driven by Eext is given for t > 0 by
2
ω0 ωpe ωpe
E(t) = E0 sin ω pe t + E0 sin ω0 t + E0 sin ω0 t
ω0 − ωpe
2 2 ω0 − ωpe
2 2
ω0 ωpe ω02
= − E0 sin ω pe t + E0 sin ω0 t
ω02 − ωpe
2 ω02 − ωpe
2
≡ Êplasma sin ωpe t + Êdriven sin ω0 t. (1.48)
The frequency dependences of the net driven response Êdriven oscillating at

frequency ω0 and of the response Êplasma oscillating at the plasma frequency
ωpe are shown in Fig. 1.6. For ω0 much less than the electron inertial or plasma
frequency ωpe , we find that Êdriven is of order − ω02 /ωpe 2
compared with the
externally applied electric field E0 sin ω0 t, and hence tends to be small. In
this limit the electrons have little inertia (ω0 << ωpe ) and they develop a
strong polarization response that tends to collectively shield out the externally
applied electric field from the bulk of the plasma. In the opposite limit ω02 >>
2
ωpe , the inertia of the electrons prevents them from responding significantly,
their polarization response is small, and the externally imposed electric field
permeates the plasma — in this limit E ' Eext since Ẽ << Eext . The singularity
at ω0 = ωpe indicates that when the driving frequency ω0 coincides with the
natural plasma oscillation frequency ωpe the linear response is unbounded. In
Chapters 7 and 8 we will see that collisions or kinetic effects bound this response
and lead to weak damping effects for ω0 ' ωpe . Nonlinear effects can also lead
to bounds on this response.
The Êplasma response in (1.48), which oscillates at the plasma frequency, is
caused by the electron inertia effects during the initial turn-on of the external
electric field. Note that it vanishes in both the low and high frequency limits —
because for low ω0 the excitation is small for the nearly adiabatic (ω0 << ωpe )
turn-on phase, while for high ω0 the electron inertial response is small during the
very brief (δt ∼ 1/ω0 << 1/ωpe ) turn-on phase. Like the driven response, the
plasma response becomes unbounded in this simple plasma model for ω0 → ωpe .
Finally, we go back and determine the conditions under which the approxi-
mation (1.35) that we made in calculating the plasma polarization induced by
an electric field is valid. Referring to the physical situation shown in Fig. 1.5, we
take the gradient scale length of the electric field perturbation |(1/|Ẽ|)∇Ẽ|−1
to be of order the spacing L between the plates. We first estimate the condition
imposed by the particle streaming indicated by the term v0 t in (1.35). To make
DRAFT 10:26
collective dielectric
shielding medium
E0
0 wpe
w
Figure 1.6: Frequency dependence of the electric field components oscillat-

ing at the driven frequency ω0 (Êdriven , solid lines) and the plasma frequency
ωpe (Êplasma , dashed lines) induced in a plasma by Eext = E0 sin ω0 t, as indi-
cated in Fig. 1.5. The driven frequency response is shielded out for ω0 << ωpe ;
it approaches the imposed electric field for ω0 >> ωpe . The plasma frequency
response is induced by the process of turning on the external electric field; it
becomes small when ω0 is very different from ωpe . The singular behavior for
ω0 → ωpe results from driving the system at the natural oscillation frequency of
the plasma, the plasma frequency; it is limited in more complete plasma models
by collisional, kinetic or nonlinear effects.
this estimate
p we take v0 to be of order the most probable electron thermal speed
vT e ≡ 2Te /me [see (??) in Section A.3]. Also, we estimate t by 1/ω. However,
since the most important plasma effects√occur for ω ∼ ωpe (see Fig. 1.6), we
scale ω to ωpe . Then, since vT e /ωpe = 2 λDe , the particle streaming part of
(1.35) leads, neglecting numerical factors, to the condition
L >> λDe (ωpe /ω) . (1.49)
That is, for ω ∼ ωpe the plasma must be large compared to the electron Debye
length.
For validity of the nonlinear condition x̃ · ∇Ẽ << Ẽ we consider a situation
where Ẽ = (Φ̃/L) sin ωt. Then, again neglecting numerical factors, we find that
to neglect the nonlinearities we must require
µ 2¶
eΦ̃ L2 ω
<< 2 2
. (1.50)
Te λDe ωpe
Since we can anticipate from physical considerations that potential fluctuations

Φ̃ are at most of order some modest factor times the electron temperature in a
plasma, the nonlinear criterion is usually well satisfied as long as the streaming
DRAFT 10:26
criterion in (1.49) is. Hence, our derivation of the plasma polarization is gen-
erally valid for ω ∼ ωpe plasma oscillation phenomena as long as the plasma
under consideration is much larger than the electron Debye length.
We can also use the preceding logic to specify the temporal and spatial
scales on which the inertial response and effects discussed in this section apply
in an infinite, homogeneous plasma — versus the conditions where the adiabatic
response in the first section of this chapter apply. (For a general discussion of
inertial and adiabatic responses — for a harmonic oscillator — see Appendix E.)
For ω ∼ ωpe , as long as the scale length L ∼ δx of interest is long compared to the
electron Debye length λDe , conditions (1.35), (1.49) and (1.50) are all satisfied.
Then, the inertial and electron plasma oscillation effects we have discussed are
relevant since ω >> vT e /δx, which is the inverse of the time required for a
thermal electron to move a distance δx. However, for low frequencies ω << ωpe
such that δx << λDe (ω/ωpe ), or for scale lengths δx << λDe with ω ∼ ωpe , the
inequality conditions become reversed and the approximations we have used
in this section break down. Then, instead of an inertial response, we obtain
an adiabatic response for ω << vT e /δx and the Debye shielding effects we
discussed in the first section of this chapter. Intermediate situations with δx ∼
λDe (ωpe /ω) ∼ vT e /ω must be treated kinetically — see Chapter 8.
1.5 Plasma as a Dielectric Medium

In general, any vector field such as the electric field perturbation Ẽ is com-
posed of both longitudinal (irrotational, ∇· Ẽ 6= 0) and transverse (solenoidal,
∇· Ẽ = 0) parts — see Section D.5 of Appendix D. From the form of (1.43) it
is clear that we have been discussing the longitudinal component of the electric
field perturbation. This component is derivable from a potential, Ẽ = −∇φ̃,
and represents the electrostatic component of the electric field perturbation.
Since we have ∇· Ẽ = −∇2 φ̃ 6= 0, we see from Gauss’s law (1.1) that these elec-
trostatic perturbation components correspond to charge density perturbations
in the plasma. Thus, the electron plasma oscillations we have been discussing
are electrostatic “space charge” oscillations in which the longitudinal component
of the electric field and plasma polarization oscillate out of phase with respect
to each other, i.e., ∂ 2 (∇· Ẽ)/∂t2 = − ωpe
2
∇· Ẽ = − ∂ 2 (∇· P̃)/∂t2 .
The polarizability of the plasma by an electric field perturbation can also be
interpreted by considering the plasma to be a dielectric medium. To illustrate
this viewpoint, we note that in a dielectric medium Gauss’s law becomes [see
(??) in Section A.2]
∇· D = ρfree , (1.51)
where
D ≡ ²E (1.52)
is the displacement vector, ρfree is the charge density of the free charges (i.e.,
those not contributing to the plasma dielectric), and ² is the dielectric constant of
DRAFT 10:26
the medium (² = ²0 for a vacuum). The electric field perturbation Ẽ induces the
polarization charge density given in (1.41) and the polarization P̃. Comparing
(1.51) with (1.41) and (1.42), we deduce that the perturbed displacement vector
D̃ is related to the polarization P̃ through [see (??) and (??)]
D̃ = ²0 Ẽ + P̃ ≡ ²0 (1 + χ̂E ) Ẽ ≡ ²̂ Ẽ, (1.53)
with
P̃ ≡ ²0 χ̂E Ẽ (1.54)
in which χ̂E is the electric susceptibility of the plasma. We have placed hats
over ² and χE to emphasize that these quantities are only defined with respect
to temporally (and later spatially) varying electric fields; that is, unlike regu-
lar dielectric media, their static, homogeneous plasma limits are divergent and
hence do not exist (see below).
For the sinusoidal electric field perturbations of the form Ẽ = Ê sin ωt that
we have been discussing, the polarization density P̃ given by (1.36) becomes
X ωps
2
P̃ = − ²0 Ẽ ≡ ²0 χ̂E Ẽ; (1.55)
s
ω2
hence, we have
X ωps
2 2
ωpe
χ̂E (ω) = − '− (1.56)
s
ω2 ω2
and
Ã ! Ã !
X ωps
2 2
ωpe
²̂I (ω) = ²0 1− ' ²0 1− 2 , inertial dielectric. (1.57)
s
ω2 ω
In obtaining this form of P̃ we have performed the integrals in (1.36) as in-

definite integrals in time and hence neglected the initial conditions — because
in determining dielectric properties of a medium one considers only the time
asymptotic response and neglects the initial transient effects.
The frequency dependence of the inertial dielectric8 ²Î (ω) in (1.57), which
represents the inertial response of a plasma, is shown in Fig. 1.7. The fact
that ²Î (ω) → ²0 for ω >> ωpe shows why the Êdriven component in (1.48)
approaches the externally applied electric field in this “vacuum” limit. Since
²Î (ω) is negative for ω < ωpe , the externally applied electric field is shielded
8 For media such as water the dielectric response function is nearly constant over most rele-
vant frequency ranges, e.g., for visible light. Hence its properties are characterized by a dielec-
tric “constant.” However, in plasmas the dielectric response function often varies significantly
with frequency (and wavenumber k). Thus, in plasmas we will usually try to avoid speaking
of a dielectric “constant;” instead we will just refer to the plasma “dielectric.” However, when
the dielectric response function is evaluated for a particular frequency (and wavenumber k),
we will often call it the dielectric “constant.”
DRAFT 10:26
²0
0
wpe w
Figure 1.7: Frequency dependence of inertial response plasma dielectric.
out of the plasma or “cut off” in this frequency range. The vanishing of ²Î for
ω = ωpe indicates that this is a “normal mode” of oscillation of the plasma, as
is evident from the plasma oscillator equation (1.44) — driven electric fields at
frequencies where the dielectric vanishes lead to unbounded resonant responses
in linear theory, as can be inferred from (1.51) and (1.52). Also, since ²Î is small
for ω close to ωpe , the transient plasma frequency response Êplasma is largest
in this frequency range. Finally, we note that ²Î (ω) is divergent in the ω → 0
or static limit. Thus, the inertial dielectric response of a plasma can only be
defined for temporally varying processes.
Because the polarization of the plasma is 180◦ out of phase with respect
to the electric field perturbations for all real ω, the inertial plasma response is
reactive (i.e., not dissipative) for all frequencies ω. That there is no dissipation
can be demonstrated explicitly by calculating the average Joule heating Ẽ · J̃
with J̃ = n0 eṽ over an oscillation period 2π/ω and showing that it vanishes.
If dissipative effects, such as collisons, are added, they lead to wave damping
through the joule heating they induce in the plasma — see Problem 1.18
The energy density of plasma oscillations is composed of two parts: the
vacuum electric field energy density ²0 |Ẽ|2 /2 and the polarization energy den-
sity wpol = − 12 P̃ · Ẽ = − 12 ²0 χ̂E |Ẽ|2 . For an electric field perturbation Ẽ
oscillating at frequency ω the polarization is given in (1.55). Thus, we find
2
wpol = (²0 /2)(ωpe /ω 2 )|Ẽ|2 . Hence, the total energy density [see (??)] in an
electrostatic plasma oscillation is given by
Ã !
2
1 ²0 2 ²0 ωpe
wE ≡ (Ẽ · D̃) = |Ẽ| + wpol = 1 + 2 |Ẽ|2 , wave energy density.
2 2 2 ω
(1.58)
For low frequencies (ω << ωpe ), for which an externally imposed electric
field is shielded out of the plasma, the polarization energy density is dominant.
DRAFT 10:26
In contrast, for high frequencies (ω >> ωpe ) the electron inertia effects cause
the polarization to be small; then, the energy density is predominantly just that
residing in the electric field perturbation itself. The fact that the energy density
caused by electric field perturbations can have a significant (or even dominant,
as occurs for ω << ωpe ) component due to the polarizability of the plasma is a
very important aspect of plasma oscillations.
1.6 Ion Acoustic Waves

In the preceding sections we have implicitly assumed that the electrons and
ions both exhibit either adiabatic or inertial responses. However, because the
ions are much heavier, they have a much lower inertial or plasma frequency
and, for the typical case where Te ∼ Ti , a much lower thermal speed than
electrons. Thus, for a given length scale δx there is an intermediate frequency
regime vT i /δx << ω << vT e /δx in which the ions respond inertially while the
electrons respond adiabatically. We will now determine the equation governing
electric field perturbations in a plasma in this regime.
The perturbed electron density for an adiabatic (ω << vT e /δx) response
induced
P by a potential perturbation φ̃ of a quasineutral plasma equilibrium
( s n0s qs = 0) is obtained from the perturbed Boltzmann relation (1.4):
eφ̃
ñe = n0e . (1.59)
Te
The perturbed ion density for an inertial (ω >> vT i /δx) response induced by
an electric field perturbation Ẽ is obtained from the ion polarization part of the
total plasma charge density given in (1.41):
Z t Z t0
²0 2 0
ñi = − ωpi dt dt00 ∇· Ẽ(t00 ). (1.60)
qi 0 0
The overall perturbed charge density in this intermediate frequency regime is

thus given by
X ñs qs Z t Z t0
ρ̃q 0 n0e e2
= =− 2
ωpi dt dt00 ∇· Ẽ(t00 ) − φ̃
²0 s
²0 0 0 ²0 Te
Z t Z t0
φ̃
= 2
+ ωpi dt0 dt00 ∇2 φ̃(t00 ) − (1.61)
0 0 λ2De
in which we have specialized to electrostatic perturbations for which Ẽ = −∇φ̃

and ∇· Ẽ = −∇2 φ̃.
Substituting this perturbed charge density into Poisson’s equation (1.1), we
obtain
µ ¶ Z t Z t0
1
− ∇2 − 2 2
φ̃ = ωpi dt0 dt00 ∇2 φ̃(t00 ).
λDe 0 0
DRAFT 10:26
Or, taking the second partial derivative with respect to time, this yields
µ ¶ 2
1 ∂ φ̃
− ∇ − 22 2
= ωpi ∇2 φ̃. (1.62)
λDe ∂t2
Considering perturbations
¡ whose
¢ scale lengths are long compared to the electron
Debye length ∇2 << 1/λ2De , the equation governing potential perturbations
in the intermediate frequency regime becomes simply
∂ 2 φ̃
− c2S ∇2 φ̃ = 0, ion acoustic wave equation, (1.63)
∂t2
in which
Te ni qi2 Zi Te
c2S = ωpi
2
λ2De = = . (1.64)
mi ne qe2 mi
As indicated in the last equality, for a plasma with a single ion component
ni qi2 = Zi ne e2 so that c2S = Zi Te /mi . The quantity cS has the units of a speed
and as we will see below is the speed of ion acoustic (or sound) waves in a
plasma. It is given numerically by
r s
Zi Te Zi Te (eV)
cS ≡ ' 104 m/s, ion acoustic speed, (1.65)
mi Ai
in which Ai is the atomic mass of the ions in the plasma: Ai ≡ mi /mp . Here,
we have used the subscript S on c to indicate that these ion acoustic waves
are the natural “sound” (S) waves that occur in a plasma. The relation of ion
acoustic waves to normal sound waves in a neutral gas are discussed at the last
of this section, and their relation to the sound waves in a magnetohydrodynamic
description of a plasma is discussed in Section 7.2.
Equation (1.63) is a wave equation. In one dimension, say the x direction,
general solutions of it are given by a linear combination of arbitrary functions
f1 , f2 of its mathematical characteristics ϕ± ≡ x ∓ cS t:
φ̃(x, t) = C1 f1 (x − cS t) + C2 f2 (x + cS t),
where C1 and C2 are arbitrary constants to be fixed by the boundary conditions.

A point of constant phase in this solution moves at the phase speed Vϕ of the
wave: dϕ± = 0 = dx ∓ cS dt =⇒ Vϕ = dx/dt = ±cS along the mathematical
characteristics x = x0 ± cS t.
For wave-like equations such as those in (1.62) or (1.63) we usually seek
solutions of the form
φ̃ (x, t) = φ̂ ei(k·x−ωt) (1.66)
in which φ̂ is a constant, k is the (vector) wavenumber and ω is the frequency

of the wave. Substituting this Ansatz (proposed form) into (1.62), we find
£ ¡ 2 ¢¡ ¢ ¤
− −k − 1/λ2De −ω 2 + ωpi 2 2
k φ̂ = 0.
DRAFT 10:26
wpi
Figure 1.8: Dispersion diagram for ion acoustic waves in an electron-proton

plasma with Te >> Ti . For kλDi << 1 the ion acoustic waves propagate at
the ion acoustic speed: ω/k ' cS . The dispersion curve ω = ω(k) is shown as
a dashed line for kλDi ∼ > 1 because in this region the ion response is no longer
inertial (kinetic effects become important) and the present analysis becomes
invalid.
For nontrivial solutions with φ̂ 6= 0, we must have

2
k 2 c2S ωpi
ω2 = 2 = , ion acoustic wave dispersion relation.
1 + k 2 λDe 1 + 1/ (k 2 λ2De )
(1.67)
This is called a dispersion relation because it givess the dependence of ω on k

— here for electrostatic ion acoustic waves propagating in a plasma.
The dispersion diagram (ω versus k) for ion acoustic waves is shown in
Fig. 1.8. For k 2 λ2De << 1 (long scale lengths compared to the electron Debye
length) we have ω/k = ±cS — the phase speed Vϕ ≡ ω/k of the wave is the
ion acoustic speed cS . Since we have assumed that the ions have an inertial
response, taking δx ∼ 1/k we must have vT i /δx ∼ kvT i << ω ' kcs . This
condition is satisfied and ion acoustic
p waves exist in an electron-proton plasma
only if the ionpacoustic speed cS ≡ Te /mi is much larger than the ion thermal
speed vT i = 2Ti /mi , which occurs only if Te >> 2Ti . As can be discerned
from (1.67), the wave frequency ω increases for increasing kλDe and asymptotes
to ωpi for kλDe >> 1. However, to satisfy the required condition for an ion
inertial response we must have kvT i << ω ∼ ωpi or kλDi << 1. We can satisfy
kλDe >> 1 >> kλDi only if Te >> Ti , which is the same as the condition noted
previously in this paragraph for the existence of ion acoustic waves.
As we discussed in the preceeding section, plasma responses can also be
described terms of the plasma giving a dielectric response ²̂. For waves of the
DRAFT 10:26
form given in (1.66) the polarization corresponding to the perturbed charge

density in (1.61) becomes
Ã !
2
ωpi 1
P̃ = ²0 − 2 + 2 2 Ẽ, (1.68)
ω k λDe
in which we have used Ẽ = −ikφ̃ and ∇· Ẽ = −∇2 φ̃ = k 2 φ̃. Using the defini-
tions for the interrelationships between P̃, χ̂E and ²̂ given in (1.53), (1.54), we
find that in the intermediate frequency regime we are considering the plasma
dielectric response is given by
Ã !
2
ωpi 1
²̂S (k, ω) = ²0 1− 2 + 2 2 , ion acoustic dielectric. (1.69)
ω k λDe
Setting this ²̂S to zero to obtain the normal modes of the plasma readily
yields the dispersion relation for ion acoustic waves given in (1.67). This di-
electric function diverges for either ω → 0 or k → 0. Thus, again, the plasma
dielectric is only a meaningful quantity for temporal and spatially varying per-
turbations, i.e., not for an infinite, homogeneous equilibrium. Also, since ²̂S
is real for all real k, ω (i.e.,the electron and ion components of the polariza-
tion are in phase or 180◦ out of phase with the electric field perturbation), this
intermediate frequency response is also totally reactive (i.e., not dissipative).
Ion acoustic waves are similar to but somewhat different from ordinary sound
waves in a neutral gas. Ordinary sound waves are compressible (∇·Ṽ 6= 0
where Ṽ is the perturbed flow velocity) mass density perturbations induced by
momentum perturbations propagated by the collisionally coupled flow of the
neutral gas molecules or atoms in response to pressure perturbations — see
SectionpA.6. They propagate
p at a “hydrodynamic” (H) phase speed given by
cH
S = Γpn /ρm = ΓTn /mn in which Γ = (N + 2)/N is the ratio of the
specific heats, N is the number of degrees of freedom, and pn , ρm , Tn and Mn
are the neutral gas pressure, mass density, temperature and mass, respectively.
In an electron-proton plasma with Te >> Ti , ion acoustic waves propagate via
longitudinal (∇· Ẽ 6= 0) electric field perturbations, which as we will see in
Section 7.2 also lead to compressible flow perturbations ∇·Ṽ 6= 0, in which
the adiabatic electron polarization charge density is balanced by an inertial
ion polarization
p charge density. Ion acoustic waves propagate at a phase speed
cS = Te /mi with the electron temperature coming from the adiabatic electron
Debye shielding and the ion mass coming from the ion inertia. Thus, the physical
mechanism responsible for ion acoustic wave propagation in a plasma is different
from that of sound waves in a neutral gas even though they are both carried by
incompressible flow perturbations — collisions couple the atoms or molecules in
a neutral gas whereas the electric field couples electrons and ions together in a
plasma. The ion acoustic speed in a Te >> Ti plasma does not, like ordinary
sound waves, depend on the ratio of specific heats or dimensionality of the
system — because it is a “one-demensional” electric field perturbation rather
DRAFT 10:26
than the collisionally-coupled flow in a neutral gas that propagates ion acoustic
waves in a plasma.
1.7 Electromagnetic Waves in Plasmas

In the preceding three sections we explored the properties of longitudinal (elec-
trostatic) electric field perturbations in an unmagnetized plasma. In this section
we develop the properties of transverse (solenoidal) electric field perturbations
for which ∇×Ẽ 6= 0 but ∇· Ẽ = 0 — see Sections A.2 and D.5. These types of
perturbations are often referred to as electromagnetic (em) waves in a plasma
and become light waves in the vacuum limit where the plasma effects are negli-
gible.
To investigate electromagnetic waves in a plasma we begin from the two
Maxwell equations that involve time-derivatives [see (??) in Section A.2]:
µ ¶
∂E
∇×B = µ0 J + ²0 , Ampere’s law, (1.70)
∂t
∂B
∇×E = − , Faraday’s law. (1.71)
∂t
We combine these equations by taking the partial time derivative of Ampere’s
law and substitute in ∂B/∂t from Faraday’s law to obtain
∂J 1 ∂2E
−∇× (∇×E) = µ0 + 2 2
∂t c ∂t
in which we have used the fact that µ0 ²0 = 1/c2 , where c is the speed of light
in a vacuum. Since −∇× (∇×E) = ∇2 E − ∇ (∇· E), for transverse electric
fields Et (∇· Et = 0) this can be written as
∂ 2 Et 1 ∂J
− c2 ∇2 Et = − . (1.72)
∂t2 ²0 ∂t
This is a wave equation for the transverse electric field Et . The inhomogeneous
term on the right represents the plasma effects. The general Green function
solution of this equation, including the plasma inertial response effects, is de-
veloped in Problem 1.23.
Because electromagnetic waves in a plasma are relatively fast (high fre-
quency) phenomena, we can anticipate that the plasma response will be in-
ertial. Thus, the current perturbation induced by the effect of an electric field
perturbation Ẽt on the charged particles in a plasma is given by
X
J̃ = n0s qs ṽs (1.73)
s
in which ṽs is the particle velocity perturbation given in (1.32). Taking the
partial derivative of this current with respect to time, we obtain
1 ∂ J̃ X n0s qs2 X
= Ẽt = 2
ωps Ẽt ' ωpe
2
Ẽt . (1.74)
²0 ∂t s
ms ²0 s
DRAFT 10:26
[This result can also be obtained by considering the plasma to be a dielectric

medium with the inertial dielectric given by (1.57) and calculating the time
derivative of the displacement current and subtracting off the vacuum contribu-
tion: (1/²0 )∂ 2 Dt /∂t2 − ∂ 2 Et /∂t2 = −ω 2 (²̂I /²0 − 1)Ẽt = ωpe
2
Ẽt .]
Substituting the resultant inertial plasma response into (1.72), we obtain
∂ 2 Ẽt 2
+ ωpe Ẽt − c2 ∇2 Ẽt = 0. (1.75)
∂t2
This equation is the same as (1.44), which we obtained for electrostatic (or
longitudinal electric field) perturbations, except for the presence of the c2 ∇2 Ẽt
term, which leads to light wave solutions for ωpe 2
→ 0. Thus, (1.75) embodies
a combination of charged particle inertial (plasma frequency) and light wave
effects in a plasma.
To explore the properties of electromagnetic waves in a plasma we consider
wave solutions of the form
Ẽt (x, t) = Êt ei(k·x−ωt) . (1.76)
Substituting this Ansatz (proposed form) for Ẽt into (1.75) yields
¡ 2 ¢
−ω + ωpe2
+ k 2 c2 Êt = 0.
Nontrivial (Êt 6= 0) solutions are possible for electromagnetic waves that satisfy
q
ω 2 = ωpe
2
+ c2 k 2 , or k = ± ω 2 − ωpe
2 / c, em wave dispersion relation.
(1.77)
This dispersion relation is plotted in Fig. 1.9. Since for these waves ω/k is
greater than the speed of light and hence, for a nonrelativistic plasma, the ther-
mal speeds of both the ions and electrons, it was valid for us to use the inertial
response that we utilized in (1.74). In the short wavelength limit (k >> c/ωpe )
the inertial plasma effects become negligible and we have regular light waves
with ω ' ±ck. For longer wavelengths (k ∼ < ωpe /c), but still high enough fre-
quency so that ω > ωpe , the waves have the dispersion characteristics shown
in Fig. 1.9. For ωpe /c >> k, the waves become electromagnetic plasma oscilla-
tions with ω ' ωpe . For ω < ωpe , the wavenumber k becomes imaginary; this
indicates that transverse electric field perturbations are spatially evanescent in
this regime. In the limit ω << ωpe we have k ' ± i ωpe /c.
To make the properties of electromagnetic waves in a plasma more concrete,
we consider the propagation of electromagnetic waves from a vacuum into a
plasma. As shown in Fig. 1.10, we consider a situation in which the infinite
half-space where x > 0 is filled with plasma while the infinite half-space where
x < 0 is a vacuum. A wave of frequency ω is launched from x = −∞ in the +x
direction toward the plasma and is incident (I) on the plasma at x = 0. It will
DRAFT 10:26
wpe
Figure 1.9: Dispersion diagram for electromagnetic waves in a plasma. For

kc << ωpe the waves become electromagnetic plasma oscillations with ω ' ωpe .
For kc >> ωpe the waves become ordinary light waves with ω ' ck.
be taken to be of the form:
incident wave: Ẽt = ÊI eikI ·x−iωt , ÊI = ÊI êy ,

kI = k0 êx , k0 = ω/c (1.78)
in which for simplicity we have assumed that the incident wave has linear po-
larization in the y direction.
In general, part of this wave will be transmitted into the plasma at the
vacuum-plasma interface at x = 0. We take the transmitted (T ) wave to be of
the form
transmitted wave: Ẽt = ÊT eikT ·x−iωt ,

ÊT = ÊT êy ,
q
kT = kT êx , kT = ω 2 − ωpe
2 /c (1.79)
in which the polarization has again been taken to be in the y direction because
the presence of the plasma does not change the wave polarization. In addition,
part of the wave will be reflected; we take the reflected (R) wave to be of the
form
reflected wave: Ẽt = ÊR eikR ·x−iωt , ÊR = ÊR êy ,

kR = −k0 êx , k0 = ω/c. (1.80)
The magnetic field accompanying each of these waves is obtained from Fara-
day’s law (1.71) for wave solutions of the form (1.76): iω B̃ = ik×Ẽt =⇒ B̃z =
êz · (k×êy ) Ẽy /ω = k Ẽy /ω. The boundary conditions at the vacuum-plasma
interface (x = 0) are that the electric field Ẽy and magnetic field B̃z must be
DRAFT 10:26
VacuumÃRegion PlasmaÃRegion
wÃ>>Ãwpe
kI kT
kR
wÃ<<Ãwpe
kI
kR
xÃ=Ã0
Figure 1.10: Propagation of an incident (I) electromagnetic wave from a vacuum

into a plasma. For ω >> ωpe the wave is transmitted (T) into the plasma with
little reflection (R); the wavenumber k is reduced from ω/c in the vacuum to
(ω 2 − ωpe2 1/2
) /c in the plasma. For ω << ωpe the wave is mostly reflected
from the plasma; the part that does penetrate into the plasma is exponentially
evanescent in the electromagnetic skin depth distance c/ωpe
.
DRAFT 10:26
continuous there. They lead to the two conditions
ÊI + ÊR = ÊT ,
³ ´
(k0 /ω) ÊI − ÊR = (kT /ω) ÊT .
Solving these equations for the relative magnitudes of the transmitted and re-
flected waves, we find
ÊT 2k0 2ω
transmitted: = = q ,
ÊI k0 + kT ω + ω 2 − ωpe
2
q
ÊR k0 − kT ω − ω 2 − ωpe
2
reflected: = = q .
ÊI k0 + kT ω + ω 2 − ωpe
2
The properties of the transmitted and reflected electromagnetic waves are

shown in Fig. 1.10 for two extreme limits: ω >> ωpe and ω << ωpe . For very
high frequencies (ω >> ωpe ) the incident electromagnetic is transmitted into the
plasma with very little reflection and only a slight reduction in the wavenumber
k. As theqfrequency of the incident wave is decreased, the wavenumber decreases
to kT ≡ ω 2 − ωpe 2 /c. At the point where ω = ω , the transmitted wave has
pe
kT = 0 and becomes just an electromagnetic plasma q oscillation. For ω < ωpe
the wavenumber becomes imaginary, kT = ±i ωpe
2 − ω 2 /c. The plus sign is
the physically relevant solution since it leads to evanescence (spatial decay not
due to a dissipative process) in space for x > 0. In the limit ω << ωpe the
incident wave is mostly reflected and the small component of the wave that is
transmitted into the plasma is given by
Ẽt ' ÊT exp[−x/(c/ωpe ) − iωt], ω << ωpe . (1.81)
This electric field perturbation is exponentially evanescent in the distance δe

given by
δe ≡ c/ωpe , electromagnetic skin depth. (1.82)
Thus, for ω > ωpe electromagnetic waves are partially reflected at the
vacuum-plasma interface and propagate into plasmas with some reduction in
the wavenumber k. However, for ω < ωpe the plasma (and in particular the
electron) inertial response to an electromagnetic wave causes the wave to be
mostly reflected at the vacuum-plasma interface and prevents
q the wave from
penetrating into a plasma more than a distance of about c/ ωpe 2 − ω 2 , which
becomes just the electromagnetic skin depth c/ωpe in the limit ω << ωpe .
A major diagnostic application of the properties of electromagnetic waves in
a plasma is their use in a microwave interferometer to determine the density of
DRAFT 10:26
referenceÃleg
attenuator
microwave
source
phase
L
detector
plasma
x
0
Figure 1.11: Schematic illustration of a microwave interferometer. The electro-

magnetic wave passing through the plasma has a smaller wavenumber (longer
wavelength) than the wave passing through vacuum in the reference leg. Thus,
there is a phase shift between the two signals arriving at the detector.
a plasma — see Fig. 1.11. The difference in the phase between the wave that
passes through a reference vacuum leg versus the wave that passes through a
leg with a plasma in it is given by
 q 
Z L Z L ω − ω 2 − ωpe
2 (x)
∆ϕ = dx [kI − kT (x)] = dx  .
0 0 c
In the limit ω >> ωpe this becomes simply

Z L 2 Z L
ωpe (x) e2
∆ϕ ' dx = dx ne (x). (1.83)
0 2ωc 2ωme ²0 c 0
Since the square of the electron plasma frequency is proportional to the local
plasma density, the measurement of this phase shift determines the line integral
of the electron density in the plasma. For example, microwave interferometers
with frequencies in the 50 − 200 GHz range are commonly used to measure the
RL
“line-average” density n̄e ≡(1/L) 0 dx ne (x) of plasmas with electron densities
in the 1018 −1020 m−3 range. For some other applications in which the properties
of electromagnetic waves in plasmas are important see Problems 1.27–1.29.
DRAFT 10:26
1.8 Plasma Definition and Responses

Now that we have elucidated the basic length (Debye length λD ) and time
(plasma period 1/ωpe ) scales for collective phenomena in plasmas, we can specify
quantitatively the criteria that must be satisfied for matter to exist in the plasma
state. As we discussed in the introduction to this chapter, a general criterion for
the existence of a plasma is that charged particle interactions be predominantly
collective rather than binary in the medium. For this general criterion to be
satisfied, we must require that it be satisfied in charged-particle interactions, as
well on the relevant length and time scales for collective phenomena:
1. nλ3D >> 1. The number of charged particles within a Debye cube (or
sphere) must be large so that: a) collective interactions dominate over
binary interactions at the mean interparticle separation distance; b) the
energy density embodied in the polarization electric field around a given
charged particle is small compared to a typical particle’s kinetic energy;
and c) the thermal noise level is small — see (1.11), (1.14) and (1.15),
respectively.
2. L >> λD . The spatial extent of a collection of charged particles must
be large compared to the collective interaction scale length for plasmas,
the Debye length λD , so that: a) the collective interactions are dominated
by bulk plasma rather than boundary effects; and b) inertial effects are
determined locally — see (1.27) and Fig. 1.3, and (1.49), respectively.
3. ωpe >> νen . The collective inertial response frequency in a plasma, the
electron plasma frequency ωpe , must be large compared to the electron-
neutral collision frequency νen , so that the fundamental inertial responses,
the electrostatic electron plasma oscillations in (1.45) and the plasma os-
cillation effects on electromagnetic waves in (1.75), are not damped by
dissipative neutral particle collision effects.
While we have derived the basic collective phenomena in an unmagnetized
plasma, the same physical phenomena occur in magnetized plasmas (primar-
ily along the magnetic field direction); hence these criteria for the existence of
the plasma state apply to magnetized plasmas as well.
Among the three criteria for existence of the plasma state, the first one,
the requirement that there are many charged particles in a Debye cube, is the
necessary condition and the most critical. After this fundamental criterion is
satisfied, the second and third criteria are just checks (sufficient conditions)
that the behavior of the medium will be dominated by collective plasma phe-
nomena on the basic plasma length and time scales. The fundamental plasma
parameter, the number of charged particles in a Debye cube, depends on the
plasma temperature and charged-particle density, i.e., nλ3D ∝ T 3/2 /n1/2 . Thus,
as shown in Fig. 1.12, we can exhibit the various types of plasmas that occur in
nature by showing where they lie relative to lines of constant nλ3D in a plot of
electron temperature versus electron plasma density. As shown in this figure,
DRAFT 10:26
PlasmaÃFrequencyÃfpeÃ(Hz)
2 4 6 8 10 12 14 16 18
10 10 10 10 10 10 10 10 10
6
10
=Ã104Ãm 101Ãm 10-2Ãm 10-5Ãm 10-8Ãm
magnetic
10-11Ãm
fusion
inertial
fusion
104
magneticÃconfinement
10
10
experiments
magneto-
earth’s
sphere
experiments
confinement
inertial
TemperatureÃ(eV)
corona
2
solar
10
5
10
0
10
solarÃatmosphere
discharges
interplanetary
glow
generators
MHD
1
flames
partially
sphere
iono-
interstellar
ionized
alkaliÃmetal
plasmas
10-2
0 5 10 15 20 25 30 35
10 10 10 10 10 10 10 10
ElectronÃDensityÃ(m-3)
Figure 1.12: Ranges of electron temperature and density for various types of lab-
oratory and extraterrestial plasmas. Also shown are the characteristic plasma
parameters: electron Debye length λDe (constant along the dashed lines), num-
ber of charged particles in a Debye cube ne λ3De (constant along solid lines), and
electron plasma frequency ωpe (constant along vertical lines). Also indicated is
the electron temperature range below which the medium is not fully ionized,
which is determined from the Saha equation [see Section A.7 and in particular
(??) and (??)].
DRAFT 10:26
the plasma state spans an enormous parameter range — 26 orders of magnitude

in density and 7 orders of magnitude in temperature! P
Almost all plasmas are electrically quasineutral (i.e., ρq = s ns qs ' 0) on
length scales longer than the Debye length λD . (Notable exceptions are the
electron-rich, non-neutral magnetized plasmas — see references listed below.)
On length scales shorter than a Debye length the local charge density and po-
tentials are dominated by the effects of the discrete charged particles. However,
on length scales longer than the Debye length the collective, plasma polariza-
tion effects dominate and the plasma is quasineutral. The use of a quasineutral
approximation for scale lengths longer than a Debye length is often called the
“plasma approximation.”
Slow processes (ω << vT /δx ∼ kvT ) in a plasma are governed by the adia-
batic response, as discussed in Section 1.1. Fast processes (ω >> vT /δx ∼ kvT )
are governed by the inertial response, as discussed in Section 1.4. Because
the electron
p thermal speed is usually much greater than the ion thermal speed
(vT e ∼ mi /me vT i ∼ 43 vT i >> vT i for Te ∼ Ti ), the electrons and ions in
a plasma can respond differently to perturbations — for kvT i << ω << kvT e
electrons respond adiabatically while the ions respond inertially, as discussed in
Section 1.4.
The response of the plasma to electric field perturbations leads to polariza-
tion of the plasma, and hence to a dielectric response for the plasma medium.
The plasma responses in the various frequency regimes can be summarized in
terms of the density and dielectric responses to small (i.e., linearizable) wavelike
perturbations of the form exp (ik · x − iωt) in an infinite, homogeneous electron-
ion plasma as follows:
Adiabatic (A) electrons and ions: ω ¿ kvT i , kvT e ; Debye shielding;

· µ ¶¸
q φ̃ 1 1 1
ñA ' − n0 ; ²̂A (k, ω) ' ²0 1 + 2 + 2 . (1.84)
T k λ2De λDi
Adiabatic electrons, inertial ions: kvT i ¿ ω ¿ kvT e ; ion acoustic waves;
" #
2 2
eφ̃ Zi eñiI ωpi 1 ωpi
ñeA ' n0e , ' − 2 ∇· Ẽ; ²̂S (k, ω) ' ²0 1 + 2 2 − 2 .
Te ²0 ω k λDe ω
(1.85)
Inertial (I) electrons and ions: kvT i , kvT e ¿ ω; plasma oscillations;
" #
2 2 2
qs ñsI ωps ωpe ωpi
' − 2 ∇· Ẽ; ²̂I (k, ω) ' 1 − 2 − 2 . (1.86)
²0 ω ω ω
As can be seen from these various responses, a plasma is an electrically ac-

tive medium with a frequency- and wavenumber-dependent polarizability and
dielectric response function. As discussed before, these responses are only ap-
plicable for spatially and temporally varying perturbations — they all diverge
for ω, k → 0.
DRAFT 10:26
Within the approximations employed in this chapter, all the basic phenom-
ena in plasmas that we have discussed are reactive with no dissipation. Dissi-
pation would be caused by polarization components that are 90◦ out of phase
with the electric field perturbations, which for exp (ik · x − iωt) perturbations
would be indicated by an imaginary part of the dielectric ²̂.
Implicitly, we have been considering the plasma to be “collisionless.” Pre-
suming collisions with neutrals are negligible (ωp >> νn ), there are two types
of effects that lead to evanescence of waves in a plasma — Coulomb collisions,
which will be discussed in Chapter 2, and wave-particle resonance effects (Lan-
dau damping), which will be discussed in Chapter 8. Since the thermal noise
fluctuation energy induced by two-particle
¡ correlations
¢ (or Coulomb collisions)
in a plasma is only a small fraction 1/ nλ3D << 1 of the thermal energy in
a plasma, we can anticipate
¡ that
¢ the average Coulomb collision frequency will
also be small: ν ∼ ωp / nλ3D << ωp . Cumulative small-angle Coulomb ¡ ¢ colli-
sions enhance the Coulomb collision rate ν by a factor of order ln nλ3D — see
Chapter 2 — but do not change the basic conclusion that the Coulomb collision
rate is slow in a plasma as long as nλ3D >> 1. The wave-particle resonance
effects will be largest when ω ∼ kvT , i.e., when pthe wave phase speed ω/k is
of order the most probable thermal speed vT ≡ 2T /m of one of the species
of charged particles in a plasma. Thus, wave-particle resonance effects will lead
to evanescence of waves (Landau damping) for ω/k ∼ vT i or vT e . These wave
phase speeds and corresponding frequencies are between the frequency ranges
we have considered in this chapter and require a kinetic plasma description.
Wave-particle resonance effects and Landau damping are discussed in Chapter
8, and in particular in Section 8.2.
REFERENCES AND SUGGESTED READING

Discussions of plasma sheath and Langmuir probe theory can be found in
D. Bohm in Characteristics of Electrical Discharges in Magnetic Fields, A.
Guthrie and R.K. Wakerling, eds. (1949) [?].
F.F. Chen, “Electrical Probes,” in Plasma Diagnostic Techniques, R.H. Huddle-
stone and S.L. Leonard, eds. (1965), Chapt. 4 [?].
L. Schott and R.L.F. Boyd in Plasma Diagnostics, W. Lochte-Holtgreven, ed.
(1968) [?].
J.D. Swift and M.J.R. Schwar, Electric Probes for Plasma Diagnostics (1971)
[?].
P.M. Chung, L. Talbot and K.J. Touryan, Electric Probes in Stationary and
Flowing Plasmas (1975) [?].
P.C. Stangeby, “The Plasma Sheath,” in Physics of Plasma-Wall Interactions
in Controlled Fusion, D.E. Post and R. Behrisch, eds. (1985), Vol. 131, p. 41
[?].
Hutchinson, Principles of Plasma Diagnostics (1987), Chapt. 3 [?].
N. Hershkowitz, “How Langmuir Probes Work,” in Plasma Diagnostics, O. Au-
ciello and D.L. Flamm, eds., (1990) [?].
DRAFT 10:26
Recent books that discuss the various types of plasmas indicated in Fig. 1.12,
some of which are beyond the scope of this book, include
Magnetically confined plasmas for controlled fusion:
Rose and Clark, Jr., Plasmas and Controlled Fusion (1961) [?].
Miyamoto, Plasma Physics for Nuclear Fusion (1980) [?].
Miyamoto, Fundamentals of Plasma Physics and Controlled Fusion (1997) [?].
Stacey, Fusion Plasma Analysis (1981) [?].
Teller, ed., Fusion (1981), Vol. I, Parts A & B [?].
Nishikawa and Wakatani, Plasma Physics, Basic Theory with Fusion Applica-
tions (1990) [?].
Hazeltine and Meiss, Plasma Confinement (1992) [?].
White, Theory of Tokamak Plasmas (1989) [?].
Goldston and Rutherford, Introduction to Plasma Physics (1995) [?].
Laser-produced plasmas for inertial confinement fusion:

Kruer, The Physics of Laser Plasma Interactions (1988) [?].
Lindl, Inertial Confinement Fusion (1995) [?].
Space plasmas:
Parks, Physics of Space Plasmas, An Introduction (1991) [?].
Gombosi, Physics of the Space Environment (1998) [?].
Cosmic plasmas:
Alfvén and Fälthammer, Cosmical Electrodynamics, Fundamental Principles
(1963) [?].
Parker, Cosmical Magnetic Fields, Their Origin And Their Activity (1979) [?].
Sturrock, Plasma Physics, An introduction to the theory of astrophysical, geo-
physical, and laboratory plasmas (1994) [?].
Choudhuri, The Physics of Fluids and Plasmas, An Introduction for Astrophysi-
cists (1998) [?].
Partially ionized plasmas and plasma processing:

Chapman, Glow Discharge Processes (1980) [?].
Lieberman and Lichtenberg, Principles of Plasma Discharges and Materials Pro-
cessing (1994) [?].
Nonneutral plasmas:
Davidson, Physics of Nonneutral Plasmas (1990) [?].
Marshall, Free Electron Lasers (1985) [?].
DRAFT 10:26
PROBLEMS
−1/3
1.1 Evaluate the scale lengths bclmin , ne and λDe for an electron-proton plasma
in a typical small-scale magnetic confinement experiment (e.g., a university-
based tokamak) with ne = 2 × 1019 m−3 , Te = Ti = 300 eV. Compare them
to the Bohr radius, de Broglie wavelength, and the classical electron radius
re = e2 /({4π²0 }me c2 ). Discuss the physical significance of each of these scale
lengths. Over what length scale will collective effects occur in such a plasma? /
1.2 Calculate the plasma parameter nλ3D for the plasma described in the preceding
problem. Estimate the effective temperature for thermal noise in such a plasma.
Compare this thermal noise temperature to normal room temperature. /
1.3 Consider a hypothetical situation in which all the electrons in a homogeneous
and quasineutral but bounded plasma are displaced a small distance x in the êx
direction. Show that in the bulk of the plasma the electric field is unchanged,
but that in a layer of width x at the plasma edge there is an electric field.
How large a displacement x induces a maximum potential change equal to the
electron temperature in the plasma? Compare this length to the electron Debye
length λDe and discuss why such a comparison is relevant. //
1.4 Determine the one-dimensional potential distribution in a plasma around an
infinite sheet charge with a one-dimensional surface “test” charge density given
by ρq = σt δ(x − xt ). ///
1.5 Show that for a two-dimensional situation of an appropriately modified form of
(1.7) the potential around a line charge in a plasma is given by
φt (x) = (2λt /{4π²0 })K0 (|x − xt |/λD )
in which λt is the line charge density (coulombs/m) for a line charge of infinite
length placed at x = xt and K0 is the modified Bessel function of the second
kind of order zero. ///
1.6 Show that the potential given by (1.10) is the Green function for the adiabatic
(Debye shielding) response to a free charge density ρfree (x) in an infinite, homo-
geneous plasma, and thus that the general potential solution is given by
Z
ρfree (x0 ) exp(−|x − x0 |/λD )
φ(x) = d3x0 .
{4π²0 } |x − x0 |
Discuss the physical scale lengths over which this Green’s function solution is
valid. Compare this result to the corresponding potential induced by a charge
density in vacuum given in (??). ///
1.7 Show that the combination of the charge of a test particle and the polarization
charge density it induces produces a vanishing net charge Q in the plasma. //
1.8 A spherical spacecraft orbiting the earth in a geostationary orbit finds itself
immersed in a plasma that typically has an electron density of about 106 m−3
and temperature of about 100 eV. Sketch the spatial variation of the electric
potential around the spacecraft, indicating the magnitudes of the potential and
spatial scale lengths involved. To what potential does the spacecraft charge up
relative to its surroundings? /
DRAFT 10:26
1.9 A spherical probe 3 mm in diameter is inserted into a fully ionized electron-

proton plasma generated by microwave heating power applied to a hydrogen gas
and has ne = 1015 m−3 , Te = 10 eV, Ti = 1 eV. What is the Debye length in
such a plasma? Is it large or small compared to the probe size? If the probe is
biased to −10 V, how much current would it draw? /*
1.10 It is proposed to put a wire screen into the plasma described in the preceding
problem and bias it so as to exclude plasma from the region behind the screen.
Taking account of sheath effects, how closely spaced must the screen wires be
so as to block a substantial fraction of the plasma? To what potential should
the probe be biased? /*
1.11 The electron temperature in low energy density plasmas can be measured with a
“double probe” — a single assembly with two identical but differentially biased
Langmuir probes that electrically floats and draws no net current from the
plasma. Show that the current flowing between the two probes is given by
I = 2 ISi tanh (e∆Φ/2Te )
in which ∆Φ is the potential difference (voltage) between the two probes. //*
1.12 The “static” electrical admittance Y (inverse of impedance Z) of a sheath is
given by ∂I/∂ΦB . At what bias potential ΦB should this partial derivative
be evaluated? Show that the sheath admittance is given approximately by
Y ' I/(Te /e). Up to what frequency will this estimate be valid? //*
1.13 For large negative “wall” potentials (|ΦW | >> Te /e) applied between two grids
in a planar diode the electrical current is limited by space charge effects. De-
rive the Child-Langmuir law for this limiting current for a grid separation d as
follows. First, using (1.27) show that when the sheath thickness xS → d and
Φ∞ → ΦW , the ion speed V∞ that corresponds to the space-charge-limited ion
flow at the sheath edge can be written in terms of the wall potential ΦW and
the grid separation d. Then, show that the (ion) current density into the sheath
region between the grids is given by
r µ ¶µ 2 ¶
4 2e ²0 |ΦW |3/2 4 λDΦ
J= = (ne eVW )
9 mi d2 9 d2
in which VW ≡ (2e|ΦW |/mi )1/2 is the ion speed at the wall, and λDΦ =
(²0 |ΦW |/ne e)1/2 is an effective Debye length. //*
1.14 In the Plasma Source Ion Implantation (PSII) technique [J.R. Conrad, J.L.
Radtke, R.A. Dodd, F.J. Worzola, N.C. Tran, J. Appl. Phys. 62, 4591 (1987)],
the target to be bombarded is inserted into a plasma with parameters ne ∼
109 cm−3 and Te ∼ 2 eV, and a natural sheath is allowed to form around it.
Then, the target is rapidly biased to a very large (≥ 30 kV) negative potential
ΦB . This expels the lighter electrons from the region around the object, which
in turn causes an “ion matrix” to be formed there. On what time scales are
the electrons expelled, and the new sheath formed? What is the approximate
maximum energy and current density of the ions bombarding the target before
the new sheath forms? Compare this current density to that given by the Child-
Langmuir law discussed in the preceding problem. Finally, estimate the fluence
(ions/cm2 ) per pulse and the number of pulses required to inject an atomic
monolayer of ions in the target. ///*
DRAFT 10:26
1.15 Consider an impure laboratory plasma composed of a number of different types

of ions: protons with np = 3 ×1019 m−3 , fully ionized carbon ions with 10% of
the proton density and iron ions that are 23 times ionized (Lithium-like charge
states) with 1% of the proton density. What is the electron density and the
overall plasma frequency in this plasma? Also, what is the dielectric “constant”
for 90 GHz electrostatic fluctuations in this plasma? /
1.16 Consider electrostatic plasma oscillations in an electron-positron plasma, such
as could occur in interstellar space, with ne− = ne+ = 106 m−3 . What is the
plasma frequency for such oscillations? Assume the electrons and positrons have
temperatures of 100 eV and that the cross-section for an annihilation interaction
between them is given by π times the square of the classical electron radius
re = e2 /({4π²0 }me c2 ), i.e., σan ∼ π re2 . What is their annihilation reaction
rate? Compare this annihilation rate to the plasma frequency. /
1.17 Consider a situation where an oscillating potential is applied across two plates
on either side of a plasma such as that described in Problem 1.9. Assume the
plates are separated by 10 cm and that the potential oscillates at 100 kHz.
What is the dielectric “constant” for these oscillations? What is the ratio of
the energy density in the plasma polarization fluctuations to that in the electric
field fluctuations? /
1.18 Calculate the weak dissipation induced by Coulomb collisions (ν << ωpe ) of ω ∼
ωpe electrostatic oscillations in a plasma as follows. Add a collisional dynamical
friction force − ν me v [cf., (??)] to (1.31) and show that for Ẽ = Ê sin ωt the
perturbed velocity of electrons is then given for t >> 1/ν by
eÊ
v' [cos ωt − (ν/ω) sin ωt].
ωme
Next, calculate the average of the Joule heating in the plasma by the oscillations
over an oscillation period 2π/ω, i.e., hJ̃ · Ẽiωt . Finally, use a wave energy balance
equation [cf., (??)] to show that
1 ∂hwE iωt hJ̃ · Ẽiωt 2ν

=− '−
hwE iωt ∂t hwE iωt 1 + ω 2 /ωpe
2
in which hwE iωt is the average of the electrostatic wave energy density in the
plasma over an oscillation period. ///
1.19 Consider a hypothetical situation in which all the electrons in a thin slab are
displaced a small distance x0 in the êx direction. Show that the electric field
induced by this displacement is given by E = (n0 e/²0 ) x êx in the region where
the electrons are displaced. Then, show from Newton’s second law that this
force causes the position of the slab of displaced electrons to oscillate at the
electron plasma frequency. //
1.20 Taking account of plasma sheath effects, sketch the spatial variation of the
potential Φ(x) between the plates of a capacitor filled with plasma assuming
the capacitor has a potential Φ ∼ 3 Te /e applied across it. Next, consider a
case where an oscillating potential Φ = Φ0 sin ωt is applied across the plasma
capacitor with Φ0 = 10 Te /e and ω = ωpe /10. If the capacitor plate separation
is L (>> λD ), how large is the electric field component oscillating at frequency
ω in the body of the plasma? //
DRAFT 10:26
1.21 Show that for a one-dimensional wave perturbation in a plasma with Ẽ(x, t) =
Ê êx sin(kx − ωt) the nonlinear terms in (1.31) are negligible in (1.32) when the
wave-induced velocity “jitter” in the particle motion, ṽjitter ≡ q Ê/mω, is small
compared to the wave phase speed ω/k, or alternatively when kx̃jitter << 1. //
1.22 Consider the propagation of ion acoustic waves in a typical hollow cathode arc
discharge composed of electrons and doubly charged Argon ions with ne = 1019
m−3 , Te = 10 eV, Ti = 1 eV. Discuss why the conditions for propagation of
ion acoustic waves are satisfied in this plasma. What is the ion acoustic speed
in this plasma? Compare it to the speed of sound in air at the earth’s surface.
With what wavelength and phase speed will externally imposed waves with a
frequency of 100 MHz propagate in this plasma? /
1.23 Show that the transverse electric field Ẽt induced by a small free current Jfree
in a plasma is governed by the equation
µ 2 ¶
ωpe 1 ∂ 2 Ẽt ∂Jfree
∇2 − 2 Ẽt − 2 = µ0 .
c c ∂t2 ∂t
Then, show that a Green’s function solution of this equation in an infinite,
homogeneous plasma which satisfies this equation is
Z · µ ¶¸
µ0 ∂Jfree (x0 , t0 )/∂t0 |x − x0 |
Ẽt = − d3x0 exp −
4π |x − x0 | c/ωpe ret
in which the square bracket [ ]ret means that the time t0 is to be evaluated at
the retarded time t0 = t − |x − x0 |/c. ///
1.24 Use the solution in the preceding problem to calculate the transverse electric
field Ẽt caused by the current qt vδ[x − x(t)] produced by a nonrelativistic test
particle moving along the trajectory x = x(t) in a plasma. Show that this
transverse electric field points in the direction of test particle motion. Also,
show that for |x − x(t)| < λD its magnitude is of order v 2 /c2 << 1 compared
to the longitudinal electric field produced by the electrostatic potential φt in
(1.10). ///
1.25 Plot the wave dispersion diagrams for electrostatic ion acoustic waves and elec-
tromagnetic plasma waves (i.e., Figs. 1.8 and 1.9) in the plasma described in
Problem 1.22 on a single ω versus k diagram with approximately linear scales.
Indicate in which regions of this diagram adiabatic and inertial responses for
the electrons and ions are applicable. /
1.26 A 140 GHz microwave interferometer set up across a 30 cm thick column of
plasma measures a phase shift of 240◦ . What is the “line-average” plasma
density in the column? /
1.27 Amateur radio operators routinely communicate via shortwave radio over long
distances around the earth. Since communication by direct line of sight is not
possible because of the curvature of the earth’s surface, the waves must be re-
flected from the ionosphere above the earth’s surface. What frequency range
corresponds to the 10 to 40 meter free space wavelength range used by ama-
teur radio operators for these communications? What is the minimum electron
density and height of the ionosphere above the earth’s surface for single-bounce
communications over the approximately 6000 km from the United States to
Western Europe? /
DRAFT 10:26
1.28 During reentry of satellites into the earth’s upper atmosphere, microwave com-
munications in the 300 MHz frequency range are “blacked out” by the plasma
formed in the heated air around the satellite. How high must the plasma den-
sity be around the satellite and how thick must the plasma be to cause the
communications blackout? /
1.29 In one type of inertial fusion experiment, intense light from a laser is shined on
a frozen hydrogen pellet. As the laser light is absorbed it heats up the pellet
and produces a plasma on its surface. Light from a Neodynium glass laser (λ =
1.06 µm) is ultimately observed to be reflected from the pellet. How high must
the density of “free” electrons be in the plasma around the pellet? Compare
this density to the original solid density of the pellet. How thick must the layer
of free electrons be to reflect (or refract) the light waves? Compare this length
to a typical pellet radius of 3 mm. /
1.30 In plasma processing of materials for the semiconductor industry an inert, low
pressure gas is partially ionized by radiofrequency waves in a vacuum chamber.
Consider a case where the initial gas is Argon at a pressure of 10−4 mm Hg (a
760 mm column of mercury corresponds to atmospheric pressure), the electron
density is 107 cm−3 , the electron temperature is 3 eV and the temperature of the
singly charged Argon ions is 0.1 eV. What is the degree of ionization in this gas?
Estimate the electron-neutral collision frequency νen using an electron-neutral
cross-section of 103 πa20 where a0 is the Bohr radius. Does this medium satisfy
all the criteria for being a plasma? How large must it be to satisfy the length
criterion? /
1.31 An oscillating potential of 3 volts at a frequency of 1 MHz is applied to a probe
inserted into the plasma described in Problem 1.9. Over what distance ranges
from the probe can adiabatic or inertial responses be used for the electrons and
for the ions in this plasma? /
1.32 What is the dielectric “constant” for externally imposed waves with a frequency
of 1 MHz and a wavelength of 5 cm in the plasma described in problem 1.9?
What would the dielectric “constant” be if the wavelength was increased to 500
cm? /
DRAFT 10:26

Chapter 1. Collective Plasma Phenomena

Uploaded by

Copyright:

Available Formats

Chapter 1. Collective Plasma Phenomena

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chapter 1. Collective Plasma Phenomena

Uploaded by

Copyright:

Available Formats

CHAPTER 1.

COLLECTIVE PLASMA PHENOMENA 1

The properties of a medium are determined by the microscopic processes in it.

1.1 Adiabatic Response; Debye Shielding

charges in a vacuum. Thus, the electrostatic potential in the plasma can be

∇· E = −∇2 φ = ρq /²0 , Poisson equation, (1.1)

ρq (x) = qt δ(x − xt ) + ρpol (x) (1.2)

in which ρpol is the polarization charge density.

n(x) = n0 e−qφ(x)/T , Boltzmann relation (adiabatic response), (1.3)

where n0 is the average or equilibrium density of these charged particles in the

plasma is electrically quasineutral:

1 X 1 X n0s q 2 1 1 n0e e2 n0i Zi2 e2

In the last expressions we have assumed a plasma composed of electrons with

The general solution of (1.7) in an infinite, homogeneous three-dimensional

qt e−|x−xt |/λD qt e−r/λD

Figure 1.1: Potential φt around a test particle of charge qt in a plasma and

where the eφ/T << 1 approximation breaks down.

ne λ3D >> 1, necessary condition for the plasma state. (1.12)

in which r ≡ |x − xt | and êr ≡ ∇r = (x − xt )/|x − xt | is a unit vector in the

Here, the dimensionless integral I is simplified using x ≡ r/λD and is given by

1.2 Boundary Conditions; Plasma Sheath

While numerical solutions of this equation can be obtained, no analytic solution

Keeping only linear terms in Φ∞ − Φ(x), (1.19) can thus be simplified to

the wall will be given by

Since in a quasineutral plasma equilibrium we must have J = 0, the plasma

This equation can be de-dimensionalized by multiplying through by e/Te . Thus,

in which δ ≡ λDe /(mi V∞2

Here, we have defined

1.3 Langmuir Probe Characteristics*

voltage, which is then attracting electrons and modifying their trajectories in

1.4 Inertial Response; Plasma Oscillations

Ẽ [x(t), t] = Ẽ (x0 , t) + (x̃ + v0 t) · ∇Ẽ |x0 + · · · (1.34)

(x̃ + v0 t) · ∇Ẽ << Ẽ. (1.35)

in which for each charged species s

Numerically, the electron plasma frequency is given by

Now, to calculate the perturbed electric field Ẽ in a plasma we need to use

Ẽh = Cc cos ωpe t + Cs sin ωpe t, (1.45)

Figure 1.5: Schematic of circuit for imposing an oscillating potential Φ(t) =

where Cc and Cs are arbitrary coefficient vectors to be fixed by the boundary

Adding together the homogeneous, particular and externally applied electric

≡ Êplasma sin ωpe t + Êdriven sin ω0 t. (1.48)

The frequency dependences of the net driven response Êdriven oscillating at

Figure 1.6: Frequency dependence of the electric field components oscillat-

L >> λDe (ωpe /ω) . (1.49)

Since we can anticipate from physical considerations that potential fluctuations

1.5 Plasma as a Dielectric Medium

D̃ = ²0 Ẽ + P̃ ≡ ²0 (1 + χ̂E ) Ẽ ≡ ²̂ Ẽ, (1.53)

In obtaining this form of P̃ we have performed the integrals in (1.36) as in-

Figure 1.7: Frequency dependence of inertial response plasma dielectric.

1.6 Ion Acoustic Waves

The overall perturbed charge density in this intermediate frequency regime is

in which we have specialized to electrostatic perturbations for which Ẽ = −∇φ̃

where C1 and C2 are arbitrary constants to be fixed by the boundary conditions.

in which φ̂ is a constant, k is the (vector) wavenumber and ω is the frequency

Figure 1.8: Dispersion diagram for ion acoustic waves in an electron-proton

For nontrivial solutions with φ̂ 6= 0, we must have

This is called a dispersion relation because it givess the dependence of ω on k

form given in (1.66) the polarization corresponding to the perturbed charge

1.7 Electromagnetic Waves in Plasmas