f

ns
• WILLL!\l\il L. KRUER
LAWRENCE LIVER MORE NATIONAL LABORATORY

o~::..''':<C'\~

" 0
"
o «,
of- PR00 ".

ADDISON-WESLEY ,PUBLISHING COMPANY

THE ADVANCED BOOK PROGRAM

Redwood City, California· Menlo Park, California

Reading, Massachusetts· New York > Amsterdam
Don Mills, Ontario· Sydney· Bonn • Madrid
Singapore· Tokyo· Boqota » Santiago· San Juan
Wokingham, United Kingdom
 xiv

.
. f description are then used to describe laser plasma interactions.
lcye"1 0 .'
s 3, 4 , and 5 treat the linear theory
.
of light wave propagation
Contents
Cliapcer . .
. lasmas, including linear mode conversion into plasma waves and col-
:Si~ual damping. The excitation of a variety of plasma instabilities by
intense light waves is then treated in Chapters 6, 7, and 8. In Chapters
9, 10, and ll, important nonlinear co~equences of the various pr~cesses
are discussed using both simple theoretical models and computer simula-
tions. The physics of electron heat transport in laser-produced plasma is
discussed in Chapter 12. Finally, some experimental observations of the
various laser plasma processes are discussed in Chapter 13.
This manuscript is based on lectures given in a graduate course in the
Department of Applied Science of the University of California, Davis, A
detailed review of laser plasma interactions is beyond the scope of this
book. However, a broad cross-section of references to the literature is g
.~
giveu, particularly in those areas of very active research. Lastly, I do not
.'
consider either implosion physics or the very rich topic of electromagnetic
wavesin magnetized plasmas,
I am grateful to numerous colleagues with whom I have worked, and I. Basic Concepts and Two-Fluid Description of P'lasmas
especially to present and former members of the plasma physics group in
1.1 Basic Plasma Concepts 2
tpe laser fusion program at the Lawrence Livermore National Laboratory:
J. Albritton, J. Denavit, K. Estabrook, R. Faehl, D. Hewett, A. B. Lang- 1.2 The Vlasov Equation 5
,~, dlo11,·B.Lasinski, W. Mead, C. Max, C. Randall, J. Thomson, E. Valeo and 6
1.3 The Moment Equations
E. Williams. I also thank many others for helpful comments on portions
of.the manuscript, including J. M. Dawson, E. M. Campbell, H. Baldis, 1.4 The Two-Fluid Description of Plasma 10
R. P. Drake, T. L. Crystal, C. S. Liu, R. Turner, D. Phillion, M. Rosen, 1.5 Plasma Waves II
J. DeGroot, J. Delettrez, L. Goldman, T. Tajima, and R. Lehmberg. I
1.6 Debye Shielding 14
acknowledge the encouragement of J. Nuckolls and J. Lindl. S. Auguadro
typed the original lecture notes. T. L. Crystal very ably produced the final
manuscript. He and A. Wylde provided just the right amount of support
and pressure to finish. I am grateful to the Lawrence Livermore National
2. Computer Simulation of Plasmas
Laboratory, and particularly to B. Quick and P. Brown for a variety of U sing Particle Codes
assistance. Finally, I warmly thank my family for generously allowing me 2.1 Basic Ingredients of a Particle Code 19
to devote many evenings and weekends to this manuscript. 2.2 A 1-D Electrostatic Particle Code 21

xv
 xvii

.•..
7. Stimulated Raman Scattering
Electromagnetic Wave Propagation in Plasmas 74
3. 7.1 Instability Analysis
27
3.1 Wave Equation for Light Waves in a Plasma 7.2 Dispersion Relation 77

3.2 WKB Solution for Wave Propagation 30 7.3 Instability Thresholds 79

in an Inhomogeneous Plasma
7.4 The 2wpe Instability 81
3.3 Analytic Solution for Plasma with 32
a Constant Density Gradient
8. Stimulated Brillouin Scattering
8.1 Instability Analysis 88
4. Propagation of Obliquely Incident Light Waves
in Inhomogeneous Plasmas 8.2 Dispersion Relation 90
38
4.1 Obliquely Incident S-polarized Light Waves 8.3 Instability Thresholds 91

4.2 Obliquely Incident P-polarized Light Waves 39 8.4 The Filamentation Instability 93
_ Resonance Absorption

9. Heating by Plasma Waves

5. Collisional Absorption of Electromagnetic 9.1 Collisional Damping 96
Waves in Plasmas
. 46 9.2 Landau Damping 96
5.1 Collisional Damping of Light Waves
9.3 Linear Theory Limitations - Trapping 100
5.2 Collisional Damping of a Light Wave 48
in an Inhomogeneous Plasma 9.4 Wavebreaking of Electron Plasma Waves 101
Collisional Absorption Including Oblique Incidence 9.5 Electron Heating by the Oscillating- Two-Stream
5.3 51
and a Density Dependent Collision Frequency and Ion Acoustic Decay Instabilities 104
52 108
5.4 Derivation of the Damping Coefficient 9.6 Plasma Wave Collapse

6. Parametric Excitation of Electron and Ion Waves 10. Density Profile Modification
58
6.1 Coupling via Ion Density Fluctuations 10.1 Freely Expanding Plasma 116
60
6.2 The Ponderomotive Force 10.2 Steepening of the Density Profile 117
61
10.3 Resonance Absorption with Density Profile Modification 121
6.3 Instabilities - A Physical Picture
62
6.4 Instability Analysis
66
6.5 Dispersion Relation
69
Instability Threshold due to Spa.tial Inhomogeneity
6.6
70
6.7 Effect of Incoherence in the Pump Wa.ve
-

XVll1

11. Nonlinear Features of Under dense Plasma Instabilities

11.1 Nonlinear features of Brillouin Scattering 127
11.2 Nonlinear Features of Raman Scattering
11.3 Nonlinear Features of the Two-Plasmon Decay
132
The Physics of
and Filamentation Instabilities 135
Laser Plasma
12. Electron Energy Transport
12.1 Electron Thermal Conductivity 144
Interactions
12.2 Multigroup Flux-Limited Diffusion 146
12.3 Other lnftuences on Electron Heat Transport 147
12.4 Heat Transport in Laser-Irradiated Targets 149

13. Laser Plasma Experiments

13.1 Density Profile Steepening 155
13.2 Absorption of Intense, Short Pulse-Length Light 156
13.3 Heated Electron Temperatures 158
13.4 Brillouin Scattering 160
13.5 Raman Scattering 162
13.6 Other Plasma Processes 167
13.7 Wavelength Scaling of Laser Plasma Coupling 168
 CHAPTER

Basic Concepts and

Two-Fluid Description
of Plasmas

The study of the interaction of intense laser light with plasmas serves
as an excellent introduction to the field of plasma physics. Both the lin-
ear and nonlinear theory of plasma waves, instabilities and wave-parti.cle
interactions are important for understanding the laser plasma coupling.
Indeed, the field is a veritable testing ground for many fundamental pro-
cesses. Numerous plasma effects have now been observed in laser plasma
experiments, and many challenging problems remain to be understood.
Since laser plasma interactions are of interest to scientists from many
different fields of expertise, little prior background in plasma physics will
be assumed. Even for those with plasma experience, it can be very in-
structive and refreshing to begin from the basics and examine a field of
a.pplications. Two levels of description will be used - a theoretical one
based on the two-fluid theory of plasmas and a numerical one based on
particle simulation codes. These two descriptions both reinforce and com-
plement one another. For example, the particle simulations allow one to
both test the theory and develop some u.nderst.anding of the nonlinear
effects.

- I
 1 Basic Concepts 1.1 Basic Pfasrne Concepts 3
2

BASIC PLASl\IIA CONCEPTS square velocity. This average rate of change is givenby (t:.v)2 times the
1.1
rate of encounters, which is n., a v. Here ni is the ioo density and a is the
Let's begin. A plasma is basically just asystem of Ncharges whichare cou-
cross-section of impact. Summing over all encounters gives
led to one another via their self-consistent electnc and magnetic fields.
~onsicler then following the evolution of these N charges. ~v~n neglectill~
,,~etic fields and electromagnetic waves, we must ill principle solve 6
ma.o"
coupled equations:
If we substitute for t:.v and integrate over impact parameters, we obtain
mir,

E (ri) 8'Tr 'n.; Z2e4 In A

m2v t,

tb where A is the ratio of the maximum and mmimum impact parame-

Here mi, q, and ri are the mass, charge and position of the i particle,
and E is the electrostatic field. This is clearly an unpromising approach ters (bmax and bmin). The maximum impact parameter is approxi-
if a nontrivial number of charges-is considered. mately the electron Debye length, since other electrons in the plasma
Fortunately a very gTeat simplification is ossible if we focus our a~- shield out the Coulomb potential over this distance. The minimum im-
tCntlOnon co 'sionless plasma behavior. We can decompose the electnc pact parameter is the larger of either the classical distance of closest
field into two fields (E and Ez) which have distinct spatial scales. The approach (bmin ::::::Ze2/mv2) or the DeBroglie wavelength of the electron
field El has spatial variations on a scale lengt much les:>t an the so- (bmin ::::::n/nw), where n is Planck's constant. Using the first, the distance
called electron Debye length, which is the length over which the field. of of closest approach, we have A ::::::9ND/Z, where ND is the number of
an individual charge is shielded out by the response of the surrounding electrons in a Debye sphere. In particular ND = ~'Tr neA6e' where ne is
charges. El represents the rapidly fluctuating microfield due to multipW the electron density and ADe is the electron Debye length. This important
and random encounters ollisions among the discrete charges. In con- length will be derived later in this chapter.
trast, z represents the field due to deviations from char e neutralit over It is convenient to define a ninety-degree deflection time (t90o) by the
space sc es [ITeater an or com Deb e len h. This field condition that the root-mean-square change in velocity becomes as large
'ves rise to "collective" or coher t motion of the char es. as the velocity. Hence
We t us have a natural separation into collisional and collective be-
havior. Not surprisingly, the collisional behavior becomes negligible whe~
the number of electrons in a sphere with a radius ual to the electron
Debye length becomes very large. To motivate this, let us c~ out. a sim-
A veraging over a Maxwellian distribution of velocities then provides us
ple calculation of electron scattering by ions. As illustrated ill FIg. 1.1,
with a convenient measure of the mean rate (1/900 == l/tgoo) at which
Weconsider an electron with velocity v, mass m and charge e streaming
electron-ion collisions scatter electrons through a large angle:
past an ion with charge Ze. The distance of closest approach is b. The
electron undergoes a change in velocity 6v which is approximately
2
6v = Z e (2b) ,
mb2 v
Here Ve JBe/m is the electron thennal velocity and Be is the electron
which is just the rnaxirnurn electrostatic force times the interaction time
("'" 2b/v)'. If we assume many randomly spaced ions, (t:.v).= 0, where
the brackets denote a.n average. However, there is a change ill the mean
 I
I
1.3 The Moment Equations 5
4 1 Basic Concepts

1.2 THE VLASOV EQUATION

The natural starting point for describing the evoilltion of 'I. collisionless
plasma is the Vlasov equation. We first introduce the phase space distri-
bution function fj(x, v , t). This is simply the function which characterizes
_----e -1

the location of the particles of species j in phase space (x. v) as a function

\
b of time. Knowing the laws of motion, we can readily derive an equation for
fj(x, v , t). Since particles are assumed to be neither created nor destroyed

@
~ Ze
as they move from one location in phase space to another (no ionization
Figure 1.1 An electron IS deflected as it streams past an ion.@. t. or recombination), fj(x, v, t) must obey the continuity equation:

temperature. We note that 0

~ J~ ~.~
e&t~; «s: 1- f) fj
et a.
+ Ox . (xfj)
f)
+ 8v . (v fj) = o. (1.2)

Z ln A 1 ..' ',·.'.·"r,,· "'. ~ t'U2:»'

From the laws of motions, we have
z: 'IQ ND'
x v

where Wpe is the electron plasma frequency, which we will see is a fre-
quency characteristic of collective electron motion. (
V
E+ -- x
c
B) ,
(1.3)

The important point we wish to make is now apparent. The fine scale,
collisional interactions can be neglected to zeroth order in the parameter where qj and mj are the charge and mass of the /h species and E and
l/ND. If we express the electron density in cm-3 and the electron tem- B are the coarse-grained fields associated with the collective behavior.
perature in eV, then ND = 1.7 x 109 (();/n,Y/2. ND can be very large Noting that x and v are independent variables and substituting Eq. (1.3)
even in a rather dense plasma, provided the electron temperature is high. into Eq. (1.2), we arrive at the Vlasov equation:
For example, if ne = 1021 cm-3 and ()e = 1 keY, ND ~ 1700. In the
collisionless limit (N D --> 00), the fine scale fluctuating microfields associ-
ated with discrete charges are completely negligible. The plasma behavior
a fj f) fj
-+v·_·+-
at Ox mj
qj (E + v Xc B) o. (1.4)

can then be investigated by solving for the motion of the charges in the
smoothed or coarse-grained fields which arise from the collective motion
of large numbers of charges.
We will develop two parallel levels of description for the collective be-
havior, One level is analytical. Starting from the Vlasov equation, we will
derive moment (fluid-like) equations for the electrons and ions by averag-
ing over the velocities of the charges. This so-called two-fluid description
will then be used extensively to describe a wide variety of laser plasma
interactions. The second level of description is numerical: the use of par-
ticle simulations. These simulations are a powerful tool for investigating
nonlinear effects and kinetic effects (effects which depend on the details
of the velocity distribution of the particles).
This equation simply says that fj( x(t), v(t), t) is a constant; i.e., the
phase space density is conserved following a dynamical trajectory. Such
an equation applies to each charge species in the plasma.
The Vlasov equation, augmented with Maxwell's equations, is a corn-
plete description of collisionless plasma behavior. In practice, we need a
more tractable description which can be obtained by averaging over the
velocities of the individual particles. By taking different velocity moments
of the Vlasov equation, we can derive equations for the evolution in space
and time of the density, mean velocity, and pressure of each species. A5
we will see, each moment brings in the next higher moment, generating an
infinite set of moment equations. However, we can fortunately trunca e
the series of equations by introducing assumptions about the heat flow.

" .
 t·~' ·e·····~·· ---------------.----

1 Basic Concepts 7
6 1.3 The l\IIoment Equations

1.3 THE MOMENT EQUATIONS The first term in Eq. (1.10) is straightforward:
Let us now derive the moment equations and motivate their truncation.
at a
First, we note that the density (n)), mean velocity (Uj), and pressure
tensor (~j) are determined by averaging the various moments of the phase
j dvv -
at -nu.
at
space distribution function over velocities:
The second term gives

a
(1.5)
j dvvv.
8f
ax Ox
j dv vv f

j (1.6)
a
nj Uj = v fj(x, v,t) dv
Ox (~ + nuu)

~j = mj j(v-Uj}.(V-Uj)fj(x,v,t)dV. (1. 7)' This result is readily obtained by rewriting the integral as

In deriving the moment equations, we will suppress the subscript i,

since it is clear that these equations will apply to each charge species. j dv (v - U + u)(v - u + u) f = ~
111
+ nu U ,
Averaging the Vlasov equation over velocity gives
since J (v - u) f dv = O. Evaluation of the last term in Eq. (1.10) yields

j dv [~ + v.~ + ! (E + v: B) . ~] O. - (1.8)
B) . Ov
af B)
The first two terms in Eq. (1.8) give
j dv v m E
q (
+
v x
-c-
n q (
= - -:;;; E +
U x
-c- ,

where we have integrated by parts. Collecting the above terms, we obtain

the equation of motion for the charged fluid:

a
-(nu)
&t
a
+-·(nuu)
Ox
= -
nq (
m
E +--xc B)U ~.~
Ox m
(1.11)

a
Ox. . (n u) .
It is convenient to rewrite the first two terms of Eq. (1.11) using the
continuity equation and to assume that the pressure is isotropic, i.e.,
~ ! p where! is the unit dyad. Then
The third term in Eq. (1.8) vanishes, as can be seen by integrating by
parts and noting that f -t 0 as lvl -t 00. Hence the first moment of the 1 ap
Vlasov equation gives the continuity equation for the particle density: (1.12)
mOx

( 1.9) Observe that each moment brings in the next higher one. The continu-
ity equation for the density involves the mean velocity; the force equation
The next' moment of the Vlasov equation is for the velocity brings in the pressure. The next moment will give us an
. equation for the pressure (energy density) which involves the heat flow.

[ afat B) . aOvf]
J dv v + V·
B]
Ox +
q (
m E +
v x
-c- O. (110)
Continuing, we would end up with an infinite set of coupled equatiolli,
hardly a practical c!escription.
'JL "IV \.VI' r;; 'I\.. t.(.. ·LC - \.l.. 'l'>'\. +- ril.. 'l:.t.- + :.v l rr",,,,,-, ~ Y\.1.l..- ::?.._l0
'\le "7)/.. 0;:: rue 'U;<' _ __ ~x
 1.4 The Two-Fluid Description of Plasma 9
8 1 Basic Concepts

A great deal of simplification results from the llS!t..O{ the lower moment
Fortunately, we can truncate the moment equations by making various equations. In particular,
assumptions about the heat flow, which gives us a so-called equation of
state. The simplest assumption is that the heat flow is so rapid that the on nmv.-
ov.
temperature of the charged fluid is a constant. In this case, we have the 2 Ot + Ot·
isothermal equation of state: p = ne, where the temperature e is a
constant. This equation of state, plus the continuity and force equations Using Eqs. (1.9) and (1.12), substituting into Eq. (1.14). and cancelling
for the fluid, and Maxwell's equations form a closed description. terms gives
The isothermal equation of state is appropriate when w/k « Vt, op Bp ov. oQ (1.15)
-+v.-+3p-+2-=O.
where w and k are the frequency and wave number characteristic of the at ox ox ox
physical process being considered and Vt is the thermal velocity of the To obtain the adiabatic equation of state, we neglect the heat flow.
particles. In the opposite limit (wfk » tIt) we can simply neglect the This assumes that oQ/ox is much less than the other terms in Eq. (1.15).
heat flow. This assumption leads to an adi~batic equation of state, which For example, demanding that oQ / ox « op/ Ot gives w p » k Q, where w .
we will now derive. and k are a frequency and wavenumber characteristic of the process being
To obtain an equation for the pressure, we multiply the Vlasov equa- considered. Clearly Q < Qmax ~ ne Vt, where Vt is the thermal velocity.
tion by the kinetic energy and average over velocity: Hence, to neglect heat flow it is sufficient to assume that wik » Vt·
With this assumption, Eq. (1.15) reduces to
(E B) . of] O.
I mv2 dv
2
[Of + v. of + J..
Ot Ox m
+ v x
c Qv
=p (1.13)
(1.16)

At this point, let us specialize to one-dimension to simplify the algebra.

The first term can be written as The continuity equation allows us to express Ov./ox as

- -0
m
2 Ot
I f (v - u + u )2 dv ov. = _
ox
(!l.. + v.~)
at ox
In n . (1.17)

The next term in Eq, (1.13) gives Substituting Eq. (1.17) into Eq. (1.16) gives

mOl
"2 Ox f ( v - v. + u ] dv
3
=
eo
ox +
30
"2 Ox (u p)
mo
+ "2 ox (n v. ) ,
3
(! + v.:x ) In p - (! + u :x ) In n
3
0,

or
Q == (m/2) J(v - u)3 f dv. The final term in Eq, (1.13) is simply
where
.: +
Ot
v.~)
ox
E-3 = O. ( 1.18)

iJ.
2
I v2 E of dv = - n q u E .
ov
n

This equation shows that, following the plasma flow, p/n3 = constant,
which is the adiabatic equation of state for motion with oue degree of
freedom. This equation of state is readily generalized to p/n'Y = constant,
Collecting terms, we obtain
where 1 = (2 + N)/N and N is the uumber of degrees of freedom.
10
__ '(p + nmu2) + -30-(up) 10
+ --(nmu 3
) + -oQ = qriuE: (1.14)
2 Ot 2 ox 2 ox ox
 1 Basic Concepts 1.5 Plasma Waves
10 11

• THE TWO-FLUID
l.~ DESCRIPTION
Finally let us summarize the fluid equations, which we have derived by
raking moments with respect to velocity of the Vlasov equation. The first
rwo equations are the continuity and force equations for the density and
mean velocity of particles with charge qj and mass mj.
OF PLASMA 1.5 PLASMA WAVES
Using the two fluid model (the electrons as one fluid, the ions as the
other), we can investigate a wide range of plasma behavior. A character-
istic feature of a plasma is its ability to support waves or collective modes
of interaction. In the simplest case, these waves correspond to charge den-
sity fluctuations at a characteristic frequency determined by the electrons
(1.19) and/or the ions. In a plasma with no large imposed magnetic fields, there
are two such plasma waves: a high frequency one called an electron plasma

n] (
BUj
&t + Uj .
0.Uj)
Ox
= nJ qj
mj
(E + Uj x
c
B ) _1_ Bp]
mj Ox
. (
1.20
)
wave and a Iow frequency one called an ion acoustic wave.
Let us first investigate the high frequency charge density fluctuations
associated with the motion of the electrons. Because this is a high fre-
The pressure of each charged fluid is related to its density by an equation
quency oscillation, we can treat the massive ions as an immobile, uniform,
of state, which depends on the characteristic frequency (w) and wavenum-
neutralizing background with density nOi. Since the wave is electrostatic
bel' (k) of the process being considered. When w I k: « Vj, the isothermal
and the relevant electron motion is along the wave vector (taken to be in
equation of state is valid:
the z-direction), a one-dimensional treatment suffices. The equations for
(1.21)'
an electron fluid with density ne, mean velocity '!.Le,and pressure Pe then
where 8j is the constant value of the temperature and "i J8j/mj. are
When w / k » Vj, the adiabatic equation of state obtains: (1.27)

Pj
--:y = constant, (1.22) (1.28)
nj

where I = (2 + N)/N and N is the number of degrees of freedom. Pe (l.29)

constant.
When w / k ~ Vj, the details of the velocity distribution of the charges are n3e
important. The fluid description is then inadequate, and we must return We are using the adiabatic equation of state under the assumption that
to the Vlasov equation. the wave has a phase velocity w/k » Ve, the electron thermal velocity.
For a plasma composed of electrons and one species of ions, Eqs. (1.19- It is straightforward to develop an equation for the fluctuations in
1.22) constitute the well-known two-fluid model. This description is com- electron density. First we take a time derivative of Eq. (1.27), a spatial
pleted by Maxwell's equations, which relate the electric and magnetic derivative of Eq. (l.28), and eliminate the term B2 ne1/.e/&t8x to obtain
fields to the charge and current densities of the plasma. In@units,
Maxwell's equations are B2ne 82 2 e B 1 B2pe (l.30)
--, - - (ne1/. ) - - -(neE) - - --, = O.
V . E = 47i P (1.23) Bt- Bx2 e m Bx me Bx-
V .B = 0 (1.24) We then use Poisson's equation to relate the electric field to the density:
1 BB
Vx E = - - - (1.25)
BE = _ (ne - ZnOi) , (1.31)
c &t 47fe
Bx
47f 1 BE
Vx B = - J + -c -{Jt , (l.26) where Z is the charge state of the ions.
c
vVe next consider small amplitude perturbations in density, velocity
Where p L J
nj qj , J = Lj nj qj Uj , and c .. the velocity-of light.
and electric field and linearize the equations, i.e., ignore products of the
12 1 Basic Concepts l.~ Plasma Waves 13

pertnrbatioIls. U we let ne = nO + n, Ue U , Pe = nOee +P , and Since W / l: « Ve. the electrons are described by the "'~thermal equation of
E = E, Eqs. (l.29-1.31) give state: pp = neee. Substituting Pe into Eq. (1.37) and letting n ; = nf) + n.,
ami E = E, we obtain the linearized equation:
P = 3m v; n (1.32)
( 1.38)
8E _
- = - 4rr en (1.33)
8x .
where no is the uniform unperturbed density of the electrons.
82n no e 8E 82p The equations for the ion fluid with density ni, mean velocity Hi, and ,
at2 m 8x 8x2 = o. (1.34)
pressure Pi are ~
8n 8 r
Substitution of Eqs. (1.32-1.33) into Eq. (1.34) then gives a wave equation
at
, + -(niui)
8x
= 0 (l.39) I
describing the small amplitude fluctuations in electron density:
8 8 2 Ze 1 8Pi
+ = (1.40)
(~
at2
- (1.35)
at (niUi) 8x (niUi)
Pi
M ni E
M 8x

J
where wpe = 4 rr e2no / me is the electron plasma frequency lor a plasma
with electron density no = ZnOi. If the density is expressed/in units of
cm -3 , t hen wpe = 5.64 x 104 ne1/2 .
Looking for a wave-like solution (n ~ ei kx-i wt) we readily obtain
- constant,
nt
where Z is the charge state and M the mass of the ions. vVe use the
adiabatic equation of state for the ions under the assumption that W / k »
Vi, the ion thermal velocity. To derive an equation for the evolution of the

from Eq. (1.35) the dispersion relation for electron plasma oscillations: ion density, we proceed as before. Take a time derivative of Eq. (l.39), a
spatial derivative of Eq. (1.40), and eliminate 82niUi/at8x to obtain
W
2
= Wpe
2
+. 3 k2 2
Ve· (1.36)
(1.41)
Note that the frequency of these. waves is essentially wpe, the electron
plasma frequency, with a small thermal correction dependent on wavenum-
ber. If kinetic effects are allowed, there will also be a small damping or We now take ni = (no/Z) + ni, Ui = Ui, Pi = PiO+Pi, and E = E, where
growth depending on the details of the electron distribution function for the superscript denotes small perturbation; further, Pi = 30ini, where 9;
velocities near the phase velocity of the wave. This damping will be dis- is the ion temperature. Substituting these expressions into Eq. (1.41) and
cussed in Chapter 9. neglecting products of the perturbed quantities, we obtain
A plasma will also support charge density oscillations at a much lower
frequency determined by the ion inertia. To investigate these oscillations,
(1.42)

g..
we need to consider the motion of both the electron and the ion fluids.
Since the frequency of these oscillations is much less than the character-
istic frequency with which electrons responsL(.wpe), we can neglect the A wave equation for the fluctuations in ion density is now readily obtained
inertia of the electrons; i.e., neglect the j~tron If we again con- by substituting from Eq, (1.38) into Eq. (1.42) and noting that ne '::: Z ni,
sider motion only along the direction of propagatio (taken to be the. x since the electrons closely follow the slow motion of the massive ions:
direction), the force equation for the electron fluid r uces to
Z ee + 3ei 82n
( 1.43)
aPe LVI 8x2' = o.
neE - - (1.37)
e ax
 1 Basic Concepts 1.6 References 15

l£w(' search for wave-like solutions (iti ~ e,kx-iwt), Eq. (1.43) readily where ADe == JBe/4 ITno e1 defines the electron DebY,e.length. If Be is ex-
. . 1 c disperSlOl1relation for ion acoustic waves: pressed in units of ev aud no in units of cm-3, A~e = 743 (Be/no)1/2
~J
\.c" ( 1 Equation (1.-17) is easily solved by Fourier-transforming and then invert-
w = ± I.: Vs, (1.44) iug, which gives
,
IjJ = -q exp (- - 1') . (1.48)
," V. = J(ZBe + 3Bi)/Af is called the ion-sound velocity. These low T ADe
whtlC , . ..1:_ ...
ey waves are the analogue of sound waves ill an orrnnary gas. The solution is readily verified by direct substitution. This result demon-
freqLlell .' .. . .
. 15 pI'ovide the inertia and fluctuations .ill the pressure provide the strates an important feature of a plasma alluded to earlier in this Chapter.
The 101 , .
. 1(f force. The electron pressure fluctuations are transnutted to the
restonlo . . The plasma electrons shield out the field of a discrete charge in a charac ..
. . by the electric field. If kinetic effects are allowed, there IS a damping teristic distance which is ADe. In general, the ions also contribute to the
~ ...
d to both the electrons and the ions as will be discussed later ill 'Ch apter shielding.
9L1~hiSdamping is small provided that w / k » Vi, which requires that
ZBe» Bi· The assumption of quasi-neutrality requires that kADe « 1.
References

1. Spitzer, L., Physics of Fully Ionized Gases. Interscience Publishers, New

1.6 DEBYE SHIELDING
York, 1962.
Finally, it is instructive to show how a plasma modifies or shields the elec~ I 2. Dawson, J. M. and M. B. Gottlieb, Introduction to Plasma Physics, Lec-
tric field of a discrete charge. We place a charge q at rest in a plasma with ture Notes, Princeton University, 1964.
an initially uniform electron density no and treat the ions as a fixed, neu- 3. Chen, F. F., Introduction to Plasma Physics. Plenum Press, New York,
tralizing background. The electrical potential (E = - \7 <p) is determined 1974.
by Poisson's equation: ) 1.: •. ~w C. ;"" .••..•• :? ~~ 4. Schmidt, G., Physics of High Temperature Plasmas. Academic Press, New
York, 1966.
\72 cjJ = - 47r q 5(x) + 47r e (ne - no) , (1.45)
5. Stix, T. H. The Theory of Plasma Waves. McGraw-Hill, New York, 1962.

where the charge is located at r = 0 for convenience. In the static limit,

the force equation for the electron fluid reduces to

Where an isothermal equation of state has been used i.e., Pe = ne Be,

whe.re Be is the electron temperature. Since E - \7cjJ, e electron
density then is

ne = no exp(~~) . (1.46)

~~i~g that ecjJ/Be « 1, we expan~ the exponential in Eq. (1.46) and

stltute ne into Eq. (1.45) to obtain

- 47r q 5(x) , (1.47)

 CHAPTER 2
.•..
Computer Simulation
of Plasmas
Using Particle Codes

Having considered a theoretical description in which the plasma is treated

as two charged fluids, let us introduce a complementary numerical descrip-
tion of plasma behavior using particle codes [1-12]. Computer simulation
of plasma using particle codes is a very direct and powerful approach, par-
ticularly for investigating kinetic and/or nonlinear effects. The approach
is extremely simple: numerically follow the motion of a large collection
of charges in their self-consistent electric and magnetic fields. The basic
cycle is illustrated in Fig. 2.1. From the positions and velocities at any
given time, compute the charge and current densities on a spatial grid
sufficiently fine to resolve the collective behavior. Using these charge and
current densities, next compute the self-consistent electric and magnetic
fields via Maxwell's equations. Then use these fields in the equations of
motion to advance the positions and velocities of the charges. Finally, con-
tinue around this basic cycle with a time step sufficiently small to resolve
the highest frequency in the problem (which is often the electron plasma
frequency) .
 2 Computer Sirnulation Using Particle Codes
Figure 2.1 The basic cycle of a particle simulation code.
Again, what makes this approach viable is that we are investigating
the collectivebehavior which occurs on a space scale greater than or com-
parableto the electron Debye length rather than the fine scale fluctuating
microfieldassociated with discrete particle collisions. As we have discussed
in Chapter 1, these microfields can be systematically ignored in a collision-
lessplasma (i.e., when the number of particles in a Debye sphere is much
greater than one). T~ is fortunate since it would be extremely difficult
in practice to use a spatial grid fine enough to resolve these microfields,
whichoccur on space scales even less than the interparticle spacing.
Alternatively, we can view our approach as using ''finite-size'' charges
(ofsizea). It is physically obvious that the behavior of a collection of such
, chargesis the same (with minor modifications) as the behavior of point
chargesfor scale lengths l » a, but fluctuations with scale lengths l « a
thu suppressed. By choosing a '" ADe (the electron Debye length), we
t US achieve with a "trick" what nature does with the use of an enormous
nllInher of particles i.e., smooth out the fine scale length microfields.
pl From a computational viewpoint, a particle code is remarkably sim-
th:' l'h~ reader can easily write his/her own code. We will first discuss
CodbasIC ingredients. Then a very simple code will be presented. This
icd.!e, and modifications easily made, will be very useful to test theoret-
a.xn calCulations and to investigate various nonlinear effects not readily
eIlableto analytic treatment.
2.1 Basic Ingredients of a Particle Code 19
2.1 BASIC INGREDIENTS OF A PARTICLE CODE
Discussion of a very simple but useful particle code Will·sufficeto illustrate
the basic ingredients of all such codes. We treat the ions as a fixed, neu-
tralizing background, assume no imposed magnetic fields, and consider
ouly electrostatic fields. In this electrostatic limit, the magnetic field gen-
erated by plasma currents is negligible, and Maxwell's equations reduce
to Poisson's equation, V' . E = 4 7r p. Here E is the electric field, p is the
charge density, and cgs units are used. Variations are allowed in only one
direction, and periodic boundary conditions are adopted.
Our first task is to compute the charge density. Our system extends
from 0 to L, as shown in Fig. 2.2. We divide this system into Ne cells and
for convenience take the cell size (8) to be unity (i.e., L = Ne). The grid
points are identified as the integer values of the position, augmented by
1 so that the counting begins with 1. Note that the Ne + 1th grid point
is then identical with the first, due to the assumed periodic boundary
conditions. Given the position of the charge, many schemes can be used
to assign it to the spatial grid. For example, we could just assign the
charge to its nearest grid point location. A better scheme is to share the
charge between its two nearest grid points. For a charge located a distance
6.x to the right of the ith grid point, we then have
6.p(i) = q (1 - 6.x)
6.p( i + 1) = q 6.x ,
where 6.p is the increment to the charge density.
Having assigned the charges and determined the charge density on
the spatial grid, we next determine the self-consistent electric field using
Poisson's equation:
8E
8x = 47rp.
The simplest approach is to finite-difference Poisson's equation,
E(i+l) = E(i) + 47r8 [P(i+l)2 + P(i)].
An alternative approach is to Fourier transform the charge density, use
Poisson's equation to find El. U k El. = 47r Plo), and invert the transform
to determine the electric field on the spatial grid.
 2 Computer Simulation Using Particle Codes 2.2 A I-D Electrostatic Particle Code 21
20

used. It is interesting to note that a numerical instability would be intro-

duced if AD e were chosen to be a small fraction or the- grid size [13]. This
Particle instability is due to aliasing, which arises from the rict that on the grid a
x=o I x=L
i i disturbance with wave number k cannot be clistinguished from spurious
I
8J- ones with wavenumbers k + 27in/8, where t5 is the grid spacing and n is
,,
I
an integer.

2.2 A I-D ELECTROSTATIC PARTICLE CODE

As is now apparent, a particle code is quite straightforward from a compu-
cell n cell n+ I tational viewpoint. Of course, particle codes become more complex when
Figure 2.2 . The charge sharing scheme for finite-size particles.
the full set of Maxwell's equations is allowed and magnetic fields act on
the particles. But the basic concepts are the same: 1iS€ of a spatial grid,
the spacing of which is chosen to resolve the collective behavior, and the
The last step in the basic cycle is to use the electric field to move the mapping between the cliscrete grid and the particles. To further illustrate
particles. The electrical force is assigned from the grid to the individual the ideas, a specific I-D electrostatic particle code will be presented. For
particles using the same scheme chosen to assign the charges to the grid. tutorial purposes, the code has intentionally been kept in a form which is
For example, considering a particle a distance 6.x to tIle right of the ith easily deciphered. Before presenting this code, it is helpful to briefly dis-
grid point and using linear interpolation, we obtain for the force F on the cuss two topics: the dimensionless units chosen and the finite-difference
particle ' solution of Poisson's equation.
F = qE(i)[l - 6.x) + qE(i+l)6.x. The basic equations under consideration are:
The velocity (v) and position (x) of each particle are then advanced 6.t dv eE
in time using a "leap frog" algorithm, i.e.,
dt m
dx
Vn-1/2 + F" 6.t v
dt
xn + vn+1/2 6.t . oE
ox
The superscripts denote time step. By defining x and v one-half time step
apart, we achieve second-order accuracy in the time step. In the initial where nay is the density of the fixed ion background and n is the den-
conclitions, x and v are defined at the same time (t = 0), but it is straight- sity of the simulation electrons with charge e and mass m. \Ve choose
forward to then clisplace the velocities backward in time using the force climensionless variables in the following way:
at t = o.
The plasma evolution is compu ed by simply continuing around the
t' Wpe t
basic cycle using a time step small ough to r olve the characteristic
, x
x
oscillations of the plasma. As discussedi ter 1, in this electrostatic 8
, v
lirnit, the highest frequency oscillation has a frequency near Wpe, the elec- v
tron plasma frequency. Hence a time step of about 0.2 w;l
is commonly
wp"o·
~sed. Of course, it is also necessary to resolve the scale lengths character- E
E'
Istic of the collective behavior. Hence a grid size of about ADe is commonly 47ien"vO'
 2 Computer Simulation Using Particle Codes 2.2 A l-D Electrostatic Particle Code 23

,f
, e b = L (system size)/NC (the number of cells), nay = N P (the For j > 1,
\dl bcr of simulation particles) / L, and wpe = (4 1f nay e-? / m) 1/2 . In these
nlJfJl j-I 1 ~
\-ari3bJcs,the equations become
E(j) E(1) + L 2 [RH(i) + RH(i + 1)]
dv' i=1
-E'
dt' Substituting, we obtain
dx' ,
v
dt'
dE' 1
[N - Nay] .
Ell) + ~ { Ell) + ~ ~ [RHli) + RHli + I)]} o.
dx' Nay
Hence
Here N = n {; is the number of particles initially assigned to each cell
and Ne» = nayS = NP/NC. In terms of these variables, the total NC j-I 1
energy(TE) of the system is NC x E(l) LL "2 [RH(i) + RH(i + l)J .
j=2 i=1
NP NC 2
m 2
TE L
i=1
-v·
2 t
+ ""
L......
;=1
Ei {;
81f Since j is simply a counting variable, we let j' = j - 1 to obtain

TE 1Y!
2 .2
wpe U
2
[ NP
""
L......vi ,2
+ Nay
NC
""
L......E /2i
1 NC

LL
j-I NC-I

LL
j'

;=1 ;=1 j=2 i=1 j'=1 ;=1

The finite-difference solution to Poisson's equation is also straightfor- Note that we can add the j' = NC term since by charge neutrality the
ward. Defining a normalized charge density RH, we have spatially-averaged charge density vanishes ('L~~ RH(i) = 0). Hence

NC .,
aE' = RH. J 1
ax' NC x E(l) = - L L"2 [RH(i) + RH(i+1)]
j'=1 ;=1
Afterthe charges are assigned to the grid, we know RH(i), (-i = 1, NC).
Droppingthe prime notation and taking S = 1 then gives With E(l) determined, each of the E(i) is readily found.
A Fortran implementation of the code is presented in Fig. 2.3. The
1
E (i + 1) = E (i) + "2 [RH(i + 1) + RH (-i)] . coding is self-explanatory; many comment cards are included. Arrays x
and v have dimensions corresponding to the number of particles N P and
arrays RH and E have dimensions corresponding to the number of cells
Couside.
th rmg 0 nl y peno
. di c b oun d ary COIldi tions
. . \
Imp~es t he eo tram . t
at E(NC + 1) = E(l). NC. For clarity of presentation, various optimization techniques, such
s?Ur procedure is to first solve for E(l), using the con .:t;i n that the as time normalization in units of fit, have not been exploited. The cells
pat1allY-averagedelectric field is zero i.e., to which a given particle is assigned are determined by truncation of its
position using fixed-point arithmetic conversion. Prior to entering the ba-
NC sic cycle, the particles are initialized to represent the initial state of the
L E(j) O. plasma for the problem under consideration. The most common diagnos-
j=1 tics are the temporal evolution of the electrostatic and kinetic energies,
 2 Computer Simulation Using Particle Codes 2.2 References 25

e e .•.•
P'IOCIt"'''''
CALL H£pI
•••• 'NIIIfPUT.
•• It
,",5" (ltl"
E E """''''Oq,,,
snapshots of the electric field as a function of position. and plots of the
00 13 1".'"
c ''i( OI/lolEHSIO",.t.l ,.,HICl( S''''Ul"rION
C
c
'f.~IOOIC
FIlED
IIOUNO"''''''
10H$
eOHOI TiONS
lJ
'1
(KI'(I("VIII'VII,
COHIll'IUE
.r - (I phase space of the particles at various times . .fInally: we note that
At r ••. US( OHLY HlLF OF E
C »c IS r •• e 'IUIoIIlf.A OF CElLS
C
C
", IS
r~(s[
'"f. NVIoI"f..
1o,,'tlUYS .'H
0" HECTAOH!!i
OF 511£ HP
rc srlGG(A
! "I r ••(
'HE IHOClf!(S
J I CO t o ~I energy conservation is an important test of the code performance.
C r"(SE
OI"'E~SIOH
••IUtAY$
XI ""f!,
ARE OF SIZE
VI
.•••e
",,'1 1ll'~IJS I1
)if
00 ~.
[11,· .• ~'(III
!" I ItC 1
Even with this simplest particle code. many instructive problems can
) I COHI !foIuE
O!iIoI[NSIOM 'hUlI'I.tIU"
C , •••It! !CLE LOO"
C 11-'£5( ••1I1t.uS
011ol£"'510H
"liE OF sr z e "'£MO
££1 I"'t. £ !I. 1 , ••• r • rEt ".,1 C ""OVE 11"1( HECliIIO"S
00 '1 I- I ."IP
be investigated. For example, we can excite electron plasma waves and
C

C
liHUl'

CI'lf
OF
0''''(Io5I::1N
INPUT
51:£

~.r •.
IoI(NO
,"lI I""
OR MC
...••
·(11'
.
,.. ...,
01-( I1I .,,,,
examine their behavior in both the linear and nonlinear regimes. The code
'Wit IrE t ~9. l' I
FOR"''''I"
•• [ •• OO~.121
crve .••C .••• '.'.II·
•• c. HP ",I"
IN n-, "1'''''1
••• • ••·1 can be easily extended to include ion motion, and the reader is encour-
FO'lC(--( I"'" 11 'OII-EI"'I' Ot.
FOA •••• r I J I SI
'fRIi£'U.!ll
fO"'''''' T I"~ Ne •
NC."'p.",,,
". I H" HP • '".
VI t I 'VI 11 .F-OACE'
"(III'XI I I '\11 I,'
01
or aged to do this as a learning experience. After becoming familiar with
C •••.•.IIE SUAE paAIICLE IS '11111'41" Of( SYSIE'"
, I ~/- .." • -.! S I
"''''TE
1'0,h••.rr-
IS.,?6,
GIV[ vu..or.r~!IIO IN Fq ~-,
,I'
11='
1"( III
III 11
l t , I llll'1.1
at
11' "HC
","'CI :t111-ItIII·A""
the basic techniques, one can easily graduate to the use of more flexi-
C "'SSIG'" CH .•.ACE TO THE CRIO~-"R(A ""EIGHTJ."C

lS
R(AOI S9. Hi
fOIt •••••r tlF'I
\/fLOT.
~I
f(HO
."I_XIII
ble and optimized particle codes, such as ES1 which is available through
,..:t!UtH.1S1
FOA•••"rt~ vTE • ·."'.H~
vTE,Or,T[NO "N''''
OX·IIII- .• H

,. r[~o • '". ". H

Of • ~."" ~I
H· ••• I NMFECC, the national computer center for magnetic fusion energy. Al-
C"U.
•• A I
"AIUt,
C"'LL
rr
t I".
••.
I'A .•IoI[
SE rCH"
2J I
25!
. ~.
~C. NP. "'P
'HE.OT.TE~O
. I. IiI , ,. ,. ""'''''1
':toH""
AHI""
COHT IHU(
'AMI "'I ~ I1
.qHIW! ·Ot
-ox I
though less accessible, large 2 and 2-idimensional (two-dimensions in
.• HC9HC
.~C ,91.1C· ,
1'1'01
\(1·';1'1 space and three in velocity) electromagnetic, relativistic particle codes
IF lit 1 I. t . NOI, GO 10 \
aHP-HP

,.,
aHOAI,,. .•.HC' .• HP KI"
LI'L'"
now exist. Such codes solve the full set of Maxwell's equations and use
..
C CO"'PuTE IH( KINEIIC EHERGY
'(la,
L'·'
(K2-'.
00 • ~ I. I . HP
relativistic particle dynamics. So-called implicit particle codes have also
been constructed [14]. These codes allow time steps greater than w;-e1 and
"" ~• E K 2 • (It 2' VIII' \I r I I
Hor_, lOT
."O,_HOT.,
C SAvE rHt: ENERGIES FOR L .• HR PLOTTING

.t,lEHo-rEHO·'. , (11 r l 11 -. ~'I Ell "E K 2'

((ILll"'E(A
C SEf PLOTTING
",PLOr-IoIEHO
"tAU
rEILII-EKrlll'EEILII are used to more economically study low frequency kinetic phenomena.
IF IHC.GT \,IPlOTl IoIPlOf-NC
C PI.OT
""''''''l
S",lpSHOfS OF r~E sYSreW EvER" 'Ill'£" T'''P
I·,.
,,'-,
00 l." \,IPLOT
C SURST\IUTE
IF
'I'OUR LOCAL
l"l.l r ",P, GO TO
PLOTTIHG
"
ROUTINES

I'llll-a\
a'(I_'
C PH Tt"l! lE
2.lU .•.
TNf:
HC
PAIn ICLf:S
""-.
CALL
CALL
ClAT\lNIHC.'I''''I'''.Y'''.'''.E.
IotAPCl' • A"'C. '1''''''''. 1•••.•11
I1
References
C "'",;£ iUAf: P,,'HICLES 11<4 SYSTEIoI
CO I ,-,. ~P CAlL SETCHII .• L.'.'.I.I."
.•I -, - ~
",AIT(II ••. 61, r
"'I!' •.•. , •.•.
NOII.W FOR"'''Tr' F-IHO 'IS e
POSITIO"",
'FIX, II.l r • 1 XIII-'III' .•.NC
r- r I wE _ ". F I' ~I
IF 'It 11 ce . "'HCI tl" "11 I' • .•.HC C"'LL
C"LL
TA"CE
I",",WE
IPL. E. «c, I. , I 1. Introductory a.rticles:
CO",T l!otUE
va-'
00 2
.
1-'. HP
V""","·12
Viol".'
'YI(
12 • 'IrE
Denavit, J. and W. L. Kruer, How to get sta.rted in pa.rticle simulation,
C"LL u.P'lI •...•. NC.V •• \H.VWAXI
v8-·vl
!
C .•. Ll A .•.NOUNI VVI
VIII .\lTE· VV_V.
C"'LL
'IIRI TEll".
SETC •.•••.•
621
2. .•.••
I
I.I •• j
Comments on Plasma Physics and Controlled Fusion 6,35 (1980);
C tNlTlALlY .•.SSIGH
00
'Hf:
J 1-I.HCI
CH .••II.Qf: 10 IHE 011.10 52 FOIU"'AT'·
1- It"'£. ••
P"""SE
-. F I'
SP'CE
. ~,
"
W. L. Kruer, Plasma simulation using pa.rticle codes, Nuclear Technology
"IHt 11.' C"LL POIHISI1. Y. "p. I. "
00 I I- I. NP CALL F-RA"'E
27, 216 (1975).
~-XII'
.•.N·'"
~ ;~~~ T ~ ~~~E M~~" ~~~ .• ~H~Lg~~~:G F !~~~ I :~~ IS
ro
C
O .•.-tlll-"''''
COUHTlNG FAOIol I TO HC
" IF
C"LL
I T LT.
CAAI
NOI 00
.•••(1'40. '!'~I"'.
5
Y •••.• lI. EE. 11 2. Birdsall, C. K. and A. B. Langdon, Plasma Physics via Computer Simu-
••••• ·1 C"LL ••••• "GII .TEHO.'!''''IH.YN'''XI
u •••• ,
lt~II't1.AHIHI-' '11
C"'LI.
'llIIIT(II
S( ICMI'
••• 61
.. 2 .'."
T
I. I. 11
lation. McGraw-Hill, New York 1985.
;tHIIoI'_i!HIIoII·O. III FOA"'ArI~ ( tHO (NEAGY vs IIN( "
CONI \NU(
CONTINUE
I'
CAll
r 1.•••
( .'
TRAC('
11 ~,
Plo Et: ..•.•(NO. t . " 3. Hockney, R. W. and J. W. Eastwood, Computer Simulation Using Parti-
AHI lIaAHI ,,,AH,,,,e' 11 C"LL ~"A'"•••
C .• 00 !~ ,"AAGE
00 6 1-'. HC
01< UNII'OIl."LY SP"C(O 10HS CAlL C
APGI'
1011 "'(."f0.tloltM.Y"'A,"II,.EJI..1I
• HHO. ''''IN. '1''''01.,"11,1
cles. McGraw-Hjll, New York 1981.
IIIM I II_IIIHI! I' ANPf .•.HC C"'LL S(lCH" .. 2 '.i.I.'.II

C
6
SOlV(
;tHI I1 'A"I
00ISSOHS
I I' "NO AN
(CUAflO'"
'IIAITEIII'.6n
FOA"''''''' lIM(lIC
T
EHEAG'I' vs f!W( -,
4. Alder, B., S. Fernback, and M. Rotenburg, eds., Methods in Computational
C nl'lo EllI " .• v(A .•GE HII'D I" r 1 IoIE " F I' ~ I
FIf:LO IS
Physics, Vols. 9 and 16. Academic Press, New York, 1970 and 1976.
W

(IIUH2-'
C"LL I ''''C( PL. (11. "'EHO.i 1. 11

00 )2 J.'. HC
C"'L L I""'IoI~
CAlL C .••"I .•••W'N(HO. '1'1011"'. nUl. lE."
EAUH'-'
00 J' 10 I. J
C"LL
C"LI.
!'UPGli
S(lCHII
.. r(No.ywIN
.. 2 .•.•. '
. ., •••.• "
.•. 11
5. Bunernan, 0., Dissipation currents in ionized media, Phys. ReI" 115, 503
'I' llt.!;1 NCI It· 1
'llIIITEIII e . a..!!1 I
¥,
31
II
EAU"
(IIUN2o(AUH1'
I.EAUH It'
EAUH'
~'I AHIII' 11I1041
It I'
I"
FO""'''T'"
T luE
Tor"'L
• ". F ,.
(,..ERGY
~ I
vs f!wE
(1959).
(111 '-EAUH2f lHC C"LL TA"C( I PL. r e • .•••(NO.
."CIoI·HC- , C .•ll FA ""',.~
6. Dawson, J. M., One-dimensional plasma model, Phys. Fluids 5, 445 (1962)
".
C"'LL EX IT
00 r
I" I. HC'"
( , , • , 1"( I I I ' I . ~' , RH I I' I1 • 11.1'11
, 1I

.. of anomalous diffusion, Phys. Fiuids

ccvevr e
CHAD
fUfC'I,o(11i
TH( CH.•.AGE O(NSI r, .• AA'"
C
C .." rH .•
11.""'00'"
GlUSSloI.N
Huw8ERS
01 STilI IWflON 7. Hockney, R. W., Computer experiment
•
00 •
IIIHtll·'.
1°'. HC I
~,
SUIROUIINE II"HOUW, X, I

9,1826 (1966) .
C coesu r e KINETIC _HO FIHO ffol(lilGtES 00 , ~ ! I. '2 0

:t
IF
[(
00
....•
12
11(,

t v r . NC
I. I. "'01' 00 10 f 1 1-I I' AH"lrOO"
"(TuA
eso
•.•
~

One-, two-, and three-dimensional

8. Morse, R. L. and C. W. Nielsen, simulation
11 £E"'f("'FIII'[111
of two-beam plasmas, Phys. Rev. Letters 23, 1087 (1969).
Figure 2.3 Listing of a one-dimensional pa.rticle code.
 2 Computer Simulation

. is 11 C. K. and I). Fuss, Clouds-in-clouds,

Using Particle Codes

clouds-in-cells physics for

CHAPTER 3
13lf<"a ,
9 ._body plasma simulation, J. Camp. Phys. 3, 494 (1969).
lOan). J P. and rv,
S
•.. V . R 0 b erts, Th e opturuzation
...
10· !30rI d,3· dimensions, 1. Camp. Phys. 4, 552 (1969).
f . le calculati .
0 par tic e ea cu auons III
Electromagnetic Wave
.) an
-. r \V. L., J. M. Dawson, and B. Rosen, Dipole expansion method for
r.
I fue,. .
I· lasma simulatIOn, 1. Camp. Phys. 13, 114 (1973) .
Propagation in Plasmas
.) ;enavit, J. and W. L. Kruer, Comparisonof numerical solutions of the Vlasov
1-· aLion with particle simulations of collisionless plasrnas, Phys. Fluids 14,
equ
1782 (1971).
13.Langdon, A. B., Effects of the spatial grid in simulation plasrnas, 1. Camp.
phys. 6, 247 (1970). J

H. For several articles on implicit particle codes, see Multiple Time Scales,
(Bfackbill, J. U., and B. L Cohen, eds.). Academic Press, Orlando, 1985.

3.1 WAVE EQUATION FOR LIGHT

WAVES IN A PLASMA
Having examined the characteristic charge density oscillations which are
supported by a plasma, let us now consider how a plasma modifies the
propagation of electromagnetic waves. Motivated by laser fusion applica-
tions, we will continue to assume that there are no large imposed (or self-
generated) magnetic.fields. We begin by examining the linearized plasma
response to a high frequency field of the form [1]

E = E(x) exp( -iwt) . (3.1)

Since the frequency w is assumed to be .<: wpe, the ions are treated as a
stationary, neutralizing background with density nOi(x). If we neglect the
terms involving Ue' VUe and u., X B as products of small quantities

21
 3.1 Light Wavf>.sin a Plasma 29
3 E-M Wave Propagation in Plasmas

(i.e., r-J IEI2), the linearized force equation for the velocity of the electron Since v x fE = (\7 x E + "i1€ x E, we obtain

fluid reduces to
8ue e
at = - m E(x) exp( -iwt) .

Since the current density is .J = - no(x) e Ue, As our first application of these results, let us derive the dispersion
relation for electromagnetic waves in a plasma with a uniform density.
8J Bue w~e(x) Since V E = 0 and V . E = 0 , the wave equations for E and B become
fit = - no(x) e at = ~ E
identical. Assuming a spatial dependence of exp(ik· x) then gives the
dispersion relation for electromagnetic waves in a plasma:
where wpe
2 47r e2no/m and no = Z nOi is the electron density. Hence

. i
J -_ t wpe E = (!
E ,
47r W
or
where the high frequency conducti:vity of the plasma is a = i w~e/ 4 7r w. W2 = w;e + k2 c2 . (3.8)
To develop wave equations for the oscillating electric and magnetic Note that wpe is the minimum frequency for propagation of a light wave in
fields, we first consider Faraday's law and Ampere's law, which become a plasma i.e., k becomes imaginary for W < wpe' Since the characteristic

tW B
response time for electrons is -z:
the electrons shield out the field of
VxE= , (3.3) a light wave when W < wpe. Hence the condition wpe = W defines
c
the maximum plasma density to which a light wave can penetrate. This
47r so-called critical density is ncr = 1.1 x 1021/.\/ cm -3, where AI-' is the
VxB = -(!E
2W
E. (3.4) 4
free-space wavelength of the light in units of microns (1J..Lm= 10- cm).
c c
To investigate some of the basic features of the propagation of light
Substituting for a into Eq. (3.4) gives
waves in an inhomogeneous plasma, let's consider plane waves normally
tW
incident onto a plasma slab. Assuming variations only in the z direction,
VxB = --EE, (3.5)
c we then have

where E = 1 -: w~e/w2 defines the dielectric function 0 the plasma. no = no(z)

Taking the curl of Eq. (3.3), substituting from Eq. (3.5), d using a E f(W,Z)
standard vector identity gives E(x) E(z) exp(-iwt). (3.9)

W2
V2 E - V(V· E) + ""2 E E = o. (3.6) In Cartesian coordinates, the wave equation for E (Eq. 3.6) becomes
c
d2 w2
The wave equation for B is developed in a similar fashion. The curl -d 2 Ex,y
z
+ ""2

c
E e.; = 0,
ofEq. (3.5) gives (3.10)

2W
V x (V x B) - - V x (fE) .
c
 3 E-M Wave Propagation in Plasmas 3.2 WKB Solution in an Inhomogeneous Plasma 31

Likewise, the wave equation for B (Eq. 3.7) reduces to To next order in the gradients,

2iw , i w 1lJ' Eo
~
~ dE dBx,y
0, - {I Eo + 0, (3.14)
E dz dz c c
dB,
o. with the solution
constant
d.::
Eo(.::) = (3.15)
We will first develop a WKB solution for the fields. Although limited
to weak density gradients, this solution provides an excellent illustration Substituting Eqs. (3.13) and (3.15) into (3.12) then gives
of how gradients in the density affect the wave propagation. Then we will
complement this analysis with an exact solution for the fields, assuming E(::) =
EFS
E1/4 exp [.iw;Z
~ ~')
Vqw,Z) dz ,] , (3.16)
a lillear variation in the plasma density.
where EFs is the value of the electric field in free space.
It is apparent from Eq. (3.16) that the amplitude of the electric field
3.2 WKB SOLUTION FOR WAVE PROPAGATION increases as the light wave propagates toward higher density. This be-
IN AN INHOMOGENEOUS PLASMA havior is readily explained physically by noting that the energy flux is
A very useful approximate solution for the wave propagation can be ob- conserved, i.e.,
tained in the limit that the fields vary slowly in space. It's most convenient Vg IE(zW cE~S
to solve for the electric field. If we take the electric vector to be in the x 87r
s;;.- , (3.17)

direction and let Ex = E, Eq. (3.10) becomes where Vg is the group velocity of the light wave in the plasma. Using the
dispersion relation (Eq. 3.8) to relate the local value of the group velocity
to the local value of the dielectric function gives vg/c = Vc(w, z). Hence,
(3.11)
Eq. (3.17) becomes
EFS
We assume a slow variation in the dielectric function of the plasma (i.e., I Eo (z) I = ~/4
c .
'
a weak gradient in density) and look for a solution of the form
in agreement with the WKB result. Since the energy flux can also be

E{z) = Eo(z) exp [. ;z

tw
~ llJ{z)" dz ] , (3.12)
expressed as C E x B /4 7r , we can easily see that the amplitude Ba of the
magnetic field of a light wave is decreased in a plasma, i.e.,

Wherethe amplitude Eo (z) and the phase IlJ (z) are slowly varying func-
t10llS. Differentiating E{z) and substituting into Eq. (3.11) yields where BFs is the value of the magnetic field in vacuum.
The validity condition for the WKB solution can be readily estimated.
Eo" + 2iw IlJ Eo' For example, in deriving Eq. (3.14), we required that
C
W W
Eo" « I
-1lJ Eo, -IlJEo·
I

Wherethe prime denotes a derivative with respect to z . To lowest order, C C

Weneglect all derivatives, obtaining Noting from Eq. (3.16) that k{z) = wllJ/c, it suffices to require that

V c{w,z). (3.13) ee;

- «kEo·
az
 3.3 Analytic Solution with Constant Density Gradient 33
3 E-M Wave Propagation in P'lasrnas

. 'ng for Eo from Eq. (3.16), we then obtain the condition

t1ttlt1
••.•• 1.0
se= Of.
,,\ OZ « 871' e , (3.18) Ai(x)

,,\{z) == 27r/k{z). In other words, the variation in the plasma den-

~here ust be sufficiently slow that the fractional change in the dielectric
SltY: in a local wavelength is very small. Note that the WKB solution
on I
~ down near the critical density where f. -+ 0 and ,,\ -+ 00. 2

3.3 ANALYTIC SOLUTION FOR PLASMA WITH

A CONSTANT DENSITY GRADIENT
The WKB solution is especially valuable because it provides us with a very
intuitive way to describe the wave .propagation. The wave is assigned a
waveOl.unberand group velocity which are defined via the local disper- Figure 3.1 A plot of the Airy function Ai(X).
sion relation. The amplitude and the phase of the wave are related to
these locally-defined characteristics in a physically obvious way. However,
This differential equation defines the well-documented Airy functions, Ai
strictly speaking, this approximation is valid only for very gentle density
and a, [3]. The general solution of Eq. (3.20) is
gradients and in particular breaks down near the cutoff density where the
wavereflects. Hence it's important to complement the vVKB solution with
a more complete solution of the wave propagation. Fortunately, an exact
solution can readily be obtained for. a plasma with a linear variation in where a and (3 are constants which are determined by matching to the
density [2]. boundary conditions.
We again consider a plane e ectromagnetic wave normally incident / On physical grounds, we expect E to represent a standing wave for
onto an inhomogeneous plasma s .b and allow variations only in the z TJ < 0 and to decay as TJ -+ 00. Since Bi(TJ) -+ 00 as TJ -+ 00, we choose
,direction. The wave equation for the eld E(z) is (3 = O. A plot of Ai{TJ) is shown in Fig. 3.1. Note that the wavelength
and the amplitude of the field increases as the reflection point (TJ = 0) is
d2E w2
dz2 + 2"f.(w,z)E = O. approached, as expected from the WKB solution. Beyond the reflection
point, the field is attenuated.
Assuming that the plasma density is a linear function of position (n = The constant a is chosen by matching the electric field with the field
n..:.z/L, where ncr = mu}/47re2 is the critical density), we obtain of the incident light wave at the interface between the vacuum and the
plasma at z = 0 i.e., at TJ = - (wL/C)2/3. If we assume that wL/c » 1
2 2 and use the asymptotic representation,
d E
-+- w (
1-- Z )
E=O. (3.19)
dz2 c2 L

A.change of variables to TJ = (w2 / c2 L) 1/3 (z - L) gives

d2E
dTJ2 - TJ E O. (3.20) ')
-2..
 3 E-l'vl Wave Propagation in Pfasrnes 3.3 References 35

to the standing wave, we then estimate IEmax/EF~L2 ~ 3.5(wL/c)1/3, a

value in reasonable agreement with the exact solution. It's also interesting
that the phase shift between the incident and reflected waves is given by
the WKB solution if we simply subtract 7r /2 to account for reflection at
the critical density surface. In other words,
N re that we can express E(z = 0) as the SUlll of an incident wave with
°plitude EFS and a reflected wave with the same amplitude but shifted
aIIl .
in phase Le., IlJ = 2 ~
c
r
la
L
Vc dz _ ~ = 4 wL
2 3 c 2

E(z=O) EFS [1 + exp -i (~WcL - ~) ] , The magnetic field of the light wave is readily calculated from the
solution for E. Noting that the electric vector is in the x direction and
provided that the wave is propagating in the z direction, we take the y-component
, (WL) of Faraday's law to obtain
a = 2 Vir --z- 1/6 .
EFS e'" .
ic aE
B = ---
HereEFs is the free-space value of the electric field of the incident light w az .
waveand 'P is just a phase factor which does not affect IEI. Hence
Changing variables from z to T) and using Eq. (3.21) gives

(3.21)
(3.23)

As can be seen from Fig. 3.1, the amplitude of the electric field
where the prime denotes a derivative with respect to T). At the reflection
reachesa maximum value at T) = 1, which corresponds to (z - L) =
point, IB(T) = 0)1 ~ 0.92 (cjwL) 1/6 EFS. Note that B decreases as E
-(c2 L/w2)1/3. This maximum amplitude (Emax) is
swells, as qualitatively shown by the WKB solution.
WL) 1/3
3.6 ( - . (3.22) References
c

WeWouldexpect a factor of four increase in E2 because a standing wave is 1. Jackson, J. D., Classical Electrodynamics. Wiley, New York, 1962.
~t up. The additional swelling is due to the decrease in the group velocity 2. Ginzberg, V. 1., The Properties of Electromagnetic Waves in Plasmas.
o the light wave as the dielectric function becomes small, Pergamon, New York, 1964.
t~A similar swelling of the peak electric field amplitude can be ob- 3. Handbook of Mathematical Functions (Abramowitz, M. and L A. Ste-
If. eelby heuristic arguments based on WKB theory. Here we use k = gun, eds.). National Bureau of Standards, Applied Mathematics Series
co~w/c) and IEI = EFS/c1/4. As e becomes smaller, the wavelength be- 55, 1964.
Pro es l?nger. Intuitively we expect the wave to average over the plasma
" ~es within at least half of its local wavelength. Hence we expect
4<l.lf Urn value of e, which is roughly the value of c averaged over
!Jrofit lOcal wavelength near the reflection point. For a linear density
~~. e, <min ~ 7r/(2kminL). Since kmin = cm1/2n w/c, we then obtain
_~( 1

-c 7rc/2wL)2/3. Including the factor of 2 in amplitude which is due

 CHAPTER 4
Propagation of Obliquely
Incident Light Waves in
Inhomogeneous Plasmas

To complete our introduction to the propagation of light waves in inho-

mogeneous plasmas, let us now consider a light wave whose propagation
vector is at an angle to the density gradient. We again consider a plane
electromagnetic wave incident onto a plasma slab with electron density
ne(z). The vacuum-plasma interface is taken to be at z = 0, where the
e
angle of incidence is defined as the angle between the propagation vector
k and the direction of the density gradient (z). Without loss of generality,
we take the plane of incidence (defined by the vectors V'n and k) to be the
y - z plane, as shown in Fig. 4.1. With this choice, there are no variations
in the x-direction (i.e., k" = 0 and :" = 0). At the vacuum-plasma
interface, we note that ky = (w/c) sine and kz = (w/c) case. As we will
show, the wave propagation now depends on whether the electric vector
E of the incident light wave lies in or OtLt of the plane of incidence.

37
 4 Propagation in Inhomogeneous Plasmas 4.2 Resonance Absorption 39

Z
density lower than the critical density i.e., where ne = ncr cos2 e. Here
-----------------£=0 the critical density is defined as ncr = mw2 /41ie2 ~ ..
- ---------------£ = sin2e
Our previous analyses can be carried out with the straightforward
substitution E(Z) ~ E(Z) - sin ' B. For example, in a plasma with a linear
density profile, ne = ncrz/L, the wave reflects at z = Lcos2e, and the
Airy function pattern occurs relative to this point rather than at z = L.
~----------------y

x 4.2 OBLIQUELY INCIDENT P-POLARIZED

Figure 4.1 i\. sketch illustrating a light ray obliquely incident onto an inho- LIGHT WAVES - RESONANCE ABSORPTION
mogeneOUS plasma slab. If the electric vector of the light wave lies in the plane of incidence, the
light wave is termed p-polarized. In this case, there's a component of the
electric vector which oscillates electrons along the direction of the density
gradient i.e., E- 'Vne =1= O. Since this oscillation generates fluctuations in
4.1 OBLIQUELY INCIDENT S-POLARIZED charge density which can be resonantly enhanced by the plasma, the wave
LIGHT WAVES is no longer purely electromagnetic. Part of the energy of the incident light
wave is transferred to an electrostatic oscillation (electron plasma wave),
If the electric vector points out of the plane of incidence, the light wave a phenomenon termed resonance absorption [1,2J.
is termed s-polarized. If we then take E = E,);, the wave equation for We again consider a plane electromagnetic wave incident with an angle
E (Eq. 3.6) becomes e onto an inhomogeneous plasma slab with density ne(z), as shown in
Fig. 4.1. Now the electric vector is taken to be in the plane of incidence i.e.,
(4.1) E = Ey Y + Ez z. It is readily seen that the field acquires an electrostatic
component. Poisson's equation gives
Since the dielectric function (E) of the plasma is a function of z alone, ky
must be conserved. Hence ky = (w / c) sin B and 'V·(eE) = 0,

Ex = E(z) exp (w =: (4.2)

where E(Z) = 1 - w~e(z)/w2 is the dielectric function of the plasma. Since
'V. (EE) = e'V·E+'VE·E,weobtain

where B is the angle of incidence. Substituting Eq. (4.2) into Eq. (4.1)
n. E = _ ~ aE E
then gives v e az z·

2 2
d E(z) w ( ) Note the resonant response when E = 0, i.e., where wpe = w.
dT + -;;'2 E(Z) - sin
2
(} E(z) = O. (4.3)
The physical interpretation is straightforward. Oscillation of electrons
It is apparent that reflection of the light wave now occurs when between regions of differing density directly creates a charge density fluc-
tuation, On, which is
e(z) = sin2 () . (4.4)
On ne(X + Xos) - ne(x)
Since 10 = 1 - w~e(z)/w2, reflection takes place where the electron plasma Xos . 'Vne ,
frequency wpe = W cos (). An obliquely incident light wave reflects at a
 4.2 Resonance Absorption 41
40 4 Propagation in Inhomogeneous Plasmas

where :>Cos is the amplitude of oscillation of an electron in the electric field electric field of the light wave in free space. The decay of the field as it
of the light wave (xos = eE/mw2). Where W = wpe, this imposed charge penetrates beyond the turning point is estimated by ",=-!3, where
~
density fluctuation is at just the frequency at which the plasma resonantly
responds. Hence an electron plasma wave is excited where I: = 0; that is, (3 = jL
L cos? e c
~ / w~e _ w2 cos? fJ dz .
at the critical density.
Even though an obliquely incident light wave reflects at a density less
For a linear density profile, f3 = (2wL/3c) sin:' B. Hence, we obtain
than the critical density, its fields still tunnel into the critical density
region and excite the resonance. To determine the energy transfer to the
excited plasma wave, we need to determine the size of the electric field B(z=L) ~ 0.9EFS
C)
( -,-,-,'L
1/6
exp
(-2wL sirr' B) (4.8)
3c
along the density gradient near the critical density. In order to evaluate
Ez, it is most convenient to work in terms of the magnetic field of the The important physical features of resonance absorption can be de-
p-polarized wave. Noting that B = xB" and using the conservation of duced from our approximate treatment for Ed [3]. Using Eq. (4.8) and
ky = (w / c) sin {},we express defining a new variable 7 = (WL/C)1/3 sin B, we obtain

iw y SinB) E EFS -I.() (4.9)

B = x B (z) exp ( -i wt + C . (4.5) d = J27r wL/c <p 7 ,

The electric field is related to the magnetic field by substituting Eq. (4.5) where ifJ(7) ~ 2.37 exp( -273/3). The driver field vanishes as 7 -+ 0, since
into Ampere's law: the component of the electric vector of the incident light wave along the
2W densi ty gradient varies as sin B. Likewise, the driver field becomes very
VxB = --I:E. (4.6)
c small for large 7, since the incident wave then has to tunnel through too
The z component of Eq. (4.6) then gives large a distance to reach the critical density surface. Between these two
limits, there is an optimium angle of incidence given approximately by
sinB B(z) (wL/C)1/3 sinB ~ 0.8.
E; = €(z) (4.7)
In Fig. 4.2 we compare our simple estimate for ifJ( 7) with the result
obtained by Denisov [2] by numerically solving the wave equation. Note
Since E, is 'strongly peaked t the critic density, we approximate the that our heuristic estimate is in reasonable agreement with the detailed
resonantly driven field as Ed/I: ), where d is evaluated at the resonance calculation. As expected, our heuristic solution is quite accurate for 7 « 1,
point. Physically, Ed is simply t onent of the electric field of the since our estimate for B (z = L) becomes exact as 7 -+ O. Our expression
light wave which oscillates electrons along the density gradient at the for ifJ(7) is qualitatively correct even for 7 ~ 1, since the dominant phys-
critical density i.e., the field driving the resonance. ical effect is then the attenuation of the incident field as it tunnels from
To evaluate Ed, we need to calculate the magnetic field at the critical the cut-off density to the critical density.
density. For our purposes, it suffices to simply estimate the value of the Having related the electrostatic field near the critical density to the
magnetic field using the insight obtained from our previous calculations of electric field and the angle of incidence of the light wave, we can now
wave propagation in inhomogeneous plasmas, Assuming a linear density calculate the energy absorption. As shown in Eq. (4.7), the resonantly-
profile (ne = ner z/ L ), we represent B (z = L) as its value at the turning driven field is E; = Ed/I:(Z). If we include a small damping of the wave
point B(;:; = L cos2 B) multiplied by an exponential decay from the turn- with frequency u , €(z) = 1 - w;e(z)/w(w + iv), as we will show in
ing point to the critical density. The value of B at the turning point is the next chapter. Hence, E~ has a resonance behavior near z = L , i.e.,
estimated using the Airy function solution for an s-polarized wave, which the maximum value of E~ is proportional to v-I and the width of the
gives B(z=Lcos28) ~ 0.9EFS (c/wL)1/6. Here EFS is the value of the
 4 Propagation in Inhomogeneous Pla;;mas 4" References 43

1.2 The integral is easily evaluated to give 11w 1I/. Hence

-- .p(T) Ginzburg

\ - - Simple estimate wL E"£1

1.0 ___
(4.13)
\ 8
\
0.8
By conservation of energy, labs = fA C Ets/811, where fA is the frac-
\ tional absorption of the incident light wave due to the excitation of an
>-
~ 0.6 \ electrostatic wave at the critical density. Substituting for Ed( Z = L) from
\ Eq. (4.9), we then obtain fA -::= cP2(T)/2, where cP(T) is the characteristic
0.4 \ resonance function describing the strength of the excitation as a function
\ of the angle of incidence and the scale length of the density gradient.
0.2
\ Our simple model emphasizes the physics of resonance absorption
\ and captures its basic features. For a linear density profile, the fractional
OL- L- ~ ~ absorption peaks at 8max ~ sin -1 [0.8( cfco]: )1/3] and is sizeable for a range
o 2 3 of angles of incidence 68 ~ 8max. The peak absorption is somewhat over-
estimated. Detailed numerical calculations for a linear density profile show
Figure 4.2 A plot of the function cP( T), which characterises the efficiency of a peak resonance absorption of about 0.5 [4-6].
resonanceabsorption. The solid line is from Ginzberq (1964).

References
resonant region is proportional to u . This feature of Ez will enable us
to compute the energy absorbed via excitation of the electrostatic wave 1. Ginzberg, V. L., The Properties of Electromagnetic Waves in Plasma.
without specifying the detailed value of v (assuming vi w « 1). Hence, Pergamon, New York, 1964.
11 can represent dissipation by electron-ion collisions, linear or nonlinear 2. Denisov, N. G., On a singularity of the field of an electromagnetic wave prop-
wave-particleinteractions, or even propagation of the wave out of the agated in an inhomogeneous plasma, Sov. Phys.- JETP 4,544 (1957).
resonantregion. 3. Kruer, W. L., Theory and simulation of laser plasma coupling; in Laser
The absorbed energy flux (Jabs) is Plasma Interactions, (R. A. Cairns and J. J. Sanderson, eds.). SUSSP
Publications, Edinburgh, 1980.

labs -
.
- 1= 0
V
E;
8"
-
dz -
s:
811
1=
0
dz E~(z)
1 12
e
(4.10)
4. Forslund, D., J. Kindel, K. Lee, E. Lindman and R. 1. Morse, Theory and
simulation of resonant absorption in a hot plasma, Phys. Rev. All, 679
(1975).
For·a linear density profile (ne = nee Z 1L), we have 5. Estabrook, K. G., E. J. Valeo, and W. L. Kruer, Two-dimensional relativistic

1(12 = (1 - ir (~r~:., + (4.11)

~~bstitutingEq. (4.11) into Eq. (4.10) and approximating
er the narrow width of the resonance function gives
simulations of resonance absorption, Phys. Fluids 18, 1151 (1975).
6. Speziale, T. and P. J. Catto, Linear wave conversion in an unmagnetized
collisionless plasma, Phys. Fluids 20, 990 (1977).
e; as constant

v Ej(.:=L) dz
(4.12)
811 (1 - z1L)2 + (vlw)2 .
 CHAPTER

Collisional Absorption
Electromagnetic Wa,
in Plasm

~) So far we have focused on collisionless plasma behavior. As disc

Chapter 1, collisional effects due to discrete particle encounters
.
-, systematically neglected as the number of particles in a Debye s
(ND) becomes very large. Even when ND is large, there are, of
some collisions, which can be iteratively included by going to first
in the expansion parameter l/ND. A collision term is then added
Vlasov equation, which becomes

et, +
_J
af- +
v._J
q-
_J
(
E + ~). afj = L (aIik)
&t Ox mj c &v k ate

where (a fjk/ at)c represents the rate of change of fJ due to collisions

the kth charge species.
The contribution of collisions to the moment equations is straig
ward. If we neglect ionization and recombination, collisions do not
the number of charges of each species. Hence f dv Lk (a fjk/ at)e
 tl

5 Collisional

_olltinuity equation is unchanged. Noting that collisions between

Absorption of E-M Waves 5.1 Collisional Damping of Light Waves

The analysis is straightforward. If we use Eqs. (5.1) and (5.2), the

41

I
I

e of the same species lead to no net change of momentum, we have lincarized force equation for the electron fluid becomes
~

(5.3)

Since the field varies harmonically with time,

(&njuj/&th denotes the rate of change in the momentum of the ph
-i e E
~es due to collisions with charge species k. Hence the force equation Ue =
m(w + i LIe,)
~es

11] (
[}&tUj + U
J
. ~u)
[Jx J
=
The plasma current density is then

(5.1)
n ('E + _UjX_B)
Jm
_qj
c
__ 1
m8x
_[JPj - '"
~
(~nu.) .
&tJJ
J ,J k#j k
where wpe is the electron plasma frequency. Note that the plasma con-
ductivity (J (J = (J E) is now complex: (J = i w;e /[41T(W + i Veil]·
Faraday's and Ampere's laws become
5.1 COLLISIONAL DAMPING OF LIG T WAVES
tW (5.4)
To investigatecollisional damping, we ag consid the linearized plasma V'xE
, -B
iwt -, c
responseto a high-frequency field of the form E(x) e- . Treating the
41T tW E tW
iODswith charge state Z as a fixed, neutralizing background with density V'xB - (JE - - EE, (5.5)
c c c
ne(z)/Z, we need only treat the dynamics of the electron fluid. For an
electronfluid with density ne and velocity lie interacting with a stationary where the dielectric function of the plasma is now
ionfluid, it is convenient to express
2
E = 1 _ Wpe (5.6)
(5.2) W(W + i Llei)

The wave equation for E is obtained by taking the curl of Eq. (5.4) and
"'hherel1ei is a collision frequency which describes electron scattering by
substituting for V' x B from Eq. (5.5):
t e ions.
<lis~his collision frequency l1ei depends on an average over the velocity
(5.7)
thtnbution of the electrons. Indeed, the form of the average depends on
e physical process under consideration, and so a more detailed treat-
~en.tis needed in order to derive the numerical value of Llei· Fortuitously, Let us first derive the dispersion relation for light waves in a spatially
~ra Maxwellian distribution of electron velocities, the electron-ion colli- uniform plasma. Taking E(x) ~ eik-x and substituting for E into Eq. (5.7)
SIonfrequency which describes the damping of a high-frequency wave is grves
~ntially the same as the characteristic electron-ion collision frequency (5.8)
ch.we estimated in Eq. (1.2). A derivation of this result will be pre-
sentedin Section 5.4.
 5 Collisional Absorption of E-M Waves 5.2 Collisional Damping 49

we have assumed that vedw « 1. The light waves are now damped. where we have taken E = E x. If we assume a linear density profile
w~
h . . .
Expressing W = Wr - IV /2, where V IS the energy damping rate, we (ne = ner z/ L) and neglect the dependence of Vei ~'ihe plasma density,

+ e
obtain 2
( Wpe - c2) 1/2 E 1 -
Z
(5.11)
2 (5.9) L (1 + iv;Jw)
Wpe
V w2 vei·
r where Vei is approximated by its value (v~) at the critical density. This is
The collisional damping has a simple physical interpretation. The rate a reasonable first approximation, since most of the collisional absorption
of energy loss from the light wave ( V E2 /8 7r ) must balance the rate at occurs near the critical density where the electron-ion collision frequency
which the oscillatory energy of the electrons is randomized by the electron- maximizes and the group velocity of the light wave minimizes. Note that
ion scattering with a frequency Vei (Vei no mV~s/2). Since Vos = eE/mwr, we are also assuming that the plasma with a density less than or equal to
this power balance zi
ance gives v = vei Wpe
2/2 Wr· the critical density is isothermal. Substituting Eq. (5.11) into Eq. (5.10)
It is instructive to also consider a spatial problem i.e., let W be real gives
and k be complex. Then, substituting k = k; + i kd2 into Eq. (5.8) and
assuming that ki « kr, we obtain

1
-
C
J W2
By changing variables, we again obtain Airy's equation
2
Wpe vei
k, =
w2 Vg (5.12)
where ki is rate at which the energy decays in space and "s is the group
1
velocity of the light wave. Note that the energy damping length (ki ) is where 1)is now a complex variable:
simply Vg/v, where v is given in Eq. (5.9).
1/3
~ i~
= z L 1+~ (5.13)
5.2 COLLISIONAL DAMPING OF A LIGHT WAVE
1)
[ o'L(1+~) ] [ (w)]
IN AN INHOMOGENEOUS PLASMA
'Let us now investigate collisional damping of a light wave propagating into As discussed in Chapter 3, the solution which satisfies physically rea-
an inhomogeneous plasma. First, we will neglect the density dependence of sonable boundary conditions is
the collision frequency and compute the absorption of a normally incident
wave both from an analytic solution for a plasma with a linear density (5.14)
profile and from a WKB treatment. Then we will use the WKB·treatment
to allow for the density dependence of the collision frequency and for where Ai(1)) is a well-documented Airy function. The constant et is chosen
by matching to the incident light wave at the vacuum-plasma interface at
oblique incidence.
We start by again considering a plane wave propagating in the z z = O. For 1111 » 1, we can evaluate Eh) using the asymptotic represen-
direction into a plasma slab with electron density ne (z). Since there is tation for Ai (1)) i.e.,
only variation in the z direction, Eq. (5.7) becomes
d2E
dz2
w2
+ -;} E(Z) E = 0, (5.10)
Ai( -1))
,fii
1
1)1/4
cos (~3 1)3/2 - ~4)-
 5 Collisional Absorption of E-M Waves 5.3 Collisional Absorption 51

_ zz: 0, E can be represented as an incident plus a reflected wave 5.3 COLLISIONAL ABSORPTION INCLUDING
~eJl~e :;pliwde is multiplied by the quantity ei<l>, where OBLIQUE INCIDENCE AND A DENSITY
",b~ DEPENDENT COLLISION FREQUENCY
<1 [ ] 3/2 7r
4> = :3 -1](Z=O) - 2 . Using WKB theory, let's now extend our calculations of collisional absorp-
tion to include obliquely-incident light waves and to allow for the depen-
is now complex, there is both a phase shift and a damping of the dence of the collision frequency on density. For definiteness, we consider
SiU,e f} 2/3
ted wave. At Z = 0, 1] = - [(wL/c)(1 + iv;;/w)]
. For v;;/w « 1, s-polarized plane waves incident onto a plasma slab with electron density
eC
rdl base shift (the real part of 4» is the same as in the collisionless ne(z). As discussed in Chapter 4, the local dispersion relation then is
tbe;ation i.e., 4>real = (4wL/3c) - 7r/2. The imaginary. part o~ 4> is
caI, £/3c, which means that the reflected wave IS decreased ill amplitude
(5.16)
4V~xp(_4V;iL/3C), or in energy by exp( - 8v;;L/3c). Hence the fractional
by
absOrption fA d ue to eollis:
sion ald ampmg
.. IS
where e' is the angle of incidence. Substituting Eq, (s.n) into Eq. (5.16)

fA = 8 -;
1 - exp ( - ~
L) . (5.15)
gIves
w 2 2
wpe ]1/2
kz = - cos 8 - . (5.17)
Let's now calculate the collisional absorption in a plasma with a linear c [ w(w + iVe;)
densityprofile using WKB theory. Here
The density dependence of Vei is easily included. Referring to Eq. (1.2)
of Chapter 1, we neglect any weak dependence on density introduced by
In A and note that the collision frequency is then simply proportional to
the plasma density. Hence we will approximate Vei = -; ne/ncr, where
where -v;i is again the collision frequency evaluated at the critical density.
k(Z') = ~ [ c(Z') ] 1/2 In the WKB approximation, the wave energy decays by e-2c5, where

The energy of the wave decreases by the factor e-2c5, where s = 2 ~m 1Lt kz(Z/) dz' ,

s = i ~ 1L ~m ( EI/2) dz' . and Lt is the turning point. If we assume a linear density profile and use

r:
Eq. (5.17), we obtain
The sYlnbol~m denotes the imaginary part, and the factor of two enters
~~e the wave is absorbed as it propagates both into and out of the ? Z1 ( 'lVeiZ
•• ')] 1/2 I

Illa. SUbstituting for E from Eq. (5.11) gives

{j - - m
[ cos" 8 - L 1 - ~ L dz ,
c 0

wh e we have assumed that v;;/w « 1. This standard integral gives

2 W r [ z' ]1/2 dz'. (j - (16v;;L/15c) cos58. Hence the fractional absorption is
s = C ~m lo 1 - L (1 + ~)
32 v: L )
~Qr fA = 1 - exp ( - e. cos5 8 . (5.18)
15c
cl... V;;/ W
«1 (j = 4 v· L/3c. Hence the energy of the light wave
"<tYs b ' e.
th~ an Y.the factor exp( -8 v;i L/3c), the same result as that given by For normal incidence (8 = 0), the density-dependence of Vei has re-
a.IYtIC solution. duced the coefficient in the exponent of Eq. (5.15) from 8/3 to 32/15
 5.4 Derivation of the Da.mping Coefficient
5 Collisional Absorption of E-M Waves

. by abont 20%. Note also that the absorption is a sensitive function

I·;-~he angle of incidence. Since an obliquely incident wave reflects at a
~owerdensity, less collisional plasma is traversed. Finally, we note that the E "'-
collisional absorption will depend in detail on the density profile of the I"
plasma. For example, for an exponential profile (ne = ner exp( - z j L)), e~ I
I
8 tJ;iL
f A = 1 - exp ( - ~ cos () .
3) I
I
I
I

5.4 D~RIVATION OF THE DAMPING COEFFICIENT

As a final topic, let us give a derivation of the collisional damping rate
of a light wave. To determine, the electron-ion collision frequency heuris-
tically introduced in Eq. (5.2), we need to start with a representation
for (afe;/at)c. The simplest description starts with the Fokker-Planck
equation Figure 5.1 A sketch of the coordinate system.

i.e., E = Eo coswt. If we substitute Eq. (5.19) into Eq. (s.i), the ki:
equation for the electrons becomes

where (6v) describes the slowing down of electrons due to electron-ion

encounters and (6v6v) their diffusion in velocity space. These coeffi-
a fe _
at m
s.
E- a fe _
&v - A &v
!-. . [ v ! v3- v v
2
a t,
. &v
1 ...L.
,Cee(fe) .
cients can be derived by computing the changes in velocity of electrons
streaming past ions and summing over encounters i.e., by a more detailed where fe is the electron distribution function and Cee(fe) denotes a .
treatment of the approach used in Chapter 1 to estimate the 90° collision but more complex operator describing electron-electron collisions.
frequency. With this motivation, we will simply give the standard result, electron-electron collisions are important for determining the form
zero-order distribution function but can be neglected otherwise.
For low-intensity light, we decompose the distribution functi
(5.19) a zero-order part which depends only on the absolute value of v ]
perturbation driven by the field i.e., f(v) = fo(v) + ft(v) cosO, i.
is the angle between v and E as shown in Fig. 5.1. The linearizedf
Here A = (271" ne Z e4jm2) In A where ne is the electron density, Z is( equation then becomes
the ion charge state and A is the ratio of the maximum and minimum ... : " .
aft e E afo 2A
)mpact parameters as discussed in Chapter 1. A detailed discussion of the
at -;;: a;; = - -;J I, .
derivation oftq. (5.19) is given in Chapter 7 of Shkarofsky et al., 1966.
To calculate the high-frequency resistivity, we consider a plasma with Here we have used the fah that the collision operator vanishes E
a uniform electron density ne and a fixed, neutralizing background of ions. function of Ivl = 11 and .
The electric field of the light wave is treated in the dipole approximation a v2 I - vv a 2A
A-· av It (v) cos o - 3"" fl (v) cos8
av v3 11
 5.4 Derivation of the Damping Coefficient 55
5 Collisional Absorption of E-M Waves

where v; is the electron thermal velocity. A modification of the usual

-G' e-,..;t, the driven solution of Eq. (5.21) is
S~J:.I() expression for In A should be noted. The uiaximum .impact parameter is
now ue/w rather than the electron Debye length .'"-Since 1/ = V el. w2pe /w2 ,
( )w+--
hu----
_ ieEo alo [ i2A]-1 (5.22) Eq. (5.26) determines the electron-ion collision frequency which describes
m OV v3
the damping of a high frequency wave in a plasma. If the density is ex-
erturbed current density is JI -e h(v) cos(j vdv, and J pressed in cm-3 and the electron temperature in ev,
Tbe P rate of the absorption of energy by the plasma is
dlta¥efa ge
(JI.E) = ~e o h(v)v [-eE
2
COS (jdV] . J(5.23)
(5.27)

However, if Z (vos/ Ve) 2 » 1, electron-electron collisions cannot equi-

'fhesywbol~e denotes the real part. Substituting for I,(v) from Eq. (5.22) librate the distribution function sufficiently rapidly. The form of the dis-
,ad integrating over angles grves tribution function becomes determined by the collisional heating [4,5]. In
. 47f 2 (XJ alo this limit, we return to Eq. (5.20) and balance olo/at with the heating·
(JI·E> = -"3Ane~vos lo dVTvg(v). (5.24) term, .
010 = / e E . a It )
Here/o = Helo, Vos = and g(v) = [1 + (2A/v3w)2]-I.
(eEo/mw), Note at \ m ov '
that (2A/v3w) ~ Vei(V)/W, where Vei(V) is a characteristic collision fre- where the brackets denote an average over angles. Noting that JI
quency for electrons with velocity v. Since Vei(V)/W «1 for all but a It (v) cos (}and averaging over angles gives
small class of electrons, we approximate g( v) ~ 1 in the integral, giving
(>0 alo _ OlD _
7ft - ~
eE 1
3v2 Qv
0 (
v
2
JI(v)
)
(5.28)
lo dv OV g(v) = - 10(0) .

Substituting for JI(v) from Eq. (5.22), approximating g(v) ~ 1, and look-
FiDally, we invoke energy balance to equate (JI ·E) with the rate of energy
ing for a self-similar solution, we find
~ from the field, which is v E2 /8 7r. If we use Eq. (5.24), the rate v at
which energy is damped by electron-ion collisions becomes
(5.29)
- 2
. wpe 87f -
V = ~ "3 A 10(0) , (5.25)

C t)
where
where .
. T wpe IS the electron plasma frequency. 'U = A3v~s 1/5 .
Tb f,he damping rate depends on the zero-order distribution function.
el: t
orm of tI:is distribution function in turn depends on whether electron-
fastefOn collisions (with frequency vee) can equilibrate the distribution
Hence the self-consistent distribution function is super-Gaussian in this
limit. Since this distribution has fewer particles than a Maxwellian near
(i.e.rifthan electron-ion collisions cause it to heat. If Vee v~ » Vei v~s v = 0, the collisional damping rate is reduced by a factor of about 2.
£Vai ~Zv~/v~) « 1 ), the distribution function remains Maxwellian.
SllltUatlon of Eq, (5.25) for a Maxwellian distribution then gives the re-
Ill;Uallyquoted in the literature

W~e 1 Z W~e
v --3 In A, (5.26)
W
2 3(27f) 3/2 nVe
56 5 Cotlisional Absorption of E-M Waves

CHAPTER
References
s,
1. Shkarofsky. 1. P., T. W. Johnston, and M. P. Backynski, The Particle Ki-
netics of Plasmas. Addison-Wesley, Mass., 1966.
2. Dawson, J. M. and C. Oberman, High frequency conductivity and the emis-
Parametric Excitation of
sion and absorption coefficients of a fully ionized plasma, Phys. Fluids 5,
517 (1962).
Electron and Ion Wav€~
3. Johnston, T. W. and J. M. Dawson, Correct values for high-frequency power
absorption by inverse bremsstrahlung in plasmas, Phys. Fluids 16, 722
(1973).
4. Langdon, A. B., Nonlinear inverse bremsstrahlung and heated-electron dis-
tributions, Phys. Rev. Letters 44,575 (1980).
5. Jones, R. D. and K. Lee, Kinetic theory, transport, and hydrodynamics of
a high-Z plasma in the presence of an intense field, Phys. Fluids 25, 2307
(1982).

As we have seen in our consideration of resonance absorption, the 0 '

lation of electrons in the direction of a spatial variation in the plasn
density drives charge density fluctuations. When the frequency of the (l
ciliation is near the electron plasma frequency, an electron plasma wave>
resonantly excited. In the case of resonance absorption, the spatial vars
tion was due to the density gradient produced by plasma expansion i.nt
a vacuum. However, the spatial variation in the density can also be d'l
to ion density fluctuations associated with ion waves.
We will first discuss this coupling of a light wave into an elect
. (llasma wave by an ion density fluctuation. Then we will show that tit
c upling can lead to unstable growth of both electron and ion waves, W
, a count is taken o~ the generation ~f ion ?ensity fluct~ations. ~a t~e
: called ponderomotive force. A physical picture of the instability will]
given, followed by a derivation from the two-fluid model.
 6.1 Coupling Via Ion Density Fluctuations 59
6 Parametric Excitation of Electron and Ion Waves

Equation (6.3) describes the excitation of an electron plasma wave by

pLING VIA ION DENSITY FLUCTUATIONS interaction of the pump field (light wave) with an ion density fluctuation.
CO . The driven solution to Eq. (6.3) is straightfo~ard. Noting that the
. ,,,te the coupling of a light wave into an electron plasma w~ve by
tlg.. fluctuation, a simple, one-dimenslOnal model IS sufficient. pump field varies as exp( -iwt), we obtain
dCIl:>ltyI the lig"ht wave as a spat!'all y homogeneous OSCla ill tiing
e nW cle b f 2
6n Eo cos kx
Vi 'el : Eo,,== xEo exp( -iwt) .. In other words, the wave num er ~ wpe
d E (6.4)
c fi I'Snezlecced on the assumption that It IS much less than k ; 'Z? -;; c(k,w)
J:••ht wave 0 . .
.~ ber of the fluctuation in ion density. Since the frequency of
ve own 'd wave, we escriib e th e
._much less than the frequency of a light where
~~ ....'- w2 + 3k2v2e + t I/ei
jGIl.. fluctuation as a static modulation m the plasma density, n - c(k,w) = 1 _ pe
denSItY . kx where no is the average density .' and 6n IS t he amp litu de 2 w w
An co:> , . hi .
density fluctuation. Finally, we t~eat the lOns.as fixed on t s time Since energy is coupled into the driven wave, the pump field is damped.
and describe the electrons as a fluid With density ne, mean velocity The energy damping rate, 1/", is determined by balancing the rate of
aDdpressure Pe· . energy lost (1/" E5/87r) with the rate of energy absorption via the driven
To derive an equation for the high freque~cy. electron. density flue- wave i.e.,
_ .••:n - we take a time derivative of the continuity equatlOn, a spatial
_lIS, 1/' E5 = ~ r271
'/k dx
deri1ativeof the force equation, and combine to obtain 87r 27r io
(6.1)
By substituting for E from Eq. (6.4) and noting that ~m c = I/ei/W, we
then obtain [1,2J
1/" 1 (6n)2 ~mc
where we have included collisions with frequency I/ei· vVe next linearize
~ = 2 ner Ic(k,w)12'
thisequation,i.e., let ne = no+6.n cos kx+n, E = Eo+E and Ue = uo+u,
where the tilde denotes a small perturbation and uo is the oscillation Here ner is the critical density determined by the condition = w2. w;e
velocityof electrons in the field Eo. If we treat n « 6.n « no, Eq. (6:1) We note that the electric field becomes very large when c(k,w) -::: 0
becomes i.e., when the pump field resonantly couples to an electron plasma wave.
Of course, our linearized analysis fails if c(k,w) becomes too small, and
e aE 2'a2n + I/ei-
an eEo·.
= - -- 6n ksmkx. (6.2) nonlinear effects must then be considered. One such nonlinear effect is
-no- 3ve- at
m a x ax2
m repeated mode coupling. When IAn/nerl > 1E(k,w) I, the driven wave
We have used an adiabatic equation of state, assuming w/k » Ve, the
el~ron thermal velocity. Then by substituting the Poisson equation
(as/f)-x::: - 47r en) into Eq, (6.2), we obtain

~a~ [{)2-- E
+
. _
w2 E _ 3v2--
, a2 E + ei: 1 47r e
2
= --E06nk
. _
smkx,
n' becomes as large as the pump. It in turn acts like a. pump to drive a
wave at 2k, which in turn can beat with the ion density fluctuation to
drive a plasma wave at 3k, and so on. A spectrum of driven waves is
obtained; the maximum wave number can be estimated by the condition
1E(Nk,w) I'" 6.n/ner. Note that even a modest density fluctuation can
efficiently couple a long wavelength plasma wave into shorter wavelength
&t2 pe e ax2
1/'-
e, at m
ones .

.) 6.n k
- w- - Eo cos -x . (6.3)
pe no
60 6 Parametr-ic Excitation of Electron and Ion Waves 6.3 Instabilities - A Physical Picture

6.2 THE PONDEROMOTIVE FORCE Observe that the electrons experience a force which pushes them <\\\'3'
The coupling of a light wave into electron plasma waves by density fluctu- from regions of high field pressure. This ponderometlve force (Fp) is pre
ations is a very basic phenomenon, which emphasizes that electromagnetic portional to the gradient of the electric field pressure i.e.,
and electrostatic waves are inherently coupled in a turbulent plasma. If
sizeable levels of ion fluctuations exist in the plasma, this coupling can
clearly be very significant. In fact, sizeable ion density fluctuations can
be self-consistently produced in the light-plasma interaction, since an ex- In a uniform plasma with density n, the ponderomotive force densi
cited plasma wave beats with the light wave to generate variations in fp can be expressed as fp = -'V(nm(u~)/2). In other words, the tim
electric field pressure. This gradient in field pressure gives rise to a force averaged energy density of motion in the electric field plays the same ro:
(the so-called ponderomotive force), which acts to generate ion density as the ordinary pressure, which represents the random or thermal energ
fluctuations. density.
To introduce the ponderomotive force, we consider the response of a
homogeneous plasma to a high frequency field whose amplitude is spatially
dependent i.e., E = E(x)sinwt, where w .:<: Wpe » Wpi. We treat the 6.3 INSTABILITIES - A PHYSICAL PICTURE
electrons as a fluid and compute their response to order E2. If we neglect
We can now easily see that a light wave can excite an instability in whir.
the electron pressure, the force equation is
both ion waves and electron plasma waves grow. An ion density fluctr
ation couples the light wave into an electron plasma wave. In turn, tn
GhIe e (
at + Ue' 'Vue = - m E x) sinwt . (6.6) electron plasma wave beats with the light wave to generate a spatial
at ion in the electric field intensity, which can enhance the ion densit
To lowest order in IEI, Ue = u" where fluctuation via the ponderomotive force. Hence, there's a feed-back loq
and instabi can re t .
au" e . A phy ical picture f the instability can be given [3J. Consider .
at = - m E(x) sin wt, a~Ton deiiSiYy'fluct-ation' in an otherwise homogeneous plasma i.
(6.7) n ;;no+- 6.n(;'osk;;. The '~fectric field of the light wave is again appron
eE
u" - m.w
cos wt. mated as a spatially homogeneous field of the form Ed = Eo sin wt. TI
electrostatic field E associated with the excited plasma wave is then giYe
The electrons are simply oscillating in the local electric field. By av- by Eq, (6.4). Explicitly including the time dependence and neglecting CG
eraging the force equation over these rapid oscillations, we obtain lisions, we then have
2
wpe 6.n.
(6.8) E= --:-..!.:..::....,::- -- Eo cos kx sin wt ,
w2 - w2ek no
where ( )t denotes an average over high frequency oscillation and u" = where w;k = w~e + 3k2v~. Since E has a spatial dependence, the tim
(ueh, E3 = (Ek Substituting for u"
from Eq. (6.7), we obtain
averaged electric field intensity has a gradient. To lowest order in t\.
small amplitude 6.n of the (thermal) ion density fluctuation, we obtain
(6.9)
 6 Parametric Excitation of Electron and Ion Waves
6.4 Instability Analysis 63

plasma frequency. We further consider only electrostatic perturbatious

f~
p /-
and restrict the analysis to one-dimensional perturbations along the di-
I \ rection of the pump field.
I
I The fluid equations for the electrons are
x x
Bn; 8neue
at + ~ o (6.11 )
8ue 8ue -!..E 1 8pe
---- (6.12)
at + Ue 8x m mne 8x
Pe
constant, (6.13)
n~
Figure 6.1 When Wo < Web the ponderomotive force acts to enhance the f-7~u...!'~ ') O".....
<1P;.r; L..] I(.-·,...o-:Lrv-/'..........)

density fluctuation, i.e., to push more plasma into regions of higher density. where v = 3 for high frequency perturbations and y = 1for low frequency
ones. Note that we have included collisions with frequency Ve to model'
, either collisional or Landau damping of the electron waves. To proceed, we
The ponderomotive force Fp is then divide the electron fluctuations into low and high frequency components

no + nee + neh
(6.14)
Vos + Uee + ueh ,

As shown in Fig. 6.1, the ponderomotive force acts to reduce the where the superscripts £ and h denote low and high frequency, respectively,
density fluctuation when W > ~ek' However, when W < Wek, the pon- no is the uniform background density, and Vos is the oscillation velocity
deromotiveforce acts to enhance the density fluctuation. Hence a purely of electrons in the pump field. To analyze for instability, we linearize by
growing(zero frequency) ion density fluctuation will spontaneously grow assuming that n~ or n~ « no and u~ or u~ « Vos and neglecting products
from the noise. As it grows in amplitude, so also does the associated of the perturbed quantities.
electronplasma wave. This instability is called the oscillating two stream. For spatially dependent electrostatic fluctuations, we can write
We next consider the ion density fluctuation associated with an ion
acoUsticwave. The fluctuation is no longer static, but has a frequency 8E
~ to kvs, where Vs is the ion sound velocity. In this case, the instability
&t + 47r J = 0, (6.15)

IS Illest easily thought of as the resonant decay of the light wave into
where J is the longitudinal part of the current density. Equation (6.15) is
an electron plasma wave plus an ion acoustic wave. The instability is
readily obtained from Poisson's equation and the equation for continuity
strongestwhen all three waves are. matched in frequency i.e., when W =
of charge. Linearizing and taking the high frequency component gives
We/c + kv •. Hence this;~tabilitylis often called the l;;nacousti~-d~c~Y-'-1
- I --.--
(6.16)
~4 INSTABILITY ANALYSIS
tioeCan derive these instabilities [4-9] from the two-fluid plasma descrip- IIIturn, the high frequency component of the linearized force equation
~ For simplicity, we consider a spatially uniform plasma driven by a becomes
P field of the form Ed = Eo coswot, where Wo is near Wpe, the electron 8u~ 3v; 8n~
Vos -- --- (6.17)
8x no 8x
 1

6.4 Instability Analysis 65

64 6 Parametric Excitation of Electron and Ion Waves

(Z m « I'd) and use an isothermal equation of state. the low frr-quency

where we have used pZ= 3mv;nZ·Taking a time derivative of Eq, 6.17
and substituting from Eq. (6.16) gives
component of the force equation becomes
~-
a 2 e
3v; a nZ
---
2 ~ El = _ -(vosuZ)
m ax
_ ve ane .
no ax (6.20)
no &tax
(6.18) In other words. the low frequency electric field transmits the ponderorno-
2
2 h: 471" e l
- wpe ue - --;;:;- ne Vos . tive force and the electron pressure to the ions.
The fluid equations for the ions with mass M and charge Z e are
The third term in Eq. (6.18) can be neglected relative to the other an a
terms. First note that &t' + ax (niUi) 0 (6.21)

a ( au~) ~ avos au; aUi

,:u + Ui
Bu,
ax
Z eE _ _1_ api
- Vi1.Li (6.22)
&t Vos ax - &t ax' u(. M NfnOi ax
Pi
- = constant, (6.23)
since the low frequency is assumed to be « Wo ~ Wpe. We next use the n3 ,
low frequency component of the continuity equation to give
where we have included collisions with frequency Vi to model either colli-
sional or Landau damping of the ion fluctuations. Neglecting the response
of the massive ions to the high frequency fields, we linearize these equa-
tions by assuming
Direct comparison shows that these terms are neglegible compared with ni nOi + ne,
the other terms of Eq. (6.18), provided that k2 « v;' w;." Ui
i
Ui
Lastly, we use the high frequency component of the continuity equa-
The linearized ntinuity and force equations become
tion to simplify the thermal correction term in Eq. (6.18). In particular,
ani aui
anZ auZ an~ -' + nOi -' = 0
0, &t ax
&t + no ax + Vas ax
auf Z eEl 3 v; an; Vi uI .
which gives &t M nOi ax
We next take a/at of Eq, (6.25), a/ax of Eq. (6.26), and eliminate the
---a nZ
2
3v; common term a2uU ax&t to obtain
no &tax

The second term on the right hand side can be neglected compared to
(471"e2/m)n~vos, provided k2>.te« 1. Hence we finally obtain
Substituting for E from Eq. (6.20) and noting that Z nf :::: n~ then gives
(6.19)

To obtain an equation for the low frequency fluctuations, we must

consider both the electron and ion responses. If we neglect electron inertia
 6 Par-arnet r ic Excitation of Electron and Ion Waves
6.5 Dispersion Relation 67

DISPERSION RELATION
6.5 Substituting Eq. (6.32) into (6.31) and rearranging, we obtain
cOllpJedequations for u~ and n~ describe the feedback which leads
1'b~.tability. To derive the dispersion relation, we assume that u~ and
lIlt> ( w 2. + uav, - k
2v;?) [( w.+ iVe)2
2 - /i 2] + Wpi-4-wo
2 '1::2v~ss = O. (6.33)
t~ ,'arY as exp(ikx - -iwt) and represent vos(t) as
lit
Both the oscillating two stream and' ion-acoustic decay instability
exP('iwot) + exp( -iwot)]
are readily determined from this dispersion relation. Let us first look
vos(t) = Vos [ 2 '
for a purely growing instability, which corresponds to the oscillating two
stream. For w =i"y,.Eq.(6.33) becomes
where Vas :=: eEo/mwo .. The Fourier transforms in space and time of
!:'''s. (6.19) and (6.28) give ,) \~ ( \
J:I'i I.,..e.;...c w)
= 0, (6.34)
(6.29)
where Vi has been taken to be zero as is appropria.te for ion Landau damp-
ing of a wave with zero phase velocity. Clearly 'Y > 0 requires that /i < 0
i.e., Wo < Wek as expected from the physical picture of this instability
(u} + uut/; (6.30) which we discussed in Section 6.3.
Expressions for the maximum growth rate are readily obtained in both
the weak and strong growth limits. In the weak growth limit, 'Y « kvs
and Eq. (6.34) becomes

where w~k w~e + 3 k2 v; .

Choosing w as low frequency, we use
+/i 2
+-41 (vas)
-Ve 2
wo/i=O. (6.35)
Eq. (6.29) to eliminate uZk(w ± wo) from Eq, (6.30). Noting that Iw ± . .
2wol':::2wo, we neglect as very off-resonant the terms n~k(w + 2wo) and
We find the mismatch 8 (and hence wavenumber k) which corresponds to
n!k(W - 2wo). Hence we obtain the dispersion relation maximum growth by taking the derivative 8/8/i of Eq. (6.35) and setting
8'Y/8/i = O. Hence /i = -(vos/ve?wo/8. Substituting /i into Eq. (6.35)
then gives the maximum growth rate, which is

1 (Vos)'2 (6.36)
'Y=-- Wo
where€(k w) - w2 + iv e W - w2ek: 8 Ve . 2
1 . -

This dispersion relation can be simplified considerably. First, note Due to the damping of the plasma wave, the amplitude of the pump
that 2 ' 2. •
(w .•..
l.A.lo) -+ v~(WtW,,) - 14:<. field must exceed a certain threshold value for net growth to occur. This
threshold value is simply given by the condition 'Y = 0:
€(k,w±wo) = (W±Wo+Wek)(W±Wo-Wek) + i(w±wo)ve. 2 .
Vos) _ 4 Ve
(6.37)
(
~lining [; = Wo - Web approximating Wo + Wek ~ 2wo, and assuming that Ve TH Wo
" Wo gives Let us next consider the regime of very strong growth. For growth
rate y » (kvs, ve), Eq. (6.34) becomes
'€(k,w ± wo) ±2wo (w ± /i) ± ~VeWO . (6.32)
O. (6.38)
 6 Parametric Excitation of Electron and Ion Waves
6.6 Instability Inhomogeneity Threshold

We again take 8/88 of Eq, (6.38) and set 8-y/88 = 0 to estimate the
The growth rate, -y = Iwl sin~, maximizes when a = 1/-./2. Hence we
b which corresponds to the maximum growth rate. After straightforward
obtain the real part of the frequency (wr) and the W·o·,vthrate:
aiO"ebra,we obtain
o
-y r::::.
(zm
M
k2v~s
-8-wo
) 1/3
, (6.39) 2
Wpi k 2 Vos
2 ) 1/3

( Wo 16
for a mismatch 8 = - -y.
The mentioned ion-acoustic-decay instability is also readily obtained J3 w2 ,2 2 ) 1/3
~ k Vos
from Eq. (6.33). We first examine the weak growth regime, assuming that
2 ( Wo 16 .
7 « kvs· Maximum growth clearly occurs when both the ion acoustic
wave and the electron plasma wave are nearly resonant. Hence we take
w == ku, + i-y and choose 8 = Wo - Wek = kvs. Substitution into Eq. (6.33)
then gives a quadratic equation for the growth rate T
6.6 INSTABILITY THRESHOLD DUE TO
SPATIAL INHOMOGENEITY
4-y2 + 2-y ()Vi+Ve + ViVe - -1 (Vos)
- 2
kvswo = 0, (6.40)
, 4 Ve The threshold for instability is often determined not by collisions but
where we have used -y « kvs. Growth again requires that the pump field by spatial inhomogeneties. In a plasma with a gradient in density, the
exceed a threshold value, which is obtained by the condition that -y= 0: oscillating-two-steam and ion acoustic decay instabilities are excited only
over a region of limited size i.e., where Wo ~ wpe. There is then a loss
VOS)2 Ve Vi mechanism, since the unstable waves can propagate out of the region in
(6.41 )
( -:;;; TH = 4 Wo kvs . which they are excited. Let us conclude our discussion of these instabilities
For a growth rate much greater than either collision frequency but still by estimating the effect of a density gradient on the threshold [10].
much less than the ion acoustic frequency, We start by considering a plasma wave driven unstable by the electric
vector of a li t wave which is normally incident onto an inhomogeneous
(6.42) plasma. For si plicity, we consider only the oscillating two stream insta-
bility and ass e that the plasma density varies linearly near the critical
For large amplitude pump fields, the frequency of the ion wave can be den . scale length L. "Where the excitation is strongest, the plasma
determined by thepump field intensity. In this limit, the ion wave is called wave has a wavenumber kll aligned with the electric vector of the light
a quasi-mode since it is not a mode of the undriven plasma. Assuming wave. However, at a lower density, the wave vector must develop a compo-
that Iwl » kv; and ignoring the damping terms, we return to Eq.(6.33) nent (kz) down the density gradient so that the increase in the frequency
to obtain due to the thermal correction balances the decrease due to lower density.
2 (2 .2 ) 2 k
2
s -_ O. v;;' Hence
W W -u +w·--- (6.43) 2 2 2 Z
pi 4 w 3 kz ve = wpe L'
:0
IS
solve Eq. (6.43), we take w = Iwl exp(i~) and and () = a Iwl, where a
a parameter to be varied to maximize the growth. The imaginary part
where z = 0 corresponds to the place where k = kll. As kz increases, the
efficiency of the coupling between the light wave and the plasma wave
of Eq. (6.43) then gives sin ~ = !.../2 - 0'2. The real part of this equation
decreases, since the plasma wave begins to propagate more and more
gives
in a direction orthogonal to the electric vector of the pump field. If we
estimate the size (e[ NT) of the interaction region by the condition k; -::::
kll'
 6 Par-ametr-ic Excitation of Electron and Ion Waves 6.7 References 71

. c: (3v~kll L/w~,J. The time that it takes for wave energy to propagate where the brackets denote an average. The effective bandwidth 6.w is
tt» 'f . . .
t of this region IS defined via the autocorrelation function:
ll/NT dz ~
oU
T =
o
-
Vg~ ,
(6.46)
-
1
6.w
=
100
0
dr (o:(t)o:(t + T)/ .
here "s= is the component of the group velocity of the plasma wave down
~edensity gradient. Noting that Vgz = 3kzv~/wpe and using Eq. (6.45) for If 6.w » /0, Eq. (6.49) gives
k«, we readily obtain T = 2kIlL/wpe. The effective damping rate (v = l/T)
b~oroes v/wpe = 1/2kIlL. Substituting this damping into Eq, (6.37) for (6.50)
(f) = j3exp(~~).
the threshold gives

(vas)
-:;:-
2
TH ':::
(6.47) Hence the growth rate is reduced by the ratio /o/6.w.
This reduction in the growth rate is readily understood. The intensity
of the pump wave is distributed over a bandwidth 6.w, and the resonance
width of the instability is the growth rate. Hence when 6.w » /0, only
6.1 EFFECT OF INCOHERENCE IN some fraction of the pump wave resonantly couples to any two given
THE PUMP WAVE unstable waves. A random modulation in the phase of the pump waves
leads to the same reduction in the growth rate [12,13]. In general, a spread
Finally let's note that either temporal or spatial incoherence in the pump
in the wave vectors of the pump field or even random turbulence in the
wavewill reduce the instability growth rate. As a simple example, consider
plasma [14] can also limit the coherence and contribute to the effective
the ion-acoustic decay instability driven by a spatially homogeneous pump
bandwidth.
field with a frequency near wpe and with a random modulation in its
amplitude [11]. In particular, we let
References
E = Eo o:(t) coswt ,
1. Dawson, J. M. and C. Oberman, Effect of ion correlations on high-frequency
where E is the electric field, w is the frequency, and 0:(t) is a stochastic plasma conductivity, Phys. Fluids 6, 394 (1963).
variable with a zero mean and a variance of unity. When the growth rate 2. Faehl, R. J. and W. L. Kruer, Laser light absorption by short wavelengthion
is much less than the ion acoustic frequency, the amplitude (f) of an turbulence, Phys. Fluids 20, 55 (1977).
unstable wave can be represented by terms of the form 3. Chen, F. F. Introduction to Plasma Physics. Plenum Press, New York,
1974.
4. Silin, V. P., Parametric resonances in a plasma, So». Phys. JETP 21, 1127
(6.48)
(1965).
5. DuBois, D. F. and M. V. Goldman, Radiation-induced instability of electron
~ere /0 is the growth rate in the absence of amplitude modulation, damp- plasma waves,Phys. Rev. Letters 14, 544 (1965).
l1l.g has been neglected, and j3 is a constant determined by the initial
6. Nishikawa,K., Parametric excitation of coupled waves1.General formulation,
COnditions.If we assume that 0:(t) is Gaussian,
J. Phys. Soc. Japan, 24, 1152 (1968).

(f) = j3 exp [~Jit it dt' dt" (o:( t')o:( t") ) ] (6.49)

7. Kaw, P. K. and J. M. Dawson, Laser-induced anomalous heating of a plasma,
Phys. Fluids 12, 2586 (1969).
8. Sanmartin, J. R., Electrostatic plasma instabilities excited by high-frequency
electric field, Phys. Fluids 13, 1533 (1970).
 -------------------"'

6 Parametric Excitation of Electron and Ion Waves

. a K. and K. Nishikawa, Parametric instabilities and wave dissipation

CHAPTER 7
9 ~ [irn . .
'. pJasrnas; in Handbook of Plasma PhYS1CS, Vol. 2, (A. A. Galeev and R.
~. Sudan, eds.), p.451-517. North HoUandPhysics Publishing, Amsterdam,
1984.
StImulated
io.
Perkins, F. W. and J. Flick, Parametric instabilities in inhomogeneous plas-
mas, Phys. Fluids 14, 2012 (1971).
Raman
11.Thornson, J . .1., W. L. Kruer,.S. E. Bodner and J. S. DeCroot, Parametric
instability thresholds and their control, Phys. Fluids 17, 849 (1974).
Scattering
12. Valeo, E. J. and C. Oberman, Model of parametric excitation by an imperfect
pump, Phys. Rev. Lett. 30, 1035 (1973).
13. Thomson, J. J. and J. I. Karush, Effect of finite-bandwidth driver on the
parametric instability, Phys. Fluids 17, 1608 (1974).
14. Williams, E. A., J. R. Albritton, and M. N. Rosenbluth, Effect of spatial
turbulence on parametric instabilities, Phys. Fluids 22, 139 (1979).

An important class of instabilities involves the coupling of a large ampli-

tude light wave into a scattered light wave plus either an electron plasma
wave (the Raman instability) or an ion acoustic wave (the Brillouin in-
stability). In this Chapter, we will consider the Raman instability and a
related instability in which a light wave couples into two electron plasma
waves, In the next chapter, we will discuss the Brillouin instability and
a related instability which can lead a beam of light to break up into fila-
ments.
The Raman instability Can be most simply characterized as the reso-
nant decay of an incident photon into a scattered photon plus an electron
plasma wave (or plasmon). The frequency and wave number matching
conditions then are
Wo W. + Wek
(7.1)
k, + k,
where Wo (w.) and ko (k.) are the frequency and wave number of the
incident (scattered) light wave, and Wek (k) is the frequency (wavenumber)
of the electron plasma wave. Since the minimum frequency of a light wave

73
 \ '

7.1 Instability Analysis 75

7 Stirnulated RaInan Scattering

potential A and the electrostatic potential cP, where B v x A and

P;y
.fIla is wpe, the electron plasma frequency, it is clear that this
requires that wo;;:: 2wpe i.e., noS ner/4, where n is the plasma
aDd ner is the critical density.
E = -c-1oA/0t Vci>. We begin with Ampere's la~
.•..
47r 1 DE
tY process, part of the incident energy is scattered, and part is V x B = ~ J + ~&t . (7.2)
thiS
IP.ted into the electron plasma wave. If we simply multiply the fre-
nlatching condition by n. (Planck's constant) and note that Iu» is Substituting for B and E and choosing V . A = 0, we obtain
~rgy of a photon or plasmon, it is clear that for each photon un-
&be eD this process, the fraction of its energy transferred to the plasma (7.3)
ing

-:00
d#tf,o is (wedwo). This portion of the energy will heat the plasma as the
plasma wave damps. As we will see, this electron plasma wave
~ havea very high phase velocity (of order the velocity of light) and so We next separate the current density J into a transverse part Jt (associ-
ated with the light waves) and a longitudinal part Je (associated with the
:: produce very en~rgetic el~c~rons wh.en i.t damps. Since such electrons
canpreheat the fuel m laser fusion applications, the Raman instability is electrostatic plasma wave). The longitudinal part of J can be related to
V cP via Poisson's equation and the equation for conservation of charge:
aparticularly significant concern.
The physics of the Raman instability is straight forward. Consider a (7.4)
lightwavewith electric field amplitude EL propagating through a plasma
whose density is rippled along the direction of propagation by the density (7.5)
fluctuationon associated with an electron plasma wave. Since the elec-
cronsare oscillating in the light wave with the velocity v L = eEL/TTUJJo, where p is the charge density. In particular, taking o/Ot of Eq. (7.4) and
a transverse current SJ = -e v L Sn is generated. If the wave numbers substituting for op/Ot from Eq. (7.5) gives
andfrequencies are properly matched, this transverse current generates a
scatteredlight wave with an amplitude SE. In turn, this scattered light
0. (7.6)
wave interferes with the incident light to produce a variation in the wave
~e: V{E2/87r) = V{EL . SE)/47r. Variations in wave pressure act
JUSt likevariations in the ordinary kinetic pressure i.e., plasma is pushed ince V . Jt = 0, we then obtain
&olD. regions of high pressure to regions of low pressure and vice versa,
IDd a density fluctuation is generated. Due to this feed-back loop, an 47rJe· (7.7)
lIIstabilityis possible. A small density fluctuation leads to a transverse
~nt which generates a small scattered light wave, which can in turn ence Eq. 7.3 becomes
rtanforcethe density fluctuation via a variation in the wave pressure.
(7.8)

tl INSTABILITY ANALYSIS
1\ If we restrict ourselves to the condition A . Vne = 0, the transverse
Uy ~ Coupled equations describing the Raman instability can be read-
current can be simply expressed as Jt = - ne eUt. Here Ut is the oscillation
thr erlVed [1-3]. For clarity, let us consider a light wave propagating
velocity of an electron in the electric field of the light wave and ne is the
!\ieOugha plasma with a uniform density and temperature. It is conve-
electron density. For lutl «c, Ut = eA/me since
Il.t to express the electric and magnetic fields in terms of the vector

e e oA
--Et (7.9)
m me fit .
 r

7 Stimulated Raman Scattering 7.2 Dispersion Relation 77

we obtain an eqnation for the propagation of a light wave II1 a no + ne, A = AL +A and d> = d> where the tilde denotes an infinitesimal
Bell("e.
quantity. Then
~ Ej2 2V 2) 41T
A = - ~
f'2 A . (7.10) ane
( [)t2 - C ne

[)t
+ noV· ii = 0 (7.15)
The scattering of Cl. large amplitnde light wave (Ad by a small amplitude aii e -
2
e
fluctuation (ne) is easily determined by substituting into Eq. 7.10 -Vrb -
(
-')-') V AL' A ) - 3v~
- _
V/1,.. (7.16)
~ ~y at m m-e- no
for A = AL + A and for n = no + ne where no is the uniform background
plasma density. 'vVethen obtain -Lt r where Ve is the electron thermal velocity. Taking a time derivative of

\->~~~
7-j>~H.ttlIt9... Eq. (7.15), then a divergence of Eq. (7.16), and finally eliminating the
2V 2 2) 2 '\ ( 7.11)
" ( atfY 2 - C + wpe
41T e
A = - ~ ne AL '-...J term a(V . iid/at gives 2

• ( ata2 + 2
Wpe -
2 2) -
3ve V ne
noe
= m2c2 V
2 ( AL' A- ) ~ (7.17)
The tight hand side is simply the _transverse current (ex neVL) which 2

produces the scattered light wave (A).

--:!> To derive an equation for the density fluctuation associated with the Here we h~ve also made use of Poisson's equation (V2~ = 41Tene) to
electron plasma wave. we treat the massive ions as a fixed, neutralizing eliminate cP. This equation describes the generation of a fluctuation in
background and describe the electrons as a warm fluid. The continuity the electron density by variations in the intensity of the electromagnetic
and force equations then are waves.

ane
at + V· ()ne Ue = 0 (7.12)
7.2 DISPERSION RELATION
a~e + Ue' VUe = ~e (E + U
e: B) (7.13) Equations (7.11) and (7.17) describe the coupling of the electrostatic and
el .ctromagnetic waves discussed in the introduction to this chapter. To
where ne, Ue and Pe are the density, velocity and pressure of the electron derive the dispersion relation for the Raman instability, we here take
fluid. (As we have shown in Chapter 1, these equations are readily derived AL = Ao cos(1<o. x - wot) and Fourier-analyze these equations:
as the first two .moments of the Vlasov equation.) Separating the veloc-
ity into longitudinal and transverse components (u, = UL + eA/me),
substituting into Eq. (7.13), and using a standard vector identity gives

2
e 1 eA Vpe
-aUL = -Vc/> - - V UL +- - -- . (7.14)
at m 2 ( me ) nem

The second term on the right hand side is the ponderomotive force and is
proportional to the gradient of the intensity of both the longitudinal and
transverse components of the electric field. 1/2
We now use the adiabatic equation of state (Pe/n~ = constant) and where Wek = ( w~e + 3k2v~ ) is the Bohm-Cross frequency and Wo and
linearize Eqs. (7.12) and (7.14.) In particular, we take UL = ii, ne = ~:oare the frequency and wave number of the large amplitude light wave.
We next use Eq. (7.18) to el.im.inate A from Eq. (7.19). Taking W c::: t.!pe
 7.3 Instability Thresholds 79
7 St irrurlace d Raman Scattering

For forward scatter at very low density, k « wu/c. Both upshifted and
. ,the terms n~(k - 2ko, W -2wo) and ne(k + 2ko,w + 2wol as downshifted light waves can now be nearly reson<l.I\!:
i.e.,
till!; we obtain the dispersion relation:
Ilant,
D(w ± Wo, k ± ko) ~ 2(wpe ± wo)bw ,
::=; ~ [D (w _ w~, k - ko) + D (w + w~, k + ko) 1 .(7
.20) where we have chosen k = wpc/c and let w = wpe + bw, where bw «wpe'
" 2 _ 1.2c2 _ ,,2 and v . is the oscillatory velocity of an
Substituting into Eq. (7.20), we readily find the maximum growth rate
.k)::==w re ~pe. 0, (bw = il):
f)(~, the large amplitude light wave. 2
wpe vos
~. bility growth rates are readily found from Eq. (7.20). For )~ (7.25)
~ta tter we can neglect the upshifted light wave as nonreso-
2V2wo ~.
(6 SldeSCa ,
Lastly, let us note that there is also a kinetic instability which rep-
&lving resents stimulated Compton scattering by the electrons [4,5]. Now the
W2 v2
. (w 22 2] _ pe
k2
os (7.21) electrostatic fluctuation is no longer a resonant mode of the plasma but
[ - wO)2 - (k - ko) ,c - wpe - 4 . rather a beat mode which interacts with the electrons. This instability can
. wth be readily derived from Eq. (7.20) if we replace w2 - w~e by w2 t(k,w)
•• 1.4
_w _
-Wek
+ u,'w where Dw «Wek, and note t h at. maximum gro where c(k,w) is the fully kinetic dielectric function. For a Maxwellian
whenthe scattered light wave is also resonant l.e., when velocity distribution, the growth rate peaks when Wo z: Ws + k Ve' Not
surprisingly, the maximum growth rate is much less than that for the
(Wek _ wo) 2 _ (k _ ko) 2c
2
- o. (7.22) Raman instability, unless the plasma wave is heavily damped. The two
processes then merge.

2 ]1/2 7.3 INSTABILITY THRESHOLDS

k Vos Wpe (7.23)
I = -4- [ Wek (wo - Wek) . Damping of the unstable waves introduces a threshold intensity for in-
stability generation. The simplest way to include the effect of damping is
Tbe wave number k is determme . d by Eq . (7 .22) . For example, for to add terms Vs (aA/at) and Ve (an/at) to Eqs. (7.11) and (7.17), where
eLect backscatter where the growth rate maximizes, Vs (ve) is the energy damping rate for the scattered light wave (the elec-
tron plasma wave). The dispersion relation remains the same as Eq. (7.20)
k = ko + Wo (1 _ 2Wpe) 1/2 (7.24) with the substitutions
C Wo
w2 - w~e ==} W (w + i Ve) - w~ ,
:: "'ave number starts from k = 2ko for n « ncr / 4, .~d goes to k = ko
n-, ncr / 4 as is apparent from the matching condition.
D(w, k) ==} w (w + i vs) - 2 2
kc - w~e .
~ l'he wav~ number and growth rate are less for any 90° sidescatter
The instability analysis proceeds as before. For example, for back
~~. v'2 ko for n « ner /4). For the more general case ?f sidescatter III or sidescatter, we again retain only the down-shifted light wave, take

'rs
''''IQl. A.. V'n =f. 0, the growth rate is further reduced since t~e electnc

of the incident and scattered light waves are no longer ali~ed. For
P!e, it is apparent from Eq. (7.17) that the growth rate will vanish
w = Wek + if, and choose k according to Eq. (7.22) to obtain maximum
growth. Then we obtain

A.. Ao = O. Hence sidescatter occurs preferentially out of the plane

larization , the case we have treated.
80 7 Stimulated R8lTlan Scattering
where le and "Is are the amplitude damping rates (half of the energy
damping rates) and 1'0 is the growth rate in the absence of damping. The
threshold condition due to damping then is
"10 2: J"Ie I's .
As an example, we consider backscatter for wpe/WO« 1/2 and assume
only collisional damping. Substituting Eq. (7.23) into Eq. (7.27) then gives
where Vei is the collision frequency discussed in Chapter 5. This threshold
intensity can be quite low. In general, Landau damping of the plasma
wave needs to be included, as will be discussed in Chapter 9.
In practice, the threshold intensity is usually determined by gradients
in the plasma density rather than by damping. Let us conclude our dis-
cussion of the linear theory of the Raman instability with a he istic cal-
culation of the threshold in a plasma with a linear density profil . Plasma
inhomogeneity limits the region over which three waves can r nantly
interact, and propagation of wave energy out of this region introduces.an.>
effective dissipation which must be overcome. Noting that the wave num-
bers are now a function of position, let us define K = kl (z) - k2 (z) - k3 (z).
At some point K = 0 (i.e., the waves are resonantly coupled), but away
from this point a mismatch develops. The resonant coupling is spoiled
when a significant phase shift develops. Hence we can estimate the size
fINT of the interaction region by the condition JOlINT x dz rv 1/2. Tay-
lor expanding about the matching point (K = 11:(0) + 11:' z) then gives
fINT rv 1/#. Propagation of wave energy out of this interaction region
introduces an effective damping rate of approximately Vgi/fINT, where vgi
is the component of the group velocity of the itb wave along the gradient.
Inserting these damping rates into Eq. (7.27) then gives the Rosenbluth
criterion for exp(27r) amplification in a plasma with linear variation in K:
~ 1,
Ill:' Vgl Vg2 1
where 1 and 2 refer to the growing waves.
As an example, we consider Raman backscatter at n « ncr/4. Since
the wave number of the electron plasma wave depends more sensitively on
7.4 The 2 wpe Instability 81
density than do the wave numbers of the transverse waves, K' ::= - ok/ax
and IVgpl\:'l::= oWpe/ox. Neglecting temperature grxdients and assuming
a locally linear variation in density with a scale length L = n/(on/ox),
oWpe/ox ::= wpe/2L. Noting that Vg2 ::= c and substituting into Eq. (7.29),
(7.27) we obtain the threshold condition:
VOS)2 > _2 .
( (7.30)
c ko L
In general, a more detailed treatment of the instability generation in
(7.28) inhomogeneous plasma is required. As the region of ncr/4 is approached,
the group velocity of the scattered light wave decreases towards zero, and
the WKB approximation fails. There the threshold becomes lower by a
factor of rv (ko L) 1/3, which is rougbly the maximum factor by which the
group velocity of a light wave decreases in an inhomogeneous plasma.
The threshold for Raman sidescatter is also lower than that given in
Eq. (7.30) by a similar factor, since the sidescattered light wave is more'
weakly affected by the gradient in density. The threshold intensity is also
substantially reduced at density maxima (where K' = 0). In all these
cases the instability can become absolute. The unstable waves do no
then limit by convection but grow in time until nonlinear effects onset.
An extensive discussion of the thresholds due to plasma inhomogeneity
and the convective or absolute nature of the instability is given in the
literature [6-11].
7.4 THE 2wpe INSTABILITY
Finally, let us briefly consider a related instability in which the laser
light decays into two electron plasma waves [12-17]. The frequency and
wavenumber matching conditions for this so-called 2wpe instability are
Wo = Wekl + Wek2 (7.31)
ko k, + k2 ,
where Wo (ko) is the laser light frequency (wave number) and Wekl (ktJ
(7.29)
and Wek2 (k2) are the frequencies (wave numbers) of the electron plasma
waves. Since Wekl and Wek2 are approximately wpe, this instability clearly
takes place at a density n ~ ncr/4. The 2wpe instability is a preheat
concern, since electron plasma waves are generated.
 -
7 Stimulated R.arrJ.anScattering 83
7.4 The 2..Jp" Instability

,... derive this instability, we can simply treat the ions as a fixed Here we have noted that ko,vos = O. These equations describe the coupling
.lD '
,ra!izing background and describe the electrons as a warm fluid. If we of electron plasma waves with wave numbers k an<r l: - ko by the laser
eXpress Ue = UL+vos where Yos = eAo/me and linearize Eqs. (7.12) light.
(7.H), we obtain The Fourier-analyzed continuity equation is next used to approximate
UL in the coupling terms as
8 «,
at + no v . UL + Yos . vne = 0 (7.32)
k ne(k, w)
lidk,w) 2 w
(7.38)
8UL e - 3 v2 k no
-vm - _e vne ""(v os' UL
-) , (7.33)
at m
'f-'
no
- V

where we are neglecting the additional term involving Vos which would
here UL, ne and if! are treated as infinitesimal quantities. We next take simply give a correction of order v~s' We then substitute Eq. (7.37) into
aWUe derivative of Eq. (7.32), a divergence of Eq. (7.33), use Poisson's .Eq. (7.36) to obtain the dispersion relation
equation,and combine to eliminate the term 8(v . lid / at. This gives
2_ 2)[( )2 2 ]_[k'voswpe[(k-ko)2_k2]]2( )
(W wek W-Wo - wek-ko - 2klk-kol .7.39
{file (2 3 2 2) - 8( Vos . V ne) 2
¥ + wpe - vev ne + at -nov (vos·lid = O. (7.34)
The coupling term has been simplified by approximating w :::: wpe and
Representing Vos = vos[exp(iko . x - iwot) + exp] -iko . x +: ot)]/2 w - wo:::: -wpe·
andFourier-analyzing Eq. (7.34) gives The growth rate is readily found by substituting w = Wek + i~( and
invoking frequency matching. Then
(_w2 +w;k)ne(k,w)

+ ~ k· Vas [ne(k - ko,w - wo) + ne(k + ko,w + wo)] (7.35)

~ k· Vos \ (k - ko)2 - k2\ (7.40)
'Y - 4 k [k - kol .
no
+ -2-
k2 [ liL(k - ko,w - wo) + lidk + ko,w + wo)] O.
Vos :
For k » ko, the growth rate maximizes at 'Y :::: kovos/4 for plasma waves
propagating at 450 to both ko and VOS'
~~uation for ne(k-ko,w-wo) can be directly obtained from Eq. (7.35) Either dissipation or plasma inhomogeneity introduce a threshold in-
_~unply replacing k,w with k - ko,w - wo. If we choose w ~ w and tensity for the instability. The collisional threshold is simply given by the
~t
obtain as 0ff-reson~t any r~ponses at w + Wo or w - 2wo, we~easily condition that 'Y = ve/2, where Ve is the energy damping rate due to
coupled equations for ne (k, w) and ne(k - ko, w - wo): either electron-ion collisions or Landau damping. The threshold due to
inhomogeneity is
(VOS)2
Ve
~ ~
- koL'
(7.36)
where L is the density scale length at ncr / 4. The inhomogeneous threshold
for the 2wpe instability is lower than that for the Raman instability at
ncr/4 unless the plasma is quite hot.
(7.37)

O.
 7.4 References 85
84 7 Stimulated Raman Scattering
16. Lasinski, B. F. and A. B. Langdon, Linear theory of the 2wpe instability
References in inhomogeneous plasmas, in Lawrence Liuertnore Laboratorij, UCRL-
50021-77, pA-49 (1978). ~
l. Drake, J. F., P. K. Kaw, Y. C. Lee, G. Schrnidt , C. S. Liu, and M. N. Rosen- I T. Powers, 1. V. and R. L. Berger, Kinetic theory of two plasmon decay, Phys.
bluth , Parametric instabilities of electromagnetic waves in plasrnas, Fluids 27,242 (1984).
Phys. Fluids 17, Ti8 (1974).
2. Forslund,D. W., J. YI. Kindel and E. 1. Lindman, Theory of stimulated scat-
tering processes in laser-irradiated plasmas, Phys. Fluids 18, 1002 (1975).
3. Thornson, J. J., Stimulated Raman scatt.er in laser fusion target chambers,
Phys. Fluids 21,2082 (1978).
4. Ott, E., W. M. Manheimer and H. H. Klein, Stimulated Compton scattering
and self-focusing in the outer regions of a laser fusion plasma,
Phys. Fluids 17,1757 (1974).
5. Lin, A. T. and J. M. Dawson , Stimulated Compton scattering of electromag-
netic waves in plasmas, Phys. Fluids 18, 201 (1975).
6. Rosenbluth, M. N., Parametric -instabilities in inhomogeneous media,
Phys. Rev. Letters 29, 565 (1972).
7. Nishikawa, K. and C. S. Liu, General formalism of parametric excitation;
in Advances in Plasma Phys'cs, Vo!. 16, (A. Simon and W. Thompson,
eds.), p.3-81. Wiley, New York, 1 6.
8. Liu, C.S., Parametric instabilities in' homogeneous un magnetized plasmas,
ibid, p.121-177.
9. Mima, K. and K. Nishikawa, Paramet 'c instabilities and wave dissipation in
plasmas; in Handbook of Plasma ysics, Vo!. 2, (A. A. Galeev and R. N.
Sudan, eds.), pA51-517. North-Holland, Amsterdam, 1984.
10. Koch, P. and E. A. Williams, Absolute growth of coupled forward and back-
ward Raman scattering in inhomogeneous plasma, Phys. Fluids 27, 2346
(1984).
11. Afeyan, B. B. and E. A. Williams, Stimulated Raman sidescattering with the
effects of oblique incidence, Phys. Fluids 28, 3397 (1985).
12. Goldman, M. V., Parametric plasmon-photon interactions, Annals of
Physics 38, ll7 (1966).
-,
13. Jackson, E. A., Parametric effects of radiation on a plasma, Phys. Rev. 153,
235 (1967).
14. Liu, C. S. and M. N. Rosenbluth, Parametric decay of electromagnetic waves
into two plasmons and its consequences, Phys. Fluids 19, 967 (1976).
15. Simon, A., R. W. Short, E. A. Williams and T. Dewandre, On the inhomo-
geneous two-plasmon instability, Phys. Fluids 26, 3107 (1983).
 CHAPTER 8
.•..
Stimulated
Brillouin
Scattering

In this chapter we will consider the Brillouin instability, which involves the
coupling of a large amplitude light wave into a scattered light wave plus
an ion acoustic wave. The physics of this instability is analogous to that
of the Raman instability, except that now the density fluctuation which
provides the coupling to the scattered light wave is the density fluctuation
associated with a low frequency ion acoustic wave. Our analysis will be
sufficiently general to show another instability, which is also associated
with the variations of plasma density induced by variations of light wave
pressure. This latter instability is called the filamentation instability, since
it can lead to the break-up of a light wave into filaments.
The Brillouin instability can be most simply characterized as the res-
onant decay of an incident photon with frequency Wo and wavenumber
ko into a scattered photon with frequency Ws and wavenumber k; plus
an ion acoustic phonon. The frequency and wave number matching con-
ditions then are
Wo Ws + w
ko k, + k ,

87
 8 Stimulated Brillouin Scattering 8.1 Instability Analysis 89

- -
where now W and k are the frequency and wave number of the ion acoustic then linearize Eq. (8.2) by letting ne = no + ne, A = Ai., +A and CD = O.
wave. Since the frequency of an ion acoustic wave is much less than Wo, it which gives "'"..
is dear that this instability can occur throughout the underdense plasma.
Furthermore, nearly all the energy can be transferred to the scattered
.!!...'Vd; =
'm
2
e ? 'V(AL'
m2c-
A) + v~
no
v«, . (8.3)
light wave. Hence this instability is a significant concern for laser fusion where Ve is the electron thermal velocity. The electrical potential transmits
applications, since the process can either degrade the absorption or change
the ponderomotive force to the ions.
its location. To treat the ion response, we describe the ions as a charged fluid with
density ni and velocity u.. The continuity and force equations are
8.1 INSTABILITY ANALYSIS ani
To obtain the coupled equations [1,2] describing the Brillouin instability, at + v· (niui) = 0
(8.4)
we again consider the response of an initially uniform plasma driven by aUi _ Ze 'V4>
+ u, . 'VUi =
a large amplitude light wave. We have already derived in the previous at 1'4 '
chapter an equation for the generation of a scattered light wave with
where we have neglected the ion pressure for simplicity. VV~nex~ linearize
vector potential A by the coupling of a large amplitude light wave with
these equations by taking ni = nOi + ni, u, = u, and 4> = 4>. Then
vector potential AL with an electron density fluctuation ne:
ani
+ nOi'V . Ui = 0 (8.5)
(8.1) at
aUi Z e -
= - - 'V4>. (8.6)
where wpe is the electron plasma frequency. Only the fluctuation in elec- at M
tron density appears in Eq. (8.1), since the ion r onse to the high fre- z: Taking a time derivative of Eq. (8.5), a divergence of Eq. (8.6) and com-
quency field of the light wave is less than the electrd n response by Zm/l'4,
where Z is the charge state, m the electron mass, d M the ion mass.
bining to eliminate the term a'V . lid gives at
For the Brillouin instability, the density fluctua . n ne is the low fre-
quency fluctuation associated with an ion acoustic wave. 0 derive an (8.7)
equation for this low frequency fluctuation, the ion motion must also be
included. We again describe the electrons as a warm fluid and separate If we substitute for ~ using Eq. (8.3), note that Znoi = no and approx-
the fluid velocity u, into longitudinal (uL) and transverse components imate Z ni ::::ne, we finally obtain an equation for the low frequency
(eA/m c). Then, as shown in Eq. (7.14), density fluctuation:

a:
uc
L = ~'V4>.-
m
~'V(UL
2
+ eA)2
mc
(8.2)
.. -
2 '<"72-
C, v ne = (8.8)

where 4>is the electrostatic potential, Pe the electron pressure, and ne the Here Cs = (ZBe/ M) 1/2 is the ion acoustic velocity. Equation (8.8) describes
electron density. Since we are now considering a low frequency fluctuation, the excitation of an ion acoustic wave by the interaction between the
We neglect the electron inertia (aur) at-> 0) and use the isothermal incident and scattered light waves.
equation of state (Pe = neE}e, where Be is the electron temperature). We
 8 Stirnul ••ted Brillouin Scattering 91
8.3 Instability Thresholds

.~DISPERSION RELATION III the strong field limit, Iwl » kc., and Eq. (8.11) then becomes a
cubic equation for w. Again choosing k as given by Eq!,,(S.14) for maximum
'fo derive the dispersion relation from the coupled equations for A and
ne' we take ~L = AL cos(ko . x - wot) and Fourier-analyze Eqs. (8.1) and growth, we now obtain
8.8) to obtam
- 4?Te
2
AL [_ W c::
k-o.)?v-os ur:
---
.')
pi
1/3 [ 1 + J33]
-
.
't - .
(8.16)
D(k,..,;) A(k,w) = ~2 ne(k-ko,w-wo)+Tte(k+ko,w+wo)] (8.9) ( 2 wo 2 2

Note that in this strong field limit, the frequency of the electrostatic
(w2 - k2c;) iie(k,w) = (8.10) wave is determined by the amplitude of the light wave. In this limit,
Znoe2 k2 AL [- - the electrostatic wave is sometimes called a quasi-mode, since it is not a
mAfc2 -2- . A(k - ko,w - wo) + A(k + ko,w + wo)] norrn al mode of an undriven plasma.
The wave number k cOlTesponding to sidescatter is less than that for
",here D(k,w) = w2 - k22 - w2pe . We next use Eq . (89)
. t 0 elimin ate backscatter, since the ion wave has to take up less momentum. For exam-
A(k - ko,w - u.;o)and A(k + ko, w + wo) from Eq. (8.10). If we choose w ple, for 90° sidescatter, k = V2 ko· As is then apparent from Eq. (8.11),
to be ,lOWfrequency (w « wo) and neglect as nonresonant the terms with the growth rate is also less for sidescatter. We again note that sidescatter
ne{k I 2ko,w ± 2wo), we obtain the dispersion relation: occurs preferentially into light waves propagating out of the plane defined
by the electric and propagation vectors of the large amplitude light wave.
w
2
- k
2
c; = k2:~s W~i[D(W _ w~, k _ ko) + D(w + w~, k + koJ . (8.11) In this c e, the electric vectors of the light waves can be aligned, maxi-
mizing e ponderomotive force. This is the case we have focused on with
Here Vos = eAL/mc, and Wpiis the ion plasma frequency which is given o unplifying assumption that A . V'ne = O.
by wpi = wpeVZm/M. There is also a kinetic version of the Brillouin instability, which rep-
10 Instability ~owth rates are readily found from Eq. (8.11). For Bril- resents stimulated scattering from the ions. In this instability, the two
~ b~ck or sideward scatter, k is of order ko and so only the down- electromagnetic waves beat together to produce an electrostatic fluctua-
shifted light wave need be retained. Then tion which resonates with the bulk of the ions i.e., Wo- Ws '" kVi, where
Vi is the ion thermal velocity. Since the electrostatic disturbance is not a

(w2 _ ec;)(w2-2wwo+2ko·kc2_k2c2) = k2:~sW~i. (8.12) normal mode of the plasma, the growth rate is much less than that for
the Brillouin instability unless the ion waves are heavily damped. This in-
As an example, we consider backscatter which has the largest growth stability can be included in the dispersion relation by replacing our fluid
rate. If we consider first the weak field limit in which w = kc + i wh description of the ions with a kinetic treatment.
-y <:<: kcs, Eq, (8.12) becomes s I, ere

2ikcs,(-2iwO/-2wokcs+2kkoC2_k2C2) = k2:~W~I. (8.13) 8.3 INSTABILITY THRESHOLDS

Damping of the unstable waves introduces a threshold intensity for in-
Ma.:·
is :um growth clearly occurs for k such that the scattered light wave stability generation. As we have discussed in the previous chapter, net
o a resonant mode. Then growth of the unstable pair of waves requires that
2wo c,
k 2 ko (8.14) (8.17)
c c
1 ko Vos Wpi
where 1 is the growth rate in the absence of damping, I. is the amplitude
1 (8.15)
2V2 Jwo ko c, damping rate of the scattered light wave, and li is the amplitude damping
 92 8 Stimulated Brillouin Scattering 8.4 The Filamentation Instability 93

rate of the ion acoustic wave. As an example, let us consider backscat- along the density gradient of a light wave propagating principally in the
ter and assume collisional damping of the light wave hs ':= veiw~e/2w5, direction orthogonal to the gradient is ~ c] ("-oL) lAl. As is apparent. from
where Vei is the electron-ion collision frequency defined in Chapter 5). Eq. (8.19), t.he threshold intensity is then reduced by a factor of order
Substituting Eq. (8.15) into Eq. (8.17) then gives ("-oL) l/3

VOS)2 2 4 Vei~. (8.18)

( 8.4 THE FILAMENTATION INSTABILITY
Ve Wo ko Cs
The dispersion relation shown in Eq. (8.11) also describes the filamenta
The damping of the ion wave is usually determined by Landau damp- tion instability [3-6], which corresponds to the growth of zero-frequency
ing, which will be discussed in a later chapter. We need only note here density perturbations (and the corresponding modulations in intensity)
that-y, < kocs, and so the threshold due to damping is usually quite low in the plane orthogonal to the propagation vector of the light wave. If we
i.e., (vos/ve)2 « 1.
assume that w = i, « Wo and k- ko = 0, we find that D(w±wo, k± 1<.0)::
In practice, the threshold intensity is usually determined by gradients ±2iwo, - k2c2 Substituting into Eq. (8.11), we obtain
in the plasma density and expansion velocity rather than by damping. As
indicated in the previous chapter, instability growth in an inhomogeneous
2 2 2) 2
k4
_c_ 4) 2 2
_ k4 Vos C ~w
2
plasma is a very rich topic. To illustrate the effects of inhomogeneity, let's h +k Cs ( ,+ 4w2
o
- 8
Wo
2 . (8.21)
again simply consider the Rosenbluth criterion for exp(27r) convective
growth in a plasma with a linear variation in the wavenum er matching:
For illustration, let us here derive the maximum growth rate in the limit
, « kcs. The wave number for maximum growth is found by differen-
(8.1 tiating Eq. (8.21) with respect to k and requiring that 8,/8k = O. The
growth rate is then evaluated for this value of k. In the limit, « kc, and
neglecting the ion temperature, we obtain
Here ",,'is the gradient of the wave number mismatch and Vgl and Vg2 are
the components of the group velocities of the unstable waves along this
gradient. As an example, we consider Brillouin backscatter in a plasma ~ (Vos)2 w~e
8 Ve Wo
with a density gradient with scale length L = n/(8n/8x). Then ",,' =
wpe Vos
8(ko - k, - k)/8x ':= 2(8ko/8x) z: wi,e/wocL, for wpe « Wo. Substituting k
2c Ve
into Eq (8.19) gives
8 We note that the density fluctuations are purely growing. They simply
(8.20)
koL correspond to the variations in plasma density driven via the ponderomo-
tive force by intensity modulations in the light beam. Whole beam self
A gradient in expansion velocity can be even more effective in lim-
focusing results from the same physical process and can be considered a
iting the region over which the coupling is resonant. In an expanding
special case of the filament at ion process. Since resonance with an ion wave
plasma, w = k(cs +vexp). Then ",,'= -8k/8x ':= kCsL;;l /(c. +vexp), where
is not involved, the filament at ion instability is not extremely sensitive to
L; = cs/ (8vexp/ 8x) is the velocity gradient scale length. Substituting ",,'
plasma inhomogeneity. The instability is often characterized by its spatial
into Eq. (8.19), we then obtain the result shown in Eq. (8.20) with the
gain coefficient «, which is the growth rate divided by the group velocity
replacement of L by L; w~e/2w5.
of the light wave. For, «kc" r: ':= (1/8)(vos/ve)2 (w~e/w5)(wo/c).
The threshold intensity due to inhomogeneity is lower for sidescatter,
Filamentation and self-focusing can also be driven by either thermal
since a sidescattered light wave spends a longer time in the region of
forces [7-10] or relativistic effects [11]. In the first case, a localized increase
interaction. If we refer to the discussion in Chapter 3, the group velocity
 8 Stimulated Brillouin Scattering
CHAPTER 9
. the intensity of a light wave raises the plasma temperature via the en-
:allced. heating. Refraction of the yght ,,:,aveinto the resulting density
depressIOnenhances the perturbation m intensity, completing the feed- Heating by
back loop. These thenn~. effects can be particularly important in dense,
old plasmas where collisional absorption is efficient. The relativistic ef- Plasma Waves
~cctcall be si~ficant fo~a very intense light wave. Since w~e = 47rne2/m,
the relativlstIc.mcrease ill the mass of an electron oscillating in the light
wavehas the same effect as a decrease in the plasma density. The light
waveis focused, enhancing its intensity.
The Wave-Particle Interaction

References

1. Drake, J. F., P. K. Kaw, Y. C. Lee, G. Schmidt, C. S. Liu, and M. N.

Rosep.bluth, Parametric instabilities of electromagnetic waves in plasmas,
Phys. Fluids 11, 778 (1974).
2. Liu, C. S., Parametric instability in an inhomogeneous unmagnetized plasma;
in Advances in Plasma Physics, Vo!. 16, (A. Simon and W. Thompson,
eds.), p. 121-177. Wiley, New York, 1976.
3. Kaw, P. K., G. Schmidt and T. Wilcox, Filamentation and trapping of elec-
tromagnetic radiation in plasmas, Phys. Fluids 16, 1522 (1973).
We have examined a number of different processes whereby intense laser
4. Litvak, A. G., Finite-amplitude wave beams in a magnetoactive plasma,
light couples either into electrostatic waves (resonance absorption and the
Sov Phys. -JETP 30, 344 (1970). '. oscillating-two-stream, ion acoustic decay and 2wpe instabilities) or into
5. Palmer, A. J., Stimulated scattering and self-focusing in laser-produced plas- both electrostatic and scattered light waves (the Raman and Brillouin in-
mas, Phys. Fluids 14, 2714 (1971).
stabilities). In order to understand the evolution and the consequences
6. Cohen, B. 1. and C. E. Max, Stimulated scattering of light by ion modes in a of these processes, it is necessary to consider how electrostatic waves
homogeneous plasma: space-time evolution, Phys. Fluids 22, 1115 (1979). are damped by the plasma particles. Since electrostatic waves are sim-
7. ~odka, M. S., A. K. Ghatak and V. K. Tripathi, Self-focusing of laser beams ply charge density fluctuations and their associated electric fields, these
III plasmas and semiconductors; in Progress in Optics, Vol. 13, (E. Wolk, waves do not readily escape from the plasma. Their energy is ultimately
ed.). North-Holland, Amsterdam, 1976.
transferred to the particles via either linear or nonlinear damping mech-
8. Perkins, F. and E. Valeo, Thermal self-focusing of electromagnetic waves in anisms. We will first discuss the damping of a small amplitude electron
plasmas, Phys. Rev Letters 32, 1234 (1974). plasma wave, which is already sufficient to illustrate important featur~ of
9. Craxt
. on, R. S. and R. 1. McCrory, Hydrodynamics of thermal self-focusing the heating via plasma waves. We will then briefly consider the damping
10 III laser plasmas, J. Appl. Phys. 56, 108 (1984). of a large amplitude electron plasma wave, Finally we conclude with a
. ;~tabrook, K. and W. L. Kruer, Two-dimensional ray-trace calculations of discussion of electron heating by parametric instabilities near the critical
1\ ermal whole beam self-focusing, Phys. Fluids 28, 19 (1985) . density, including a brief discussion of plasma wave collapse.
. ~ax, C. E., J. Aarons and A. B. Langdon, Self-modulation and self-focusing
electromagnetic waves in plasmas, Phys. Rev. Letters 33, 209 (1974).

95
96 9 Heating by Plasma Waves 9.2 Landau Damping 97

9.1 COLLISIONAL DAMPING over a distribution of particles. Finally we will invoke energy balance to
determine the rate at which the field damps or grows.due to its interaction
Electron-ion collisions provide the simplest mechanism for the damping
with the particles.
of an electron plasma wave. Our discussion of this collisional damping
Since we consider an electrostatic wave and neglect any magnetic
is quite analogous to that previously given for an electromagnetic wave.
fields, a one dimensional treatment is sufficient. The particle dynamics
The coherent motion of oscillation of electrons in the electric field of the
are determined by
wave is converted to random (or thermal) motion at the rate at which
electron-ion collisions occur. To balance the energy dissipated, the energy
x = .!iE sin{kx - wt) , (9.2)
m
of the wave then damps at the rate l/ i.e.,
where q and m are the charge and mass and E is the amplitude of the
electric field. We compute the dynamics by expanding about the free-
= nmv2 w_
l/' __
(9.1) streaming motion of a particle with initial position Xo and initial velocity
e> 2 '
Vo. In particular, we assume that

where vw = eE/rru..;, E is the amplitude and w is the frequency of the

electric field, n is the plasma density, and l/ei is the electron-ion collision x Xo + Vo t + Xt + X2 (9.3)
frequency. Hence l/ = W~el/edW2; where wpe is the electron plasma fre- v Vo + VI + V2 . (9.4)
quency. The value of l/ei is the same as that derived in Chapter 5 for e
collisional damping of a light wave. For an electron plasma wave, W -:= wp The subscript 1 denotes a first-order correction which is proportional to E;
and so l/ -:= l/ei. the subscript 2 denotes a second-order correction proportional to E2. As
As discussed in previous chapters, collisional damping can determine will become apparent, our expansion parameter is kbx, where bx is the
the threshold intensity for instabilities, an effect which can be quite si - change between the free-streaming position of a particle and its actual
nificant for dense, low temperature, high Z plasmas. However, colli- position.
sional thresholds are often greatly exceeded, particularly in laser-fusion To compute the energy changes to order E2, we must simply compute
applications with very intense and/or long wavelength laser light. Other the motion to second order. IT we substitute Eqs. (9.3) and (9.4) into
mechanisms for the wave damping must then be considered. Eq. (9.2) and expand, we obtain

Vt = qE sin(kxo - Ot) (9.5)

9.2 LANDAU DAMPING m
qE
An electrostatic wave can be damped even in the absence of collisions. V2 - kXI cos(kxo - Ot) , (9.6)
m
This so-called collisionless or Landau damping can be qualitatively un-
derstood rather simply. Consider an electrostatic wave with electric field where 0 = w - kvo. Several trivial integrations of Eq. (9.5) give
E sin(kx - wt). Most particles are non-resonant i.e., have a velocity v
much different than w/k, the phase velocity of the wave. These particles vi = ;~ [cos(kxo - Ot) - cos kxo] (9.7)
simply oscillate in the field and experience no secular gain or loss in en-
ergy. In contrast, resonant particles with v -:= w/k experience a nearly Xl - qE [sin(kxo - Ot) - sinkxo + Otcoskxo], (9.8)
COnstant field and so can be efficiently accelerated or decelerated. These m02
Particles do exchange energy with the wave.
Substitution of Eq. (9.8) into Eq, (9.6) gives
A very straight forward and physical treatment of Landau damping
c~ be given [1]. We will first calculate the changes in the energy of par- 2
tIcles moving in a given field. Then we will average these energy changes V2 = - kq2
m202E cos{ho - Ot) [sin{ho - Ot) - sin ho + Ot cos ho ] . (9.9)
 9.2 Landau Damping 99
9 Heating by Plasma Waves

particles with velocity slightly greater than w / k lose energy. If the veloc-
We ne~t compute. th~ rate of change of the energy (8£) of a set of ity distribution decreases with velocity, the particles gain energy from the
ides WIth random initial positions. First, note that (8£1) = (mvOVl) = wave. If the slope of the distribution function is ~verted, the particles
where ( ) denotes the average over initial positions. To second order
lose energy to the wave.
obt~ (8£2) = mVO(V2) + m(vlvl)' Substituting from Eqs. (9.7) and By energy conservation, the rate of change in the energy of the parti-
.9) gwes cles must be balanced by a growth or damping of the wave. Specializing
to an electron plasma wave, we have
q2 E
2m
2
[Sin
n-Dt. + kvo ( .
0.2 smDt - Dt cosDt)
]
. (9.10)
(9.16)
Considerable simplification results if we now take the long-time limit
f1d express the results ill terms of a delta function. A useful representation where I is the rate at which the electric field grows or damps. Substituting
iJ a delta function is from Eq. (9.15) gives
sin Dt 1"'('(/)"'2.

1T0. . (9.11) " - -~/~--

-: J 2-
I ~wpe
2
a-(W) (9.17)
i'- ",.'-
w = 2 k2 avf k '
.
(8£2) = ~
1Tq2 E2 [
0(0.) - kvo
a
aD 0(0.) ] . (9.12)
where f = nf and wpe is the electron plasma frequency with density n.
Note that I depends on the slope of the distribution function evaluated
at the phase velocity of the wave. This Landau damping (or growth) rate
e 8(w - kvo) = Ikj-l o(vo - w/k), we then obtain can also be readily derived directly from the Vlasov equation. It,
If we consider as an example a Maxwellian distribution with thermal 'I
. 1Tq2E2 a [ w ]
(9.13) velocity ve, I,
(0£2) = 2mlkl Ovo vo o(vo - k)
I F w~ew ( W2) (9.18) .\

11. ~tly, we average the rate of the energy change over a distribution
:; = - V 8' IPlv~ exp - 2k2v~ . 11
1I
IlUtial velocities, f (vo). Then Note that the Landau damping of an electron plasma wave is a strong I
function of its phase velocity. The damping becomes sizeable whenever
d,; :0£10...)
w/k ~ 3ve i.e., when kADe ~ 0.4 where ADe is the electron Debye length 'j
(9.14)
and w = (w~e + 3k2v~)1/2.
~I
l...1 '::_I/,t1.\ :1
'ilere th T~ } d . . Let us conclude our discussion of linear Landau damping with a sim-
IIIto Eq e ar enot:s the a:erag~ over velocities. Substituting Eq. (9.13)
ple mechanical analogy. Consider a group of boxes translating along at a
It~ . (9.14) and integrating gives the rate at which the particles gain
velocity equal to w/k. Inside the boxes are uniformly-distributed parti-
energy:
2 2 cles, some moving slightly slower than w/k, some moving slightly faster.
(0£) = _ 1Tq E ~ af (~) (9.15) As illustrated in Fig. 9.1, those particles moving slower than w / k are over-
2 2rri Ikl k Ov k .
taken by the wall to their left and gain energy as they are bounced off.
1lo~~Uation (9 .15) illustrates some very important features of the wave- Likewise, those particles moving faster than w / k overtake the right wall
. :u
:"'liq .
lllteraction in the collisionless limit. The energy exchange is de-
elo by the resonant particles (those with Vo r::: w / k) and depends on
and lose energy as they are reflected. For a time less than the transit time
of a particle through the box, the net energy change simply depends on
~e of the velocity distribution at the phase velocity of the wave. whether more particles are initially moving faster or slower than w/k.
lCUlar, partrc. les wi velocity
es WIth .. slightly less than w / k gain energy;
 100
9 Heating by Plasma Waves
9.4 Wave breaking of Electron Plasma Waves 101

Wbt « 1. Alternatively, the linear theory requires that I » Wb i.e., that

the wave damp before electrons can oscillate in the ~~~ghs.
In the opposite limit b « Wb), we encounter trapping of electrons
in the potential troughs of the electrostatic wave. Since the motion of
the resonant particles (both trapped and untrapped) becomes periodic,
we then expect the amplitude of the wave to oscillate, as it first gives
cdk -
and then recovers energy from the particles. In other words, first there
Figure 9.1 A mechanicalanalogue for Landau damping. are more electr moving slightly slower than the phase velocity of the
wave. The wave d ps as the slower electrons gain energy. This leads to a
situation in which ere are now more electrons moving slightly faster than
w/k, and the wave egains energy from the particles. Such an oscillation in
the energy exch ge is also present in the mechanical analogy discussed in
9.3 LINEAR THEORY LIMITATIONS - TRAPPING the previous section. Electrons bouncing off one wall and gaining energy
Let us now note some important, restrictions on the linear theory of the clearly lose this energy after they transit the box and bounce off the other
wav~particle interaction. Our expansion about the free-streaming orbit wall.
requires that kbx « 1, a condition that fails after a finite time which Of course, electrons in a sinusoidal potential trough actually have
de~ends on the wave amplitude. This limitation on the linear theory can bounce frequencies which depend on their initial positions. Hence the pe-
easily be seen by examining (k2xi). Using Eq. (9.8), we obtain riodic interchange of energy between the wave and the particles gradually
phase mixes away, as the slope of the distribution function flattens in
2 2 2
2 ke E [ the neighborhood of the phase velocity (2). It should also be noted that
(k xi) = 2 m2!l4 (1 - cos !It)2 + (!It - sin nt)2] . the trapped electrons can also generate the so-called sideband instability,
which leads to an exponentiation of nearby waves (3).
The condition (k2xi) « 1 is most stringent for a resonant particle (!l = 0)'
(k2e2E2/8m2)t4 « 1. Defining a frequency Wb = (eEk/m)I/2 we then
have Wbt« 81/4. '
9.4 WAVEBREAKING OF ELECTRON
Physically, Wb is a characteristic frequency with which trapped elec-
PLASMA WAVES
trons oscillate in the potential troughs of the wave. Consider the motion
o.fa resonant electron in the field E sin( kx - wt). If we use the transforma- Let us now discuss a useful picture which illustrates some important qual-
tion ~ = x - (w / k)t to change to a frame moving with the phase velocity itative features of the nonlinear wave-particle interaction. As linear theory
Eq. (9.2) becomes ' has shown, a small amplitude wave is damped only by those particles with
e a velocity quite near its phase velocity. However, in a large amplitude elec-
~ - m E sink~ . (9.19)
tron plasma wave, the oscillation velocity of an electron in the field can
For electrons near the bottom of the potential troughs (i.e., well trapped be large enough to bring even an initially cold, main body particle into
electrons), sink~ ':::'k~. Hence, resonance with the field. That is, when (eE/mw) ':::' w/k, numerous
particles can "resonantly interact" with the wave. A strong, nonlinear
.. ekE
~= --c
m
(9.20)
damping results as electrons are efficiently accelerated by the wave. The
wave amplitude is often referred to as the amplitude at which breaking
desCri.bingharmonic motion with a bounce frequency Wb = (eEkjm)I/2. occurs in a cold plasma [4].
The linear theory only describes the early phase of this motion i.e., for At the wave breaking amplitude, large numbers of formerly nonreso-
nant particles become strongly "trapped." The wave energy is suddenly
 I
9.4 Wavebreaking of Electron Plasma Waves 103

I
9 Heating by Plasma Waves

Here Ve is the electron thermal velocity and no is the density of the uni-
cd as these slow particles are accelerated by falling into the potential
form, unperturbed plasma. Substituting Eq. (9.23) .••.into Eq. (9.24), we
~Pbs of the wave. At this wavebreaking, wt, = (eEk/m)1/2 = w. Since
~f~s the characteristic time for resonant particles to move in the field
obtain
2eeP
Wb hence take energy from it, the energy exchange to the particles takes (9.25)
d mv2 p
/Ill
every rapiidly.
pta.cThe amplitude of the field at which particles are nonlinearly brought where {3 = 3v2 /v2. By differentiating Eq. (9.25) with respect to u, it
e p _ 1/4
. to resonance with a wave (i.e., are strongly trapped) is significantly is easy to see that eP has an extremum (ePer) when u/vp - (3 . The
~uced [5]in a warm plasma for several reasons. Faster electrons are more
corresponding potential is
easilybrought into resonance, and the sizeable pressure force associated
withthe density fluctuation of the wave gives an additional acceleration. 2 e ePer (9.26)
We can crudely model the effect of plasma temperature on wavebreak- - mv2
ingby considering a water bag model, which corresponds to replacing a
Maxwelliandistribution with a velocity distribution which is constant be- This simply corresponds to the condi .on that the net energy of the fastest
tween ±J3ve. Such an idealized distribution is convenient since it has electron be zero in the wave frame.
I
i
the same pressure as does a Maxwellian distribution with thermal ve- To determine the critical value of the electric field, we consider Pois- 1'<'
locity ve, yet there is a well-defined maximum initial velocity of J3ve. son's equation: 82eP/8x2 = 47r e - no). Multiplying by 8eP/8x and !
Although there are particles with an arbitrarily high velocity present in using Eqs. (9.23) and (9.24) gives (in the wave frame)
a Maxwellian distribution, the number of particles is not sizeable until
v ~ 2ve. Hence the water bag distribution can be expected to roughly
modelthe condition that significant numbers of particles are nonlinearly (9.27)
broughtinto resonance.
In this model which assumes fixed ions, the average density (n) and
velocity(u) satisfy the same equations as those for a warm electron fluid,
as is apparent from taking moments of the Vlasov equation. Hence the
The constant has been evaluated by noting that ;p = 0 when eP= ePcr·The
continuityand force equations are
maximum electric field (Emax = - 4>max) obtains when eP = 0:

=0, (9.21) 1 + 2 v0 - ~{31/4 - f!.. . (9.28)

3 3
-~E _ 1 ap (9.22)
m m n Bx '
The maximum field amplitude is plotted in Fig. 9.2 as a function of
:ince We are considering a high frequency electron plasma wave, the J3 ve/vp. For Ve = 0, the cold plasma result is recovered. Note. the size-
~essUre p is determined by the adiabatic equation of state. Introduc- able decrease of the maximum field as the plasma temperature increases .
.,; E:::: -8eP/8x and transforming to the wave frame with velocity =v» For example, for vp = 5ve, (eEmax/mwpevp) '::::0.3.
•••\res

(9.23)

2
2eeP
U - + (9.24)
m
 104 9 Heating by Plasma Waves 9.5 Electron Heating 105

1.0 eEo
=0.5
0.1 mCllpe V le

0.8 w., = 1.04 Cllpe

~ •
a. .:.t.
>Cl) c:
w
a. 0.6
~
Cl>
a ~
E '"ul 0.01
0.4
.,

0.2
0

(b)
0.6 0.8 1.0 3.40
{3velvp

Figure 9.2 The wavebreakingamplitude as a function of thermal velocity.

2.79

o
W
9.5 ELECTRON HEATING BY THE ~
W 2.17
OSCILLATING-TWO-STREAM AND ~
ION ACOUSTIC DECAY INSTABILITIES
We will continue our consideration of electron heating via plasma waves
with a discussion of some particle simulations of a simple but instructive 1.56
model problem. A plasma with a uniform density is driven by .an imposed
spatially-independent pump field (Eo sinwot) with a frequency (wo) near
the electron plasma frequency. Such a pump field models the electric field
of a light wave near its critical density under the assumption that the wave 0.94oL_.l.--12LO--L--24~0~--L-=:.-.-J--~480
number of the light wave is negligible compared with the wave numbers
of the plasma waves which are excited. Since the unstable plasma waves
Wpet
preferentially grow along the electric vector of the pump field, a great
deal can be learned by using a one-dimensional electrostatic particle code Figure 9.3 Computed evolution of (a) the electron plasma ~av~ energy and
[6-8J. (b) the total energy of a plasma driven by an electric field oscillating near the
electron plasma frequency (from Kruer et al., 1970).
 9.5 Electron Heating
107
9 Heating by Plasma Waves

A few sample results from a simulation illustrate important features

of the collective heating]"]. In this example Wo = 1.04wpe, eEo/mwpe =
O.SVe, and the ion-electron mass ratio is 100 which is sufficient to clearly
separate the electron and ion time scales. Fig. 9.3(a) shows the evolution
of the energy in plasma waves, and Fig. 9.3(b) shows the evolution of f\
the total energy of the simulated plasma. At first there is essentially no
plasma heating, retlecting the fact that the plasma is nearly collisionless.
Meanwhile the plasma waves are exponentiating in amplitude. Finally
these waves saturate, concomitant with the onset of a rapid plasma heat-
6- -
ing due to the acceleration of plasma particles by the large amplitude
plasma waves. An effective collision frequency corresponding to the col-
lective heating is very large; i/" ~ 0.06wo, where v· describes the rate at
whichthe plasma energy increases with time in the nonlinear state.
The computer simulations emphasize another very important feature
ofthe anomalous heating. Figure 9.4 shows a typical heated electron veloc-
ity distribution calculated with the particle code. The heating has been
~ -
principally a production of very high velocity tails on the distribu ion ai4-
function [9]. This generation of very high velocity electrons takes pace E.
since large amplitude electron plasma waves readily accelerate part' les .•..
out to their phase velocity.
The physics of the nonlinear saturation can be very ri e are a
number of different regimes depending on the pump field intensity. When
the plasma is strongly driven (eEo / mwpeVe .:G 1), the dominant process
is simply strong electron trapping in the most unstable plasma wave. 2e-
-
Strong trapping occurs when many electrons are nonlinearly brought into
resonance with the wave. A large energy transfer then occurs, as the
electrons are efficiently accelerated by the wave.
. We can use the calculations of the wave breaking amplitude to es-

J
timate the saturation in the simulations, considering an example in the
tt~PPing regime: eEo/mwove = 1.0 and Wo = 1.04wpe.Linear theory ap-
plied to this case predicts that the most unstable plasma wave has a I I

~ve number ~ 0.25wpe/ve for the electron-ion mass ratio of 0.01 used in -9 ~ -3 0 3 6 9
this simulation. Equation (9:28) then predicts that strong trapping on- VIVT
sets when eE/mwpeve ~ 0.8, which compares reasonably well with the
cOInputedvalue of eE/mwpeve ~ 0.6 at saturation. Figure 9.4 A typical heated electron velocity distribution from a particle

\\I A. simple estimate of the anomalous heating rate can also be given. simulation of a plasma driven by an electric field oscillating near wpe·
pI e estimate the energy transfer from the external driver to the electron
(~a:Inaoscillations as 2, (E?v) / 47r,where, is the linear growth rate and
IV) /47r is the energy density of the plasma oscillations. The transfer of
 108
9 Heating by Plasma Waves
energy to the particles is given by our definition ~f the anomalous heating
rate as v· EJ/87f. When the plasma waves saturate, these energy flows
balance. Hence, we estimate u" as
(9.29)
where (Elv) is the mean square amplitude of the electric field at satura-
tion. For the example discussed above, Eq. (9.29) predicts v" ':.:::! O.04wpe,
again comparing reasonably with the computed value of v· = O.06wpe.
There are other nonlinear regimes. A particularly important one ob-
tains when the amplitude of the pump field is weaker. Then the excited
plasma waves obtain an amplitude E '" Eo without trapping. Hence they
in turn act like efficient "pumps" to drive even shorter wavelength plasma
waves, and so on. The net r~t is a cascade (collapse) of energy from
long wavelength waves to short wavelength ones which Landau damp.
Again the saturated state is characterized by a steady transfer of energy
from the pump field to plasma waves to a heated tail of electrons. This
nonlinear transfer of energy to shorter w~elength waves is also important
in the evolution of beam-driven instabi ties [lG-12].
A red,:"ced ~escripti?n was shown t ,reprod~1 these moderately-
pumped simulations quite well [13]. In t~cription, the two fluid
equations were used to describe the coupled evolution of the electron
plasma waves and the ion fluctuations. Simultaneously the electron dis-
tribution function (and hence the Landau damping) was evolved by solv-
ing a diffusion equation with the diffusion coefficient made a function of
the electric field amplitudes. And even though the field structures locally
became quite spiky [14], test particle calculations showed that diffusion
was a reasonable approximation for the coarse-grained evolution of the
distribution function [15].
9_6 PLASMA WAVE COLLAPSE
The tendency of intense plasma waves to cascade to higher wave numbers
or to collapse to shorter scale lengths is a very important property of this
turbulence. Weak turbulence theory only includes processes such as the
ion acoustic decay instability and stimulated scattering on the particles,
which down-shift the wave frequency and so transfer energy to longer
wavelength (higher phase velocity) waves [16J. In the absence of sufficient
9.6 Plasma Wave Collapse 109
collisional damping, wave energy in a driven plasma would indefinitely
accumulate at long wavelengths. However, when E~f47fnBe ::p k2 Ab.,
nonlinear contributions to the wave dispersion relation begin to exceed
thermal corrections. Here E is the electric field and k a typical wavenum-
ber of the plasma oscillation, Be is the electron temperature, and ADe is
the electron Debye length. Weak turbulence theory then no longer applies
and wave energy can indeed be coupled into shorter wavelength oscilla-
tions, where Landau damping provides an energy sink. The oscillating
two stream instability discussed in Chapter 6 is an excellent example of
this generation of higher wavenumber plasma waves.
It's very instructive to consider the nonlinear processes in space rather
than in a Fourier representation. A local region of intense field expels
plasma via the ponderomotive force, forming a density cavity which fur-
ther localizes and intensifies high frequency oscillations. In two or three ~
dimensions (or in strongly-driven one-dimensional plasmas), the resulting
cavity plus its self-consistent high frequency oscillation continues to col-
lapse until efficient damping of the high frequency oscillation onsets. This
Landau (or transit-time) damping onsets when the cavity size is of order
1G-20Ar>e.
The basic theory of electron plasma wave collapse was developed by
Zakharov [17]. The analysis is based on a generalization of the coupled
equations for the plasma waves and the ion waves discussed in Chapter
6. First we explicitly remove the high frequency time dependence at wp< ,
i.e., let
Ueh = ne Ueh X,
<0 - ( t) e-iw!"'t ,
where Ueh is the oscillation velocity of the electron fluid in the high fre-
quency electrostatic field. Equation (6.19) is then readily generalized:
. a 3 v; 2 ) _ wpe nel _
( at + -2 -
t -
wpe
V Ueh = -
2
-
n
Ueh . (9.30)
Note that the driving term on the right hand side of this equation simply
represents the coupling of the plasma wave with nel, the low frequency
fluctuation in the electron density. Likewise, Eq, (6.28) becomes
(9.31)
where Vs is the ion sound velocity. Equations (9.30) and (9.31) are called
the Zakharov equations. The physics of the coupling is clear from our
 9 Heating by Plasma Waves 9.6 References III

discussion in Chapter 6, where a linearized version of these equations was References

used to derive the oscillating two stream and ion acoustic decay instabil-
ities. 1. Stix, T. H., The Theory of Plasma Waves. McGra~-Hill, New York, 1962.
If ion inertia is neglected, Eq, (9.31) shows that nel/n = -u;h/4v;. 2. O'Neil, T. M., Collisionless damping of nonlinear plasma oscillations, Phys.
Substitution for net! into Eq. (9.30) then gives a nonlinear SchrOdinger Fluids 8, 2255 (1965).
equation [18J. However, neglect of ion inertia is an extremely restrictive 3. Kruer, W. L. and J. M. Dawson, Sideband instability, Phys. Fluids 13,
assumption which is quickly violated in the collapse process. The rate (le) 2747 (1970).
at which the high frequency field localizes is proportional to its intensity, 4. Dawson, J. M., Nonlinear electron oscillations in a cold plasma, Phys. Rev.
which in turn is proportional to the depth of the cavity in this limit. In 113, 383 (1959).
particular, Iclwp ~ u;h/4v; ~ neR./n. Localization of the fields to the
5. Coffey, T. P., Breaking of large amplitude plasma oscillations, Phys. Fluids

A
cavity requires that the decrease of the plasma wave frequency due to
the depression in density be compensated by the increase due to thermal
.. 14, 1402 (1971).
6. Kruer, W. L., P. K. Kaw, J. M. Dawson and C. Oberman, Anomalous high-
.\6e ~
dispersion i.e., k2 ne/./n. The condition for ignoring ion inertia is
frequency resistivity and heating of a plasma, Phys. Rev. Letters 24, 987
le« kvs. If we substitute the above estimates for le and k, this condition (1970).
becomes ne£! n « Z m/M, where m/M is the electron-ion mass ratio and
7. Kruer, W. L. and J. M. Dawson, Anomalous high-frequency resistivity of a
Z is the ion charge state.
plasma, Phys. Fluids IS, 446 (1972).
The Zakharov equations admit of solitary wave solutions in one di-
8. DeGroot, J. S. and J. 1. Katz, Anomalous plasma heating induced by a very
mension [19-21J. The width of the soliton is related to its amplitude since
strong high-frequency electric field, Phys. Fluids 16, 401 (1973).
the nonlinearity is balanced by thermal dispersion. However, such solitons
9. Dreicer, H., R. Ellis and J. Ingraham, Hot electron production and anomalous
are unstable to two-dimensional perturbations [22,23J. Numerical studies
microwave absorption near the plasma frequency, Phys. Rev. Letters 31,
show a collapse to smaller scale lengths, followed by so-called burn-out
426 (1973).
due to damping by the particles. After burn-out of the high frequency
10. Thode, L. E. and R. N. Sudan, Two-stream instability heating of plasmas by
field, the unsupport ca ity breaks up into ion acoustic waves which
relativistic electron beams, Phys. Rev. Letters 30, 732 (1973).
serve as a seed for ditional coupling in a driven plasma. A variety of
self-similar solutio have been derived to describe the collapse stage, and 11. Kainer, S., J. M. Dawson, and T. Coffey, Alternating current instability
simulations to isola e the collapse have been carried out [24]. produced by the two-stream instability, Phys. Fluids IS, 2419 (1972).

Localized regions intense high frequency fields within cavities have 12. Papadopoulous, K., Nonlinear stabilization of beam plasma interactions by
been observed in a number of experiments in low density laboratory plas- parametric effects, Phys. Fluids 18, 1769 (1975).
mas [25-30J. In some of these experiments [30], the three-dimensional col- 13. Thomson, J. J., R. J. Faehl and W. L. Kruer, Mode-coupling saturation of
lapse of beam-driven plasma waves has been measured. Although much the parametric instability and electron heating, Phys. Rev. Letters 31,
remains to be understood, the general picture of plasma wave turbulence 918 (1973).
as a set of randomly occurring collapsing cavities is clearly a very fruitful 14. Valeo, E. J. and W. L. Kruer, Solitons and resonant absorption, Phys. Rev.
one. Several reviews of the ongoing work on plasma wave collapse and Letters 33, 750 (1974).
Strong plasma wave turbulence are now available [31-35J. 15. Katz, J. 1., J. Weinstock, W. L. Kruer, J. S. DeGroot, and R. J. Faehl,
Turbulently heated distribution functions and perturbed orbit theory, Phys.
Fluids 16, 1519 (1973).
16. Tsytovich, V. N., Nonlinear Effects in Plasma. Plenum. Press, New York
1970.
17. Zakharov, V. E., Collapse of Langmuir waves, Sou. Phys. JETP 35,908
(1972).
112 9 Heating by Plasma Waves 9.6 References 113

18. Morales, G. S., Y. C. Lee and R. B. White, Nonlinear Schrodinger equation 34. Shapiro, V. D. and V. 1. Shevchenko, Strong turbulence of plasma osciUations,
model of the oscillating two stream instability, Phys. Rev. Letters 32,457 ibid, p.123-182. "--
(1974). 35. Russell, D., D. F. DuBois, and H. A. Rose, Collapsing caviton turbulence in
19. Rudakov, L. L, Deceleration of electrons with a high level of Langmuir tur- one dimension, Phys. Rev. Letters 56, 838 (1986).
bulence, Soo. Phys Dokl. 17, 1166 (1973).
20. Kingsep, A. S., 1. L Rudakov and R N. Sudan, Spectra of strong Langmuir
turbulence, Phys. Rev. Letters 31, 1482 (1973).
21. Nishikawa, K., H. Hojo and K. Mima, Coupled nonlinear electron-plasma
and ion-acoustic waves, Phys. Rev. Letters 33, 148 (1974).
22. Denavit, J., N. R Pereira and R. N. Sudan, Two-dimensional stability of
Langmuir solitons, Phys. Rev. Letters 33, 1435 (1974).
23. Degtyarev, L. M., V. G. Nakhankov and 1. L Rudakov, Dynamics of the
formation and interaction of Langmuir solitons and strong turbulence, Sou.
Phys. JETP 40, 264 (1975).
24. Anisimov, S. L, M. A. Berezovskii, M. F. Ivanov, L V. Petrov, A. M.
Rubenchik and V. E. Zakharov, Computer simulation of Langmuir collapse,
Phys. Letters 92A, 32 (1982).
25. Kim, H. C., R L. Stenzel and A. Y. Wong, Development of cavitons and
trapping of RF fields, Phys. Rev. Letters 33, 886 (1974).
26. Antipov, S. V., M. V. Nezlin, E. N. Snezhkin and A. S. Trubnikov, Excitation
of Langmuir solitons by monoenergetic electron beams, Sov. Phys. JETP
49,797 (1
27. Egglesto D., A. Y. Wong and C. B. Darrow, Development of two-
dimensio al structure in cavitons, Phys. Fluids 25, 257 (1982).
28. Cheung, . Y., A. Y. Wong, C. B. Darrow and S. J. Qian, Simultaneous
observation 0 viton formation, spiky turbulence, and electromagnetic
radiation, Phys. Rev. Letters 48, 1348 (1982).
29. Leung, P., M. Q. Tran and A. Y. Wong, Plasma wave collapse generated by
the interaction of two oppositely propagating electron beams with a plasma,
Plasma Phys. 24,567 (1982).
30. Wong, A. Y. and P. Y. Cheung, Three dimensional self-collapse of Langmuir
waves, Phys. Rev. Letters 52, 1222 (1984).
31. Goldman, M. V., Strong turbulence of plasma waves, Rev. Mod. Phys. 56,
709 (1984).
32. Rubenchik, A. M., R Z. Sagdeev and V. E. Zakharov, Collapse versus cavi-
tons, Comm. Plasma Phys. Cont. Fusion 9, 183 (1985).
33. Zakharov, V. E., Collapse and self-focusing of Langmuir waves; in Handbook
of Plasma Physics, VoL II (A. Galeev and R. N. Sudan, eds.), p.81-122.
North Holland, Amsterdam 1984.
 CHAPTER 10
Density
Profile
Modification

l.,

Studies of light-plasma interactions in a plasma with a uniform density

are very valuable for understanding many aspects of the physics and have
some direct applications to heating of low density, magnetically-confined
plasmas, However, laser-produced plasmas are typically rather inhomoge-
neous. The light-plasma interactions take place in an expanding corona
blowing off from an irradiated target. The plasma inhomogeneity affects
the mix of the interaction processes. In turn, the interactions can signifi-
cantly modify the plasma inhomogenity and temperature.
Steepening of the density profile by intense laser light near its critical
density is an important example of the interplay between the plasma and
the light. As normally incident light reflects at the critical density, twice
its pressure is transmitted to the plasma via the ponderomotive force.
The plasma expansion is perturbed, leading to a local steepening of the
density profile. The profile modification can be enhanced for obliquely-
incident, p-polarized light by the ponderomotive force of the resonantly-
generated electrostatic waves. As we will see, the profile modification can

115
116 10 Density Profile Modification 10.2 Steepening of the Density Profile 117

be substantial and can play an important role in determining both the A self-similar solution describing the plasma expansion can be readily
mix and the scaling of interaction processes near the critical density. found from Eqs. (10.2) and (10.4) by letting n = 1~/t) and 1.1 = g(xlt). ~
To gain insight into the profile modification, we will first consider a where I and 9 are functions to be determined. These equations then IQvP.
freely expanding plasma and then develop a simple model for profile steep-
ening by normally incident light reflecting at its critical density. Lastly we I '( =:;X) +Ig , o (lO.S)
will briefly examine some simulations of resonance absorption including
density profile modification.
9
'( 9 -
X)t + 2
Cs fI' = 0, (10.6)

10.1 FREELY EXPANDING PLASMA where the prime denotes the derivative with respect to xlt.
Straightforward manipulations yield 9 = (xlt) + c, and f' I 1= _c;l
We begin by deriving a self-similar solution to describe one-dimensional Hence the self-similar solution is
expansion of a planar, isothermal plasma. We again use the two-fluid
x
equations to describe the electrons and ions. With the neglect of electron 1.1 = Cs + (10.7)
inertia, the electron momentum equation simply determines the electric t
field: n no exp( -~) (10.8)
cst
ne e E = - \7 Pe , (10.1)

where ne is the electron density and Pe the electron pressure. This electric where no is the density at x = O. Note that the density gradient length
field transmits the electron pressure to the ions. increases with time i.e., L = nl( Irn] ox) = cst. Note also that, in a frame
The continuity and force equations for the ion fluid give moving with a point of constant density, 1.1 = Cs. In other words, the
plasma flows through a point of constant density at the sound speed.
on 0
at + ox (nu) = 0 (10.2)
01.1 01.1 ZeE \7Pi 10.2 STEEPENING OF THE DENSITY PROFILE
+ 1.1- (10.3)
at ox M nM' If this expanding plasma is pushed on at a preferred location (for example,
at n = ncr), the density profile will be locally steepened. The simplest
were n and 1.1are the ion density and flow velocity, Z the ion charge
example of this local steepening is that due to the momentum deposition
stat M the ion mass and Pi the ion pressure. We next substitute from
of normally incident light reflecting at its critical density. The basic idea
Eq. (1. . to Eq. (10.3), neglect the ion pressure relative to the electron
is that twice the pressure of the light wave is taken up by the plasma
pressure, and take ne ~ Zn, which is an excellent approximation for length
near the reflection point, and this local momentum deposition steepens
scales much greater than the electron Debye length. With an isothermal
the density profile near the critical density [1].
equation of state for the electrons, Eq. (10.3) becomes
It is instructive to develop a simple model of this profile steepening. 10
01.1 01.1 210n particular, we consider a normally incident light wave reflecting from the
- + 1.1- = -c -- (10.4) critical surface of an isothermal, freely-expanding, collisionless plasma.
at ox • n ox '
Again adopting a two-fluid description and assuming planar geometry,
where c. = (ZTe/M)I/2 is the well-known ion sound velocity and Te is we easily obtain equations for the density (n) and flow velocity (tL) of the
the electron temperature. plasma [2-4]. The analysis parallels that discussed in the previous section,
with the inclusion of the ponderomotive force exerted by the light wave
on the plasma.
 10 Density Profile Modification 10.2 Steepening of the Density Profile 119

If we include the ponderomotive force, Eq. (10.1) becomes

E "- Standing
(10.9) EM wave

where Vw is the velocity of oscilla.tion of an electron in the electric field of

the light wave. Eq. (10.4) now becomes
ncr Density
profile
(10.10)

where c, is the ion sound velocity, Z the ion charge state, and M the ion
mass.
Anticipating that the profile will be steepened from a density n <.
ner to ~ ~ens~ty n2 .> ner, we use Eqs. (10.2) and (10.10) to exp;ess
the variation m density and flow velocity in the frame moving with the
steepened surface. In this frame, we have
f)
o ns to ner with a density scale length of L and express the electric field E
f)x (n u)
Figure 10.1 A schematic of the ponderomotively-steepeneddensity profile.
To make further progress, we must now relate the density at the sonic
point to the critical density (ncr) by considering the solution for the stand-
ing electromagnetic wave. A crude treatment will allow us to obtain ana-
lytic estimates. We approximate the density profile as locally linear from
(10.11) by the well-known Airy function solution discussed in Chapter 3:

!(~2) _c2 ~lnn

S f)x
Zm
4M
f)
f)x
2
Vw . (10.12)
(10.15)
If we normalize the flow velocity to the sound speed and substitute from
Eq. (10.11) into Eq. (10.12), we obtain (
where a is a constant determined by fitting to the incoming light wave.
This locally linear assumption would be a reasonable approximation when-
(1 - u 2
) ~
n Bx
an + ~ (
f)x
v;)
4v~.'
= 0 (1\13))
~
ever vos/ve « 1, where Vos is the oscillation velocity of an electron in the
free space value of the electric field of the light. The assumption fails
:h~re Ve ~ t~e ~lectron thermal velocity. Since the density gradient re- when (vos/ve)2 » 1, since the jump in density becomes too large. Match-
~ finite, It IS clear that the sonic point (u = 1) must be at the ing the peak of the Airy function solution to the field at the sonic point,
lllaxunum of the field ofthe standing light wave i.e., where fJvw/ f)x = o. we then obtain v;(n = ncr) ~ O.44v!ax' In addition, we note that the
Integrating Eqs. (10.11) and (10.13), we then readily obtain . sonic point and the critical point are separated by ~x z: (2L/w2)1/3.
Equation (10.14) then becomes
n;2 + 2 In (~) + v; = 1 + v!ax . (10.14)
n ns 2v~ 2v~
(10.16)
liere ns is the density at the sonic point and Vmax is the value of v at the
~ '. w
tl axunum of the st~ding wave. There are two solutions of Eq. (10.14): 1/3
1 < ns (uJ > 1) which corresponds to a lower density plateau and n2 > where v!ax ~ 3.7v~ ( wL/c ) and ncr/L
~; (U2 < 1), which c~rresponds. to the upper density shelf. A schematic
the steepened density profile IS shown in Fig. 10.1.
120 10 Density Profile Modification 10.3 Resonance Absorption 121

The solutions for nI, n3, and n2 are now-straightforward. (For nl and field are given in Ref. [2].
n2, Vw = 0.) For weak fields (vos/ve .!s 0.1) analytic results can be given: Since a freely expanding, planar plasma flows tlu:.,9ugha point of con-
stant density at the sound speed, the momentum deposition at the critical
:: ~ 1 _ 0.77 (::8) O.B density resonantly perturbs the flow. Hence the profile is steepened over
a significant range of densities even for relatively low intensity light. As
n. ~ 1 + 0.97 (Vos)O.B we will see in the next section, such a profile modification can have a
nl Ve significant effect on the coupling processes near the critical density.
(10.17)
~: "J 1 - 0.97 (:: fB
wL "J 1.5 (vos)-1.2 10.3 RESONANCE ABSORPTION WITH
DENSITY PROFILE MODIFICATION
C Ve

Note that the jump in the density scales as Cl: fractional power of the in- Let us conclude our discussion of density profile modification with a more
tensity. For more intense fields, numerical solutions of the transcendental complicated example, which illustrates the nonlinear interplay between
equations are required. Results for the steepened scale length as a function resonance absorption of an obliquely incident, p-polarized light wave and
of vo. are shown in Fig. 10.2: In one interesting regime that is typical of profile modification. In this case, the steepening of the density profile is
many current applications (0.1 < (v08/ve) < 1.), (wL/c) "J 2{v08/ve)-1. generated both by the pressure of the reflecting, obliquely incident light
Numerical solutions for nl and n2 including the detailed structure of the wave and by the pressure of an intense, resonantly-generated electrostatic
field near the critical density [5-10]. We can see the essential features
of the nonlinear evolution by examining some computer simulations of
resonance absorption.
24 These simulations [5] are carried out with a two-dimensional code
which solves the complete set of Maxwell's equations and includes rela-
tivistic particle dynamics. Plane light waves are propagated from vacuum
18 into an inhomogeneous slab of plasma. Variations are followed both along
the propagation vector of the light and along its electric vector, which
wL allows for resonance absorption and for the generation of parametric in-
-c
12 stabilities. Reflected light waves are allowed to freely pass out of the sys·
tem. Particle boundary conditions are chosen to model a freely expanding
plasma adjacent to a reservoir of constant temperature plasma. The ini-
6 tial density varies with x (the direction normal to the slab) from zero to
a supercritical value. A region of vacuum is included adjacent to the low
density boundary to allow for free expansion of the plasma. Particles im-
1.0
pinging on the high density boundary are replaced with equal incoming
0.0
flux distributed according to vxfm{v), where Vx is the component of the
v
os velocity normal to the boundary and f m (v) is the initial Maxwellian ve-
-. locity distribution. The plasma evolution is followed until a quasi-steady
Figure 10.2 Model predictions for the steepened density scale length as a state has been established.
function of vo./ve. See Estabrook and Kruer, (1983). A typical simulation will again illustrate the principal effects. In this
example, p-polarized light is incident at an angle of 240 onto an initial
 10 Den.sity Profile Modification 10.3 Resonance Absorption 123
122

density profile which rises linearly from 0 to 1.7 ncr in a distance of 3Ao
(where AO is the free space wavelengths). The free space amplitude of the
electric field of the light is eE/mwoc = 0.09, which corresponds to an
'mtensity of 1A6 ~ 1016 W-J-L2/cm'!.. The initial electron temperature is 4
keV, and the ion-electron mass ratio is WO.
After the light wave penetrates to its turning point, an electrostatic
field is resonantly excited at the critical density. The magnitude of this
field initially grows linearly in time, becoming more and more localized to
the critical density surface, as expected from the discussion of resonance
absorption in Chapter 4. Finally the resonantly-driven field becomes suf-
ficiently intense and localized that electrons can be accelerated through it
in one oscillation period, a process called wavebreaking. Physically, wave-
breaking corresponds to the onset of strong electron "trapping" in the
localized oscillating field. At wavebreaking , electrons which enter the os-
cillating field with the proper phase are efficiently heated, taking energy
from the driven field and saturating its growth.
The feedback of these intense fields (and the concomitant localized
heating) on the plasma density profile is a crucial feature of the long-time
evolution of the coupling. The pronounced profile modification is demon-
strated in Fig. 10.3, which shows three snapshots of the density profile
as it evolves from its initial linear profile to a quasi-steady, very steepened
profile. The ponderomotive force due to the intense, localized electrostatic
field ejects plasma, digging a hole in the plasma density at the critical
surface. The plasma ejected towards the vacuum eipands away, leaving
a locally steepened density profile which is supported by the pressure of
both the localized electrostatic wave and the reflecting light wave.
This profile steepening has important consequences for the mix of
absorption processes. In particular, resonance absorption becomes impor-
tant for a wide range of angles of incidence. This effect is demonstrated
in Fig. lOA, which is a plot of the fractional absorption of p-polarized
light (after the profile steepening) versus angle of incidence as computed
in a series of simulations with the same initial plasma conditions as the
sample simulation. Note that the absorption peaks at about 50% for a
sizable angle of incidence (Bmax ~ 24°) and is quite large over a broad
range of angles (LlB'" Bmax). This is qualitatively as expected from our
simple theoretical discussion of resonance absorption. In addition, para-
metric instabilities near the critical density (discussed in the previous
0.6 1 I I
•
0.5 - -

•
0.4 f- -
•
A 0.3 r- • -

0.2 f- -

•
0.1 f- -

0 I I I

0 15 30 45 60
Figure 10.3 The ion density profile at three different times from a simulation e (degrees)

of resonance absorption: (a) the initial profile, (b) the profile after the resonantly- Figure 10.4 The fractional laser light absorption after profile steepening ver-
driven field has grown, and (c) the asymptotic profile which shows a characteristic sus angle of incidence as computed in a series of simulations. See Estabrook et
step-plateau feature. See Estabrook et al., (1975). al., (1975).
 10 Density Profile Modification
124
chapter) are strongly limited, since there's a very small region. of plasma
in which these instabilities can operate. Note that the absorptlOn IS only
about 15% for normally incident light.
Finally, the profile steepening strongly reduces the heated electron
energies due to the resonantly-generated wave. As the wave beco.meslarge
enough to nonlinearly interact with the electrons, a small fraction of the
electrons (those entering the wave with the proper phases) are s~ron~ly
heated to an effective temperature of order mv~ where Vw is the oscillation
velocity of an electron in the resonantly-driven wave (vw. = e~/mwo~. As
is apparent from our discussion of nonlinear wave-partIcle mteractlOns,
the resonantly-driven field decreases in amplitude as the ~rofile steepens.
Physically the wave then has a smaller spatial extent which corresponds
to a lower' effective phase velocity. Hence it "traps" electrons at a lower
amplitude and heats them to a lower ener~. .
As expected, the heating via the localized electron. plasma oscilla-
tion produces a population of suprathermal electrons. FIgure 10.5 shows
u.
Q)
Electron energy - keV
Figure 10.5 The heated electron distribution function from a simulation of
resonance absorption.
10.3 References
125
the electron distribution computed in the steepened, nonlinear state in a
sample simulation. The distribution is composed of &,[elatively cold main
body plus a quasi-Maxwellian heated tail.
Self-consistent steepening of the density profile can play an important
role in many other laser plasma processes. For example, the two-plasmon-
decay instability occurs for a narrow range of densities near one-fourth the
critical density. As will be discussed in the next chapter, a local steepen-
ing of the profile can help limit this instability. Calculations of collisional
absorption must also take profile steepening into account. Even neglect-
ing ponderomotive forces, the density profile is modified by temperature
changes driven by the localized heating which occurs on a length scale
comparable to the collisional absorption length. This ablative steepening
depends on the details of the electron transport.
References
L Kidder, R. E., Interaction of intense photon beams with plasmas (II);
in Proceedings of the Japan- U.S. Seminar on Laser Interaction with·
Matter, (C. Yamanaka, ed.), Tokyo International Book, Tokyo, 1973.
2. Lee, K., D. W. Forslund, J. M. Kindel and E. L. Lindman, Theoretical
derivation of laser-induced plasma profiles, Phys. Fluids 20, 51 (1977).
3. Takabe, H. and P. Mulser, Self-consistent treatment of resonance absorption
in a streaming plasma, Phys. Fluids 25, 2304 (1982).
4. Estabrook, K. and W. L. Kruer, Parametric instabilities near the critical
density in steepened density profiles, Phys. Fluids 26, 1888 (1983),
5. Estabrook, K. G., E. J. Valeo and W. L. Kruer, Two-dimensional relativistic
simulations of resonant absorption, Phys. Fluids 18, 1151 (1975).
6. Forslund, D., J. Kindel, K. Lee, E. L. Lindman and R. L. Morse, Theory and
simulations of resonant absorption in a hot plasma, Phys. Rev. All, 679
(1975).
7. DeGroot, J. S. and J. Tull, Heated electron distributions from resonant ab-
sorption, Phys. Fluids 18, 672 (1975).
8. Albritton, J. R. and A. B. Langdon, Profile modification and hot electron
temperature from resonant absorption at modest intensity, Phys. Rev. Lett.
45, 1794 (1980).
9. Chen, H. H. and C. S. Liu, Soliton generation at resonance and density
modification in laser-irradiated plasma, Phys. Rev. Lett. 39, 1147 (1977).
10. Morales, G. J. and Y. C. Lee, Generation of density cavities and localized
electric fields in a nonuniform plasma, Phys. Fluids 20, 1135 (1977).
 CHAPTER 11
.•..

N onlinear Features of
Underdense Plasma
Instabilities

In Chapter 10, we discussed some nonlinear phenomena in the neighbor-

hood of the critical density. Let us now examine some nonlinear features
of of the light-driven instabilities which can take place in a plasma whose
density is significantly below the critical density. In particular, we consider
the Brillouin, Raman, two-plasmon decay, and filamentation instabilities.
These processes can become significant when large regions of underdense
plasma are produced, as is expected for reactor targets in laser fusion. As
we will see, both the absorption and the preheat can be strongly affected.

11.1 NONLINEAR FEATURES OF BRlLLOUIN

SCATTERING
Let us first consider Brillouin scattering. This scattering can most simply
be described as the resonant decay of an incident photon into a scattered
photon plus an ion sound wave. Hence

Wo = Ws + Wi, (11.1)

127
 11 Nonlinear Features of Underdense Plasma Instabilities 11.1 Nonlinear Features of Brillouin Scattering 129
128

where wo(w.) is the frequency of the incident (reflected) light wave and wher~ th~ reflectivity r == £;(O)/E;(O). Defining y =;=.Er(x)IE;(O) and
Wi is the frequency of the ion sound wave. As is apparent from the fre- substituting from Eq. (11.5) into Eq. (11.3), we obtain
quency matching conditions, this process occurs throughout the under-
dense plasma. In addition, since Wo » Wi, nearly all the energy of an ay w~e 8n /r-------
incident photon undergoing this process is transferred to the scattered
-8
x
= - 4k
OC
2 -
no
V y2 + 1 - r (11.6)

photon. Hence Brillouin scattering can significantly impact the absorp-

The solution to this standard differential equation is
tion.
As shown by linear theory, gradients in either expansion velocity or
density inhibit the onset of Brillouin scattering. Intensity thresholds due
to gradients have been discussed in Chapter 8. These thresholds are very
y = ~'Sinh
.

[
-Wpe
---(x-xo)
2
4koc2 no
Sn.
1' (11.7)

useful for identifying regimes for which Brillouin scatter is not a concern.
However these threshold intensities are often far exceeded, particularly where Xo is a constant ?f integration which is determined by noting that
when large regions of underdense plasma are irradiated. The nonlinear at x = 0, y = Jr. We finally obtain the reflectivity by assuming that
behavior [1-9] of this instability then becomes an important issue. y c:= 0 at x = L i.e., Er(L) » E;(O). Then
To illustrate nonlinear aspects of the Brillouin instability, let us con-
sider a very simple model problem: the backscattering of a light wave
propagating through a slab of underdense plasma with a uniform density.
r = tanh 2 ( w;"L 8n ) (1l.8)
4ko2 no .
If we postulate that the density fluctuation associated with the ion wave
is nonlinearly saturated at some value bn, we can readily calculate the
In order to illustrate the magnitude of the reflectivity, let's estimate
reflectivity [2]. The wave equation which describes the propagation of a ~helevel t.owhich the density fluctuation can be nonlinearly driven. Strong
light wave with amplitude E through a plasma with density ne is Ion trapping (or wave breaking) is one effect commonly invoked to limit
the ion wave amplitude. The basic idea has already been discussed in
(11.2)
Chapter 9 for electron plasma waves. As the amplitude of the ion wave
~creases, its po~ential becomes large enough to nonlinearly bring ions
into resonance WIth the waves. Since such ions are efficiently accelerated
where w2 = 47rnee2/m. We decompose E into an incident and reflected
e by the wave, a strong damping results, which serves to restrict the ion
part with slo . ~g amplitudes E; (x) and Er (x), respectively, and
wave. ~p~tu~e from further increase. If the ions are cold, the trapping
let ne = no 8nsin(kix - Wit). If we substitute ne into Eq. (11.2) and
condition IS SImply Zed: = Mv~/2, where <jJ is the potential, M the ion
assume fr uency and wave number matching, we obtain
mass, ~d vp the phase velocity of the wave. Neglecting Debye length
se; CO~ect.lOns,the trapping condition corresponds to 8nlno c:= e<jJIBec:= 1/2,
_ w~e 8n E, (11.3)
= which IS a large amplitude.
ax 4ko2 no
It is important to realize that even a small ion temperature signifi-
ee. _ w~e 8n Er . (11.4) c~tly red~ces the trapping amplitude. This temperature effect is readily
=
ax 4ko2 no estimated If one assumes a so-called waterbag velocity distribution for the
ions. In one-dimension, such a distribution is constant with velocity be-
Since Wi « Wo, we have approximated k; = 2ko, where ko is the wave
tween ±J3v; (Vi is the ion thermal velocity) and zero elsewhere. Sincethe
number of the incident light wave. majority of the ions in a Maxwellian distribution have velocities < 2v'
A conservation law is apparent (aE; 1ax = aE; 18x), which then gives
the waterbag distribution gives a reasonable first approximation for the
E; - E; = E;(O) (1 - r) , (11.5)
r
~ -e. rh

E----;;.
l' M ~2.
2-
4-lT
c
=?
d I'\.. i!: e.
._C '-"
d'"
/'If\.. 0
~

V 1/ " .
 11.1 Nonlinear Features of Brillouin Scattering 131
11 Nonlinear Features of Underdense Plasma Instabilities

onset of strong trapping. Trapping now occurs when the fastest ion is
0.8 0.5
nonlinearly brought into resonance with the wave i.e.,

Ze4> M (
2" Vp - V3 i )2
U 0.6

(11.9)
8n
no
-
1
2
(F¥£Bi
1+- - {¥£Bi)
ZBe -ZBe 2
0.4

Here 8i(8e) is the ion (electron) temperature, Z is the ion charge, and 0.2
the Debye length correction to the phase velocity has been neglected. For
8;jZ8e = 0.2, Eq. (11.9) predicts a fluctuation amplitude of 8n/no z: 0.12.
Clearly the ion temperature serves to significantly reduce the amplitude, o~--~----~--~ 0.2 '-_-'----_---l'--_-l

but note that the trapping amplitude is still of order 10%, unless the ions
o 1.0 2.0 3.0 1.0 2.0 3.0

are quite hot (i.e., B;/ZBe '" 0(1)). Strong trapping does not, in general, wot.l0' Wo t, 10'

limit the fluctuation amplitude to a small value.
Using the ion trapping estimate for bn, we can now calculate a reflec-
tivity. As an example, consider a 30AOslab of plasma with a uniform den-
sity of no = 0.33ncr and an ion-electron temperature ratio of BelZBi= 0.2.
Here AOis the free-space wavelength of the light. Substituting from Eq.
(11.9) into Eq. (11.8), we obtain r '::::.94%. Even a modest fluctuation
amplitude can lead to a sizable reflectivity in a large underdense plasma.
Wavebreaking (or strong trapping) arguments only give an estimate of
the amplitude at which a strong damping onsets due to wave-particle in-
teractions. When a significant number of ions are accelerated and sizeable
tails develop on the ion distribution function, a more complex description
of the nonlinear wave . icle interaction is needed. To gain insight into
these effects, let's n consider some computer simulations of Brillouin
scattering.
Figure 11.1 sh s the temporal evolution of the Brillouin back reflec-
tion and the energy' the ions computed with a one-dimensional code [3]
which treated the io icles and the electrons as a fluid. In this ex-
ample, light with an intensity of IA5 = 3 x 1015 W -J.L2 I cm2 is incident onto
a 30AOslab of plasma with an initially uniform density of no = 0.33ncn
an electron temperature of 3 keY, and an ion-electron temperature ratio
of 0.2. In the simulation, ions reaching the right plasma boundary were
re-emitted with the initial thermal temperature, modeling transport of
heated ions to a higher density plasma. As shown in Fig. 11.1(a), the re-
flectivity rapidly proceeds to a large level of about 65%, as the unstable
Figure 11.1 Evolution of the reflectivity (time-averaged over many cycles)
and the mean ion energy from a computer simulation of Brillouin backscatter.
ion wave grows to a large amplitude and traps ions. However, concomitant
with this large reflectivity is a substantial heating of the ions as shown in
Fig. 11.1(b). In detail, this heating consists of the formation of a sizable
tail of energetic ions as expected from our discussion of collisionless wave
particle heating in Chapter 9. This strong self-consistent distortion of the
ion velocity distribution in turn enhances the damping of the ion wave
and lowers its amplitude. Finally a quasi-steady state is reached in which
the heating of ions by the wave is balanced by their transport out of
the underdense plasma. The reflectivity drops to a more modest value of
about 25%.
This ion heating or energetic tail formation is an intrinsic feature
of the Brillouin scattering. In the scattering process a fraction wi/wo of
the reflected light energy is deposited into the ion wave and then into
the heated ions when the wave damps. We first estimate an effective
v;
tem~erat~e of the heated ions as M /2 which is the energy of an ion
moving With the phase velocity (vs) of the wave. The density nh of the
ion tail is then estimated by balancing the energy flux into the ion wave
with that carried away by the heated ion tail; i.e.,
Wi
r I -- (11.10)
Wo
132 11 Nonlinear Features of Underdense Plasma Instabilities
Here the flux carried by the heated ions has been described in a free-
streaming limit. For the example discussed above, Eq. (11.10) predicts
(nh/no) ~ 0.4 in the nonlinear state, which compares well with the sim-
ulation.
The ion heating is a clear manifestation of the damping of the ion
wave in the nonlinear state. If we represent the scattering as a reflection
from a heavily damped wave, the reflectivity can be readily calculated
in terms of this damping. A heuristic estimate for the damping in the
nonlinear state is Landau damping on the self-consistent ion tail (or on
the heated main body when the heating is very strong). Estimates of the
reflectivity obtained in this way [2,4] compare favorably with simulation
results, as well as illustrate some of the gualitative trends. For example,
for a given size of underdense plasma, the reflectivity tends to saturate
with intensity. The increase in light intensity is balanced by an increase
in the self-consistent damping associated with the greater ion heating.
Finally there are other nonlinear mechanisms for limiting the Brillouin
instability. Harmonic generation [4-6], quasi-resonant decay of ion waves
[7], nonlinear frequency shifts [8,9], and profile steepening can play a sig-
nificant role in some regimes. Quantitative calculations [10,11] ofBrillouin
scattering also require consideration of the noise sources as well as of the
detailed profile of the density and expansion velocity of the plasma. A
significant complication is that partial reflection of the light wave from
the critical surface can serve as a noise source in an expanding plasma
[12,13]. The angular distribution of the scattering and the competition
with other underdense plasma processes such as inverse bremsstrahlung
and filamentation are other important issues.
11. NONLINEAR FEATURES OF RAMAN
SCATTERING
To comp ent this brief discussion of Brillouin scattering, let us now
consider some nonlinear aspects [1,14-20] of the Raman instability. The
Raman instability can be most simply characterized as the resonant decay
of an incident photon into a scattered photon plus an electron plasma
wave. The frequency and wave number matching conditions then are
Wo = w~ + Wpe, (11.11)
11.2 Nonlinear Features of Raman Scattering 133
where Wo (w~) and ko (ks) are the frequency and wave number of the
incident (scattered) light wave, and wpe (kp) is the ~~ency (wave num-
ber) of the electron plasma wave. Since the minimum frequency of a light
wave in a plasma is wpe, the electron plasma frequency, it is clear that
this process requires that Wo ;G 2wpe i.e., n:;; ncr/4, where n is the plasma
density and ncr is the critical density. In this process, part of the incident
energy is scattered, and part is deposited in the electron plasma wave.
This latter portion of the energy in turn will heat the plasma as the elec-
tron plasma wave damps. Since the plasma frequency is much greater
than the ion acoustic frequency, the Raman instability is clearly not as
efficient in scattering the incident laser light as is the Brillouin instability.
However, the electron plasma wave which is generated can have a high
phase velocity (of order c) and so can produce very energetic electrons
when it damps. Since such electrons can preheat the fuel in laser fusion
applications, the Raman instability is a particularly significant concern.
As discussed in Chapter 7, the intensity threshold due to a density
gradient is rather high but can clearly be exceeded in a large underdense
plasma irradiated with intense light. Again, we will use simulations to
give us some estimates of what to expect in the nonlinear regime. To most
simply explore the nonlinear effects, consider a 1-~ dimensional particle
simulation [14] in which laser light with intensity I A5 = 2.5 x 1015 w-
p.2/cm2 is propagated through a 127Ao region of plasma with a uniform
density of O.lncr, an electron temperature of 1 keY, and an electron-ion
temperature ratio of three. In this simulation, the back reflection due to
the Raman instability builds up to about 15%, accompanied by strong
tail heating of the electrons by the electron plasma wave associated with
the scatter. (There is also a modest back reflection of ~ 20% due to
the Brillouin instability in this example.) The resulting heated electron
distribution is shown in Fig. 11.2. Note the heated tail, which is roughly
Maxwellian in shape with a characteristic temperature eH::: 13 keY. Such
energetic tail formation is characteristic of heating via a large amplitude
electron plasma wave. A useful rule of thumb estimate for the heated
temperature found in these strongly-driven simulations is eH ~ mv2/2,
which is simply the energy of an electron accelerated to the phase velocity
vp of the plasma wave. As can be seen from the frequency and wave
number matching conditions, such a temperature depends on both the
density and background electron temperature and can easily be of order
50-100 keY, even for backscatter.
 11.3 Two-Plasmon Decay and Filw:nentation 135
11 Nonlinear Features of Uoderdense Plasma Instabilities

reflectivity is sufficient to drive the instability into a regime in which a

significant fraction. of the electrons are resonant with. the plasma wave.
A ball-park estimate of the Raman back reflection in the sample sim-
ulation can be obtained by estimating the damping of the plasma wave
104 associated with the hot electron generation. If the damping is crudely
s Raman heated modeled as Landau damping on the heated tail, the reflectivity due to
g
this damped plasma wave is readily calculated, giving r := 10% for the
electrons
u

'"
0; 103
above example. This back-of-the-envelope model illustrates some impor-
a
0; tant features of the nonlinear evolution: the self-consistent generation of
.0
E hot electrons and their feedback on the instability.
:J 102
Z Ion fluctuations can also play an important role [16-21] in Raman
scattering. The ion fluctuations are produced either by the Brillouin in-
10' stability or by collapse of the Raman-generated plasma wave. As discussed
in Chapters 6 and 9, an ion fluctuation efficiently couples plasma waves
provided 8n/n > tlw/wpe, where 8n is the amplitude of the density fluc-
10°
40 60 80 tuation and tlw is the frequency mismatch between the plasma waves.
0 20
Electron energy (keV)
This energy transfer from the Raman-driven plasma wave into shorter
wavelength plasma waves both reduces the level of the primary wave
Figure 11.2 The heated electron distribution from a computer simulation of and produces less energetic heated tails. Frequency shifts in the primary
Raman backscatter. See Estabrook et al., (1980). plasma wave due to the ion waves may also be significant.
Even when the Raman instability is not operative, an incident light
wave can still undergo stimulated scattering on the electrons [22,23].
This self-consistent electron heating is a significant feature of the non-
There also are important multi-dimensional effects, including Raman side-
linear evolution and can play a role in restricting the scatter. It is instruc-
tive to estimate the size of the heated tail in the nonlinear state by bal- scattering and filamentation of the incident light wave. Two-dimensional
simulations [24] using a very intense beam of light emphasize the im-
ancing the energy flux deposited in the electron plasma wave (and hence
into heated electrons when the wave damps) with the energy flux carried portance of Raman sidescattering and even show filamentation due to
away by the hea tail. Neglecting background thermal effects and using relativistic effects. The latter is accentuated by self-generated magnetic
fields driven by the Weibel instability of the heated electrons.
a free-str estimate of the hot electron transport gives

(11.12) 11.3 NONLINEAR FEATURES OF THE

TWO-PLASMON DECAY AND
FILAMENTATION INSTABILITIES
Where r is the reflectivity and nh is the tail density. If we consider
backscatter, asSUIDewpe/WO » 1/2, and use our estimate of eH, we obtain Let's now consider another instability involving electron plasma waves:
the two-plasmon decay or 2wpe instability. As discussed in Chapter 7, this
tu,
- ~ 19r
(vos)2
-
(ncr)2
- (11.13)
instability represents the resonant decay of a light wave into two electron
plasma waves. The frequency matching condition clearly requires that
n c n
cr
Wo := 2wpe i.e., n := n /4. The feedback mechanism leading to instability
~here n is the plasma density, Vos
is the oscillation velocity of the electrons is similar to that already discussed for the Raman instability, except now
Ill. the light wave, and c is the velocity of light. Clearly even a modest
 11 Nonlinear Features of Underdense Plasma Instabilities 11.3 Two-Plasmon Decay and Fila.rn.entation 131
136

both growing waves are electron plasma waves. The m~um !?:owth linear evolution. The importance of ion density fluctuations is also em-
rate in a uniform plasma is the same as that for the Raman instability at phasized by the nonlinear theory [27-29]. '\.
ncr/4, but now a broad spectrum of plasma waves is unstable. . . Finally we conclude with a brief discussion of the filamentation insta-
Because the 2wpe instability is confined to a narrow range of densities bility, which can have an important effect on the mix of coupling processes.
near ncr/4, a local nonlinear steepening of the density profile can play As discussed in Chapter 6, this instability represents the development of
an especially important role in the nonlinear evolution. A density pro~e filamentary structure in the intensity profile of a light wave. The insta-
from an illustrative simulation [25,26] of the 2wpe instability is shown ID bility occurs throughout the underdense plasma and is related to whole-
Fig. 11.3. In the two-dimensional simulation, laser light with an intensity beam self-focusing. Both filamentation and whole-beam self-focusing can
of I)"~ = 1016 W_J.L2/cm2 is incident onto an initially inhomogeneous be driven by ponderomotive, thermal, or even relativistic effects [30]. The
plasma slab. The initial electron temperature is 1 k~V, an~ the electron- processes can be accentuated by resonantly-enhanced fields [31]. For sim-
ion mass ratio is 0.01. Note the pronounced steepenmg which takes place plicity, we will here concentrate on ponderomotive filamentation.
near ncr/4 due to the instability-generated plasma waves. . To illustrate the rich possibilities introduced by filamentation, con-
In the simulation the instability occurs in bursts, as the density profile sider a two-dimensional simulation [32] in which the temporal evolution
steepens and relaxes: The averaged absorption in the steepened profile is of an intense light wave is followed in a doubly-periodic plasma. The
modest (of order 10%). During periods of instability generation, hot elec- background plasma density is 0.31ncr, the electron temperature is 4 keY,
tron tails are formed with an effective temperature of about 100 keV for and the intensity of the light wave is I ~ ~ 2.3 X 1016 W-J.L2/cm2. The
this strongly-driven example. Ion fluctuations driven by beating of the electron-ion mass ratio is 0.01, and the ion temperature is large in order
unstable plasma waves are observed to play an important role in the non- to suppress the competing effects of the Brillouin instability. A small si·
nusoidal density modulation perpendicular to the direction of the wave
0.40
propagation and along the direction of the electric field of the light wave
serves as an initial perturbation for the growth. This density perturbation
0.35 and the corresponding modulation in the intensity of the light wave grows
"
~ in time. When the density in the channel has been depressed to about 0.25
0.30
,;: ncr, the laser light decays into intense electrostatic fields, which in turn
heat the electrons. A contour plot of the electrostatic potential in the
0.25
simulation at this time is shown in Fig. 11.4. Note that the electrostatic
.
~c 0.20 fields are concentrated in the channel. In other simulations, another type
of decay analogous to stimulated Raman scattering was observed.
0.15
Profile at Wo t = 750 The competition of filamentation with other processes is a very rich
0.10
topic. In some two-dimensional simulations [33], it has been found that
intense Brillouin sidescattering can suppress filamentation. In other calcu-
0.05 lations [34], self-focusing of a light wave was arrested by intense Brillouin
backscattering which onset as the intensity of the light wave increased. In
0.0
5 10 15 20 25 30 35
addition, calculations have shown that filaments can be unstable [35] to
kox bending along the direction of their propagation. A general picture ofthe
role of filaments in laser plasma interactions has not yet emerged.
Figure 11.3 A density profilefrom simulation of the 2wpe instability. Langdon
In summary, in this chapter we have illustrated some important ef·
et al. (1979). fects produced by intense laser light in a plasma with a density belowthe
 11.3 References 139
11 Nonlinear Features of Underdense Plasma Instabilities

References
I I I I
l. Forslund, D. W., J. M. Kindel, and E. L. Lindman, Plasma simulation studies
of stimulated scattering processes in laser-irradiated plasmas, Phys. Fluids
-
15 f-
18, 1002 (1975).
~ 2. Kruer , 'vii. L., Noulinear estimates of Brillouin scatter in plasma, Phys.

-
r=---- -.... --
...--/
Fluids 23, 1273 (1980).
3. Kruer , W. L., E. J. Valeo, and K. G. Estabrook, Limitation
scattering in plasmas, Phys. Rev. Lett. 35, 1076 (1975).
of Brillouin

~y
10
~G-
--
~
~
4. Kruer , W. L. and K. G. Estabrook, Nonlinear behavior of stimulated scatter
in large underdense plasmas; in Laser Interaction and Related Plasma
Phenomena, Vol. 5, p.783-800 (H. Schwarz, H. Hora, M. Lubin, and B.
Yaakobi, eds.). Plenum Press, Ne~ York, 1981.
:---: =-
p===== 5. Silin, V. P. and V. T. Tikhonchuk, Nonlinear saturation of 5MBS in a rarefied
C

nonisotherrnal plasma, JETP Letters 34, 365 (1981).

- <z:.>
-
~~
6.Heikkinen , J. A., S. J. Kartunnen, and R. R. E. Saloman, Ion acoustic non-
linearities in stimulated Briilouin scattering, Phys. Fluids 27, 707 (1984).
5 I---
7. Kartunnen, S. J., J. N. McMullin, and A. A. Offenberger, Saturation of
stimulated Brillouin scattering by ion wave decay in a dissipative plasma,
Phys. Fluids 24, 447 (1981).
~
- 8. Casanova, M., G. Laval, R. Peilat, and D. Pesme, Self-generated loss of

-
-
- I
-
coherency in Brillouin scattering and reduction of reflectivity, Phys. Rev.
Lett. 54, 2230 (1985).
9. Ikezi, H., K. Schwarzeneggar, A. L. Simons, Y. Ohsawa, and T. Kamimura,
3 4
o 2
Nonlinear self-modulation of ion-acoustic waves, Phys. Fluids 21, 239
wpe X
(1978).
C
10. Ramani, A. and C. E. Max, Stimulated Brillouin scattering in an inhomoge- ;t
Figure 11.4 Contour plot of the electrostatic potential in the simulation when neous plasma with broad bandwidth thermal noise, Phys. Fluids 26, 1079 j

channel density has been depressed to about ncr / 4. From Langdon and Lasinski
(1975).
critical density. Our discussion emphasizes that crucial features of the cou-
pling can depend on the size of the underdense plasma in a laser-irradiated
target. The possible consequences of sizable regions of underdense plas~a
include significant degradation of the absorption and/or the generatlOn
of very energetic electrons, which can complicate fusion target design.
(1983).
11. Colombant, D. G. and W. M. Manheimer, A model of anomalous absorption,
backscatter and flux limitation in laser-produced plasmas, Phys. Fluids I'
13, 2512 (1980). I
!
12. Randall, C. J., J. R. Albritton, and J. J. Thomson, Theory and simulation
of stimulated Brillouin scatter excited by non absorbed light in laser fusion
systems, Phys. Fluids 24, 1474 (1981).
13. Rand all , C. J. and J. R. Albritton, Chaotic nonlinear stimulated 8rillouin
scattering, Phys. Rev. Lett. 52, 1887 (1984).
14. Estabrook, K. G., W. L. Kruer, and 8. F. Lasinski, Heating by Raman
backscatter and forward scatter, Phys. Rev. Letters 45, 1399 (1980).
 11 Nonlinear Features of Underdense Plasma Instabilities 11.3 References 141
140

15. Biskamp, D. and H. Welter, Stimulated Raman scattering from plasmas 29. Shapiro, V. D. and V. L Shevchenko, Strong turbulence of plasma oscillations;
irradiated by normally and obLiquely incident laser light, Phys. Rev. Lett.
N. Sudan, eds.). North Holland, Amsterdam, 1984.
t
in Handbook of Plasma Physics, Vol II, p.123-18 lA. A. Galeev and R.
34, 312 (1975).
16. Bonnaud, G., Ion mobility influence on stimulated Raman scattering in homo- 30. Max, C. E., Physics of the coronal plasma in laser fusion fusion targets;
geneous laser-irradiated plasma, Laser and Particle Beams 5, 101 (1987). in Laser-Plasma Interaction, (R. Balian and J. C. Adam, eds.). North
Holland, Amsterdam, 1982.
17. Estabrook, K. and W. 1. Kruer, Theory and simulation of one-dimensional
Raman backward and forward scattering, Phys. Fluids 26, 1892 (1983). 31. Joshi, C., C. E. Clayton, and F. F. Chen, Resonant self-focusing of laser light
in a plasma, Phys. Rev. Lett. 48, 874 (1982).
18. Aldrich •. C. H., B. Bezzerides, D. F. DuBois, and H. A. Rose, Langmuir
nucleation and collapse in stimulated laser light scatter, Comm. Plasma 32. Langdon, A. B. and B. F. Lasinski, Filamentation arid subsequent decay of
Phys. 10, 1 (1986). laser light in plasrnas, Phys. Rev. Lett. 34, 934 (1975).
19. Rozmus, W., R. P. Sharma, J. C. Samson, and W. Tighe, Nonlinear evolu- 33. Estabrook, K. G., Critical surface bubbles and corrugations and their impli-
tion of stimulated Raman scattering in homogeneous plasmas, Phys. Fluids cations to laser fusion, Phys. Fluids 19, 1733 (1976).
(1987). 34. Rand all, C. J., Simultaneous self-focusing and Brillouin backscattering of
20. Barr, H. C. and G. A. Gardner, Harmonic emission from the quarter critical Gaussian laser beams; in Lawrence Livermore National Laboratory
density surface of laser-produced plasmas; in Proceedings of the Interna- UCRL -50021-79, p.3-45 (1980).
tional Conference on Plasma Physics Vol I1, p.265 (Q. '!'ran and R. J. 3.5. Valeo, E. J., Stability of filamentary structures, Phys. Fluids 17, 1391
Verbeek, eds.). Ecole Polytechnique de Lausanne, Lausanne, 1984. (1974).
21. Barr, H. C. and F. F. Chen, Raman scattering in a nearly resonant density
ripple, Phys. Fluids 30, 1180 (1987).
22. Ott, E., W. M. Manheimer, and H. H. Klein, Stimulated Compton scattering
and self-focusing in the outer regions of a laser fusion plasma, Phys. Fluids
17, 1757 (1974).
23. Lin, A. T. and J. M. Dawson, Stimulated Compton scattering of electromag-
netic waves in plasma, Phys. Fluids 18, 201 (1975).
24. Forslund, D. W., J. M. Kindel, W. B. Mori, C. Joshi, and J. M. Dawson,
Two-dimensional simulations of single-frequency and beat-wave laser-plasma
heating, Phys. Rev. Lett. 54, 558 (1985).
-,
25. Langdon, A. B., B. F. Lasinski and W. 1. Kruer, Nonlinear saturation and
recurrence of the two-plasmon decay instability, Phys. Rev. Letters 43, 133
(1979).
26. Langdon, A. B., B. F. Lasinski, and W. L. Kruer, ZOHAR simulations of
two-plasmon-decay; in Lawrence Lioermore National Laboratory UCRL-
50021-85, p.2-43 (1986).
27. Chen, H. H. and C. S. Liu, Soliton formation and saturation of decay insta-
bility of an electromagnetic wave into two plasma waves, Phys. Rev. Lett.
39, 881 (1977).
28. Kartunnen, S. J., Saturation of parametric instabilities by the nonlinear decay
of electrostatic daughter wave, Plasma Phys. 22, 151 (1980).
 CHAPTER 12
't,.

Electron
Energy
-Tr anspor t

In the previous chapters, we have discussed a number of different processes

by which laser light heats the plasma surrounding a laser-irradiated tar-
get. As we have seen, the energy is deposited primarily into electrons. The
rate at which the electrons in turn transport this energy to the higher den-
sity colder plasma determines both the efficiency of the implosion and the
plasma conditions in the region of deposition. Not surprisingly, the trans-
port of large fluxes of energy in inhomogeneous plasmas is itself a rich
and complex topic. We will begin with the classical diffusive calculation
of electron heat conduction and then discuss its extension to multigroup,
flux-limited diffusion. Lastly, we will indicate some of the additional com-
plications due to self-generated magnetic fields or ion acoustic turbulence
and conclude with a brief summary of what experiments have indicated.

143
 12 Electron Energy Transport 12.2 Multigroup Plux-Lirnit ed Diffusion 145
144

12.1 ELECTRON THERMAL CONDUCTIVITY 'Ne next assume that electron-electron collisions maintain the zero-order
distribution fo(v) as a Maxwellian with a local teIIliler'ature Be(z):
It's instructive to begin with the classical Spitzer-Harm calculation [1,2]
of electron thermal conductivity in a plasma with no magnetic fields.
Neglecting hydrodynamic motion and density gradients, we start with n [mv2] (12.6)
fo(v) = (27T)3/2 v~(z) exp - 2Be(z) ,
the kinetic equation introduced in Chapter 5:

where v;(z) = Be(z)/m. Equation (12.5) is now readily integrated to give

-afat + v . -af -
eE . -af = A- a . [v21-
- 3'
vv af]
~_ + Cee(f) . (12.1)
&x m fJv fJv v uv
e E = _~ aBe(z) (12.7)
2 az
In this expression, f is the electron velocity distribution function, A =
(2rinZe4/m2) lnA, and Cee(f) denotes a similar but more complex opera- Hence fr becomes
tor giving electron-electron collisions. Further, n is the plasma density, Z
the ion charge-state, and In A is the Coulomb logarithm. In order to carry
heat , the distribution function must be warped in the direction of the
h
= fo 4mA
v
4
[8v;(z) -
v
2

v:(z)
] see
az .
(12.8)
heat flow. Adopting spherical coordinates, we represent the distribution
function as the first two terms of an expansion in Legendre polynomials: The heat flow Q is obtained by using fr (v) to evaluate

f = fo(v) + fr(v)cosB, (12.2)

(12.9)
where v =JvJ and B is the angle between v and the direction of the heat
flow (also the direction of E). Substituting for f(v) into Eq. (12.1) and After straight-forward integration, we obtain,
collecting the terms proportional to cos B, we obtain an equation for f 1 (v):
ee, (12.10)
Q -K--

a fr + v a fo _ !... E a fo = _ 2A fr . az '
(12.3)
at Bz m av v3
K- =
4 B~/2
(12.11)
Here we have assumed for simplicity a high Z plasma and neglected the
effect on fr (v) of electron-electron collisions relative to electron-ion ones. Note that the thermal conductivity K- is independent of density and is
The steady-state solution is proportional to B~/2. A convenient representation is K- -:= 14nv;/vei, where
Vei is the collision frequency which describes collisional damping of a light
(12.4) wave as discussed in Chapter 5.
For small to moderate values of Z, it becomes important to directly
include electron-electron collisions which, of course, reduce the conduc-
The electric field is determined by the condition that charge neutrality be tivity. A simple approximation to the numerical results of Spitzer-Harm
J
preserved i.e., Jz = -e jJ-fr jJ-V dv = 0, where jJ- = cos B. This condition is given by multiplying K. given in Eq. (12.11) by g(Z) -:= (1 + 3.3/Zt1
gives
{CO dv v7
lo
(eEmv afoBv _ afo)
az
= o. (12.5)
 12 Electron Energy Transport
MULTIGROUP FLUX-LIMITED DIFFUSION
Although an instructive point of departure, this classical calculation of
diffusive heat flow needs to be extended in several important ways. First,
since the mean free-path is energy-dependent, a conductivity which av-
erages over a distribution of disparate velocities is clearly inadequate to
properly treat the transport of energy into a target. Hence the heat trans-
port is usually modeled as multigroup diffusion [3]. In this description, the
electrons are divided into energy groups, with the lowest energy group de-
scribed as a thermal one, with a Maxwellian velocity distribution. Each
gTOUpis assigned a diffusion coefficient, and the groups are coupled to
one another by self-consistent electric fields imposed by charge neutral-
ity. Such a description is especially important when high energy tails are
produced by collective absorption processes.
In general, a second extension is also needed, since the diffusive calcu-
lation of the heat flow fails for'strong temperature gradients. For example,
when (Ve/Vei) .:G O.lLT (LT1 = alnBe/az), Eq. (12.10) gives Q .:G nBeve·
This result is clearly unphysical, since electrons cannot carry an energy
flux greater than their energy density times some typical velocity. In fact,
the Spitzer-Harm calculation is expected to fail [4,5] for even smaller heat
fluxes (i.e., for longer LT). We first note from Eqs. (12.8) and (12.9) that
electrons with velocities up to about 4-5 Ve contribute significantly to the
heat flow. Substituting Eq. (12.10) into Eq. (12.8) gives
i, -
-
fo Q
321fnBeve
(V)4
-
Ve
[(V)2-Ve 8 ]
Demanding that Ihl < fo for v ~ 4ve then requires that Q :S nQeve/5.
From a physical standpoint, the breakdo~ of Eq. (12.10) for strong
temperature gradients represents the transition to collisionless behavior in
which electrons simply free-stream rather than diffuse. Heuristic attempts
to match onto this collisionless regime have been made by simply limiting
the heat flow to a maximum flux. To illustrate, we again return to the
simple example in which electrons are characterized by a temperature Be.
Then Q = min[/"\: aBe/az, f nBeve], where f is the so-called flux limit.
Initially f was chosen to be about 0.6, since the maximum energy flux
(neglecting fields) carried by electrons with a Maxwellian distribution of
velocities is about 0.6 nBeve. However, clearly the flux limit is just a crude
but efficient attempt to describe the heat flow as the classical theory fails.
To properly describe the heat flux in this limit, both numerical calcu-
lations of the full kinetic equation (including electron-electron collisions)
12.3 Other lniluences on Electron Heat Transport 147
and Monte-Carlo calculations have been carried out [6-15]. In comparison
to the Spitzer-Harm treatment, more angular structure and/or modifica-
tions to the zero-order distribution functions have been included. Indeed,
the modification of fo(v) is obviously essential to include in many appli-
cations, since classical absorption can generate super-Gaussian velocity
distributions as discussed in Chapter 5. FUrthermore, the transport itself
can strongly modify the zero-order distribution function. These calcula-
tions have indicated that the heat flux tends to saturate at a value of
order 0.1 nBeve but also emphasize that a single flux limit is in general
too simple a parameterization. A significant challenge is to now incorpo-
rate these insights into improved transport models which are sufficiently
economical for routine use in design codes.
12.3 OTHER INFLUENCES ON ELECTRON
HEAT TRANSPORT .\
Thus far we have considered the electron heat transport as determined
by Coulomb collisions in a plasma with no magnetic fields. The heat
transport becomes an even more complex calculation when one includes
self-generated magnetic fields. There are many source terms [16-18] for
such fields; the best known is that which arises when the density and tem-
perature gradients are not parallel. This effect can be simply illustrated.
If we treat the electrons as a fluid and neglect their inertia, the force
equation gives the electric field necessary to preserve charge neutrality:
uxB
E= (12.12)
ne c
Here u is the plasma flow velocity, Pe the electron pressure, and B the
magnetic field. Substitution of E into Faraday's law then gives
~ aB = \l
C at
x (u xc B + \lnepe) . (12.13)
Equation (12.13) shows that a magnetic field is generated whenever the
condition \l x (\lpe In) j 0 holds i.e., when \In x \lBe j o.
To show that a sizable inhibition is possible via these fields, con-
sider here a crude order-of-magnitude estimate. Setting aB/at = 0 and
denoting all gradient lengths by some L, we find with Eq. (12.13) that
IBI ~ (c/-u)(Be/eL). The energy flux carried by electrons across the field
148 12 Electron Energy Transport
can be estimated as le ~ nBe D / L, where D is the diffusion coefficient.
Since the B fields can be quite inhomogeneous, we conservatively take
Bohm-like diffusion (D ~ WeeT~e' where Wee is the electron cyclotron fre-
quency and Tee the gyroradius). Combining these estimates then gives
le ~ nBeu, that is, the characteristic energy flow speed is the plasma flow
velocity which is typically of the order of the ion sound velocity.
For quantitative results, it is essential to address many important
questions - the size and extent of the self-generated fields, how electrons
diffuse across them and other mechanisms for their generation. The B
fields can also be generated by other mechanisms such as instabilities
produced by the anisotropic heated electron distribution or by velocity
anisotropies associated with the heat How. The resulting B fields can be
quite inhomogeneous, and so their influence [19J on the electrons is itself
a very rich topic.
Lastly the transport coefficient can also be modified by ion turbulence
in the plasma. The most commonly invoked mechanism for producing
this turbulence is the ion-acoustic drift instability driven by the heat How
[20-22]. The basic idea is very simple. The electron distribution function
carrying a heat flux Q is skewed, as shown by Eq. (12.8). In particular,
the low-energy electrons have a drift Vd ~ Q/nBe relative to the ions.
Physically this drift is produced by the self-consistent electric field nec-
essary to draw a return current to compensate for the How of the hotter
electrons which carry the heat How. When this drift exceeds the threshold
for the ion-acoustic drift instability, ion-acoustic waves are driven unsta-
ble. It has been hypothesized that the ion turbulence then so effectively
scatters the electrons that the heat flow is locked into a value near the
instability threshold. In other words, Qmax ~ nBevs, when ZBe » Bi. Here
Vs is the ion sound velocity and Bi is the ion temperature. There has been
considerable controversy over whether the ion-acoustic turbulence can be
this effective, particularly for electrons of very high energy. Computer
simulations [23J have suggested that the ion turbulence does not strongly
limit the heat flux.
12.4 Heat Transport in Laser-Irradiated Targets 149
12.4 HEAT TRANSPORT IN LASER-IRRADIATED
TARGETS
Theoretical description of large electron heat fluxes in laser-irradiated
targets is clearly a challenging problem. Let's conclude this chapter with
a very brief synopsis of the experimental feedback. In laser-irradiated
targets, the heat transport has been inferred from a variety of different
measurements, including x-ray images of the heated plasma, the ratio
of the energy in fast and slow ion expansion, the implosion and mass
ablation efficiencies, the density profile in the underdense plasma, and
the burn-through rate of thin films and layered targets.
As an example, let's consider experiments [24J in which Al disks coated
with a layer of CH were irradiated with a 100 ps pulse of 1.06tl laser
light with a peak intensity of 1OlSW / cm2. The thickness of the CH layer
was varied, and the x-ray emission at energies between 1 and 3 keV was
measured as a monitor of the energy transported to the Al substrate. AB
shown in Fig. 12.1, a rather thin layer of CH led to an abrupt decrease
in the x-ray emission, indicating poor electron transport. Calculations of
electron energy transport model this data by using a flux limit of f ~ 0.0l.
A similar inhibition of the heat transport (f ~ 0.01-{).04) has been
inferred from many other experiments [25-29] on disk targets. The trans-
port inhibition appears to decrease as the intensity and/or the wave-
length of the light is reduced. Furthermore, in at least some experiments
[30-36J in which spherical targets are rather uniformly irradiated, the
electron energy transport appears to saturate at a value close to that ex-
pected from numerical calculations of the Fokker-Planck equation (i.e.,
Q ~ 0.1 nBeve). This suggests that B field generation and/or lateral en-
ergy transport may be playing a significant role in the disk experiments.
Many of the recent transport experiments and calculations are reviewed
in Ref. 15.
Although much remains to be understood, there has clearly been sig-
nificant progress in both characterizing and understanding electron en-
ergy transport in laser irradiated targets. More experiments are needed
to clarify the dependence of the heat transport on such critical features
as target geometry, uniformity of irradiation, and laser light wavelength
and intensity, Improved. measurements of the underdense plasma condi-
tions are also needed. These conditions are an important indicator of the
transport and directly influence the coupling processes, which serve as
source terms driving the energy transport.
150 12 Electron Energy Transport 12.4 References 151

5. Shvarts, D., J. Delettrez, R. L. i'v1cCrory,and C. P. Verdon , Self-consistent re-

6 duction of the Spitzer-Harrn electron thermal heat flux in steep temperature
4
gradients in laser-produced plasmas, Phys. Rev. Led. 47, 247 (1981).
1 to 3 keV x-rays
I D. Crystal spectrograph
6. Bell, A. R., R. G. Evans, and D. J. Nicholas, Electron energy transport in
steep temperature gradients in laser-produced plasmas, Phys. Rev. Lett.
:2 o Pin diode 46, 243 (l981).
Q;
2
T 7. Mason, R. J., Apparent and real thermal inhibition in laser-produced plas-
El
S 10-3
c
0 8
t Q
mas, Phys. Rev. Lett. 47, 652 (1981);
8. Khan, S. A. and T. D. Rognlein, Thermal heat flux for arbitrary collisionality,
VI
Q)
Phys. Fluids 24, 1442 (1981).
6
:i 9. Matte, J. P. and J. Virmont , Electron heat transport down steep temperature
0
~ 4 gradients, Phys. Rev. Lett. 49, 1936 (1982).

t
'<r
0 D. 10. Matte, J. P., T. W. Johnston, J. Delettrez, and R. L. McCrory, Electron heat
C
2 transport with inverse brernsstrahlung and ion motion, Phys. Rev. Lett. 53,
VI
~::l 0 1461 (1984).
0
-:l 11. Luciani, J. F., P. Mora, and J. Virmont, Nonlocal heat transport due to steep
10-4
8 2 0
temperature gradients, Phys. Rev. Lett. 51, 1664 (1983).
12. Luciani, J. F., P. Mora, and R. Pellat, Quasistatic heat front and delocalized
6 D.
heat flux, Phys. Fluids 28, 835 (1985).
4 13. Albritton, J. R., Laser absorption and heat transport by non-Maxwell-
0 0.5 1.0
Boltzmann electron distributions, Phys. Rev. Lett. 50,2078 (1983).
CH thickness (urn)
14. Albritton, J. R., K A. Williams, 1. B. Bernstein, and K. P. Swartz, Nonlo-
Figure 12.1 The fraction of incident laser energy converted in Al line radi- cal electron heat transport by not quite Maxwell-Boltzmann distributions,
ation (.6.) and int~ 1-3Kev x-rays (0) as a function of the thickness of the CH Phys. Rev. Lett. 57, 1887 (1986).
overlayer. See Young et al., (1977). 15. For a review of calculations and experiments on transport, see J. Delettrez,
Thermal electron transport in direct-drive laser fusion, Can. J. Phys. 64,
932 (1986).
16. Haines, M. G., Magnetic-field generation in laser fusion and hot-electron
References transport, Can. J. Phys. 64, 912 (1986).
17. Stamper, J. A. et al., Spontaneous magnetic fields in laser-produced plasmas,
1. Spitzer, L. and R. Harm, Transport phenomena in a completely ionized gas, Phys. Rev. Lett. 26, 1012 (1971).
Phys. Rev. 89, 977 (1953). 18. Forslund, D. W. and J. U. Brackbill, Magnetic-field-induced surface transport
2. 1. P. Shkarovsky, T. W. Johnston, and M. P. Backynski, The Particle on laser-irradiated foils, Phys. Rev. Lett. 48, 1614 (1982).
Kinetics of Plasmas. Addison-Wesley, Reading, Mass., 1966. 19. Max, C. K, W. M. Manheimer, and J. J. Thomson, Enhanced transport
3. Zimrnerman, G. P. and W. L. Kruer, Numerical simulation of laser-irradiated across laser generated magnetic fields, Phys. Fluids 21, 128 (1978).
fusion, Comments Plasma Phys. Controlled Fusion 2, 51 (1975).
20. Forslund, D. W., Instabilities associated with heat conduction in the solar
4. Gray, D. R. and J. D. Kilkenny, The measurement of ion acoustic turbulence wind and their consequences, J. Geophys. Res. 75, 17 (1970).
and reduced thermal conductivity caused by a large temperature gradient in
21. Bickerton, R. J., Thermal conduction-amitations in laser fusion, Nucl. Fu-
a laser heated plasma, Plasma Physics 22, 81 (1980).
sion 13, 457 (1973).
152 12 Electron Energy Transport

CHAPTER
22. Manheimer, W. M., Energy flux limitation by ion ~coustic turbulence in laser
fusion schemes, Phys. Fluids 20, 265 (1977).
23. Lindman, E. L., Absorption and transport in laser plasmas, Journal
Physique 38, Colloque C6, 9 (1977).
de
Laser
24. Young,F. C. et al., Laser-produced-plasma energy transport through plastic
films, Appl. Phys. Lett. 30,45 (1977).
Plasma
25. Yaakobi, B. and T. C. Bristow, Measurement of reduced thermal conduction
in (layered) laser-target experiments, Phys. Rev. Lett. 14, 350 (1977).
Experime ts
26. Malone, R. C., R. L. McCrory, and R. L. Morse, Indications of strongly flux-
limited electron thermal conduction in laser-target experiments, Phys. Rev.
Lett. 34, 721 (1975).
27. Campbell, E. M., R. R. Johnson, F. J. Mayer, L. V. Powers, and D. C. Slater,
Fast-ion generation by ion-acoustic turbulence in spherical laser plasmas,
Phys. Rev. Lett. 39, 274 (1977).
28. Yaakobi, B. et al., Characteristics of target interaction with high power UV
laser radiation, Opt. Commun. 39,175 (1981).
29. Mead, W. C. et al., Characteristics of lateral and axial transport in laser ir-
radiations oflayered-disk targets at 1.06j..Lm and O.35j..Lm wavelengths, Phys.
Fl7~ids 27, 1301 (1984).
30. Goldsack, T. J. et al., Evidence for large heat fluxes from the mass ablation
rate of laser-irradiated spherical targets, Phys. Fluids 25, 1634 (1982). In previous chapters, we have discussed a variety of mechanisms for laser
plasma coupling, ranging from collisional absorption to excitation of many
31. Yaakobi, B. et al.; Thermal transport measurements in l.05j..Lm laser irradi-
different instabilities. Figure 13.1 illustrates the rich variety of coupling
ation of spherical targets, Phys. Fluids 27, 516 (1984).
processes as a function of the plasma density. Near the critical density
32. Tarvin, J. A., W. B. Fechner, J. T. Larsen, P. D. Rockett, and D. C. Slater,
(ncr), we have resonance absorption and instabilities leading to the exci-
Mass-ablation rates in a spherical laser-produced plasma, Phys. Rev. Lett.
tation of electron and ion waves. Near ncr/4, we have the 2wpe instability.
51, 1355 (1983).
The Raman instability operates for densities ;:; ncr/4. The Brillouin and
33. Fechner, W. B., C. L. Shephard, G. E. Busch, R. J. Schroeder, and J. A.
filament at ion instabilities take place throughout the under dense plasma,
Tarvin, Analysis of plasma density profiles and thermal transport in laser-
irradiated spherical targets, Phys. Fluids 27, 1552 (1984). as does inverse bremsstrahlung absorption. Throughout the underdense
plasma there can be self-generated magnetic fields or elevated levels of
34. Haver, A. et al., Measurement and analysis of near-classical thermal trans-
ion turbulence driven by a variety of processes associated with the plasma
port in one-micron laser-irradiated spherical plasmas, Phys. Rev. Lett. 53,
heating and expansion.
2563 (1984).
The mix of coupling processes depends on the intensity, wavelength,
35. Yaakobi, B. et al., Thermal transport measurements in six-beam, ultraviolet
and beam quality of the laser light and upon the gradient lengths, plasma
irradiation of spherical targets, J. Appl. Phys. 57,4354 (1985).
composition, and other plasma conditions. In turn, these depend on the
36. Jaanimagi, P. A., J. Delettrez, B. L. Henke, and M. C. Richardson, Tem-
mix of coupling processes. An understanding of this coupled nonlinear
poral dependence of the mass-ablation rate in UV-laser irradiated spherical
targets, Phys. Rev. A 34, 1322 (1986).

154 13 Laser Plasma Exper-iments
Self -generated magnetic
Brillouin back and side scatter
self-focusing and filamentation
/~----~ '------~
Parametric
L nCR
Resonant
n where 'min' denotes mirumum, T is expressed in nanoseconds, and R
Ion turbulence
~ distinction in mind, we will first examine some short-pulse-length exper-
Raman back and side scatter
x pulse-length ones.
Figure 13.1 A schematic of the density profile in the underdense plasma
illustrating the many processes which affect laser plasma coupling.
problem requires close collaboration between theory, computer simula-
tion and experiments. Although the understanding is far from complete,
calculations.have at least qualitatively predicted many important plasma
effects observed in experiments. These include steepening of the density
profile, hot electron generation, resonance absorption, stimulated Raman
and Brillouin scattering, and favorable wavelength scaling. In this final
chapter, let us briefly consider some of the experimental evidence for var-
ious laser plasma processes.
The characteristic size (L) of the underdense plasma is a very use-
ful parameter to consider when discussing experiments. If L is small
(L/ >"0 ;S 0 (10), where >"0 is the free space wavelength of the laser light),
then many of the coupling processes are either before threshold or weakly
occurring, and we are primarily concerned with how light is absorbed near
the critical density surface. On the other hand, if there is an extensive
region of underdense plasma [L/>..o ~ 0(100)], theory indicates that pro-
cesses such as Brillouin and Raman scattering, filamentation and inverse
bremsstrahlung can begin to play a significant role.
We can estimate the size of the underdense plasma in laser-irradiated
targets as the minimum of C*T /2 or R where c* is a typical plasma ex-
pansion velocity, T is the pulse length of the laser light, and R is the focal
spot radius. To give some feeling for the numbers, let us take an expansion
-,,*
-',
13.1 Density Profile Steepening 155
velocity of 3 X 107 cm/sec, which is approximately the ion sound speed in
fields a hydrogen plasma with an electron temperature o~r keV. Then
IDlll [ 1.5 X 102 T(ns), R(jJ.) ]
L
>"0(jJ.)
>"0
instabilities
absorption
and >"0 are expressed in micrometers. Hence experiments with 1.06jJ.m
light and pulse lengths of ;S 30ps have rather small underdense plas-
rnas whereas experiments with pulse lengths .<. ins and large focal spot
have large underdense plasmas. Note also the scaling as T / >"0. With this
iments (with small underdense plasmas) and then consider some longer
13.1 DENSITY PROFILE STEEPENING
As we discussed in Chapter 10, calculations show a pronounced steepening
of the density profile near the critical density. This steepening is impor-
tant because the scale length near the critical density affects the mix of
absorption processes and the heated electron temperatures. This profile
steepening has been confirmed by interferometric measurements [1-~] of
the density of a laser-heated plasma. In one experiment [1] a 41jJ.m diam-
eter glass micro balloon was irradiated with a 30ps, l.06jJ.m laser pulse at
an intensity of 3 x 1014W/cm2. An interferogram was taken with a 15ps,
O.26jJ.m probe beam and Abel-inverted to determine the axial electron
density profile plotted in Fig. 13.2. In both experiment and simulat~ons
the profile is steepened to an upper density nu that is roughly determined
by pressure balance:
where Vos is the oscillation velocity of an electron in the laser light field, Ve
is the electron thermal velocity and ncr is the critical density. The profile
is steepened down to a lower density that is determined by how the light
pressure and localized heating darns the plasma flow. This.lower density
typically appears to be somewhat less in experiments than ill the part icle
simulations, perhaps because of energy-transport inhibition.
156 13 Laser Plasma Experiments 13.2 Absorption of Intense, Short Pulse-Length Lig ht 157

2. 0 r-~-~-.--~------,

--
N
o
u
Cl>
1.0
o
,....

on

~ 0.3 Qc::: 1.6 urn />

c
o
•..u
•...
Cl>

w 0.10 6 12 35 o 35
Transverse radius, pm
.Axial distance from initial wall, urn
Figure 13.3 A transverse density profile measured by interferometry. See
Figure 13.2 A plasma density profile in a laser-irradiated target measured
Attwood et al., (1978).
by interferometry using a 0.26,lLm light pulse. See Attwood et al. (1978).

electron temperature) and as the scale length of the plasma, it is rela-

Calculations with a focused light beam [4) show an additional, re-
tively weak in these experiments. However, as discussed in Chapter 4,
lated effect: cratering of the critical density surface. Physically, the den-
computer simulations showed that there would still be a sizeable absorp-
sity surface is preferentially pushed in where the light intensity is greatest.
tion due principally to resonance absorption with some additional absorp-
This effect has also been observed in experiments. Figure 13.3 shows an
tion due to nonlinearly-generated ion fluctuations. These early simulation
Abel inverted transverse density contour measured in an experiment [1]
results assumed plane waves incident onto a plasma slab. In practice, a
in which a disk target was irradiated with 1.06j.Lmlight with an intensity
focussed light beam (say, with hot spots) both craters and ripples the
of c:=3 x 1014 W /cm2. The density cavity has a transverse scale length
critical density surface as discussed in the previous section. These sur-
approximately equal to that of the incident light beam. A smaller scale
face ripples average the absorption over angle as well as change part of
rippling of the critical density surface has also been inferred in experi-
the p-polarized light into s-polarized light and vice versa [8-10). A sim-
ments using higher intensity light. These ripples are probably due to hot
ple theory [8) was used to extend the ideal simulation results to crudely
spots in the incident light beam and/or to a critical surface instability
include this additional critical surface rippling. The result for the absorp-
found in computer simulations [5-7].
tion as a function of polarization and angle of incidence is shown by the
black line in Fig. 13.4.
13.2 ABSORPTION OF INTENSE, SHORT The absorption has been measured in detail in numerous experi-
PULSE-LENGTH LIGHT ments [11-17). In some of these experiments [11), plastic disks were irra-
diated with about 30ps pulses of 1.06j.Lmlight with an intensity of 1015 -
Many important features of the absorption measured in experiments with 1016W /cm2. The measured absorption, denoted by the circles in Fig. 13.4,
short pulse length, high intensity laser light can be understood in terms was both polarization-dependent and broad in angle. The absorption of
of resonance absorption in a steepened density profile. In such experi- p-polarized light peaked at approximately the predicted angle, and the
ments, the underdense plasma has both a high temperature and a small absorption of s-polarized light monotonically decreased with the angle of
. .
spatial extent. Because collisional a bsorption
. . as 8-e3/2 (8e IS
vanes . t he incidence. The magnitude of the absorption was also in reasonable agree-
 158 13 Laser Plasma Experiments 13.3 Heated Electron Temperatures 159

0.60 r-r-r-' , ,
0.50 L P potarrz a tion S polar rzation The generation of an electron distribution with at least two character-
I -0-< istic temperatures is supported by measurements of the-z-ray spectrum in
~
OAO many different experiments using short-pulse-length laser light. As an ex-
ample, Fig. 13.5 shows the x-ray spectrum observed in an experiment [20)
in which a plastic disk was irradiated with a 80ps pulse of L06~m light
at a peak intensity of z: 2 x 1015W j cm2. The low energy x-rays indicate ji
an electron temperature of ~ 700eV, and the high energy x-rays a tem- ;1:'
, 1

perature of ~ 8 keY. A third, higher temperature component has been ill

!I

1i'li;
o 10 20 30 40 50 60 o 10 20 30 40 50 60 observed in experiments with small underdense plasmas for sufficiently l
Angle of incidence, V, degrees intense irradiation [16,17J.

Figure 13.4 Laser light absorption as a function of angle of incidence, shown

here [or (a) p-polarization and (b) s-polarization. Circles denote the absorption
measured in a series of experiments in which plastic disks were irradiated with
30ps pulses of l.06~m light. See Manes et al., (1977).
ment. The principal discrepancy is an additional 10-15% absorption,
rather independent of angle of incidence and polarization. This addi-
tional absorption may be due to a number of effects such as inverse
bremsstrahlung, critical surface rippling, or self-generated magnetic fields.
At very high intensities (D.5 .:G 1017W_~2jcm2), the measured fractional
absorption increases significantly [16]. This effect has been attributed to
the development of a highly turbulent critical surface.
Figure 13.6 shows the heated electron temperature inferred from the
high energy x-rays in short pulse experiments over a wide intensity regime.
:,l
, ,
The various symbols with error bars represent data from a series of exper-
iments [21J in which disks or microballoons were irradiated with L06~m
light with pulse lengths in the range of about 50-20Ops. The open x's are
values of the resonantly-heated electron temperature calculated in a se-
ries of two-dimensional simulations [18) of resonance absorption. Both the
magnitude and intensity scaling of the heated electron temperatures are
in reasonable agreement, especially in view of the fact that the simulations
are quite ideal and do not include the space-and time-averaging inherent
in the experiments. Experiments using laser light with other wavelengths
10'6

13.3 HEATED ELECTRON TEMPERATURES

10'4
The heated velocity distribution is a very important feature of the ab-
sorption process. The light primarily heats electrons, since tlieir motion
in the oscillating fields is much larger than that of the massive ions. As >
Q) 10'2
""
;;
discussed in the Chapter 9, absorption via plasma waves in general leads Q)
- Rei cale
to high energy tails on the electron distribution function. Physically this is ""
-' 10
'0
because plasma waves tend to preferentially heat the faster (more nearly I Exp

resonant) electrons. The simulations of resonance absorption discussed in

loB
Chapter 10 predict that a roughly two-temperature distribution will re-
sult [18,19J. The lower temperature is that typical of electrons streaming 0.1 1.0 10.0 100.0
into the absorption region and is determined by how the heat transports hv, keV

to higher density. The hot temperature is that characteristic of the elec- Figure 13.5 The x-ray spectrum measured in an experiment in which a plastic
trons heated by resonance absorption, which in the simulations are found disk was irradiated by a 80ps pulse of L06~m light with an intensity of about
to have a quasi-Maxwellian velocity distribution. 2 X 1015 W/cm2. See Haas et al., (1977).
 13 Laser Plasma Experiments 13.4 Brillouin Scattering 161.
160

Experiments (23-37] with longer scale length plasmas and l.06/-LIl1

light show that significant Brillouin scattering is possible. In these exper-
o P2 O/W on cyclops !l Kinetic simulations
irncnts, sizeable underdense plasrnas (L/ >'0 > 30) were formed in various
o Parylene on janus with Geoid = 1 keV
ways: by using a prepulse, a long pulse length in the sense of T / >'0, or it
(:; Parylene on cyclops
100 preformed plasma. Let's consider a specific example in which a larger un-
••. 5i02 Microshell targets
>., derdense plasma was created by use of a prepulse. As shown in Fig. 13.7,
-'"
the addition of a prepnlse about 2115prior to a 75ps main pulse was found
to increase the backscatter of the main pulse which was normally incident
10 m=0.39 onto a CH slab [23]. As the prepulse energy was increased, the fraction of
the main pulse which was back-reflected increased from ::: 15% to ::: 40%
and the net absorption decreased from c- 50% to ::: 20%. For a fixed ratio
of prepulse energy to main pulse energy, the backscattered light increased
with the intensity of the main pulse and was rather insensitive to-the angle
at which the targets were tilted. In addition, it was shown that the light
Peak intensity. W/cm2
rays retraced their path. All these features are as expected if the light
Figure 13.6 The heated electron temperature as a function of incident laser reflection is due to the Brillouin instability in the underdense plasma.
light intensity. The symbols with error bars are values inferred from the high Experiments have also shown evidence for Brillouin sidescatter. This
energy x-rays in experiments in whichdisks or rnicroballoonswereirradiated with instability preferentially scatters light out of the plane of polarization. As
l.06/-Lmlight. The open x's are values calculated in a series of two-dimensional shown in Fig. 13.8, the sideward scattering out of the plane of polarization
simulations. See Manes et al., (1977). has been observed to greatly exceed that into the plane of polarization
in experiments with large focal spots [27]. Typically, the polarization-
dependent, sideward scattering observed to date has been .!S 30% of the
have given quite similiar results. One simply has to scale the intensity as
incident laser energy.
1..\6 (where .Ao is the free-space wavelength of the light), as theoretically
expected. ~ 50
c 5-10 X 1015W/cm2
~
0 40
<.l f/l.9 Normal inc.
13.4 BRlLLOUIN SCATTERING <1J
;;:: CH slab
;.
e
se
30
We have been discussing short-pulse-Iength experiments which are char- 1-= Main pulse

_:~edJ---
o
eo
.D
acterized by a small region of plasma with density less than ncr· These 20
:;: }- --------- 16 !: 2%
experiments have been typical of the exploding-pusher target experiments ::J
a. 10
carried out in the early days of the laser fusion program. Laser plasma c (a) -~ 2 nsec
coupling is more complex in long-pulse-Iength experiments with larger
regions of underdense plasma. Such experiments are more characteristic
- '"
:2 OU-~----~~------~~-----L~----~
o 10-1

of the ablative compressions needed to compress fuel to a high density to Prepulse energy/Incident energy

achieve high gain [22]. In the longer scale length plasmas, effects such as Figure 13.7 The fraction of the main pulse energy which was backscattered
collisional absorption, Brillouin and Raman scattering, and filamentation into an f /1.9 lens versus the fraction of the energy into a prepulse. See Ripin et
can all play a much more important role. Let's now discuss experiments al., (1977).
on some of these processes.
162 13 Laser Plasma Exp cr irncut s 13.5 Rama.n Scattering 163

Dia = 450 urn Dla = 15O~

geucrat ion has been observed. Let us brietiy mention several of these ex-
peruueuts. '.
In early experiments which showed a significant1evel of Raman scat-
10
tering [47], a large region of underdense plasma was formed by irradiating
a thin CH foil (7000A thick) with a 900ps, 3kJ pulse of 1.064j1m light.
5 Computer calculations predicted this foil to go underdense and expand
through a density of a few tenths of the critical density somewhat before
the peak of the laser pulse. When this happens, the laser light is propa-
0° 40° 80° 120° 160° gating through a relatively fiat region of underdense plasma. If we simply
Polar angle, e (0) use the radius of the focal spot as a measure of the distance over which
Figure 13.8 The angular distribution of the light scattered from a Au disk the plasma density is reasonably fiat, the characteristic plasma size is of
irradiated with l.06j1mlight focussedto an intensity of about 3.X 1014 W /cm2 order 200j1m, which is sufficient to produce sizeable Raman scatter. In-
The focal spot diameters are 150j1m and 450j1m. See Rosen et al., (1979). deed, about 10% of the light was observed to be Raman scattered. Figure
13.9 shows the measured angular distribution of the spectrally integrated
Raman scattered light. Most of the scattering was into the rear hemi-
Quite detailed studies of the nonlinear aspects of Brillouin scattering sphere, but even a small fraction of the incident light was scattered into
have been ca..r.riedout in microwave experiments with low density plasmas the forward hemisphere.
[38,39]. In some of these experiments, the amplitudes of both the driven
ion wave and its harmonic were measured via Thomson scattering of mil-
limeter waves. In addition, a heated ion tail was observed and its density
measured. These experiments emphasize that ion tail formation is a very
potent saturation mechanism. 200

The behavior of Brillouin scattering in laser experiments needs further Raman light energy
study. Identification of the scattering by its frequency spectrum is often angular distribution

uncertain due to Doppler shifts in the expanding plasma. Often the ob-
served scattered light with frequency near Wo does not appear to exhibit 100
any well-defined growth or saturation [32]. In experiments to date with
overdense targets and short wavelength light, the backscattering is quite
modest (.:S 10%).
0° 60° 90°

Angle
13.5 RAMAN SCATTERING
Figure 13.9 The angular distribution of the spectrally-integrated Raman
There are numerous experiments [40-53] on Raman scattering in laser-
scattered light measured in experiments in which thin foils were irradiated
irradiated plasmas. This process is relatively simple to identify since the
with l.06}-Lm light. The differentsymbols denote two separate experiments. See
scattered light is down-shifted by an electron plasma frequency. In exper-
Phi/lion et al., (1982).
iments with a large region of underdense plasma, the energy in Raman-
scattered light has been measured to be as large as 10-20% of the inci-
dent laser energy. III addition, the expected correlation with hot electron
164 13 Laser Plasma Experiments 13.5 Raman Scattering Hl5

Subsequent experiments [49] with thin foils irradiated with O.S3J.Land
O.26ft light have shown comparable levels offi.aman scattering. As ex-
pected on the basis of collisional damping, the scattering was found to
decrease dramatically for Au foils irradiated with O.26J.L light. In other
thin foil experiments using O.S3J.Llight, both the up- and down-shifted
components expected from Raman forward scattering were observed [SO].
In experiments [SI] at the University of Alberta, Raman backscatter
was observed when a rather uniform low density plasma in a solenoid was
irradiated with CO2 laser light. The plasma density was about 1/40 of
the critical density, the background electron temperature about 80 eV,
and the interaction length about 3=, as estimated from the depth of
focus of the laser light. As shown in Fig. 13.10, back reflection due to
the Raman instability was observed to onset at an intensity of ~ 4 x
1010 W/cm2, which was calculated to be the expected threshold intensity.
When the intensity was further increased, the reflectivity from this rather
low density plasma saturated at a value of about 0.7%.
I I
Raman reflected light has also been observed in experiments with
rather inhomogeneous plasma blowing off from an ~('!rdens(' t<l.rw't.. The
spectra measured in such experiments [4S,46] have indicated Raman ~(M-
tering from the region near ncr/4 as well as from the plasma at lower
densities down to about 0.05-0.1ncr' The level of the scattered light in
these experiments is often quite low: in the range of 10-6-10-4 of the
incident light. However, this level has been found to increase rapidly as
the focal spot size and pulse length is increased, leading to more gentle
gradients. For example, in experiments [47,S2] in which thick disk targets
were irradiated with 1.06J.L or 0.53J.L light, up to several percent of the
incident laser light was observed to be Raman scattered.
Finally a correlation of Raman scattering with hot electron genera-
tion has been observed in experiments [52] in which Au disks were irra-
diated with Ins pulses of 0.S3j.tm light. In these experiments, the laser
energy varied from 0.S-4.0kJ and the nominal intensity from about 1014_
2 x 1016W /cm2. The slope of the high energy x-rays indicates hot electrons
with a temperature of about 30keV. Figure 13.12 shows the fraction of
the laser energy deposited into hot electrons as inferred from the level
of the hard x-rays versus the measured fraction of the laser energy in

Z'
.~
~
o
10-2 r-

1J r- H
-

>Q)
.x
1014

1013
Shot 93100603

• Data
-Spectrum
~Q)
:;
Cl:
=.
Q)

1012

1 Q)
<J
c
Q)
:J
u:::
1011

I I
10-3
0 2 3
1010
0 200
Intensity (1011 W/cm2)
X-ray energy (keV)
Figure 13.10 The back reflection due to the Raman instability measured in
Figure 13.11 Hard x-ray spectrum from Au disk irradiated with a 3.6kJ, 1ns
experiments in which a low density plasma in a solenoid was irradiated with
pulse of O.53J.Lm light focussed onto 740J.Lm spot. See Drake et al., (19R4).
1O.6J.Lm light. See Offenberger et al., (1982).

..
loO 13 Laser Plasrna Exp er irne nt s
gc 10-3
000
u o
'"
Qi o
o eo
I o
o The 3wo/'2 emission is frequently diagnosed in laser plasma exper-
Raman scattered light fraction
Figure 13.12 The fraction of the laser energy absorbed into hot electrons
versusthe fraction in Raman-scattered light in Au disks irradiated by Ins pulses
ofO.53).1mlight. See Drake et al., (1984).
Raman-scattered light. Note the impressive correlation. The solid line
represents the expected correlation using the Manley-Rowe relations with
the measured mean value of the frequency of the scattered light. Because
of the error bars, it is quite possible that other processes such as the 2wpe
instability are also contributing to hot electron generation.
Although many trends in the observations agree with expectations,
there are also challenging puzzles [54]. There is usually a gap in the fre-
quency spectrum, showing that Raman scattering is much weaker than
expected for a narrow range of densities near ncr/4. In addition, a low
level of Raman backscattering is .often observed below the nominal inten-
sity threshold. Both these puzzles may indicate that Raman scattering
is being seeded by an enhanced level of plasma waves excited by other
processes. For example Simon and Short [55,56] postulate that bursts of
hot electrons due to the 2wpe instability preferentially excite the plasma
waves in the lower density region. Below the Raman instability threshold,
we would then have ordinary Thornson scattering from enhanced fluctu-
ations. Above threshold, the instability grows from the enhanced levels.
13.6 Other Plasma Processes 167
13.6 OTHER PLASMA PROCESSES
..
There is also experimental evidence for many other plasma processes in
laser-produced plasmas. Electron plasma waves due to the 2wpe instability
have been directly observed by Thoruson scattering in experiments with
1O.6~mllight. The growth rate. the local density profile steepening, and
the generation of ion waves in the nonlinear state have been measured
ill some detail [57-59]. A useful signature of this instability is emission
near 3wo/2, which arises from the coupling of the incident and reflected
light wave with a plasma wave near ncr/4. Unfortunately the level of the
instability is difficult to estimate from this signal, since the emission only
indirectly indicates the level of part of the spectrum of driven plasma
waves.
iments [60-62]. For example, in some experiments at the University of
Rochester [62], CH spheres were irradiated with a 60G-700ps pulse of
O.35).1mlight. The 3wo/2 emission was observed to onset at an intensity
of about '2 x 1014W /cm2, the estimated threshold intensity of the 2wpe
instability. The level of the emission increased with the intensity of irra-
diation but then saturated at rather a low level when the intensity was
about 3 x 1014 W/cm2. Hard x-rays indicating suprathermal electrons
with a temperature of about 35keV were observed to be correlated with
the 3wo/2 emission. The inferred fraction of the laser energy in these
suprathermal electrons saturated at a low value of about 10-4 of the in-
cident energy in these experiments with L/ AO ;S 150.
Filamentation of laser light is perhaps the least characterized of the
plasma processes we have discussed. Much of the evidence is rather in-
direct: inferrences from structure [63-65] in x-ray pictures of the heated
plasma or in images of the back-reflected light. Filamentation has also
been inferred from the angular distribution [45]of the half-harmonic light
or from frequency shifts [66] in the reflected light. Filamentary structures
have been directly observed by using optical shadowgraphy [67]' by irnag-
ing the second harmonic emission [68], and by Thomson scattering from
electron plasma waves generated in the walls of the filament [69].
Parametric instabilities near the critical density have been inferred
from frequency shifts in the second harmonic emission [70-72] as well
as from Thornson scattering measurements [73] of ion acoustic waves in-
duced by a 10.6).1mlaser. The unstable waves and the plasma heating
have been measured in some detail in microwave experiments with low
density plasmas [74-77]. Excitation of waves near the critical density has
168 13 Laser Plasma Experiments 13.7 Wavelength Scaling of Laser Plasma Coupling 169

also been studied extensively in ionospheric heating experiments [78,79].

In addition, self-generated magnetic fields with values up to 106 Gauss
have been measured in laser experiments using Faraday rotation of a
probe beam [80,81]. Since we've already considered a number of different
.
~
c:
100

.2 80
plasma processes, we will now proceed to the important topic of wave- a.
length scaling. (;
Cl) 60
.0
III

E 40
S!'
13.7 WAVELENGTH SCALING OF LASER ~
C1J
PLASMA COUPLING Cl) 20
j
As both calculations and experiments have amply demonstrated, laser 0
plasma coupling can be influenced by a rich variety of collective plasma 1012
effects. Many of these collective processes either decrease the absorption Laser Irradiance (Watts/cm 2 )
or give absorption into a tail of very energetic electrons. The advantages
of enhancing collisional absorption and reducing collective effects have Figure 13.13 The absorption versus intensity measured using laser light with
placed a premium on the use of short wavelength laser light [82,83]. wavelengths ranging from l.05I-Lm to O.26pm. See Ripin and Kruer, (1986).
When the laser wavelength (AO) is decreased, the light wave penetrates
to higher density plasma since the critical density increases as A02. For
a given absorbed intensity, the heated plasma is both denser and lower
in temperature and hence much more collisional. In addition to being
reduced by this greater collisionality, the collective processes are more
weakly driven by short wavelength light. For a given intensity, the oscil- 1014,--.---,--,---,--.---,--,---,
lation velocity of an electron in the light wave is proportional to the laser 1013
• 1.064 urn
wavelength. >(1) 1012 •. 0.355 urn
Many experiments [84-88] with 0.53jlm, 0.35jlm and 0.26jlm light ~
have demonstrated that important features of the coupling improve as :>
(1)
1011
the wavelength decreases. Figure 13.13 shows a compendium [89] of the ~ 1010
absorption as a function of intensity measured in a variety of experiments (1)
U 109
using laser light with wavelengths ranging from 1.05jlm to 0.26jlm. In C
(1)

these experiments on CH targets, the pulse lengths varied from lOOps to :::l 10B I
u:::
Ins, but the focal spot size was typica:lly rather sma:ll. Note the dramatic 107
increase in absorption as the wavelength decreases, as expected since in- 106~~--~--~--~--~--L-~--~
verse bremsstrahlung depends strongly on wavelength. A very strong de- o 10 20 30 40 50 60 70 80
crease in hot electron generation has also been shown in such experiments.
Energy (KeV)
Figure 13.14 shows x-ray spectra measured in experiments [90]in which a
600ps pulse of light was focused to an intensity of about 3 x 1014 W /cm2 Figure 13.14 The x-ray spectra from an Al disk irradiated with v-- 600ps
onto an Al disk. The level of the high energy x-rays decreased by sev- pulses of 1.064I-Lmand 0.355pm light. The absorbed intensity was about 1.5 X
eral orders of magnitude as the wavelength of the light was changed from 1014 W/cm2 See Campbell, (1984).
170 13 Laser Plasma Exp er ime nts
1.U6J.llll to O.355J.llU, illustrating a significant reduction in collective plasma
interactions.
, ot surprisingly, there has been a strong trend to the use of short-
wavelength lasers. Most large laser fusion facilities now operate or plan
to operate at wavelengths ~ O.5J.l. These wavelength scaling experiments
are being extended to include long-scale-length plasmas, which are more
characteristic of reactor targets. Such experiments will be able to quantify
the regimes of intensity and wavelength for optimum coupling.
Finally, increased attention is being focussed on techniques to smooth
the intensity profile of a laser beam [91-94]. Smoother beams will allow
more uniform illumination of fusion targets and clarify the role of beam
structure in experiments. Induced spatial incoherence (ISI) is one tech-
nique for beam smoothing [91]. In the usual embodiment of ISI, a pair of
reflecting, echelon-like mirrors is used to divide a broad bandwidth laser
beam into many independent bearnlets. The echelons introduce time de-
lays between the beamlets which are longer than the laser coherence time.
The beamlets are overlapped onto a target, producing a very smooth in-
tensity profile when averaged over time scales long compared to the coher-
ence time. Experiments [94] with induced spatial incoherence are showing
that the use of a smoother beam further improves the coupling of short
wavelength laser light with targets.
In summary, laser plasma coupling is a rich and challenging area of
applied physics. Many different processes can compete to determine the
coupling, ranging from collisional absorption to a variety of plasma insta-
bilities. As lasers have become more energetic and targets larger, more of
these plasma processes have been shown to indeed play a role in experi-
ments. Much has been learned through the fruitful and close interaction
between theory and experiment, and many challenging problems remain
to be understood. There are many important theoretical issues, includ-
ing the multidimensional and nonlinear behavior of the instabilities, the
competition of the various coupling processes, and the heat transport in
strongly driven plasmas. There are likewise many important experiments
yet to be done to better diagnose the plasma and irradiation conditions
and to extend the data base to the larger plasmas which are more char-
acteristic of reactor targets. Laser plasma coupling continues to be an
exciting and important area of research for laser fusion applications.
13.7 References 171
References
1. Attwood, D. T., D. W. Sweeney, J. M. Auerbach, anjp. H. Y. Lee, Interfero-
metric confirmation of radiation-pressure effects in laser-plasma interactions,
Phys. Rev. Lett. 40, 184 (1978).
2. Fedosejevs, R., 1. V. Tomov, N. H. Buruett , G. D. Enright, and M. C.
Richardson, Sell-steepening of the density profile of a C02-laser-produced
plasma, Phys. Rev. Lett. 39, 932 (1977).
3. Fechner, W. B., C. L. Shepard, Gar. E. Busch, R. J. Schroeder, and J. A.
Tarvin, Analysis of plasma density profiles and thermal transport in laser-
irradiated spherical targets, Phys. Fluids 21, 1552 (1984).
4. Randall, C. J. and J. DeGroot, Effect of crater formation on the absorption
of focused laser light, Phys. Rev. Lett. 42, 179 (1979).
5. Estabrook, K. G., Critical surface bubbles and corrugations and their impli-
cation to laser fusion, Phys. Fluids 19, 1733 (1976).
6. Nishimura, H. et al., Resonance absorption and surface instability at a critical
density surface of a plasma irradiated by a C02 laser, Plasma Phys. 21,
69 (1980).
7. Bezzerides, B. et al., Recent developments in understanding the physics of
laser-produced plasmas; in Proceedings of the Sixth International Con-
ference on Plasma Physics and Controlled Fusion Research, Vol. 1, p.
123. International Atomic Energy Agency, Vienna, 1977.
8. Thomson, J. J., W. L. Kruer, A. B. Langdon, C. E. Max, and W. C. Mead,
Theoretical interpretation of angle- and polarization-dependent laser light
absorption measurements, Phys. Fluids 21, 707 (1978).
9. Cairns, R. A., Resonant absorption at a rippled critical surface, Plasma
Phys. 20, 991 (1978).
10. David, F. and R. Pellat, Resonant absorption by a magnetic plasma at a
rippled critical surface, Phys. Fluids 23, 1682 (1980).
11. Manes, K. R., V. C. Rupert, J. M. Auerbach, P. Lee, and J. E. Swain,
Polarization and angular dependence of 1.06J.lm light absorption by planar
piasmas, Phys. Rev. Lett. 39, 281 (1977).
12. Godwin, R. P., P. Sachsenmaier, and R. Sigel, Angle-dependent reflectance
of laser-produced plasmas, Phys. Rev. Lett. 39, 1198 (1977).
13. Pearlman, J. S. and M. K. Matzen, Angular dependence of polarization-
related laser-plasma absorption processes, Phys. Rev. Lett.39, 140 (1977).
14. Balmer, J. E. and T. P. Donaldson, Resonance absorption of 1.06J.lm laser
radiation in laser-generated plasma, Phys. Rev. Lett. 39, 1084 (1977).
 172 13 Laser Plasma Experiments 13.7 References
173

15· V'U
I eneuve, D . M ., G .. D Enright , and M. C. Richardson, Features ofRlateralA 31. Massey, R., K. Berggren and Z. Pietryzk, Observation of st imulat eri Brillouin
energy transport in C02-laser-irradiated microdisk targets, Phys. ev. backscattering from an underdense plasma. PkYs·. Re». Lctt. 36. %3
(1976).
27, 2656 (1983). . .
16. Bach, D. R. et al., Intensity-dependent absorption in 10.61-Lmlaser illumi- :12. Coldrnan, L. M., W. Seka, K. Tanaka, R. Short. and 1\. Simon, The use of
nated spheres, Phys. Rev. Lett. 50, 2082 (1983). . laser harmonic spectroscopy as a target diagnostic, Can. 1. Ptujs. 64.969
( 1986).
17. Burnett, N. H. and G. D. Enright, Hot electron gener~tion and transport ID

high-intensity laser interaction, Can. J. Phys. 64, 9_0 (1986). 3.3. Gorbunov, L. M., Vu. S. Kasyanov, V. V. Korobkin, A. N. Polyanich-v. and
18. Estabrook, K. G. and W. L. Kruer, Properties of resonantly heated electron A. P. Shevelko, Spectral and temporal measurements of radiation of Ught
distributions, Phys. Rev. Lett. 40, 42 (1978). backscattered by a laser plasma, Sou. Phys. JETP 27,226 (197 )

19. Forslund, D. W., J. M. Kindel, and K. Lee, Theory of ~ot electron spectra 34. Maaswinkel, A., K.Eidmann, and R. Sigel, Comparative reflectance measure.
at high laser intensity, Phys. Rev. Lett. 39, 284 (197f). ments of laser-produced plasmas at 1.06 and 0.53f.Lm, Phys. Rev. Lett. 42.
1625 (1979).
20. Haas, R. et al., Irradiation of parylene disks with a l.06f.Lm laser, Phys.
Fluids 20, 322 (1977).
35. Nakai, S. et al., Nonlinear interaction processes between a CO2 laser and a
plasma, Phys. Rev. A17, 1133 (1978).
21. Manes, K. R, H. G. Ahlstrom, R H. Haas, and J. F. Holzrichter, Light-
plasma interaction studies with high-power glass lasers, J. Opt. Soc. Am. 36. Yamanaka, C., T. Yamanaka, T. Sasaki, J. Mizui, and H. B. Kang, Brillouin
backscattering and parametric double resonance in laser-produced plasmas,
67,717 (1977). .
Phys. Rev. Lett. 32, 1038 (1974).
22. Nuckolls, J., L. Wood, A. Thiessen, and G. Zimmerman, L~r compression of
matter to super high densities: thermonuclear (CTR) applications, Nature 37. Basov, N. G. et al., Generation of high power light pulses of wavelengths
1.06 and 0.53f.Lm and their application in plasma heating, Sou. 1. Quant.
239, 139 (1972).
Electron. 2, 439 (1973).
pm, B .. H et al ., Enhanced backscatter
23 · Rini with a structured laser pulse,
38. Mase, A. et al., Stimulated-Brillouin-scattering studies in low density plas-
Phys. Rev. Lett. 39, 611 (1977). .
mas using microwave sources; in Proceedings of the Eighth International
- 24. Phillion, D. W., W. L. Kruer, and V. C. Rupert, Brillouin scatter ill laser-
Conference on Plasma Physics and Controlled Nuclear Fusion Re.
produced plasmas, Phys. Rev. Lett. 39, 1529 (1977).
search, Vol. Il, p.745. International Atomic Energy Agency, Vienna, 1981.
25 N A L Pitt D. Salzmann and A. A. Offenberger, Saturation of stimulated
39. Pawley, C. J., N. C. Luhmann Jr., and W. L. Kruer, Microwave simulation
· g, .,."
Brillouin backscatter in C02-laser-plasma interaction,
Ph
Vs.
R eu. Le.tt 42 ,
of laser plasma intractions; in Proceedings of the Eleventh International
307 (1979). Conference on Plasma Physics and Controlled Nuclear Fusion Re.
26 M ayer, F . J ., G ..E Busch , C. M. Kinzer, and K. G. Estabrook, L Measurements
44 1498 search. International Atomic Energy Agency, Vienna, 1987.
· of short-pulse backscatter from a gas target, Phys. Rev. ett. ,
40. Bobin, J. L., M. Decroisette, B. Meyer, and Y. Vitel, Harmonic generation
(1980). .. . . and parametric excitation of waves in a laser-created plasma, Phys. Rev.
27 · R osen, M .. D et al ., The interaction of 1.06f.Lm laser radiation WIth high Z Lett. 30, 594 (1973).
disk targets, Phys. Fluids 22, 2020 (1979). .
41. Watt, R. G., R. D. Brooks, and Z. A. Pietryzk, Observation of stimulated
28 T aka K et al Brillouin scattering, two-plasmon decay, and self-focusmg Raman backscatter from a preformed, underdense plasma, Phys. Rev. Lett.
· i: und~rd~nse ui~raviolet laser-produced plasmas, Phys. Fluids 28, 2910 41, 170 (1978).
(1985). . ., . 42. Elazar, J., W. Toner, and E. R. Wooding, Backscattered radiation at wo/2
29. Turner, R E. and L. M. Goldman, Evidence for multiple Brillouin modes III from neodymium laser plasma interactions, Plasma Phys. 23,813 (1981).
laser-plasma backscatter experiments, Phys. Flmds 24, 184 (1981): .
43. Joshi, C., T. Tajima, J. M. Dawson, H. A. Baldis, and N. A. Ebrahim,
30. H er b st, M . J ., C .., E Clayton and F. F. Chen,
) Saturation of Brillouin Forward Raman instability and electron acceleration, Phys. Rev. Lett.47,
backscatter, Phys. Rev. Lett. 43, 1591 (1979 . 128.5 (1981).
 13.7 References 175
13 Laser Plasma Experiments

59. Meyer, J and H. Houtrnan, Experimental investigation of the two-plasmon

.14. Figueroa, H., C. Joshi, H. Azechi, N. A. Ebrahirn, and K. G. Estabrook,
decay instability in a C02-laser-produced plasma, Pkys. Fluids 28, 1549
Stimulated Raman scattering, two-plasmon-decay and hot electron genera-
(1985). .••.
tion from underdeuse plasnias at .35J-Lm, Phys. Fluids 27, 1887 (1984).
60. Turner, R. E., D. W. Phillion, B. F. Lasinski, and E. M. Campbell, Hall and
.15. Tanaka, K. et al., Stimulated Raman scattering from UV-Iaser-produced
three-halves harmonic measurements from laser-produced plasmas, Phys.
plasmas, Phys. Rev. Lett.48, 1179 (1982).
Flu'ids 27, 511 (1984) .
.16. Seka, W. et al., Collective stimulated Raman scattering instability in UV
61. Carter, P. D., S. M. 1. Sim, H. C. Barr , and R. G. Evans, Time-resolved
laser plasmas, Phys. Fluids 27, 2181 (1984).
measurements of the three-halves harmonic spectrum from laser-produced
47. Philliou, D. W., D. L. Banner, E. M. Camp bell, R. E. Turner, and K. G.
plasmas, Phys. Rev. Lett. 44, 1407 (1980).
Estabrook, Stimulated Raman scattering in large plasmas, Phys. Fluids
62. Keck, R. L., R. L. McCrory, W. Seka, and J. M. Soures, Absorption physics
25, 1434 (1982).
at 351nm in spherical geometry, Phys. Rev. Lett. 54, 1656 (1985).
48. Phillion , D. W., E. M. Campbell, K. G. Estabrook, G. E. Phillips, and F.
63. Haas, R. A., M. J. Boyle, K. Manes, and J. E. Swain, Evidence of localized
Ze, HIgh energy electron production by the Raman and 2wpe instabilities in
heating-in CO2-laser-produced plasmas, J. Appl. Phys. 47, 1318 (1976).
a 1.064j1m laser-produced underdense plasma, Phys. Rev. Lett. 49 1405
(1982). ' 64. Max, C. E. et al., Scaling of laser-plasma interactions with laser wavelength
and plasma size; in Laser Interaction and Related Plasma Phenomena,
49. Turner, R. E. et aI., Evidence-for collisional damping in high-energy Raman
Vol. 6, p.507 (H. Hora and G. Miley, eds.). Plenum, New York, 1984.
scattering experiments at .26j1m, Phys. Rev. Lett. 54, 189 (1985).
65. Ng, A., D. Saltzmann, and A. A. Offenberger, Filamentation of C02-laser
50. Turner, R. E. et al., Observation of forward Raman scattering in laser-
radiation in an underdense hydrogen plasma, Phys. Rev. Lett. 43, 1502
produced plasmas, Phys, Rev. Lett. 57, 1725 (1986).
(1979).
51. Offenberger, A., R. Fedosejevs, W. Tighe, and W. Rozmus, Stimulated Ra-
66. Tanaka, K., L. M. Goldman, W. Seka, R. W. Short, and E. A. Williarns,
man backscatter from a magnetically confined plasma column, Phys. Rev.
Spectroscopic study of scattered light at around the fundamental wavelength
Lett.49, 371 (1982).
in UV laser-produced plasmas, Phys. Fluids 27, 2960 (1984).
52. Drake, R. F:et al., Efficient Raman sidescatter and hot-electron production
67. Willi, 0., P. T. Rumsby, and Z. Lin, Filamentation instability in laser-
m laser-plasma interaction experiments, Phys. Rev. Lett. 53, 1739 (1984).
produced plasmas; in Laser Interaction and Related Plasma Phenom-
53. Shephard, C. L. et al., Raman scattering in experiments with planar Au ena, Vol. 6, p.633 (H. Hora and G. Miley, eds.). Plenum, New York, 1984.
targets irradiated with 0.53j1m laser light, Phys. Fluids 29, 583 (1986).
68. Herbst, M. J., J. A. Stamper, R. R. Whitlock, R. H. Lehmberg, and B. H.
54. Kruer, W. L., Laser-driven instabilities in long scalength plasmas; in Laser Ripin, Evidence from x-ray, ~wo, and 2wo emission for laser filamentation
Plasma Interactions 3, (M. B. Hooper, ed.) p.79-103. SUSSP Publications in a plasma,' Phys. Rev. Lett. 46, 328 (1981).
Edinburgh, 1986. '
69. Baldis, H. A. and P. B. Corkum, Self-focusing of 1O.6j1m radiation in an
~S'· unon, A . and R. 1. Short, New model of Raman spectra in
. laser-produced
underdense plasma, Phys. Rev. Lett. 45, 1260 (1980).
plasmas, Phys. Rev. Lett. 53, 1912 (1984).
70. Basov, N. G. et al., Compression of laser irradiated hollow microspheres:
56. ~imon, A., W. Seka, L. M. Goldman, and R. W. Short, Raman scattering in in Plasma Physics: Nonlinear Theory and Experiments, pA7 (H. Wil-
inhomogeneous laser-produced plasmas, Phys. Fluids 29, 1704 (1986).
helmsson, ed.). Plenum, New York, 1977.
57'· Baldis, H. A., J. C. Simpson,
.' and P. B. Corkum, Two-plasmon decay and 71. Yamanaka, C., T. Yamanaka, T. Sasaki, J. Mizui, and H. B. Kang, Brillouin
profile modification produced by 1O.6j1m radiation at quarter-critical density, backscattering and parametric double resonance in a laser-produced plasma,
Phys. Rev. Lett. 41, 1719 (1978). Phys. Rev. Lett. 32, 1038 (1974).
~Bald'
· IS,..HA .
and C. J. Walsh, Growth and saturation of the two-plasmon 72. Tanaka, K. et al., Evidence of parametric instabilities in second harmonic
decay instability, Phys. Fluids 26, 1364 (1983). >. spectra from 1054nm laser-produced plasmas, Phys. Fluids 27, 2187
,
'! (1984).
176 13 Laser Plasma Experiments 13.7 References 177

73. Walsh, C. J., H. A. Baldis, and R. G. Evans, Plasma fluctuations at critical 89. Ripin, B. H. and W. L. Kruer, Inertial confinement fusion systems; in Plasma
density in a C02 laser plasma interaction, Phys. Fluids 25, 2326 (1982).
Press, Washington, 1986.
,
and Fluids .- Physics Through the 1990'.5, p.221-236. National Academy
74. Chu, T. K. and H. W. Hendel, Measurements of enhanced absorption of
electromagnetic waves and effective collision frequency due to parametric 90. CampbeU, E. :V!., Dependence of laser-plasma interaction physics on laser
decay instability, Phys. Rev. Lett. 29, 634 (1972). wavelength and plasma scalelength; in Radiation in Plasmns, Vol. I, p.579
75. Dreicer, H., R. F. Ellis, and J. C. Ingraharn, Hot electron production and (B. McNamara, ed.). World Scientific, Singapore, 1984.
anomalous microwave absorption near the plasma frequency, Phys. Rev. 91. Lehmberg, R. H. and S. P. Obenschain, Use of induced spatial incoherence
Lett. 31, 426 (1973). for uniform illumination of laser fusion targets, Opt. Comm. 46, 27 (1983).
76. Porkolab, M., V. Arunasalam, N. C. Luhmann Jr., and J. Schmitt, Parametric 92. Kato, Y. et al., Random phasing of high-power lasers for uniform target
instabilities and plasma heating in an inhomogeneous plasma, Nucl. Fusion acceleration and plasma instability suppression. Phsjs. Rev. Lett. 53, 10.'i7
16,269 (1976). (1984).
77. Mizuno, K., J. S. DeGroot, and K. G. Estabrook, Electron heating caused 93. Ximing Deng, Xiangchun Liang, Zezun Chen, Wenyan Yu, and Renyong
by parametrically driven turbulence near the critical density, Phys. Fluids Ma, Uniform illumination of large focal targets using a lens array, Chin. 1.
29, 568 (1986). Lasers 12, 257 (1985).
78. Fejer, J. A., Ionospheric modification and parametric instabilities, Rev. Geo- 94. Obenschain, S. P. et al., Laser-target interaction with induced spatial inco-
phys. Space Phys. 17, 135 (1979). herence, Phys. Rev. Lett. 56, 2807 (1986).
79. Wong, A. Y., J. Santoru, C. Darrow, L. Wang and J. G. Roederer, Radio
Science 18, 815 (1983).
80. Stamper, J. A. and B. H. Ripin, Faraday-rotation measurements ofmegagauss
magnetic fields in laser-produced plasmas, Phys. Rev. Lett. 34, 138 (1975).
81. Raven, A. et al., Dependence of spontaneous magnetic fields in laser produced
<plasmas on target size and structure, Appl. Phys. Lett. 35, 526 (1979).
82. Max, C. E. and K. G. Estabrook, Wavelength scaling in laser fusion from a
plasma physics viewpoint, Comments on Plasma Physics and Controlled
Fusion 5, 239 (1980).
83. Kruer, W. L., Laser plasma coupling in reactor-size targets, Comments on
Plasma Physics and Controlled Fusion 6, 167 (1981).
84. Garban-Labaune, C. et al., Effect of laser wavelength and pulse duration
on laser-light absorption and back reflection, Phys. Rev. Lett. 48, 1018
(1982).
85. Seka, W. et al., Measurements and interpretation of the absorption of .35t-tm
laser radiation on planar targets, Opt. Comm. 40,437 (1982).
86. Slater, D. C. et al., Absorption and hot-electron production for 1.05 and
0.53t-tm light on spherical targets, Phys. Rev. Lett. 46, 1199 (1981).
87. Nishimura, H. et ol., Experimental study of wavelength dependences of laser-
plasma-coupling, transport, and ablation processes, Phys. Rev. A 23, 2011
(1981).
88. Mead, W. C. et al., Laser irradiation of disk targets at 0.53t-tm wavelength,
Phys. Fluids 26, 2316 (1983).

ablative steepening, 125
ablative compression, 160
absorption processes, 3!t-43, 45-55, 58,
59, 74, 81, 121-125, 132-136, 153,
156-158, 165-170
adiabatic equation of state, 8-10
Airy function, 33, 39, 40, 49, 119
aliasing, 21
bandwidth, 70, 71, 170
Bohm-Gross frequency, 77
bounce frequency, 100, 101
Brillouin instability, 73, 87-93,
127-132, 135, 137, 160-162
Brillouin scattering,
see Brillouin instability
Index
charge density fluctuations, 11-14, 39,
57-59, 95
collective behavior, 2, 5, 17, 18, 20
collision frequency, 2-4, 46-48, 52-55,
106-108,145
collisional absorption, 45-55, 94, 125,
153-158, 168
collisional behavior, 2-4, 45-55,
96, 144-147, 168
collisional damping, 45-48, 67-68, 80,
83, 92, 96, 109, 164
collisionless plasma, 2, 4-5, 18, 146
computer simulations, 17-25, 104-108,
121-125, 130-138, 148
continuity equation, 6, 10, 46
critical density, 29
critical surface rippling, 157, 158

lRO
Debye length, 2, 3, 14, 15
Debye sphere, 3
density profile modification, 115-125.
];16,155-157
density scale length. 115-117, 120,
1.54-156
dielectric function, 28-35, 38, 39,
47, 49
electromagnetic waves, 27-35, 37-43,
46-52, 74-76, 88, 93
electron heating, 45-55, 95-108,
121-125, 132-137, 158, 159, 165-168
electron plasma frequency, 4, 12 '
electron plasma waves, 11, 12, 39, 40,
57-83,95-110, 122-125, 132-136, 167
electron transport, 134, 143-150
electron-electron collisions, 53, 144-146
electron-ion collisions, 2-4, 45-55, 96,
143-145
electrostatic waves, 11-14, 95
energy transport, 143-150
equation of state, 8-10
filamentation, 87, 93, 94, 132, 135, 137,
153, 154, 167
finite-size simulation particles, 18
fluid equations, 6-10
flux limit, 146, 147, 149
Fokker-Planck, 52-55, 144-145, 149
force equation, 7, 10, 46
forward scattering, 79, 164
freely-expanding plasma, 116, 117
frequency matching, 62, 73, 81, 87
Gaussian statistics, 70
grid in particle codes, 19-21
group velocity, 31, 70
Index
harmonic generation of ion waves, 112,
162
heat flux, 143-150
hot electron generation, 74, 101, 106,
107, 124, 125, 133-136,158-160,
165-168
incoherent pump, 70, 71, 170
induced spatial incoherence, 170
inhibited transport, 149
instabilities, 61-83, 87-94, 101,
104-108, 123, 124, 127-138, 148,
153, 156, 160-168
inverse bremsstrahlung, 45-55, 125,
153, 154, 168
ion acoustic decay instability, 62-7l,
95, 104-108, 167
ion acoustic drift instability, 148
ion acoustic frequency, 14
ion acoustic velocity, 14, 89, 116
ion acoustic wave, 11-14,57-71,
87-93, 104-110, 127-132, 135-137,
148, 162, 167
ion plasma frequency, 90
ion sound speed,
see ion acoustic velocity
ion sound wave,
see ion acoustic wave
ion tail formation, 130-132, 162
isothermal equation of state, 8, 10
kinetic effects, 4, 5, 12, 14, 17, 55, 79,
91,96-99, 146, 147
Landau damping, 96-101, 132, 135
laser beam smoothing, 170
light wave propagation, 27-43, 46-52,
76, 88, 93, 128
long-pulse-length experiments, 160-168
Index
Maxw=ll's equations, 10
Monte Carlo calculation, 147
rnultigroup diffusion, 146
nonlinear wave particle interactions,
100-108, 110, 121-125,129--137
non linear behavior of instabilities,
104-110,127-138,
obliquely incident light wave, 37-43, 51,
52,121, 156-158
oscillating two stream instability,
61-70,104-108
oscillatory velocity, 40
p-polarized light wave, 39-43, 121-125,
156-158
particle codes, 17-25
phase space distribution, 6
phase velocity, 10-12,96-101
plane of incidence, 37
plane of polarization, 78
plasma expansion, 116-121, 154, 155
plasma wave collapse, 108-110
Poisson's equation, 10
polarization-dependent absorption,
37-43, 156-158
ponderomotive force, 6(}-{)2, 76, 88, 89,
93,109,117-121,123,137
preheat, 74, 81, 127, 133
pressure tensor, 6
profile steepening, 117-125, 155, 156
pulse length, 154, 155
quasi-mode, 68, 91
quasi-neutrality, 14
Raman instability, 73-81, 127,
132-135,137,153,154,162-166
l1l 1
Raman scatteri~g; .
see Raman instability
relati vistic self-Iocusi ng, 93, 94, 137
resonance absorption, 39-43, sr. ns,
121-12;:;,153, 154, 156-160
resonant particles, 96-102. 106, 129.
UO, n5
rippled critical surface, 156
s-polarized light wave, 38, 39, 51, 157.
158
second harmonic generation, 167
self-focusing, 93, 94, 137
self-generated magnetic fields, 13.5,
147-149, 153, 158, 168
short-pulse-length experiment, 155-160
sideband instability, 101
sidescatter, 78, 79, 81, 91, 92, 161
stimulated Compton scattering, 79
stimulated scattering, see Raman and
Brillouin instabilities
super-Gaussian distribution, 147
suprathermal electrons,
see hot electrons
swelling of the electric field, 31, 34
thin foil targets, 163, 164
Thomson scattering, 162, 166, 167
three-halves harmonic emission, 167
threshold condition for instability,
67-70, 79-81, 83, 91-93, 164, 167
transport experiments, 149
trapped particles, 100-102, lOO, 108,
122, 129, 130
two-fluid equations, 4, 6-10
two-plasmon decay instability, 81-84,
125, 135-137, 153, 166, 167
underdense plasma, 88, 127-138,
153-155, 160-167
182 Index

velocity distribution function, 4, 5,

52-55, 98, 102, 106, 124, 125, 129-
136,144-148,158, 159, 165-167
velocity gradient, 92
Vlasov equation, 4-8, 10, 45, 76, 98, 99

wave breaking, 101-104, 106, 122, 129,

130
wave equation, 12, 13, 27-30, 32, 38,
47-49, 58, 64, 65, 75-77, 82, 88, 89,
109, 128
wavelength scaling, 154, 159, 160,
168-170
wavenumber matching, 73, -80, 81, 87,
92
weak turbulence theory, 108, 109
Weibel instability, 135

x-ray emission, 149, 150, 159, 160,

165, 167-169

Zakharov equations, 109, no