Analytical estimate of phase mixing time of longitudinal Akhiezer-Polovin waves

Arghya Mukherjee and Sudip Sengupta

Citation: Physics of Plasmas (1994-present) 21, 112104 (2014); doi: 10.1063/1.4901194

View online: http://dx.doi.org/10.1063/1.4901194
View Table of Contents: http://scitation.aip.org/content/aip/journal/pop/21/11?ver=pdfcov
Published by the AIP Publishing

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
14.139.114.54 On: Mon, 10 Nov 2014 14:22:22
 PHYSICS OF PLASMAS 21, 112104 (2014)

Analytical estimate of phase mixing time of longitudinal Akhiezer-Polovin

waves
Arghya Mukherjeea) and Sudip Sengupta
Institute for Plasma Research, Bhat, Gandhinagar 382 428, India
(Received 5 August 2014; accepted 27 October 2014; published online 10 November 2014)
Phase mixing of a longitudinal Akhiezer-Polovin wave subjected to a small amplitude longitudinal
perturbation and its eventual breaking is studied analytically. It is well known that longitudinal
Akhiezer-Polovin wave subjected to arbitrarily small longitudinal perturbation breaks via the pro-
cess of phase mixing at an amplitude well below its limiting amplitude [Verma et al., Phys. Rev.
Lett. 108, 125005 (2012)]. We analytically show that the phase mixing time (breaking time, xpsmix)
scales with b (phase velocity) and um(maximum fluid velocity) as xp smix  2pb 2
3d 1=um  1=4,
where d is the amplitude of velocity perturbation and xp is the non-relativistic plasma frequency.
This analytical dependence of phase mixing time on b, um, and d is further verified using numerical
simulations based on Dawson sheet model. V C 2014 AIP Publishing LLC.

[http://dx.doi.org/10.1063/1.4901194]

I. INTRODUCTION the wave to break.16–19 This process of wave breaking

through gradual phase mixing is not exhibited by a pure
The space-time evolution of nonlinear oscillations and
Akhiezer-Polovin wave.
waves in cold plasmas has been an interesting and funda-
Recently, Verma et al.20 constructed a longitudinal
mental topic of research over several decades. Beginning
Akhiezer-Polovin travelling wave solution from the exact
with the pioneering work of Akhiezer and Polovin1 and
space time dependent solution of Infeld and Rowlands, using
Dawson,2 this subject has retained its importance till date
appropriate choice of initial conditions. It was further shown
because it serves as a useful paradigm to elucidate the under-
by the same authors,21 that even a longitudinal Akhiezer-
lying physics behind a wide range of physical phenomena.
Polovin wave breaks through the gradual process of phase
Examples range from laboratory based laser/beam plasma
mixing at an amplitude well below its wave breaking limit,
interaction experiments3–11 to some astrophysical phenom-
when it is subjected to arbitrarily small longitudinal pertur-
ena,12,13 where large amplitude oscillations and waves are
bations. It was found through extensive numerical simula-
usually encountered.
tions that phase mixing time decreases with increasing
In 1956, Akhiezer and Polovin1 obtained an exact one-
um(for a fixed d) and increasing d(for a fixed um), where um
dimensional longitudinal travelling wave solution in a cold
and d are, respectively, the velocity amplitudes of the
plasma, including relativistic mass effects. This travelling
Akhiezer-Polovin wave and the applied perturbation. In this
wave solution is obtained by solving the cold plasma relativ-
paper, we present a weakly relativistic calculation which
istic fluid-Maxwell equations in a stationary wave frame.
analytically brings out the scaling of phase mixing time with
The maximum electric field amplitude (EWB) of this wave is
the parameters of the Akhiezer-Polovin wave (um and b; b is
limited pbyffiffiffi the wave breaking limit which is given by
pffiffiffiffiffiffiffiffiffiffiffiffiffi
1=2 1,14 the phase velocity of the Akhiezer-Polovin wave) and the
eEWB
¼ 2 ðcph  1Þ : Here, cph ¼ 1= 1  b2 is the
mxp c amplitude of velocity perturbation d. We have further veri-
Lorentz factor associated with the normalized phase velocity
fied this scaling numerically, using a code based on Dawson
b (b ¼ vph/c) of the Akhiezer-Polovin wave, e and m are,
sheet model.2,22
respectively, the charge and mass of an electron, c represents
In Sec. II, we briefly describe the Dawson sheet model
the velocity of light in free space, and xp is the non-
and construct a longitudinal Akhiezer-Polovin travelling
relativistic plasma frequency. In 1989, Infeld and
wave solution using this model.19 This derivation, although
Rowlands15 presented an exact space-time dependent solu-
equivalent to that presented in Refs. 20 and 21 is physically
tion of the relativistic cold plasma fluid-Maxwell equations
more transparent. It is presented here for the sake of com-
in Lagrange coordinates. The solution presented by them
pleteness. Section III is devoted to weakly relativistic analyt-
shows explosive behaviour for all initial conditions except
ical calculation of phase mixing time of longitudinal
for the one, which are needed to excite Akhiezer-Polovin
Akhiezer-Polovin wave subjected to a small amplitude sinu-
waves. Physically, this explosive behaviour arises due to rel-
soidal perturbation. In Sec. IV, analytical predictions are
ativistic variation of electron mass, which causes the charac-
compared with numerical findings. Finally, Sec. V contains a
teristic electron plasma frequency to acquire a spatial
discussion and summary of our results.
dependence. Due to this dependence, neighbouring electrons
gradually move out of phase and eventually cross, causing
II. RELATIVISTIC TRAVELLING WAVE SOLUTION
According to the Dawson sheet model description of a
a)
Electronic mail: arghya@ipr.res.in cold plasma, electrons are assumed to be infinite sheets of

1070-664X/2014/21(11)/112104/6/$30.00 21, 112104-1 C 2014 AIP Publishing LLC

V

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
14.139.114.54 On: Mon, 10 Nov 2014 14:22:22
 112104-2 A. Mukherjee and S. Sengupta
charge embedded in a cold immobile positive ion back-
ground.22 Evolution of any coherent mode can be studied in
terms of oscillating motion of these sheets about their equi-
librium positions. Let xeq and n(xeq, s), respectively, be the
equilibrium position and displacement from the equilibrium
position of an electron sheet. In terms of xeq and n(xeq, s), the
associated fluid quantities, viz., number density, velocity,
and self-consistent electric field can, respectively, be written
_ and
as nðxeq ; sÞ ¼ n0 =ð1 þ @nðxeq ; sÞ=@xeq Þ; vðxeq ; sÞ ¼ n,
Es ðxeq ; sÞ ¼ 4pen0 nðxeq ; sÞ, where e is the charge of an
electron and n0 is the equilibrium density of electrons. Here,
dot represents differentiation w.r.t Lagrange time s. These
expressions can further be represented in terms of Euler
coordinates (x,t) using the transformations x ¼ xeq þ
nðxeq ; sÞ and t ¼ s. Thus, with the knowledge of n(xeq, s), the
problem of space-time evolution of a coherent mode is
essentially solved. n(xeq, s) can be obtained by solving the
relativistic equation of motion of a sheet which may be writ-
ten as
::
n
 3=2 þ n ¼ 0:
2 nap ðxeq ; 0Þ ¼
1  n_
Here, we have used Gauss’s law and the following normal-
x n _
ization: s ! xp s; n ! cp ; n_ ! nc , Es ! mx
eEs
, where xp is
pc
the nonrelativistic plasma frequency. The rest of the paper is
_ we get
in normalized units. Multiplying Eq. (1) by n,
n2
1 Eða; r Þ  r 0 Fða; rÞ ¼  :
 1=2 þ ¼ aðxeq Þ:
2 2
1  n_
Here, “a(xeq)” corresponds to the total energy of a sheet.
Substituting
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n ¼ 2ða  1Þ sin a (3)
in Eq. (2), solution of Eq. (1) becomes
2
s¼ Eða; rÞ  r0 Fða; rÞ þ Uðxeq Þ:
r0
This gives a as an implicit function offfi s and xeq, where r ¼
pffiffiffiffiffiffiffiffiffiffiffiffi
½ða  1Þ=ða þ 1Þ1=2 and r0 ¼ 1  r2 . E(a, r) and F(a, r)
are incomplete elliptic integrals of second and first kind,23
respectively, and Uðxeq Þ is an integration constant. Equation
(3) along with Eq. (4) describes the motion of an electron
sheet about its equilibrium position for a given set of initial
conditions Uðxeq Þ and r(xeq). Frequency x of an electron
sheet is obtained by integrating Eq. (2) between two turning
points (n_ ¼ 0Þ and is given by
p r0
x¼  :
2 2Eðr Þ  r 02 K ðr Þ
Here, E(r) and K(r) are complete elliptic integrals of second
and first kind,23 respectively. It is evident from Eq. (5) that for
arbitrary set of initial conditions, “x” is in general a function
of “xeq.” Since any coherent mode is made up of a large num-
ber of electron sheets oscillating about their equilibrium
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
14.139.114.54 On: Mon, 10 Nov 2014 14:22:22
Phys. Plasmas 21, 112104 (2014)
positions, this spatial dependency of x causes the neighbour-
ing electron sheets to gradually go out of phase with time.
This eventually leads to crossing of electron sheet trajectories
resulting in singularities in the electron density profile. This is
the phenomenon of phase mixing leading to wave breaking.
For a sinusoidal initial density profile and for wave like initial
conditions, the phenomenon of phase mixing leading to wave
breaking is convincingly demonstrated in Refs. 15–17, and 24.
As stated in the Introduction, in Ref. 20, it is shown that
it is possible to choose a special set of initial conditions which
excites a propagating solution with phase velocity b, which
does not phase mix and break. This propagating solution is
nothing but a longitudinal Akhiezer-Polovin wave. Absence
of phase mixing implies, from Eq. (5), that “a” (energy of an
oscillating sheet) should be independent of “xeq.”
Furthermore, propagation with a fixed phase velocity b deter-
mines the functional form of Uðxeq Þ as Uðxeq Þ ¼ xeq =b. This
form of Uðxeq Þ is obtained by choosing n (hence a) to be
entirely a function of w ¼ x(t  x/b). Thus, the initial condi-
tions for exciting a longitudinal Akhiezer-Polovin wave are
(1) 2r
sin a; (6)
r0
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
_n ap ðxeq ; 0Þ ¼ 2r cos a 1  r sin a ; (7)
r02 1 þ 2r 2
r02 cos a
2
along with aðxeq ; 0Þ implicitly given by
2 xeq
(8)
(2) r0 b
Following Akhiezer-Polovin’s work,1 we now choose
um(maximum fluid velocity) and b as independent parame-
ters instead of a(or r) and b. a and um are relatedffi to each
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
other through Eqs. (1) and (2) as a ¼ 1= 1  um 2 . In Sec.
III, we add a small perturbation to nap ðxeq ; 0Þ and n_ ap ðxeq ; 0Þ
which leads to phase mixing and subsequent breaking of lon-
gitudinal Akhiezer-Polovin wave.
(4) III. ESTIMATION OF PHASE MIXING TIME
Adding a small amplitude sinusoidal perturbation with
velocity amplitude d and wavelength kap (same as the longi-
tudinal Akhiezer-Polovin wave) to nap and n_ ap , we get
d
nper ¼ nap  sinðkap xeq Þ; (9)
xap
n_ per ¼ n_ ap þ d cosðkap xeq Þ: (10)
Here, nap and n_ ap are the required initial conditions for excit-
ing a longitudinal Akhiezer-Polovin wave20 and nper, n_ per are
the perturbed initial conditions. Here, xap is the frequency
(5)
(normalized to nonrelativistic plasma frequency xp) of the
Akhiezer-Polovin wave. The perturbed initial conditions are
equivalent to adding a small amplitude sinusoidal density
perturbation propagating with phase velocity b, to longitudi-
nal Akhiezer-Polovin wave. In the weakly relativistic limit,
keeping terms linear in d, the energy associated with an elec-
tron sheet becomes
 112104-3 A. Mukherjee and S. Sengupta Phys. Plasmas 21, 112104 (2014)

" #  
nap d 2pb 1 1
aper aþ  sinðkap xeq Þ þ dn_ ap cosðkap xeq Þ ; (11) smix   : (16)
xap 3d u2m 4

2 It is clear from the above expression that phase mixing time

where a  1 þ n2 =2 þ n_ =2. This in turn gives
scales directly with b, inversely with d and has a u2
m depend-
" # ence on um.
2 2 2r 2 nap _
rper  r  2 d sinðkap xeq Þ  n ap cosðkap xeq Þ : In Sec. IV, we verify these predictions numerically
a  1 xap using Dawson sheet simulation.
(12)
IV. NUMERICAL RESULTS
Now using Eq. (5), and expanding the complete elliptic inte-
2
grals upto order rper , the frequency of oscillation in the per- Using a code based on Dawson sheet model, we numeri-
turbed case can be approximately written as cally verify the process of phase mixing of a large amplitude
Akhiezer-Polovin wave perturbed by a small amplitude sinu-
2 soidal perturbation. We first load Akhiezer-Polovin type ini-
3rper
xper  1  : (13) tial conditions with a sinusoidal perturbation of velocity
4
amplitude d in a one-dimensional relativistic sheet code con-
2
Finally, substituting rper in the above equation, the frequency taining 10 000 electron sheets. Using these initial condi-
of oscillation in the weakly relativistic limit stands as tions and using periodic boundary conditions, the equation
of motion for each sheet is then solved using fourth order
3r2 Runge-Kutta scheme. At each time step, ordering of sheets is
xper  1 
4 checked for sheet crossing (electron trajectory crossing).
" #
3r 2 nap Phase mixing time is measured as the time taken by any two
þ d sinðkap xeq Þ  n_ ap cosðkap xeq Þ : adjacent sheets to cross over. We terminate our code at this
2ða2  1Þ xap
time because the expression for self-consistent electric field
(14) (Es ¼ 4pen0n) used in equation of motion (Eq. (1)) becomes
invalid beyond this point.2,25
It is clear from the above expression that the frequency of
Fig. 1 shows the space time evolution of the electron
the wave is dependent on the equilibrium position of the
density profile of the resultant structure. As time progresses,
electrons which leads to the phenomena of phase mixing.21
the density profile becomes more and more spiky as energy
Following Dawson’s argument,2 the phase mixing time
is irreversibly transferred to higher and higher harmonics.
(smix) depends on the spatial derivative of frequency as
Due to longitudinal perturbation, the energy which was ini-
smix  p=2nmax ðdxper =dxeq Þ. Differentiating Eq. (14) w.r.t
tially loaded on the Akhiezer-Polovin mode goes to higher
xeq and, respectively, using Eqs. (6)–(8) to evaluate the
harmonics, then partially returns to the original mode, again
derivatives dnap =dxeq ; d n_ ap =dxeq , and da/dxeq, the maximum
goes to higher harmonics and so on. This partial back-and-
value of dxper/dxeq (at a  p/2) is given by
forth sloshing of energy between different harmonics eventu-
ally results in accumulation of energy at higher harmonics.
dxper 6d r 3 1 This is nothing but phase mixing type of damping of the
 : (15)
dxeq b r 0 a2  1 original Akhiezer-Polovin mode which ultimately breaks at
an amplitude well below the Akhiezer-Polovin limit.16,17,26
Finally, substituting the above expression in Dawson for- A manifestation of this process is also seen in the Fourier
mula and using nmax ¼ 2r=r0 , (at a  p/2), the phase mixing spectrum where we see a broadening of the spectrum as time
time (smix, upto order u2m ) is given by progresses (Fig. 2). Thus, the energy which was loaded in

FIG. 1. Space-time evolution of elec-

tron density for an Akhiezer-Polovin
wave with velocity amplitude
um ¼ 0.55, with perturbation amplitude
d ¼ 0.1 and with phase velocity
b ¼ 0.9995.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
14.139.114.54 On: Mon, 10 Nov 2014 14:22:22
 112104-4 A. Mukherjee and S. Sengupta Phys. Plasmas 21, 112104 (2014)

FIG. 2. Fourier spectrum of an Akhiezer-Polovin wave with velocity amplitude um ¼ 0.55, with perturbation amplitude d ¼ 0.1 and phase velocity b ¼ 0.9995
at different time steps.

FIG. 3. Momentum distribution of an

Akhiezer-Polovin wave with velocity
amplitude um ¼ 0.55, with perturbation
amplitude d ¼ 0.1 and with phase ve-
locity b ¼ 0.9995 at different time
steps.

FIG. 4. Analytical (solid) and numeri-

cal (circles) scaling of phase mixing
time for a finite amplitude Akhiezer-
Polovin wave for um ¼ 0.20(4a),
0.55(4b) and b ¼ 0.9995 as a function
of perturbation amplitude (d).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
14.139.114.54 On: Mon, 10 Nov 2014 14:22:22
 112104-5 A. Mukherjee and S. Sengupta Phys. Plasmas 21, 112104 (2014)

FIG. 5. Analytical (solid) and numeri-

cal (circles) scaling of phase mixing
time as a function of velocity ampli-
tude of Akhiezer-Polovin wave (um) in
the presence of a finite perturbation
(d) ¼ 0.01 (5(a)), 0.1 (5(b)) and
b ¼ 0.9995.

FIG. 6. Analytical (solid) and numeri-

cal (circles) scaling of phase mixing
time as a function of phase velocity b
for a fixed amplitude of Akhiezer-
Polovin wave um ¼ 0.2, d ¼ 0.01 (6(a))
and um ¼ 0.1, d ¼ 0.1 (6(b)).

FIG. 7. Analytical (solid) and numeri-

cal (circles) scaling of phase mixing
time as a function of density amplitude
of Akhiezer-Polovin wave D in the
presence of a finite perturbation
(d) ¼ 0.01 (7(a)), 0.1 (7(b)), and
b ¼ 0.9995.

the primary mode eventually distributes over higher modes.
Interaction of these high “k” modes with the particles
(sheets) accelerates the particles, causing the initial delta-
function momentum distribution to spread. Fig. 3 shows that
as time progresses, the momentum distribution function
gradually spreads generating multi-stream flow; a clear indi-
cation of phase mixing leading to breaking. Figures 4–6,
respectively, show the variation of phase mixing time with
respect to d, um, and b for fixed values of the other two pa-
rameters. In all the cases, points represent the simulation
results and the solid line represents our scaling obtained
from Eq. (16). We note here that our analytical expression
for phase mixing time has been derived in the limit of weak
relativistic effects. As a result, we have kept terms which are
linear in d and in addition have neglected terms which are of
order higher than u2m . Thus, the expression for phase mixing
time is approximate in d and um, but exact in b. Therefore,
the numerical results in Fig. 4 (variation of smix with d) for
small values of d (upto  0.12) and in Fig. 6 (variation of
smix with b) fit the analytical expression quite well, whereas
the numerical result in Fig. 5 matches the analytical expres-
sion reasonably upto um  0.4. Hence in all the cases, the an-
alytical expression (Eq. (16)) shows a reasonably good fit to
the observed numerical results, thus vindicating our weakly
relativistic calculation.
V. DISCUSSION AND SUMMARY
The phenomenon of phase mixing is a manifestation of
spatially dependent plasma frequency.17 It is well known
that a large amplitude longitudinal Akhiezer-Polovin wave
breaks via the process of phase mixing at an amplitude well

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
14.139.114.54 On: Mon, 10 Nov 2014 14:22:22
 112104-6 A. Mukherjee and S. Sengupta Phys. Plasmas 21, 112104 (2014)

7
below the breaking amplitude for Akhiezer-Polovin wave H. Schwoerer, S. Pfotenhauer, O. J€ackel, K. U. Amthor, B. Liesfeld, W.
pffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffi Ziegler, R. Sauerberg, K. W. D. Ledingham, and T. Esirkepov et al.,
ð 2 cph  1Þ, when subjected to an arbitrarily small longi-
Nature 439, 445 (2006).
tudinal perturbation.21 We have derived an expression for 8
J. Faure, C. Rechatin, A. Norlin, A. Lifschitz, Y. Glinec, and V. Malka,
phase mixing time which brings out its dependence on um, b, Nature (London) 444, 737 (2006).
9
and d. Our weakly relativistic calculation indicates that the N. H. Matlis, S. Reed, S. S. Bulanov, V. Chvykov, G. Kalintchenko, T.
Matsouka, P. Rousseau, V. Yanovsky, A. Maksimchuk, S. kalmykov, G.
phase mixing time scales linearly with b, inversely with d
Shvets, M. C. Downer et al., Nat. Phys. 2, 749 (2006).
and has 1=u2m dependence on um. We have verified this scal- 10
C. Rechatin, J. Faure, A. Ben-Ismail, J. Lim, R. Fitour, A. Speckam, H.
ing using numerical simulations. Videau, A. Tafzi, F. Burgy, and V. Malka, Phys. Rev. Lett. 102, 164801
We note here that, the dependence of phase mixing time (2009).
11
E. Esarey, C. B. Schroeder, and W. P. Leemans, Rev. Mod. Phys. 81, 1229
on D (density amplitude, ð~n =n0 Þmax ¼ ðn=n0  1Þmax ) can be
   (2009).
2 12
G. J. J. Botha, T. N. Arber, V. M. Nakariakov, and F. P. Keenan, Astron.
2pb 1þD 1
obtained from Eq. (16) as smix  3d bD  4 by elimi- Astrophys. 363, 1186 (2000).
13
Y. Voitenko, J. Andries, P. D. Copil, and M. Goossens, Astron. Astrophys.
nating um using n ¼ n0b/(b  v). This shows that for D  1, 437, L47 (2005).
smix is essentially independent of D and for D  1, smix 14
P. Gibbon, Short Pulse Laser Interaction with Matter (Imperial College
scales as 1/bdD2. Figs. 7(a) and 7(b) show the dependence Press, 2005).
15
of phase mixing time on D. We emphasize here that for D  E. Infeld and G. Rowlands, Phys. Rev. Lett. 62, 1122 (1989).
16
S. Sengupta, V. Saxena, P. K. Kaw, A. Sen, and A. Das, Phys. Rev. E 79,
1, for a given b and d  D, our expression for phase mixing 026404 (2009).
time exhibits 1/D3 scaling, in conformity with the results 17
S. Sengupta, P. Kaw, V. Saxena, A. Sen, and A. Das, Plasma Phys.
presented in Refs. 15,16, and 27. 18
Control. Fusion 53, 074014 (2011).
C. Maity, A. Sarkar, P. K. Shukla, and N. Chakrabarti, Phys. Rev. Lett.
110, 215002 (2013).
1 19
A. I. Akhiezer and R. V. Polovin, Sov. Phys. JETP 3, 696 (1956). S. Sengupta, AIP Conf. Proc. 1582, 191–200 (2014).
2 20
J. M. Dawson, Phys. Rev. 113, 383 (1959). P. S. Verma, S. Sengupta, and P. Kaw, Phys. Plasmas. 19, 032110 (2012).
3 21
T. Tajima and J. M. Dawson, Phys. Rev. Lett. 43, 267 (1979). P. S. Verma, S. Sengupta, and P. Kaw, Phys. Rev. Lett. 108, 125005
4
A. Modena, Z. Najmudin, A. E. Dangor, C. E. Clayton, K. A. Marsh, C. (2012).
22
Joshi, V. Malka, C. B. Darrow, C. Danson, D. Neely, and F. N. Walsh, J. M. Dawson, Phys. Fluids 5, 445 (1962).
23
Nature 377, 606 (1995). M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions
5
V. Malka, S. Fritzler, E. Lefebre, M. M. Aleonard, F. Burgy, J. P. (Dover Publications, INC, New York).
24
Chambaret, J. F. Chemin, K. Krushelnick, G. Malka, S. P. D. Mangles, Z. J. F. Drake, Y. C. Lee, K. Nishikawa, and N. L. Tsintsadze, Phys. Rev.
Najmudin, M. Pittman, J. P. Rousseau, J. N. Scheurer, B. Walton, and A. Lett. 36, 196 (1976).
25
E. Dangor, Science 298, 1596 (2002). S. Sengupta and P. K. Kaw, Phys. Rev. Lett. 82, 1867 (1999).
6 26
B. M. Hegelich, B. J. Albright, J. Cobble, K. Flippo, S. Letzring, M. P. K. Kaw, A. T. Lin, and J. M. Dawson, Phys. Fluids 16, 1967 (1973).
27
Paffett, H. Ruhl, J. Schrciber, R. K. Schulze, J. C. Fernandez et al., Nature L. M. Gorbunov, A. A. Frolov, E. V. Chizkonkov, and N. E. Andreev,
439, 441 (2006). Plasma Phys. Rep. 36, 345 (2010).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
14.139.114.54 On: Mon, 10 Nov 2014 14:22:22