Self-generation of megagauss magnetic fields during the

expansion of a plasma
Cédric Thaury, P Mora, A Héron, Jean-Claude Adam

To cite this version:

Cédric Thaury, P Mora, A Héron, Jean-Claude Adam. Self-generation of megagauss magnetic fields
during the expansion of a plasma. Physical Review E : Statistical, Nonlinear, and Soft Matter Physics,
2010, pp.016408. �10.1103/PhysRevE.82.016408�. �hal-01166850�

HAL Id: hal-01166850

https://hal.science/hal-01166850
Submitted on 23 Jun 2015

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci-
entific research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
 PHYSICAL REVIEW E 82, 016408 共2010兲

Self-generation of megagauss magnetic fields during the expansion of a plasma

C. Thaury, P. Mora, A. Héron, and J. C. Adam

Centre de Physique Théorique, Ecole Polytechnique, CNRS, 91128 Palaiseau, France
共Received 4 December 2009; revised manuscript received 13 April 2010; published 27 July 2010兲
The expansion of a plasma slab into a vacuum is studied using one-dimensional and two-dimensional
particle-in-cell simulations. As electrons transfer their longitudinal kinetic energy to ions during the expansion,
the electron temperature becomes anisotropic. Once this anisotropy exceeds a threshold value, it drives the
Weibel instability, leading to magnetic fields in the megagauss range. These fields induce energy transfer
between the longitudinal and transverses directions, which influences the expansion. The impact of a cold
electron population on this phenomenon is also investigated.

DOI: 10.1103/PhysRevE.82.016408 PACS number共s兲: 52.35.⫺g, 52.38.Fz, 52.38.Kd, 52.65.Rr

Plasma expansion is a fundamental process which occurs
in very different fields, such as astrophysics 关1,2兴, laser-
plasma ion acceleration 关3–5兴 and thin-film deposition 关6兴.
This phenomenon is usually described by simple one-
dimensional models 关7–9兴. Yet, even when the system is
translation-invariant along the plasma surface, several effects
共e.g., Coulomb collisions 关10兴兲 can induce momentum trans-
fer between the longitudinal and transverse directions. The
purely one-dimensional 共1D兲 description is thus, in general,
the slab. The mass ratio is me / mi = 1836 with Z = 1, and the
electron temperatures are T⬜0 = T储0 = 5 keV. The code is run
with a time step ⌬t = 0.2␻−1 pe , a mesh size ⌬x ⬇ c⌬t, and there
are 2 ⫻ 104 particles in each mesh.
Figure 2共a兲 displays the time evolution of the total longi-
tudinal 共Ek兲 and transverse 共E⬜
储
k 兲 kinetic energies, and of the
total magnetic energy 共E M 兲. It shows that the global aniso-
tropy E⬜
储
k / Ek − 1 rises as the plasma expands. At t = t0
⬇ 28␻ pi , the magnetic energy starts to increase, which indi-
−1

inaccurate. In this paper, we show that self-generated mag-
netic fields can lead to such momentum transfer during the
expansion of a collisionless plasma slab. This study is of
particular interest in the context of laser-plasma ion accelera-
tion, where an intense laser pulse is focused on a thin foil to
create a hot electron population that transfers progressively
its energy to ions via the ambipolar electric field at the
plasma surface 关11兴.
We assume here that the electron distribution is initially
Maxwellian with an isotropic temperature. As the plasma
expands, the longitudinal temperature T储 decreases and the
anisotropy parameter A = T⬜ / T储 − 1 increases, which eventu-
ally leads to the growth of the Weibel instability 关12–18兴.
The most unstable modes are obtained for k = kxex, where ex
is a unit vector normal to the plasma surface. In this case, the
maximum unstable wave vector is
km 冑
x = A共␻ pe/c兲,
where ␻ pe is the plasma frequency 关12兴 关Fig. 1共a兲兴. A more
general expression for km can be obtained from the linear
dispersion relation, considering that k = 共kx , ky , 0兲 makes an
angle ␪ with ex 关19兴. The result depends on the orientation of
cates that the anisotropy is sufficient for the Weibel instabil-
ity to grow. Then, for t ⬎ ti ⬇ 40␻−1 pi , the magnetic field drives
the isotropization of the electron temperature, which eventu-
ally leads to the saturation of the instability at t ⬇ 48␻−1 pi .
According to Eq. 共1兲, unstable modes should develop as soon
as kmx ⱖ 2␲ / L, that is A ⱖ 0.016 for L = 50c / ␻ pe. If one esti-
mates the anisotropy based on the mean kinetic energies, the
instability should start to grow around ␻ pit = 4, which is in-
consistent with Fig. 2共a兲. To understand this discrepancy, we
focus on Fig. 2共b兲, where A and Bz are plotted as a function
of x, at ␻ pit = 40. The anisotropy is observed to be much
larger on the sides of the plasma than in its center 共the limit
between these two parts being the rarefaction front兲. This is
because the center of the plasma is mainly occupied by slow
electrons which do not give any energy to ions 共see Fig. 3 in
Ref. 关20兴兲. As a result of this inhomogeneous anisotropy,
unstable modes grow on the sides of the plasma, as shown by
共1兲 Fig. 2共b兲.
We now take into account this effect to determine when
the instability is triggered. We assume that the expansion is
(a) (b)
A=1 1
0.6

the magnetic field B which can grow either along the z-axis,
or in the 共ex , ey兲 plane 共B␪ field兲. In Fig. 1共b兲, we observe
kx c/ωpe

A=0.6
Γ/ωpi

that, in both cases, kmx decreases with ␪ faster than cos ␪. This 0.3
m

means that the largest km are obtained for ␪ = 0. We notice A=0.2 Bz

however that for small angles km x 共␪兲 ⬇ kx 共0兲, for both orien-
m
Bθ
tations of B. 0.0
cos θ
0
To study the growth of the Weibel instability in the con- 0.0 0.5 1.0 0 20 40 60
text of plasma expansion, we first use 1D2V relativistic col- kxc/ωpe θ (degree)

lisionless particle-in-cell 共PIC兲 simulations, for which k
= kxex, and B = Bzez. Initially, ions occupy a slab of thickness
L = 50c / ␻ pe, while electrons are in Maxwell-Boltzmann equi-
librium with the self consistent electrostatic potential. A
vacuum region extends for about 2500c / ␻ pe on each side of
FIG. 1. Growth rate and maximum wave vector of the Weibel
instability. 共a兲 Growth rate ⌫ as a function of kx for 3 different
anisotropy parameters A and T⬜0 = 5 keV. 共b兲 Maximum wave vec-
tor of the unstable modes Bz and B␪, as a function of the angle
between k and the plasma normal ex, for A = 1.

1539-3755/2010/82共1兲/016408共4兲 016408-1 ©2010 The American Physical Society

THAURY et al. PHYSICAL REVIEW E 82, 016408 共2010兲

Kinetic energies (arb. units)

Magnetic energy (arb. units)

10
-2 50
1 (a) (c)
50
-3
10

ωpi t0
⎪ ⎢
-4 1D2V
Ek EM 10 2D3V
⊥ 0 0
(a) Ek
-5
0 10

Bz / (mωpe/c)
0 50 100 150
ωpit
0.01 0.02
A 1
1D2V

m
Bz (arb. unit)
(b) 2D3V (d)
Anisotropy

Bz
1 0.00 0.00
0 100 200 0.0 0.1 0.2
Lωpe/c v⊥0 / c

(b) FIG. 3. 共Color online兲 Simple model. 共a-b兲 Influence of the

0 0
-50 0 50
plasma width L on the triggering time of the instability t0, and on
xωpe/c the peak magnetic field Bzm, for v⬜0 = 0.1c. 共c-d兲 Influence of the
initial thermal velocity v⬜0 on t0 and Bzm, for L = 10c / ␻ pe. The dots
FIG. 2. Growth of the Weibel instability during a 1D2V expan- result from PIC simulations.
sion. 共a兲 Integrated kinetic 共E⬜
储
k , Ek兲 and magnetic 共E M 兲 energies as
a function of time, in linear and logarithmic scales, respectively. − m ev ⬜ 2
/ 2kBT⬜兲 is divided in isotropic and anisotropic parts,
The same normalization is used for both Ek and E M . 共b兲 Anisotropy f e = f e + f Ae , with f 0e ⬀ exp共−mev2 / 2kBT储兲. The magnetic field
0
parameter A and magnetic field Bz as a function of space, at ␻ pit does not perturb f 0e . In contrast, it deflects electrons associ-
= 40.
ated to f Ae in different directions, depending on the sign of
their transverse velocity vy. Actually, for Bz ⬎ 0 the Lorentz
self-similar and that the width of the part of the plasma on longitudinal force −evyBz / m, pushes electrons with vy ⬍ 0 in
which the instability grows is LA = ␤cst, where cs the positive x direction, and electrons with vy ⬎ 0 in the op-
= 共kBT储0 / mi兲1/2 is the ion acoustic velocity and ␤ ⬇ 3 is cho- posite direction. At most, all electrons of f Ae with vy ⬍ 0 共re-
sen to reproduce simulation results 关see e.g., Fig. 2共b兲兴. Then spectively vy ⬎ 0兲 can pull out of a part of the plasma where
we use results of Refs. 关20,21兴, to obtain an estimate of the they are replaced by electrons with vy ⬎ 0 共respectively, vy
anisotropy A around the center of mass of the anisotropic ⬍ 0兲, resulting in the local distribution function f m 0
e = fe
area 共at x = ⫾ L / 2兲. According to these papers the distribution + 2H共⫾vy兲f e , where H is the heaviside step function. The
A
function in the expansion direction is, to first order in t, peak current jmy is simply obtained by averaging −evy over
储 储 fme,
f a共x, v,t兲 ⬇ 关1 + 共4 − 8u/3兲ucst/L兴f e0共v兲 共2兲
储
where f e0共v兲 is the initial Maxwellian distribution function, jmy = neev⬜0冑2/␲关1 − 共A + 1兲−1/2兴. 共4兲
u = 关mev2储 / 2 − e⌽共x , t兲兴kBT储0 the normalized total electron en- For the sake of simplicity, we now approximate ne by its
ergy, and ⌽共x , t兲 = −关1 + 共兩x 兩 −L / 2兲 / cst兴kBT储0 / e the electro- mean value over LA 共ne ⬇ ne0 / ␤, where ne0 is the total initial
static potential. From Eq. 共2兲, the longitudinal temperature electron density兲, and we assume that j y is sinusoidal. As
储
T储共t兲 is computed by averaging mev2储 over f e. For x = ⫾ L / 2, both j y and Bz vanish on the sides of LA, Bz must be of the
the potential is ⌽共x , t兲 = −kBT储0 / e, and the averaging leads to form Bzm关1 + cos kmx 共x − x0兲兴 / 2 were Bz = 2␮0 j y / kx , and x0 is
m m m

T储共t兲 ⬇ T储0共1 − ␣cst / L兲 with ␣ = 32/ 3. Thus the anisotropy for the center of the anisotropic area. Substituting for jmy , we get
t Ⰶ L / cs is A = T⬜0 / T储共t兲 − 1 ⬇ ␣cs / L. Note that in Fig. 2共b兲, A
is slightly lower than ␣cst / L because at this time it has
stopped to increase.
Bzm =
2
␤
冑 2 m␻ pe v⬜0 −1/2
␲ e c
A 关1 − 共A + 1兲−1/2兴. 共5兲
The instability starts to grow when 2␲ / LA = km x , that is
when 2␲ / ␤cst = 冑␣cst / L␻ p / c. From this condition, we de- This function is plotted in Fig. 3共b兲 and 3共d兲 for A
rive the triggering time of the instability, = ␣cst0 / L, together with PIC results. The agreement is sur-
prisingly good, considering the crude approximations made.
t0 = ␻−1
pi
c
v⬜0
冉 ␲2␻ peL
␣␤c
冊 1/3
, 共3兲
Equation 共5兲 actually provides both the order of magnitude
of Bzm and its behavior with L and v⬜0. We remark in par-
ticular that Bzm depends weakly on L. For a given tempera-
where v⬜0 = 共kBT⬜0 / me兲1/2. Figures 3共a兲 and 3共c兲 show that ture, the maximum field is Bzmax / 共me␻ pe / e兲 ⬇ 0.16v⬜0 / c, that
this simple formula reproduces quite well PIC simulation is Bzmax ⬇ 20v⬜0 / c MG, for ne0 = 1021 cm−2. Note that during
results, for different values of L and v⬜0. the expansion, the magnetic layer always occupies the area
This analysis can be pushed further to obtain an estimate between the rarefaction front and the vacuum. After the satu-
of the peak magnetic field. We first compute the maximal ration of the instability, the amplitude of Bz thus tends to
current jmy that can be driven by the instability. To this end, decrease.
the electron distribution function f e ⬀ exp共−mev2储 / 2kBT储 Up to now we have not discussed the growth rate of Bz.

016408-2
SELF-GENERATION OF MEGAGAUSS MAGNETIC FIELDS …
30
(a)
30
yωpe/c
yωpe/c
0 0
-30 -30
-5 0 5 -5
xωpe/c
1
-2/3
t (c)
8
vf / cs0
Te / Te0
4 x-axis. The existence of such modes is not surprising. Actu-
1D 3D
0.1 0
1 10 100 0 200
ωpit
FIG. 4. 共Color online兲 2D3V simulations. 共a兲 Magnitude of B, at
t = 26␻−1
pi . From blue 共light gray兲 to red 共dark gray兲, 兩B兩 increases up
to 0.018me␻ pe / e. 共b兲 Angle ␪ between B and ey in the 共ey , ez兲 plane.
It goes from 80 degrees 共light blue兲 up to 160 degrees 共dark red兲.
共c兲, 共d兲 Electron temperature and ion front velocity as a function of
time. The straight line is the result of an electromagnetic collision-
less simulation, and the dashed line of an electrostatic collisional
one. Collisions are inhibited in the gray area.
From Eq. 共3兲 and the definition of A, we find that A共t0兲
⬀ L−2/3, which means that the instability is triggered by
smaller anisotropy for large L. According to Fig. 1共a兲, the
field should therefore grow more slowly when L increases.
This effect is clearly seen in PIC simulations. We also
checked that the growth rates provided by these simulations
are consistent with theoretical values. For L = 50c / ␻ pe for
instance, we have at t = 40␻−1 pi , LA = 12c / ␻ pe and we estimate
from Fig. 2 that A ⬇ 0.6. Figure 1 indicates that for A = 0.6
and k = 2␲ / LA = 0.52␻ pe / c, ⌫ = 0.33␻ pe, which is close to the
value directly measured in PIC simulation ⌫ PIC = 0.3␻ pe.
Before concluding this 1D analysis, we note that Eqs. 共3兲
and 共5兲 are valid only if t0 is smaller than the disassembly
time ␶ = L / 2cs. For very small L the rarefaction wave can
reach the center of the plasma before the instability grows. In
this case Bz develops in the whole plasma, and the model has
to be slightly modified. The length under which this happens
is computed from Eq. 共3兲, and is given by L0 = ␲ / 2共c / ␻ pe兲,
that is L0 ⬇ 260 nm for ne0 = 1021 cm−2.
We now turn to two-dimensional in space and three-
dimensional velocity 共2D3V兲 simulations. The plasma is a
semi-infinite slab with L = 20c / ␻ pe and 500 particles by cell,
surrounded on both sides by 600c / ␻ pe of vacuum 共⌬t
= ⌬x / c = 0.2␻−1
pe 兲. Periodic conditions for both the fields and
the particles are used in the transverse direction 共y axis兲,
which is Ly = 60c / ␻ pe width. Figure 4共a兲 displays the magni-
tude of the magnetic field in the 共x , y兲 plane, when it is
maximum. It shows that B develops on the two sides of the
plasma, just as in 1D2V simulations, and reaches the same
peak value Bm ⬇ 0.018me␻ pe / e as in 1D. The growth of B
starts at the same time t0 ⬇ 17␻−1 pi in the two cases. This
indicate that Eqs. 共3兲–共5兲 are still valid in 2D, which is con-
firmed by additional simulations 关see Figs. 3共a兲 and 3共b兲兴.
There is however a significant difference between 1D and
PHYSICAL REVIEW E 82, 016408 共2010兲
(b) two-dimensional 共2D兲 simulations, as shown by Fig. 4共b兲: in
2D the orientation of B depends on y, which means that
some modes grow with nonpurely longitudinal wave vectors.
For instance in the case of Fig. 4, the magnetic field is made
of three main modes and can be approximated by
B ⬇ 关B1ez + B2 cos ky共y − y 0兲e␾兴cos kx共x − x0兲, 共6兲
0 5
xωpe/c with B1 ⬇ 0.013me␻ pe / e, B2 ⬇ 0.004me␻ pe / e, e␾ = cos ␾ ey
+ sin ␾ ez, ␾ ⬇ 52°, kx ⬇ ␻ pe / c, and ky = 0.1␻ pe / c. This shows
(d) that the plasma supports two non-longitudinal modes which
have wave vectors making an angle of ␪ ⬇ ⫾ 5.7° with the
ally, since for small angles km x 共␪兲 ⬇ kx 共0兲 关see Fig. 1共b兲兴,
m
EM
Collisional these modes are triggered at the same time t0 than longitudi-
nal modes. In contrast, large angle modes are forbidden be-
ωpit cause they require larger anisotropy to grow. As A begins to
decrease soon after t0, such modes can indeed never be trig-
gered. To check this analysis, we performed simulations with
different transverse width Ly, and we observed that for Ly
⬍ 36c / ␻ pe, i.e., ␪min ⲏ 10°, only longitudinal modes develop.
As in realistic cases, the transverse width of the plasma is
much larger than 36c / ␻ pe, modes with ␪ ⬎ 0 can generally
grow during the expansion.
In general, the orientation of the magnetic field therefore
varies in the transverse directions. Unexpectedly, this effect
simplifies the physics of the expansion, avoiding the trapping
of particles, which occurs in 1D2V and some 2D geometries
关14–17兴. As a result, though the magnetic field drives com-
plex electron trajectories, its only macroscopic effect is to
isotropize the electron temperature. This isotropization is ef-
fective as soon as the cyclotron frequency ␻m c = eB / me is
m
greater than the inverse of the expansion characteristic time
2cs / L, that is L ⲏ 0.3c / ␻ pe. When this condition is fulfilled,
the expansion is divided in two stages: while t ⱕ t0 electrons
transfer only their longitudinal energy to ions, then at the
time ti which is of the order of t0, they start to deliver also
their transverse energy, and the expansion becomes isotropic
关T⬜ = T储 = Te共t兲 , ∀ t ⱖ ti兴.
To check that this is the only macroscopic effect of B, we
compare in Figs. 4共c兲 and 4共d兲, the evolution in time of the
electron temperature and of the ion front velocity, obtained
when Ly = 60c / ␻ pe, with the ones provided by a 2D3V elec-
trostatic collisional code. In this code, the thermal collision
frequency was forced to ␯ = 0.3共v / v⬜0兲−3␻ pe, which is, ac-
cording to Ref. 关10兴, sufficient to ensure an isotropic expan-
sion. Further, to mimic the electromagnetic simulation, we
inhibited the collisions for t ⬍ ti = 17.4␻−1 pi . In Figs. 4共c兲 and
4共d兲, the agreement between the two codes is almost perfect,
indicating that the magnetic field has the same influence on
the expansion as do elastic Coulomb collisions. We note in
particular in Fig. 4共c兲 that Te decreases as t−2/3, instead of t−2
in the purely 1D case. This is because ions take their energy
from both the longitudinal and the transverse directions 关10兴.
This important result shows that if the plasma is sufficiently
thick 共L ⲏ 0.3c / ␻ pe兲, ions will always end up taking all the
electron energy, just as if the plasma was collisional.
We now briefly discuss the case of ion acceleration on the
rear side of thin targets interacting with intense laser pulses
关11兴. The electron distribution in this case generally differs
016408-3
THAURY et al.
Electronic density (norm.)
1 1
nc
B simulations is still quite good. In the case of Fig. 5 for in-
0.1
nh
0.01
1E-3
-20 -10 0 10
xωpeh/c
FIG. 5. Growth of the Weibel instability in a bi-Maxwellian
plasma. The electronic densities nh and nc are plotted together with
B, at t = 35␻−1
pih. In this 1D simulation, we took Th0 / Tc0 = 100 with
Th0 = 5 keV, f h = 0.1, and L = 20c / ␻ peh.
from a Maxwellian, because a large number of electrons are
accelerated by the laser. To take into account this energetic
tail, the plasma is generally modeled by a bi-Maxwellian
distribution with a cold population that describes thermal
electrons, and a hot population associated with accelerated
electrons. To facilitate the comparison with the one-
temperature case, we define the partial electron 共ion兲 plasma
frequency ␻ peh = f 1/2
h ␻ pe 共␻ pih = f h ␻ pi兲 and the partial ion
1/2
acoustic velocity csh = 共ZkBTh0 / mi兲1/2. In the following, we
consider that the fraction of hot electrons is small f h
= nh / 共nh + nc兲 Ⰶ 1, and that the cold temperature is steady
共this is supported by Fig. 8 in Ref. 关9兴兲. Under this last con-
dition, we easily derive a new dispersion relation, from
which we find that unstable modes have wave vectors kx
⬍ km c = 共f hA兲 . This shows that cold electrons tend to stabi-
1/2
lize the plasma. For this reason, the magnetic field in a bi-
Maxwellian plasma develops only in its “hot-dominated”
part, as illustrated by Fig. 5.
If Th / Tc Ⰷ 1, we can neglect the motion of cold electrons,
and assume that the length of the anisotropic layer is LA
= 2csht 共that is ␤ = 2 rather than ␤ = 3兲. Doing so we find that
关1兴 U. Samir, K. H. Wright, Jr., and N. H. Stone, Rev. Geophys.
Space Phys. 21, 1631 共1983兲.
关2兴 K. E. Lonngren, Planet. Space Sci. 38, 1457 共1990兲.
关3兴 R. A. Snavely et al., Phys. Rev. Lett. 85, 2945 共2000兲.
关4兴 E. L. Clark et al., Phys. Rev. Lett. 84, 670 共2000兲.
关5兴 A. Maksimchuk, S. Gu, K. Flippo, D. Umstadter, and V. Y.
Bychenkov, Phys. Rev. Lett. 84, 4108 共2000兲.
关6兴 R. K. Singh and J. Narayan, Phys. Rev. B 41, 8843 共1990兲.
关7兴 A. V. Gurevich, L. V. Pariǐskaya, and L. P. Pitaevskiǐ, Zh.
Eksp. Teor. Fiz. 49, 647 共1965兲.
关8兴 P. Mora, Phys. Rev. Lett. 90, 185002 共2003兲.
关9兴 P. Mora, Phys. Rev. E 72, 056401 共2005兲.
关10兴 C. Thaury, P. Mora, J. C. Adam, and A. Héron, Phys. Plasmas
16, 093104 共2009兲.
关11兴 S. C. Wilks, A. B. Langdon, T. E. Cowan, M. Roth, M. Singh,
S. Hatchett, M. H. Key, D. Pennington, A. MacKinnon, and R.
A. Snavely, Phys. Plasmas 8, 542 共2001兲.
关12兴 E. S. Weibel, Phys. Rev. Lett. 2, 83 共1959兲.
PHYSICAL REVIEW E 82, 016408 共2010兲
the triggering time is slightly increased by the presence of a
cold population. The agreement between the model and PIC
Magnetic field (norm.)
stance, we measured ␻ piht0 = 21 in agreement with Eq. 共3兲.
Note that in a bi-Maxwellian plasma, this equation is valid
even for very small L because the rarefaction wave which is
mainly governed by cold electrons is very slow 关9兴. If Eq. 共3兲
fits well PIC results, Eq. 共5兲 significantly underestimates Bm.
This is probably because B deviates strongly from a sine. Be
0 that as it may, we observed that the maximum magnetic field
20
is enhanced by the presence of cold electrons. For example,
in the case of Fig. 5 we measured Bm ⬇ 0.033me␻ peh / e ⬇ 4
MG for ne0h = 1021 cm−2, instead of Bm ⬇ 2 MG with only
hot electrons.
In conclusion, high-amplitude magnetic fields can de-
velop at the surface of plasma slabs expanding into a
vacuum. This phenomenon has important repercussions on
the plasma expansion. In particular, the self-generated mag-
netic fields can keep the electron distribution isotropic during
the expansion. Coulomb collisions can have a similar effect,
but according to Ref. 关10兴, the collisional isotropization is
effective only when 共ne / 1019 cm2兲共L / ␮m兲 / 共Te / eV兲2 ⲏ 1. As
a result magnetic fields are much more efficient than colli-
sions to transfer the electron transverse energy to ions in a
moderately dense and warm plasma. We showed that the
presence of a cold electron population does not affect signifi-
cantly the development of the Weibel instability. This study
in the outer plasma can, therefore, apply to the specific case
of ion acceleration at the rear side of a thin foil irradiated by
an intense laser pulse. Similar effects should also occur in
dusty or magnetized plasmas. In this last case, the presence
of a static magnetic field B0 should influence both t0 and Bm,
for kmx is a function of B0 关12兴.
We gratefully acknowledge T. M. Antonsen for helpful
discussions and remarks on the manuscript. This work was
partly supported by the AXA Research Fund and by Agence
Nationale de la Recherche under Project No. ANR-06-
BLAN-0392.
关13兴 K. Bendib, A. Bendib, and A. Sid, Laser Part. Beams 16, 473
共1998兲.
关14兴 R. L. Morse and C. W. Nielson, Phys. Fluids 14, 830 共1971兲.
关15兴 H. H. Kaang, C.-M. Ryu, and P. H. Yoon, Phys. Plasmas 16,
082103 共2009兲.
关16兴 A. Stockem, M. E. Dieckmann, and R. Schlickeiser, Plasma
Phys. Controlled Fusion 51, 075014 共2009兲.
关17兴 L. Palodhi, F. Califano, and F. Pegoraro, Plasma Phys. Con-
trolled Fusion 51, 125006 共2009兲.
关18兴 D. V. Romanov, V. Y. Bychenkov, W. Rozmus, C. E. Capjack,
and R. Fedosejevs, Phys. Rev. Lett. 93, 215004 共2004兲.
关19兴 A. Bret, M.-C. Firpo, and C. Deutsch, Phys. Rev. E 70,
046401 共2004兲.
关20兴 T. Grismayer, P. Mora, J. C. Adam, and A. Héron, Phys. Rev.
E 77, 066407 共2008兲.
关21兴 P. Mora and T. Grismayer, Phys. Rev. Lett. 102, 145001
共2009兲.
016408-4