Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Magnetic field effects on gas discharge plasmas

Natalia Sternberg, Valery Godyak, and Daniel Hoffman

Citation: Phys. Plasmas 13, 063511 (2006); doi: 10.1063/1.2214537


View online: http://dx.doi.org/10.1063/1.2214537
View Table of Contents: http://pop.aip.org/resource/1/PHPAEN/v13/i6
Published by the American Institute of Physics.

Related Articles
Role of ion mass in the generation of fluctuations and poloidal flows in a simple toroidal plasma
Phys. Plasmas 19, 072306 (2012)
Influence of magnetic field strength on potential well in the ionization stage of a double stage Hall thruster
Phys. Plasmas 19, 073511 (2012)
High speed vacuum ultraviolet telescope system for edge fluctuation measurement in the large helical device
Rev. Sci. Instrum. 83, 10E513 (2012)
Study of applied magnetic field magnetoplasmadynamic thrusters with particle-in-cell and Monte Carlo collision.
II. Investigation of acceleration mechanisms
Phys. Plasmas 19, 073108 (2012)
On the space-charge formation in a collisional magnetized electronegative plasma
Phys. Plasmas 19, 073507 (2012)

Additional information on Phys. Plasmas


Journal Homepage: http://pop.aip.org/
Journal Information: http://pop.aip.org/about/about_the_journal
Top downloads: http://pop.aip.org/features/most_downloaded
Information for Authors: http://pop.aip.org/authors

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
PHYSICS OF PLASMAS 13, 063511 共2006兲

Magnetic field effects on gas discharge plasmas


Natalia Sternberg
Department of Mathematics and Computer Science, Clark University, Worcester, Massachusetts 01610
Valery Godyak
OSRAM Sylvania, Beverly, Massachusetts 01915
Daniel Hoffman
Applied Material, Sunnyvale, California 94086
共Received 20 February 2006; accepted 23 May 2006; published online 30 June 2006兲
A study of the fluid model for cylindrical weakly ionized quasineutral plasmas in an axial magnetic
field is presented. The model takes into account ionization, ion and electron inertia, as well as
frictional forces for ions and electrons. The behavior of the plasma parameters for arbitrary
magnitudes of the magnetic field, arbitrary gas pressure, and plasma size is presented, making the
model applicable for a wide range of discharge conditions. A magnetic field parameter is introduced,
which specifies a parameter range for the magnetic field, gas pressure, and plasma size where the
Boltzmann equilibrium with the ambipolar field for the electron distribution is satisfied. In addition,
a parametric relation for the magnetic field, gas pressure, and plasma size is obtained, which
separates the region of weak magnetic field effects from the region of strong magnetic field effects.
For strongly magnetized plasmas, an asymptotic solution with nonzero plasma density at the plasma
boundary is presented. Analytical approximations for the ionization frequency and the plasma
density at the plasma boundary are found for arbitrary external discharge parameters. The theoretical
results are supported by numerical computations. © 2006 American Institute of Physics.
关DOI: 10.1063/1.2214537兴

I. INTRODUCTION magnetic field may cause low frequency instabilities that af-
fect the plasma transport and the plasma global confinement.
The behavior of plasmas in a magnetic field is among the There is, however, much evidence of low frequency instabili-
oldest problems in plasma physics. It has been central in ties and turbulence caused by a magnetic field in low tem-
plasma fusion research since the early experiments on perature gas discharge plasmas.
plasma confinement by a magnetic field in the 1950s, and it There is a large body of literature about the effects of a
has remained of great interest in plasma fusion studies that magnetic field on hot collisionless plasmas and on collision
use contemporary sophisticated devices.1 It is also an impor- dominated gas discharge plasmas. Different approaches are
tant problem for many plasma discharges used in processing used to study those two very different plasmas. The guided
semiconductor materials where the application of a magnetic center approximation is usually considered for fusion
field results in enhancement of some desirable features of plasmas,1,2 whereas the diffusion approximation is usually
specific plasma sources.2 considered for gas discharge plasmas.2,3 The guided center
Various effects are induced by a magnetic field on gas approximation assumes no collisions, no ionization, and a
discharge plasmas. The magnetic field affects the plasma strong magnetic field with a weak nonuniformity. The diffu-
transport to the chamber wall, changing the ionization bal- sion approximation neglects ion inertia and ionization in the
ance and the plasma spatial distribution. In fact, due to re- plasma momentum transfer, and is applicable only for suffi-
duction in the plasma loss to the wall, the application of a ciently high gas pressures, when the condition for constant
magnetic field to gas discharge plasma is accompanied by a ion mobility 共linear diffusion兲 is satisfied.4,5 Those approxi-
fall in both, the ionization frequency and the electron tem- mations, however, are not suitable for bounded gas discharge
perature. In this case, the magnetic field causes the same plasmas in a weak or moderate magnetic field with gas pres-
effects that can be observed by increasing the gas pressure. A sures ranging from low to intermediate, where neither ion-
multipole magnetic field 共magnetic bucket兲 on the plasma ization, nor ion inertia, collisions, or the magnetic field can
periphery enhances the plasma confinement, causing a rise in be neglected. For such low pressure discharges, typical in
the plasma density and an essential improvement in the laboratory studies and commercial plasma processing reac-
plasma uniformity. In rf discharges, the application of a mag- tors, the analysis of bounded plasmas in a magnetic field
netic field changes the plasma electrodynamics. Anisotropic requires the consideration of the full hydrodynamic model
modification of the plasma conductivity as well as the propa- without the assumptions of the limiting approaches men-
gation and absorption of electromagnetic waves are typical tioned earlier.
in helicon and electron cyclotron resonance plasma sources A hydrodynamic model for a semiinfinite plasma in an
when the electron cyclotron frequency is near or exceeds the oblique magnetic field was introduced in Ref. 6, where ion-
discharge driving frequency. In collisionless fusion plasma, a ization and collisions were not taken into account. Assuming

1070-664X/2006/13共6兲/063511/14/$23.00 13, 063511-1 © 2006 American Institute of Physics

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-2 Sternberg, Godyak, and Hoffman Phys. Plasmas 13, 063511 共2006兲

that the ion motion in the plasma is aligned with the mag- and its reversal occur whereas the plasma density decays
netic field, this model yields a semiinfinite presheath 共Cho- monotonously toward the boundary. This is obviously in-
dura layer兲 between the uniform plasma at infinity and the compatible with the Boltzmann equilibrium for the electron
space-charge sheath at the wall. According to Ref. 6, the ions distribution.
enter the presheath when their velocity component in the In the overwhelming majority of work concerning mag-
direction of the magnetic field reaches the ion sound speed, netic field effects on gas discharge plasmas, the models used
and they exit the presheath when their velocity component assume the constant ion mobility regime.1–3,19–22 This as-
normal to the wall reaches the ion sound speed. In Ref. 7, the sumption, however, is valid only for hot ions, or for rela-
Chodura layer was eliminated by accounting for ion colli- tively high gas pressures in the case of cold ions.23 For cold
sions with gas atoms, whereas in Ref. 8, it was eliminated by ions and low to moderate gas pressures, the ion-neutral col-
accounting for ionization. The approaches from Refs. 7 and 8 lisions are dominated mainly by charge exchange, and the
were combined in Ref. 9 where both, ionization and colli- linear diffusion approximation cannot be assumed.4,23
sions were taken into account, and no Chodura layer has In Ref. 24, a two-dimensional model for a bounded cy-
been found. It is therefore not surprising that the Chodura lindrical plasma with an axial magnetic field was studied
theory could not be corroborated experimentally where ion- without the assumption of the Boltzmann equilibrium. In that
ization and ion collisions, now matter how small, are always work, the competition between the axial plasma transport
present.10 along the magnetic field and the radial plasma transport
In many works on the plasma sheath transition, includ- across the magnetic field was analyzed. The validity of the
ing those with a magnetic field,6,7,11–16 the models assume a analysis in Ref. 24 was illustrated by numerical computa-
plasma bulk at infinity and a transitional semiinfinite tions performed for specific conditions of a particular experi-
presheath. In such models, ionization and bounded plasma ment with a very strong magnetic field so that the axial
effects that are essential for real plasmas cannot be consis- plasma transport became dominant.
tently included in the analysis. Indeed, at any finite ioniza- The present article is concerned with the study of the
tion frequency, such models generate an infinite ion flux to general model for an infinite cylindrical plasma immersed
the wall. Attempts found in the literature to balance the par- into an axial magnetic field. The model takes into account
ticle flux to the wall by artificially introducing particle injec- ionization, ion and electron inertia as well as ion and electron
tion from undisturbed plasma, or by introducing ionization collisions with gas atoms, and does not assume the Boltz-
that is balanced in the bulk plasma by recombination, make mann equilibrium for the electrons. Our main goal is to bet-
those models irrelevant to low pressure gas discharge plas- ter understand the onset of strong magnetic field effects, and
mas. A more detailed explanation of this issue can be found the dependence of the plasma characteristics on the external
in Refs. 17 and 18. discharge parameters such as gas pressure, plasma size and
All the models in Refs. 6–9 consider a planar plasma in the magnetic field.
an oblique magnetic field and assume the Boltzmann equi- The article is organized as follows. In Sec. II, we formu-
librium with the ambipolar field for the electron distribution. late and discuss the mathematical model of the cylindrical
On the other hand, it was pointed out in Ref. 12 and reiter- plasma in an axial magnetic field. In Sec. III, the model is
ated in the literature13–16 that the Boltzmann equilibrium solved numerically, and the dependence of the plasma char-
does not hold in a strong magnetic field when the electron acteristics on the magnetic field and gas pressure is dis-
Lorentz force becomes comparable with the electron gradient cussed. In Sec. IV, we derive a magnetic parameter that
force. In order to understand when the Boltzmann equilib- specifies a parameter range for the magnetic field, gas pres-
rium can be assumed, one has to include the electron mo- sure and plasma size where the Boltzmann equilibrium is
mentum equation into consideration of the plasma model. valid. In Sec. V, we present an asymptotic solution for
This has been done in Refs. 12–16, but without accounting strongly magnetized plasmas with nonzero plasma density at
for ionization. To avoid an infinite presheath triggered by the the plasma boundary. Further, we introduce a parametric re-
neglect of ionization, it has been assumed that the ions are lation between the magnetic field, gas pressure, and plasma
injected into the presheath with some initial velocity which is size that yields a criterion for the separation of the weakly
quite arbitrary. This assumption made the presheath size fi- magnetized plasma region from the strongly magnetized
nite, but also arbitrary, and lead to results of little relevance plasma region. We find analytical expressions for both, the
to a real gas discharge plasma of a given finite size. plasma density at the plasma boundary and the frequency of
An analysis of magnetic field effects on cylindrical gas ionization, for arbitrary magnitudes of the magnetic field, gas
discharge plasmas at low gas pressure was given in Refs. 19 pressure and plasma size. We conclude the article with a
and 20 without assuming the Boltzmann equilibrium and ac- summary of the results in Sec. VI.
counting naturally for the plasma ionization balance by in-
cluding ionization and the plasma size into consideration. In
that work, a significant reduction of the ambipolar electric II. BASIC EQUATIONS
potential and even its reversal has been observed. Both phe-
The plasma-wall problem can be described by the fol-
nomena were repeatedly rediscovered by others in recent
lowing hydrodynamic equations for ions and electrons.
years. The potential reversal is caused by a strong magnetic
Continuity equations for the ion and electron flux:
field when the ion diffusion exceeds the electron diffusion
across the magnetic field. The ambipolar potential reduction div共nivi兲 = Zne , 共1兲

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-3 Magnetic field effects on gas discharge plasmas Phys. Plasmas 13, 063511 共2006兲

div共neve兲 = Zne . 共2兲 共i兲 the plasma cylinder is of radius R with infinite length;
thus, there is no variation with respect to z;
Momentum equations for ions and electrons: 共ii兲 the magnetic field is directed along the plasma axis:
div共Mnivi 丢 vi + kTiniI兲 + eni共ⵜ␸ − vi ⫻ B兲 + niFi = 0, B = 共0 , 0 , B兲;
共iii兲 the plasma is axially symmetric; thus there is no
共3兲 variation with respect to ␪;
共iv兲 the plasma is quasineutral: ni = ne = n, vir = ver = vr,
div共mneve 丢 ve + kTeneI兲 − ene共ⵜ␸ − ve ⫻ B兲 + neFe = 0. where vir is the radial ion velocity and ver is the radial
共4兲 electron velocity.
With those assumptions, system 共1兲–共5兲 yields the fol-
The Poisson equation: lowing cylindrical plasma model.
Continuity equation:
⑀0ⵜ2␸ = − e共ni − ne兲. 共5兲
In the previous equations, k is the Boltzmann constant, ␸ is d
共rnvr兲 = rZn. 共10兲
the potential, m is the electron mass, M the ion mass, ne is dr
the electron density, ni the ion density, Te is the electron
temperature, Ti is the ion temperature, ve is the electron Momentum equations for ions:
transport velocity, and vi the ion transport velocity. Further, Z dvr dn d␸
is the frequency of ionization, ⑀0 is the permittivity of free Mnvr + kTi + en + MZnvr − envi␪B + nFir = 0,
space, e is the elementary charge, B is the magnetic field, I is dr dr dr
the identity matrix, Fi is the ion frictional force, and Fe the 共11兲
electron frictional force.
Equations 共1兲–共5兲 describe a weakly ionized plasma un- d v i␪
der the action of a magnetic field. We assume direct ioniza- Mnvr + MZnvi␪ + envrB + nFi␪ = 0. 共12兲
dr
tion 共i.e., Z is constant兲, no heat transfer 共i.e., constant elec-
tron temperatures兲, and cold ions 共i.e., Ti ⬇ 0 is negligible兲. Momentum equations for electrons:
The electron frictional force is given by
dvr dn d␸
vTe mnvr + kTe − en + mZnvr + enve␪B + nFer = 0,
F e = m ␯ ev e ⬇ m ve , 共6兲 dr dr dr
␭e
共13兲
where ␯e is the electron transport collision frequency with
gas atoms, ␭e is the electron mean free path, and
d v e␪

冉 冊 1/2 mnvr + mZnve␪ − envrB + nFe␪ = 0, 共14兲


kTe dr
vTe = 共7兲
m
where vi = 共vr , vi␪兲, ve = 共vr , ve␪兲, Fi = 共Fir , Fi␪兲, and Fe
is the electron thermal velocity. = 共Fer , Fe␪兲. The symmetry of the problem is reflected in the
In the present article, we consider low to moderate pres- following boundary conditions at the plasma center r = 0:
sure discharges,4 i.e.,
n共0兲 = n0, vi共0兲 = ve共0兲 = 0, ␸共0兲 = 0. 共15兲
␭i Tg
艌 , 共8兲 Further, according to 共6兲 and 共9兲,
R Te
␲ ␲
M vr冑vr2 + vi2␪ , M vi␪冑vr2 + vi2␪ ,
where Tg ⬇ Ti is the gas temperature, R is the characteristic
Fir = F i␪ =
plasma size, and ␭i is the ion mean free path. In this case, the 2␭i 2␭i
ion-atom charge-exchange collisions are dominant, and it is 共16兲
therefore sufficient to use the ion frictional force in the ion- Fer = m␯evr, F e␪ = m ␯ ev e␪ .
charge-exchange regime4,5,23:
Using the facts that m Ⰶ M, Ti Ⰶ Te, and m␯e Ⰶ MZ 共the
␲ last inequality holds as the electron mobility is much larger
Fi = M兩vi兩vi , 共9兲
2␭i than the ion mobility25兲, and denoting the ion and electron
cyclotron frequencies by ␻ci = eB / M and ␻ce = eB / m, we can
where 兩vi兩 is the magnitude of vi. rewrite the previous plasma equations in explicit form as
The plasma-wall problem describes the bulk quasineutral follows:


plasma that is transiting into the thin sheath near the wall. In
this article, we are studying only the quasineutral region of a dv kTe/M + vr2 vr
cylindrical plasma. =Z + ␻ci共ve␪ − vi␪兲
dr kTe/M − vr2 kTe/M − vr2


To model a cylindrical plasma under the action of a mag-
netic field, we introduce cylindrical coordinates 共r , ␪ , z兲 and −
kTe 1 Fir
· + , 共17兲
make the following assumptions: M r M

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-4 Sternberg, Godyak, and Hoffman Phys. Plasmas 13, 063511 共2006兲

dn
dr
=−
n
2
kTe/M − vr
2Z v r −
vr2
r

+ ␻ci共ve␪ − vi␪兲 +
Fir
M
, 册 dui␪
d␰
u i␪
ur
f i␪
= − S − bi − .
ur
共27兲

共18兲 In the previous equations, the ion magnetic parameter is


given by
d␸
dr
=−
kTe/e
kTe/M − vr2
2Z v r −
vr2
r
+ ␻ 冋
ci
vr2
kTe/M
v e␪ − v i␪ 冉 冊 bi =
R
␳i
, 共28兲

+
Fir
M

, 共19兲 where ␳i = vs / ␻ci is the ion cyclotron radius at the ion sound
speed, and
R
d v e␪ v e␪ S= Z 共29兲
= ␻ce − 共Z + ␯e兲 , 共20兲 vs
dr vr
is the normalized ionization frequency, which is the eigen-
d v i␪ v i␪ F i␪ value of the problem. Further,
= − Z − ␻ci − . 共21兲
dr vr M vr ␲
f ir = ␣iur冑ur2 + ui2␪ ,
2
Observe that the radial component of the electron frictional
force, Fer, does not appear in the plasma equations 共17兲–共19兲,

and can therefore be neglected in the momentum equation f i␪ = ␣iui␪冑ur2 + ui2␪ , 共30兲
共13兲. However, the electron frictional force in the azimuthal 2
direction cannot be neglected even at very low gas pressures.
Note that Eqs. 共17兲–共19兲 have a singularity when the f e␪ = ␣ eu e␪ ,
radial velocity component vr reaches the ion sound speed
where ␣i and ␣e are the ion and electron collision parameters

冉 冊
kTe 1/2 given by

冉冊
vs = .
M R R M 1/2
␣i = , ␣e = . 共31兲
We will consider this singular point r = R to be the plasma ␭i ␭e m
boundary,8,9 as in the case without a magnetic field.26 In dimensionless coordinates, initial condition 共15兲 at the
In order to simplify the numerical computations, we in- center ␰ = 0 becomes
troduce the following dimensionless variables:
y共0兲 = 1, ui共0兲 = ue共0兲 = 0, ␩共0兲 = 0. 共32兲
r e n
␰= , ␩共␰兲 = − ␸共r兲, y共␰兲 = , The plasma boundary is located at ␰ = 1 where the radial
R kTe n0 velocity component reaches the ion sound speed, i.e., ur = 1.
As we have mentioned earlier, at the boundary, the plasma
1 1 equations have a singularity. In numerical computations, we
u i共 ␰ 兲 = vi共r兲, u e共 ␰ 兲 = ve共r兲, uir = uer = ur , 共22兲 can avoid this singularity by considering ur as the indepen-
vs vs
dent variable. The plasma equations, however, have a second
singularity at the plasma center ␰ = 0, where ur = 0, and that
R R singularity cannot be eliminated. In order to solve the plasma
fi = F i, fe = Fe .
M vs2 mvs2 equations, we therefore need to start with the following ini-
tial conditions near the plasma center 共␰ ⬇ 0兲, which can be
Then, Eqs. 共17兲–共21兲 become: obtained by differentiating 共23兲–共27兲 and then using Taylor
dur
d␰
=S
1 + ur2
2 +
ur
1 − ur 1 − ur2 bi共ue␪ − ui␪兲 −
1


+ f ir , 册 共23兲
expansions:
S
u r共 ␰ 兲 = ␰ , u i␪共 ␰ 兲 = −
bi
␰, u e␪共 ␰ 兲 =
biM/m
␰,
2 3 3 + 2␣e/S
d␩
=
1
d␰ 1 − ur2
2Su r −


ur2
+ bi共ur2ue␪ − ui␪兲 + f ir , 册 共24兲
y共␰兲 = 1, ␩共␰兲 = 冉
␰2 3 2 b2i
S + . 冊
共33兲

2 4 3

dy
d␰
=−
y
1 − ur2
2Su r −


ur2
+ bi共ue␪ − ui␪兲 + f ir , 册 共25兲
The normalized plasma equations 共23兲–共27兲 with the
boundary condition 共33兲 can now be solved numerically by
choosing ur 苸 关0 , 1兴 as the independent variable and ␰
苸 关0 , 1兴 as the dependent variable such that ␰ = 0 for ur = 0,
due␪
=−S
u e␪ M f e␪
+ bi − , 共26兲 and ␰ = 1 for ur = 1. The normalized ionization frequency S is
d␰ ur m ur then found by an iteration process.

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-5 Magnetic field effects on gas discharge plasmas Phys. Plasmas 13, 063511 共2006兲

FIG. 1. Normalized plasma density distribution y for different values of the magnetic parameter be and the collision parameter ␣i; 共a兲 ␣i = 1, 共b兲 ␣i = 10, and
共c兲 ␣i = 100.

III. NUMERICAL RESULTS cross section in a Ramsauer gas is falling with gas pressure,
and therefore, ␭e is not proportional to ␭i. Based on experi-
There are four external independent parameters in the
mental data obtained in argon inductively coupled plasma for
formulation of problems 共23兲–共27兲: the mass ratio M / m, the
a wide range of pressures 27, we use the approximation ␣e
ion magnetic parameter bi = R / ␳i, the ion collision parameter
␣i = R / ␭i, and the electron collision parameter ␣e = 50␣1/2i if ␣i ⫽ 0, and ␣e = 30 if ␣i = 0. Note that although the
= 共M / m兲1/2R / ␭e. As electrons are much more sensitive to the collisionless limit for ions 共␣i = 0兲 is quite adequate for low
magnetic field than ions, it is more convenient to use the pressures 共␣i ⬍ 1兲, one cannot assume collisionless electrons,
electron magnetic parameter as the electron frictional force in the azimuthal direction may

冉冊 1/2
not be negligible for ␣i ⬍ 1.
R M In this article, the computations were performed for ␣i
be = = bi 共34兲
␳e m = 0 ; 1 ; 10; 100 and be between 0 and 300, which corresponds
as a parameter of the problem rather than bi. Here, ␳e to a wide range of gas pressure and magnetic field. For an
= vTe / ␻ce is the electron cyclotron radius. We consider the argon plasma of radius R = 3 cm, this range of dimensionless
benchmark argon gas 共M / m = 7.344⫻ 104兲 that is character- parameters corresponds to gas pressure between approxi-
ized by a well pronounced Ramsauer effect. For the sake of mately 1 and 100 mTorr and to a magnetic field between 0
simplicity, we reduce the number of independent external and approximately 1000 G. The upper limit of argon pres-
parameters to two: ␣i that represents the gas pressure and be sure marginally satisfies the condition for the ion-atom
that represents the magnetic field. charge exchange process 共␣i 艋 Te / Tg ⬇ 100兲. For higher pres-
Usually, ␭i Ⰶ ␭e. In general, ␭i is inversely proportional sures 共␣i ⬎ 100兲, the regime of constant ion mobility should
to gas pressure p 共i.e., ␭i ⬀ p−1兲. Due to the reduction in the be considered.4,5
electron temperature with gas pressure, the electron-atom The normalized radial plasma density distribution y

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-6 Sternberg, Godyak, and Hoffman Phys. Plasmas 13, 063511 共2006兲

FIG. 2. Dependence of the normalized ionization frequency S on the mag- FIG. 3. Dependence of the normalized plasma density at the plasma bound-
netic parameter be for different collision parameters ␣i. ary y 1 on the magnetic parameter be for different collision parameters ␣i.

n共R兲 ZR S
= 具n典 or y 1 = 具y典, 共35兲
= n / n0 for different values of the magnetic and collision pa- n0 2 v sn 0 2
rameters is shown in Fig. 1. For ␣i = 1, the plasma density
distribution does not change much when be 艋 3 and is close where 具 典 denotes the average value over the cylinder cross
to the plasma density distribution in the collisionless limit section. The average value 具y典 is practically independent of
without a magnetic field25 共i.e., ␣i = 0 and be = 0兲. As the mag- gas pressure and falls from about 0.73 in the case of no
netic field is increased 共be ⬎ 10兲, the plasma density distribu- magnetic field to about 0.43 in the case of a strong magnetic
field. Thus, y 1 / S ⬇ const. over the whole parameter range.
tion experiences great changes, and for strong magnetic
As suggested by Figs. 2 and 3, the dependence of y 1 and
fields 共be = 100兲, it approaches the Bessel distribution
S on gas pressure is different for high and low magnetic
J0共2.405r / R兲, which is typical for plasmas without a mag-
fields. Without a magnetic field, both, y 1 = y 01 and S = S0 fall
netic field in the regime of linear diffusion 共constant ion
with increasing gas pressures and can be approximated as
mobility兲.25 We will discuss in more detail the behavior of follows4:
the plasma density for strong magnetic fields in Sec. V.
As expected, an increase in gas pressure 共␣i 艌 10兲 dimin- y 01 = 0.8共4 + ␣i兲−1/2 ⬃ p−1/2, S0 = 2.2共4 + ␣i兲−1/2 . 共36兲
ishes the effect of the magnetic field on the plasma density In the following sections, we will show that for a given gas
distribution. In fact, for ␣i = 10, an appreciable effect of the pressure, y 1 and S are little affected by sufficiently low mag-
magnetic field on the plasma density distribution starts at netic fields 共i.e., weakly magnetized plasma兲 and can be ap-
be ⬎ 10, and for ␣i = 100 at be ⬎ 30. At ␣i = 100, the plasma proximated by the relationship 共36兲, whereas for high mag-
density distribution does not change too much even for be netic fields 共i.e., strongly magnetized plasmas兲, y 1 = y B1 and
= 100, and has the shape typical for the plasma density dis- S = SB can be approximated as follows:
tribution in the regime of nonlinear diffusion 共charge ex-
␣e ␣e
change regime兲.4,5,23 y B1 = 1.25 ⬃ p1/2B−2, SB = 共2.405兲2 . 共37兲
Figure 2 shows the normalized ionization frequency 共ei- b2e b2e
genvalue of the problem兲, S = RZ / vs, and Fig. 3 shows the The plasma potential distribution at different collision and
value of the normalized plasma density at the plasma bound- magnetic parameters, ␣i and be, is illustrated in Fig. 4. For
ary, y 1 = n共R兲 / n0, for different gas pressures and magnetic small ␣i and large be, the potential distribution is signifi-
fields. Observe that as the pressure parameter ␣i increases, cantly affected by the magnetic field. In this case, the mag-
the variations with respect to the magnetic field in the ion- netic field reduces the ambipolar field, pushing it out toward
ization frequency and in the boundary value of the plasma the plasma boundary. For sufficiently large magnetic fields,
density become smaller, and for high pressures 共␣i = 100兲 are this effect may cause electric field and potential reversal 19,20
practically negligible. The corresponding graphs for S and y 1 if the ion diffusion exceeds the electron diffusion across the
are very similar. In general, for a given gas pressure and a magnetic field, which cannot happen in our model with cold
given magnetic field, the relationship between S and y 1 can ions. At high gas pressures 共␣i = 100兲, the magnetic field has
be obtained by integrating continuity equation 共10兲 共i.e., bal- little effect on the plasma potential distribution.
ancing the plasma generation in the volume with its loss to The reduction of the ambipolar field caused by the mag-
the wall兲: netic field results in a reduction of the radial plasma velocity

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-7 Magnetic field effects on gas discharge plasmas Phys. Plasmas 13, 063511 共2006兲

FIG. 5. Normalized radial velocity ur for different values of the magnetic


FIG. 4. Normalized plasma potential ␩ for different values of the magnetic parameter be and the collision parameter ␣i; 共a兲 ␣i = 1 and 共b兲 ␣i = 100.
parameter be and the collision parameter ␣i; 共a兲 ␣ = 1 and 共b兲 ␣i = 100.

␻ce ␭e
v e␪ ⬇ vr = vr . 共39兲
␯e ␳e
for low gas pressures 共␣i = 1兲 as can be seen in Fig. 5. For
large gas pressures 共␣i = 100兲, this effect is negligible. Our numerical computations have shown that Eq. 共39兲 is
The normalized electron azimuthal velocity, ue␪ = ve␪ / vs, quite accurate in most of the plasma, but not at the boundary.
is shown in Fig. 6 for low 共␣i = 1兲 and high 共␣i = 100兲 gas For example, for ␣i = 1 and be = 100, Eq. 共39兲 yields a relative
pressures. Observe that the electron azimuthal velocity is in- error in the electron azimuthal velocity in most of the plasma
creasing toward the boundary 共i.e., dve␪ / dr ⬎ 0兲. Therefore, of less than 1%, but grows sharply near the boundary reach-
from Eq. 共20兲, ing about 140% at the boundary. For ␣i = 100 and be = 100,
the relative error in most of the plasma is less than 2% and
about 45% at the boundary. Nevertheless, our computations
␻ce d v e␪ ␻ce ␭e have shown that using 共39兲 instead of Eq. 共20兲, i.e., ignoring
v e␪ = vr − vr ⬍ vr = vr . 共38兲 electron inertia, has little effect on the numerical solution of
Z + ␯e dr ␯e ␳e
the problem. This is due to the small size 共compared with the
plasma size兲 of the region near the plasma boundary where
It is often assumed in the literature 共see e.g., Refs. 20, 12, the representation 共39兲 is not accurate.
and 24兲 that the electron inertia in the azimuthal direction is The azimuthal electron velocity grows toward the
negligible, which means that the azimuthal electron velocity plasma boundary together with vr. For low gas pressure and
is determined by the equilibrium between the Lorentz and high magnetic field, the azimuthal electron velocity ap-
frictional forces: proaches the electron thermal velocity vTe at the plasma

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-8 Sternberg, Godyak, and Hoffman Phys. Plasmas 13, 063511 共2006兲

FIG. 6. Normalized azimuthal electron velocity ue␪ for different values of FIG. 7. Normalized azimuthal ion velocity ui␪ for different values of the
the magnetic parameter be and the collision parameter ␣i; 共a兲 ␣i = 1 and 共b兲 magnetic parameter be and the collision parameter 共a兲 ␣i = 1 and 共b兲 ␣i
␣i = 100. = 100.

show that the behavior of the plasma characteristics depends


boundary. With increasing gas pressure, the electron azi- on the magnetic field, gas pressure and the plasma size. In
muthal velocity is reduced by the electron frictional force the following sections, we will derive a parametric relation
caused by electron-atom collisions. which takes this into account and yields a criterion for sepa-
The normalized ion azimuthal velocity, ui␪ = vi␪ / vs, is ration of the weakly magnetized plasma from the strongly
shown in Fig. 7. It is quite small: 兩ui␪兩 ⬍ 0.1 for ␣i = 1, and magnetized plasma.
兩ui␪兩 ⬍ 3 ⫻ 10−3 for ␣i = 100. Thus, it is negligible compared
to ue␪ and ur. Note that away from the center, ui␪ is practi- IV. BOLTZMANN EQUILIBRIUM AND WEAKLY
cally constant in the collision dominated regime 共␣i = 100兲. MAGNETIZED PLASMA
At first glance, this disappearance of the radial dependence
of ui␪ looks strange, as the ion azimuthal velocity is caused The Boltzmann distribution for electrons
by the Lorentz force that is proportional to ur, and ur grows
sharply near the plasma boundary 共see Fig. 5兲. This can be
explained by the fact that the ion friction force 关see Eqs. 共16兲
ne = n0 exp 冉 冊
e␸
kTe
共40兲

and 共30兲兴 is defined by the modulus of the ion velocity. As has been used in the majority of works on low temperature
ui␪ Ⰶ ur, both, the Lorentz force and the azimuthal compo- plasmas in a magnetic field. In essence, the Boltzmann dis-
nent of the ion friction forces are proportional to ur, which tribution describes the distribution of hot electrons in an
for high pressures results in ui␪ ⬇ const. except near the cen- electron repelling electric field, and is the consequence of the
ter where ur is small and ui␪ ⬃ ␰ 关see 共33兲兴. equilibrium between the plasma gradient force and the elec-
The numerical results we have discussed in this section tric force:

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-9 Magnetic field effects on gas discharge plasmas Phys. Plasmas 13, 063511 共2006兲

kTe ⵜ ne = ene ⵜ ␸ . 共41兲 ably well for be 艋 1 and in this case ␤ 艋 0.02Ⰶ 1. For ␣i
= 100 共i.e., ␣e = 500兲 the Boltzmann equilibrium is satisfied
In the bounded plasma problem 共1兲–共5兲 without a magnetic reasonably well if be 艋 10, and in this case ␤ 艋 0.2Ⰶ 1. Our
field 共B = 0兲, Boltzmann equilibrium 共41兲 is obtained from computations have shown that when the Boltzmann equilib-
electron momentum equation 共13兲 by neglecting electron in- rium holds, the axial magnetic field has practically no effect
ertia, the ionization term and the electron friction force. The on the plasma characteristics. In this case, it is sufficient to
neglect of those terms is justified by the facts that m / M Ⰶ 1, consider the nonmagnetized problem.
Ti Ⰶ Te, and that the electron mobility is much larger than the The dependence of the normalized ionization frequency,
ion mobility.25 With the same justifications, when an axial S, and the boundary plasma density, y 1, on the parameter ␤
magnetic field is present, we obtain by combining equations for different collision parameters ␣i is shown in Fig. 10. One
共18兲 and 共19兲 the following relationship for the plasma equi- can distinctly see two regions of different asymptotic behav-
librium in the radial direction: ior. One region, ␤ Ⰶ 1, is the region of weakly magnetized
dn d␸ plasma. In this parameter range, the magnetic field effects
kTe − en + enve␪B = 0. 共42兲 are negligible, the Boltzmann equilibrium holds, and the ion-
dr dr
ization frequency and boundary plasma density do not de-
A comparison of Eq. 共42兲 with the momentum equation for pend on the magnetic field and can be approximated by the
electrons 共13兲 shows that in the radial direction, we can ne- corresponding expressions in 共36兲. The second region, ␤
glect the electron frictional force Fer, the ionization repre- Ⰷ 1, is the region of strongly magnetized plasma. In this
sented by the term mZvr and the electron inertia mvrdvr / dr. region, the magnetic field is dominant, the Boltzmann equi-
Equation 共42兲 describes the balance between the plasma librium does not hold, the ionization frequency and the
gradient, electric force, and the Lorentz force acting on the boundary plasma density do not depend on gas pressure and
electrons. Obviously, 共42兲 reduces to Boltzmann equilibrium can be approximated by the corresponding expressions in
共41兲 if the Lorentz force can be neglected. The question is 共37兲. Asymptotic approximations 共36兲 and 共37兲 for weakly
under what conditions is the Boltzmann equilibrium obtained and strongly magnetized plasmas are represented in Fig. 10
when an axial magnetic field is present. According to Refs. by straight lines.
12 and 13, in fully ionized plasmas the Boltzmann equilib- Inequality 共46兲 gives a rough idea in what parameter
rium holds for an infinite presheath when ␻ce / ␯e = ␭e / ␳e Ⰶ 1. range one can expect the Boltzmann equilibrium to hold. In
This criterion is determined solely by the electrons, and does the next section, we will refine this result by deriving a cri-
not depend on the ions or the plasma size. In order to find the terion that separates the weakly magnetized plasma where
corresponding criterion for bounded active plasmas, we inte- the Boltzmann equilibrium holds from the strongly magne-
grate Eq. 共42兲: tized plasma, where the Boltzmann equilibrium is not satis-

ln 冉冊 n
n0
=
e
kTe
␸−
1
␳ iv s
冕0
r
ve␪ dr 共43兲
fied.

V. STRONGLY MAGNETIZED PLASMA


and obtain
n
n0
= exp
e
kTe
冉 冊 冉 冕 冊
␸ exp −
1
␳i 0
r
v e␪
vs
dr . 共44兲
In the previous section, we have found the parameter
range for ␤ = b2e / ␣e, namely ␤ Ⰶ 1, that yields a weakly mag-
netized plasma where the electric force is dominant whereas
Using estimates 共38兲 and 共39兲 together with the fact that vr the Lorentz force is negligible. Neglecting the Lorentz force
艋 vs, we find in Eq. 共42兲 yields the Boltzmann equilibrium. In this section,
we consider the second limiting case ␤ Ⰷ 1, when the Lor-
1
␳i

0
r
v e␪
vs
dr 艋
␭e
␳ e␳ i
冕0
r
vr
vs
dr 艋 r
␭e
␳ e␳ i
艋R
␭e
␳ e␳ i
. 共45兲
entz force is dominant and the electric force is negligible. In
this case, Eq. 共42兲 yields

Hence, if dn
kTe + enve␪B = 0. 共48兲
dr
R␭e b2e
␤= Ⰶ 1 or equivalently ␤ = Ⰶ 1, 共46兲 Using representation 共39兲 for ve␪, we find from 共48兲
␳ e␳ i ␣e
D⬜ dn
then vr = − , 共49兲

冉 冕 冊 r
n dr
1 v e␪
exp − dr ⬇ 1 共47兲 where D⬜ = ␳2e ␯e is the classical diffusion coefficient across
␳i vs
0
the magnetic field. Substituting 共49兲 into continuity equation
and the Boltzmann equilibrium is obtained. Thus, inequality 共10兲, we find that the plasma density distribution satisfies the
共46兲 describes the relationship between the magnetic field, following Bessel equation:
gas pressure and the plasma radius which yields the Boltz- d2n 1 dn Z
mann equilibrium. This is illustrated in Fig. 8 for low pres- 2 + + n=0 共50兲
dr r dr D⬜
sures and in Fig. 9 for high pressures. Note that for ␣i = 1
共i.e., ␣e = 50兲, the Boltzmann equilibrium is satisfied reason- which yields

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-10 Sternberg, Godyak, and Hoffman Phys. Plasmas 13, 063511 共2006兲

FIG. 8. Deviation from the Boltzmann equilibrium at low pressures 共␣i = 1兲 for different magnitudes of the magnetic field: 共a兲 be = 1, 共b兲 be = 3, 共c兲 be = 10, and
共d兲 be = 100.

n
n0
= J0 r冉冑 冊 Z
D⬜
. 共51兲 n共R兲vs = − D⬜
dn
dr
共R兲. 共54兲

According to Fig. 1, for strong magnetic fields, the value of Substituting 共52兲 into 共54兲 and using the Taylor expansion
the plasma density at the plasma boundary, y 1 = n共R兲 / n0, is about ␦ = 0, we find
close but not equal to zero. Therefore, we can use the fol- D⬜ 1
lowing representation for the plasma density distribution ␦= = Ⰶ1 共55兲
v sR ␤
n
n0

= J0 2.405共1 − ␦兲
r
R
冊 共52兲 which together with 共52兲 and 共53兲 yields 共37兲 for the plasma
density at the plasma boundary and the normalized ioniza-
tion frequency.
for a sufficiently small ␦ that is to be determined. In particu- Using the numerical solution, we have found for ␣i = 1
lar, equating 共51兲 and 共52兲 yields for ␦ Ⰶ 1 and be = 100 that the representation for the plasma density by
the Bessel function in 共52兲 is quite accurate and yields a
冉 冊Z
D⬜
1/2
=
2.405共1 − ␦兲 2.405
R

R
. 共53兲
relative error of about 0.6% at the plasma boundary.
Consider once more Fig. 10. Observe that for a fixed
pressure, the two approximations for y 1 given by formulas
In order to determine ␦, first note that Eq. 共49兲 yields the 共36兲 and 共37兲 intersect at some critical value ␤*. This value
following relationship at the plasma boundary: satisfies the following relationship:

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-11 Magnetic field effects on gas discharge plasmas Phys. Plasmas 13, 063511 共2006兲

FIG. 9. Deviation from the Boltzmann equilibrium at high pressures 共␣i = 100兲 and different magnitudes of the magnetic field: 共a兲 be = 10, 共b兲 be = 30, and 共c兲
be = 100.

1.25 共approximately G 艌 2兲. In the weakly magnetized region, the


0.8共4 + ␣i兲−1/2 = 共56兲 effect of the magnetic field is negligible, the Boltzmann equi-
␤*
librium holds, and for a fixed ␣i, the boundary plasma den-
or equivalently, sity y 1共G兲 ⬇ const. is controlled only by ion inertia, ioniza-
0.64共4 + ␣i兲−1/2␤* = 1. 共57兲 tion, and ion charge-exchange. This is the region where y 1
and S can be approximated according to 共36兲. In the strongly
Thus, for a fixed pressure, the critical value ␤* separates the magnetized region, the plasma is strongly affected by the
parameter domain of weak magnetic field effects from the magnetic field, the Boltzmann equilibrium does not hold, y 1
domain of strong magnetic field effects. Note that ␤* is a and S fall rapidly as the magnetic field increases, and the
function of ␣i. For example, ␤* = 3.5 for ␣i = 1 and ␤* = 15.9 plasma confinement is controlled by the classical diffusion
for ␣i = 100. Recall that the Boltzmann equilibrium holds for across the magnetic field. This is the region where the effect
␣i = 1 at ␤ 艋 0.02, and for ␣i = 100 at ␤ 艋 0.2. Introducing the of ion collisions with gas atoms is negligible and for a fixed
parametric relation magnetic field, y 1 and S can be approximated by formula
G = 0.64共4 + ␣i兲−1/2␤ 共58兲 共37兲. We have interpolated the two limiting representations
共36兲 and 共37兲 of y 1 and S and found the following approxi-
provides a criterion G = 1 for the separation of the two pa- mation of y 1 and S in the whole parameter range, from weak
rameter domains. Figure 11 shows the dependence of y 1 and to strong magnetic fields:
S on the parameter G for different gas pressures. One can
y 1 = y 01共1 + G + G2兲−1/2 共59兲
distinctly see two regions: the weakly magnetized region
共approximately G 艋 0.5兲, and the strongly magnetized region and

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-12 Sternberg, Godyak, and Hoffman Phys. Plasmas 13, 063511 共2006兲

FIG. 10. Dependence of 共a兲 the normalized ionization frequency S and 共b兲
the normalized plasma density at the plasma boundary y 1 on the magnetic
parameter ␤ for different collision parameters ␣i. For low and high magnetic
fields, the asymptotic solutions given by Eqs. 共36兲 and 共37兲 are indicated by
the straight lines.

S = S0关1 + 0.59G + 共0.59G兲2兴−1/2 . 共60兲


FIG. 11. Dependence of 共a兲 the normalized plasma density at the plasma
In Fig. 11 approximations 共59兲 and 共60兲 of y 1 and S are boundary y 1 and 共b兲 the normalized ionization frequency S on the magnetic
indicated by an asterisk. parameter G for different collision parameters ␣i. The approximations given
by 共a兲 Eq. 共59兲 and 共b兲 Eq. 共60兲 are indicated by an asterisk.

VI. CONCLUSIONS
共1兲 We have used the nonlinear ion friction force, which
In the present article, we have studied the model for a is more suitable for low pressure discharges with cold ions
cylindrical plasma in an axial magnetic field. The model is than the linear ion friction force used in Refs. 20.
applicable to a wide range of discharge conditions, as ioniza- 共2兲 We were able to express the plasma characteristics
tion, ion and electron inertia, as well as ion and electron through external parameters of the problem such as gas pres-
collisions with gas atoms were taken into account. Using sure, plasma size, and the magnetic field. This makes our
numerical solutions of the model, we have analyzed the ef- representation transparent and convenient for practical appli-
fects of arbitrary magnetic fields on plasma characteristics cations. In contrast, in Refs. 20 and 24, the plasma charac-
for various gas pressures. teristics were expressed through a combination of external
Although our model is close to the ones found in Refs. parameters and the eigenvalue of the problem 共ionization
20 and 24, our approach and results are significantly differ- frequency兲, which is less practical. One can certainly argue
ent. Our results can be summarized as follows. that the electron temperature used in our normalization is not

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-13 Magnetic field effects on gas discharge plasmas Phys. Plasmas 13, 063511 共2006兲

really an external parameter. However, unlike the ionization


frequency, the electron temperature can readily be measured
or estimated, and it does not change much over the whole
range of discharge conditions.
共3兲 We have obtained numerical solutions for a wide
range of external parameters, whereas in Refs. 20 and 24,
numerical solutions for some specific parameter values were
presented. In this way, we have provided scaling laws for
basic discharge parameters as functions of the external pa-
rameters.
共4兲 There is no consensus in the literature on the crite-
rion for the onset of significant magnetic field effects on gas
discharge plasmas. Some assume the criterion to be ␳e Ⰶ R,
as electrons are the most sensitive plasma component with
respect to magnetic field effects. Others believe the criterion
to be ␳i Ⰶ R, as, without a magnetic field, the plasma trans-
port to the wall is governed by ion inertia and ion friction
force. In Ref. 12, the authors used the criterion ␻ec / ␯e ⬎ 1,
FIG. 12. BR ⫻ pR diagram for an argon discharge plasma at kTe = 3 eV. The
whereas in Ref. 21, ␻ci␻ce / 共␯e␯i兲 ⬎ 1 was suggested, where curve G = 1 separates the region with strong magnetic fields effects, from the
␯i is the ion collision frequency with gas atoms. We have region with no magnetic field effects. For G ⬍ 0.5 and G ⬎ 2 the correspond-
found analytically that for a bounded active plasma in an ing asymptotic solutions are valid.
axial magnetic field, significant magnetic field effects occur
only if

␤=
R␭e
␳ e␳ i
Ⰷ 1,
G = 0.64共4 + ␣i兲−1/2␤ = 0.64 4 + 冉 冊 R
␭i
−1/2
R␭e
␳ e␳ i
,

which separates the region of weak magnetic field effects


and in this case, the Boltzmann equilibrium relation is not from the region of strong magnetic field effects. The transi-
applicable. This shows that the alternative criteria used by tion from the region with no magnetic field effects to the
others are only partially true. For ␤ Ⰶ 1, the effect of the region with strong magnetic field effects occurs in the pa-
magnetic field is negligible. Observe that unlike the findings rameter range 0.5⬍ G ⬍ 2. For practical applications, it is
in Refs. 12 and 21, our magnetic field parameter ␤ depends useful to represent the different plasma regions 共where the
on the plasma size. asymptotic approximations are valid兲 in the BR ⫻ pR plane.
共5兲 We have shown that the plasma solution has two Such a representation is shown in Fig. 12 for an argon dis-
asymptotic approximations. One is for the weakly magne- charge plasma. As one can see, the region of no magnetic
tized plasma 共␤ Ⰶ 1兲 where there are no magnetic field ef- field effects 共G 艋 0.5兲 is separated from the region of strong
fects, and the Boltzmann equilibrium holds. In this case, the magnetic field effects 共G 艌 2兲 by the curve G = 1.
boundary plasma density y 1 and the ionization frequency S 共7兲 For the whole parameter range, from weak to strong
can be approximated by magnetic fields, we have found the following analytical ex-
pressions for the plasma density at the plasma boundary and
y 01 = 0.8共4 + ␣i兲−1/2 ⬃ p−1/2, S0 = 2.2共4 + ␣i兲−1/2 the ionization frequency as functions of the external dis-
charge parameter G:
which were obtained without a magnetic field.4
y 1 = y 01共1 + G + G2兲−1/2
The second asymptotic approximation is for the strongly
magnetized plasma 共␤ Ⰷ 1兲 where the Boltzmann equilibrium and
is not applicable. In this case, the plasma density distribution
S = S0关1 + 0.59G + 共0.59G兲2兴−1/2 ,
is represented by the Bessel function:
which are in very good agreement with the corresponding
n
n0
冉 冉 冊冊
= J0 2.405 1 −
1 r
␤ R
numerical results.
Although the results presented in this article for the one-
dimensional analysis allow to evaluate the magnetic field
which accounts for the nonzero plasma density at the plasma effects on bounded gas discharge plasmas, the applicability
boundary. In strongly magnetized plasmas, y 1 and S can be of those results have certain limitations. Even for long cylin-
approximated by drical plasmas when the plasma length is much larger than
the plasma radius, the axial plasma diffusion along the mag-
1.25 5.78 netic field may dominate the plasma transport.24 Moreover,
y B1 = ⬃ p1/2Bz−2, SB = . the magnitude of the maximal magnetic field is limited by
␤ ␤
the onset of plasma instabilities that can result in the anoma-
共6兲 We have found a parametric relation, G = 1, where lous 共Bohm兲 diffusion.

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions
063511-14 Sternberg, Godyak, and Hoffman Phys. Plasmas 13, 063511 共2006兲

13
ACKNOWLEDGMENT I. I. Beilis and M. Keidar, Phys. Plasmas 5, 1545 共1998兲.
14
D. D. Tskhakaya, P. K. Shukla, B. Eliasson, and S. Kuhn, Phys. Plasmas
The authors thank R. N. Franklin for helpful discussions 12, 103503 共2005兲.
15
and comments, and especially for suggesting to include Fig. R. F. Fernsler, S. P. Slinker, and G. Joyce, Phys. Rev. E 71, 026401
12. 共2005兲.
16
B. Alterkop, S. Goldsmith, and R. L. Boxman, Contrib. Plasma Phys. 45,
1 485 共2005兲.
F. F. Chen, Introduction to Plasma Physics 共Plenum, New York, 1974兲. 17
2 R. N. Franklin, Phys. Plasmas 13, 024701 共2006兲.
M. A. Lieberman and A. J. Lichtenberg, Principles of Plasma Discharges 18
D. Tskhakaya, B. Eliasson, P. K. Shukla, and S. Kuhn, Phys. Plasmas 13,
and Material Processing 共Wiley, New York, 2005兲.
3 024702 共2006兲.
F. Vidal, T. Wyatt, J. Margot, M. Chaker, and O. Pauna, IEEE Trans. 19
J. R. Forrest and R. N. Franklin, Br. J. Appl. Phys. 17, 1061 共1966兲.
Plasma Sci. 27, 727 共1999兲. 20
4
V. A. Godyak and V. N. Maximov, Vestnik Moscovskogo Universiteta, R. N. Franklin, Plasma Phenomena in Gas Discharges 共Clarendon, Ox-
Fisika i Astronomia 5, 51 共1977兲. ford, 1976兲.
21
5
V. A. Godyak, ESCAMPIG-90, edited by B. Dubreuil 共European Physical V. E. Golant, A. P. Zhilinsky, and I. E. Sakharov, Fundamentals of Plasma
Society, Orleans, France, 1990兲, contributed papers, p. 415. Physics 共Wiley, New York, 1980兲.
22
6
R. Chodura, Phys. Fluids 25, 1628 共1982兲. V. A. Rozhansky and L. D. Tsendin, Transport Phenomena in Partially
7
K.-U. Riemann, Phys. Plasmas 1, 552 共1994兲. Ionized Plasma 共Taylor and Francis, London, 2001兲.
8 23
N. Sternberg and J. Poggie, IEEE Trans. Plasma Sci. 32, 2217 共2004兲. J. Poggie and N. Sternberg, Phys. Plasmas 12, 023502 2005.
9 24
R. N. Franklin, J. Phys. D 38, 2790 共2005兲. A. Fruchtman, G. Makrinich and J. Ashkenazy, Plasma Sources Sci.
10
G. H. Kim, N. Hershkowitz, D. A. Diedbold, and M.-H. Cho, Phys. Plas- Technol. 14, 152 共2005兲.
25
mas 2, 322 共1995兲. S. A. Self and H. N. Ewald, Phys. Fluids 9, 2488 共1966兲.
26
11
M. Keidar, I. Beilis, R. L. Boxman, and S. Goldsmith, J. Phys. D 29, 1973 K.-B. Persson, Phys. Fluids 5, 1625 共1962兲.
27
共1996兲. V. A. Godyak, R. B. Peijak, and B. M. Alexandrovich, J. Appl. Phys. 85,
12
I. I. Beilis, M. Keidar, and S. Goldsmith, Phys. Plasmas 4, 3451 共1997兲. 3081 共1999兲.

Downloaded 29 Jul 2012 to 132.236.27.111. Redistribution subject to AIP license or copyright; see http://pop.aip.org/about/rights_and_permissions

You might also like