You are on page 1of 7

Home Search Collections Journals About Contact us My IOPscience

Ion streaming instability in a plasma sheath with multiple ion species

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2012 Plasma Phys. Control. Fusion 54 082001

(http://iopscience.iop.org/0741-3335/54/8/082001)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 130.15.241.167
The article was downloaded on 03/05/2013 at 18:37

Please note that terms and conditions apply.


IOP PUBLISHING PLASMA PHYSICS AND CONTROLLED FUSION
Plasma Phys. Control. Fusion 54 (2012) 082001 (6pp) doi:10.1088/0741-3335/54/8/082001

BRIEF COMMUNICATION

Ion streaming instability in a plasma


sheath with multiple ion species
Zhike Zhang and Xiaogang Wang
State Key Lab of Nuclear Physics & Technology and School of Physics, Peking University, Beijing
100871 People’s Republic of China
E-mail: xgwang@pku.edu.cn

Received 21 February 2012, in final form 27 April 2012


Published 25 May 2012
Online at stacks.iop.org/PPCF/54/082001

Abstract
The ion streaming instability in a dc plasma sheath with multiple ion species is studied as an
eigenvalue problem. By numerically solving the eigenvalue equations of sheath plasma
variables, we find a weak ion streaming instability induced in the sheath with a growth rate of a
few per cent of the proton plasma frequency. The instability may provide a mechanism to
enhance the collisional friction between different species of ions and result in different ions
leaving the plasmas nearly at the same velocity. The effect of the species ratio on the growth
rate of the instability is also discussed.

1. Introduction the velocities of various species of ions into the same system
ion sound speed. In recent experiments such instabilities with
The plasma sheath containing multiple species of ions plays a a growth rate of a few per cent of the proton plasma frequency
critical role in a broad range of plasma applications including have indeed been observed [17, 18, 25, 26].
transport process of plasmas in fusion devices [1–4], plasma Nevertheless, in a recent simulation in multiple fluid
processing of materials [5], space plasmas [6] and diagnostics model for both dc and rf sheaths, the ion–ion streaming
[7]. With only a single ion species in the plasma sheath, the instability was not seen [27]. Different from the experiments
Bohm criterion can be applied, for ions entering the sheath and the later theoretical model [16–26], it was found in the
with
√ a velocity equal to or greater than the ion acoustic velocity simulation that each species of ions indeed entered the sheath
Te /mi , also called the Bohm velocity [8–12]. Similar to the at their own Bohm velocities, as predicted by the earlier theory
Bohm criterion for the single ion species sheath, the previous [13–15]. Thus, the ion–ion streaming instability is still an issue
theory of plasma sheath with multiple ion species assumed in our understanding of the plasma sheath properties.
that each species of ions entered the sheath edge at different The analysis of ion–ion streaming instability in theory
velocities by satisfying their own Bohm criterion [13–15]. was based on the dispersion relation derived in the theory of
However, this assumption was challenged by experiments two-stream instability in uniform plasmas [22–24]. It was
[16–21]. In these experimental studies, He–Ar plasmas and thus valid in the presheath, where the characteristic scale of
Ar–Xe plasmas in a low-temperature regime were investigated. the inhomogeneity was much longer than the wavelength of
It was then found that each species of ions entered the sheath the mode. The instability was then a possible mechanism
edge nearly
√ at the same velocity, the ‘system’ ion sound speed to accelerate the ion to its ‘system’ Bohm velocity in the
Cs = (n1 /n0 )VB1 + (n2 /n0 )VB2 , where n1 , VB1 and n2 , VB2 presheath. Nevertheless, the characteristic length scale of
are the densities and Bohm velocities for the corresponding the plasma sheath is more or less on the same order of the
species, rather than their individual Bohm velocities. To disturbance wavelength, i.e. both are scaled by the Debye
resolve the issue, certain theoretical models were developed length. It is then clearly a typical eigenvalue problem for
[22–24]. In those models, the ion–ion streaming instability the instability in the sheath region. In this work, we revisit
was introduced for enhancing the collisional friction between the problem in a new model of eigenvalue equations only
the different species of ions. The friction force then dragged concerning the sheath and a transition region of ∼100 Debye

0741-3335/12/082001+06$33.00 1 © 2012 IOP Publishing Ltd Printed in the UK & the USA
Plasma Phys. Control. Fusion 54 (2012) 082001 Brief Communication

lengths. The eigenvalue equations are derived from the vj2s (x) eφs (x) vj2 (0) eφ(0) 2
VBj
multiple fluid model similar to that in [27]. Nevertheless, + = + ≈ , (9)
2 mj 2 mj 2
to focus on finding the instability, we study a collisionless  
dc sheath. The equations are then solved numerically with ∂ 2 φs (x) 
appropriate boundary conditions, and a weak ion steaming ε0 = e nes (x) − nj s (x) . (10)
∂x 2 j
instability is found with the growth rate on the order
 of a few 
per cent of the proton plasma frequency ωpi = n0 e2 /mi ε0 , Here, VBj = Te /mj is the Bohm velocity of the j th ion. Also
where mi is the proton mass. φ(0) ≈ 0 is applied for the electric potential at the position of
This paper is organized as follows. In section 2, the basic x = 0 in the transition region.
equations of the multiple fluid model are presented and steady- Then, for convenience, we normalize equations (1)–(10)
state solutions of the equations are derived. The linearization by introducing the following dimensionless quantities:

and the eigenvalue equations are then shown in section 3. In nj 0 VBj mi V Te mi
ηj = , βj = = , βe = =
section 4, the numerical results are obtained, together with n0 VBi mj VBi me
analysis. The paper is concluded in section 5 with a brief x ω
summary and discussion. x̄ = , t¯ = ωpi t, ω̄ = ,
λD ωpi
ne nj ve vj eφ
Ne = , Nj = , ue = , uj = , = ,
2. Model equations n0 n0 VBi VBi Te
nes (x) nj s (x)
In the model, we consider that the plasma consists of electrons N̄e (x̄) = , N̄j (x̄) = ,
n0 n0
and two species of cold ions [27, 28]. To focus on finding the
ves (x) vj s (x) eφ (x)
instability, we have neglected the source and collision terms in ūe (x̄) = , ūj (x̄) = , ( ¯ x̄) = s .
the model. For simplicity, only the one-dimensional problem VBi VBi Te
is considered, and the assumption of the two species of ions Here n0 is the plasma density,  n j 0 is the density of the j th ion
both being singly ionized is applied. Then the multiple fluid in the plasma
 region, λ D = T ε
e 0 /n 0 e 2 is the Debye length,

equations for electrons and ions are ωpi = n0 e2 /mi ε0 is the plasma frequency √ of the proton with
mi being the proton mass, VBi = √ Te /mi = ωpi λD is the
∂ne ∂ Bohm velocity of the proton, VTe = Te /me is the electron
+ (ne ve ) = 0, (1)
∂t ∂x thermal velocity. Equations (1)–(10) then become
  ∂Ne ∂
∂ve ∂ve ∂φ ∂(ne Te ) + (Ne ue ) = 0, (11)
me ne + ve = ene − , (2) ∂t ¯ ∂ x̄
∂t ∂x ∂x ∂x
1 ∂ue 1 ∂ue ∂ 1 ∂Ne
∂nj ∂ + ue = − , (12)
+ (nj vj ) = 0, (3) βe2 ∂ t¯ βe2 ∂ x̄ ∂ x̄ Ne ∂ x̄
∂t ∂x ∂Nj ∂
+ (Nj uj ) = 0, (13)
∂vj ∂vj e ∂φ ∂ t¯ ∂ x̄
+ vj =− , (4)
∂t ∂x mj ∂x 1 ∂uj 1 ∂uj ∂
2 ∂ t¯
+ 2 uj =− , (14)
  βj βj ∂ x̄ ∂ x̄
2 
∂ φ
ε0 2 = e ne − nj  , (5) ∂ 2 
∂x = N e − Nj ; (15)
j ∂ x̄ 2 j
with the subscript j = 1, 2 for the two species of ions, and me , with their steady-state solutions
ne , ve , and mj , nj , vj for the electron mass, density, velocity, ¯
N̄e (x̄) = e(x̄) , (16)
and ion mass, density, velocity, respectively; φ stands for the
ūe (x̄) = 0, (17)
electrostatic potential, ε0 for the vacuum permittivity and e
ηj βj ηj
for the element charge. The isothermal assumption is applied N̄j (x̄) = = , (18)
for the electron with Te being a constant, while the ions are ūj (x̄) ¯ x̄)
1 − 2(

assumed to be cold.
ūj (x̄) = βj 1 − 2( ¯ x̄), (19)
In comparison with ions, the electron mass can be
neglected. Thus, we can further assume massless electrons 2¯ x̄)
d (
=¯  (x̄) = N̄e (x̄)
satisfying the Boltzmann distribution in the steady state. Then, dx̄ 2
the steady-state solution of the model can be obtained easily.  ¯ 1
We write the steady solutions of equations (1)–(5) as − N̄j (x̄) = e(x̄) −  . (20)
j
¯ x̄)
1 − 2(
ves (x) = 0, (6) Obviously, by integrating equation (20), we can obtain the
steady-state distribution of the electrostatic potential and then
nes (x) = ne0 exp[eφs (x)/Te ], (7) steady-state distributions of the ion densities and velocities
by substituting the steady-state distribution of the electrostatic
nj s (x)vj s (x) = nj (0)vj (0) = nj 0 VBj , (8) potential into equations (18) and (19).

2
Plasma Phys. Control. Fusion 54 (2012) 082001 Brief Communication

3. Linearization and eigenvalue equations

In nonuniform plasmas, with a perturbation to the steady-


state equilibrium, a function of f (x, t) can be divided into
two parts as f (x, t) = fs (x) + f˜(x, t), with the perturbation
being written in the form of f˜(x, t) = f˜(x)e−iωt . In the low-
frequency regime considering only the response of ions, we
can obtain

Nj (x̄, t¯) = N̄j (x̄) + Ñj (x̄, t¯) = N̄j (x̄) + Ñj (x̄)e−iω̄t¯, (21)

uj (x̄, t¯) = ūj (x̄) + ũj (x̄, t¯) = ūj (x̄) + ũj (x̄)e−iω̄t¯, (22)
¯ x̄) + (
(x̄, t¯) = ( ˜ x̄, t¯) = (
¯ x̄) + (
˜ x̄)e−iω̄t¯. (23)
Substituting (21)–(23) into equations (13)–(15), we derive the
linearized equations
Figure 1. Steady-state distribution of the electric potential in the
−iω̄Ñj (x̄) + ηj [1 − 2( ¯ x̄)]−1/2 transition and sheath regions.
 
dũj (x̄)
× ¯ −1 ¯ 
+ [1 − 2(x̄)]  (x̄)ũj (x̄)
dx̄ equations, we get the steady-state distributions of densities and
¯ x̄)]1/2 velocities of the electron and the ions from equations (16)–(19),
+βj [1 − 2(
  shown in figure 2. Thus, the steady-state solution is obtained.
dÑj (x̄) Due to the negative electric potential applied on the wall,
× − [1 − 2( ¯ x̄)] 
−1 ¯ (x̄)Ñj (x̄) = 0, (24)

dx̄ the electron and both ion densities decrease rapidly near the
wall. In particular, the electron density is nearly zero, as
1 1 dũ (x̄) shown in figure 2. Applying the solutions in figures 1 and 2 to
−iω̄ ¯ x̄)]1/2 j
ũj (x̄) + [1 − 2(
βj2 βj dx̄ equations (24)–(26), we can easily solve the linear eigenvalue
problem. To consider instabilities in the sheath region, we
1 ˜ x̄)
d(
− [1 − 2( ¯  (x̄)ũj (x̄) = −
¯ x̄)]−1/2  , (25) integrate the equations from a position in the transition region
βj dx̄ at about x̄ = 60 far from the sheath region. Then the boundary
˜ x̄)
d2 (  conditions are
2
=− Ñj (x̄). (26) 
dÑj 
dx̄ j
Ñj (60) = 0,  = 0,
With appropriate boundary conditions, we can solve equations dx̄ 
(24)–(26) to determine the eigenfunctions of Ñj (x̄), ũj (x̄) and  x̄=60
dũj 
˜ x̄) with the eigenvalue ω̄ at the same time. If the imaginary ũj (60) = 0, = 0;
dx̄ x̄=60
(
part of the eigenvalue ω̄ = ω̄R + iω̄I is positive, there is then
an instability due to ion streaming. Otherwise, the sheath is ˜
while (60) ˜
= 0, and (L) on the wall is set free but limited
stable, at least in the low-frequency regime where the fluid in the linear regime with (L)˜  1. Then, we integrate
approximation is valid. equations (24)–(26) and obtain the linear eigenfunction
solutions, as shown in figures 3 and 4, with the eigenvalue
4. Numerical results ω̄ = +i0.0436. It is imaginary with a positive ω̄I = 0.0436.
Clearly, for parameters we have chosen, there is an unstable
We then solve the eigenvalue equations numerically. ion steaming mode in the sheath and the transition region
With the scale length of λD , we integrate equations (24)– with a growth rate of a few per cent of the proton plasma
(26) in a domain of L = 100 to include the sheath and a frequency ωpi .
transition region in our simulation. The boundary x̄ = 0 is To further study the instability, we redo the simulations
in the transition region, and x̄ = L is on the wall. Then the with various parameter sets. First, we change the ratio of
boundary conditions (0) = 0 and (L = 100) = −30 can the two species with other parameters kept the same as in the
be applied, with an assumed electron temperature of Te = 1 eV previous case. Varying η1 from 0 to 1 with η2 = 1 − η1 ,
and a −30 V bias applied
 on the wall. Also, the typical plasma
we repeat the previous simulation to get the growth rate as a
density is chosen as j nj (0) = ne (0) = n0 = 1 × 1016 m−3 . function of η1 , shown in figure 5. We can find that the growth
Based on these conditions, the steady-state distribution of rate γ increases almost linearly with η1 . Since m1 < m2 ,
the electric potentials is easily obtained from equation (20) by the mode becomes more unstable if the proportion of the
shooting from both boundaries, as shown in figure 1, where heavier ion decreases (η2 = 1 − η1 ). This indicates that the
an indistinct interface can be seen between the sheath and the lighter ion species is more responsible for the instability, since
transition region. the lighter ion streaming is faster. To further investigate the
Taking m1 = mi , m2 = 4mi and η1 = η2 = 0.5 as characteristics of the instability, we keep η1 = 0.5 and change
in [27], and substituting the solution shown in figure 1 into the the parameter m2 from 4mi to 100mi . The result is shown in

3
Plasma Phys. Control. Fusion 54 (2012) 082001 Brief Communication

Figure 2. Steady-state distributions of densities and velocities of the electron and ions in the transition and sheath regions: (a) the density of
species 1 ions (n2 = n1 not shown); (b) electron density; (c) comparison of the electron and ion densities (here ni = n1 + n2 ); (d) velocities
of the two species of ions.


of the growth rate shown in figures 5 and 6. The ∼1/ mi
dependence is also in accordance with the heavier ion mass
effect shown in figure 6.

5. Summary and discussion

In this paper, we have investigated the problem of a plasma


sheath containing two species of cold ions in a multiple fluid
model. The eigenvalue equations of the perturbations of the
electric potential, ion densities and ion velocities were derived.
Numerical simulations based on these equations were carried
out, and an ion streaming instability was observed in the
simulations. The growth rate of the instability was found to be
on the order of a few per cent of the proton plasma frequency.
The effect of parameters on the growth rate of the instability
was also discussed. It was found that the heavy ion mass can
˜
Figure 3. Eigenfunction distribution of the potential perturbation ,
with the eigenvalue ω̄I = 0.0436.
significantly stabilize the mode.
There is still a question to be answered: why is the
figure 6. In accordance with figure 5, where the growth rate of instability not seen in the previous simulation using a similar
the instability decreases as the heavier ion portion increases, model? In fact, in the simulation in [27], a collision rate of
the growth rate γ decreases when m2 increases in figure 6. 0.04ωpi between each ion species and the neutrals was applied.
There is another interesting point shown in figure 5. For a It is on the same order of the growth rate of the mode we have
single ion species, e.g. when η1 = 0 for the second species only found. With such a collision rate, the mode should easily be
and η1 = 1 for the first species only, the instability still exists. damped. Nevertheless, the collision effect is a complicated
Therefore, the instability is clearly not an ion–ion streaming issue. To take it into account in the model, we also need to
mode, but mainly due to the ion–electron steaming. In fact, introduce the collision between the two ion species. Thus, the
the growth rate of two-stream modes in uniform plasmas with friction between the two ion species would have been in place
streaming ions and resting electrons can be roughly estimated already.
to be γ ∼ kV0 ωpi /ωpe , where V0 is the streaming velocity. For Another important effect we have not taken into account
our simulation parameters, √ kV0 ∼ VBi /λD ∼ ωpi , we can get in the paper is the ionization. In the low-frequency regime,
the estimate of γ /ωpi ∼ me /mi ∼ 10−2 , on the same order ionization instability has been found in low-temperature

4
Plasma Phys. Control. Fusion 54 (2012) 082001 Brief Communication

Figure 4. Eigenfunction distributions of Ñ1 , Ñ2 , ũ1 and ũ2 , with the eigenvalue ω̄I = 0.0436.

Figure 5. Growth rate of instability γ (=ω̄I ) versus the species ratio Figure 6. Growth rate of instability γ (=ω̄I ) versus the ion mass
parameter η1 , with stars for numerical results and the solid line for ratio m2 /m1 , with stars for numerical results and the solid curve for
the fitting. the fitting.

plasma discharges and theoretical analyses [29–31]. The


[2] Brooks J N 1996 Phys. Plasmas 3 2286
instability can be another mechanism to excite the friction
[3] Xu X Q, Umansky M V, Dudson B and Snyder P B 2008
between various ion species. We leave the investigation to Commun. Comput. Phys. 4 949
a future work. [4] Li B, Rogers B N, Ricci P and Gentle K W 2009 Phys.
Plasmas 16 082510
[5] Lieberman M A and Lichtenberg A J 2005 Principles of
Acknowledgment Plasma Discharges and Materials Processing (New York:
Wiley)
This work is supported by NSFC, Grant Nos 40731056 and [6] Barjatya A, Swenson C M, Thompson D C and Wright K H
10975012. 2009 Rev. Sci. Instrum. 80 041301
[7] Fernsler R F 2009 Plasma Sources Sci. Technol. 18 014012
References [8] Bohm D 1949 The Characteristics of Electrical Discharges in
Magnetic Fields ed A Guthrie and R K Wakerling
[1] Void E L, Najmabadi F and Conn R W 1991 Phys. Fluids B (New York: McGraw-Hill) chapter 3
3 3132 [9] Riemann K U 1991 J. Phys. D: Appl. Phys. 24 493

5
Plasma Phys. Control. Fusion 54 (2012) 082001 Brief Communication

[10] Riemann K U1997 Phys. Plasmas 4 4158 [22] Baalrud S D, Hegna C C and Callen J D 2009 Phys. Rev. Lett.
[11] Oksuz L and Hershkowitz N 2002 Phys. Rev. Lett. 89 145001 103 205002
[12] Severn G D, Wang X, Ko E and Hershkowitz N 2003 Phys. [23] Baalrud S D, Callen J D and Hegna C C 2010 Phys. Plasmas
Rev. Lett. 90 145001 17 055704
[13] Riemann K U 1995 IEEE Trans. Plasma Sci. 23 709 [24] Baalrud S D and Hegna C C 2011 Phys. Plasmas 18 023505
[14] Franklin R N 2000 J. Phys. D: Appl. Phys. 33 3186 [25] Yip C S, Hershkowitz N and Severn G 2010 Phys. Rev. Lett.
[15] Franklin R N 2001 J. Phys. D: Appl. Phys. 34 1959 104 225003
[16] Wang X and Hershkowitz N 2006 Phys. Plasmas 13 053503 [26] Hershkowitz N, Yip C S and Severn G D 2011 Phys. Plasmas
[17] Hershkowitz N, Ko E, Wang X and Hala A M A 2005 IEEE 18 057102
Trans. Plasma Sci. 33 631 [27] Xiang N, Hu Y and Ou J 2011 Plasma Sci. Technol. 13 385
[18] Hershkowitz N 2005 Phys. Plasmas 12 055502 [28] Xiang N 2004 Phys. Plasmas 11 4213
[19] Lee D, Hershkowitz N and Severn G D 2007 Appl. Phys. Lett. [29] Akhiezer A I, Akhiezer I A and Angeliko V V 1970 Sov.
91 041505 Phys.—JETP 30 476
[20] Lee D, Oksuz L and Hershkowitz N 2007 Phys. Rev. Lett. [30] Johnson J C, D’Angelo N and Merlino R L 1990 J. Phys. D:
99 155004 Appl. Phys. 23 682
[21] Oksuz L, Lee D and Hershkowitz N 2008 Plasma Sources Sci. [31] Wang X G, Bhattacharjee A, Gou S K and Goree J 2001 Phys.
Technol. 17 015012 Plasmas 8 5018

You might also like