High-frequency electrical breakdown in electronegative gases with fast electron attachment

and detachment processes

This article has been downloaded from IOPscience. Please scroll down to see the full text article.
1997 J. Phys. D: Appl. Phys. 30 1403
(http://iopscience.iop.org/0022-3727/30/9/016)
Download details:
IP Address: 152.135.235.188
The article was downloaded on 16/11/2012 at 21:36
Please note that terms and conditions apply.
View the table of contents for this issue, or go to the journal homepage for more
Home Search Collections Journals About Contact us My IOPscience
J. Phys. D: Appl. Phys. 30 (1997) 14031407. Printed in the UK PII: S0022-3727(97)73274-1
COMMENT
High-frequency electrical breakdown
in electronegative gases with fast
electron attachment and detachment
processes
N L Aleksandrov, A P Napartovich and A M Okhrimovskyy
Department of Physical Mechanics, Moscow Institute of Physics and Technology,
141700, Dolgoprudny, Russia
Troitsk Institute for Innovation and Fusion Research, 142092 Troitsk,
Moscow Region, Russia
Received 1 April 1996
Abstract. Breakdown in a high-frequency uniform electric eld in electronegative
gases with fast electron attachment and detachment processes is theoretically
studied. A process of electron pseudo diffusion introduced earlier by Aints et al is
analysed, and a breakdown criterion is regularly derived using the great difference
in rate scales for collisional processes. The criterion is generalized to account for
an arbitrary number of negative ion species including those which are not involved
in electron detachment reactions. The importance of the electron pseudo diffusion
is illustrated for specied gases (air, N
2
O, etc). It is argued that the application of
the concept of pseudo diffusion to the breakdown in air under the conditions of the
experiment conducted previously by Aints et al is invalid.
1. Introduction
The development of high-frequency electrical breakdown
in a gas depends on a variety of parameters such as the
eld strength and frequency, the gas properties and the
electrode geometry (MacDonald 1966, Meek and Craggs
1978, Raizer 1991). At sufciently high gas densities when
the eld angle frequency is lower than the frequency
of the electron momentum relaxation
m
electrons execute
an oscillatory drift motion with the amplitude A =
e E
0
/(m
m
), where e and m are the electron charge and
mass and E
0
is the eld strength amplitude. If A < d/4
(d is the distance between electrodes), the main mechanism
of electron loss which governs the breakdown voltage in
an electropositive gas is the free electron diffusion to the
boundaries of the discharge volume. In an electronegative
gas, electron attachment to neutral species also becomes
important.
Aints et al (1995) have shown that if fast electron
detachment from negative ions compensates for the electron
loss through attachment to neutral species, the breakdown
voltage is lowered at A d and resulting losses could
be described as due to the process of pseudo diffusion of
electrons with a rate that might be up to several orders
of magnitude greater than that of normal diffusion. This
was invoked to explain a decrease in the radio frequency
(RF) breakdown voltage with increasing and d which
was observed experimentally in atmospheric-pressure air
(Aints et al 1995). According to this effect, the calculation
gave a value of the breakdown voltage which was 20% less
than the measured value. The authors supposed that this
discrepancy could be attributed to using one sort of negative
ion approximation. However, it is not quite clear whether
the concept of electron pseudo diffusion is applicable to
breakdown in air.
It should be noted that similar pseudo diffusion has
been considered for a few systems in a steady electric eld.
These are: electrons (or holes) transiting between energy
valleys with different effective electron masses in a solid
state plasma (Shockley et al 1966, Ohmi and Hasuo 1969,
Matulionis and Starikov 1981); electrons moving through
a gas and captured by molecules to form unstable negative
ions (Aleksandrov and Napartovich 1986), and stable ions
(Aleksandrov and Okhrimovskii 1995); and the transport
of ions in a gas with fast forwardbackward ionmolecule
reactions (Gatland 1974, Iinuma and Takebe 1992, Iinuma
et al 1994).
The aim of this paper is to derive the criterion
of breakdown in a high-frequency uniform electric eld
in electronegative gases with fast electron attachment
and detachment processes. A more regular procedure
based on perturbation theory is proposed to derive an
0022-3727/97/091403+05$19.50 c 1997 IOP Publishing Ltd 1403
Comment
explicit expression for the electron pseudo diffusion
coefcient. We clearly show the limits of validity of
the approximations which are used for the description of
electron pseudo diffusion and generalize the discussion in
order to take into account an arbitrary number of negative
ion species including those which are not involved in
electron detachment processes. The role of electron pseudo
diffusion is discussed both for air under the conditions of
the experiment conducted by Aints et al (1995) and for
more general conditions.
2. Basic equations
We consider breakdown in plane-to-plane electronegative
gas gaps in a uniform electric eld E = E
a
cos t ,
neglecting the eld of space charge because of the low
density of charged particles and assuming A d. Let
all collisional electron and ionmolecule processes in
the gas be classied into two typesfast processes and
slow processesand all negative ion species be classied
into two categoriesshort-lived ions and long-lived ions.
Unlike the latter, short-lived ions are involved in fast
ionmolecule processes which are completed by electron
detachment. The fast processes are electron attachment to a
molecule to form a short-lived negative ion, ionmolecule
reactions to convert one short-lived ion into another and
electron detachment from short-lived negative ions. The
slow processes are ionization of molecules by electron
impact, electron attachment to a molecule to form a long-
lived negative ion and ionmolecule reactions to convert a
short-lived negative ion into a long-lived one. In addition,
we assume that the reciprocal of the time it takes for an
electron to drift between the electrodes in a steady eld E
a
obeys the inequalities

f min
d/w
a

s max
(1)
where w
a
is the electron drift velocity in the eld E
a
,
f min
is the smallest of the frequencies of the fast processes
and
s max
is the greatest of the frequencies of the slow
processes.
The continuity equations for electrons and an arbitrary
number (k 1) of short-lived negative ion species are
n
t
=

Bn gE
a
cos t
n
1
x
+

Cn (2)
where
n =
_
_
_
_
n
1
n
2
.
.
.
n
k
_
_
_
_
g =
_
_
_
_
1
0
.
.
.
0
_
_
_
_

B =
_
_
_
_
R
1

21
. . .
k1

12
R
2
. . .
k2
.
.
.
.
.
.
.
.
.
.
.
.

1k

2k
. . . R
k
_
_
_
_
and

C =
_
_
_
_

ion

1
0 . . . 0
0
2
. . . 0
.
.
.
.
.
.
.
.
.
.
.
.
0 0 . . .
k
_
_
_
_
.
Here, n
i
is the number density of electrons (i = 1) and of
the short-lived negative ions (2 i k), is the electron
mobility,
ij
is the reaction frequency per ion (electron)
from species i to j,
R
i
=
k

j=1
j=i

ij
is the total frequency of removal of ith species through
fast processes,
ion
is the ionization frequency,
1
is the
frequency of electron attachment to molecules to form long-
lived ions and
i
is the frequency (per ion) of the reaction
which converts the short-lived negative ion of species i
into the long-lived one. In that notation, R
1
is a sum of
the frequencies of electron attachment to molecules to form
short-lived negative ions and
i1
is the frequency of electron
detachment from the short-lived negative ion of species i.
Equation (2) neglects electron diffusion and ion drift and
diffusion in comparison with the electron drift which is
directed along the x axis.
We assume that, with the exception of
ion
, the
quantities which are contained in matrices

B and

C and the
value of are time independent. This is true if
u
,
where
u
is the frequency of the electron energy relaxation.
The reverse inequality
u
is more typical for the RF
range of at atmospheric gas pressure; therefore, in this
case, the last assumption should be considered as a model
one.
It follows from (1) that on the right-hand side of
equation (2) the rst term is much greater than the second
one which in turn is much greater than the third term.
However, the third term cannot be neglected because it
determines the breakdown threshold.
In a short time of about
1
f min
, the densities of electrons
and short-lived ions reach the quasi-steady state
n
0
= n

_
_
_
_
p
1
p
2
.
.
.
p
k
_
_
_
_
(3)
where n
0
is the eigenvector of matrix

B corresponding to
the zero-eigenvalue, that is

Bn
0
= 0. (4)
Here, n

is the sum of the densities of electrons and short-

lived ions which is determined by the slow processes; p
i
are the relative densities of electrons (i = 1) and short-
lived ions (1 < i k) which are determined as (Iinuma et
al 1993)
p
i
=
(i, i)

k
i=1
(i, i)
(5)
where (i, i) is the cofactor of the (i, i) element in matrix

B.
From equations (3) and (5) a proportionality between the
electron density n
1
and n

follows. Therefore, in order

to derive the breakdown criterion, the continuity equation
for the time-averaged value of n
1
can be replaced by that
for the time-averaged value of n

. The latter equation is

1404
Comment
obtained by adding the scalar equations in system (2) and
averaging the result over the period of the electric eld.
Then, the rst terms on the right-hand side of equation (2)
cancel each other and the second terms corresponding to the
electron drift oscillations vanish on averaging at A d.
As a result, the breakdown threshold is governed by the
slow collisional processes.
Aints et al (1995) have shown that, under considered
conditions, there is an additional mechanism of electron
loss through pseudo diffusion which affects the breakdown
criterion. We study this point by perturbation theory which
is based on inequalities (1). When solving equation (2)
we account for the second term on the right-hand side in
perturbation theory and neglect the third term. To the rst
order, the quasi-steady solution of equation (2) should be
sought in the following form
n = n
0
+acos t +b sin t (6)
where a and b change only slowly with time. Substituting
equation (6) into (2), we have
a = w
a

B(

B
2
+
2

I)
1
g
n
e
x
(7)
and
b = w
a
(

B
2
+
2

I)
1
g
n
e
x
(8)
where w
a
= E
a
, n
e
= n

p
1
and

I is the identity matrix.

In order to derive the breakdown criterion we add the
equations in system (2) and average the resulting equation
over the eld period; in so doing we should account for
the small correction to the second terms on the right-hand
side of the equations in system (2) which follows from (6).
This correction is much less than the second terms which
vanish on averaging but is not less than the third terms
which survive. As a result, we have
n

t
=
ion
n
e
+

2
x
2
D

n
e

i=1

i
n
i
(9)
where the overline denotes time averaging
(t ) =
2
T
_
t +T/2
t
() d
(T = 2/ is the eld period) and
D

=
w
2
a
2
[

B(

B
2
+
2

I)
1
]
11
. (10)
In deriving equation (9) we assumed
ion
, that is n
e
varies only slightly over the period and

ion
n
e

ion
n
e
.
The second term on the right-hand side of equation (9) is
of a diffusion nature because it can be interpreted in terms
of a divergence of the electron ux which is proportional
to the gradient of n
e
along the x axis. The coefcient D

of this so-called pseudo diffusion depends on w

a
, and
the frequencies of the fast processes.
Using the relation between n

, n
e
and n
i
which follows
from (3) on averaging and assuming n
e
(x = 0) = n
e
(x =
d) = 0 we obtain from equation (9) for a uniform-eld gap
in the one-dimensional case (MacDonald 1966, Meek and
Craggs 1978)
n
e
(t ) = n
e
(0) exp
___
(
ion

d
)p
1

i=1

i
p
i
_
t
__
(11)
where

d
=

2
d
2
D

.
This is true if the sizes of the plane electrodes are much
greater than d. The breakdown criterion is written as

ion
=
d
+
k

i=1

i
p
i
p
1
. (12)
In the absence of long-lived negative ions the breakdown
threshold corresponds to the balance between the average
ionization and the loss of electrons due to pseudo diffusion
to the electrodes, whereas electron detachment compensates
for the loss of electrons due to attachment to molecules
(Aints et al 1995). The second term on the right-hand side
of equation (12) shows that the formation of long-lived
negative ions results in the additional loss of electrons.
It is interesting to note that according to equation (11)
both ionization and attachment rates are rescaled by the
relative density of electrons, p
1
< 1. This means that the
ionization and attachment rates effectively decrease.
3. The pseudo diffusion coefcient: particular
cases
Accounting for only one sort of short-lived negative ions
(k = 2), it follows from equations (5) and (10) that
p
1
=

21

12
+
21
p
2
=

12

12
+
21
(13)
and
D

=
w
2
a
2

12

2
+(
12
+
21
)
2
. (14)
Formula (14) has been derived earlier (Aints et al 1995)
and in the limit 0 it reduces to that in the steady
electric eld E = E
a
/

2 (Gatland 1974, Aleksandrov and

Napartovich 1986, Iinuma and Takebe 1992).
For the coefcient of common electron diffusion, we
have D
2

m
, where is the electron mean free
path for momentum transfer. Equation (14) shows that
the pseudo diffusion coefcient D

A
2

12
at >

12
+
21
, where A = w
a
/ is the amplitude of electron
drift oscillations. The mechanism of the pseudo diffusion
process may be explained in the following way (Aints et al
1995). An electron which executes oscillatory drift motion
with amplitude A along the eld attaches to a molecule
to form a short-lived negative ion which is essentially at
rest. Then, at a random moment the electron detaches from
the ion and oscillates around the new midpoint which is
displaced by a distance of the order of A along the x
1405
Comment
axis because the electric eld is now phase shifted. If
A d the phenomenon described for this electron is
repeated many times. The direction of the displacement
along the eld direction is also random. Therefore, the
considered transport of electrons (and short-lived negative
ions) can be treated as pseudo diffusion of charged particles.
The breakdown criterion (equation (12)) in a gas
comprising one sort of short-lived negative ion and an
arbitrary number of sorts of long-lived ones reduces to

ion
=
d
+
1
+
12

2
/
21
.
That is, the loss of electrons is due to pseudo diffusion,
direct formation of long-lived ions (with the frequency

1
) and indirect formation with a frequency which is the
product of the frequency of electron attachment to form
a short-lived negative ion (
12
) and the possibility of
converting the short-lived ion into a long-lived one (
2
/
21
in the considered case when
2

21
). It should be
noted that the effect of the indirect formation of long-lived
negative ions on the breakdown in a steady eld when there
is no pseudo diffusion electron loss has been studied before
(see, for instance, Eletskii and Smirnov 1990, Wetzer and
Wen 1991).
In a gas with two species of short-lived negative ions
(k = 3), instead of equations (13) and (14) we have
p
i
= S
i
/S
and
D

=
w
2
a
2
R
1

2
+S
2
(R
3
+
23
) +S
3
(R
2
+
32
)

4
+
2
(R
2
2S) +S
2
(15)
where
S
1
= R
2
R
3

23

32
S
2
= R
3
R
1

31

13
S
3
= R
1
R
2

12

21
S = S
1
+S
2
+S
3
R = R
1
+R
2
+R
3
.
If the frequencies of reactions involving short-lived ions of
the third species are small (
13
=
31
=
23
=
32
0)
equation (15) reduces to (14). There are also other cases
when equation (15) reduces to (14). For instance, let us
assume that an electron can attach to a molecule to form the
negative ion of the second species (i = 2) which does not
participate in the electron detachment reaction, yet converts
into the negative ion of the third species (i = 3) which
is involved in the fast electron detachment reaction, i.e.

13
=
21
=
32
0. Then, from (15)
D

=
w
2
a
2

12
(
2
+
2
23
+
23

31
+
2
31
)
[
4
+
2
(
2
23
+
2
31
+
2
12
)
+(
23

31
+
12

23
+
31

12
)
2
]
1
. (16)
In the limit of fast conversion of an ion of the second
species into that of the third species (
23

31
,
12
)
equation (16) reduces to
D

=
w
2
a
2

12

2
+(
12
+
31
)
2
.
This looks like equation (14) because a large value of
23
implies that the apparent frequency of electron attachment
to form an ion of third species is
12
and the intermediate
ion of the second species can be neglected.
4. The effect of electron pseudo diffusion on RF
breakdown in real gases
We consider conditions when electron pseudo diffusion
governs RF breakdown in gases. To estimate the order
of magnitude let us take
ij
= and
i
= for any i and
j. Then, the condition that the rst term on the right-hand
side of equation (12) is much greater than the second one
reduces to


2
2
_
w
a
d
_
2
1
1 +(k/)
2
. (17)
Considering that the pseudo diffusion concept is true only
for amplitudes of electron drift oscillations which are much
less than a quarter of the distance between the electrodes
(A = w
a
/ d/4), from (17) it follows that a necessary
condition for pseudo diffusion to be important to breakdown
is


2
32
1
1 +(k/)
2
. (18)
That is, electron attachment is followed by electron
detachment almost every time without the formation of
long-lived negative ions.
Aints et al (1995) observed that the measured value of
the RF breakdown voltage in air under normal conditions
is 20% lower than that calculated from equating the
frequencies of ionization and electron attachment. This was
interpreted in terms of the total compensation for electron
attachment loss by electron detachment and, therefore, the
breakdown criterion was written as the equality between the
ionization frequency and rate of electron loss due to pseudo
diffusion. However, this approach gave a breakdown
voltage 20% lower than the measured value.
Let us analyse the effect of the electron pseudo
diffusion on RF breakdown under the conditions of the
experiment performed by Aints et al (1995). Based on the
observation of time-resolved avalanche current waveforms,
Wetzer and Wen (1991) derived the apparent rates of
electron detachment from short-lived negative ions and
conversion of short-lived negative ions into long-lived ions
in air as functions of E/N and gas pressure. These data
show that the conditions used by Aints et al correspond
to 1. A similar result ( 0.5) follows from a
study by Akishev et al (1994) of the spatial structure of a
glow discharge in atmospheric-pressure dry air; they have
shown that the O

ion can be considered as a short-lived

ion species and O

2
and O

3
ions as long-lived ion species.
The above estimates of are inconsistent with
inequality (18); that is, the electron pseudo diffusion
has little or no effect on RF breakdown in air, but,
more importantly, the concept of pseudo diffusion fails
in air because of slow electron detachment from short-
lived negative ions. This contradicts the interpretation of
experimental data by Aints et al (1995). Our estimates
1406
Comment
show that the observations by Aints et al (1995) can be
interpreted (without pseudo diffusion) in terms of electron
detachment which partially compensates for the loss of
electrons due to attachment to molecules. This effect
is described by the second term on the right-hand side
of equation (12). Unfortunately, this approach fails to
calculate the breakdown voltage in air exactly because of
the uncertainty in the rates of ionmolecule reactions.
However, the inuence of electron pseudo diffusion
on high-frequency breakdown can become more important
as the rate of electron detachment increases. This can
be observed for repetitive breakdown in air when a
substantial number of active neutral particles such as
O atoms, radicals and electronically excited molecules
are accumulated from impulse to impulse (Kozlov and
Kudravtsev 1985). Collisions between active particles
and negative ions result in fast electron detachment in air
(Kossyi et al 1992). Another way of increasing the rate of
electron detachment in air is to heat the gas to 1000 K.
This leads to fast electron detachment from the O

2
ion
(Massey 1976, Aleksandrov 1986).
High-frequency breakdown can also be controlled by
electron pseudo diffusion in N
2
O. In this case at the
breakdown voltage electrons attach to the N
2
O molecules
due to the process (Massey 1976)
e +N
2
O O

+N
2
followed by the fast processes (Parkes 1972, Meek and
Craggs 1978, Aleksandrov and Konchakov 1992)
O

+N
2
O NO

+NO
and
NO

+N
2
O e +NO +N
2
O.
The removal of the NO

ion also proceeds through the slow

(at not too high a pressure) reactions
NO

+2N
2
O N
3
O

2
+N
2
O
and
NO

+N
2
O NO

2
+N
2
.
Therefore, O

and NO

ions can be considered as short-

lived ions and N
3
O

2
and NO

2
ions as long-lived ions.
Then, the criterion of high-frequency breakdown in N
2
O
can be described by equation (12) where the pseudo
diffusion coefcient is given by equation (15).
A similar effect is anticipated to take place in gas
mixtures in which the dominant electron attachment
process results in the formation of O

if the mixture
contains components (NO, CO, H
2
, SO
2
, C
2
H
4
, etc) which
effectively destroy O

(Massey 1976, Smirnov 1982).

In the above discussion the applied electric eld was
assumed to be uniform. Aleksandrov and Okhrimovskii
(1996) have shown that fast electron attachment and
detachment processes in nonuniform elds result in the
appearance of another electron ux which is proportional
to the gradient of the eld. In the steady eld case they
derived an expression for the coefcient which describes the
new electron ux and found that it is of the same order of
magnitude as the electron pseudo diffusion coefcient. It is
clear that this additional electron transport in a nonuniform
eld, as well as pseudo diffusion, might be important
for high-frequency breakdown in gases with fast electron
attachment and detachment processes.
5. Conclusions
The criterion of breakdown is derived in a high-frequency
uniform eld in gases with fast electron attachment and
detachment processes. A regular procedure based on
perturbation theory is proposed to derive an explicit
expression for the coefcient of electron pseudo diffusion
which was introduced earlier by Aints et al (1995). We
generalize the consideration in order to account for an
arbitrary number of negative ion species including those
which are not involved in electron detachment processes.
It is shown that the concept of pseudo diffusion is not
applicable to a description of RF breakdown in air under
the conditions of the experiment performed by Aints et al;
however, it is anticipated that electron pseudo diffusion
should govern RF breakdown in air under other conditions
and in some gases and gas mixtures.
References
Aints M, Haljaste A, Kudu K and Adamson V 1995 J. Phys. D:
Appl. Phys. 28 81
Akishev Yu S, Derygin A A, Elkin N N, Kochetov I V,
Napartovich A P and Tryshkin N I 1994 Fiz. Plazmy 20 487
Aleksandrov N L 1986 Sov. Phys. Tech. Phys. 56 835
Aleksandrov N L and Konchakov A M 1992 High. Temp. 29 850
Aleksandrov N L and Napartovich A P 1986 Fiz. Plazmy 12
1101
Aleksandrov N L and Okhrimovskii A M 1995 Pisma v ZhTF 21
46
1996 Fiz. Plazmy 22 82
Eletskii A V and Smirnov B M 1990 Zh. Tekh. Fiz. 60 192
Gatland I R 1974 Case Studies Atom. Phys. 4 369
Iinuma K, Hamano T and Takebe M 1994 J. Chem. Phys. 101
2949
Iinuma K, Sasaki N and Takebe M 1993 J. Chem. Phys. 99 6907
Iinuma K and Takebe M 1992 Can. J. Phys. 70 1604
Kossyi I A, Kostinsky A Yu, Matveyev A A and Silakov V P
1992 Plasma Sources Sci. Technol. 1 207
Kozlov S I and Kudravtsev V P 1985 Kosmich. Issledovaniya 23
426
MacDonald A D 1966 Microwave Breakdown in Gases (New
York: Wiley)
Massey H S W 1976 Negative Ions (Cambridge: Cambridge
University Press)
Matulionis A and Starikov E 1981 Hot Electron Diffusion ed J
Pozhela (Vilnius: Mokslas) p 35 (in Russian)
Meek J M and Craggs J D 1978 Electrical Breakdown in Gases
(New York: Wiley)
Ohmi T and Hasuo S 1969 Electron. Commun. Japan 52C 423
Parkes D A 1978 J. Chem. Soc. Faraday Trans. I 11 2103
Raizer Yu P 1991 Gas Discharge Physics (Berlin: Springer)
Shockley W, Copeland J A and James R P 1966 Quantum
Theory of Atoms, Molecules and the Solid State (New York:
Academic)
Smirnov B M 1992 Negative Ions (New York: McGraw-Hill)
Wetzer J M and Wen C 1991 J. Phys. D: Appl. Phys. 24 1964
1407