Vous êtes sur la page 1sur 6

JOURNAL OF APPLIED PHYSICS 106, 013301 2009

Measurement of the electron density in a subatmospheric dielectric barrier


discharge by spectral line shape
Lifang Dong,a Yuyan Qi, Weiyuan Liu, and Weili Fan
College of Physics Science and Technology, Hebei University, Baoding 071002, China

Received 22 December 2008; accepted 8 June 2009; published online 7 July 2009
The electron density in a subatmospheric dielectric barrier discharge by using argon spectral line
shape is measured for the first time. With the gas pressure increasing in the range of 1
104 Pa 6 104 Pa, the line profiles of argon 696.54 nm are measured. An asymmetrical
deconvolution procedure is applied to separate the Gaussian and Lorentzian profile from the
measured spectral line. The gas temperature is estimated by using rotational temperature of N+2 . By
subtracting the van der Waals broadening and partial Lorentzian instrumental broadening from the
Lorentzian broadening, the Stark broadening is obtained and used to estimate the electron density.
It is found that the electron density in dielectric barrier discharge increases with the increase in gas
pressure. 2009 American Institute of Physics. DOI: 10.1063/1.3159891
I. INTRODUCTION

Dielectric barrier discharges DBDs, also referred to as


silent discharges, are characterized by the presence of at least
one insulating dielectric layer in contact with the discharge
between two planar or cylindrical electrodes connected to an
ac power supply. In recent years, it has been a subject of
great interest due to their potential industrial applications
including large-area flat plasma display panels, surface modification of polymers, reduction of pollutants, and generation
of UV and vacuum UV VUV radiation.14 In order to optimize this kind of plasmas for the industry applications, it is
necessary to know the plasma parameters, such as electron
density and electron temperature, which mainly determine
the characteristics of the discharge.
Plasma-broadened and shifted spectral lines have been
used as an important noninterfering plasma diagnostic technique. In previous work, we measured the electron density in
individual microdischarge channel by stark broadening.5,6
Balcon et al.7 measured the average electron density for the
filamentary mode in dielectric barrier discharge by H stark
broadening. However, to the best of our knowledge, there is
no report on the study of the variations of electron density
with the gas pressure in subatmospheric dielectric barrier discharge up to now.
In this paper, we measure the electron density in a subatmospheric dielectric barrier discharge by using Stark
broadening. The results show that the electron density varies
from 8.4 1014 cm3 to 1.5 1015 cm3 with the gas pressure increasing from 1 104 Pa to 5 104 Pa.
II. THEORY

In experiments, the spectral lines emitted from plasma


are subject to various broadening mechanisms including
natural broadening, Doppler broadening, instrumental broadening, and pressure broadening, which include resonance,
van der Waals, and Stark broadening. The natural broadening
a

Electronic mail: donglf@mail.hbu.edu.cn.

0021-8979/2009/1061/013301/5/$25.00

and resonance broadening are generally negligible in high


density plasma.8 Doppler broadening originates from the statistical velocity distribution of the emitting atoms. The related intensity profile follows a Gaussian distribution if the
emitting atom has a Maxwell velocity distribution. The van
der Waals broadening is caused by the dipolar interaction
between excited atom and the induced dipole from the neutral perturber, whose profile follows a Lorentzian function.
The Stark broadening is determined by electron impact
broadening and plasma ion impact broadening. The electron
impact broadening gives a symmetrical Lorentzian profile,
and the smaller contribution of ion broadening is asymmetrical in nature. Thus the Stark profile is the combination of
these two broadenings and exhibits an asymmetrical Lorentzian profile. The apparatus induces another broadening on
the line profile, which depends upon the width of the slit on
the monochromator and the dispersion of its diffraction system. In many cases, the apparatus function can be approximated by a Voigt function composed by Gaussian profile and
Lorentzian profile.

A. Stark broadening

The Stark broadening and shift of a certain spectral line


spontaneously emitted from atoms in the plasma allows the
determination of electron density in a rapid and inexpensive
way.
The Stark effect is determined by electrons impact and
plasma ions impact. Its profile is described as an asymmetrical Lorentzian spectral line profile j,
j =

WRd
,
1 + 0 de/e 4/322

where WR represents the microfield strength distribution


function, depending upon the dimensionless parameter R that
accounts for the Debye shielding and ion-ion correlations.

106, 013301-1

2009 American Institute of Physics

013301-2

J. Appl. Phys. 106, 013301 2009

Dong et al.

The full width at half maximum FWHM of Stark


broadening t and Stark shift dt are complex functions of the
electron density Ne and electron temperature Te, expressed
as911
1/6 1/2
t = 2 1 + 1.75 104N1/4
e 1 0.068Ne Te

1016eNe .

Due to the asymmetry of plasma-broadened atom line,


the shift at the half width of the spectral line is slightly different from the one measured at the peak of line profile and
can be calculated from911
FIG. 1. Schematic diagram of the experimental setup.

1/6 1/2
dt = de 3.2 104N1/4
e e 1 0.068Ne Te

1016Ne .

In Eqs. 13, e and de are electron impact Stark


broadening and shift, respectively, is the ion-broadening
parameter, and Ne and Te are the electron density cm3 and
temperature K.

B. van der Waals broadening

Van der Waals broadening results from the dipole interaction of an excited atom with the induced dipole of a neutral
ground state atom of number density N. This is a short-range
C6 / r6 interaction. Griems9 estimation for the FWHM v can
be written12 as
R22/5Tg/3/10N,
vcm = 8.18 10122a

where
2
R2 = RU
RL2 .

R2 is the difference of the squares of coordinate vectors in


a0 units of the upper and lower level, Tg is the gas temperature, and is the atom-perturber reduced mass in a.m.u.
= 19.97 for excited Ar perturbed by Ar atoms, and N can
be obtained from the equation of the ideal gas. Values of a,
the mean atomic polarizability of the neutral perturber, are
tabulated for different elements by Allen:13 for argon a
= 16.54 1025 cm3. If the required value of a is not tabulated, it can be estimated either from the expression given by
Allen13 or by Griem:9
a = 9/2a303EH/4EEXC2 ,

where EH is the ionization potential of hydrogen


109 737.32 cm1 and EEXC is the energy cm1 of the first
excited level of the perturber. In the Coulomb approximation, the values of RU and RL in Eq. 5 may be calculated
from
1
R2j = n2
5n2
j + 1 3l jl j+1,
2 j

where the square of effective quantum number nj is


n2
j = EH/EIP E j,

and EIP is the ionization potential of the studied element and


E j is the energy of the upper or lower levels of the transition.

III. EXPERIMENTAL SETUP

The experimental device is shown in Fig. 1. Two cylindrical containers, with diameters of 65 mm, sealed with 1.5
mm thick glass plates are filled with water. A metallic ring
immerses in the water of each container and is connected to
a power supply. Thus, the water acts as a liquid electrode. A
glass frame with the thickness of 1.5 mm is placed between
the dielectric layers, serving as the lateral boundary. Thus,
the discharge gap is 1.5 mm. A sinusoidal ac voltage with a
frequency of 50 kHz is applied to the electrodes. All of the
apparatus are enclosed in a big chamber filled with argon.
The voltage applied to the electrode is measured with a Tektronix high voltage probe ratio 1:1000 connected to an oscilloscope Tektronix TDS 3054, 500 MHZ. The discharge
gas is argon with a purity of about 99.92%. Optical emission
spectra from the plasma are collected with a converging lens
and an optical fiber and detected by a monochromator ACTON SP-2758, 2400 groove/mm grating, resolution 0.01 nm
with a charge coupled device 1340 400 pixels. The
opening of the slit input of the monochromator is set to
50 m. The calibration of the instrumental function is made
with a HeNe laser 632.8 nm line and is found to be a
Lorentzian component L = 0.002 38 nm and a Gaussian
component G = 0.016 68 nm.
IV. RESULTS AND DISCUSSION

As is well known, breakdown is the most critical aspect


of the discharge, since essentially all the energy, which is put
into the plasma electrons, is delivered in this phase. It is also
the critical phase for determining what chemical reactions
occur. It is necessary to investigate the plasma parameters at
the phase of breakdown. So, the electron density in discharges at critical breakdown voltage named as critical discharge in the context below is investigated.
The critical discharges undergo two modes, a diffuse
mode and a filamentary mode. When the gas pressure is
varying in the range of 1 104 Pa 5 104 Pa, the gas between electrodes ignites in a diffuse mode, in which a plasma
completely fills the cross section of the discharge space.
However, the gas ignites in a form of filamentary mode, the
discharge no longer permeate on the whole electrodes, when
the gas pressure is increased to 6 104 Pa.
It is found that the spectral lines profiles change with the
gas pressure. Figure 2 gives the profiles of argon 696.54 nm

013301-3

J. Appl. Phys. 106, 013301 2009

Dong et al.

FIG. 2. The total profiles of Ar I 696.54 nm as a function of gas pressure.

at different gas pressures. As a reference source of unshifted


argon spectral line, a low pressure argon discharge in a small
tube at gas pressure of about 10 Pa is used. It can be clearly
seen that the broadenings and shifts of the line profile increase with the increasing of gas pressure.
In many cases, the shift of the line profile is mostly
caused by Stark shift, while the van der Waals shift is relatively small and can be negligible. Djurovic et al.14 found
that the contributions of van der Waals shift to the shift of
argon spectral line are in the range 1%9% in plasma of an
atmospheric pressure wall stabilized argon arc with electron
densities of 0.74 2.9 1022 m3 and electron temperatures of 928010750 K. So, in our experiment, the increase
in the spectral line shift indicates the increasing in the electron density with the gas pressure.
In order to get the Stark broadening, the Lorentzian profile must be known. Here we used the method for deconvolution of asymmetric line profiles.5 As is well known, the
measured line profiles are the results of convolution of
Lorentzian profile and Gaussian profile. The Lorentzian profile comprises two parts: a symmetrical part and an asymmetrical part. The symmetrical Lorentzian profile comprises
van der Waals Lorentzian and partial instrumental Lorentzian
component. The asymmetrical one is Stark broadening
Lorentzian profile. The Gaussian type induces the Doppler
broadening and a majority of instrumental broadening. Thus,
the total spectral line profile K is the convolution of
Gaussian G and Lorentzian profile L, described by
K =

G yLydy =

exp
dy,

2y 2
2
G

G/2

WRd
1 + y 0 d/w 4/322
9

where G, w, d, and are fitted to the experiment data.


We compiled a deconvolution procedure to separate L
profile from the measured spectral lines. The Lorentzian
broadening thus can be obtained. Figure 3 gives a deconvolution result for Ar I 696.54 nm of discharge at the gas pressure of 5 104 Pa.
The Stark broadening can be obtained after subtracting
the calculated van der Waals broadening from the gas temperature and measured partial instrumental broadening from
the deconvolved Lorentzian profiles.

FIG. 3. Typical deconvolution result for Ar I 696.54 nm in argon discharge


at the pressure of 5 104 Pa. C-convolution profile, G-Gaussian profile,
L-Lorentzian profile.

For the argon plasma in our experiment with argon as the


perturber, the reduced mass is equal to 19.97 and the parameter a is 16.54 1025 cm3. Accordingly, the Eq. 4 can
be written in terms of the gas temperature as

v696.54 =

1.52 a
nm,
Tg0.7

10

where a is a ratio of discharge gas pressure to atmospheric


pressure.
The gas temperature in discharge is usually estimated
from analysis of the rotational spectra of molecular species
present in the plasma.1517 However, it is no longer suitable
for pure argon discharge. For estimating the gas temperature,
the rotational temperature in air/argon mixture discharge are
studied by analyzing the first negative band of N+2 and shown
in Fig. 4. It is found that the gas temperature varies from 420
to 460 K with the gas pressure changing from 1 104 Pa to
5 104 Pa
Figure 5 shows the variations of Lorentzian widths, van
der Waals broadenings, Stark broadening, and the electron
density with gas pressure increasing, respectively. It reveals
that the electron density in the discharge domain increase
from 8.4 1014 cm3 to 1.5 1015 cm3 with gas pressure
increasing from 1 104 Pa to 5 104 Pa. The relative error
of the electron density is estimated to be about 15% by considering the error induced by deconvolution method and the
experimental errors. The results prove the electron density in
dielectric barrier discharge is lower than that in dc glow discharge measured by Penache et al.,18 which increases from
9 1014 cm3 to 5 1015 cm3 when the gas pressure
changes from 5 103 Pa to 4 104 Pa.
It is worth pointing out that the electron density obtained
here is much higher than that estimated by discharge current
or power balance.7 In order to explain the discrepancy, another experiment was carried out in dielectric barrier discharge in argon at pressure in the range of 1 104 5
104 Pa. In the experiment, the light emission from total
discharge area and light emission of a small area of 0.25 mm
in diameter were measured by two photomultiplier tubes
RCA7265, respectively, and recorded by an oscilloscope

013301-4

Dong et al.

J. Appl. Phys. 106, 013301 2009

FIG. 5. Variations of the Lorentzian broadening, the Stark broadening, the


van der Waals broadening of argon 696.54 nm spectral line, and the electron
density with the increase of gas pressure.

FIG. 4. The measurement of gas temperature in dielectric barrier discharge


by emission spectrum. a The spectrum of the first negative band of N+2 in
dielectric barrier discharge at gas pressure of 2 104 Pa. b The rotational
temperature is estimated by the slope of lnI / J + J + 1 vs JJ + 1. c
The gas temperature in dielectric barrier discharge with the gas pressure
increasing from 1 104 to 7 104 Pa in different argon concentrations in
gas mixture.

Tektronix TDS 3054, 500 MHZ. The waveforms of voltage


and discharge current were also recorded. From Fig. 6, it can
be clearly seen that the discharge in the small area is not
ignited every half cycle of the applied voltage. In other
words, the discharge actually does not filled the entire area at
any time although the image of discharge exposed over a
long time many half cycles is homogeneous. The actual
discharge area here in each half cycle is much less than the
exposure area in image, which is generally used in the electron density estimation by discharge current or power balance. The electron density will be much less than the true
value if discharge area is overestimated. On the other hand,

FIG. 6. Color online a Image of diffuse discharge at 1 104 Pa. Discharge area is 30 30 mm2. Exposure time is 66.7 ms. b From top to
bottom the curves are waveforms of voltage, current, light from total discharge area, and light from a small area of 0.25 mm in diameter,
respectively.

013301-5

J. Appl. Phys. 106, 013301 2009

Dong et al.

the spectral line profile should vary with time in each discharge current because the electron density changes with
time. A narrow profile corresponds to small electron density
while a wide profile corresponds to a large electron density.
It is obvious that the recorded spectral line profile is a temporal integral and the widest profile is usually recorded. Thus
the electron density estimated by Stark broadening should be
the peak value of the electron density.
V. CONCLUSION

In this work, the electron density in a subatmospheric


DBD is estimated by using the Stark broadening of atomic
spectral line. With the gas pressure increasing in the range of
1 104 Pa 6 104 Pa, the line profiles of argon 696.54 nm
are measured. An asymmetrical deconvolution procedure is
applied to separate the Gaussian and Lorentzian profiles
from the measured spectral line. The gas temperature is estimated by using rotational temperature of N+2 . By subtracting the van der Waals broadening and partial Lorentzian instrumental broadening from the Lorentzian broadening, the
Stark broadening is obtained. It is found that the electron
density in dielectric barrier discharge increases with the increase in gas pressure.
ACKNOWLEDGMENTS

This work is supported by the Natural Science Foundation of China under Grant Nos. 10575027 and 10775037, the
Specialized Research Fund for the Doctoral Program of
Higher Education of China Grant No. 20050075001, and

the Natural Science Foundation of Hebei Province, China


Grant Nos. A2006000950 and A 2008000564.
1

H. S. Uhm, E. H. Choi, G. Cho, and K. Whang, J. Plasma Phys. 67, 49


2002.
2
B. Eliasson and U. Kogelschatz, IEEE Trans. Plasma Sci. 19, 309 1991.
3
Z. Fang, Y. Qiu, and Y. Luo, J. Phys. D: Appl. Phys. 36, 2980 2003.
4
S. K. Dhali and I. Sardja, J. Appl. Phys. 69, 6319 1991.
5
L. F. Dong, J. X. Ran, and Z. G. Mao, Appl. Phys. Lett. 86, 161501
2005.
6
L. F. Dong, Y. Y. Qi, Z. C. Zhao, and Y. H. Li, Plasma Sources Sci.
Technol. 17, 015015 2008.
7
N. Balcon, A. Aanesland, and R. Boswell, Plasma Sources Sci. Technol.
16, 217 2007.
8
S. Pellerin, K. Musiol, B. Pokrzywka, and J. Chapelle, J. Phys. B 29, 3911
1996.
9
H. R. Griem, Plasma Spectroscopy McGraw-Hill, New York, 1964, p.
492.
10
R. Konjevic, Plasma Sources Sci. Technol. 10, 356 2001.
11
D. Nikoli, S. Djurovi, Z. Mijatovi, R. Kobilarov, B. Vujii, and M.
irian, J. Quant. Spectrosc. Radiat. Transf. 86, 285 2004.
12
D. E. Kelleher, J. Quant. Spectrosc. Radiat. Transf. 25, 191 1981.
13
C. W. Allen, Astrophysical Quantities The Athlone Press, London, 1964
p. 87.
14
S. Djurovi, D. Nikoli, Z. Mijatovi, R. Kobilarov, and N. Konjevi,
Plasma Sources Sci. Technol. 11, A95 2002.
15
X. Duten, A. Rousseau, A. Gicquel, K. Hassouni, and P. Leprince, J. Phys.
D: Appl. Phys. 35, 1939 2002.
16
B. A. Cruden, M. V. V. S. Rao, S. P. Sharma, and M. Meyyappan, Appl.
Phys. Lett. 81, 990 2002.
17
Q. Wang, F. Doll, V. Donnelly, D. J. Economou, N. Sadeghi, and G. F.
Franz, J. Phys. D: Appl. Phys. 40, 4202 2007.
18
C. Penache, M. Miclea, A. Bruning-Demian, O. Hohn, S. Schssler, T.
Jahnke, K. Niemax, and H. Schmidt-Bcking, Plasma Sources Sci. Technol. 11, 476 2002.

Vous aimerez peut-être aussi