The simplest equivalent circuit of a pulsed dielectric barrier discharge and

the determination of the gas gap charge transfer

A. V. Pipa, J. Koskulics, R. Brandenburg, and T. Hoder

Citation: Rev. Sci. Instrum. 83, 115112 (2012); doi: 10.1063/1.4767637

View online: http://dx.doi.org/10.1063/1.4767637
View Table of Contents: http://rsi.aip.org/resource/1/RSINAK/v83/i11
Published by the American Institute of Physics.

Related Articles
Study of fast electron jet produced from interaction of intense laser beam with solid target at oblique incidence
Phys. Plasmas 19, 113118 (2012)
Raman shifted third harmonic generation of upper hybrid radiation in a plasma
Phys. Plasmas 19, 112116 (2012)
Propagation of radiation in fluctuating multiscale plasmas. II. Kinetic simulations
Phys. Plasmas 19, 113304 (2012)
L- and M-shell absorption measurements of radiatively heated Fe plasma
Phys. Plasmas 19, 113302 (2012)
Modification of the collective Thomson scattering radiometer in the search for parametric decay on TEXTOR
Rev. Sci. Instrum. 83, 113508 (2012)

Additional information on Rev. Sci. Instrum.

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

Downloaded 30 Nov 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
 REVIEW OF SCIENTIFIC INSTRUMENTS 83, 115112 (2012)

The simplest equivalent circuit of a pulsed dielectric barrier discharge

and the determination of the gas gap charge transfer
A. V. Pipa,1,a) J. Koskulics,2 R. Brandenburg,1 and T. Hoder1
1
Leibniz Institute for Plasma Science and Technology (INP Greifswald), Felix-Hausdorff-Str. 2,
Greifswald 17489, Germany
2
Light and Life Laboratory, Department of Physics and Engineering Physics, Stevens Institute of Technology,
Hoboken, New Jersey 07030, USA
(Received 31 August 2012; accepted 1 November 2012; published online 28 November 2012)
The concept of the simplest equivalent circuit for a dielectric barrier discharge (DBD) is critically
reviewed. It is shown that the approach is consistent with experimental data measured either in large-
scale sinusoidal-voltage driven or miniature pulse-voltage driven DBDs. An expression for the charge
transferred through the gas gap q(t) is obtained with an accurate account for the displacement current
and the values of DBD reactor capacitance. This enables (i) the significant reduction of experimen-
tal error in the determination of q(t) in pulsed DBDs, (ii) the verification of the classical electrical
theory of ozonizers about maximal transferred charge qmax , and (iii) the development of a graphical
method for the determination of qmax from charge-voltage characteristics (Q-V plots, often referred as
Lissajous figures) measured under pulsed excitation. The method of graphical presentation of qmax is
demonstrated with an example of a Q-V plot measured under pulsed excitation. The relations between
the discharge current jR (t), the transferred charge q(t), and the measurable parameters are presented
in new forms, which enable the qualitative interpretation of the measured current and voltage wave-
forms without the knowledge about the value of the dielectric barrier capacitance Cd . Whereas for
quantitative evaluation of electrical measurements, the accurate estimation of the Cd is important.
© 2012 American Institute of Physics. [http://dx.doi.org/10.1063/1.4767637]

I. INTRODUCTION that the charge transfer through the gas gap is larger then the
measured charge, that is in contradiction to Ref. 5.
Dielectric barrier discharges (DBDs) have been studied
The aim of the present work is to review the concept of
extensively due to numerous applications.1–3 The electrical
the simplest equivalent circuit to resolve the apparent contra-
theory of DBDs was developed for large-scale sinusoidal-
diction with the classical theory, and to suggest an approach
driven industrial ozonizers.1–6 In these reactors, measured
for the determination of the gas gap charge transfer.
charge-voltage characteristics, Q-V plots, appear as a parallel-
ogram and the maximum charge transferred through the gas
gap, qmax , can be easily deduced from a Q-V plot as the vari-
ation of measured charge during the discharge period.5 II. SIMPLEST EQUIVALENT CIRCUIT
The form of Q-V plots for DBDs under short-pulsed ex-
citation is significantly different from classical parallelogram A. Derivation of the equivalent circuit
and has not been used for qmax determination. The limited All DBD reactors contain three main parts: metallic elec-
applicability of the classical theory for miniature laboratory trodes where the applied voltage, current, and charge can be
DBDs driven by fast pulsed voltages was discussed, for ex- measured; dielectric barriers that prevent the flow of the con-
ample, in Ref. 7. ducted current through the reactor; and the gas gap where
A simplified model of a DBD reactor by a lumped- the discharge takes place, see Fig. 1. The discharge current
element equivalent circuit has been found to be useful for through the gas gap and the gas gap voltage Ug (t) cannot be
the electrical characterization in numerous numbers of works, measured directly.
see for example, Refs. 8–17. This approach can be applied to An equivalent circuit model of a DBD discharge can
pulsed DBDs as well as for sinusoidal-voltage driven DBDs be used to infer discharge properties from measured volt-
to determine such instantaneous discharge characteristics as age V (t) and current i(t) waveforms. The choice of the sim-
the gas gap voltage, Ug (t), and the discharge current, jR (t), plest equivalent circuit can be guided by experimental charge-
from voltage, V (t), and current, i(t), waveforms measured at voltage characteristics or Q-V plots. Let us consider, for ex-
the external terminals of the discharge arrangement. The re- ample, the classical Q-V plots which are measured in large-
view of the simplest equivalent circuit and the final equations scale sinusoidal-voltage driven reactors. They appear as a
reveal some discrepancies and uncertainties. For example, the parallelogram,1–5 schematically shown in Fig. 2. The varia-
opportunity to deduce the charge transferred through the gas tion of the charge on a capacitor is proportional to the varia-
gap was analyzed in Refs. 10 and 17. There it was concluded tion of voltage C = dQ/dV . The appearance of two distinct
slopes on the Q-V plot indicates the two values of effective
a) Electronic mail: [email protected]. discharge capacitance.

0034-6748/2012/83(11)/115112/7/$30.00 83, 115112-1 © 2012 American Institute of Physics

Downloaded 30 Nov 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
115112-2 Pipa et al. Rev. Sci. Instrum. 83, 115112 (2012)

FIG. 1. Schematic presentation of dielectric barrier discharge cell.

Due to dielectric barriers, DBD reactors necessarily have

capacitive properties that may be reconciled with the effec-
tive values indicated by the Q-V plot. According to the tradi-
tional interpretation of the DBD capacitance alternation, the
capacitance of the discharge reactor is determined only by the
FIG. 3. Simplest equivalent circuit of a DBD.
capacitance of the dielectric barriers Cd , when the discharge
channel bridges the gas gap; whereas in the absence of dis-
charge, the DBD capacitance is significantly lower and equal the measured charge during the active discharge phase. The
to reactor cell capacitance Ccell . This smaller capacitance Ccell segment “c” represents the variation of the applied voltage in
is calculated as a series connection of gas gap Cg and dielec- the active phase. The segment “a” is the difference between
tric barrier Cd capacitances: qmax and Q0 . Taking into account that the slope of B-C is Cd ,
Cd Cg Eq. (2), and that the slope of A-B is Ccell , Eq. (3), it is possible
Ccell = . (1) to derive the following relations from Fig. 2:
C d + Cg
In the active discharge phase, corresponding to the sides B-C qmax qmax − Q0
= Cd and = Ccell , (5)
and D-A in the parallelogram in Fig. 2, the classical Q-V plot c c
can be described by the following equation:4, 5 expressing qmax from relations (5) Eq. (4) can be obtained.
 
Q(t) = Cd V (t) ± Ug (t) , (2) The dielectric layers are commonly treated as ideal ca-
pacitors. For simplicity, the effective capacitance of both the
where the gas gap voltage remains constant and equal to the dielectrics is considered as a single lumped element Cd equal
effective breakdown voltage Ug (t) = Ub . In the dark phase, to the serial connection of two single layer capacitors. In his
A-B and C-D sides of the parallelogram can be described by work (Ref. 4) Manley concludes, “To treat the air space as
the equation:7 an ohmic resistance led to no useful result.” It is clear that
Q0 a variable ohmic resistance would not explain the change in
Q(t) ± = Ccell V (t), (3) the effective capacitance of the discharge reactor. A variable
2
capacitance is also not suitable, because it will not describe
where Q0 is related to the maximum charge transferred
the non-zero gas gap voltage Ug (t) during the active phase.
through the gas gap qmax as follows:
  The parallel connection of a variable resistor R and the gas
Ccell gap capacitor Cg , see Fig. 3, is the simplest equivalent circuit
Q0 = 1 − qmax . (4)
Cd that can be used for the analysis of the electrical properties of
DBDs.
In the previous paper (Ref. 7) we mentioned that When the applied voltage rises to the value such that
Eq. (4) can be deduced graphically from a classical Q-V plot. Ug (t) = Ub , the resistance of R falls to a certain value (larger
The present work is focused on the determination of qmax and then zero) and the charge transfer through the gas gap begins.
relation (4) is essential. Thus, Fig. 2 is specially prepared According to the equivalent circuit in Fig. 3, the effective re-
for the demonstration of relation (4). The value of qmax is actor capacitance is Cd . The voltage on the capacitance Cd
marked in accordance to classical theory5 as a variation of is lower then applied voltage V (t) by the value Ug (t). Equa-
tion (2) can be directly written from the equivalent circuit.
The sum of the charges on capacitance plates Cg and Cd con-
nected by the internal node is equal to the charge transferred
through the resistance R. After the positively applied voltage
peak the charge stored on the internal plates of Cg and Cd is
+qmax /2. After the negative voltage peak it is −qmax /2. Conse-
quently, the maximum charge transferred over one half period
is qmax .
When the gas gap voltage is low that Ug (t) < Ub , R has an
infinite resistance and there is no charge transfer through the
gas gap. The reactor capacitance is thus equal to Ccell in accor-
dance to the equivalent circuit in Fig. 3. At this moment Ug (t)
FIG. 2. Schematic presentation of Q-V plot from classical electrical theory
depends on the applied voltage V (t) and on the charge trans-
of ozonizers. ferred through the resistance R in the previous cycle qmax /2.

Downloaded 30 Nov 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
115112-3 Pipa et al. Rev. Sci. Instrum. 83, 115112 (2012)

A part of the transferred charge is accumulated on the capac- and D-A sides of the classical parallelogram in Fig. 2 is due
itance Cd : to a large number of repetitive gas gap breakdowns. The small
  decrease of Ug (t) due to a single breakdown is quickly com-
Ccell qmax
q˜d = 1 − (6) pensated by continuous increase of the applied voltage V (t)
Cd 2
and thus Ug (t) can be assumed to be constant.
and the other part on capacitance Cg : Finally, it can be concluded that the simplest equiva-
Ccell qmax lent circuit can qualitatively describe the main peculiarities of
q˜g = . (7) electrical measurements of DBDs. The description by the sim-
Cd 2
plest equivalent circuit should be applicable for either clas-
The sum of the charges on Cg and Cd capacitance plates con- sical ozonizer reactor or laboratory pulsed DBDs. This ap-
nected by the internal node remains equal to the value qmax /2 proach is widely used for DBD diagnostics but the details of
until a new charge transfer occurs through the gas gap. The the circuit are sometimes presented differently.
distribution of the charge, Eqs. (6) and (7), can be obtained
in the time t = τ , when V (τ ) = 0, (the charge should be re-
distributed between capacitance Cd and Cg in order to keep
Ug (τ ) + Ud (τ ) = V(τ ) = 0). Then the instantaneous charge C. Relevance of the simplest equivalent circuit
on the capacitance Cd can be presented as the superposition The presentation of the gas gap as parallel connection of
of the charge induced by the external voltage and stored on R and Cg was also used in Refs. 8, 15, and 16. Two elements
the dielectric barriers: of Cd in the equivalent circuit presented in Ref. 8 seem to be
qd (t) = Ccell V (t) ± q˜d . (8) a misprint, because the text of the article assumes only one
element Cd , as in Fig. 3. Zener diodes2 can be considered as
However, the charge qg (t) on the capacitance Cg cannot be a specific case of variable resistor with denoted rules and pro-
measured. Even in asymmetrical DBDs with one barrier, the vide the simplest circuit model for ozonizer reactor where the
charge measurements on the gap node will be disturbed by gas gap voltage is constant Ug (t) = Ub during the active phase.
discharge current jR (t). The measured charge represents the For pulsed DBDs the resistor rules should be more complex.
charge on capacitance Cd , Q(t) = qd (t). Comparing Eqs. (3) Variable impedance12 is the most general consideration, but
and (8) one may see that q˜d = Q0 /2 and that Eqs. (4) and (6) the motivation of including capacitive and inductive proper-
demonstrate the same relation between measured charge and ties is unclear. The introduction of a current source9, 10 con-
charge stored on the dielectric barriers. Therefore, the equiv- flicts with the fact, that a current source is an active element
alent circuit is in agreement with the passive part of Q-V plot which can deposit energy into the electrical circuit. Neverthe-
as well as with the active part. The presence of qmax /2 on di- less, the change of the current source to a variable resistor
electric surfaces is seen by external charge measurements as will not make an influence on the consideration in Refs. 9 and
Q0 /2 in Eq. (3) and Fig. 2. 10. Actually R in Fig. 3 can be considered as a “black box,”
whose properties are of interest.
The simplest equivalent circuit necessarily assumes a sin-
B. Additional comments on pulsed DBD
gle potential on the internal node. Comparing Figs. 1 and 3
The shape of Q-V plots measured for fast pulsed-driven one can see that the internal node corresponds to the entire
DBDs is significantly different from the classical parallelo- internal surface of the dielectric barriers. The potential on the
gram. The main reason is that for pulsed excitation the gas dielectric surface is determined by the charge distribution. So,
gap voltage Ug (t) is not constant during the active discharge the potential on the internal node and the gas gap voltage Ug (t)
phase.7 The variation of Ug (t) during gas gap conduction leads are effective values which are averaged over the electrode sur-
to a complex form of the Q-V plot. It was shown that the faces. The real charge distribution can not be accounted unless
plot of maximum measured charge as a function of the max- additional complexity is introduced. The limitation of this as-
imum voltage (Qmax − Vmax plot) indicates the alternation of sumption was not yet explored or discussed. The relevance
the effective discharge capacitance analogous to the classical of the equivalent circuit description for the determination of
theory.7 In case of the pulsed-driven DBDs, Vmax represents electric field in the gas gap is unclear yet.18
the value of the applied voltage at which the maximum charge The charge distribution on the dielectric surfaces can play
transfer is reached Vmax = V (Qmax ). Thus, the active phase of an important role in DBD formation. Appearance of the self-
pulsed DBDs can be described by Eq. (2) if the time depen- organized patterns in DBDs19–21 was explained by the pres-
dence of Ug (t) is taken into account. ence of non-homogeneous charge distribution over the dielec-
In case of large scale reactors driven with the sinusoidal tric surfaces. Surface breakdown in traditional volume DBD
voltage, the discharge consists of a large number of microdis- configurations can be responsible for time scattering of the
charges or current filaments.1–3 The charge is transferred current pulses,22 which is a sign for the complex processes
through the gas gap in small portions during the slow increase on the dielectric surfaces. The pattern of residual charges can
of applied voltage V (t). Fluctuations of Ug (t) could not be stay for a long time on the dielectric surface, sufficient for
seen due to the very small influence of a single microdis- scanning by an electrostatic probe, see for example, Ref. 23.
charge on Ug (t) as it is an average value over the large surface Nevertheless, the simplest equivalent circuit enables the inter-
of the reactor electrodes. Low bandwidth of the measurement pretation of the measured electrical characteristics in a sim-
equipment would smooth the measurements as well. So, B-C ple way for a reliable comparison of the characteristics of

Downloaded 30 Nov 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
115112-4 Pipa et al. Rev. Sci. Instrum. 83, 115112 (2012)

different DBDs. In the present work, the selection of the Equation (15) has two advantages. First, it expresses the
equivalent circuit is based on the description of empirical Q-V discharge current through the values that can be directly mea-
plots. sured in an experiment, whereas Eq. (14) contained Cg , which
In Ref. 24 the gas gap voltage in a homogeneous helium evaluation from Cd and Ccell values, see Eq. (1), would be as-
discharge was determined in two ways: (i) the integral of the sociated with a significant increase of the experimental error.
electric field over the gas gap (the electric field was deter- Second, it indicates that the difference of the measured current
mined by optical emission spectroscopy) and (ii) evaluated and scaled voltage derivative is proportional to the discharge
from the measured current and voltage waveforms under the current. The difference is easy to see when the measured cur-
assumption of an equivalent circuit. Both the values are in rent and scaled voltage derivative are plotted together as in
agreement. Thus, in case of uniform discharge the parameter Ref. 7, that enables qualitative estimation of the jR (t) without
Ug (t) has a transparent physical meaning. knowledge of the Cd .
The displacement current component Ccell dV (t)/dt in
III. GAS GAP VOLTAGE AND DISCHARGE CURRENT Eq. (15) (in contrast to Cg dV (t)/dt in Eq. (14)) can be mea-
sured without discharge ignition for the same form of the ap-
The values of applied voltage V (t), current i(t), and plied voltage V (t)25, 26 or can be intuitively estimated espe-
charge Q(t) on the reactor electrodes (Q(t) = qd (t)) and the cially in sinusoidal-voltage driven DBDs.27 Nevertheless, it
capacitances Ccell and Cd can be measured.7 The measured is necessary to account the factor (1 − Ccell /Cd )−1 , which was
values can be related to the parameters of the equivalent cir- omitted in Refs. 25–27. In certain discharge geometry, the dis-
cuit such as the gas gap voltage Ug (t) and the discharge cur- regard of the factor can bring a significant error. For example
rent jR (t) by following the ideas presented in Refs. 8–10. Four in Ref. 25 the ratio Ccell /Cd is about 0.5 and the discharge cur-
coupled equations based on the definition of capacitance and rent jR (t) is underestimated by a factor 2, however, the charge
Kirchhoff’s laws can be written from Fig. 3: values were evaluated correctly. The works (Refs. 26 and 27)
Q(t) do not provide the capacitance values and, therefore, the error
Ud (t) = , (9) could not be estimated.
Cd
Obtained Ug (t) and jR (t) can be used for the determination
of the instantaneous power P(t) and energy E(t) coupled into
Ug (t) = V (t) − Ud (t), (10) discharge:
P (t) = jR (t)Ug (t), (16)
dUg (t)
jg (t) = Cg , (11)
dt
 t
E(t) = P (τ )dτ. (17)
jR = i(t) − jg (t), (12) 0

where Ud (t) is the voltage across dielectric barriers and jg (t) The presented consideration of the equivalent circuit in-
is the current through the gas gap capacitance Cg . Substituting cludes only energy losses in the discharge and disregards all
Eq. (9) into (10) gives an expression for the gas gap voltage: other types of dissipation, e.g., losses in the dielectric bar-
Q(t) riers or radiation into open space. In the framework of this
Ug (t) = V (t) − . (13)
Cd approach, the electrical energy delivered into a reactor cell
over one period (T) is equal to the energy dissipated in the
Substituting Eq. (13) into (11), and taking into account discharge. It can be expressed as follows:
that the derivative of the measured charge is the measured  T  T
current dQ(t)/dt = i(t) , followed by substitution of Eq. (11) i(τ )V (τ )dτ = jR (τ )Ug (τ )dτ. (18)
into (12), an expression for the discharge current jR (t) can be 0 0
obtained after a simple rearrangement in the form suggested Relation (18) can be proved analytically for any periodic sig-
in Refs. 8 and 9: nals where Q(0) = Q(T) and V (0) = V (T ), see Appendix.
 
Cg dV (t) The energy dissipated in the discharge over one period can be
jR (t) = 1 + i(t) − Cg . (14) obtained by the integration of the Q-V plot too. This does not
Cd dt
require either knowledge of the capacitance values or detailed
Lomaev8 has assumed Cg /Cd 1 and omitted the fac- consideration of the equivalent circuit.
tor in brackets of Eq. (14). Massines et al.12 have also sug-
gested the equation for jR (t) without this factor, whereas Liu
IV. CHARGE TRANSFERRED THROUGH THE GAS
and Neiger9 intently noted that this factor is indeed important.
GAP
Our point of view is that the simplification in Refs. 8 and 12
does not have a practical justification. Nevertheless, we prefer The instantaneous charge transferred through the gas gap
to rewrite Eq. (14) in another form: q(t) can be obtained by the integration of Eq. (15):
 
1 dV (t) 1
jR (t) = i(t) − Ccell . (15) q(t) = [Q(t) − Ccell V (t)] + q0 . (19)
1 − CCcell
d
dt 1− Ccell
Cd

Downloaded 30 Nov 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
115112-5 Pipa et al. Rev. Sci. Instrum. 83, 115112 (2012)

Equation (19) expresses q(t) through the measured values of V. qmax FOR OZONIZER DISCHARGE
charge on the reactor electrodes Q(t), applied voltage V (t),
First we will use Eq. (19) for the evaluation of the max-
and measurable capacitances Ccell and Cd . The integration
imum charge transferred through the gas gap qmax in case of
constant q0 can be found if q(t) is known at zero time.
classical Q-V plots and verify the conclusion in Ref. 5. Ver-
For pulsed-driven DBDs with sufficiently long time be-
tices of the parallelogram A, B, C, and D in Fig. 2 define the
tween voltage pulses the zero time can be selected such as
moments at which these points on Q-V plot have been mea-
V (0) = Q(0) = q(0) = q0 = 0. In the case of large scale re-
sured. The difference of the instantaneous transferred charges
actors driven by the sinusoidal voltage with measured clas-
at moments C and B can be expressed using Eq. (19):
sical Q-V plot as shown in Fig. 2, taking into account dis-
cussions in Sec. II, can be asserted the following. If the zero
q(C) − q(B)
time is selected in the way that V (0) = 0 then Q(0) = ±Q0 /2;
assume q(0) = 0 and substitute these values to Eq. (19) will 1
= [Q(C) − Q(B) − Ccell (V (C) − V (B))].
obtain q0 = ∓qmax /2 through Eq. (4). 1− Ccell
Cd
In Refs. 25 and 27 the transferred charge q(t) was ob-
tained as the integral of the discharge current. The discharge (21)
current was determined accounting the displacement compo-
nent but disregarding the factor (1 − Ccell /Cd )−1 . However, The variation of the charge due to the displacement current
the authors of Ref. 25 include this factor to the final charge Ccell (V (C) − V (B)) = a is noted in Fig. 2. The difference be-
determination in contrast to Ref. 27. tween the variation of the total measured charge Q(C) − Q(B)
The relation between the transferred and measured and “a” is equal to Q0 , see Fig. 2. Thus, it is clearly seen that
charges reported in Ref. 17 is in contradiction to Eq. (19). the value in the square brackets in Eq. (21) is Q0 . Comparing
The authors of Ref. 17 came to the same relation between Eqs. (4) and (21) one may see that q(C) − q(B) = qmax .
jR (t) and measured parameters as shown in Eqs. (14) or (15). If Eq. (19) is written for the vertices C and A one got:
Then they neglect the term for the derivative of applied volt-
age. Thus they come to the conclusion that the ratio q(t)/Q(t) q(C)−q(A)
is only a function of the DBD reactor capacitance. Only if
1
V (t) and q0 are set at zero in Eq. (19) the equation suggested = [Q(C)−Q(A) − Ccell (V(C) − V(A))].
in Ref. 17 will be derived. 1− Ccell
Cd
Ignoring the displacement component, authors (Ref. 10) (22)
come to the same confusing conclusion that internal charge
transfer is larger then external charge transfer by a factor of
The value Ccell (V (C) − V (A)) = b is also noted in Fig. 2 and
(1 − Ccell /Cd )−1 . However, this conclusion was not used for
it is obvious that the term in the square brackets in Eq. (22) is
the charge determination. They expressed qmax through the
again Q0 . This is in agreement with the statement of classical
gas gap voltages in moments of discharge ignition and extinc-
theory that A-B side of the Q-V plot corresponds to the phase
tion. How one can determine these moments experimentally
when no charge transfer through the gas gap is occurring.
was not mentioned.
Therefore, for the classical Q-V plots we confirm the state-
Equation (19) derived in present work has principle pe-
ment of Ref. 5 that the variation of the measured charge Q(t)
culiarities: (i) it is written for instantaneous charge q(t) rather
during the discharge period is equal to the maximum charge
then for qmax ; (ii) it expresses the q(t) through measured pa-
transferred through the gas gap qmax :
rameters in one step; (iii) accounts accurately for the displace-
ment current and for the DBD capacitance; and (iv) it does not
qmax = q(C) − q(B) = q(C) − q(A) = Q(C) − Q(B).
require any additional assumptions except the equivalent cir-
(23)
cuit itself.
Additional conclusions can be made if Eq. (19) is rewrit-
ten in a slightly modified form:
VI. qmax FOR PULSED DRIVEN DBD
 
Ccell Q(t) The determination of qmax requires the knowledge of the
q(t) = − V (t) + q0 . (20)
1 − CCcell Ccell
d
value of Cd . The classical Q-V plot, Fig. 2, gives informa-
tion about Cd and thus qmax can be easily denoted graphically.
Here in brackets the applied voltage is compared with scaled In case of Q-V plots measured under pulsed excitation the
measured charge. Their difference is proportional to the trans- statement of Ref. 5 fails. Equation (19) helps to clarify this
ferred charge. Thus qualitative conclusion about q(t) can be question and to evaluate graphically the qmax for pulse excited
done immediately from the figure, where applied voltage and DBD.
scaled charge are displayed together. Such kind of figure Figure 4 shows the Q-V plot measured in Ref. 7 in a
can be obtained directly from measured current and voltage coaxial barrier discharge reactor in argon at 100 mbar driven
waveforms.7 It can be used complementary to the compari- by square voltage pulses with a rise time of 20 ns. Taking
son of the current with the scaled voltage derivative, Eq. (15), into account that Q(A) = 0, Q(C) = Qmax , V (A) = 0, and
for visual representation and qualitative interpretation of the V (C) = Vmax the charge transferred through the gas gap be-
experimental data. tween time moments A and C can be written on the basis of

Downloaded 30 Nov 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
115112-6 Pipa et al. Rev. Sci. Instrum. 83, 115112 (2012)

lent circuit. The expression obtained for charge transferred

through the gas gap is in agreement with the classical theory5
and enables an accurate determination of the instantaneous
charge transfer through the gas gap in case of pulsed-driven
DBDs. The method of graphical evaluation of maximum
transferred charge from a Q-V plot measured in pulsed DBD
is presented for the first time.
It was noted that the product of the total reactor capac-
itance and derivative of the applied voltage Ccell dV (t)/dt,
which represents the displacement current in absence of
the discharge, has physical meaning after discharge ignition
too. Namely, the difference between measured current and
Ccell dV (t)/dt is proportional to the discharge current. Thus,
the qualitative estimation of the discharge current can be
made without knowledge about Cd . This has a practical sig-
FIG. 4. Q-V plot measured under pulsed excitation in Ref. 7.
nificance because: (i) the experimental determination of the
Ccell is much less challenging and (ii) sometimes the term
Eq. (19): Ccell dV (t)/dt can be measured separately. Also, it was shown
1 that the difference of the scaled measured charge Q(t)/Ccell
qmax = q(C) − q(A) = [Qmax − Ccell Vmax ]. (24) and applied voltage V (t) is proportional to the charge trans-
1− Ccell
Cd ferred through the gas gap.
For graphical presentation of qmax on Q-V plot two straight
lines should be drawn. The first one is through the point A
with the slope of Ccell and the second is through the point C ACKNOWLEDGMENTS
with the slope of Cd . The values Ccell = (2.4 ± 0.3) pF and Cd The authors are thankful to Ulrich Kogelschatz for useful
= (15.7 ± 0.4) pF were determined in Ref. 7 and the corre- discussions. The work was partly supported by Federal Min-
sponding lines are plotted in Fig. 4 together with the experi- istry of Education and Research of Germany (Grant No. FKZ
mental error (represented by the dotted lines). The cross point 03FO1072).
between the Cd and Ccell lines, the point B, defines qmax .
Indeed, this approach can be confirmed by means of
the following discussion. The value Ccell Vmax = b is marked APPENDIX: ENERGY DISSIPATED IN ONE PERIOD
in Fig. 4. The term in the square brackets of Eq. (24)
For the equivalent circuit in Fig. 3 with periodically ap-
can be defined as Q0 in analogy with Fig. 2. The varia-
plied voltage with period T, where V (0) = V (T ) and Q(0)
tion of the applied voltage between moments C and B is
= Q(T), it can be shown that
V (C) − V (B) = c, which, in contrast to classical Q-V plot,
has no transparent physical meaning but relates Q0 and qmax  T  T
through Eqs. (4) and (5) in the same way. i(τ )V (τ )dτ = jR (τ )Ug (τ )dτ. (A1)
0 0
The moment 1 in Fig. 4 was recognized as the discharge
ignition in Ref. 7. It means that the variation of the measured The product of the discharge current jR (t) and gas gap volt-
charge during the active phase is Qmax − Q(1). One may see age Ug (t), which are expressed by Eqs. (13) and (15) can be
from Fig. 4 that in contrast to the classical case the variation of written as
the measured charge during the active phase is lower then the 
1 dV (t)
transferred charge Qmax − Q(1) < qmax . On the other hand, in jR (t)Ug (t) = V(t)i(t) − Ccell V(t)
contrast to the conclusion in Ref. 10 the maximum of trans- 1 − Cd
Ccell dt
ferred charge is smaller then maximum of measured charge 
1 Ccell dV (t)
qmax < Qmax . − Q(t)i(t) + Q(t) . (A2)
Cd Cd dt
Thus Eq. (19) obtained in the present work allows an ac-
curate determination of q(t) and qmax and confirms the ap- Thus the integral of the product of jR (t) and Ug (t) can be split
proach of the simplest equivalent circuit as consistent for on a sum of the corresponding integrals. Two of them are
both, ozonizer reactor and pulsed-driven DBD. equal to zero if they are integrated over period:
 T
dV (τ ) 1 2
VII. CONCLUSION V (τ ) dτ = V (T ) − V 2 (0) = 0 (A3)
0 dτ 2
The description of DBDs by means of a simplest equiv-
alent circuit has been reviewed. The present work empha-  T  T
sizes that the main peculiarities of electrical characteristics dQ(τ )
Q(τ )i(τ )dτ = Q(τ ) dτ
of DBDs measured in either large-scale sinusoidal-voltage 0 0 dτ
driven reactors or miniature laboratory reactors under short 1
pulse excitation can be described in terms of this equiva- = Q2 (T ) − Q2 (0) = 0. (A4)
2

Downloaded 30 Nov 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
115112-7 Pipa et al. Rev. Sci. Instrum. 83, 115112 (2012)

One of them can be integrated by parts: 10 S. Liu and M. Neiger, J. Phys. D: Appl. Phys. 36, 3144–3150 (2003).
 T 11 F. Massines, A. Rabehi, P. Decomps, R. B. Gadri, P. Segur, and C. Mayoux,
dV (τ ) J. Appl. Phys. 83, 2950–2957 (1998).
Q(τ ) dτ = [Q(T )V (T ) − Q(0)V (0)] 12 F. Massines, N. Gherardi, N. Naude, and P. Segur, Plasma Phys. Controlled
0 dτ
Fusion 47, B577–B588 (2005).
 T  T 13 N. Naude, J.-P. Cambronne, N. Gherardi, and F. Massines, J. Phys. D: Appl.
− V (τ )dQ(τ ) = − V (τ )i(τ )dτ. (A5) Phys. 38, 530–538 (2005).
0 0 14 L. Mangolini, C. Anderson, J. Heberlein, and U. Kortshagen, J. Phys. D:
Integrating (A2) over period T and taking in to account ex- Appl. Phys. 37, 1021–1030 (2004).
15 G. Nirsisyan and W. G. Graham, “Characterization of a dielectric bar-
pressions (A3)–(A5) it can be obtained: rier discharge operating in an open reactor with flowing helium,” Plasma
 T Sources Sci. Technol. 13, 582–587 (2004).
16 M. Laroussi, X. Lu, V. Kolobov, and R. Arslanbekov, J. Appl. Phys. 96,
jR (τ )Ug (τ )dτ
0 3028–3030 (2004).
  T 17 J. Drimal, K. V. Kozlov, I. V. Gibalov, and V. G. Samoylovich, Czech. J.
1 Ccell Phys., Sect. B 38, 159–165 (1988).
= 1− V (τ )i(τ )dτ. (A6) 18 U. Kogelschatz, private communication (2010).
1− Ccell
Cd
Cd 0 19 L. Stollenwerk , J. G. Laven, and H.-G. Purwins, Phys. Rev. Lett. 98,

Thus, relation (A1) is confirmed. 255001 (2007).

20 E. L. Gurevich, A. L. Zanin, A. S. Moskalenko, and H.-G. Purwins, Phys.

Rev. Lett. 91, 154501 (2003).

1 U. Kogelschatz, Plasma Chem. Plasma Process. 23, 1–46 (2003). 21 W. Breazeal, K. M. Flynn, and E. G. Gwinn, Phys. Rev. E 52, 1503–1515
2 U. Kogelschatz, in Non-Equilibrium Air Plasmas at Atmospheric Pres- (1995).
sure edited by K. H. Becker et al. (Institute of Physics, Bristol, 2004), 22 Y. S. Akishev, G. Aponin, A. Balakirev, M. Grushin, V. Karalnik, A.

pp. 68–75. Petryakov, and N. Trushkin, Eur. Phys. J. D 61, 421–429 (2011).
3 H.-E. Wagner, R. Brandenburg, K. V. Kozlov, A. Sonnenfeld, P. Michel, 23 A. Kumada, S. Okabe, and K. Hidaka, J. Phys. D: Appl. Phys. 42, 095209

and J. F. Behnke, Vacuum 71, 417–436 (2003). (2009).

4 T. C. Manley, Trans. Electrochem. Soc. 84, 83–96 (1943). 24 S. S. Ivkovic, B. M. Obradovic, N. Cvetanovic, M. M. Kuraica, and J. Puric,
5 Z. Falkenstein and J. J. Coogan, J. Phys. D: Appl. Phys. 30, 817–825
“Measurement of electric field development in dielectric barrier discharge
(1997). in helium,” J. Phys. D: Appl. Phys. 42, 225206 (2009).
6 V. G. Samoilovich, I. V. Gibalov, and K. V. Kozlov, Physical Chemistry of 25 M. Kettlitz, H. Höft, T. Hoder, S. Reuter, K.-D. Weltmann, and R. Bran-

the Barrier Discharge (DVS-Verlag, Düsseldorf, 1997). denburg, J. Phys. D: Appl. Phys. 45, 245201 (2012).
7 A. V. Pipa, T. Hoder, J. Koskulics, M. Schmidt, and R. Brandenburg, Rev. 26 P. Bletzinger and B. N. Ganguly, J. Phys. D: Appl. Phys. 36, 1550–1552

Sci. Instrum. 83, 075111 (2012). (2003).

8 M. I. Lomaev, Atmos. Oceanic Opt. 14, 1005–1008 (2001). 27 P. Reichen, A. Sonnenfeld, and Ph. Rudolf von Rohr, J. Phys. D: Appl.
9 S. Liu and M. Neiger, J. Phys. D: Appl. Phys. 34, 1632–1638 (2001).
Phys. 43, 025207 (2010).

Downloaded 30 Nov 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions