Eur. Phys. J. Appl. Phys.

47, 22805 (2009)

DOI: 10.1051/epjap/2009064 THE EUROPEAN
PHYSICAL JOURNAL
APPLIED PHYSICS
Regular Article

Recent advances in the understanding of homogeneous dielectric

barrier discharges
F. Massines1,a , N. Gherardi2,3 , N. Naudé2,3 , and P. Ségur2,3
1
CNRS-PROMES, Tecnosud, Rambla de la thermodynamique, 66100 Perpignan, France
2
Université de Toulouse, UPS, INPT, LAPLACE – Laboratoire Plasma et Conversion d’Énergie, 118 route de Narbonne,
31062 Toulouse Cedex 9, France
3
CNRS, LAPLACE, 31062 Toulouse, France

Received: 22 December 2008 / Accepted: 26 January 2009

Published online: 28 April 2009 – 
c EDP Sciences

Abstract. This paper is a state of the art of the understanding on the physics of homogeneous dielectric
barrier discharges at atmospheric pressure. It is based on the analysis of present and previous work about
the behavior of these discharges and the conditions to get them. Mechanisms controlling the homogeneity
during gas breakdown and discharge development are successively discussed. The breakdown has to be a
Townsend one, the ionization has to be slow enough to avoid a large avalanche development. During the
breakdown, the discharge homogeneity is related to the ratio of the secondary emission at the cathode
(γ coefficient) on the ionization in the gas bulk (α coefficient). Higher is this ratio, higher is the pressure ×
gas gap product (P d) value for which a Townsend breakdown is obtained. Among the phenomena enhancing
the secondary emission there is the negative charge of the dielectric on the cathode surface, the trapping
of ions in the gas and the existence of excited state having a long lifetime compared to the time between
two consecutive discharges. The first phenomenon is always present when the electrodes are covered by
a solid dielectric, the second one is related to the formation of a positive column and the third one is
specific of the gas. During the discharge development, the homogeneity is mainly controlled by the voltage
or the current imposed by the electrical circuit/electrode configuration and by the gas ability to be slowly
ionized. Larger is the contribution of a multiple step ionization process like Penning ionization, higher will
be the working domain of the discharge. A decrease of the gas voltage during the discharge development is
a solution to enhance the contribution of this process. After 20 years of research a lot of mechanisms have
been understood however there is still open questions like the nature of the Inhibited homogeneous DBD,
surface energy transfers, role of attachment and detachment. . .

PACS. 52.50.Dg Plasma sources – 51.50.+v Electrical properties (ionization, breakdown, electron and ion
mobility, etc.) – 52.80.Dy Low-field and Townsend discharges – 52.80.Hc Glow; corona

1 Introduction comprising many short (<100 ns), narrow (<200 μm) cur-
rent filaments, in general randomly distributed in time
Dielectric barrier discharge (DBD) is an easy and robust and space over the dielectric surface. Even if several au-
solution to generate low temperature discharge at atmo- thors [3,4] reported that a.c. discharges in helium can also
spheric pressure [1]. This discharge has a lot of appli- manifest a pulseless “glow”, in which the atmospheric-
cations since more than one century [2]. Its configura- pressure discharge assumes a diffuse, non-filamentary ap-
tion uses dielectric barriers to maintain current densities pearance, it is the work of Okazaki and their collabora-
below the threshold for glow-to-arc transition. The di- tors since 1987 [5] which has been at the origin of a real
electrics trap charges on the surfaces during a pulse which investigation of homogeneous DBD. Okazaki, Kogoma et
in turn generate a self-induced field in the gas gap that al. studied a lot of different reactor configurations [5–8]
inhibits the discharge before the current density exceed to extend the gases panel leading to homogeneous DBD
the threshold. which at first was limited to helium. They also have shown
It was long thought that for gas gap in the millimetric the interest of these discharges for gas and surface chem-
range or larger, DBD only occurs in a “filamentary” mode, istry. In parallel, since 1992 in France, Massines, Ségur
and their collaborators [9] have focused on the under-
a standing of these discharges physics mainly in He [10–12]
e-mail: francoise.massines@univ-perp.fr

Article published by EDP Sciences

 The European Physical Journal Applied Physics

and N2 [13–15] and their application to polymer surface by photo-ionisation or photo-emission. The time scale of
treatments [16,17] and thin film coatings [18–20]. At the this breakdown mechanism is in the range of nanoseconds
same time in United States, Roth and collaborators have and in a DBD, when the streamer has connected the two
worked on all the possible applications [21–24] including electrodes, a micro-discharge having a radius of typically
sterilisation [25,26] and gas flow control [27,28]. one or two hundred of μm develops.
These three teams generated a confusing list of names DBD can be separated in two groups: those due to a
for homogeneous DBD: Okazaki and Kogoma always used streamer breakdown and those due to a Townsend break-
atmospheric pressure glow discharge (APGD) because in down. The first one usually leads to a filamentary dis-
He, the gas they studied at first, the glow of the homo- charge while the second one leads to a homogeneous dis-
geneous DBD is significantly larger than that of a fila- charge. However, the connection between the breakdown
mentary DBD. Roth chose one atmosphere uniform glow mechanism and the discharge homogeneity is not fully sat-
discharge plasma (OAUGDP) to be more descriptive and isfied: when the density of seed electrons increases, the
Massines suggested that the name of these pulsed dis- density of large size avalanches also increases, inducing
charges should give an indication about the highest ionisa- avalanches interactions, streamer coupling and more or
tion level they can reach (as detailed later) defining Glow less homogeneous discharges. This needs a preionisation
DBD (GDBD or APGD) and Townsend DBD (TDBD or process like photo-ionisation, and these discharges are only
APTD). stable under nanosecond excitation which ensures the si-
When the scientific community has been convinced of multaneous development of the first avalanches. Thus we
the homogeneous DBD robustness, more and more teams, will not discuss these specific discharges.
often driven by applications, have begun to work on these Before going into the physics of the discharge, we have
discharges. Consequently, over the last decade, these dis- to define the homogeneous DBD. Most of the times, the
charges have been intensively studied, finding new con- different studies are driven by teams previously working
figurations which help the discharge stabilisation [29–31], either on atmospheric pressure thermal plasmas or on
increasing our knowledge on the discharge physics and low-pressure non-equilibrium plasmas, and that have con-
chemistry [32–46], and showing the advantage of high verged in the field of non-equilibrium atmospheric pressure
(atmospheric) pressure operation with (i) high degree of discharges. Under the influence of Okazaki’s work, they
discharge uniformity over cm length scales and (ii) thermal usually consider that a glow discharge is an homogeneous
and chemical non-equilibrium. Numerous new applica- discharge but they defined the glow discharge according to
tions including thin film deposition [18–21,47–59], bio- their own experience: for those who come from the ther-
decontamination [26,60–65], and aerodynamic flow con- mal plasmas, an atmospheric pressure glow discharge is a
trol [28,66–68] have been demonstrated. discharge not so conductive as an arc i.e. characterised by
The creativity in the field of homogeneous DBD is large a relatively low current under high voltage; for those hav-
and as the aim of this paper is the understanding of these ing a low-pressure plasma background, a glow discharge
complex discharges and not their applications, we will means a high current under low voltage discharge having
mainly focus on the simplest configuration: 2 plane paral- a specific structure from the cathode to the anode includ-
lel electrodes, a low frequency excitation allowing only one ing cathode fall, Faraday dark space and positive column.
breakdown each half cycle and a gas gap large enough to Thus, the approach to try to get a homogeneous discharge
get filamentary discharges. The gas gap, d, is important will be different in the two cases: the first ones will try
because it is not only the pressure, P , but the product to limit the current, the second ones to get a Townsend
(P d) which determines the ionisation mechanisms: the ra- breakdown similar to low-pressure plasma. However, for
tio of electron emission at the cathode over the ionisation all of us a homogeneous DBD is free of streamer and micro-
in the gas bulk is determined by the ratio of the elec- discharge.
tron mean free path over the gas gap. More precisely, for At the end, the two approaches are necessary to get a
low pd product value, typically lower than 30 Torr cm, stable homogeneous DBD. The mechanisms leading to a
the breakdown is a Townsend breakdown defined by the reliable homogenous DBD can be separated in those con-
γ emission of electron at the cathode and the gas bulk ion- trolling the breakdown and those controlling the discharge
isation defined by the effective ionisation coefficient (α). development. The first condition is to get a Townsend
The time scale of this breakdown is in the range of μs and breakdown, the second one to limit the discharge devel-
the discharge area is equal to that of the electrodes. For opment in order to avoid an ionisation level that would
higher P d values (P d > 200 Torr) breakdown is a streamer be too high. To fulfill the first condition, the “γ emis-
one due to the formation of one electronic avalanche large sion” has to be enhanced compared to the “α ionization”.
enough to create a sufficient amount of ions (∼108 ) to get a As for the second condition, the rise and the amplitude
space charge field of the order of magnitude of the applied of the current have to be limited i.e. the power supply
field. If this condition is fulfilled, breakdown can occur and the electrode configuration has to be designed to slow
without a significant contribution of the cathode surface down the ionisation. Before detailing our understanding
and the transition from Townsend to streamer breakdown of these issues, we will discuss the different homogeneous
occurs: the positive space charge induces a localization DBD regimes and highlight the memory effect from one
of the electric field which accelerates the electrons cre- discharge to the next and the consequence it has on the
ated between the cathode and the positive space charge breakdown conditions.

22805-p2
 F. Massines et al.: Recent advances in the understanding of homogeneous dielectric barrier discharges

1,2
Air
1,0 4
He - N2 N2

Gas voltage (kV)

Gas voltage (kV)

0,8

0,6
2
0,4 Ar - NH3

0,2

0,0 0
0 1 2 3 4 0,00 0,05 0,10 0,15 0,20 0,25 0,30
Current density (mA/cm²) Current density (mA/cm²)
Fig. 1. Gas voltage as a function of the current density during the current increase for He APGD; Ar-NH3 APGD; N2 APTD;
air APTD.

Table 1. Order of magnitude of the different species in the two well characterized homogeneous DBD regimes.
APTD–TDBD APGD–GDBD
−3
Electron density maximum (cm ) 107 –108 1010 –1011
Ions density maximum (cm−3 ) 1010 1011
Neutral plasma formation (n+ = n− ) No Yes (positive column)
Metastable of the dilution gas 1013 1011
Current density (mA/cm2 ) 0.1 to 10 10 to 100
Gas voltage variation around the Constant (Townsend plateau) Decrease (cathode fall formation)
current maximum
Dilution gases N2 , Air, N2 O Penning mixture in He, Ar, Ne
Typical gas gap <2 mm >2 mm
Typical frequency range <10 kHz >1 kHz
Power for a 10 kHz excitation Some W/cm3 Some 0.1 W/cm3
Solutions to observe micro-discharge Current (cf. Fig. 3), optical methods Photomultiplier, ICCD

2 The different homogeneous DBD charge in N2 or in air has been considered as a Townsend
DBD or APTD. Differences between APTD and APGD
To be out of equilibrium, the DBD is an inherently pulsed are summarized in Table 1. One of them is that the APTD
discharge and each discharge can be considered as a di- is a discharge which does not generate an electrically neu-
rect current discharge only related to the others by the tral ionized gas like a positive column. In the APTD the
memory effect from one discharge to the following one. maximum electron density is 3 orders of magnitude lower
The discharge regime is then determined by the max- and the mean power dissipated is typically one order of
imum ionisation rate reached by each discharge. Fig- magnitude larger than in the APGD. Indeed, in a pulsed
ure 1 compares different voltage-current characteristics discharge, the power is determined by the maximum en-
obtained during the current increase of one discharge pulse ergy but also by the pulse duration: when the current is
in different gases. Conditions leading to these discharges lower the pulse duration is usually longer and the power
are described in [5] for He, [69,70] for Ar-NH3 , [13] for N2 does not significantly change.
and [8,71] for air. Another family of homogeneous discharge has been ob-
Before breakdown, the current is related to the mem- tained adding specific component to the set-up like a re-
ory effect which can be considered as an external source sistive layer on the electrode [29] or a choke coil in series
of electrons. During the current increase induced by the with the discharge working in the saturation regime [31].
gas breakdown, first the voltage continues to increase, and The choke coil prevents the fast current increases and the
then it stabilizes to the Townsend plateau value. If a pos- resistive material induces a local decrease of the gas volt-
itive column is formed, the voltage decreases while the age proportional to the local current. Up to our knowl-
current continues to increase [72]. This is the case for He edge, there is no information available about the physics
and Ar+NH3 but not in N2 and air. Thus the homoge- of these discharge and the densities of their main species.
neous discharge in He and Ar+NH3 are usually APGD To reach a better understanding, a fluid model of a He-N2
but the homogeneous DBD in N2 and air are not usually APGD has been developed and the influence of a resistor
APGD. The discharge stops before reaching the ionisation in series with the electrodes has been studied. As shown
level necessary to localize the cathode fall. As the voltage in Figure 2, as the resistance value increases, the ion den-
reaches the Townsend plateau value, the homogeneous dis- sity decreases, the positive column disappears. However,

22805-p3
 The European Physical Journal Applied Physics

12 12
20 10 20 10

11 11
10 10

Electric field (kV/cm)

Electric field (kV/cm)

15 15

Densities (cm )

Densities (cm )
10 10
10 10

9 9
10 10 10 10
Electrons
+
He
8 + 8
10 He2 10

-3

-3
5 Electrons 5
+ N2+
He 7
10 10
7
+
He2
0 N2+ 6
0 6
10 10
0,00 0,02 0,04 0,06 0,08 0,10 0,00 0,02 0,04 0,06 0,08 0,10
K Position (cm) A K Position (cm) A
(a) R = 0 kΩ (b) R = 100 kΩ

Fig. 2. Influence of a resistor in series with a He-N2 APGD: electric field, electrons and ions distribution from the anode to
the cathode when the current is maximum, for (a) R = 0 kΩ and (b) R = 100 kΩ (fluid model results).

50,0 10 1,0
1,0
100 4 0,8
37,5
0,6
25,0 5 0,5

Current (mA)
Voltage (kV)

Current (mA)

2 0,4
Current (mA)

Voltage (kV)
50

Voltage (kV)
12,5 0,2

0,0 0 0 0,0 0 0,0

-0,2
-12,5

-50 -2 -0,4
-25,0 -5 -0,5
-0,6
-37,5 -4 -0,8
-100
-50,0 -10 -1,0 -1,0
0 200 400 600 800 1000 60 80 100 120 140 160 60 80 100 120 140 160

Time (µs) Time (µs) Time (µs)

(a) (b) (c)

Fig. 3. (Color online) First discharges in (a) N2 APTD, (b) He-N2 APGD, (c) He-N2 APGD with a current limitation.
10 cm2 electrode area.

due to the resistor the electrical field decreases when the In conclusion, the physics of APGD and APTD is now
current increases, then even if there is no positive col- well understood that of inhibited homogeneous DBD are
umn, the discharge is not a Townsend discharge (which still misunderstood.
should be obtained under a higher voltage than the glow
one). It is a discharge mainly controlled by different ion-
ization processes as proved by the larger contribution of 3 Memory effect evidences
N+ + +
2 compared to He or He2 . The decrease of the voltage
enhances the contribution of ionization process occurring As illustrated in Figure 3, whatever the conditions, the
under low field compared to those needing a larger field. role of the memory effect is pointed out by the evolution of
This discharge is neither an APTD nor a real APGD, at the first discharges after ignition: the first two discharges
this stage we propose Inhibited APGD. are different from the following ones showing that one
Another approach to get a homogeneous DBD con- DBD pulse depends on the previous one and that some
sists in inserting a metallic mesh between the dielectric discharges are necessary to reach the equilibrium.
and the electrode [8,32,71,73–75]. Lindmayer et al. sus- In the case of the Townsend DBD in N2 , there is no
pected an influence of partial discharges occurring around doubt about the interpretation of the current measure-
the mesh [73,74]. Trunec et al. thought about a resistive ments: the first discharge is filamentary, the second one is
behavior of the metallic mesh [32]. Our experience shows a mix of a Townsend discharge and micro-discharges and
that an important point is the epoxy resin which is used to the third one is fully a Townsend one. Note that the cur-
fix the mesh. Indeed, similar results have been obtained rent of the APTD is low (even for a 10 cm2 electrode) com-
without the mesh, by sticking the electrode directly on pared to that of a streamer. Therefore, in the APTD case,
the dielectric with the resin, which is an insulating ma- without any current limitation, the current measurements
terial [71]. In that case, the discharge obtained in air has are sufficient to determine if there are micro-discharges or
the same behavior as in N2 . not (Tab. 1).

22805-p4
 F. Massines et al.: Recent advances in the understanding of homogeneous dielectric barrier discharges

40
1,0

20
0,5

Voltage (kV)
Current (mA)
0 0,0

-0,5
-20

-1,0
-40
0,00 0,01 0,02 0,03 0,04 0,05
Time (ms)

Fig. 4. (Color online) Current and voltage oscillograms as well as gas gap picture of an unstable discharge in Ar-NH3 mixtures.
Instabilities are clearly observed on the picture but not on the current oscillogram.

In the case of noble gases (Fig. 3b), homogeneous DBD mainly the secondary emission of electrons previous and
reaches higher ionisation level and then higher current am- during the breakdown.
plitude which are of the same order of magnitude or even
larger than micro-discharge ones. Consequently, the anal-
ysis of the current measurement is delicate. Nevertheless, 4 Memory effect and cathode secondary
to exhibit a reproducible glow current, the discharge needs electron emission enhancement
several half cycles. If a current limitation is added (Fig. 3c)
no micro-discharge can be observed but there is a clear Cathode secondary emission depends on the γ secondary
evolution from the first to the third current peak. Very emission coefficient of the surface and on the flux of ener-
similar behaviour has been observed [31] in Ar by adding getic species reaching the cathode surface. These species
a choke coil in series with the discharge and using specific can be ions, excited states and photons. Emission spec-
dielectrics. However, as shown in Figure 4, in argon, troscopy has shown that photon emission is very low at
a large current pulse is not a proof of the absence the first step of the breakdown. Thus, we will focus on
of discharge localisation and/or the presence of micro- “memory” ions and long lived states created during one
discharges mixed with the APGD [69,70,76]. discharge and still present when the next breakdown oc-
At this stage, the important point is that whatever curs. The influence of the memory effect on the secondary
the discharge regime and the power supply characteristics, emission coefficient of the surface and on the ions and long
there is a clear evolution of the current discharge over the lived state species will be successively discussed. Then af-
3 first half cycles and, at least when the current is not ter a synthesis, other mechanisms still under investigations
limited by the power supply, the memory effect is neces- will be presented.
sary to get a stable and homogeneous DBD. Note that
this evolution is very large compared to that observed at
the 10s or minute time scale [77–79] which have been at- 4.1 Surface charge and secondary emission coefficient
tributed to the variation of the energy stocked in the gas
and at the surface [77], to the variation of the secondary Knowing the γ secondary emission coefficient of a dielec-
emission coefficient of the cathode related to the surface tric is not trivial because it depends on the surface charge.
transformations due to discharge interaction [79] or to the In a DBD under alternative excitation, the surface which
temperature increased [78]. is on the cathode side during the current increase was on
Memory effect has to be based on species created dur- the anode side during the previous discharge: hence it has
ing one discharge and still present when the next break- been charged by an electron flux. These electrons, which
down occurs. Obviously, these species largely influence the are trapped on the dielectric shallow traps, are easier to
breakdown and then the transition between streamer and remove than the intrinsic ones. The deepness of shallow
Townsend breakdown defined by the relative contribution traps is typically 1–2 eV from the void level [80] while
of the surface γ emission of electron and the gas bulk α the valence band is at 6–8 eV. Therefore, this trapping
ionisation. After the breakdown, the flux of ions becomes increases the γ value at the beginning of the discharge
dominant whatever the discharge regime but before the and could explain why the nature of the dielectric does
breakdown, energetic species created during the previous not play a dominant role (except for the first discharges
discharge and trapped in the gas can play a dominant role when the memory effect is not yet established [31]) and
allowing a slow ionisation under a too low field to form a why a dielectric on each electrode helps to stabilize the
large electronic avalanche. In the following part, we will discharge [70].
discuss the memory effect mechanisms which are now un- This contribution is proportional to the charge of the
derstood showing how they influence the breakdown and solid dielectric. During the current increase, γ gradually

22805-p5
 The European Physical Journal Applied Physics

Fig. 5. (Color online) Space (anode to cathode) and time (one half cycle) distribution of the ions in a APGD in He calculated
with a fluid numerical modelling [81].

decreases because of the progressive surface charge neu- is no positive column and ions drift to the cathode at a
tralization of the cathode. Synchronization between the time scale of μs: they cannot be involved in the memory
neutralization of this memory charge of the cathode sur- effect. This is confirmed by the stability of the discharge
face and the current maximum is often observed. A possi- under low excitation frequency. In conclusion, if a positive
ble explanation is that when γ decreases, if α does not in- column (or neutral plasma) is formed during the discharge
crease then the current decreases in good agreement with development and if the time between two consecutive dis-
the reduction of the ionisation. This phenomenon also ex- charges is not too long, ions created during one discharge
plains why the voltage necessary to get the same current induce secondary emission of electrons from the cathode,
level is higher during the APTD extinction than during previous and during the next breakdown.
its development (current increase) [70]. If the principle
is clear experiments are necessary to be able to quantify
these effects of surface charge on the γ coefficient and a 4.3 Gas properties and “memory” long live state
deeper study of emission mechanism and discharge-surface species
energy transfers is also necessary.
The energy required to remove an electron from a shallow
trap is about 2 eV which is low in comparison with that of
4.2 Positive column and “memory” ions most of the long live state species. Then, these species can
actively participate to the cathode secondary emission as
Previous studies have shown that it is possible to trap ions far as their flux to the surface is large enough.
from a discharge to the following one [28]. To do so, one The lifetime of noble gases metastables is rather short
needs a neutral plasma zone like the positive column in (μs range) because of the dimmer formation [33]. Con-
an APGD and a high enough excitation frequency [11]. sequently in noble gases, the contribution of “memory”
The electric field and the ions velocity are low in a posi- long live state species is negligible. However in other gases
tive column. Moreover, due to the ambipolar diffusion, the some metastable states does not react with the gas in its
positive column disappearance between two discharges is fundamental level and even at atmospheric pressure their
slow and ions can still be present when the next break- life time can be long compared to the time between two
down occurs. This is illustrated in Figure 5. According to discharges. As example, in N2 , measurements of the abso-
numerical modelling [11], when the breakdown occurs, the lute density of N2 (A3 Σu+ ) metastable have been done by
ions density close to the cathode is 109 /cm3 and their flux optical-optical double resonance-LIF (OODR-LIF) [82].
at the cathode is around 8 × 1013 /cm2 s allowing a signif- Densities as high as 1013 /cm3 are detected and the experi-
icant emission of electrons from the cathode and then the mental quenching rate in N2 is 2.4 × 104 /s. This long-lived
development of a lot of small electronic avalanches under state then represents a reservoir of energy (potential en-
low field and a Townsend breakdown. ergy of 6.3 eV) that can be released in numerous reactions
Of course ions density decreases with time and such a producing electrons.
mechanism is not efficient if the excitation frequency de- Initially [83] a volume Penning ionisation involving
creases too much (typically <1 kHz). In an APTD, there the two metastables N2 (A3 Σu+ ) and N2 (a 1 Σu− ), and

22805-p6
 F. Massines et al.: Recent advances in the understanding of homogeneous dielectric barrier discharges

producing N+ 4 + e, was guessed, but later excluded in the duced by N2 (A3 Σu+ ) [85] taken individually reproduce the
post-discharge on the basis of the too efficient quenching experimental results observed in N2 . Whatever the mech-
of N2 (a 1 Σu− ) in nitrogen [36]. This was verified by adding anism, this tends to show that the key point in obtaining
Xe to N2 : indeed Xe efficiently quenches N2 (a 1 Σu− ) but a homogeneous discharge is the generation of electrons be-
not N2 (A3 Σu+ ). Thus, if Penning ionisation plays a domi- tween two discharges. The role of N2 (A3 Σu+ ) is confirmed
nant role, the addition of Xe in N2 should destabilize the by showing that a small adding of quenchers like O2 or
discharge. However, it was observed to have no significant NH3 considerably limits the minimum frequency under
effect on the APTD behaviour. Even if Penning ionisa- which the APTD occurs. Surprisingly, even if N2 O is a
tion contributes to the gas ionisation at the first stages of more efficient quencher of N2 (A3 Σu+ ) than O2 , it can be
the discharge initiation, this mechanism becomes negligi- added to N2 with no destabilisation of the discharge [88].
ble very quickly after the breakdown because of the rapid A possible explanation is the role of electrons attachment
variation of the first Townsend coefficient of N2 with the and detachment which have not been really considered
electric field. So the strong structural similarity with a up to now because of the low density of electrons in an
Townsend discharge suggests to bring into play surface APTD [44]. However, there is a possibility of attachment
effects at the cathode acting as a source of secondary enhancement because the electrons which are continuously
electrons. Spontaneous electron desorption was proposed created (through one of the previously considered mech-
in [84] leading to a continuous emission of electrons from anisms) accumulate when the electric field is very low.
the cathode. Later, this explanation was excluded because Thus, according to the numerical modelling [89], the max-
the discharge should work when a N2 (A3 Σu+ ) quencher like imum density of electrons is about 7 × 107/cm3 during the
NH3 or O2 is added [85]. Then, secondary electron emis- discharge and about 3 × 108 /cm3 when the electric field
sion by N2 (A3 Σu+ ) impact on the dielectric surface was polarity is reversed. At this instant, the attachment prob-
considered. The discharge being a Townsend discharge, ability is high (low electric field) and the detachment can
metastables are mainly created near the anode and con- occur later (when the electric field increases again). If the
tinuously diffuse to the surface. After the end of the dis- detachment occurs for an electric field lower than the one
charge, when the gas voltage polarity reverses, i.e. when needed for the direct ionisation, this mechanism could cre-
the dielectric which was on the anode side becomes on ate electrons under low field, allowing a slow ionisation.
the cathode side, the metastable flux onto the cathode is Work is still under progress to check this hypothesis and
rather large: ∼5 × 1014 /cm2 s [85]. As metastables induce the possible contribution of electron attachment and de-
secondary emission [86,87], this mechanism can continu- tachment to the Townsend breakdown.
ously produce electrons between two discharges.

5 Gas ionisation slowdown

4.4 Synthesis and possible role of electron attachment
and detachment
The previous part explains why (i) a dielectric layer on
According to all these results, memory effects allowing to each electrode is better to get a diffuse DBD; (ii) the ex-
increase the contribution of the secondary emission pre- citation frequency influences the stability of an APGD
vious and during the first step of the breakdown can be (iii) the gas purity has to be high enough to get an APTD
separated in (i) those always present as soon as the elec- but not why a Penning mixture is necessary to get an
trode is covered by a dielectric layer which is the surface APTD and how the electrical circuit can influence the dis-
charge leading to an increase of the γ coefficient of the charge stability. The experimental parameters discussed
surface; (ii) those related to the formation of a positive up to now are related to the enhancement of γ emission,
column or neutral plasma in which ions can be trapped those we will discuss now are related to the gas ionisation
from one discharge to the following one as far as the ex- slowdown, condition to avoid a large electronic avalanche
citation frequency is high enough; (iii) those depending and thus a streamer and a microdischarge.
on the gas properties, namely the existence of an excited
state having, at atmospheric pressure, a long lifetime com-
pared to the time between two discharges. These mecha- 5.1 Penning ionisation
nisms explain the electron current observed between two
discharges due to the continuous creation of electrons by A Townsend breakdown is associated to a slower and lower
secondary emission. They ensure a slow breakdown under ionisation of the gas than for a streamer breakdown. There
a low electrical field and hence a Townsend breakdown. is two solutions to slowdown the ionisation at the dis-
However, the role of memory effect is still not fully charge ignition: (i) the enhancement of the two steps pro-
understood, as example we are not able to explain why a cess ionisation (ii) the reduction of the voltage breakdown
Townsend breakdown and a Townsend DBD are obtained value.
in pure N2 O. One point that should be emphasised is that, Among two steps ionisation processes, Penning ionisa-
according to numerical modelling, in a case where APTD tion is the more common. In a Penning mixture of M
has been largely studied (N2 ), the three scenarios: Penning with a small concentration of A, the direct ionisation:
ionisation occurring between two discharges [83], sponta- e+M → M + +2e is in competition with the Penning ioni-
neous electron emission [84], and electrons emission in- sation: e+M → M ∗ +e followed by M ∗ +A → A+ +e+M .

22805-p7
 The European Physical Journal Applied Physics

Penning ionisation pushes back the probability of mak- 60

ing a large electronic avalanche and a streamer, because
ionisation due to a two steps process is slower than a di-
rect ionisation. Moreover, the energy needed to create the
metastable state M ∗ is lower than that needed to directly 40

Current (mA)
create the ion. As example, in He the ionisation threshold
is 24.6 eV while that for He* is 19.8 eV. Then, this pro-
cess allows ionisation for a field lower than the breakdown
voltage. This slow ionisation process promotes Townsend 20
breakdown because the created ions have time to reach the
cathode and then to induce secondary emission through
bombardment before the ionisation level is high enough to
localise the field. This is particularly true for He and Ne, 0
their metastables being able to ionise numerous gases, and 60 80 100 120
in particular atmospheric gases, which are always present Time (µs)
as residual gases in discharge vessels. It was also shown Fig. 6. Microdischarges can develop after a Townsend break-
that in Ar, the adding of ketone [8] or NH3 helps to get down. Current oscillogram of a N2 DBD starting like an APTD
APGD in conditions for which only localised discharges but with microdischarges which develop after the current max-
are obtained in pure Ar [69], and this under a voltage imum, when the γ coefficient is minimum and the current fixed
value down to 50%. by the power supply, the electrodes and the solid dielectrics.
The relative contribution of this mechanism depends
on the variation of the ionisation rate with the electric
field and on the variation of the electric field with time.
The faster the variations are, the lower the contribution characteristics. As shown in Figure 6, if this current is too
of Penning ionisation will be. Therefore, this contribu- high, microdischarges can develop even after a Townsend
tion is more important during breakdown when the field breakdown. This has been also observed in Ar [76]. This
is low [89]. Nevertheless, experimental conditions can en- is explained by the fact that the current fixed by the
hance Penning ionisation during the current increase. In- power supply being too high, the ionisation in the gas
deed, as it can be seen in Figure 2, the addition of a re- has to increase too much and leads to large size electronic
sistance in series with a He APGD decrease the electric avalanches. In the case where the ion density created by
field when the current increases resulting in an enhance- the first step of the discharge development is too low (typ-
ment of the Penning ionisation (represented by N+ 2 ions in ically ≤1010 /cm3 ) to localise the field (like APTD), the
Fig. 2) compared to the direct ionisation (represented by space charge is negligible. The field is uniform and a large
He+ and He+ 2 ions) reducing the ionisation speed and then avalanche can always occur and induce the development
the probability of making a large electronic avalanche. of microdischarges.
Hence, Penning ionisation, because it is a two steps
process, contributes to slowdown the gas bulk ionisa- As described in part 2, the adding of a choke coil [31]
tion. Its contribution is significant under low field, when or resistive material [29] increases the working domain of
metastable excited states are preferentially created com- homogeneous DBD. It is interesting to push deeper the
pared to ions. This is the case before and during break- analyses of the role of these components. In a DBD, the
down, and during the discharge development when the dielectric is laid-out to avoid the transition from micro-
electrical field is limited by the electrical circuit. discharge to arc. The coil or the resistive materials have
The contribution of Penning ionisation is clear. How- to avoid the transition from the streamer to the micro-
ever other several steps ionisation which have to be discharge or even from large avalanches to streamer by de-
pointed out could be involved. creasing the gas voltage as soon as or before the ionisation
level becomes high enough to localize the field. The time
scale for arc transition is in the ms range, the time scale
5.2 Power supply and electrodes configuration for streamer formation is in the ns range. Then, taking
into account that a micro-discharge development is very
The power supply and the electrodes configuration (de- fast and induces rather high current level, one can imagine
fined by the shape of the electrodes and the materials that a component decreasing the gas voltage like the inte-
covering them) fix the current or the voltage applied to gral of the current will react too slowly (case of a capaci-
the gas after breakdown. For example, in a DBD, the solid tor), a resistance reacting proportionally to the current is
dielectric imposes a voltage decrease proportional to the better and one sensitive to the derivative is the best (case
dielectric surface charge [2]. In an APTD, the gas volt- of an inductor). Then, the voltage decrease induced by
age is more or less constant during the discharge and the these components will stop the discharge or enhanced low
current is thus fixed and equal to Cds dVa /dt where Cds field ionization mechanism if there is. As shown Figure 2
is the equivalent capacitance of the solid dielectric and where the resistor in series with the discharge decreases
Va is the voltage applied to the electrodes [90]. In par- the gas voltage and then increases the contribution of N+ 2
ticular, the current maximum is independent of the gas due to Penning ionization (He∗ + N2 → N+ 2 + e + He).

22805-p8
 F. Massines et al.: Recent advances in the understanding of homogeneous dielectric barrier discharges

6 Conclusion 16. F. Massines, G. Gouda, J. Phys. D: Appl. Phys. 31, 3411

(1998)
17. F. Massines, G. Gouda, N. Gherardi, M. Duran, E.
Homogeneous DBD are atmospheric pressure discharges Croquesel, Plasmas Polym. 6, 35 (2001)
free of microdischarge. For a given P d product, they are 18. N. Gherardi, S. Martin, F. Massines, J. Phys. D: Appl.
obtained under low frequency alternative excitation when Phys. 33, L104-8 (2000)
the experimental conditions allow a Townsend breakdown 19. S. Martin, F. Massines, N. Gherardi, C. Jimenez, Surf.
and when the discharge development is sufficiently slow. Coat. Technol. 177-178, 693 (2004)
Townsend breakdown is a condition to get a stable and re- 20. F. Massines, N. Gherardi, A. Fornelli, S. Martin, Surf.
producible homogeneous discharge. It is obtained if there Coat. Technol. 200, 1855 (2005)
is enough seed electrons. According to all the results, these 21. P. Tsai, L. Wadsworth, J.R. Roth, Textile Res. J. 67, 359
seed electrons result from a continuous emission of elec- (1997)
trons from the cathode, due to a high γ coefficient ensured 22. J.R. Roth, Z. Chen, P.P.-Y. Tsai, Acta Metallurgica Sinica
by the presence of a dielectric on the cathode and due to 14, 391 (2001)
a flux of energetic species impinging the surface previous 23. J.R. Roth, Phys. Plasmas 12, 5-2 (2005)
to the breakdown. If a positive column is formed these 24. J.R. Roth, J. Rahel, X. Dai, D.M. Sherman, J. Phys. D:
species can be ions, if the life time of metastable is large Appl. Phys. 38, 555 (2005)
enough it can be metastable. Negative ions may also play 25. K. Kelly-Wintenberg, T.C. Montie, T.C. Brickman, J.R.
a role in the seed electrons enhancement in the case of Roth, A.K. Carr, K. Sorge, L.C. Wadsworth, P.P.-Y. Tsai,
electronegative gases. At each step, the amplification in J. Ind. Microbiol. Biotechnol. 2, 69 (1998)
the gas has to be limited to avoid the formation of a large 26. T.C. Montie, K. Kelly-Wintenberg, J.R. Roth, IEEE
electronic avalanche. This is obtained by the control of the Trans. Plasma Sci. 28, 41 (2000)
current or voltage increase and by the enhancement of a 27. J.R. Roth, D.M. Sherman, S.P. Wilkinson, AIAA J. 38,
1166 (2000)
slow ionisation process like Penning ionisation.
28. J.R. Roth, Phys. Plasmas 10, 5 (2003)
29. M. Laroussi, I. Alexeff, J.P. Richardson, F. Dyer, IEEE
Trans. Plasma Sci. 30, 1 (2002)
References 30. T. Terajima, H. Koinuma, Appl. Surf. Sci. 223, 259 (2004)
31. E. Aldea, M.C.M. Van de Sanden, P. Peeters, H. de Vries,
1. U. Kogelschatz, B. Eliasson, W. Egli, Pure Appl. Chem. Surf. Coat. Technol. 200, 46 (2005)
71, 1819 (1999) 32. D. Trunec, A. Brablec, F. Stastny, Contrib. Plasma Phys.
2. U. Kogelschatz, Plasma Chem. Plasma Process. 23, 1 38, 435 (1998)
(2003) 33. F. Tochikubo, T. Chiba, T. Watanabe, Jpn J. Appl. Phys.
3. A. Von Engel, R. Seeliger, M. Steenbeck, Z. Phys. 85, 144 38, 5244 (1999)
(1933) 34. S.F. Miralai, E. Monette, R. Bartnikas, G. Czeremuszkin,
4. R. Bartnikas, J. Appl. Phys. 40, 1974 (1969) M. Latreche, M.R. Wertheimer, Plasmas Polym. 5, 63
5. S. Kanazawa, M. Kogoma, T. Moriwaki, S. Okazaki, J. (2000)
Phys. D: Appl. Phys. 21, 838 (1988) 35. T. Nozaki, Y. Miyazaki, Y. Unno, K. Okazaki, J. Phys. D:
6. S. Okazaki, M. Kogoma, H. Uchimaya, in Proc. 3rd Int. Appl. Phys. 34, 3383 (2001)
Symp. High Pressure Low Temp. Plasma Chem., Hakone 36. Yu.B. Golubovskii, V.A. Maiorov, J. Behnke, J.F. Behnke,
III, Strasbourg, France (1991), p. 101 J. Phys. D: Appl. Phys. 35, 751 (2002)
7. N. Kanda, M. Kogoma, H. Jinno, H. Uchiyama, S. Okazaki, 37. M. Madani, A. Bogaerts, R. Gijbels, D. Vangeneugden,
in Proc. 10th Int. Symp. Plasma Chemistry, Vol. 3 (1991), in Proc. 8th Int. Symp. High Pressure Low Temp. Plasma
p. 2 Chem., Hakone VIII, Puhajarve, Estonia (2002), pp. 130–
8. S. Okazaki, M. Kogoma, M. Uehara, Y. Kimura, J. Phys. 133
D: Appl. Phys. 26, 889 (1993) 38. I. Radu, R. Bartnikas, M.R. Wertheimer, J. Phys. D: Appl.
9. R. Messaoudi, F. Massines, A. Younsi, B. Despax, C. Phys. 36, 1284 (2003)
Mayoux, in Proc. 10th Int. Conf. Gas Discharges and their 39. Yu.B. Golubovskii, V.A. Maiorov, J. Behnke, J.F. Behnke,
Appl. (GD’92), Swansea, UK, 1992, pp. 318–321 J. Phys. D: Appl. Phys. 36 (2003)
10. F. Massines, C. Mayoux, R. Messaoudi, A. Rabehi, P. 40. T. Callebaut, I. Kochetov, Yu. Akishev, A. Napartovich,
Ségur, in Proc. 10th Int. Conf. Gas Discharges and their C. Leys, Plasma Sour. Sci. Technol. 13, 245 (2004)
Appl. (GD’92), Swansea, UK, 1992, pp. 730–733 41. L. Mangolini, C. Anderson, J. Heberlein, U. Kortshagen,
11. F. Massines, A. Rabehi, Ph. Decomps, R. Ben Gadri, P. J. Phys. D: Appl. Phys. 37, 1021 (2004)
Ségur, Ch. Mayoux, J. Appl. Phys. 38, 2950 (1998) 42. Peng-Zhang, U. Kortshagen, IEEE Trans. Plasma Sci. 33,
12. I. Enache, N. Naudé, J.P. Cambronne, N. Gherardi, F. 318 (2005)
Massines, Eur. Phys. J. Appl. Phys. 33, 15 (2006) 43. J.J. Shi, D.W. Liu, M.J. Kong, Appl. Phys. Lett. 89,
13. N. Gherardi, F. Massines, IEEE Trans. Plasma Sci. 29, 081502 (2006)
536 (2001) 44. R. Brandenburg, V.A. Maiorov, Yu.B. Golubovskii, H.-E.
14. F. Massines, P. Ségur, N. Gherardi, C. Khamphan, A. Wagner, J. Behnke, J.F. Behnke, J. Phys. D: Appl. Phys.
Ricard, Surf. Coat. Technol. 174-175, 8 (2003) 38, 2187 (2005)
15. N. Naudé, J.P. Cambronne, N. Gherardi, F. Massines, J. 45. A. Bogaerts, K. de Bleecker, V. Georgieva, I. Kolev, M.
Phys. D: Appl. Phys. 38, 530 (2005) Madani, E. Neyts, Plasma Process. Polym. 3, 110 (2006)

22805-p9
 The European Physical Journal Applied Physics

46. R. Bartnikas, I. Radu, M.R. Wertheimer, IEEE Trans. 69. N. Gherardi, E. Croquesel, F. Massines, in Proc. 16th Int
Plasma Sci. 35, 1437 (2007) Symp. Plasma Chem. (ISPC 16), Taormina, Italy, 2003,
47. T. Yokoyama, M. Kogoma, S. Kanazawa, T. Moriwaki, S. p. 797
Okazaki, J. Phys. D: Appl. Phys. 23, 374 (1990) 70. F. Massines, N. Gherardi, N. Naudé, P. Ségur, Plasma
48. Y. Sawada, S. Ogawa, M. Kogoma, J. Phys. D: Appl. Phys. Phys. Control. Fusion 47, B577 (2005)
28, 1661 (1995) 71. N. Naudé, S. Okazaki, N. Gherardi, F. Massines, in Proc.
49. K. Schmidt-Szalowski, W. Fabianowski, Z. Rzanek- 18th Int. Symp. Plasma Chem. (ISPC 18), Kyoto, Japan,
Boroch, J. Sentek, J. Chem. Vap. Deposition 6, 183 (1998) 2007
50. R. Prat, Y.J. Koh, Y. Babukutty, M. Kogoma, S. Okazaki, 72. Yu. P. Raizer, in Gas Discharge Physics (Springer-Verlag,
M. Kodama, Polymer 41, 7355 (2000) Berlin, 1991)
51. O. Goossens, E. Dekempeneer, D. Vangeneugden, R. Van 73. J. Tepper, M. Lindmayer, J. Salge, in Proc. Hakone VI,
de Leest, C. Leys, Surf. Coat. Technol. 142–144, 474 Cork, Ireland, 1998, pp. 123–127
(2001) 74. J. Tepper, P. Li, M. Lindmayer, in Proc. XIV Int. Conf.
52. A. Sonnenfeld, T.M. Tun, L. Zajıckova, K.V. Kozlov, H.- on Gas Discharges and their Appl., Liverpool, 2002, pp.
E. Wagner, J.F. Behnke, R. Hippler, Plasmas Polym. 6, 4 1–4
(2001) 75. X. Wang, H. Luo, Z. Liang, T. Mao, R. Ma, Plasma Sources
53. H.-E. Wagner, R. Brandenburg, K.V. Kozlov, A. Sci. Technol. 15, 845 (2006)
Sonnenfeld, P. Michel, J.F. Behnke, Vaccuum 71, 417 76. N. Naudé, F. Massines, IEEE Trans. Plasma Sci.,
(2003) doi: 10.1109/TPS.2008.923899 (2008)
54. R. Foest, F. Adler, F. Sigeneger, M. Schmidt, Surf. Coat. 77. P. Decomps, F. Massines, Ch. Mayoux, in Proc. 23rd Int.
Technol. 163-164, 323 (2003) Conf. on Phenomena in Ionized Gases (XXIII ICPIG),
55. L.J. Ward, W.C.E. Schofield, J.P.S. Badyal, A.J. Goodwin, Toulouse, France, 1997, pp.106–107
P.J. Merlin, Chem. Mater. 15, 1466 (2003) 78. N. Naudé, N. Gherardi, Et. Es-Sebbar, J.P. Cambronne,
56. D. Trunec, Z. Navratil, P. Stahel, L. Zajickova, V. F. Massines, in Proc. 9th Int. Symp. on High Pressure Low
Bursikova, J. Cech, J. Phys. D: Appl. Phys. 37, 2112 Temp. Plasma Chem. (Hakone IX), Padova, Italy, 2004
(2004) 79. J. Shin, L.L. Raja, J. Phys. D: Appl. Phys. 40, 3145 (2007)
57. S.E. Alexandrov, M.L. Hitchman, Chem. Vap. Deposition 80. Kai Wu, L.A. Dissado, IEEE Trans. Dielectr. Electr. Insul.
11, 457 (2005) 12, 655 (2005)
58. F. Fanelli, F. Fracassi, R. d’Agostino, Plasma Process. 81. R. Ben Gadri, Ph.D. thesis, University of Toulouse, 1997
Polym. 2, 688 (2005) 82. G. Dilecce, P.F. Ambrico, S. De Benedictis, Plasma
59. E. Njatawidjaja, M. Kodoma, K. Matsuzaki, K. Yasuda, Sources Sci. Technol. 16, 511 (2007)
T. Matsuda, Plasma Process. Polym. 3, 338 (2006) 83. N. Gherardi, G. Gouda, E. Gat, A. Ricard, F. Massines,
60. M. Laroussi, IEEE Trans. Plasma Sci. 24, 1188 (1996) Plasma Sources Sci. Technol. 9, 340 (2000)
61. T. Akitsu, H. Ohkawa, M. Tsuji, H. Kimura, M. Kogoma, 84. Y.B. Golubovskii, V.A. Maiorov, J. Behnke, J.F. Behnke,
Surf. Coat. Technol. 193, 29 (2005) J. Phys. D: Appl. Phys. 35, 751 (2002)
62. M.K. Boudam, M. Moisan, B. Saoudi, C. Popovici, N. 85. C. Khamphan, P. Ségur, F. Massines, M.C. Bordage, N.
Gherardi, F. Massines, J. Phys. D: Appl. Phys. 39, 3494 Gherardi, Y. Cesses, in Proc. 16th Int. Symp. on Plasma
(2006) Chem., Taormina, Italy, 2003
63. M.K. Boudam, M. Moisan, B. Saoudi, C. Popovici, N. 86. P.F. Little, Encyclopedia of physics, Electron Emission gas
Gherardi, F. Massines, J. Phys. D: Appl. Phys. 39, 3494 discharges I, Vol. XXI, edited by S. Flugge (1956)
(2006) 87. D.A. Bosan, T.V. Jovanovic, D.M. Krmpotic, J. Phys. D:
64. H. Ohkawa, T. Akitsu, M. Tsuji, H. Kimura, M. Kogoma, Appl. Phys. 30, 3096 (1997)
K. Fukushima, Surf. Coat. Technol. 200, 5829 (2006) 88. N. Gherardi, N. Naudé, Et. Es-Sebbar, I. Enache, H.
65. G. Fridman, G. Friedman, A. Gutsol, A.B. Shekhter, Caquineau, F. Massines, inProc. 10th Int. Symp. on High
V.N. Vasilets, A. Fridman, Plasma Process. Polym. 5, 503 Pressure Low Temp. Plasma Chem. (Hakone X), Saga,
(2008), doi: 10.1002/ppap.200700154 Japan, 2006, pp. 21–24
66. J. Pons, E. Moreau, G. Touchard, J. Phys. D: Appl. Phys. 89. P. Ségur, F. Massines, in Proc. 12th Int. Conf. on Gas
38, 3635 (2005) Discharges and their Appl. (GD 2000), Glasgow, Scotland,
67. A. Seraudie, E. Aubert, N. Naudé, J.-P. Cambronne, in 2000, p. 15
Proc. 3rd AIAA Flow Control Conf., San Francisco, USA, 90. E. Croquesel, N. Gherardi, S. Martin, F. Massines, in Proc.
2006, Paper 2006-3350 8th Int. Symp. on High Pressure Low Temp. Plasma Chem.
68. J.P. Boeuf, Y. Lagmich, T. Unfer, T. Callegari, L.C. (Hakone VII), Greifswald, Germany, 2000, pp. 88–92
Pitchford, J. Phys. D: Appl. Phys. 40, 652 (2007)

22805-p10