Received: 12 March 2020 | Revised: 20 July 2020 | Accepted: 21 July 2020

DOI: 10.1002/ppap.202000053

FULL PAPER

Atmospheric pressure plasma jet powered by piezoelectric

direct discharge

Dariusz Korzec | Florian Hoppenthaler | Dominik Burger |

Thomas Andres | Stefan Nettesheim

Relyon Plasma GmbH, R&D Department,

Regensburg, Germany Abstract
The piezoelectric direct discharge sustained using a high‐performance piezo-
Correspondence
Dariusz Korzec, Relyon Plasma GmbH,
electric transformer (PT) with a voltage transformation ratio of >1,000 was used
Osterhofener Straße 6, 93055 Regensburg, for the generation of an atmospheric pressure plasma jet (APPJ). The ionization
Germany. gases used are ambient air and compressed dry air. The APPJ was characterized
Email: d.korzec@relyon-plasma.com
using capacitive probe measurement, ozone concentration measurement, and
activation area determination.
The activation experiments
were conducted on a high‐
density polyethylene. The acti-
vation area was visualized using
pure formamide test ink
(58 mN/m) and captured by a
digital camera. The influence of
gas flow, PT power, and the
distance between the PT and
the substrate were investigated.

KEYWORDS
atmospheric pressure plasma jet (APPJ), CeraPlas™ F, piezoelectric direct discharge (PDD),
piezoelectric transformer (PT), plasma surface treatment

1 | INTRODUCTION generators,[13–15] or plasma for thrusters.[16] The resonantly

driven PTs[17] with multilayer structure[18] reached trans-
The piezoelectric[1] transformers (PT)[2] of Rosen type[3] formation rates sufficient for sustaining an atmospheric
are broadly used as a high‐voltage source for power pressure discharge in noble gases.[19,20]
electronics.[4,5] The earlier generations of PTs allowed for On the basis of such PTs, the atmospheric pressure
the ignition of low‐pressure gaseous discharges, such as plasma jets (APPJ) operated with noble gases have been
neon lamps,[6] cold cathode fluorescent lamps (CCFL),[7,8] developed.[21–25] The first handheld plasma device,[26] the
silent discharge exciters in nitrogen,[9] oxygen‐containing piezobrush® PZ1 was introduced in 2009. It was based on
gas mixtures for ozone generation,[10–12] ionic wind PT working with a resonant frequency of 135 kHz, an input
-------------------------------------------------------------------------------------------------------------------------------------------
This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits use and distribution in any
medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made.
© 2020 The Authors. Plasma Processes and Polymers published by Wiley‐VCH GmbH

Plasma Process Polym. 2020;e2000053 www.plasma-polymers.com | 1 of 14

https://doi.org/10.1002/ppap.202000053
2 of 14 | KORZEC ET AL.

power of 20 W, and an output voltage of 1.8 kV, which 2.1 | Plasma production
allows for an efficient discharge in noble gases, primarily in
helium. It is a useful tool for surface activation before the 2.1.1 | Microdischarges
processing steps such as gluing, painting, printing, casting,
foaming, coating, or siliconizing. An increase in wettability, The PDD in air, like the dielectric barrier discharge,[39]
printability, or adhesion has been achieved on consists of bursts of microdischarges,[40] which ignites if the
polymers[27,28] and ceramics.[29,30] The activation allowed electric field at the PT tip exceeds the breakdown field. The
for grafting of different chemically active films. For in- last one depends on pressure, temperature, and humidity
stance, silica capillary nozzles were treated before dipping and ranges from 5 to 15 kV/mm. The microdischarges de-
into a silanisation solution[31] or polymer brushes were velop as positive or negative streamers,[41] depending on the
grafted on the plasma‐treated polytetrafluoroethylene, polarization of the electric charge of the PT tip. A typical
polypropylene, and polyethylene surfaces for allowing bet- duration of the microdischarge in the air is 1–10 ns.[42]
ter wettability of these surfaces.[32] The electric resistivity of the PZT ceramics is quite
From 2006 to 2009, EPCOS in Deutschlandsberg gath- high, in MΩ⋅cm order of magnitude,[43] allowing high
ered extensive experience in development[33,34] and pro- voltages between different regions of the PT tip surface to
duction of PTs with voltage transformation ratio, which is rise. Consequently, it is possible that one region initiates
high enough for ignition of atmospheric pressure discharges already a microdischarge, when other regions are still
in the air even for small primary voltages in the range of accumulating the charge to reach the breakdown field.
5–25 V. The result of the high‐performance PT development Therefore, several independent microdischarges can be
is the CeraPlas™ F[35] allowing for piezoelectric direct dis- ignited during the same half‐period of the PT oscillation.
charge (PDD). This PT enabled the development of the first
handheld piezoelectric device worldwide, generating plas-
ma with ambient air as ionization gas.[36] Starting from 2.1.2 | Piezoelectric transformer
2014, this APPJ was commercialized by Relyon Plasma
GmbH as piezobrush® PZ2 and used for a large variety of The PT used here is constructed as a two‐zone cuboid‐
applications.[37,38] shaped Rosen type,[45] PZT (lead–zirconate–titanate,
The piezobrush® PZ3 device, used for most experiments Pb[Zrx Ti1− x ]O3(0 ≤ x ≤ 1))‐based multilayer PT. Its
in this study, is the further developed version of piezo- most important physical parameters are summarized in
brush® PZ2. It has the same operating principle and uses Table 1, where LPT , w PT , and dPT are length, width, and
the same PT. The control unit supplemented with an in- thickness of the PT, respectively.
telligent machine–human interface (MHI) is new. It allows The primary zone consists of a multilayer stack of al-
for setting the operation modus, power level and treatment ternately laid copper and PZT layers, as shown in Figure 1b.
time, and display of diagnostic information. A higher The PZT layers are prepolarized in direction perpendicular
plasma processing efficiency is reached by a moderate re- to the layers. They are polarized alternately to increase the
duction of the airflow. The aim of this study is to char-
acterize this plasma device for different potential
TABLE 1 The basic data of CeraPlas™ F[35]
application fields. Three kinds of characterization are used
for this purpose: measurement of a PT‐related voltage, Parameter description Parameter value
evaluation of ozone concentration in the plasma gas, and Operating frequency (kHz) 50
determination of activation area and speed on high‐density
Weight (g) 8.0 ± 1.6
polyethylene (HDPE).
LPT × w PT × dPT (mm) 72 × 6 × 2.8
Material PZT (lead–zirconate–titanate)
2 | EXPERIMENTAL DETAILS Maximum operating 8.0
power (W)
In this section, the PDD production method was de-
Input voltage (V) 12–24
scribed. Thereafter, three PDD jet characterization
methods were presented. The first is the electric mea- Output voltage (kV) <15
surement using a large area capacitive probe. The second Input capacitance Cin (µF) 2.0 ± 0.4
is the measurement of ozone concentration in the plasma Output capacitance Cout (pF) 3.0 ± 0.6
gaseous products using the ultraviolet (UV) absorption
Ozone production rate in air <100
spectroscopy. The third is the determination of the acti-
(mg/hr)
vation area using test ink and digital camera.
KORZEC ET AL. | 3 of 14

(a) Δx of the PT material as a function of a relative axial position

x ∕LPT for five harmonic frequencies is shown in Figure 2.
For efficient energy transfer from the primary zone to the
secondary zone, a maximum axial displacement by resonant
oscillation should be placed at the connection plane between
both zones. The first harmonic frequency is not usable
because, if the largest displacement occurs in the middle of
the PT length, there will be no oscillation, and consequently
no voltage will be induced at the tip of the secondary zone of
the PT. On the contrary, if the displacement is maximum at
the PT tip, the displacement at the interface between the PZT
(b) zones will be zero. The resonant oscillation with maximum
displacement at the center and on the end tip is possible only
for even harmonics. The lowest frequency, which can be
used for voltage transformation is the second harmonic.
The nth harmonic frequency of the PT depends on its
length LPT and longitudinal sound speed in the PZT,
v PZT [7] accordingly to the following equation:

v PZT (1)
fn = n .
2LPT

Using a typical sound speed of 3,600 m/s for PZT

obtained from the literature,[47] the second harmonic
frequency of 50 kHz, used for the PT excitation, can be
F I G U R E 1 Structure of the multilayer resonant obtained.
piezoelectric transformer[44]: (a) Top view. (b) Cross‐section. The PT bar oscillating longitudinally with the second
The arrows symbolize the prepolarization of the harmonic resonant frequency f2 has the points of mini-
lead–zirconate–titanate (PZT) material mal longitudinal displacement at 1/4th and 3/4th of its
length. This fact is used to place the electric connections
strength of the converse piezoelectric effect.[46] Each second at the LPT ∕4 position, as shown in Figure 2, assuring a
copper layer is contacted to a PT input electrode on one minimum possible fatigue of the electrical connection
narrow side of the PT block. The remaining copper layers (soldering) and a minimized dumping of the oscillation.
are contacted to the second input electrode on the opposite
narrow side of the PT block. The input voltage Vin is applied
between these two electrodes as shown in Figure 1a. The
PZT layer stacked between two neighboring copper layers
undergoes an elastic deformation when the voltage is ap-
plied. The sum of the PZT layer deformations contributes to
the global deformation of the PT. As the secondary zone of
the PT is prepolarized longitudinally, a longitudinal de-
formation across the connection plane between the two
zones causes a build‐up of the voltage on the tip of the PT. If
the voltage applied to the input electrodes is periodic with a
period corresponding to one of the frequencies of the pie-
zoelectric longitudinal resonance, the PT starts to oscillate,
and a high voltage is induced on its tip.

2.1.3 | Operation frequency of the PT F I G U R E 2 The longitudinal displacement Δx of the

piezoelectric transformer (PT) material during a resonant
The piezoelectric resonance of the PT occurs at multiple oscillation with nth harmonic frequency, as a function of a
standing‐wave frequencies fn . The longitudinal displacement relative axial position x ∕LPT
4 of 14 | KORZEC
2.1.4 | Voltage transformation ratio
The PT output voltage is proportional to the input voltage
whereupon the proportionality factor is the voltage
transformation ratio. It depends on the number of pri-
mary layers nlayers , the thickness of the stack, in the
current case dPT , the PT length LPT , and the frequency‐
dependent PZT material coefficient g (ω), as follows[7]:
LPT ⋅ nlayers (2)
G V = g (ω ) .
dPT
The amplitude of the input oscillation is limited by a
maximum allowed power density, which is stress limited.
The peak power density for some PZT samples was cal-
culated to be 330 W∕cm3 at 100 kHz.[48] The formulas
describing the dependence of the resonant frequency, the
voltage transformation ratio, and other electric properties
of a PZT‐based PT on geometrical and material data are
described by Fukunaga et al.[49] Using Equation (2) for
the PT data from Table 1, the voltage transformation ratio
of about 1,000 can be obtained.
2.1.5 | Resonance frequency control
To obtain a high voltage transformation ratio, the PT
input signal must have a frequency, which is equal to the
resonant oscillation frequency. However, different factors
cause variations in the speed of sound and consequently,
in the resonant oscillation frequency. Such factors are the
(a) (b)
ET AL.
PT temperature and electrical load.[50] To sustain the
resonant conditions, the excitation frequency must be
permanently adjusted to the changing resonant condi-
tions. For this purpose, a 32‐bit microcontroller is
applied.[51] It uses the input voltage and current to de-
termine the input impedance of the PT. The frequency
that maximizes this impedance and hence the voltage
transformation ratio is searched.
2.2 | Capacitive probe
2.2.1 | Probe configuration
To analyze the changes of the unbalanced electric charge
on the tip of the PT, a large area capacitive probe
(Figure 3a) consisting of two electrodes made of two
20‐µm‐thick pieces of aluminum foil being superimposed
on both sides of a 1‐mm‐thick alumina plate was used.
Both electrodes are aligned and have an area of
S = 25 cm2. The measured capacitance of the capacitive
probe Cp is 187 ± 0.2 pF and is in good agreement with
the calculated one. The front electrode faces the PT and
remains electrically floating. The back electrode is con-
nected to the resistive shunt R and the oscilloscope vol-
tage probe. The voltage over the shunt was measured
using the Tektronix P6015A voltage probe with an input
resistance of 100 MΩ and the input capacitance of less
than 3 pF connected to the Tektronix DPO3034 phosphor
oscilloscope operated with 5,000,000 samples per mea-
surement and with the sample duration of 0.4 ns.
F I G U R E 3 Setup for measurement of
the voltage generated by piezoelectric
discharge using capacitive large area probe.
(a) The geometrical configuration. (b) The
equivalent electric circuit
KORZEC ET AL.
Critical for the probe measurement is the distance d
between the probe front electrode and the PT tip. The
smaller it is, the stronger is the measured signal. The
limiting factor for decreasing the distance is the occur-
rence of sparks between the PT and the front electrode,
which would influence the PT operation and spoil the
measurement. The minimum spark‐free distance for
maximum power investigated is d = 30 mm . This value
is used as a standard distance for all electrical measure-
ments in this study.
The shunt resistor R has the resistance of 9.92 kΩ.
The value of approximately 10 kΩ was determined em-
pirically as a result of a compromise between two ten-
dencies. A noise contribution to the signal becomes too
high for R ≪ 10 kΩ. For a resistance much higher than
10 kΩ the plasma signal is “filtered away” together with
noise. Both effects can be analyzed using the electrical
equivalent circuit of the measurement setup presented in
Section 2.2.2.
2.2.2 | Electric equivalent circuit
The measurement setup of the capacitive probe can be
modeled by the equivalent electric circuit shown in
Figure 3b. Assuming a sinusoidal signal with angular
frequency ω on the PT output, this circuit represents a
complex voltage divider, which allows expression of the
voltage on the input of oscilloscope in phasor convention[52]
VOSC as a function of the PT output voltage VPT :
R (3)
VOSC e ωt = VPT e ωt = | ZA | VPT e ωt + jθA ,
R + jXC
with attenuation magnitude
R (4)
| ZA | = ,
R2 + XC2
and attenuation phase angle
⎛ ⎞
XC
θA = −arctan ⎜⎜ ⎟,
⎟
(5)
2
⎝ R2 + XC ⎠ 1,000 ppm (volume).
where XC is the reactance of the serial connection of Cg
and Cp representing the capacity of the gap between the
PT and the probe, and the probe capacity, respectively.
Using a simulation program Quickfield, version 6.4, the
approximated value of the capacitance Cg of 0.3 pF
is calculated, which is on three orders of magnitude
less, compared with Cp. Consequently Cg ∥ Cp ≈ Cg and
| 5 of 14
1 (6)
XC = − .
ωCg
As XC depends on the signal frequency, also the
magnitude ∣ ZA ∣ and phase angle θA are frequency‐
dependent. For the angular frequency of the PT oscilla-
tion ω = 2π × 50 × 103/s and R ≈ 10 kΩ the attenua-
tion magnitude is 0.0009 and the phase angle is ≈90∘. It
means, the PT output voltage amplitude can be estimated
by multiplying the oscilloscope voltage amplitude
by 1,000.
From Equation (4) it is obvious that the oscilloscope
voltage, and consequently, the signal‐to‐noise ratio, will
increase with resistance R . On the contrary, an increase
in resistance R results in an increase in the time constant
RCg , which is 3 µs for 10 kΩ, causing suppression of high‐
frequency signals, such as those originating from the
microdischarges (compare with empirical findings in
Section 2.2.1). As the attenuation calculated using
Equation (4) is frequency‐dependent, for signals related
to microdischarges, different magnitudes, and phase an-
gles must be calculated.
2.3 | Characterization of ozone
production
The gaseous discharge in air produces a large number of
chemically active and excited species.[53] They play a
crucial role in all chemical reactions between plasma and
the treated surfaces; therefore, it is advantageous to
maximize their concentrations. Most of them are short‐
lived, and consequently, quite difficult for quantitative
analysis. One considerably stable product of the cold air
plasma is ozone.
There are a number of measurement techniques for
the determination of the ozone concentration in the gas
phase, which can be implemented for the characteriza-
tion of PDD. For the ozone concentration measurement
in this study, a desktop instrument based on UV
absorption spectroscopy,[54,55] Ozone Analyzer Model
UV‐100 of ECO Sensors, Inc. was used. It allows the
ozone concentration measurement in the range of 0.01 to
To obtain exact values of the ozone production rate,
exact control of airflow is needed. For this purpose, the
experimental setup with mass flow meter and needle
valve for control of compressed dry air (CDA) flow was
used, as shown schematically in Figure 4.
The ozone concentration is not a suitable value for
the characterization of the PT efficiency because the re-
sult depends on airflow (see Section 3.2.2). That which is
6 of 14 | KORZEC
more suitable for such purpose is the production rate
defined as the mass of ozone produced per time unit.
Using an airflow fCDA , the production rate of ozone
Rprod is calculated from the ozone concentration NO3 as
follows:
MO3 (7)
Rprod = ⋅f ⋅ NO3,
VA CDA
where VA is the molar volume and MO3 is the molar mass
of ozone (48 g/mol).
2.4 | Activation area determination
To increase the surface energy of polymers, two para-
meters are of importance: (i) The maximal surface energy
reached after a treatment and (ii) the achievable treat-
ment speed.
The standard method for surface energy measure-
ment is the droplet test conducted with at least two test
liquids. Another method is using the test inks calibrated
for different surface energies, typically in steps of, for
example, 2 mN/m. When using the PDD‐based APPJ for
surface activation, the maximum surface energies are
reached in a very short time, and it is difficult to use these
values for the quantitative characterization of the APPJ
performance. Much more suitable for quantitative eva-
luation is the area of the activated surface. The method
applied in this study is based on the determination of the
activated area visualized on the HDPE plates using a
58‐mN/m test ink.
ET AL.
F I G U R E 4 Setup for measurement
of ozone concentration in CDA flow.
CDA, compressed dry air; MFM, mass flow
meter; PT, piezoelectric transformer;
UV, ultraviolet
The reasons for the choice of HDPE for plasma
treatment are: its wide spread in the industry, frequent
use as a reference material for plasma activation, and
weak hydrophobic recovery. The measurements of the
present study using a Krüss droplet test instrument
showed that the surface energy reached after PDD plas-
ma treatment remains practically unchanged after 4 hr in
the environmental air.
For plasma treatment, the plasma device is positioned
at a required distance d between the PT tip and the
substrate surface. The default distance and treatment
time are 4.5 mm and 10 s, respectively.
The test inks gauging the surface energies from 0 to
58 mN/m are typically mixtures of formamide and ethy-
lene glycol monoethyl ether.[56] As the ethylene glycol
monoethyl ether is much more volatile than the for-
mamide, the proportion of these liquids and conse-
quently the gauged surface energy of the test ink can
change in time. To avoid such aging effect, for activation
spot visualization, the 58‐mN/m test ink consisting of
pure formamide is selected. The additional advantage of
the value 58 mN/m is that it assures a high dynamic
range of the evaluation method because it is positioned in
the middle between the surface energy value of non-
treated HDPE (36 mN/m) and the maximum reachable
value for a plasma‐treated surface, of 72 mN/m.
It can be observed that after covering the activated
surface with test ink its area decreases within seconds.
The reason for this effect is the absorption of water from
the environmental humidity into the test ink, resulting in
an increase in the gauged surface energy and a decrease
in the visualized surface. The dependence of the
KORZEC ET AL.
visualized activated area on time elapsed after the ink
application is then called the shrinkage plot.
To assure the accuracy and repeatability of the
shrinkage plot, the pictures of the ink spot are taken in
short intervals using a digital camera. The contour of the
ink spot is automatically recognized and the ink spot area
is calculated from the pixel count using a specialized
software. A typical shrinkage plot is shown in Figure 5.
The shrinkage speed of the test ink spot decreases with
time. It is almost 1% per second, 7 s after test ink appli-
cation, and only 0.13%/s, 2 min after test ink application.
As a reference time in this study, 10 s is used. It is a
compromise between the large uncertainty of the surface
determination for very short times and an increasing
relative error by determination of the area of test ink
spots after very long shrinkage time.
The temperature and humidity have a significant in-
fluence on the surface tension of liquids.[57] According to
the American Society for Testing and Materials
(ASTM)[56] for test ink application they should be
25 ± 2°C and 50% ± 5%, respectively. Therefore, the re-
lative air humidity and temperature are measured by
electronic sensors at a distance of about 10 cm from the
ink spot and recorded together with the ink area data to
assure the validity of the method.
3 | R E S U L T S AN D D I S C U S S I O N
3.1 | Probe measurement results
3.1.1 | Voltage signal
A 50‐µs excerpt of the time‐dependent voltage signal
obtained from the shunt resistor is shown in Figure 6. It
can be interpreted as an overlap of a more‐or‐less sinu-
soidal signal and short nonperiodic pulses being a
F I G U R E 5 A typical shrinkage plot of the 58‐mN/m test
ink spot on the high‐density polyethylene plate treated for 10 s
| 7 of 14
F I G U R E 6 Voltage signal as a function of time for
piezoelectric transformer operated with 8.3‐W power
response on microdischarges. The frequency of the
sinusoidal component corresponds to the second har-
monic of the resonant PT oscillation and can be char-
acterized by amplitude VOSC which is, as shown in
Section 2.2.2, directly related to the PT output voltage by
a complex attenuation factor. This signal can be observed
even if no plasma is present at the PT tip, when, for
example, the PT input voltage is too low for plasma ig-
nition or when the PT is casted in a silicon rubber.
The voltage pulses in Figure 6 are caused by changes
in the charge distribution at the tip of the PT. During
streamer propagation toward the capacitive probe, the
effective gap between PT and front electrode decreases.
Consequently, the capacitance Cg (see Section 2.2.2) in-
creases followed by a decrease or reactance XC and an
abrupt increase in the measured voltage. After extin-
guishing of the microdischarge, the measured voltage
returns to its quasi‐sinusoidal value.
As the lifetime of the microdischarge corresponds to
frequencies in GHz range, the phase angle of such signal
after passing the complex voltage divider is different
when compared with 50 kHz of piezo‐oscillation. If the
microdischarge signal is represented by a sinusoidal
voltage with frequency f = 0.1 GHz, equivalent to the
angular frequency of ω = 2π × 0.1 × 109/s, the attenua-
tion factor calculated according to Equation (3) will have
the magnitude of 0.8 and the phase angle of 28∘. Because
the attenuation of the microdischarge signal is on several
orders of magnitude weaker than for the piezo‐oscillation
signal, the magnitudes related to the microdischarges and
the piezo‐oscillations in Figure 6 cannot be directly
compared. Also, the phase position of the microdischarge
pulses in this plot does not represent the factual one due
to the frequency dependence of the phase angle.
8 of 14 | KORZEC
The impulse responses are lagging with respect to the
piezo‐oscillation signal by about 60∘. Also, the duration of
the voltage pulses does not mirror the duration of the
microdischarges due to dispersion. The only information,
which can be extracted directly from the microdischarge
responses, is their number per cycle. This value will be
used for evaluation of the PT efficiency in Section 3.1.2.
The positive and negative microdischarge responses
are different. Typically, the positive (anodic) peaks have a
higher magnitude than the negative (cathodic) ones.
Such asymmetry is due to the difference in the physics of
the positive and negative streamers.[58] A positive strea-
mer needs only a half as a large electric field as a negative
streamer for its propagation. It reaches also much larger
dimensions than the negative one. An important factor is
also the duration of the microdischarge pulses, which, as
shown before, has an influence on the signal attenuation;
the shorter the pulse, the larger the attenuation
magnitude.
3.1.2 | Influence of power
The amplitude of the piezo‐oscillation‐related component
of the voltage measured on the shunt resistor is, as shown
in Section 2.2.2, proportional to the PT output voltage.
The dependence of this amplitude on the input power of
the PT is displayed in Figure 7. For power higher than
1.5 W, the voltage amplitude increases monotonously
with power. Assuming the attenuation magnitude esti-
mated from Equation (3), the PT output voltage varies in
the investigated power range of 10 to 17 kV.
In Figure 7, the dependence of the number of anodic
microdischarges per cycle of the PT oscillation is plotted
as a function of PT input power. Each plot point re-
presents the mean value of numbers counted in 12 ran-
domly selected cycles. The higher the PT input power,
F I G U R E 7 Amplitude of the voltage signal and the mean
number of anodic microdischarges per cycle as a function of the
piezoelectric transformer (PT) input power
ET AL.
the larger the microdischarge number per cycle. Below
1.2 W, no microdischarges can be observed despite high
voltage amplitude still being related to the PT oscillation.
The value 1.2 W can be treated as an onset value for
plasma production. The microdischarge number per cy-
cle increases slower than linearly with power (see the fit
line in Figure 7) and reaches more than six micro-
discharges per cycle for 12 W.
3.1.3 | Spectral analysis
The analysis of the time‐dependent PT voltage shows that
its periodic component is not an ideal sinus function but
shows deformation. Expressed in the term of frequency‐
domain signal such deformation means that the spectrum
of the PT signal contains not only the resonant‐second
but also higher harmonics. For a better understanding of
the PDD physics, the time domain signals, collected as
described in Section 2.2, are used for calculation of their
frequency spectra using fast Fourier transform (FFT).[59]
The time–domain data vector consisting of 5,000,000
voltage samples, each of 0.4 ns duration, obtained with
PT input power of 8 W, is used for calculation of the
vector in frequency domain ranging from 500 Hz to
10 MHz with 500‐Hz step. The excerpt from 500 Hz to
500 kHz out of this spectrum is shown in Figure 8.
The spectrum reached the maximum magnitude at
the second harmonic frequency described in Sec-
tion 2.1.3. All harmonics seen in the spectra are the even
ones (Figure 2) because only the standing waves of even
harmonics show strong charge variations at the end of
the PT and can be detected by the capacitive probe. All
higher harmonics are much weaker than the second one.
Considering the entire spectrum (not shown in Figure 8),
the spectrum magnitudes for frequencies above 8.5 MHz
are below the noise level.
3.1.4 | Plasma detection
The deformation of the originally sinusoidal signal, as
displayed in Figure 6, is due to the influence of plasma on
the PT oscillation. The presence of plasma causes an in-
crease in the PT load capacitance. But this increase is
present only when the microdischarges are ignited.
Consequently, the load impedance is variable within each
oscillation cycle, resulting in generation of higher har-
monics in the piezoelectric oscillation, getting visible in
the measured voltage signal.
A strong influence of plasma presence on the fre-
quency spectra in kHz and MHz range is demonstrated in
Figure 9. An increase in the output capacitance caused by
KORZEC ET AL. | 9 of 14

plasma reduces the resonant frequency and the voltage

transformation ratio of the PT.[44] Therefore, the max-
imum related to the second harmonic frequency shifts
from 53.0 to 51.5 kHz, as shown in Figure 9a. The mag-
nitude of the second harmonic is lower for signal of PT
without plasma because such conditions can be estab-
lished only for very low power (Figure 7).
Figure 9b shows the influence of plasma on the
spectrum at harmonic frequencies higher than the sec-
ond. In the spectrum of PT oscillating without plasma, all
spectral line magnitudes are much smaller and vanish for
frequencies higher than 2 MHz. The plasma‐influenced
PT oscillation generates harmonics in the entire fre-
quency range observed, up to 5 MHz. The largest mag-
nitude in the no‐plasma spectrum at sixth harmonic
F I G U R E 8 Excerpt of a typical frequency spectrum of the frequency is about six times smaller than the corre-
piezoelectric transformer (PT) operated with a PT input power sponding magnitude in the plasma spectrum.
of 8 W To separate the influence of power level and the
plasma presence, the magnitude ratio of the sixth and
second harmonic frequencies was investigated in a broad
(a) power range. The results are shown in Figure 10. The
spectrum magnitude at the sixth harmonic frequency
increases on one order of magnitude with the PT input
power increasing from plasma ignition value to 12 W. At
the same time, the magnitude of the second harmonic is
only doubling. Consequently, the second‐to‐sixth har-
monic magnitude ratio is very high for very small power
values. The power range in which this ratio is larger than
60, correlates with the range with no or very few mi-
crodischarges, as shown in Figure 7. The curve trends of
the microdischarge number and the sixth harmonic
magnitude versus power correlate to each other. The
practical significance of this fact is that the magnitude at
the sixth harmonic frequency can be used for detection of
the plasma presence or even for evaluation of the plasma
(b) intensity.

F I G U R E 9 Frequency spectra with and without plasma: (a) F I G U R E 1 0 The magnitude of the sixth harmonic and the
Second‐harmonic only and (b) higher harmonics without the second‐to‐sixth harmonic magnitude ratio as a function of the
second one piezoelectric transformer (PT) input power
10 of 14 | KORZEC
3.2 | Ozone production
In this section, the influence of PT power and airflow on
the ozone production rate was investigated. The drift of
the ozone concentration was analyzed to avoid erroneous
measurements.
3.2.1 | Ozone concentration drift
After switching on the PT power supply, a rapid increase
in the ozone concentration can be observed, as shown for
five CDA flows in Figure 11. For flows below 6 SLM the
maximum concentrations are reached after about 60 s.
Then, the ozone concentration decreases with time. The
lower the flow, the stronger the drop of ozone con-
centration and the longer the time it takes to reach a
stable value. For 3 SLM, the concentration drops on 15%
from its maximum value after 5 min from switching on
the power and the asymptotic value is not reached yet.
For 5 SLM, the drop is less than 5% and the stable value is
reached after 3 min. The proposed physical reason for
this effect is an increase in the PT temperature. The PT is
heated mainly due to the mechanical and also, electro-
static losses.[60] As the PT block is made of PZT ceramics
with poor thermal conductivity (≈1.4 W⋅m−1⋅K−1),[61] and
high specific heat (≈100 J⋅mole−1⋅K−1)[62] at room tem-
perature, it takes several minutes for the heat from the
inner part of the PT to reach the surface and a thermal
balance can be established. The time needed to reach the
thermal equilibrium increases with a decrease in CDA
flow due to degraded cooling of the PT. The ozone con-
centrations shown in further sections are measured after
60 s from cold start, to minimize the influence of the PT
drift.
F I G U R E 1 1 Concentration of ozone as a function of
operation time for five compressed dry air flows. The
piezoelectric transformer power is 8.3 W
ET AL.
F I G U R E 1 2 Concentration and production rate of ozone as
a function of the compressed dry air (CDA) flow
3.2.2 | Ozone dilution
The concentration of ozone as a function of CDA flow is
presented in Figure 12. According to the fitting curve of
this plot, the ozone concentration decreases almost in-
versely proportional to the CDA flow. The higher the
flow, the stronger the dilution grade of ozone, and
the lower the expected activation efficiency. To maximize
the activation efficiency, the ozone concentration must be
maximized by minimizing the airflow. The limiting factor
for minimization of the airflow is the PT cooling
requirements, which are dependent on the power
coupled into the system.
Using the ozone concentrations from Figure 12 with
Equation (7), the production rate was calculated and
plotted versus airflow in Figure 12. For the flow over
5 SLM, only minor variation of the production rate was
observed. With flow decreasing to below 5 SLM, the
ozone production rate decreased. This can be explained
by a PT temperature increase due to insufficient cooling.
The temperature rise causes a decrease in the piezo-
electric oscillation quality and consequently, in the vol-
tage transformation ratio.[63] The flows below 3 SLM
were not investigated to avoid the risk of PT thermal
damage such as the unsoldering of the electric connec-
tions, local depolarization of the PZT, or thermo-
mechanical break of the PT.
3.2.3 | Influence of power on ozone
production rate
In Figure 13, the production rate as a function of the PT
input power is shown for the CDA flow of 8 SLM. The
production rate increases with power in the entire in-
vestigated power range. It reaches the maximum value of
73 mg/hr for 8 W. The trend of this curve is very similar
KORZEC ET AL.
F I G U R E 1 3 Ozone production rate for the compressed dry
air gas flow of 8 SLM and activation area as a function of the
piezoelectric transformer (PT) input power. Distance between
PT tip and the substrate surface is 4.5 mm. The treatment time
is 10 s
to that of the mean value of the microdischarge number
per cycle as a function of power (Figure 7).
The increase in ozone production rate is faster for
power values below 5 W and slows down for higher va-
lues, which can be interpreted as a loss of the ozone
production efficiency (production rate per energy unit).
This effect can be explained by increasing the tempera-
ture of the PT tip and the resulting loss of the PT voltage
transformation ratio. An additional mechanism, which
should be considered, is the enhanced decomposition of
the ozone with increasing temperature.
3.3 | Surface activation
The surface activation efficiency, described by activation
area and activation speed, depends on the operating
conditions of the plasma tool. The major control para-
meters are (i) the distance from the substrate, (ii) the
time of interaction between plasma and the surface, and
(iii) the power coupled into the plasma. The influence of
these parameters is investigated in the following sections
of this study.
3.3.1 | Influence of the substrate
distance
The dependence of the activation area on the distance
between the PT tip and the substrate surface is shown in
Figure 14. The activation area decreases with the distance
increasing over 3.5 mm. This tendency can be explained
by three mechanisms: (i) Increasing dilution of the che-
mically active species with increasing distance from the
| 11 of 14
F I G U R E 1 4 Dependence of the activation area on the
distance between the substrate and the tip of the piezoelectric
transformer (PT). The PT input power is 8 W. The treatment
time is 10 s
nozzle opening blowing the plasma gases, (ii) a decrease
in the reactivity of the chemically active species in
the airflow with elapsing time, and (iii) a decrease in the
electric field at the substrate surface with distance from
the PT tip. The maximum of the activation area is
reached at a distance of 3.5 mm from the PT tip. The
slight increase in the value for the distance of 1.5 mm can
be explained by chocking the airflow by small slot be-
tween the substrate and the edge of the nozzle (see the
position of the nozzle edge in Figure 14). For distance
greater than 13 mm, the activation spot splits into two
small subspots, which can be considered as insufficient
activation.
3.3.2 | Dependence on treatment time
A further parameter, which can be set for static treat-
ment, is the treatment time. The activation area increases
monotonously with treatment time, as shown in
Figure 15. However, a saturation tendency for longer
treatment time is observed. It can be explained by
mechanisms restricting a radial spreading of the chemi-
cally active species, which are subjected to dilution and
chemical losses. Another factor to be considered is the
decrease in the electric field at the substrate surface with
increasing radial distance from the PT tip.
3.3.3 | Activation speed
For dynamic treatment, the activation speed can be cal-
culated as a product of the activation width and the
movement velocity, where movement refers either to the
plasma tool or the substrate. The static treatment requires
12 of 14 | KORZEC
F I G U R E 1 5 Dependence of activation area and activation
speed on the treatment time. The piezoelectric transformer's
(PT) input power is 8 W. The distance between the tip of PT and
the treated surface is 4.5 mm
different activation speed definition. It can be ex-
pressed as:
Streat (8)
vact = ,
ttreat
where Streat and ttreat are the static activation area and the
treatment time, respectively.
As the activation area increases slower than linearly with
time, the activation speed is higher for shorter treatment
times, as demonstrated in Figure 15. The activation speed is
about 0.5 cm2∕s for 10‐s treatment and increases to 4 cm2∕s
for 0.5‐s treatment. For some treatment time 0 < t0 < 0.5 s,
the activation should vanish and the activation speed should
reach zero. Between this point and the high value for
ttreat = 0.5 s, an activation speed maximum can be expected.
3.3.4 | Influence of the PT input power
Figure 13 shows that the activation area increases almost
linearly with power, about 51 mm2 per 1 W in the power
range of 3.6 to 8 W. The monotonous increase of the
activation area with power correlates with those of
the microdischarge number per cycle (Figure 7) and of
the ozone concentration (Figure 13).
4 | CONCLUSION
The performance of the piezoelectric direct discharge‐
based plasma jet was evaluated using three methods:
Capacitive probe measurement, ozone concentration
measurement, and activation area evaluation.
The capacitive probe aided in the estimation of the PT
output voltage reaching about 15 kV, and in determining
ET AL.
the mean number of anodic microdischarges per cycle of
piezoelectric oscillation reaching 4.5 per cycle, both for
8‐W PT input power.
A strong influence of the microdischarges on the
frequency spectrum of the PT voltage was observed. The
plasma presence caused the shift of the main resonance
frequency of the PT to lower values and the occurrence of
additional spectral lines of higher harmonics. The mean
number of anodic microdischarges per cycle correlates
with the magnitudes of the sixth harmonic.
The chemical activity of the generated plasma was
evaluated on the base of the ozone concentration. It in-
creases with power and decreases almost inversely pro-
portional to the CDA flow. The production rate for 8‐W
PT input power reached 75 mg/hr and is only weakly
dependent on airflow.
The LDPE surface activation area correlates with the
microdischarge number per cycle of piezoelectric oscil-
lation and with the production rate of ozone, which all
increase with PT input power. The growth of activation
area is linear, showing the growth rate of 51 mm2∕W .
The activation area increases slower than linearly
with treatment time resulting in high treatment speed
values for short treatment times. For 0.5 s, the treatment
speed of 4 cm2∕s was reached, which is 10 times faster
than for 10‐s treatment.
The surface activation was possible at a distance of up to
13 mm from the PT tip. The largest activation area, five
times larger than for the distance of 13 mm, was reached at
a distance ranging from 1.5 to 3.5 mm and is about 4 cm2.
The plasma‐generating devices based on piezoelectric
transformers allow for a very reproducible activation re-
sults and are proven to be a suitable reference for acti-
vation area determination.
For the first time, the characteristics of the PDD‐
based plasma jets are disclosed, which can be of high
interest to members of the fast‐growing community of
PDD users, but they can also be a source of useful in-
formation for researchers intending to apply the PDD in
their own setups.[64]
ACKN OWLEDGMENTS
The authors thank Markus Puff from TDK Electronics
GmbH, Austria, for providing the CeraPlas™ F devices
and CeraPlas™ drives used in this study.
ORCID
Dariusz Korzec http://orcid.org/0000-0002-6297-0163
RE FER E NCES
[1] A. J. Moulson, J. M. Herbert, Electroceramics: Materials,
Properties, Applications, 2nd ed., John Willey & Sons Ltd.,
Chichester, England 2003, p. 340.
KORZEC ET AL.
[2] A. V. Carazo, Actuators 2016, 5, 1.
[3] C. A. Rosen, K. A. Fish, H. C. Rothenberg, US Patent
2830274, 1958.
[4] S. Bronstein, Ph.D. Thesis, Ben‐Gurion University of the Ne-
gev, Department of Electrical and Computer Engineering
(Beer‐Sheva, Israel) 2005.
[5] R.‐L. Lin, Ph.D. Thesis, Virginia Polytechnic Institute and
State University (Blacksburg, VA) 2001.
[6] P. J. M. Smidt, J. L. Duarte, PESC '96 record: 27th Annual
IEEE Power Electronics Specialists Conference, Institute of
Electrical and Electronics Engineers, Piscataway, New Jersey,
US 1996, pp. 310–315. http://books.google.com/books?id=
3UNGAQAAIAAJ. Accessed July 2020.
[7] E. Wells, Analog Appl. J. 2002, 1Q, 12.
[8] Y.‐K. Lo, C.‐H. Lin, K.‐J. Pai, IEEE International Sympo-
sium on Industrial Electronics, 2006 July 2006, Institute of
Electrical and Electronics Engineers, Montréal, Québec,
Canada 2006, pp. 892–896. http://ieeexplore.ieee.org/
servlet/opac?punumber=4035454
[9] K. Teranishi, S. Suzuki, H. Itoh, Jap. J. Appl. Phys. 2001, 40,
5766.
[10] K. Teranishi, S. Suzuki, H. Itoh, Jap. J. Appl. Phys. 2004, 43, 9733.
[11] H. Itoh, K. Teranishi, S. Suzuki, Plasma Sources Sci. Technol.
2006, 15, S51.
[12] K. Teranishi, N. Shimomura, S. Suzuki, H. Itoh, Plasma
Sources Sci. Technol. 2009, 18, 045011.
[13] M. Johnson, M. MacDonald, D. B. Go, Proc. ESA Annu. Meet.
Electrostatics 2015, California State Polytechnic University,
Pomona, CA 2015.
[14] M. Johnson, D. B. Go, J. Appl. Phys. 2015, 118, 1.
[15] M. Johnson, D. B. Go, Front. Mech. Eng. 2016, 2, 1.
[16] B. T. Hutsel, Ph.D. Thesis, University of Missouri (Columbia,
MO) 2012.
[17] A. Benwell, S. Kovaleski, M. Kemp, 2008 IEEE International
Power Modulators and High‐Voltage Conference, 2008, In-
stitute of Electrical and Electronics Engineers, Las Vegas, NV
2008, pp. 113–116.
[18] Y.‐H. Hsu, C.‐K. Lee, W.‐H. Hsiao, Smart Mater. Struct. 2003,
12, 373.
[19] T. Martin, F. Pigache, C. Nadal, T. Callegari, 2012 IEEE
International Ultrasonics Symposium, Dresden, Germany,
Oct. 7‐10, 2012, Institute of Electrical and Electronics
Engineers, 2012, pp. 1–5. http://ewh.ieee.org/conf/ius_2012/
Accessed July, 2020.
[20] M. Johnson, D. R. Boris, T. B. Petrova, S. G. Walton, IEEE
Trans. Plasma Sci. 2019, 47, 434.
[21] M. Teschke, J. Engemann, Proc. 18th Int. Symp. Plasma Chem.
(ISPC), International Plasma Chemistry Society, Kyoto, Japan
2007, pp. 1–4.
[22] M. Teschke, Contrib. Plasma Phys. 2009, 49, 614.
[23] M. Teschke, Ph.D. Thesis, Bergische Universität Wuppertal
(Wuppertal, Germany) 2009.
[24] H. Kim, A. Brockhaus, J. Engemann, Appl. Phys. Lett. 2009, 95,
211501.
[25] J. Engemann, M. Teschke, US Patent Application US 2009/
0122941 A1, 2009.
[26] M. Bisges, T. Krüger, European Patent EP 2 141 968
B12008, 2008.
| 13 of 14
[27] S. Neuhaus, Ph.D. Thesis, ETH Zürich (Zürich, Switzerland) 2011.
[28] B. Stawarczyk, H. Thrun, M. Eichberger, M. Roos, D. Edelhoff,
J. Schweiger, P. Schmidlin, Eur. J. Prosth. Dentis. 2015, 114,
666.
[29] Y. Ito, T. Okawa, T. Fukumoto, M. Tanaka, J. Osaka Dent.
Univ. 2016, 50, 79.
[30] Y. Ito, T. Okawa, T. Fukumoto, A. Tsurumi, M. Tatsuta,
T. Fujii, J. Tanaka, M. Tanaka, J. Prosthodon. Res. 2016, 60,
289.
[31] S. Kemmerling, J. Ziegler, G. Schweighauser, S. A. Arnold,
D. Giss, S. A. Müller, P. Ringler, K. N. Goldie, N. Goedecke,
A. Hierlemann, H. Stahlberg, A. Engel, T. Braun, J. Struct.
Biol. 2012, 177, 128.
[32] S. Neuhaus, C. Padeste, N. D. Spencer, Plasma Processes
Polym. 2011, 8, 512.
[33] I. Kartashev, T. Vontz, H. Florian, Measure. Sci. Technol. 2006,
17, 2150.
[34] I. Kartashev, T. Vontz, Measure. Sci. Technol. 2009, 20, 055108.
[35] TDK Electronics, Cold plasma from a single component, 2014,
Applications & Cases. https://www.tdk-electronics.tdk.com/en/
373562/tech-library/articles/applications-cases/applications-
cases/cold-plasma-from-a-single-component/1109546. Accessed
July, 2020.
[36] S. Nettesheim, D. Korzec, D. Burger, G. Kuegerl, M. Puff,
F. Hoppenthaler, US Patent 9788404B2, 2017.
[37] J. R. Büttler, T. Pham, Proc. Europe/Africa Conf., AIP Conf.
Proc. 2055, Polymer Processing Society PPS, Dresden,
Germany 2017, p. 050008–1–5.
[38] D. Ujino, H. Nishizaki, S. Higuchi, S. Komasa, J. Okazaki,
Appl. Sci. 2019, 9, 1.
[39] U. Kogelschatz, P. Baessler, Plasma Chem. Plasma Process.
2003, 23, 1.
[40] B. Eliasson, U. Kogelschatz, IEEE Trans. Plasma Sci. 1991, 19,
309.
[41] R. S. Sigmond Electrical Breakdown of Gases (Eds: J. M. Meek,
J. D. Craggs), John Wiley & Sons, Ltd., New York 1978,
pp. 319–383.
[42] W. E. U. Kogelschatz, B. Eliasson, J. Physique IV Colloque
1997, 07, C4.
[43] M. W. Hooker, Properties of PZT‐Based Piezoelectric Ceramics
between −150°C and 250°C, National Aeronautics and Space
Administration—NASA Technical Report NASA/CR‐1998‐
208708, 1998.
[44] M. Day, B. S. Lee, Analog Appl. J. 2002, 4Q, 12.
[45] C. A. Rosen Philips Application Book (Ed: H. W. Katz), John
Wiley & Sons, Inc., London, England 1959, pp. 170–197.
[46] A. J. Moulson, J. M. Herbert, Electroceramics, 2nd ed., John
Wiley & Sons Ltd., Chichester, England 2003, pp. 77–79.
[47] T. Ogawa Acoustic Wave Velocity Measurement on Piezoelectric
Ceramics F. Ebrahimi IntechOpen, London, UK 2013.
[48] A. M. Flynn, S. R. Sanders, IEEE Trans. Power Electron. 2002,
17, 8–14.
[49] H. Fukunaga, H. Kakehashi, H. Ogasawara, Y. Ohta, 29th
Annual Power Electronics Specialists Conference, 1998. PESC
98 Record, Vol. 2, Institute of Electrical and Electronics
Engineers, Inc., 1998, pp. 1504–1510.
[50] C.‐Y. Lin, Ph.D. Thesis, Virginia Polytechnic Institute and
State University (Blacksburg, VA) 1997.
14 of 14 | KORZEC
[51] EPCOS AG, Evaluation kit CeraPlas®HF Driver for Cer-
aPlas®series, 2018, Data sheet, https://www.mouser.de/datasheet/
2/400/ceraplas-driving-circuit-user-guide-1487527.pdf. Accessed
July, 2020.
[52] J. W. Nilsson, S. A. Riedel, Electric Circuits, 7th ed., Pearson/
Prentice Hall, Upper Saddle River, NJ 2004, p. 396.
[53] R. Dorai, M. J. Kushner, J. Phys. D: Appl. Phys. 2003, 36, 666.
[54] L. T. Molina, M. J. Molina, J. Geophys. Res. 1986, 91, 14,501.
[55] D. Daumont, J. Brion, J. Charbonnier, J. Malicet, J. Atmos.
Chem. 1992, 15, 145.
[56] ASTM D2578‐04a, Standard Test Method for Wetting Tension
of Polyethylene and Polypropylene Films, ASTM Interna-
tional, West Conshohocken, PA, 2004, http://www.astm.org.
Accessed July, 2020.
[57] J. L. Pérez‐Díaz, M. A. Álvarez‐Valenzuela, I. Valiente‐Blanco,
S. Jimenez‐Lopez, M. Palacios‐Cuesta, O. Garcia, E. Diez‐
Jimenez, J. Sanchez‐Garcia‐Casarrubios, C. Cristache, Proc.
ASME, Sand Diego, CA, November 2013.
[58] R. L. Burden, J. D. Faires High Voltage Engineering:
Fundamentals–Technology–Applications (Ed: VDI), Springer
Vieweg, Berlina/Heidelberg, Germany 2017, p. 187.
[59] R. L. Burden, J. D. Faires, Numerical analysis, Brooks/Cole,
Thomson Learning, Pacific Grove, CA 2001, pp. 537–548.
ET AL.
[60] D. M. Cuong, Piezoelectric Transformer Integration Possibility
in High Power Density Applications, Verlag der Wissenschaften
GmbH, Dresden, Germany 2008, pp. 25–38.
[61] S. N. Kallaev, G. G. Gadzhiev, I. K. Kamilov, Z. M. Omarov,
S. A. Sadykov, L. A. Reznichenko, Phys. Solid State 2006, 48, 1169.
[62] S. B. Lang, J. C. Lashley, K. A. Modic, R. A. Fisher, W. M. Zhu,
Z. G. Ye, 10th IEEE Int. Conf. Solid Dielectrics, 4‐9 July, In-
stitute of Electrical and Electronics Engineers, Potsdam,
Germany 2010, pp. 1–3.
[63] Y.‐H. Su, Ph.D. Thesis, L'Ecole Normale Superieure de Cachan
(Cachan Cedex, France) 2014.
[64] K. V. Artemev, N. N. Bogachev, N. G. Guseinzade,
T. V. Dolmatov, L. V. Kolik, E. M. Konchekov, S. E. Andreev,
Russ. Phys. J. 2020, 62, 2073.
How to cite this article: Korzec D, Hoppenthaler
F, Burger D, Andres T, Nettesheim S. Atmospheric
pressure plasma jet powered by piezoelectric direct
discharge. Plasma Process Polym. 2020;e2000053.
https://doi.org/10.1002/ppap.202000053