D 63612
Vol. 10 Number
12
December 2013
www.plasma-polymers.org
Editors-in-Chief
Riccardo d’Agostino, Bari
Pietro Favia, Bari
Christian Oehr, Stuttgart
Michael R. Wertheimer, Montreal
ISSN 1612-8850 Plasma, 10, No. 12 (2013)
Full Paper
Examining the Role of Ozone in Surface
Plasma Sterilization Using Dielectric Barrier
Discharge (DBD) Plasma
Navya Mastanaiah, Poulomi Banerjee,y Judith A. Johnson, Subrata Roy*
Dielectric barrier discharge (DBD) devices are known ozone generators. Authors have
previously demonstrated a DBD surface plasma source, operating in air at atmospheric
pressure, to achieve killing of vegetative cells in 2–3 min and sterilization in 20 min (bacterial
spores). The aim of this paper is to examine the role of the ozone in surface DBD plasma
sterilization. The role of ozone in plasma killing is examined by a) characterizing the rate of
production/decay of ozone during DBD plasma
generation, b) studying the effect of exposing
bacteria (Escherichia coli) solely to the ozone thus
produced. Our results indicate that while ozone
plays a major role, the energy flux delivered to
the electrodes is also crucial in the process of
plasma sterilization.
1. Introduction
Plasma sterilization offers advantages (short processing
times, low operational temperatures, safety, and versatility) compared to conventional sterilization methods such
as autoclaving, ethylene oxide fumigation, etc. Plasma
sterilization can be classified into two types: (i) volume
plasma sterilization, wherein plasma is generated between
a powered and grounded electrode and contaminated
samples are placed in between, so that they are immersed
in the plasma, (ii) surface plasma sterilization, wherein
plasma is generated atop a surface (made of a dielectric
N. Mastanaiah, P. Banerjee, S. Roy
Department of Mechanical and Aerospace Engineering, Applied
Physics Research Group (APRG), University of Florida, Gainesville,
FL 32611-6300, USA
E-mail: roy@ufl.edu
J. A. Johnson
Department of Pathology, Immunology and Laboratory Medicine,
College of Medicine and Emerging Pathogens Institute, University
of Florida, Gainesville, FL 32610-0009, USA
y
Present address: RCF Colony, Chembur, Mumbai 400074, India.
1120
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
material) embedded with electrodes on either side, such
that one is powered and the other is grounded. In the latter,
contaminated samples are placed directly atop the surface
on which plasma is generated.
Earlier experiments in plasma sterilization were mostly
conducted in the low to medium pressure ranges, wherein
the role of vacuum UV (VUV) radiation (wavelength
<200 nm) in the process of plasma sterilization was
debated.[1] During plasma generation at atmospheric
pressure, the spectroscopic signature exhibits prominent
intensity peaks in the wavelength range of 300–400 nm.[2,3]
The intensity peaks in this wavelength range pertain
mostly to transitions taking place in the 2nd positive
system of nitrogen (N2). More importantly, the spectroscopic signature of plasma at atmospheric pressure shows
no intensity peaks in the wavelength range of 100–280 nm,
which pertains to the UV-C regime. UV-C wavelengths are
most lethal to bacterial species because of their ability to
induce dimerization in the DNA of bacteria, thus arresting
bacterial replication.[4] However, the absence of UV-C
wavelengths in the spectroscopic signature of plasma at
atmospheric pressure implies that cell damage due to UV
radiation is likely not a major factor in plasma sterilization.
wileyonlinelibrary.com
DOI: 10.1002/ppap.201300108
Examining the Role of Ozone in Surface Plasma Sterilization
The other agents that might play a role in plasma
sterilization are reactive chemical species (charged particles,
neutrals)[5,6] and temperature.[7] The role of each of these
agents in plasma sterilization remains highly debated.
The type of plasma discussed in this paper and in our
previous sterilization experiments[3] is known as dielectric
barrier discharge (DBD) plasma. This type of plasma is
generated by applying a potential difference between
two electrodes embedded on opposite sides of a dielectric
(insulator) surface and is characterized by a large number of
tiny filamentary micro-discharges.[8] The gap between
electrodes is of the order of a few millimeters. The ease of
generating DBD plasma at atmospheric pressure as well
as the ability of plasma generation on surfaces makes
DBD plasma ideal for applications involving surface
sterilization. However, DBD plasma devices are inherent
ozone generators. Hence, before DBD plasma is applied
for surface sterilization technologies, it is important to
characterize the trends of ozone production during DBD
plasma generation as well as understand the role of ozone
in DBD plasma-based sterilization. Understanding the
trends of ozone production during DBD plasma generation
will also provide an insight into methods required to control
the amounts of ozone produced.
Ozone formation in pure oxygen (O2) is a two-step
process that starts with the dissociation of O2 molecules by
the electrons in a micro-discharge[9]
O2 þ e ! 2O þ e
ð1Þ
Followed by a three-body reaction
O þ O2 þ M ! O 3 þ M
ð2Þ
where M is a third reaction partner. M can be a pure O2
species or a N2 species, acting as a catalyst. Ozone formation
in air is slightly different from ozone formation in pure O2,
since energy provided by the electrical discharge in air is
distributed between N2 and O2.[10] Hence in air, ozone
synthesis is also influenced by contributions from N2 in
the form of mostly atomic N, which reacts with O2 to
form atomic O, which further recombines with molecular
O2 to form ozone (as shown in Equation 2). Typically,
ozone formation in air takes longer than ozone formation in
pure O2.
Ozone decomposition is often coupled to ozone formation. The basic reactions for the process are[10,11]
Decomposition : O3 þ M , O2 þ O þ M
ð3Þ
Recombination : O þ O3 ! 2O2
ð4Þ
Typically, when ozone dissociates and creates atomic O,
the atomic O immediately reacts with molecular O2 present
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
to immediately form ozone. Thus, ozone formation
(Equation 2) is very fast and usually able to balance
ozone decomposition (Equation 3), if initial concentration
of ozone is low. The ozone decomposition time depends
exponentially on temperature and initial concentration of
ozone.[10]
Ozone is known to have bactericidal properties. Broadwater et al.[12] determined the minimum lethal concentration of ozone in water for three bacterial species, using
a contact time of 5 min. Lethal threshold concentrations
for Bacillus cereus, Bacillus megaterium, and E. coli were
determined to be 0.12, 0.19, and 0.19 mg L 1, respectively,
wherein 1 mg L 1 corresponds to 1 ppm of ozone.1
Efremov et al.[13] discussed reaction rate constants for
production and destruction of ozone formed during
plasma generation. They conjectured that the antiseptic
property of the excited dry air flowing out of a discharge
chamber was determined by its ozone concentration and
demonstrated that exposing microorganism concentrations to discharge-excited air, even for a short while,
substantially reduced their amount.
Dobrynin et al.[14] pursued a different approach in
isolating the role of ozone in plasma sterilization. They
measured the ozone concentration produced by a DBD
discharge in room air at 60% relative humidity as
28 ppm. Consequently, they used an ozone generator to
produce the same concentration of ozone and examined
the inactivation effect of this ozone concentration on
E. coli and skin flora. They noted no inactivation. Furthermore, they also compared inactivation efficiency in two
cases, one in which the measured ozone concentration was
zero (DBD plasma produced in gas by diluting a mixture of
nitric oxide and N2 with O2 gas, which inhibited ozone
production) and one in which the measured ozone
concentration was not zero (DBD plasma produced using
a mixture of pure N2 and O2). They noted that the
inactivation efficiency in both cases remained the same,
leading them to conclude that ozone does not play a major
role in bacterial inactivation.
Vaze et al.[15] used a dielectric barrier grating discharge
(DBGD) to study the inactivation of airborne E. coli inside a
closed air circulation system. They studied the effect of
charged and short-lived species and the effect of ozone.
They concluded from their experiments that ozone caused a
reduction in bacterial load, but it may not be one of the
major inactivating factors in the plasma. In contrast,
comparing plasma treatment of mammalian breast epithelial cells with ozone treatment, Kalghatgi et al.[16]
found that ozone treatment was qualitatively different
from non-thermal DBD plasma. They concluded that ozone
1
According to conversion factors found on http://www.ozonesolutions.com/info/ozone-conversions-equations.
www.plasma-polymers.org
1121
N. Mastanaiah, P. Banerjee, J. A. Johnson, S. Roy
treatment did not play a role in the observed effects of
plasma on mammalian cells.
Hence, the role of ozone and more broadly, the role of
reactive oxygen species (ROS) in plasma sterilization is
still widely debated. This paper concentrates on isolating
and examining the role of ozone in DBD surface plasma
sterilization. DBD plasma devices operating at 14 kHz and
low input voltage are used to generate plasma. These
plasma devices are enclosed within acrylic sterilization
chambers of different volumes. Rates of production/decay
of ozone during and after plasma generation using these
devices are measured. The dependence of this rate of
production/decay on the volume of the sterilization
chamber is also examined. Additionally, different substrates inoculated with E. coli are exposed to the ozone
produced during plasma generation. In doing so, the
average ozone levels as well as exposure times required
for significant reduction in bacterial concentration are
determined. When ozone production during plasma
generation is inhibited through two different methods,
the reduction in E. coli concentration is not significant
leading us to conclude that ozone produced during
plasma generation is indeed responsible for bacterial
inactivation. The remainder of the paper is written as
follows: Section 2 lists the experimental protocols and
materials used. Section 3 presents and discusses the
results of the various studies described above and Section 4
draws conclusions.
2. Experimental Protocols
2.1. Plasma Generation and Diagnostics
The experimental setup used for DBD plasma generation in this
paper has been described in detail in a previous paper.[3] The plasma
device consists of a dielectric layer (1.6 mm thick), either side of
which an electrode is embedded. Two types of dielectric material
are tested: FR4 (flame retardant 4) and a hydrocarbon ceramic
laminate (RogersW 3003 C), which will henceforth be known as
semi-ceramic (SC). FR4 is commonly used for making printed circuit
board and has a dielectric constant (e) of 4.7, while SC has a dielectric
constant of 3.00 0.04 (according to manufacturer specs). The
electrodes embedded on either side of this dielectric layer are made
of copper coated with a tin finish. The top (powered) electrode has a
comb-like pattern as shown in Figure 1 and is exposed to the air. The
bottom (grounded) electrode is a square sheet and measures the
same in surface area as the top electrode (2.4 2.4 cm2) and is not
exposed to air during plasma generation.
To generate plasma, as shown in Figure 1 above, a sinusoidal
wave of 14 kHz frequency is amplified such that the final input
signal being fed into the top (powered) electrode measures 12 kV
peak–peak (unless otherwise mentioned). The potential difference
between the top and bottom electrodes leads to DBD plasma
generation on the surface of the comb-like electrode, as shown in
Figure 1.
1122
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 1. An image of the top (powered electrode) of the DBD
plasma device, during plasma generation.
Ozone levels during and after plasma generation are measured
using a 2B TechW Ozone Monitor (Model 202). Measurement of
ozone is based on absorption of UV light (at 254 nm) and
subsequent comparison of the quanta of light reaching the detector
before and after absorption by ozone. Ozone levels are measured
in units of ppb (or ppm). Air inside the chamber is sampled every
10 s and measured ozone levels are saved to a computer via a
LabViewW interface.
2.2. Microbiological Testing
Cultures are maintained frozen at 80 8C in broth with 25% glycerol
and inoculated onto fresh plates weekly. E. coli C600 is grown
on Luria-Bertani (LB) agar or broth at 37 8C. Before each experiment,
the optical density (OD) of the microbial sample is measured
using an Ultrospec 10 cell density meter (GE Healthcare BioSciences Corp., Piscataway, NJ) to estimate the density of the
culture. An OD of 1 corresponds to approximately 5 108 colony
forming units (CFU). Cultures are diluted as needed to ensure
that approximately 106 CFU are inoculated onto any device. For
each experiment, the plasma devices are inoculated with 20 ml of
E. coli sample spread uniformly over the entire electrode surface
area with a sterile inoculating loop. Only the top (powered)
electrode of each plasma device is inoculated. After an experiment,
the number of remaining test organisms left on each device is
determined by spreading serial dilutions on appropriate agar
plates as described. Plate counts are also performed on the
inoculum to determine the exact concentration of organisms
and an inoculated device not exposed to ozone is processed as
control for the loss of viable counts due to drying or adherence
to the device. Experiments are performed in triplicate unless
otherwise noted.
3. Results and Discussion
3.1. Characterization of Trends of Ozone Production/
Decay During DBD Plasma Generation
In order to study the trend of ozone production/decay
during and after DBD plasma generation, a clean FR4
DOI: 10.1002/ppap.201300108
Examining the Role of Ozone in Surface Plasma Sterilization
plasma device is enclosed in acrylic sterilization chambers
of different volumes. In each chamber, the plasma device
is placed, powered for 2 min (120 s) and then allowed to
rest for an additional 5 min (300 s). Ozone levels are
measured every 10 s throughout this 7 min (420 s) interval.
The ozone probe used to sample ozone levels is placed 2.500
above the chamber floor and 500 to the right of the device
(measured from the center-point of the device).
The dimensions of each of these chambers are given
below. Here l, b, and h denote length, breadth, and height
respectively.
Chamber #1 – l ¼ 1200 ; b ¼ 1000 ; h ¼ 700 – 840 in3.
Chamber #2 – l ¼ 1200 ; b ¼ 1000 ; h ¼ 1400 – 1 680 in3.
Chamber #3 – l ¼ 2400 ; b ¼ 1000 ; h ¼ 1400 – 3 360 in3.
Chamber #4 – l ¼ 4800 ; b ¼ 23.500 ; h ¼ 2400 – 27 072 in3.
Thus, Chamber #2 is twice as high as Chamber #1 and
Chamber #3 is twice as long as Chamber #2. Chamber #4 is
huge in all three dimensions, as compared to the other three
chambers. The volume ratio of the chambers is 1:2:4:32
with respect to the smallest chamber (#1). Three types of
aspect ratios (AR) can be defined for these chambers – l:b, l:h,
and b:h denoted by AR1, AR2 and AR3, respectively For
Chamber #2 and #3, AR3 is the same while AR1 and AR2
for Chamber #3 are double that of Chamber #2. Figure 2
shows the production/decay profiles obtained in
this manner for all four sterilization chambers using the
same DBD device.
In Figure 2, seven different plots are shown. The first
four plots (only markers) show the variation of measured
ozone concentrations throughout the 7-min (420 s) interval
in the four different chambers. For these four plots, error is
plotted as standard deviation of measured ozone levels at
each time point. A closer study of these four plots enables
the identification of three distinct phases during and after
plasma generation in each chamber. These three distinct
phases are labeled in the figure as the ozone production
phase, the ozone diffusion phase, and the ozone decay
phase. In the production phase, while plasma is generated
for 2 min (120 s), measured ozone concentrations show a
gradually increasing trend. As soon as the plasma is turned
off at 120 s, a sharp dip in measured ozone concentration is
observed from Figure 2. This sharp decrease in measured
ozone concentration lasts for a very short time (30 s) and
is therefore identified as the diffusion phase. After 160 s,
Figure 2 shows that the measured ozone concentration in
each chamber gradually begins to decrease (as opposed
to the sharp reduction in ozone concentration at 120 s).
Appropriately, this phase is identified as the decay phase.
Essentially, in this phase, produced ozone slowly decomposes. Diluted ozone at room temperature is quite stable[10]
and hence decomposes on the timescales of minutes. We
delineate the two phases after t ¼ 120 s as diffusion and
decay (or decomposition) phase based on the timescales
of decrease in ozone concentrations. During the diffusion
phase, a very sharp reduction in ozone concentration
occurs whereas during the decay phase, the decrease in
Figure 2. Comparison of ozone concentrations produced during and after plasma generation in all four chambers. Plasma device is powered
at 0 s and turned off at 120 s, after which setup is allowed to rest for another 300 s.
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.plasma-polymers.org
1123
N. Mastanaiah, P. Banerjee, J. A. Johnson, S. Roy
Table 1. Values of the constants used in Equation (5a–c).
Chamber #
a [ppm]
1
b [ppm]
197
c [s]
186.4
[O3]0 [ppm]
89.13
0.073
168.8
2
64.85
58.57
47.78
0.028
31.3
3
48.66
36.72
21.4
0.062
24.91
ozone concentration is gradual. The timescale of reduction
in ozone concentration is similar to reported trends[11]
leading us to believe that the sudden decrease in ozone
concentrations during the diffusion phase is more likely
due to flow effects (discussed later).
Analyzing the measured ozone concentrations in
each phase, using trend-fitting tools, each phase can be
further identified by a characteristic equation. These
Equation (5a–c) are listed below.
Production phase
½O3 ¼ a
be
t=c
0 s t 120 s
ð5aÞ
Diffusion phase
Decay phase
d2 ½O3
¼ D ðppm s 2 Þ 120 s < t 160 s
dt2
ð5bÞ
½O3 ¼ ½O3 0 e
t=t
160 s < t 420 s
ð5cÞ
In the above equations, [O3] is the calculated ozone
concentration and [O3]o is the ozone concentration present
initially at the beginning of the decay phase. Both are
expressed in units of ppm, where 1 ppm is equal2 to a
density of 2.648 1013 cm 3. In Equation (5a) (production
phase), ‘‘a’’ and ‘‘b’’ are production coefficients (ppm) while
‘‘c’’ is the production time constant (seconds). In
Equation (5b) (diffusion phase), ‘‘D’’ is the diffusion
coefficient (ppm s 2). In Equation (5c) (decay phase), ‘‘t’’ is
the decay time constant (seconds). The values of all these
coefficients and time constants for each chamber except
Chamber #4 are listed below in Table 1. Calculating
and plotting the ozone concentration versus time using
these equations for Chambers #1–#3 gives rise to the
latter three plots shown in Figure 2, denoted by the word
‘‘curve fit’’ at the end.
In Figure 2, it is observed that ozone concentrations
calculated using Equation (5a) agree very well with
measured ozone concentrations in the case of Chambers #1
and #2. However starting with Chamber #3, the measured
ozone concentrations start to deviate from the expected
2
According to conversion factors found on http://www.lenntech.
com/calculators/ppm/converter-parts-per-million.htm.
1124
D [ppm s 2]
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
t [s]
410
618.047
1 017.087
trend. This deviation is very prominent in the production
phase for Chamber #3. As mentioned before, Chamber #3
is twice as long as Chamber #2 or #1. Thus it seems
that compared to Chamber #1, increasing the height of
the chamber does not affect the expected trend of ozone
concentrations, but increasing the length makes it more
difficult to accurately predict the ozone concentrations
according to Equation (5a). This is why coefficients or
constants have not been calculated for Chamber #4, due
to its much larger proportions as compared to the other
chambers.
Plotting the measured ozone levels with the corresponding chamber volume, at different time points during the
7-min (420 s) interval produces a correlation such as
the one given below in Figure 3. In Figure 3, Figure 6
time points have been plotted: 60 s, 120 s (during the
production phase), 160 s (during the diffusion phase)
and 240, 360, and 420 s (during the decay phase). Both
chamber volume (X-axis) and ozone levels (Y-axis) are
plotted on a log 2 scale for easier comparison.
There are some interesting points to be noted in Figure 3.
First as expected, at all times, measured ozone concentration in Chamber #1 is the highest. Secondly, during the
production phase, it seems that measured ozone levels at 60
and 120 s for both Chambers #2 and #3 are similar. Even
from Figure 2, observing ozone measurements during the
production phase for Chambers #2 and #3, it is noted that
the values for both chambers are very similar. Again, for
Chambers #2 and #3, AR3 is the same while AR1 and AR2
for Chamber #3 are double that of Chamber #2. Thus,
the similar levels of ozone noted during the production
phase for Chambers #2 and #3 in Figure 3 imply that
the ozone produced during the production phase is
dependent on AR3. Once the plasma is turned off, both
in the diffusion phase and the decay phase, Figure 3
once again reinforces the dependence of measured ozone
concentrations on the volume of the sterilization chamber.
However, this brings up another important point. The
same clean FR4 device, when operating at the same input
voltage and frequency (i.e., same input power), when
placed in different chambers gives rise to different
concentrations of measured ozone in each chamber. As
the ozone is being produced during the production phase,
the smallest chamber leads to confinement of the produced
ozone, hence enabling the ozone probe to detect it in
DOI: 10.1002/ppap.201300108
Examining the Role of Ozone in Surface Plasma Sterilization
Figure 3. Correlation of measured ozone levels with chamber volumes. The different plots at time points (60, 120, 160, 240, 360, and 420 s)
represent ozone measurements in each chamber at that particular time point.
very high concentrations. However as the volume of
the chamber is increased, the produced ozone has
more and more volume to diffuse (or spread) into. Thus,
an ozone probe placed at the exact same position in
each chamber begins to detect lesser and lesser concentrations of ozone as the chamber volume increases.
However from the discussion of expected trends above
(Equation 5a–c), it seems that the length of the chamber
is more important in determining how the ozone
produced during the production phase diffuses inside
the chamber, i.e., longer the length, higher the spread of
produced ozone and more difficult it is to predict the
expected ozone concentrations.
Much of the discussion of the results in Figure 2 and 3
indicate that the measurement of ozone concentration
during and after plasma generation is highly dependent on
the chamber volume. Since ozone levels discussed here
are being measured using a single ozone probe at a single
position, it is also important to evaluate the dependence of
measured ozone levels on the position of the ozone probe.
In order to do this, a plasma device is placed in the largest
Chamber #4. The device is powered over a time interval of
2 min and the ozone levels at different locations (along both
the X-axis and Y-axis) of the chamber are measured. The
various locations at which ozone is measured inside the
chamber are shown in Figure 4. For all measurements in
Figure 4, the ozone probe is positioned at a height of 500
above the floor of the chamber.
The aim of such an experiment was to get an idea of the
spatial variation of ozone levels inside Chamber #4. This
spatial variation is given below in Figure 5a and b. Figure 5a
shows the spatial variation of ozone production along the
X-axis of the chamber while Figure 5b shows the spatial
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
variation of ozone production along the Y-axis of the
chamber. While ozone data is sampled every 10 s, for the
sake of simplicity only the data sets at 30, 60, 90, and 120 s
have been plotted and shown.
Figure 5a demonstrates that the highest amount of
ozone is produced in the upper right quadrant of the
chamber (as labeled in Figure 4). Typically, the levels
of ozone noted on the right hand side (RHS) of the chamber
are more than those on the left hand side (LHS). Figure 5b
indicates that the distribution of ozone produced along
the Y-axis does not follow a clear-cut trend, as along the
X-axis. This bias in ozone levels toward the RHS of the
Figure 4. Schematic of Chamber #4. The gray square in the middle
represents the plasma device. The black short lines represent
the different locations at which ozone measurements are
taken. These locations are uniformly spaced (400 apart), along
the X- and Y-axis.
www.plasma-polymers.org
1125
N. Mastanaiah, P. Banerjee, J. A. Johnson, S. Roy
Figure 5. Spatial variation of ozone distribution inside the sterilization chamber. (a) Along the X-axis. (b) Along the Y-axis.
chamber can be explained due to the directional bias of
the electrodynamic body force (~
F ¼ q~
E ) produced as
a result of the plasma generation, where ‘‘F’’ is the electrodynamic force, ‘‘q’’ is the charge and ‘‘E’’ is the produced
electric field. Previous research conducted by our
group[17,18] elucidates further on the generation of this
electrodynamic force and its dependence on the various
plasma input parameters as well as its effect on fluid
momentum. Most importantly, the direction of this force is
from the powered electrode to the grounded electrode.
Because of this electrodynamic force (from left to right), the
flow is ‘‘pushed’’ to the right, as shown in Figure 6a. When
Figure 6. (a and b) The two different configurations in which the device is placed. It is noted that in (a), flow is ‘‘pushed’’ toward the right
(along the X-axis). In (b), when the configuration is flipped, flow is ‘‘pushed’’ toward the left.
1126
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/ppap.201300108
Examining the Role of Ozone in Surface Plasma Sterilization
the configuration is flipped, as shown in Figure 6b, the
direction of force is from right to left and thus the flow
would be pushed to the left in such a configuration.
This ‘‘pushing of the flow’’ effect may also explain the
sudden dip in ozone concentrations in the diffusion phase
(Figure 2). During plasma generation, the ozone levels
are continuously pushed toward the RHS of the chamber.
However at 120 s, as soon as plasma is turned off, due to the
sudden absence of a ‘‘pushing force,’’ produced ozone may
diffuse out toward the LHS of the chamber, thus causing a
sudden reduction in ozone concentrations.
In the above section, an overview of the trends of ozone
production/diffusion/decay during DBD plasma generation is provided. Once a FR4 plasma device is powered for
2 min (120 s) and left to rest, a plot of the produced ozone
concentrations versus time can be divided into three
distinct phases: production, diffusion, and decay. Each of
these phases can be defined by a characteristic equation.
However the predictability of measured ozone concentrations according to these characteristic equations is
dependent on the chamber dimensions. As the length of
the chamber increases, it becomes more difficult to predict
ozone concentrations using these characteristic equations.
Ozone concentrations measured during each phase are also
dependent on chamber dimensions. Three ARs, AR1–AR3
have been defined. Results (Figure 2–3) show that
during the production phase, ozone concentrations are
dependent on AR3 but during diffusion and decay
phases, they are dependent upon AR1. Not only are these
measured ozone concentrations dependent upon chamber
volume, but also upon the placement of the ozone probe
(Figure 5–6).
Hence, it is apparent that any substrate inoculated with
E. coli and placed at one constant position in each chamber
is exposed to varying concentrations of ozone. The next
section focuses on understanding the effect of exposing
E. coli concentrations to the produced ozone.
3.2. The Effect of Ozone Produced During DBD
Plasma Generation on E. coli
DBD plasma generation produces a plethora of species: UV
photons, charged particles, and neutrals. The role of each
of these species in the mechanism of plasma sterilization
is highly debated. However, one major factor that has
been considered is the ROS produced during DBD plasma
generation. Since DBD plasma produces high concentrations of ozone, it is necessary to first isolate and evaluate
the role of ozone in plasma sterilization.
To do so, different substrates are inoculated with E. coli
(20 ml corresponding to 106–108 CFU) and placed next to a
clean plasma device generating plasma. Both inoculated
and clean devices are placed inside Chamber #1. The
schematic for such tests is shown in Figure 7.
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 7. Experimental schematic for the exposure tests. On the
left is the plasma generator, which is used to generate plasma
and produce ozone inside the chamber. On the right is the
inoculated device/glass slide, which is inoculated with 20 ml of
E. coli and exposed to the ozone produced by the plasma
generator.
In the schematic below, the ‘‘plasma generator’’ is the
clean plasma device generating plasma and thus, ozone for
the required time. The ‘‘inoculated device/slide’’ represents
the substrates inoculated with E. coli. Two types of plasma
generators are tested: FR4 and SC plasma devices. Three
types of inoculated substrates are tested: a FR4 plasma
device, a SC plasma device, and a glass cover-slide. The
inoculated substrates are not exposed to direct plasma, but
instead to the reactive species produced during plasma
generation, especially ozone. Judging by ozone distributions in Figure 5, the ozone probe is positioned on the RHS
of the chamber and about 100 away from the inoculated
device.
On comparing measured ozone concentrations during
and after plasma generation, it is observed that the ozone
production in the case of FR4 is marginally higher than
that in the case of SC. This comparison is shown in Figure 8.
In Figure 8, similar production/diffusion/decay phases are
noted in the case of SC plasma devices. It is noted that the
average ozone concentration in the case of FR4 is 26.05%
higher than in the case of SC.
The three kinds of inoculated substrates described
above are exposed individually to ozone produced by both
a FR4 and SC plasma generator for 7 min (420 s). During
this 420 s interval, plasma is generated (i.e., ozone is
actively produced) for 2 min (120 s) and for the rest of
time, the inoculated substrate is simply exposed to the
residual ozone concentrations. The results of such an
experiment are given in Figure 9.
In Figure 9a, N denotes the number of E. coli survivors
after ozone exposure. Error is listed in terms of the
standard deviation of the mean of log10 (N) over a number
of trials. A single log10 reduction implies a 1/10th decrease
in E. coli concentration, i.e., 107 CFU reduces to 106 CFU.
Initial concentration of E. coli is 108 CFU. In Figure 9a,
for the FR4 plasma generator, two additional exposure
www.plasma-polymers.org
1127
N. Mastanaiah, P. Banerjee, J. A. Johnson, S. Roy
Figure 8. Comparison of ozone production during 7 min (420 s) for FR4 versus SC. The plasma device is powered at 0 s and turned off at 120 s,
after which the setup is allowed to sit for another 300 s.
Figure 9. Inactivation plots due to ozone exposure with (a) FR4 plasma generator and (b) SC plasma generator.
1128
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/ppap.201300108
Examining the Role of Ozone in Surface Plasma Sterilization
times were tested: 2 and 32 min. The results for the 2-min
time point are shown in Figure 9a while the results for
the 32-min time point are not shown. It is observed
that after 7 min of ozone exposure using a FR4 plasma
generator, E. coli concentrations in the case of the inoculated
FR4 and SC plasma devices are significantly reduced. E. coli
concentrations in the case of the inoculated glass slide are
not affected significantly. Even after a 32-min exposure
using a FR4 plasma generator, E. coli concentrations
are completely inactivated in the case of inoculated
FR4/SC plasma devices and not significantly affected in
the case of inoculated glass slide. However as is evident
from Figure 9b, when inoculated substrates are exposed to
ozone concentrations produced using a SC plasma generator, E. coli concentrations are not affected significantly.
The average ozone concentration is derived by calculating the average of measured ozone levels over a 420 s
interval and is used as an empirical parameter to compare
the dependence of bacterial inactivation on ozone concentrations. During a 7-min interval of exposure, any
inoculated substrate is exposed to an average of 90 ppm
of ozone in the case of FR4 plasma generator as compared to
an average of 67 ppm of ozone in the case of SC plasma
generator. Thus, an average of 90 ppm of ozone is required
for significant reduction (7 log10) in E. coli concentrations
exposed to ozone produced during DBD plasma generation.
In the case of inoculated glass slides, E. coli concentrations
are not significantly affected by ozone exposure, using both
a FR4 plasma generator as well as a SC plasma generator.
This indicates that the effect of ozone on E. coli is also
dependent on the type of substrate used for inoculation.
When glass slides are inoculated with 20 ml of E. coli, it is
visibly evident that the bacterial sample deposited on the
glass slide clumps into random droplets on the surface of
the glass slide. This is because glass is a far less hydrophilic
surface than FR4 or SC, thus making it difficult for liquid to
adhere to it. Hence this uneven clumping of E. coli on the
glass slide might be leading to shielding of the underlying
bacteria, which explains the insignificant drop in bacterial
concentration in the case of glass slides.
Figure 10 below shows inactivation results obtained
when the same ozone exposure experiment is repeated
with an inoculated FR4 plasma device, using a FR4 plasma
generator in different chambers. When an inoculated FR4
plasma device is placed at a constant position in different
chambers and exposed to ozone produced during plasma
generation, the inoculated substrate is exposed to varying
levels of ozone.
Again, E. coli concentrations in Chambers #2–#4 are
reduced only by 1–3 log10. E. coli concentrations are
only significantly reduced (7 log10) in Chamber #1. E. coli
concentrations are exposed to average ozone concentrations of 90, 41, 27, and 9 ppm in chambers #1–4
respectively over a 7-min (420 s) interval. Thus bacterial
inactivation results from both Figure 9 and 10 indicate
that there is a threshold level of ozone required for injuring
E. coli lethally.
The next step was to evaluate whether ozone produced
during plasma generation is truly responsible for the
bacterial inactivation noted in Figure 9a and 10. In order
to evaluate this, ozone produced was inhibited in the
following two ways: (i) using activated charcoal to inhibit
ozone production and (ii) generating plasma using N2 as
the discharge gas.
A fixed amount of activated charcoal (MarineLand Black
DiamondW) is placed on the plasma generator and the
plasma generator subsequently operated. In doing so, the
produced ozone is directly adsorbed by the charcoal and
ozone levels are immediately reduced by around 98%. For
subsequent tests, care is taken to adjust this amount of
activated charcoal on the plasma device to maintain the
same reduced levels of ozone.
Figure 10. Inactivation plots due to ozone exposure in the different chambers using a FR4 plasma generator and an inoculated FR4
substrate.
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.plasma-polymers.org
1129
N. Mastanaiah, P. Banerjee, J. A. Johnson, S. Roy
Figure 11. Comparison of ozone production with and without charcoal for Chamber #1. Ozone concentration on the Y-axis is shown on a
log10 scale.
This comparison of levels of ozone produced with and
without charcoal for Chamber #1 is shown in Figure 11. For
both cases, a clean FR4 plasma device is placed in Chamber
#1, powered at 0 min and turned off at 2 min (120 s), after
which the setup is allowed to rest for another 5 min (300 s).
Since there is a huge difference between ozone levels in both
cases, for the sake of simplicity, ozone concentration (on the
Y-axis) is shown on a logarithmic scale.
As is evident from Figure 11, the addition of charcoal on
top of the device during plasma generation leads to a
reduction of ozone concentration by a factor of 100. For all
the sterilization tests testing the effect of exposing E. coli
concentrations to ozone produced in cases with and
without charcoal, the amount of charcoal on the device
was adjusted to maintain the reduced ozone concentration
as shown in Figure 11.
As per protocol, an inoculated FR4 plasma device is
placed next to the plasma generator (a clean FR4 device)
covered with charcoal. For the purpose of this paper, such
a configuration will be referred to as a ‘‘modified plasma
generator.’’ Chamber #1 is used for these tests.
When the inoculated substrate is exposed to this
modified plasma generator for 7 min (420 s) and then
post-processed, negligible reduction in E. coli concentration
is observed, thus proving that the ozone produced during
plasma generation is responsible for inactivation of E. coli.
This inactivation effect with and without charcoal, in the
case of an inoculated FR4 substrate, is shown in Figure 12.
The wide disparity in bacterial inactivation is immediately
evident and proves that the reduced ozone concentrations
due to adsorption by activated charcoal do not have as
lethal an effect on E. coli, when compared to the case of no
charcoal. While Figure 11 demonstrates that ozone produced during DBD plasma generation is responsible for
bacterial inactivation, an alternative experiment to prove
this was also conducted.
Figure 12. Sterilization curve with and without charcoal.
1130
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/ppap.201300108
Examining the Role of Ozone in Surface Plasma Sterilization
DBD Plasma is generated using N2 as the working gas. For
the same device, plasma generation parameters such as
input frequency and voltage, and thus the generated
electric field, are kept similar. A smaller vacuum chamber
(9.300 7.400 5.600 ) is used as the sterilization chamber. The
chamber is then evacuated to an absolute pressure of
0.0978 atm, following which N2 gas (Airgas, Inc., UN1066,
99.0% N2) is introduced into the chamber until the pressure
in the chamber is restored to atmospheric pressure. This
process is repeated four times to maintain a majority N2
environment. The aim of such an intensive method of
flushing out all the air and filling the chamber with N2 is to
ensure that very low O2 levels remain in the chamber in
order to inhibit ozone production during DBD plasma
generation. Using laws of partial pressures, this percentage
of N2, at the end of four flushes is calculated to be
99.998 þ 0.78%. Hence the aim of maintaining a pure N2
environment is accomplished fairly well.
The experimental protocol in these tests, unlike the
previous ozone exposure tests, consists of inoculating
select FR4 devices with 20 ml of E. coli (initial concentration
¼ 108 CFU), placing them in the sterilization chamber,
sealing the chamber, flushing the chamber with N2 four
times and then powering the device for the requisite time
interval (Dt). Owing to the rigorous nature of these tests,
only FR4 substrate is tested. The time intervals tested are
Dt ¼ 60 s, 120 s. Each time interval test is replicated thrice,
using the same E. coli sample to ensure repeatability. The
comparison of sterilization results from plasma generation
using discharge gas as air versus N2 is given below in
Figure 13.
In Figure 13, in the case of air, complete bacterial
inactivation is noted within 120 s. In the case of N2, starting
from an initial concentration of 106 CFU, at the end of 120 s
of plasma generation, only a 1 log10 reduction in E. coli
concentration is noted. However, the initial concentration
of E. coli used is 108 CFU, which leads us to conclude that the
flushing of the chamber four times as well as evacuation of
the chamber to extremely low pressures causes a 2 log10
reduction in E. coli concentration. Nevertheless, even with a
pre-plasma concentration of 106 CFU, it is evident that
plasma generation in N2 (hence, plasma generation in the
absence of ozone) does not cause much of a reduction in
E. coli concentration. Thus from the results of the charcoal
tests as well as N2 tests, we conclude that ozone produced
during plasma generation is capable of inactivation of
E. coli, on prolonged exposure times.
Plasma generation in air produces a huge number of
reactive chemical species (both neutral molecules and
charged particles). Our experimental setup measures
concentrations of one of these reactive chemical species,
i.e., ozone. Since ozone is produced in such great quantities
during DBD plasma generation, it is prudent to ask what
role it plays in DBD plasma sterilization. In order to do that,
results with experiments involving the exposure of E. coli to
ozone generated during and after plasma production are
discussed. From the results of these experiments, it is
evident that a threshold level of ozone is required to kill
E. coli. This threshold level is calculated to be an average
ozone concentration of 90 ppm. Consequently, the next
prudent question to ask is whether the obtained results
indicate that the observed reduction in bacterial concentration is due to the produced ozone. Inhibiting produced
ozone (using activated charcoal) and repeating exposure
experiments leads to insignificant reduction in E. coli
concentrations, proving that ozone produced during and
after plasma exposure is indeed responsible for inactivating
E. coli. Additionally generating DBD plasma at atmospheric
pressure in a N2-filled environment, using inoculated
FR4 devices proves that sterilization due to DBD plasma
in a N2-filled environment is not as effective as sterilization
due to DBD plasma in air. This not only reinforces the theory
Figure 13. Plasma sterilization curve, using air and nitrogen as the working gases.
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.plasma-polymers.org
1131
N. Mastanaiah, P. Banerjee, J. A. Johnson, S. Roy
that the reduction of ozone (all the air in the chamber is
evacuated and filled with N2) leads to insignificant
reduction in E. coli concentrations but also proves that
N2, which is a major component of air and a prominent peak
on the spectroscopic signature[3] does not play a dominant
role in DBD surface plasma sterilization.
However from Figure 9, it is evident that after 2 min of
exposing E. coli concentrations to produced ozone, a 3–
4 log10 reduction is obtained, starting from an initial
concentration of 108 CFU. We have shown previously[3]
that 2 min of direct plasma treatment of E. coli using DBD
surface plasma produces complete sterilization, starting
from an initial concentration of 106–7 CFU. Obviously the
timescales of inactivation due to ozone exposure and
inactivation due to direct DBD surface plasma treatment do
not match. However, considering that during direct DBD
surface plasma treatment, E. coli present on the surface of
the DBD plasma device is directly exposed to produced
ozone, it is possible that E. coli concentrations on the surface
of the DBD plasma device are exposed to even higher levels
of ozone as compared to an inoculated substrate placed 100
away from the plasma generator.
At this time, the mechanisms of how activated charcoal
reacts with and removes, if any, other ROS are not well
understood. However, considering the lifetime of other ROS
(milliseconds to seconds) as compared to ozone (minutes),
experiments with inoculated substrates placed at a
distance away from the plasma generating source substantiate the majority role of ozone in stimulating bacterial
inactivation. Since the inoculated substrate is placed at a
distance from the plasma generating source, ROS with their
relatively short lifetimes will not be able to reach the
inoculated substrate in time to cause inactivation. Future
research in this area may detect and measure the
concentrations of absorbed ROS giving insight into which
(and how) other species produced during DBD plasma
generation acts in synergy with ozone to produce complete
sterilization.
Another way to characterize the produced ozone is by
using an intrinsic parameter that is capable of predicting
the ozone level produced by powering a plasma device with
a certain set of input parameters. Additionally, this
predicted ozone level should be independent of the volume
of the sterilization chamber in which the device is enclosed.
Consider an input voltage of 12 kV pp and an input
frequency of 14 kHz used to power a clean FR4 plasma
device in different chambers. The same device (same input
power) is producing ozone in all four chambers.
P
Power ‘‘P’’ is calculated by the formula P ¼ 1=N N
i¼1 VxI.
Thus for a sterilization time of ‘‘t’’ seconds, the average
absorbed input power over that interval can be calculated
by the formula
Pave ¼
N
1X
Pi
N i¼1
ð6Þ
where N ¼ t/Dt and Pi is measured input power at a single
time point. Thus for a sterilization time t ¼ 2 min ¼ 120 s,
if power Pi was measured at intervals of 15 s, then
N ¼ 120/15 ¼ 8. Once this average power is calculated, then
the average power density (Pden) can be calculated by
the formula
Pden ¼
Pi
A
ð7Þ
where A is the electrode surface area of the plasma device
(5.76 cm2). Pden is represented in units of W cm 2.
Finally input energy flux (J cm 2) is calculated by
multiplying Pden with the treatment time.[19] The ratio of
the measured ozone levels at each time point, divided by the
surface area of the plasma device is further divided by the
Figure 14. Comparison of specific ozone density versus input energy flux for the different sterilization chambers.
1132
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/ppap.201300108
Examining the Role of Ozone in Surface Plasma Sterilization
input energy flux at that particular time point for the same
plasma device. This gives rise to specific ozone density
(ppm J 1). This specific ozone density is plotted versus input
energy flux in Figure 14. In this figure, an almost-linear
correlation is observed between the calculated specific
ozone density and the calculated input energy flux. The
average specific ozone density for Chambers #1–#4 are 0.12,
0.055, 0.054, and 0.023 ppm J 1, respectively. This can be
made further independent of volume of the sterilization
chamber by dividing the average specific ozone density by
the volume of the sterilization chamber. Thus given a device
of a particular surface area, the above given trend should be
able to identify and predict measured ozone concentrations,
depending on volume of sterilization chamber. The input
energy flux is also important in the context of sterilization.
Complete sterilization of E. coli is observed at a threshold
input energy flux of 280–290 J cm 2. Thus future work will
also concentrate on evaluating the significance of the
specific ozone density in determining the point of complete
sterilization, using the threshold input energy flux as an
indicator.
bacterial inactivation following direct plasma exposure as
compared to exposure to ozone produced during plasma
generation, we conclude that while ozone plays a primary
role in the process of plasma sterilization, it is not the only
agent responsible for sterilization. Future work will involve
trying to understand which other factor complements
ozone production and enhances the sterilization process.
Acknowledgements: The authors are highly grateful to Sestar
Medical for their generous financial support in funding this
research. Sestar Medical had no role in study design, data
collection and analysis, decision to publish, or preparation of
the manuscript. Authors would also like to thank Raul.A.Chinga
(Doctoral Candidate, Electrical Engineering, UF) for his constant
help and support in manufacturing the plasma devices as well as
his work on the plasma generation setup.
Received: August 20, 2013; Revised: September 18, 2013; Accepted:
September 27, 2013; DOI: 10.1002/ppap.201300108
Keywords: dielectric barrier discharge (DBD); ozone; reactive
oxygen species (ROS); sterilization; surface plasma
4. Conclusion
The objective of this paper was twofold: (a) understand the
lethality of ozone produced during plasma sterilization,
using E. coli as the test pathogen and (b) understanding the
role of ozone in the process of plasma sterilization itself.
The dissipation rates of ozone, during and after plasma
generation using the different FR4/SC substrates as well as
using chambers of different volumes, are determined. Three
distinct phases are identified and the best-fit equations for
each are given. The spread of ozone concentrations
measured during each of these phases is shown to follow
an aspect ratio (AR) dependence.
Ozone exposure tests are conducted in which an initial
concentration of E. coli is exposed to ozone produced during
plasma generation. Tests comparing the rate of inactivation
using a FR4 and SC plasma generator determine a threshold
value (90 ppm) required for significant reduction in E. coli
concentrations. Furthermore, E. coli inactivation is shown to
also depend on the type of substrate inoculated.
It was also necessary to prove that bacterial inactivation
on exposure to air excited by plasma generation was due to
the amount of ozone in this air, produced during plasma
generation. In order to do so, two independent tests
inhibiting ozone are conducted. In both of these tests, it
is found that inhibiting ozone leads to ineffective bacterial
inactivation.
However owing to the highly coupled nature of the
problem, it is a little harder to understand the role of ozone
in the process of plasma sterilization itself. Based on rates of
Plasma Process. Polym. 2013, 10, 1120–1133
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[1] H. Halfmann, B. Denis, N. Bibinov, J. Wunderlich, P.
Awakowicz, J. Phys. D: Appl. Phys. 2007, 40, 5907.
[2] M. Laroussi, F. Leipold, Int. J. Mass Spectrom. 2004, 233, 81.
[3] N. Mastanaiah, J. Johnson, S. Roy, PLoS ONE 2013, 8, e70840.
[4] M. Kong, G. Kroessen, G. Morfill, T. Nosenko, T. Shimizu, J. Van
Dijk, J. Zimmermann, New J. Phys. 2009, 11, 115012.
[5] E. Stoffels, Y. Sakiyama, D. Graves, IEEE Trans. Plasma Sci.
2008, 36, 4.
[6] M. Leduc, D. Guay, R. Leask, S. Coulombe, New J. Phys. 2009, 11,
115021.
[7] H. Ayan, G. Fridman, D. Staack, A. Gutsol, V. Vasilets, A.
Fridman, G. Friedman, IEEE Trans. Plasma Sci. 2009, 37, 1.
[8] U. Kogelschatz, Plasma Chem. Plasma Process. 2003, 23, 1.
[9] B. Eliasson, M. Hirth, U. Kogelschatz, J. Phys. D: Appl. Phys.
1987, 20, 1421.
[10] A. Fridman, Plasma Chemistry, Cambridge University Press,
New York 2008.
[11] J. Wen, M. H. Thiemens, J. Geophys. Res. 1991, 96, 10911.
[12] W. Broadwater, R. Hoehn, P. King, Appl. Microbiol. 1973, 26,
391.
[13] N. Efremov, B. Adamiak, V. Blochin, S. Dadashev, K. Dmitriev,
O. Gryaznova, V. Jusbashev, IEEE Trans. Plasma Sci. 2000, 28, 1.
[14] D. Dobrynin, G. Fridman, G. Friedman, A. Fridman, New J.
Phys. 2009, 11, 115020.
[15] N. Vaze, M. Gallagher, S. Park, G. Fridman, V. Vasilets, A.
Gutsol, S. Anandan, G. Friedman, A. Fridman, IEEE Trans.
Plasma Sci. 2010, 38, 3234.
[16] S. Kalghatgi, A. Fridman, J. Azizkhan-Clifford, G. Friedman,
Plasma Process. Polym. 2012, 9, 726.
[17] K. Singh, S. Roy, J. Appl. Phys. 2007, 101, 123308.
[18] K. Singh, S. Roy, J. Appl. Phys. 2008, 103, 013305.
[19] S. Kalghatgi, C. Kelly, E. Cerchar, B. Torabi, O. Alekseev,
A. Fridman, G. Friedman, J. Azizkhan-Clifford, PLoS ONE 2011,
6, e16270.
www.plasma-polymers.org
1133