You are on page 1of 13

Home Search Collections Journals About Contact us My IOPscience

Experimental investigation of hydrogen peroxide RF plasmas

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2016 Plasma Sources Sci. Technol. 25 025003

(http://iopscience.iop.org/0963-0252/25/2/025003)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 152.8.78.7
This content was downloaded on 05/07/2017 at 00:42

Please note that terms and conditions apply.

You may also be interested in:

Spatially resolved Langmuir probe diagnostics in a capacitively coupled radio frequency argon and
oxygen plasma
C Kllig, Th Wegner and J Meichsner

On the E-H transition in inductively coupled radio frequency oxygen plasmas: I. Density and
temperature of electrons, ground state and singlet metastable molecular oxygen
Th Wegner, C Kllig and J Meichsner

SF6 RFplasma reactor for textile treatment


C Riccardi, R Barni, M Fontanesi et al.

Determination of ne and Te in a capacitively coupled RF discharge in neon


Z Navrtil, P Dvok, O Brzobohat et al.

A Langmuir probe system for high power RF-driven negative ion sources on high potential
P McNeely, S V Dudin, S Christ-Koch et al.

Characterization and global modelling of low-pressure hydrogen-based RF plasmas suitable for


surface cleaning processes
Nikola koro, Nevena Pua, Saa Lazovi et al.

Discharge equilibrium of a helicon plasma


Isaac D Sudit and Francis F Chen

Spatial density profiles ofN2(A 3u + ,B 3g,C 3u)


B Krames, T Glenewinkel-Meyer and J Meichsner
Plasma Sources Science and Technology

Plasma Sources Sci. Technol. 25 (2016) 025003 (12pp) doi:10.1088/0963-0252/25/2/025003

Experimental investigation of hydrogen


peroxide RF plasmas
RBarni1, ADecina1, SZanini1, ADOrazio2 and CRiccardi1
1
Dipartimento di Fisica G.Occhialini, Universit degli Studi di Milano-Bicocca, p.za della Scienza 3,
I-20126 Milano, Italy
2
Dipartimento di Ingegneria Astronautica, Elettrica ed Energetica, Sapienza Universit di Roma,
via Eudossiana 18, I-00184 Roma, Italy

E-mail: claudia.riccardi@mib.infn.it and ruggero.barni@mib.infn.it

Received 11 May 2015, revised 30 November 2015


Accepted for publication 14 December 2015
Published 27 January 2016

Abstract
This work reports a detailed experimental study of the plasma properties in low pressure
RF discharges in hydrogen peroxide and a comparison with argon under the same operating
conditions. H2O2 plasmas have been proposed for sterilization purposes. Electrical properties
of the discharge were shown to be similar, as for the RF and DC voltages of the driving
electrode. Bulk plasma volume remains stable, concentrated in an almost cylindrical region
between the two facing electrodes. It was found that the electron temperature is almost
uniform across the plasma and independent of the power level. This is higher than in argon
discharges: Te=4.60.9eV versus Te=3.31.1eV. The plasma density increases almost
linearly with the power level and a substantial negative ion component has been ruled out in
hydrogen peroxide. Dissociation in the plasma gas phase was revealed by atomic hydrogen
and hydroxyl radical emission in the discharge spectra. Emission from hydroxyl and atomic
oxygen demonstrates that oxidizing radicals are produced by hydrogen peroxide discharges,
revealing its usefulness for plasma processing other than sterilization, for instance to increase
polymer film surface energy. On the other hand, argon could be considered as a candidate for
the sterilization purposes due to the intense production of UV radiation.
Keywords: radiofrequency plasma, hydrogen peroxide, plasma sterilization

(Some figuresmay appear in colour only in the online journal)

1.Introduction Plasma processing is considered to be a promising phys-


ical method for sterilization [5]. It is generally performed
In modern medicine and in the pharmaceutical industry, as by producing oxidizing radicals in the plasma state, starting
well as in other sectors, the efficient sterilization of tools and with oxygen or water vapor in the gas phase mixture feeding
surfaces is of central importance [1]. The main challenge for the electrical discharge [5]. Within this approach, the pros-
these applications is selecting a reliable sterilization agent. pect of plasma sterilization employing directly hydrogen per-
Currently, sterilization methods based on the gas phase are oxide has been proposed [6, 7] and applied as a commercial
drawing significant interest, since a large number of critical sterilization device, referred as the STERRAD NX method,
devices (according to the classification of Spaulding [2]) by Advanced Sterilization Products [8].
cannot be subjected to thermal treatments. In this framework Moreover, besides these very important decontamination
the use of hydrogen peroxide as a chemical decontaminant and sterilization applications, hydrogen peroxide may also be
and sterilant is widespread, mainly because of its simplicity used as a reactive agent, both in vapor and in plasma, aimed
and its environmentally friendly decontamination products, at more general material surface processing, for instance the
i.e. water vapor and oxygen [3]. In the past decade, decon- deposition of oxide thin films [9].
tamination by hydrogen peroxide has been used frequently for Despite the hydrogen peroxide plasma sterilization process
sterilization purposes in a wide range of applications [4]. and its applications being well characterized in literature, little

0963-0252/16/025003+12$33.00 1 2016 IOP Publishing Ltd Printed in the UK


Plasma Sources Sci. Technol. 25 (2016) 025003 R Barni et al

is known concerning such plasmas [10]. Most of the modern pressure. This means that the chamber could be considered as
research focuses on the production of hydrogen peroxide, a well-mixed reactor [18]. This also holds trivially when there
mainly from water vapor and using atmospheric pressure is no pumping.
plasmas [1115]. No data exist so far in the literature con- The RF source consisted of a stainless steel disk (diam-
cerning key physical plasma parameters, such as the electron eter 30mm and thickness 2mm) shielded by a grounded alu-
temperature and density, not to mention the flux and energy of minium cup (diameter 35mm and thickness 2mm) separated
ions impinging on an exposed surface. Furthermore, there is from the source by an alumina hollow disk (1mm thick).
little information regarding the gas phase chemical composi- This prevented the formation of plasma in the part of the
tion, like the dissociation degree or the density and fluxes of chamber above the source electrode [19]. The source was cou-
(oxidizing) radicals towards such surfaces or the UV emission pled through a vacuum feed-through on the top flange to a
involved. Even more disturbing is the lack of data showing semi-automated RF matching unit which could be fed by a
how the plasma state is influenced by operating parameters 13.56 MHz RF supply (PFG-300 by Huettinger). The mainly
like discharge power level, pressure range and so on. This is capacitive RF coupling of the discharge was enhanced by the
mainly due to the lack of suitable plasma diagnostics installed presence of a counter-electrode, a stainless steel disk, diam-
in the reactors used in the processing. In general, the poor eter 80mm and thickness 2mm, facing the source at a fixed
knowledge, control and understanding of plasma processes, distance, typically 43mm. In this configuration the electrode
explain why, despite their promising properties and the claims system capacitance in vacuum was measured to be 91 pF
so abundant in scientific literature, they rarely find real appli- (Z=129 at 13.56 MHz).
cations. It is worth noting, for instance, that a recent paper On the front flange a 100mm quartz window is mounted,
discusses openly the existence of a true plasma decontami- whereas on the rear flange three vacuum ports accommodate
nation process [16]. the two gas inlets and an optical fiber vacuum feed-through.
In this work a characterization of radiofrequency dis- Hydrogen peroxide was prepared from a commercial water-
charges in hydrogen peroxide was undertaken, by means of peroxide solution by Bresciani (130vol. corresponding to
a dedicated plasma reactor, equipped with an extended set of 35 wt %). Due to the lower vapor pressure of hydrogen
plasma diagnostics, including RF electrical probes, movable peroxide (180 Pa versus 2470 Pa at 20 C [20]), the solution
Langmuir probes, optical emission spectroscopy (OES) with a concentration could be raised by distillation, pumping away
spectrometer and photomultiplier tube (PMT). This provided the mainly water vapor gas phase in a sealed bottle. About 30g
an insight into the role of different agents acting in the plasma of the solution (11g of H2O2) was reduced to 4.64g, which
processing. was used in the whole experimental campaign.
A concentration of 855 wt % was estimated by repeat-
edly measuring the equilibrium vapor pressure at 20 C [20].
2. Experimental set-up As already stated, for each experiment a fixed quantity (about
40 l) of the enriched solution was syringed into a sealed
A layout of the experimental set-up is displayed in figure1. bottle and frozen by immersion in liquid nitrogen (Dewar).
It consists of a cross-shaped vacuum chamber equipped with Residual air in the bottle was then pumped away before
a RF plasma source. The chamber, with a diameter 100mm melting occurred. Subsequently, the whole volume of the per-
and two arms both 215mm long, had several ports on the oxide solution was evaporated into the main chamber. After
four CF-100 flanges and can be evacuated through the bottom a brief pumping, the chamber was isolated. This allowed it
flange by a rotary pump (Varian SD-301, maximum pumping to reach a gas phase in the reactor at a fixed pressure (about
speed S=4.4 L s1) up to a residual pressure of 3.3101 50 Pa in these experiments) with a fairly high concentration.
Pa. Taking into account the conductance of the flow duct and As a diagnostic tool of the plasma state Optical Emission
the decrease in the pumping speed, an effective speed was esti- Spectroscopy (OES) was employed. OES of the discharge
mated to be Seff=2 L s1 at a 10 Pa pressure. The chamber could be used to gain insight into the excited species produced
volume was estimated to be 3.4L. A constant flow of Ar or in the source [21, 22]. Light was collected inside the vacuum
He could be provided and controlled by a flowmeter (Aera chamber with 50 m UV enhanced fiber optics (UV-50-FC
FC-7700C), whereas hydrogen peroxide in a sealed Pyrex by Avantes) with a numerical aperture NA=0.22. The fiber
bottle was evaporated into the chamber through a micrometer view-field was restricted with an alumina pipe in order to col-
needle valve. When pumped away, a liquid nitrogen trap was lect light only from the central plasma region. The conical
used, in order to prevent mixing of peroxide with pumping oil. view-field had a sectionof about 1cm radius at the center of
Total pressure in the chamber was measured by a Pirani gauge the vacuum chamber (see figure1). In this way, the contrib
(Thermovac TTR-90 by Leybold). The diffusion coefficient ution of the bright plasma sheath near the RF electrode was
of hydrogen peroxide (D0=0.19cm2 s1) could be used to removed and the contribution of the side regions minimized.
estimate the diffusion timescale in the chamber, D=1.3ms, This ensured the imaged plasma volume was approximately
and the diffusion speed, vD=15 m s1, at a 10 Pa pressure constant under the different discharge conditions and the
[17]. The corresponding values in argon are comparable changes in light intensity can be traced to those in the plasma
(D0=0.179cm2 s1). On the other hand, from the pumping state. Light was delivered through a vacuum feed-through
speed data a residence time of 1.7s and a flow velocity of and another fiber to a wide band, low resolution spectrometer
25cm s1 in the main chamber could be estimated at the same (PS-2000 by Avantes, with a holographic grating 600 lines mm1

2
Plasma Sources Sci. Technol. 25 (2016) 025003 R Barni et al

Figure 1. Schematic view of the experimental setup and diagnostics of the plasma reactor. The optical fiber have been connected to a
UVVis spectrometer or to a phototube.

blazed at 400nm) which allowed the measurement of the potential, and so the accuracy of the measure of the plasma
emission spectra. The spectra covered the UVVisNIR range parameters.
from 175 to 1100nm with a resolution of about 0.6nm. Line In general a fixed voltage ramp was applied in a 100ms
intensity was measured and emitting atoms and molecules time scan [29] and the probe current was measured with an
identified [23, 24]. The data on the emission line intensities equivalent resistance of 100 by an Agilent scope with a
of argon were also used to estimate the electron temperature 100kHz sampling rate. The Langmuir characteristics were
and the concentration of argon metastables, as presented in obtained by averaging 20 voltage ramps. Analysis of the curve
section4 [25]. was performed using dedicated software, in order to extract
A calibrated voltage divider was used to measure the the plasma density and potential, as well as the electron
RF and DC components of the voltage on the RF electrode temperature [29]. In a number of instances, the scans were
and inserted at the position of the RF feed-through [26]. limited to the measure of the floating potential or the ion satur
A RF-compensated Langmuir probe was used to measure ation current profiles which were performed directly with the
plasma profiles along the horizontal axis of the device. The scope. As for the former, the probe was directly connected to
probe scan covered the whole region across the electrodes (see the 1 M input, whereas, for the latter, the voltage drop on the
figure1). The probe had an exposed tungsten tip, 5mm length shunt resistance after imposing a constant bias (36V) to the
and 1mm diameter. The probe tip connection was insulated probe tip was measured.
and shielded in a probe holder, 15cm in length, made of two
coaxial alumina tubes. RF compensation was achieved using
the method proposed long ago by Godyak et al [27] and Chen 3. Plasma state
[28]. A miniature RF choke (made of a ferrite core, Q200
with an impedance of about 85 k) was embedded near the The electrode system was tested initially with argon. A RF
probe tip. A floating electrode (a copper foil, 10mm wide, discharge is easily ignited for pressure above 2.5 Pa and up
3mm diameter, glued on the probe alumina tube 3mm before to 200 Pa and sustained previously at supply power as low as
the tip) exposed to the plasma was coupled through a large 2W. The breakdown voltage was measured to be 1438V
capacitor to the probe tip (100 pF) to ensure an almost con- at P=50 Pa [30]. This is a smoothly decreasing function
stant impedance load during the probe voltage scan to record (about15%) of the pressure in the 10200 Pa range. It started
the Langmuir characteristics [28]. to rise substantially only below 10 Pa. The discharge ignition
The Langmuir probe was calibrated by fine tuning the was marked by a sharp drop in the DC electrode voltage. By
choke capacitance and thus the frequency to match the 13.56 increasing the supply power level stepwise the transition could
MHz RF supply. The probe was previously employed in other be caught many times measuring the actual RF breakdown
plasma discharges and checked against traditional Langmuir voltage. In the whole pressure range the power level measured
probes [21]. In most of the experimental conditions dis- at the RF supply remained proportional to the square of the
2
cussed here it has been observed that the probe compensation RF voltage amplitude measured at the electrode (W = a VRF )
improves, albeit marginally, the shape of the Langmuir char- with a constant almost independent from the pressure, at least
acteristics curve, increasing the reported value of the floating for pressure below 70 Pa. This corresponds to an effective

3
Plasma Sources Sci. Technol. 25 (2016) 025003 R Barni et al

Figure 2. Radiofrequency and the absolute value of the DC electrode potential as a function of the power level in hydrogen peroxide and
argon discharges at 50 Pa pressure. The dotted line shows a power-law fit with a close to 0.5 exponent.

impedance of about 5 k. As might be expected, because of positive with respect to the grounded walls (possibly draining
the electrode setup, a negative DC voltage appeared on the RF electrons from the gas phase) was independent from the power
electrode [31, 32]. Its absolute value was nicely proportional level, about 0.160.01. The measured trends are displayed
to the RF voltage, in the whole pressure range, i.e. somewhat in figure2.
larger the lower the pressure was. At lower power levels, near the breakdown, the discharge
The behavior of RF and DC voltage with applied power glow was concentrated just below the RF electrode, forming
also holds for discharges in peroxide in experiments at a fixed a sheath separating it from the electrode surface. The sheath
pressure, about 50 Pa. This value was chosen mainly because thickness was mainly controlled by the pressure level, since the
it was the one used in the most advanced plasma sterilization conditions match those of a collisional sheath because of the
method [8]. However, it is also quite favorable because break- intermediate pressure range of these experiments [31]. Using
down happens easily at this pressure. This extends to nearby equation (11.2.54) of [31] and electron density and temper
pressures, at least within the 20100 Pa range. Moreover, ature values discussed below, a mean sheath thickness of
although the pressure level is not exactly reproducible in dif- 4.6mm for the argon discharge was estimated at WRF=50W
ferent experimental scans and it derives slightly, especially at and P=50 Pa. The discharge became brighter as the power
a high power level, it was checked that the discrepancies intro- level is increased. The glowing region, however, expanded
duced into the electrical characteristics of the discharge are downwards filling quite uniformly the whole region between
weak. The breakdown voltage in peroxide was measured to be the two facing plates, already at a power level of about 25W, in
1285V. As for argon, it decreased slightly as a function of hydrogen peroxide. Conversely, its lateral extension appeared
the pressure, about 10% from 20 to 100 Pa. The value being almost untouched, remaining comparable to the RF electrode
slightly lower than in argon could probably be related to the diameter. However, the estimate of the effective plasma exten-
lower ionization potential of hydrogen peroxide (10.63 with sion based only on emission light intensity was not conclusive.
respect to 15.8eV [32, 33]) which requires less energetic elec- Two pictures of the discharge taken at a power level of 10 and
trons thus possibly compensating the expected larger electron 50W are displayed in figure3. In order to access the plasma
losses due to H2O2 electronegativity. properties the characteristics curve of the Langmuir probe was
The RF voltage increased nicely as the root of the power also analyzed. A typical example is shown in figure4, together
level. Its values were only slightly smaller (934% over the with the fit function used to extract the plasma parameters
whole power range) than the corresponding ones in argon. [29]. However, a rough picture of the plasma distribution
The effective impedance of the discharge and electrodes was could be reached initially by inspecting the spatial profiles of
5.80.1 k. The same reduction with respect to argon was the ion saturation current and of the floating potential, which
also observed in the absolute values of the DC potential. are displayed in figure5. The bell shaped profiles of the ion
A comparatively smaller DC bias with respect to argon was saturation current are nicely reproduced by Gaussian curves.
observed only near the breakdown at a power level less than The FWHM was 37mm for H2O2 and 39mm for argon, nicely
10W. The ratio between the DC and the RF voltage values was matching the RF electrode diameter of 35mm. Whereas the
the same in both H2O2 and Ar, being 0.870.02 constant in amplitude increased with the power level, the profile shape
the whole power range WRF>10W. Since the instantaneous was only slightly affected; the FWHM also staying compa-
electrode potential was the sum of the DC and RF voltages, rable with the electrode diameter at higher power. The floating
the fraction of the RF period when the electrode potential was potential showed a much shallower spatial dependence, with

4
Plasma Sources Sci. Technol. 25 (2016) 025003 R Barni et al

Figure 3. Pictures of the discharge glow in hydrogen peroxide at a power level of (a) 10W, near the breakdown and 50W (b). In the latter
the RF electrode and the shielding cup can be viewed on the upper side.

Figure 4. A typical Langmuir characteristics curve in gray together with the fitted dotted line, with a function containing the ion and
electron density and the electron temperature as free parameters. On the right axis, the second derivative of the probe current with a vertical
log-scale.

a FWHM of 90mm. Neglecting the effect of a possible spatial in repeated Langmuir probe measurements. Consistent
dependence of the electron temperature, the sign of the radial with the trend of the ion saturation current, plasma density
electric field agreed with that prescribed by ambipolar diffu- appeared to rise with the power level. The trend is somewhat
sion. So, possibly electron and ions radial fluxes were almost linear, more so in hydrogen peroxide than in argon. At low
balanced, at least for the bulk plasma region. power levels their density was similar, below 5 109 cm3,
At this stage the full Langmuir characteristics had been although with a great uncertainty due to the small size of the
measured. Its analysis allowed evaluation of the electron ion saturation current. The plasma density become larger in
temperature, together with the ion and electron density and argon with respect to hydrogen peroxide at a large power
the plasma potential. However, the analysis also introduced level. For instance, it became about three times larger at the
larger uncertainties which partly obscure, for instance, the center of the vacuum chamber with a power level of 100 W:
plasma potential spatial profile shape. Results are displayed in here there was an estimated a density of 3.20.7 1010 cm3
figures68. In order to have an estimate of the level of vari- versus 1.00.2 1010 cm3. Electron density could be evalu-
ations in the values, repeated measurements were performed ated from the probe current (subtracting the ion saturation
under the same conditions which are reported as multiple contribution) at the plasma potential [29]. In these experi-
points on the plots. Discrepancies of up to 50% were observed ments, electron and positive ion densities were similar, equal

5
Plasma Sources Sci. Technol. 25 (2016) 025003 R Barni et al

Figure 5. Horizontal axis profile of the floating potential and of the ion saturation current in hydrogen peroxide (100W) and argon (50W)
discharges at 50 Pa pressure. The dotted line shows a Gaussian fit function with a FWHM of 37mm.

Figure 6. Electron temperature and ion density measured from the Langmuir characteristics fit as a function of the power level in hydrogen
peroxide and argon discharges at 50 Pa pressure.

within the measurement of uncertainty, which implies that floating potential with an exponential function yielded meas-
even in the somewhat electronegative hydrogen peroxide the urement of the electron temperature. This was fairly flat over
negative ion density was lower than the electron one. This the power level range, both in H2O2 and in argon. It was some-
holds in the whole power range. It is at variance with other what larger only at the lowest values of the power. Averaging
observations, albeit made in a different electrode geometry, values for WRF>15W obtained Te=3.31.1eV for argon
using more electronegative gases, where negative ion den- and Te=4.60.9eV for hydrogen peroxide. For argon,
sity largely exceeded the electron one [29]. Langmuir char- also only a slight decreasing dependence from the pressure
acteristics curves, after RF compensation, showed good in the 5200 Pa range was found. Finally, the value of the
exponential behavior just above the floating potential, which DC plasma (space) potential could be obtained either from the
implies that the bulk electron energy distribution (EEDF) floating potential plus electron temperature or from the inflec-
is remarkably Maxwellian [29]. This could also be grasped tion in the characteristics curve. Both methods produced com-
from figure 4, where the logarithm of the second derivative parable results, although these show discrepancies of the order
of the (SavitzkyGolay smoothed) probe current, proportional of a few volts (larger than those in floating potential alone, see
to the EEDF, is displayed [29]. This is roughly parabolic figure7). This prevented their use to estimate, for instance, the
between the floating and the plasma potential. This observa- radial electric field from the spatial profile data or the accurate
tion holds for both argon and peroxide plasmas investigated shape of the profile. Plasma potential, however, was invariably
here. Fitting the region of the characteristics curve above the positive, about 1525V, in the whole power level range for

6
Plasma Sources Sci. Technol. 25 (2016) 025003 R Barni et al

Figure 7. Horizontal axis profile of the electron temperature and of the plasma potential, estimated from the inflection point in the probe
current, in hydrogen peroxide (25W) and argon (100W) discharges at 50 Pa pressure.

Figure 8. Horizontal axis profile of the ion and electron density in a hydrogen peroxide discharge (25W) at 50 Pa pressure. The dotted line
shows a Gaussian fit function with a FWHM of 37mm.

both argon and hydrogen peroxide. This happened notwith- were flat; practically constant over the whole spatial profile.
standing the fact the RF voltage and the DC potential at the Averaging over the whole profile yielded a Te=4.80.5eV
electrode changed by about one order of magnitude. It could for hydrogen peroxide, compared with Te=2.80.3eV for
be noted, however, that space potentials were smaller than the argon. Allying this to previous data, it can be safely stated
maximum positive voltage reached by the electrode potential that the plasma region fills a constant portion of the vacuum
during the RF period (see figure4) in the whole power level chamber when the power level exceeds a few tens of W.
range. So, even for a small fraction of the time, electrons were This corresponds probably to an almost cylindrical volume
drained to the electrode across the sheath, balancing the pre- where height is the inter-electrode distance and the base area
vailing positive ion flux [30]. matches the RF electrode extension. These considerations are
Finally, the spatial profiles of the plasma parameters were of importance in order to estimate the effective processing
measured (see figure 8). Plasma density profiles reproduced region where materials can be directly exposed to the plasma.
remarkably well the ion saturation current ones. Their bell Finally, data measured concerning the plasma density and the
shape was nicely Gaussian, with a FWHM of 33mm in argon electron temperature, together with those about the DC poten-
and 37mm in hydrogen peroxide. Their symmetry with respect tial, could be used to estimate the effective power absorbed
to the center of the chamber points towards a substantial radial by the plasma, by evaluating the stochastic and ohmic heating
uniformity of the plasma region with respect to the inter- power, as discussed in section 11.2 of [31]. For argon dis-
electrode axis. On the other hand, the electron temperature charges, a fraction between 8 and 15% of the input power
and, within a larger fluctuation band, the plasma potential level was estimated in the 5100W power range.

7
Plasma Sources Sci. Technol. 25 (2016) 025003 R Barni et al

Figure 9. A typical OES spectrum in a hydrogen peroxide discharge (200W) at 50 Pa pressure. The main emission lines are identified on
the spectrum.

Figure 10. Intensity of a few emission lines in hydrogen peroxide discharges at 50 Pa pressure as a function of the power level. The dotted
lines show linear fit functions to the data.

4. Optical emission spectroscopy of peroxide As already stated, the OES view-field was chosen in such a
plasma way to be almost independent from any changes in the dis-
charge volume. Since both the emitting states could be pro-
The typical discharge spectra of a peroxide discharge is dis- duced in electron dissociative interactions on the peroxide
played in figure9. It refers to a power level (200W) and total molecule, it is straightforward to infer that the trend could
pressure (50 Pa) similar to those already considered in the be understood as a linear increase in the electron density in
previous section. Discharge spectra were dominated by the the plasma gas phase, if the electron temperature and the per-
Balmer atomic hydrogen emission lines, with H the most oxide partial pressure were constant. This can be taken as a
prominent, atomic oxygen [24] and hydroxyl [23] ones. confirmation of the direct findings based on Langmuir probe
A clear signature of impurities, mainly CO band emission analysis and could be extended to the highest power level
was significant too [23]. It was decieded that only lines which possible with the current RF supply. This was, fairly com-
could be assigned unambiguously and the intensity of which monly, observed in RF discharges of other electronegative
is large enough to withstand the noise level should be ana- gases [17]. This could be correlated to the linear increase with
lyzed. The intensity of a hydrogen and a hydroxyl line as a the power level displayed by the total discharge current too.
function of the power level is shown in figure10. Both dis- Direct excitation by electron impact on ground state hydrogen
play a remarkable linear trend up to the highest power level. atoms or hydroxyl radicals, a concurrent process leading to

8
Plasma Sources Sci. Technol. 25 (2016) 025003 R Barni et al

Figure 11. Intensity normalized to the atomic hydrogen H line of a few emission lines in hydrogen peroxide discharges at 50 Pa pressure
as a function of the power level.

light emission, would have caused much greater difficulties in H intensity was also flat within errors in the whole dataset,
explaining the observed trend [22]. In fact, ground state rad- R=0.220.03. This also pointed towards a common origin
ical and atom concentration were expected to be proportional of both excitation mechanisms.
to electron density, being built up by electron impact dissocia- The trend shown by atomic oxygen emission lines was dif-
tion too. This would have led to a stronger than linear func- ferent, as can be better seen in figure11, where the intensities
tional dependence on electron density for the excited radical were normalized to that of the most intense line, the H at
concentration and thus of the emitted light intensity. In order 656nm. Atomic oxygen is another important agent for sterili-
to bring back OES data to the observed purely linear trend, zation and in general as the typical reactive oxidizing species
it was necessary to speculate a somewhat ad hoc relation- (ROS) in plasma gas phase [5]. It is clear that the linear trend
ship between the power level and the electron density. This observed above no longer exists. This indicates that direct for-
also hinted at the point that electron temperature cannot vary mation of excited oxygen states from electron impact on water
very much with the power level. Indeed the rate of dissocia- peroxide cannot be the main excitation channel. Light emis-
tive interactions (and of direct excitation too) were strongly sion could result from excitation of oxygen atoms produced
dependent on the electron temperature (usually exponentially, by the dissociation of OH and O2H radicals. However, since
for temperatures of a few eV, as it was the case with RF dis- radical formation is somewhat proportional to the dissociation
charges [31]). This would translate to a strong dependence of rate, and so to electron density, it could be expected that a
emitted light intensity. Again, to bring back OES data to the stronger than linear functional dependence on electron density
observed purely linear trend, it was necessary to speculate an and hence on power level would be observed. This was ruled
ad hoc relationship between the electron temperature and the out by the data gathered in this experiement. Excited oxygen
electron density as the power increased. An almost constant atoms, then, have to be quenched effectively at a high power
electron temperature agreed with the direct findings based level. Only this could explain why saturation was observed. As
on Langmuir probe analysis and such a conclusion could be further evidence, it was also observed that the ratio between
extended to the highest power level possible with the current the 844 and 777nm atomic oxygen emission line intensity was
RF supply, where RF compensation of the Langmuir probe independent from the power level: R=0.240.03. Further,
fails. This was also independently confirmed by the observa- in figure11, it can be appreciated that the impurities relative
tion that the ratio between the H and H intensity was not emission is quite flat, which implies a somewhat linear trend
correlated, within 5%, with the power level: R=0.470.02. with the power level. This was observed for the clearly iden-
The same conclusion could be obtained by comparing different tifiable CO lines, as well as for other minor lines, for instance
oxygen atom lines, albeit with a greater uncertainty. A signifi- the one at 288nm, already visible in the figure 9 spectrum.
cant role of energetic electrons in the bulk plasma region, for This points towards a constant impurity concentration level,
instance accelerated by the RF sheath, could be disfavored too, determined mainly by the absolute density of water peroxide.
since again the voltage across the sheath and then the electron It is fully consistent with impurities coming from the walls
energy gain would lead, on all evidence, to a stronger than or, more probably, from the electrode surfaces, as a conse-
linear functional dependence on the power level. A significant quence of the diffusion flow and surface reaction of peroxide.
deviation from a purely Maxwellian energy distribution was As could be expected, the good reproducibility displayed by
already ruled out by the analysis of the Langmuir characteris- H, O and OH emission intensity in different power scans is
tics at a low power level. The ratio between OH-309nm and not preserved by impurities. It was somewhat comforting that

9
Plasma Sources Sci. Technol. 25 (2016) 025003 R Barni et al

Figure 12. Comparison of the electron temperature measured by the Langmuir curve fit and by OES data in argon discharges at 50 Pa
pressure as a function of the power level.

impurities emission intensity declined as ageing of the device parameters allowed the extraction of the electron temperature
proceeds. and the two metastable concentrations. As discussed in pre-
Experiments involving OES using argon the gas phase were vious published papers [34, 35], this procedure is more sen-
also performed. As already discussed, argon discharges bear sible, reducing the systematic errors which could affect the
many similarities with hydrogen peroxide ones. OES showed proposed analytical solution [25]. Using the measured intensi-
that argon emission lines largely dominated the spectra, with ties of thirteen lines, a fair reproduction of the relative intensity
only small impurities, notably OH presumably arising from pattern was obtained, within 30%, which corresponds to a 2
water vapor wall outgassing. Light intensity was broadly ~ 0.1. The overall agreement with the Langmuir probe meas-
linear with the power level and also to the total pressure, urement was comforting, as can be grasped from figure 12.
at least in the range around 50 Pa. Besides this, the rich emis- This confirms more directly that the authors suggestion that
sion line spectrum of atomic argon in VIS and near IR (RF electron temperature is mostly independent from the power
argon spectra were dominated by the system of 3p54p4s level, based on Langmuir probe analysis, holds and could be
transitions, the 2px system in the Paschen notation, in the red extended to the highest power level possible with the current
and in the near IR [24]) could be used to measure plasma RF supply.
parameters such as the electron temperature and metastable A photomultiplier tube (PMT) was used to collect light
density [25]. Here, the proposed method [34, 35] is recalled emitted during different phases of the RF oscillation. A PMT
only briefly. The core of the method was to exploit the dif- by Hamamatsu (H10721-210) was used for its fast temporal
ferent electron energy dependence of the cross section of response and high sensitivity (rise time is 0.57 ns, gain ~
the 2px excited levels and the large differences in the prob- 2106). Its ultra bialkali photocathode is sensitive to light
ability of electron impact excitation of such levels starting with a wavelength in the range 230700nm and the radiant
from metastable argon atoms [25]. The proposal was to use sensitivity curve almost flat between 300 and 500nm, sub-
the intensities of the whole set of emission lines of the 2px stantially matching the emission spectra range [36]. Both
system, provided that their intensity was not negligible and the number of PMT counts and the intensity of the collected
no ambiguity could arise (especially because of the limited light increased linearly with the power level, with a slope in
wavelength resolution) to define a sort of 2 function: argon almost twice that of in H2O2. This trend correlated to
that already observed for individual emission lines. With quite
2 = 2 N (N 1) [log 2(Rn /Rm )],
a large dataset of events it was possible to assess how the
n < m = 1, N
light emission was correlated with the different phases of the
Rn = (In /bn )/ k n(Te )(1 x *) + k*n(Te )x * ,
electrode potential RF cycle. In particular, the distribution in
PMT peak amplitude and duration were almost symmetrical
where I is the measured intensity of the selected emission between the positive and negative parts of the RF cycle and
line and b is the branching ratio of that particular transition almost no modulation was observed during the RF period.
from the upper excited level involved. In the latter formula k This is reported in figure 13. This is not entirely expected
and k* are the electron impact collision rate to the excited level since the effective collision frequency at 50 Pa is larger than
corresponding to the nth emission lines, respectively from the 13.56 MHz RF one [32]. The independence from the RF
the ground state and from the two metastable states, while cycle phase points towards a substantial uniformity in time
x* are their concentrations. Minimization with respect to the of the emission light, at least from the central plasma region.

10
Plasma Sources Sci. Technol. 25 (2016) 025003 R Barni et al

Figure 13. Histogram of the PMT peak number in hydrogen peroxide (at two power levels) and argon discharges at 50 Pa pressure as a
function of the phase during the RF cycle.

Incorporating the previous considerations, it can be postulated power level. Any substantial modulation of the light emission
that, even in hydrogen peroxide, the plasma region, away during the RF cycle period was also exlcuded.
from the electrode sheath, is substantially stable during the All these findings point towards a common picture, where
RF oscillation period. hydrogen peroxide plasmas are easily produced by RF dis-
charges and their gas phase is composed of highly oxidizing
radicals, partly different from those prevailing in water vapor
5.Conclusions plasmas, as well as from the composition of the initial neu-
tral gas phase. Hydrogen peroxide discharges, then, have the
A rather detailed experimental study of the plasma properties capability to deliver different plasma processing and surface
in low pressure RF discharges in hydrogen peroxide has been treatments and are potentially interesting for many applica-
performed and a comparison made with the much more inves- tions, not confined purely to improved sterilization.
tigated argon ones.
Discharges were easily ignited and sustained in hydrogen
peroxide, just as in argon. Due to the lower vapor pressure Acknowledgments
of hydrogen peroxide with respect to water, it is possible to
We are pleased to thank M Piselli for his technical support.
obtain highly concentrated gas phases of almost pure H2O2
by differential pumping. Electrical properties of the discharge
appeared quite similar indeed, as did the RF and DC voltages References
of the driving electrode.
In a wide region of power level, from the breakdown up to [1] BennettJE, DolinR and BlaserMJ 2014 Principles and
300W, the bulk plasma volume remained stable, concentrated Practice of Infectious Diseases (Philadelphia, PA: Elsevier)
in an almost cylindrical region between the two facing elec- [2] SpauldingEH 1972 J. Hosp. Res. 95
trodes. The electron temperature was probably uniform across [3] KlapesNA and VesleyD 1990 Appl. Environ. Microbiol.
the plasma (only its uniformity along the radial distance from 565036
[4] RadlS, OrtnerS, SungkornR and KhinastJG 2009 J. Pharm.
the discharge axis was checked) and almost independent from Innov. 451
the power level. It was higher than on argon discharges since [5] LaroussiM 2005 Plasma Process. Polym. 2391
Te=4.60.9eV versus Te=3.31.1eV. The DC comp [6] KyiMS, HoltonJ and RidgwayGL 1995 J. Hosp. Infect.
onent of the space potential was possibly quite uniform too 31275
in the bulk plasma. Plasma density increased almost linearly [7] KrebsMC, BcasseP, VerjatD and DarbordJC 1998 Int. J.
Pharm. 16075
with the power level, being smaller than in argon at the same [8] Diab-ElschahawiM, BlackyA, BachhofnerN and KollerW
power. A substantial negative ion component had been ruled 2010 Am. J. Infect. Control. 38806
out in the hydrogen peroxide plasmas, notwithstanding the [9] TaylorMP, TimmsPL, AllenGC and ChurchSR 1998
more than negligible electro-negativity of the parent molecule. J. Mater. Chem. 81769
Dissociation in the plasma gas phase was revealed by atomic [10] VenugopalanM and ShihAL 1981 Plasma Chem. Plasma
Process. 1191
hydrogen and hydroxyl radical emission in the discharge [11] VaskoCA, LiuDX, Van VeldhuizenEM, IzaF and
spectra. Atomic oxygen was produced too, besides other BruggemanPJ 2014 Plasma Chem. Plasma Process.
impurities, but its relative emission was suppressed at a large 341081

11
Plasma Sources Sci. Technol. 25 (2016) 025003 R Barni et al

[12] ShibataT and NishiyamaH 2012 Int. J. Plasma Environ. Sci. National Institute of Standards and Technology)
Technol. 62539 (http://physics.nist.gov/asd3)
[13] LeeHW, LeeHW, KangSK, KimHY, WonIH, JeonSM [25] MariottiD, ShimizuY, SasakiT and KoshizakiN 2006
and LeeJK 2013 Plasma Sources Sci. Technol. 22055008 Appl. Phys. Lett. 89201502
[14] OConnorN, CahillO, DanielsS, GalvinSc and [26] SobolewskiMA 1998 Appl. Phys. Lett. 721146
HumphreysH 2014 J. Hosp. Infect. 8859 [27] GodyakVA, PiejakRB and AlexandrovichBM 1992
[15] Mai-ProchnowaA, MurphyaAB, McLeanKM, KongMG Plasma Sources Sci. Technol. 136
and OstrikovKK 2014 Int. J. Antimicrob. Ag. 43508 [28] SuditID and ChenFF 1994 Plasma Sources Sci. Technol.
[16] ShintaniH, SakudoA, BurkeP and McDonnellG 2010 3162
Exp. Ther. Med. 17318 [29] RiccardiC, BarniR and FontanesiM 2001 J. Appl. Phys.
[17] RiccardiC, BarniR, De ColleF and FontanesiM 2000 903735
IEEE Trans. Plasma Sci. 28278 [30] KorolovI, DerzsiA and DonkoZ 2014 J. Phys. D 47475202
[18] SmithJM 1987 Introduction to Chemical Engineering [31] LiebermanMA and LichtenbergAJ 1994 Principles of Plasma
Thermodynamics (New York: McGraw-Hill) Discharges and Materials Processing (New York: Wiley)
[19] GoedheerWJ 2000 Plasma Sources Sci. Technol. 9507 [32] RaizerYP 1991 Gas Discharge Physics (Heidelberg: Springer)
[20] Van LaarJJ 1910 Z. Phys. Chem. 7272351 [33] LitorjaM and RuscicB 1998 J. Electron Spectrosc. Relat.
[21] BarniR, ZaniniS and RiccardiC 2008 Vacuum 82217 Phenom. 97131
[22] FantzU 2006 Plasma Sources Sci. Technol. 15S137 [34] DonnellyVM 2004 J. Phys. D 37R217
[23] PearseRWB and GaydonAG 1976 The Identification of [35] CroccoloF, BarniR, ZaniniS, PalvariniA and RiccardiC
Molecular Spectra (New York: Wiley) 2008 J. Phys.: Conf. Ser. 100062023
[24] RalchenkoY, KramidaAE, ReaderJ and NIST Team 2008 [36] BiganzoliI, BarniR and RiccardiC 2013 J. Phys. D
NIST Atomic Spectra Database (Gaithersburg, MD: 46025201

12

You might also like