Multi-diagnostic approach to energy transport in an atmospheric pressure helium-oxygen plasma jet
Tristan Winzer, Natascha Blosczyk, Jan Benedikt, Judith Golda
MMulti-diagnostic approach to energy transport in anatmospheric pressure helium-oxygen plasma jet
Tristan Winzer , Natascha Blosczyk , Jan Benedikt ,Judith Golda , Institute of Experimental and Applied Physics, Kiel University, Germany. Plasma Interface Physics, Ruhr-University Bochum, Germany.E-mail: [email protected]
Abstract.
The energy balance of a plasma holds fundamental information notonly about basic plasma physics, but it is also important for tailoring plasmasto specific applications. Especially RF-driven atmospheric pressure plasmajets (APPJs) operated in helium with oxygen admixture have high applicationpotential in industry and medicine. Many types of plasma jets have been studiedup to now, leading to the challenge how to compare results from various sources.We have developed a method for measuring the power deposited in the plasmaas the parameter to compare different sources and gas mixtures with each other.Furthermore, we studied energy transport as a function of this input power andmolecular gas admixture in a newly developed APPJ based on the COST-referencejet with a capillary as a dielectric in between the electrodes. The gas temperature,atomic oxygen density, ozone density and absolute emission intensity in the visiblewavelength range have been determined. Combining the results gave an energybalance with most of the energy deposited into gas heating. Production of finalchemical products made up a small amount of the deposited power while radiationwas negligible for all combinations of external parameters studied.
Keywords : atmospheric pressure plasma, µ -APPJ, dissipated power, optical emissionspectroscopy, oxygen actinometry, molecular beam mass spectrometry a r X i v : . [ phy s i c s . p l a s m - ph ] F e b nergy transport in an atmospheric pressure plasma jet
1. Introduction
Atmospheric pressure plasma jets (APPJs) havegained increased interest in recent years, due totheir application potential in medicine and industrialprocesses [1–5]. These plasmas exhibit strong non-equilibrium characteristics with high electron energiesat low gas temperature, especially when driven withRF-frequencies. The energy is in this case selectivelycoupled to the electrons and is then dissipated viadifferent physical processes to other species present inthe plasma, e.g. via drift, diffusion, elastic and inelasticcollisions. To channel the energy inside the dischargeinto the desired processes, fundamental understandingof the influence of external parameters on the energybalance is necessary.An energy balance answers the question of howthe energy introduced into the plasma system isdistributed among the species and processes. Inhistory, the Lawson criterion is probably one of themost prominent examples for successful considerationof an energy balance for a plasma in thermalequilibrium [6]. As early as 1957, Lawson theoreticallycompared the energy produced in a fusion reactionwith the energy lost to the environment. In doing so,he defined a minimum required value for the productof the plasma density and the confinement time thatresults in a net energy output. Another example for athorough analysis of an energy balance is the modelof equi-operational plasmas by Jonkers et al. from2003, where the particle balance, the energy balance,and the relative population of the ground state wereused to derive three characteristic plasma parameters:characteristic size, power density and pressure. Whenthese parameters are identical, two plasmas usingdifferent feed gas mixtures are comparable [7].Here, we experimentally studied the energybalance of an atmospheric pressure plasma using anRF-driven capacitively coupled APPJ with a capillaryas dielectric between the electrodes. The first step wasto measure how much energy is delivered to the plasma.The following analysis focuses on the loss mechanismsof the electron energy. We took into account elasticand inelastic collisions. Losses by elastic collisions ofelectrons with ions and neutrals are transferred intotranslational energy of these atoms and molecules, i.e.heat. This energy fraction leaving the plasma with thegas flow is represented by the neutral gas temperature.Inelastic collisions lead to excitation, ionization anddissociation. Part of this energy is initiating chemicalreactions in the plasma, for example the generationof atomic oxygen. The excitation is representedby quantitatively characterizing the radiation emittedduring de-excitation.
2. Experimental setup
For studying the influence of different externalparameters on the energy transport in the plasma,a setup which ensures stable plasma operation overa broad range of parameters and on long timescalesis needed. As reproducibility is also very importantfor comparing the results of different measurementseries, we employ a derivative of the COST ReferenceMicroplasma Jet. The COST-Jet has shown todeliver reproducible and stable operating conditions.However, it is limited in terms of the maximum inputpower and molecular gas admixture. At input powersbetween 1 . . side view of assembled jet vertical cross section horizontal cross sectiongrounded shieldingmetalelectrodesglas capillary ceramic insulation40 mm 14 mm39 mm 14 mmgas flow11.5 mm 12.5 mm Figure 1.
Schematic drawing of the capillary jet device. Theside view of the jet is shown in the upper left corner. The twoother views are the cross sections at the positions of the dashedlines.
The so called capillary plasma jet features asimilar discharge geometry and excitation scheme asthe COST-Jet [8], see Figure 1. The plasma isgenerated inside a glas capillary with a quadratic innercross section of 1 mm and a wall thickness of 0 . . . One of the electrodes isgrounded and the other one is connected to the RF-generator via a matching network. Both stainless steelelectrodes are held in place by insulating ceramic plateson two sides which are fixed by grounded stainlesssteel metal shielding. Electrical probes for voltageand current are implemented by small circuit boardsscrewed to the grounded metal shielding. The voltage nergy transport in an atmospheric pressure plasma jet . .
999 %) and 0 sccm to 20 sccmoxygen (99 . . . Because of losses e.g. in the matchbox or cables, thegenerator power is always larger than the depositedpower. A setup for simultaneously measuring thevoltage and current signals at the plasma jet withhigh temporal resolution is needed to derive the powerdelivered to the plasma. The setup is adapted fromthe COST-Jet, where it has delivered stable andreproducible results for the discharge power [9]. Asketch of the setup is shown in figure 2.
L-type matching network13.56 MHz RF-generator electrode electrode Ω . Ω V Ω Vpick-up voltage probeoscilloscopechannel 2 oscilloscope channel 1
Figure 2.
Schematic diagram of the electrical setup used forpower measurements.
The voltage is supplied to one electrode by a13 .
56 MHz generator (dressler Cesar) connected to aL-type matching network (Barthel Matching Cube i-300). This voltage is measured using a pick-up antennaand a 2 . / s oscilloscope (Yokogawa DL9140l).Calibration of the pick-up voltage was performedusing a commercial voltage probe (Tektronix P5100A).A second voltage is measured over a 4 . P avg can be calculated from the measured voltage andcurrent signals. For sinusoidal voltages, this can besimplified to P avg = U rms I rms cos(∆ ϕ ) , (1)where U rms and I rms are the root-mean-squarevalues of the voltage and current respectively and∆ ϕ is the phase shift between them. Therefore, anestablished technique to calculate the power is to fitsine waves to the measured waveforms and extractthe necessary values. However, at higher voltages,the processes in the plasma become increasinglynonlinear, leading to distortion of the measured signals[11]. Consequently, for measurements in higher powerregimes, it is necessary to account for these distortions.Therefore, we developed a more elaborate techniquebased on the cross-correlation method to enable powercalculation by point-wise multiplication of voltage andcurrent waveforms.First, the phase shift between the probesmeasuring voltage and current has to be determined.For a perfect capacitor, this phase difference is 90 ° .Most capacitively coupled setups without plasma donot resemble perfect capacitors. This has to beaccounted for by measuring a reference phase shift∆ ϕ ref without plasma. A powerful method forcalculating the time shift between two arbitrary signalsis the cross-correlation method. This time shift ∆ t ref is directly related to the phase shift ∆ ϕ ref .Calculating the cross correlation coefficient Ψ UI for two infinite, continuous, sinusoidal signals analyti-cally leads to a cosine-shaped functionΨ UI ( τ ) = U rms I rms cos( ω (∆ t − τ )) . (2)The shift ∆ t ref between the two signals canbe calculated from the lag τ max of the maximumcorrelation coefficient closest to zero lag, because thecosine has its highest value at τ = ∆ t . In reality,the measured waveforms are finite and discretelysampled by the oscilloscope. The so-called unbiasedcorrection of the cross-correlation compensates for thefinite nature. To overcome the limitation of theaccuracy of ∆ t ref due to the time resolution of theoscilloscope, a second order polynomial fit to themaximum correlation coefficient near zero lag is used.The lag τ max is calculated using the vertex formulawith the coefficients of the fit. After correcting the timeshift between voltage and current without plasma by∆ t ref , the voltage and current waveforms with plasmacan be multiplied point by point to obtain the period-average power supplied to the plasma P avg = 1 N N (cid:88) n =1 U n I n , (3) nergy transport in an atmospheric pressure plasma jet N is the number of sampled data points.As the reference time shift from the polynomial fit ismore accurate than the time resolution of the sampledsignals, the current is linearly interpolated after thecorrection to allow point-wise multiplication.The average plasma input power as a functionof the root-mean-square voltage in a range between200 V rms and 600 V rms is shown in blue in figure 3.The gas flow consisted of helium with 0 . rms , whichis indicated by a jump of the power from 0 W to0 . .
75 W at 400 V rms . At higher voltages,the dependence of the power on the voltage changesfrom linear to quadratic. This change is induced bya change from the so-called α -mode to the γ -mode asalready observed and discussed e.g. for the COST-Jet by Golda et al. [12]. A change of the plasmaemission from a homogeneous glow in the middlebetween the electrodes to a bright emission directlyat both electrodes is observed in this power region.Ohmic heating of the electrons in the plasma bulk isdominant in the α -mode, so it is also referred to as Ω-mode [13]. The γ -mode is dominated by secondaryelectrons emitted from the electrodes or created byPenning processes in the sheaths. For this reason, itis also called Penning-mode in the literature [14, 15].In contrast to the COST-Jet, which transitions to aconstricted discharge at the electrode tips when furtherincreasing the voltage, the dielectric in the capillaryjet prevents glow-to-arc transition and the plasma isstable also at high voltages [16, 17]. The maximummeasured power in this work is 10 W. The orange datais the power without the plasma. This curve has beenmeasured with ambient air inside the capillary, whichleads to a much higher breakdown voltage. The poweris zero over the whole voltage range, confirming thatthe correction of the time shift is working as expected.The uncertainty of the measured powers ∆ P avg depends partly on the systematic uncertainties ofthe used resistors and the calibration of the voltagemeasured by the pick-up antenna to the actual voltageat the electrode. This fraction is calculated usingGaussian error propagation. The other part of ∆ P avg depends on the uncertainty of the measured referencephase shift ∆(∆ t ref ). It is calculated using theuncertainty of the lag-value ∆( τ max ) of the maximumcross correlation coefficient. The uncertainty of the lagis derived from the vertex formula and the polynomialfit coefficients using Gaussian error propagation. Theuncertainty values are smaller than the size of themarkers in figure 3 and are therefore not shown here.
200 300 400 500 600 rms voltage / V po w e r / W plasma onplasma off Figure 3.
Power coupled to the plasma as a function of theroot-mean-square voltage measured with the cross-correlationmethod. The gas mixture consisted of 1000 sccm helium and5 sccm oxygen. plasma jetgas flow rubber seal thermocouple glas tube gas flow
10 mm spectrometer Echelle spectrometer
Figure 4.
Schematic diagram of the experimental setup usedfor the optical measurements parallel and perpendicular to thegas flow direction and the measurement of the gas temperaturein the effluent.
The optical setup used in this work consisted ofa combination of two spectrometers measuring theplasma emission parallel and perpendicular to the gasflow direction (figure 4). Parallel to the gas flow, theplasma was imaged into an optical fiber connectedto a relatively calibrated spectrometer (Ocean OpticsHR4000) by a lens with a focal length of 20 mm.One-to-one imaging of the plasma perpendicularto the gas flow was performed using two lenses.The first lens had a focal length of 35 mm and thesame distance to the plasma. Therefore, the plasmaemission was parallelized and afterwards focused intoan optical fiber in the focal point of a second lenswith a focal length of 50 mm. This optical fiber wasconnected to an absolutely calibrated high-resolutionbroadband Echelle spectrometer (LLA InstrumentsESA4000 plus).An example spectrum of the plasma in γ -modewith 0 . and traces of argon taken parallel tothe gas flow is shown in figure 5 with the dominantemission features marked by labeled arrows. For bettervisibility, the intensities below 650 nm have been scaled nergy transport in an atmospheric pressure plasma jet
200 300 400 500 600 700 800 900 wavelength / nm i n t en s i t y / a r b . u . O 777.3 nmHe 706.5 nmAr 750.4 nmN intensity x8 Figure 5.
Relatively calibrated spectrum of the plasma for 5 Winput power and a gas mixture of 1000 sccm helium, 0 . up by a factor of 8. The emission is dominated byhelium and atomic oxygen lines with the only clearlyvisible argon line at 750 . As a second non-invasive method, we used thermo-couple measurements in the effluent to obtain the gastemperature. The setup is also shown in figure 4.A thermocouple (Type-K) was positioned in front ofthe capillary and sealed using a glas tube to ensurea laminar gas flow pattern. A rubber seal was posi-tioned around the capillary to avoid ambient intrusioninto the glass tube. The signal of the thermocouplewas recorded and converted to temperature by an in-house-built measurement electronic originally designedfor passive thermal probe measurements [18]. This al-lowed measurements with 90 Hz repetition rate for in-vestigation of fast changes in the gas temperature. Aswe could only measure in the effluent about 11 mmaway from the plasma, we had to use an extrapolationtechnique to estimate the temperature in the plasma.With a distance variation, we could extrapolate themeasured gas temperature at the end of the electrodesto be approximately 21 % higher than in the effluentwith respect to a room temperature of 20 ° C. We used molecular beam mass spectrometry (MBMS)to obtain absolute densities of atomic oxygen and ozoneon the central axis of the plasma effluent. Sampling ofthe species from atmospheric pressure was performedusing a setup consisting of three differential pumpingstages. Gas enters the first stage through an orificewith a diameter of 50 µ m, which has been laser drilledinto a molybdenum plate of 250 µ m thickness. Thefirst stage is pumped by a scroll pump, while the second and third stage are evacuated using turbo-molecular pumps. A moving chopper with a skimmerwas used in the first pumping stage to achieve highbeam-to-background ratios and to further reduce thebackground pressure. This setup is able to reach thenecessary UHV-conditions in the third stage housingthe mass spectrometer. Details on the setup and themoving skimmer concept can be found in the literature[19–21]. This applies also to the setup and basics of thequadrupole mass spectrometer (HIDEN EPIC1000)used in this study [22]. Instead of the rotating skimmerconcept shown in the literature, we used a linearlymoving chopper here.As we expected gas temperatures of up to a fewhundred ° C based on preliminary measurements, astainless steel cooling plate was mounted onto theorifice plate to ensure constant temperature conditions.This introduced limitations on the jet setup, sosampling of the plasma species through the orifice wasperformed about 15 mm behind the plasma. To avoidmixing with ambient air and consequently turbulenceresulting in enhanced recombination of reactive species,the capillary extended 14 mm of this distance. Theresulting gap between the capillary tip and the orificewas 1 mm. This was achieved by moving the electrodesshown in figure 4 4 mm to the left.Calibration of the measured atomic oxygen signalhas been performed by admixing small amountsof neon (0 . in the ionizer, we usedthreshold ionization mass spectrometry (TIMS) withelectron energies of 15 eV and 25 eV for atomic oxygenand neon, respectively. Ozone density calibration wasrealised in the same way by admixing 0 .
3. Results and discussion
The following discussion of the experimental results isbased on the assumption, that the power dissipatedin the plasma is coupled to the electrons. Thisassumption is valid, because the ion with the highestdensity in He/O -discharges is O , which has amuch lower plasma frequency than the RF-frequency[2]. Energy loss by ions accelerated in the sheathsis also neglected. Due to the high collisionalityin the sheath region, the ion energy is directlyconverted to a temperature increase of the feed gas. nergy transport in an atmospheric pressure plasma jet . α -mode) and 5 W ( γ -mode).Oxygen admixtures ranged from 0 % to 2 % due tolimitations of the setup. Although energy transfer of electrons to atoms andmolecules by elastic collisions is negligible at lowpressures, it has to be taken into account at highpressure due to high collision frequencies. This is whywe focused on heat as one significant component ofenergy transport.In addition to thermocouple measurements inthe effluent with the setup described in section 2.4,we used the evaluation of molecular rotationalspectra to derive the rotational temperature andget a second estimate for the gas temperature inthe plasma. The high resolution of the Echelle-spectrometer allowed for identification of individuallines in rotational spectra. For atmospheric pressureplasmas, typically the hydroxyl radical (OH) or thenitrogen molecular ion (N ) are used [23]. Comparingthe rotational temperatures of both species, thetemperature determined from the nitrogen ion emissionwas way higher than the one from OH. It is wellknown, that especially for helium plasmas, the nitrogenlevels often interact with those of helium, which leadsto an overestimation of the gas temperature [24, 25].Therefore, we decided for the rotational temperature ofOH as an estimate for the gas temperature and used abubbler system to admix a controlled amount of watervapor to the feed gas. Details about the bubbler-setupand the control of the water admixture to the plasmafeed gas can be found in the literature [26]. To avoidany influence on the plasma discharge, the admixturewas always kept around 10 ppm. The low admixtureis also important as otherwise an overpopulation ofhigher rotational states is probable [27]. The analysisof the OH spectra was realized using the software massiveOES by Vor´aˇc et al. [28–30] to fit a syntheticrotational spectrum to the measured one. power / W t e m pe r a t u r e / ° C rotational temperature of OHextrapolated effluent gas temperature oxygen admixture / % t e m pe r a t u r e / ° C rotational temperature of OHextrapolated effluent gas temperature Figure 6.
Gas temperature in the plasma extrapolated from theeffluent ( ◦ ) compared to the rotational temperature from the A-X transition of the OH-radical ( × ) as a function the dissipatedpower at constant admixture of 0 . (upper figure) andoxygen admixture at constant input power of 1 W (lower figure).The rest of the feed gas consisted of 1000 sccm helium and0 . The temperatures obtained using both methodsare shown in figure 6 for different input powersat constant admixture and different admixtures atconstant power. Overall, the gas temperature increasesnearly linearly with increasing input power in theobserved range from 50 ° C at 0 . ° C at 6 W.The slope of the linear increase is lower for powersabove 3 . nergy transport in an atmospheric pressure plasma jet ° C. Thetemperature increase between no admixture and 0 . ° C lower than the extrapolatedeffluent gas temperature. A possible explanation is theposition of the measurement. The extrapolation fromthe effluent gives the gas temperature at the end of theelectrodes where the gas has the highest temperature[31]. In contrast, the spectra were recorded at aposition about 1 cm before the end of the electrodes,leading to a shorter residence time of the gas in theplasma and therefore lower temperature. The gastemperature under variation of the oxygen admixturefor 5 W input power isn’t shown here, because theobserved trend is the same as for low input power apartfrom a higher overall temperature and an even lowerOH emission intensity leading to a larger uncertaintyof the rotational temperature.The gas temperature measured with the thermo-couple is used in the energy balance calculations, asit has a lower uncertainty than the rotational temper-ature and is a more accurate estimate for the totaltemperature increase along the plasma channel.
We measured the atomic oxygen density in thedischarge using actinometry and molecular beam massspectrometry. For the analyzed oxygen admixtures,we assume oxygen dissociation being the dominantchemical reaction in the plasma initiated by electrons.Therefore, measuring the density of atomic oxygengives a good estimation for the power investedin plasma chemical processes. As the subsequentformation of ozone from atomic and molecular oxygenis efficient in oxygen-containing discharges, the ozonedensity has been studied as well and was included inthe estimations for power invested in generation of finalchemical products.
Actinometry is based on therelation between the emission intensity of an opticaltransition and the ground-state density of the emittingspecies. As the emission also depends on the electronenergy distribution in the energy range for excitationinto the emitting state, a actinometer gas is addedto the gas mixture. Argon is used in our case withadmixtures as low as 0 .
04 %. More details on thetheoretical basics and validity range of the methodcan be found in the literature [35–37]. The densityof atomic oxygen is calculated via n O = I O I Ar ν Ar ν O k ∗ e,Ar k ∗ e,O a ik,Ar a ik,O n Ar − k ∗ de,O k ∗ e,O n O . (4)The quantities in equation 4 are the emissionintensities I O of the oxygen transition at 844 . I Ar of the argon transition at 750 . ν O and ν Ar of the transitions, the effective excitationrate coefficients k ∗ e , the effective optical branchingratios a ik , the effective dissociative excitation ratecoefficient k ∗ de and the densities n Ar and n O of argonand molecular oxygen respectively. Calculation of theeffective rate coefficients k ∗ requires knowledge aboutthe averaged electron energy distribution function(EEDF). The determination of the EEDF is beyond thescope of this work, so we used the rate coefficient ratioscalculated by Niemi et al. [37]. The emission intensitieswere derived from the spectra measured parallel to thegas flow. The atomic oxygen line lies outside of thewavelength range of the Echelle spectrometer, so itcould not be used for this measurement. power / W a t o m i c o xy gen den s i t y / m - Figure 7.
Atomic oxygen density from actinometry as afunction of plasma input power. The gas flow consisted of1000 sccm helium, 0 . The upper graph in figure 9 shows the atomicoxygen density calculated from the optical emissionspectra as a function of the plasma input power. Forpowers up to 4 W, the density rises linearly from8 × m − to 2 . × m − . Above 4 W, the slopechanges and the density increase is less steep, leadingto a density of 2 . × m − at 6 W. Comparableabsolute densities have been determined by Knake etal. using TALIF and Niemi et al. using diagnosticbased modeling for similar discharges [37, 38]. As bothauthors do not give the plasma input power, directcomparison of the observed trends with our results isdifficult. The change in the slope is not induced bythe plasma mode change, as this happens at muchlower powers (see section 2.2). A possible explanationfor the overall sub-linear increase of the density is the nergy transport in an atmospheric pressure plasma jet oxygen admixture / % a t o m i c o xy gen den s i t y / m - Figure 8.
Atomic oxygen density from actinometry as afunction of oxygen admixture for constant input power of 1 W( × , α -mode) and 5 W ( ◦ , γ -mode). The main gas mixtureconsisted of 1000 sccm helium and 0 . Figure 8 shows the dependence of the atomicoxygen density measured by actinometry on the oxygenadmixture for constant input power of 1 W ( × ) and5 W ( ◦ ). At 1 W, the atomic oxygen density shows asteep increase up to 1 . × m − at 0 . . × m − at1 . . × m − at 2 %. For an input powerof 5 W, the density rises to 2 × m − between noadmixture and 0 . × m − at 1 . . We used MBMS as a second methodto verify the atomic oxygen densities obtained byactinometry and to measure the ozone density in theeffluent.The atomic oxygen density under variation ofthe input power measured by MBMS ( × ) is shownin figure 9. The atomic oxygen density increasesalmost linearly with the power up to 7 × m − at1 . . × m − above 5 W.The absolute densities are about a factor of 20lower than the values measured by actinometry whileshowing comparable trends. An explanation is the power / W a t o m i c o xy gen den s i t y / m - o z one den s i t y / m - Figure 9.
Atomic oxygen density from actinometry as afunction of plasma input power. The gas flow consisted of1000 sccm helium, 0 . measurement in the effluent with a distance of 15 mmto the plasma. This leads to a density decrease ofa factor of 5 to 6 compared to inside the plasmaas deduced from literature with a controlled heliumatmosphere [10, 34] and with an enclosed effluent asit is the case in our setup [40]. When comparingour results with the given literature, the lower gasflow in our case has to be taken into account. Thisdifference in flow velocity leads to increased atomicoxygen recombination in the effluent. Also, thesystematic uncertainty of values measured by MBMSis about a factor of 2 resulting from the uncertaintyof ionization cross-sections and the density calibration[10]. Another explanation is an overestimation ofthe density by actinometry, most likely due to theuse of effective rate coefficients from the literatureand the resulting assumptions about the EEDF. Also,actinometry as an optical method always probes theplasma volume where most excitation processes takeplace. This may also be the volume in which the mostchemical reactions happen, so the the obtained atomicoxygen density is not volume-averaged in contrast tothe MBMS measurements. In conclusion, the actualatomic oxygen density inside the plasma is expected tobe between the density obtained by actinometry as anupper limit and the extrapolated density from MBMSin the effluent as the lower limit.The ozone density ( (cid:5) ) rises steeply with dischargeignition to 1 . × m − at 0 . × m − above 3 W. Thisdecrease is most likely induced by the rising gastemperature (see figure 6). Ellerweg et al. measuredozone densities up to 5 × m − with similar oxygenadmixture under discharge voltage variation for theCOST-Jet corresponding to the low end of ourobserved power range [10]. The different distances to nergy transport in an atmospheric pressure plasma jet oxygen admixture / % a t o m i c o xy gen den s i t y / m - o z one den s i t y / m - oxygen admixture / % a t o m i c o xy gen den s i t y / m - o z one den s i t y / m - Figure 10.
Atomic oxygen ( × ) and ozone ( (cid:5) ) densitiesmeasured by MBMS as a function of the oxygen admixturefor constant input power of 1 W and 5 W. The main gas flowconsisted of 1000 sccm helium. Traces of argon were added forMBMS of ozone (0 . . The atomic oxygen density ( × ) as a function ofthe oxygen admixture measured using MBMS is shownin figure 10. Again, the input power has been keptconstant at 1 W and 5 W to study α - and γ -modebehaviour. At 1 W, the density of atomic oxygen risessteeply for low admixtures up to 5 . × m − at0 . . × m − at 0 . × m − at 2 %. The absolute densities lie below the values measured at an inputpower of approximately 0 . . × m − at 1 . γ -mode is still not observed, eventhough Park et al. suggested a shift of the maximumobserved at low input power to higher admixtures withincreasing power based on simulations [41]. The atomicoxygen density for a plasma jet in γ -mode and forthis admixture range has to our knowledge not beenreported before.The ozone density as a function of the oxygenadmixture in figure 10 shows an over-linear increasewith no maximum being reached in the observedadmixture ranges for both input powers. Thisincrease of the ozone density also explains the decreaseof atomic oxygen observed at 1 W above 0 . . × m − is reached at the highest admixture of2 %. The ozone density at 5 W input power and 2 %admixture is about a factor of 10 lower than at lowinput power. A possible explanation is the overalllower gas density due to the higher gas temperatureand increased ozone loss reactions. The highest ozonedensity of 4 . × m − for 5 W input power isreached at the maximum oxygen admixture of 4 %.Comparable trends have again been measured byEllerweg et al. in the COST-Jet [10]. At 1 W inputpower, the absolute ozone densities are about a factorof 2 lower than in the COST-Jet. Possible reasons forthis difference have been discussed above. Inelastic collisions of electrons with other plasmaspecies also lead to excitation of these species. Oneof the main spontaneous relaxation processes is theemission of radiation. As the measurement ofabsolute VUV/UV photon fluxes is very challengingand technically demanding, we limited the followingmeasurements to the visible wavelength region usingthe absolutely calibrated Echelle spectrometer. Theemission intensity has been calculated as the sumover all photons per second I i in a wavelength range∆ λ i emitted from an effective plasma volume V eff .This volume takes into account the solid angle fromwhich radiation is detected by the spectrometer. Theassociated formula is nergy transport in an atmospheric pressure plasma jet I rad = V p V eff (cid:88) i I i ∆ λ i . (5)Multiplication with the plasma volume V p scalesthe intensity detected from the effective volume up tothe overall emitted intensity. In the case of the Echellespectrometer, the covered wavelength range is 200 nmto 780 nm. power / W i n t en s i t y / pho t on s s - Figure 11.
Absolutely calibrated intensity of visible radiationas a function of input power. The gas mixture consisted of1000 sccm helium, 0 . Figure 11 shows the intensity in photons s − emitted in the visible wavelength range as a functionof the input power in a range up to 6 W. Theintensity rises approximately linear with the inputpower from 0 . × photons s − at 0 . × photons s − at 6 W. This behaviour isexpected, since the plasma becomes visibly brighterwith rising power.From the spectrum in figure 5 it is evident, thatline radiation dominates. Therefore low emitted poweris expected in the measured wavelength range. We alsodo not expect high radiation power in the VUV/UV-region, as the spectrum is also mostly dominated byline radiation despite of the helium excimer continua.However, the intensity of this continuum radiationdrops with oxygen admixture and should be negligiblefor all admixtures observed in this work [42]. If theatomic oxygen lines in the VUV/UV-region made upapproximately 1 % of the input power, the number ofexcited oxygen atoms would have to be in the order of1 × m − . We estimated the density of the excited S ◦ -state of oxygen to be around 1 × m − using aMaxwellian electron energy distribution with 2 eV andcross-sections from Laher and Gilmore [43]. As thecalculated densities of the other excited states fromwhich emission is observed in this wavelength regionare orders of magnitude lower than of the S ◦ -stateunder the same conditions, we assume energy loss viaVUV/UV-radiation to be negligible. Jonkers et al. even omitted energy losses due to radiation in theenergy balance for their analytical model because of itssmall contribution [7]. Possible sources of uncertaintyfor the measured powers are the absolute calibrationof the spectrometer, the calculation of the effectivelyobserved plasma volume and the gaps in the Echelle-spectrum [44]. Because of the expected very smallcontribution of radiation to the overall power balance,we did not calculate the uncertainty here. oxygen admixture / % i n t en s i t y / pho t on s s - Figure 12.
Absolutely calibrated intensity of visible radiationas a function of oxygen admixture for constant input power of1 W. The main gas mixture consisted of 1000 sccm helium and0 . The emission intensity as a function of the oxygenadmixture up to 2 % is shown in figure 12 for aconstant input power of 1 W. The highest intensityof 3 . × photons s − is measured for no oxygenadmixture. The intensity then drops steeply tojust below 1 × photons s − at 0 . . × photons s − at2 %. Even though the plasma input power has beenkept constant for all admixtures by increasing theapplied voltage, the plasma became darker with risingadmixture. This qualitatively confirms the observedtrend of the intensity. As stated above, we expect theradiation in the VUV/UV-region to follow the sametrend with negligibly small absolute values. We alsomeasured the radiation power for various admixtures ata constant power of 5 W and observed the same trendwith higher absolute values as expected from the inputpower variation. Using the the plasma input power as a reference, wewere able to combine the results of the presentedmeasurements to obtain a power balance of the plasmain the capillary jet under variation of the externalparameters power (figure 13) and oxygen admixture(figure 14). The three measured contributions tothe overall dissipated power (dashed line) are heat nergy transport in an atmospheric pressure plasma jet ◦ ), generation of atomic oxygen and ozone as finalchemical products ( + ) and radiation in the visiblewavelength range ( ∗ ). The sum of the contributions isalso shown ( × ). If the power balance was complete, thesum of the contributions would give the input power(dashed line).To calculate the power put into heating of the feedgas, we took into account the heat Q = c He m ∆ T (6)stored in the helium fraction of the gas and theresidence time t = pp s T s T V p Φ s (7)of the gas in the plasma channel that can becalculated from the gas flow Φ s under standardconditions. The other quantities are the heat capacityof helium c he , the mass of helium in the plasma volumeat the current gas temperature m , the temperaturechange ∆ T , the pressure p , the current temperature T (under standard conditions p s and T s ) and the plasmavolume V p . The resulting equation for the power is P heat = Qt = c he m he ∆ T p s Φ s k B T s (8)with the atomic weight of helium m he = 4 u andthe Boltzmann constant k B .From the measured atomic oxygen and ozonedensities, we calculated the fraction of the power whichis dissipated by the electrons in inelastic collisionswith oxygen molecules leading to oxygen dissociationand subsequently to ozone production. The atomicoxygen densities obtained by actinometry were usedhere to estimate the maximum power put into oxygendissociation. This may be an overestimation for thisspecific reaction, but as we neglected other chemicalreactions, e.g. with impurities like nitrogen or water, itmay be a good estimation for the overall power put intoplasma chemistry. The energy E chem needed to createthe number of oxygen atoms and ozone molecules inthe discharge volume V p is calculated via E chem = (cid:18) n O E d n O E f (cid:19) V p N A . (9)The other quantities in equation 9 are the densityof atomic oxygen n O , the Avogadro-constant N A , thedissociation energy E d of molecular oxygen and theformation energy E f of ozone in J mol − . Dividingthis value by the residence time t of the species in theplasma (equation 7) gives the dissipated power P chem = (cid:18) n O E d n O E f (cid:19) N A p s p TT s Φ s . (10) Using the number of photons per second I i ina wavelength range ∆ λ i , the power dissipated byradiation P rad = V p V eff hc (cid:88) i I i ∆ λ i λ i (11)can be calculated. adjusted power / W po w e r / W totalgas heatingfinal productsradiation Figure 13.
Power balance of the capillary plasma jet as afunction of the input power. The gas mixture consisted of1000 sccm helium, 0 . With variation of the plasma input power in arange from 0 . . . . nergy transport in an atmospheric pressure plasma jet . have been admixed tothe feed gas. The most atomic oxygen per W inputpower is produced at 0 . × − W. Moravej et al. calculatedthe emitted power density of an atmospheric pressuredischarge with an electron temperature of T e = 2 eVand an electron density of n e = 1 × m − to beapproximately 4 × − W m − [45]. The values forthe electron temperature and density are in a realisticrange for a discharge similar to the capillary jet [39].Taking into account the plasma volume in this study,we obtain power densities around 2 . × − W m − .Due to the uncertainties in the calculation of theemission intensity and power, a difference of one orderof magnitude is still a good agreement. oxygen admixture / % po w e r / W totalgas heating final productsradiation Figure 14.
Power balance of the capillary plasma jet as afunction of the oxygen admixture for constant input power of1 W. The rest gas flow consisted of 1000 sccm helium and0 . The variation of the oxygen admixture shows thesame distribution of the input power between the threemeasured contributions (figure 14). This distributionseems basically independent of the molecular oxygenadmixture in the observed range. This holds true foradmixtures higher than 0 .
4. Conclusion electron heating plasma powerheat chemical products radiation generator powerelastic collisions inelastic collisionselectric losses > 90%< 10%100%> 60 % < 10 % << 1 % final productsgas heating additional losses < 30 %
Ohmic heating dissociation/ionizationrecombination excitationmeasured quantities
Figure 15.
Energy transport diagram of the atmosphericpressure RF-driven capillary plasma jet.
The energy transport in the atmospheric pressurerf-driven capillary plasma jet is summarised in thediagram shown in figure 15. The power measurementenabled us to estimate the fraction of the generatorpower that is coupled into the plasma, to be below10 %. This is confirmed by results from Stewig et al.for a comparable plasma device featuring a capillary asdielectric [46]. The power is selectively coupled to theelectrons and distributed between the plasma speciesvia elastic and inelastic collisions. Between 60 % and100 % of the power are converted to heat while lessthan 10 % is used to initiate chemical processes. Powerlosses due to radiation make up a negligible amountof the dissipated power. We identified 0 % to 30 %of the dissipated power to be additional losses whichwere not quantified. These losses are most likely due toheat transfer from the gas to the setup and surroundingair. For this reason, active cooling of the setup wouldlead to lower gas temperatures when needed for specificapplications. Using a shorter plasma channel or highergas flow may also reduce the gas temperature butmight have an influence on the generation of atomicoxygen as well. Overall, the parameters power andoxygen admixture do not substantially change theenergy transport into the different channels. To furtherexplore the possibility of energy transport control,an excitation voltage waveform or frequency variationmight be promising via influencing the EEDF.
5. Acknowledgements
The authors would like to thank Luka Hansen andHolger Kersten from the Plasma Technology Groupat Kiel University for help with the temperaturemeasurements. Volker Rohwer, Michael Poserand Mario Kn¨uppel are gratefully acknowledged for
EFERENCES
Cold atmospheric plasmasfor material synthesis: production of silicon quantumdots and particle free thin-film deposition (project-IDBE 4349/7-1) and the project
PlasNOW - Plasmagenerated Nitric Oxide in Wound healing (project-IDSCHU 2353/9-1). JG acknowledges funding by theFaculty of Mathematics and Natural Sciences, KielUniversity, Germany, in the framework of the Awardfor Young Women Scientists 2019.
References [1] U. Kogelschatz, “Atmospheric-pressure plasmatechnology”, Plasma Physics and ControlledFusion , B63–B75 (2004).[2] C. Lazzaroni, P. Chabert, M. A. Lieberman,A. J. Lichtenberg, and A. Leblanc, “Analytical–numerical global model of atmospheric-pressureradio-frequency capacitive discharges”, PlasmaSources Science and Technology , 035013(2012).[3] J. Ehlbeck, U. Schnabel, M. Polak, J. Winter, T.von Woedtke, R. Brandenburg, T. von dem Ha-gen, and K.-D. Weltmann, “Low temperature at-mospheric pressure plasma sources for microbialdecontamination”, Journal of Physics D: AppliedPhysics , 013002 (2011).[4] D. B. Graves, “The emerging role of reactiveoxygen and nitrogen species in redox biologyand some implications for plasma applicationsto medicine and biology”, Journal of Physics D:Applied Physics , 263001 (2012).[5] G. Y. Park, S. J. Park, M. Y. Choi, I. G. Koo,J. H. Byun, J. W. Hong, J. Y. Sim, G. J. Collins,and J. K. Lee, “Atmospheric-pressure plasmasources for biomedical applications”, PlasmaSources Science and Technology , 043001(2012).[6] J. D. Lawson, “Some Criteria for a Power Pro-ducing Thermonuclear Reactor”, Proceedings ofthe Physical Society, Section B , 6–10 (1957).[7] J. Jonkers, M. van de Sande, A. Sola, A.Gamero, and J. J. A. M. van der Mullen,“On the differences between ionizing heliumand argon plasmas at atmospheric pressure”,Plasma Sources Science and Technology , 30–38 (2003).[8] J. Golda, J. Held, B. Redeker, M. Konkowski,P. A. C. Beijer, A. Sobota, G. M. W. Kroesen,N. S. J. Braithwaite, S. Reuter, M. M. Turner,T. Gans, D. O’Connell, and V. Schulz-von derGathen, “Concepts and characteristics of the ‘COST Reference Microplasma Jet’”, Journal ofPhysics D: Applied Physics , 029503 (2016).[9] J. Golda, F. Kogelheide, P. Awakowicz, andV. Schulz-von der Gathen, “Dissipated electricalpower and electron density in an RF atmosphericpressure helium plasma jet”, Plasma SourcesScience and Technology , 095023 (2019).[10] D. Ellerweg, J. Benedikt, A. von Keudell, N.Knake, and V. Schulz-von der Gathen, “Charac-terization of the effluent of a He/O2 micro-scaledatmospheric pressure plasma jet by quantitativemolecular beam mass spectrometry”, New Jour-nal of Physics , 059501 (2010).[11] J. Park, I. Henins, H. W. Herrmann, G. S.Selwyn, and R. F. Hicks, “Discharge phenomenaof an atmospheric pressure radio-frequencycapacitive plasma source”, Journal of AppliedPhysics , 20–28 (2001).[12] J. Golda, J. Held, and V. Schulz-von der Gathen,“Comparison of electron heating and energy lossmechanisms in an RF plasma jet operated inargon and helium”, Plasma Sources Science andTechnology , 1–13 (2020).[13] T. Hemke, D. Eremin, T. Mussenbrock, A.Derzsi, Z. Donk´o, K. Dittmann, J. Meichsner,and J. Schulze, “Ionization by bulk heating ofelectrons in capacitive radio frequency atmo-spheric pressure microplasmas”, Plasma SourcesScience and Technology , 015012 (2013).[14] L. Bischoff, G. H¨ubner, I. Korolov, Z. Donk´o,P. Hartmann, T. Gans, J. Held, V. Schulz-von der Gathen, Y. Liu, T. Mussenbrock, andJ. Schulze, “Experimental and computationalinvestigations of electron dynamics in microatmospheric pressure radio-frequency plasma jetsoperated in He/N 2 mixtures”, Plasma SourcesScience and Technology , 125009 (2018).[15] A. R. Gibson, Z. Donk´o, L. Alelyani, L. Bischoff,G. H¨ubner, J. Bredin, S. Doyle, I. Korolov,K. Niemi, T. Mussenbrock, P. Hartmann, J. P.Dedrick, J. Schulze, T. Gans, and D. O’Connell,“Disrupting the spatio-temporal symmetry ofthe electron dynamics in atmospheric pressureplasmas by voltage waveform tailoring”, PlasmaSources Science and Technology , 01LT01(2019).[16] J. J. Shi, D. W. Liu, and M. G. Kong,“Plasma stability control using dielectric barriersin radio-frequency atmospheric pressure glowdischarges”, Applied Physics Letters , 081502(2006). EFERENCES ,031505 (2007).[18] M. Stahl, T. Trottenberg, and H. Kersten, “Acalorimetric probe for plasma diagnostics”, TheReview of Scientific Instruments , 023504(2010).[19] J. Benedikt, D. Ellerweg, and A. von Keudell,“Molecular beam sampling system with veryhigh beam-to-background ratio: The rotatingskimmer concept”, The Review of scientificinstruments , 055107 (2009).[20] J. Benedikt, D. Ellerweg, S. Schneider, K.R¨ugner, R. R, H. Kersten, and T. Benter, “Massspectrometry of positive ions and neutral speciesin the effluent of an atmospheric pressure plasmawith hexamethyldisiloxane and oxygen”, Journalof Physics D: Applied Physics , 464017 (2013).[21] S. Große-Kreul, S. H¨ubner, S. Schneider, D.Ellerweg, A. von Keudell, S. Matejˇc´ık, and J.Benedikt, “Mass spectrometry of atmosphericpressure plasmas”, Plasma Sources Science andTechnology , 044008 (2015).[22] J. Benedikt, A. Hecimovic, D. Ellerweg, and A.von Keudell, “Quadrupole mass spectrometry ofreactive plasmas”, Journal of Physics D: AppliedPhysics , 403001 (2012).[23] P. J. Bruggeman, N. Sadeghi, D. C. Schram, andV. Linss, “Gas temperature determination fromrotational lines in non-equilibrium plasmas: Areview”, Plasma Sources Science and Technology , 023001 (2014).[24] A. Ionascut-Nedelcescu, C. Carlone, U. Ko-gelschatz, D. V. Gravelle, and M. I. Boulos, “Cal-culation of the gas temperature in a throughflowatmospheric pressure dielectric barrier dischargetorch by spectral line shape analysis”, Journal ofApplied Physics , 063305 (2008).[25] S. Hofmann, A. F. H. van Gessel, T. Verreycken,and P. J. Bruggeman, “Power dissipation, gastemperatures and electron densities of coldatmospheric pressure helium and argon RFplasma jets”, Plasma Sources Science andTechnology , 065010 (2011).[26] J. Benedikt, D. Schr¨oder, S. Schneider, G.Willems, A. Pajdarov´a, J. Vlˇcek, and V.Schulz-von der Gathen, “Absolute OH and Oradical densities in effluent of a He/H2O micro-scaled atmospheric pressure plasma jet”, PlasmaSources Science and Technology , 045013(2016). [27] P. J. Bruggeman, F. Iza, P. Guns, D. Lauwers,M. G. Kong, Y. A. Gonzalvo, C. Leys, andD. C. Schram, “Electronic quenching of OH(A)by water in atmospheric pressure plasmas and itsinfluence on the gas temperature determinationby OH(A–X) emission”, Plasma Sources Scienceand Technology , 015016 (2010).[28] J. Vor´aˇc and P. Synek, massiveOES , 2017.[29] J. Vor´aˇc, P. Synek, L. Potoˇcˇn´akov´a, J. Hnil-ica, and V. Kudrle, “Batch processing of over-lapping molecular spectra as a tool for spatio-temporal diagnostics of power modulated mi-crowave plasma jet”, Plasma Sources Science andTechnology , 025010 (2017).[30] J. Vor´aˇc, P. Synek, V. Proch´azka, and T. Hoder,“State-by-state emission spectra fitting for non-equilibrium plasmas: OH spectra of surfacebarrier discharge at argon/water interface”,Journal of Physics D: Applied Physics , 294002(2017).[31] S. Kelly, J. Golda, M. M. Turner, and V.Schulz-von der Gathen, “Gas and heat dynamicsof a micro-scaled atmospheric pressure plasmareference jet”, Journal of Physics D: AppliedPhysics , 444002 (2015).[32] F. Riedel, J. Golda, J. Held, H. L. Davies,M. W. van der Woude, J. Bredin, K. Niemi,T. Gans, V. Schulz-von der Gathen, and D.O’Connell, “Reproducibility of ‘COST referencemicroplasma jets’”, Plasma Sources Science andTechnology , 095018 (2020).[33] J. Y. Jeong, J. Park, I. Henins, S. E. Babayan,V. J. Tu, G. S. Selwyn, G. Ding, and R. F.Hicks, “Reaction Chemistry in the Afterglowof an Oxygen − Helium, Atmospheric-PressurePlasma”, The Journal of Physical Chemistry A , 8027–8032 (2000).[34] D. Ellerweg, A. von Keudell, and J. Benedikt,“Unexpected O and O3 production in theeffluent of He/O2 microplasma jets emanatinginto ambient air”, Plasma Sources Science andTechnology , 034019 (2012).[35] R. E. Walkup, K. L. Saenger, and G. S. Sel-wyn, “Studies of atomic oxygen in O2+CF4 rfdischarges by two–photon laser–induced fluores-cence and optical emission spectroscopy”, Jour-nal of Chemical Physics , 2668–2674 (1986).[36] H. M. Katsch, A. Tewes, E. Quandt, A. Goehlich,T. Kawetzki, and H. F. D¨obele, “Detection ofatomic oxygen: Improvement of actinometry andcomparison with laser spectroscopy”, Journal ofApplied Physics , 6232–6238 (2000). EFERENCES , 151504(2009).[38] N. Knake, K. Niemi, S. Reuter, V. Schulz-vonder Gathen, and J. Winter, “Absolute atomicoxygen density profiles in the discharge core ofa microscale atmospheric pressure plasma jet”,Applied Physics Letters , 131503 (2008).[39] J. Waskoenig, K. Niemi, N. Knake, L. M. Gra-ham, S. Reuter, V. Schulz-von der Gathen,and T. Gans, “Atomic oxygen formation in aradio-frequency driven micro-atmospheric pres-sure plasma jet”, Plasma Sources Science andTechnology , 045018 (2010).[40] N. Bibinov, N. Knake, H. Bahre, P. Awakowicz,and V. Schulz-von der Gathen, “Spectroscopiccharacterization of an atmospheric pressure μ -jetplasma source”, Journal of Physics D: AppliedPhysics , 345204 (2011).[41] G. Y. Park, Y. J. Hong, H. W. Lee, J. Y. Sim, andJ. K. Lee, “A Global Model for the Identificationof the Dominant Reactions for Atomic Oxygen inHe/O2 Atmospheric-Pressure Plasmas”, PlasmaProcesses and Polymers , 281–287 (2010).[42] J. Golda, B. Biskup, V. Layes, T. Winzer, andJ. Benedikt, “Vacuum ultraviolet spectroscopy ofcold atmospheric pressure plasma jets”, PlasmaProcesses and Polymers , 1–12 (2020).[43] R. R. Laher and F. R. Gilmore, “UpdatedExcitation and Ionization Cross Sections forElectron Impact on Atomic Oxygen”, Journal ofPhysical and Chemical Reference Data , 277–305 (1990).[44] N. Bibinov, H. Halfmann, P. Awakowicz, and K.Wiesemann, “Relative and absolute intensity cal-ibrations of a modern broadband echelle spec-trometer”, Measurement Science and Technology , 1327–1337 (2007).[45] M. Moravej, X. Yang, G. R. Nowling, J. P.Chang, R. F. Hicks, and S. E. Babayan,“Physics of high-pressure helium and argonradio-frequency plasmas”, Journal of AppliedPhysics , 7011–7017 (2004).[46] C. Stewig, T. Urbanietz, L. Chauvet, M.B¨oke, and A. v. Keudell, “Dedicated setup toisolate plasma catalysis mechanisms”, Journal ofPhysics D: Applied Physics54