electric field measurements in nanosecond pulse discharges ......electric ﬁeld measurements in...

Electric field measurements in nanosecondpulse discharges in air over liquid watersurface

Marien Simeni Simeni1, Edmond Baratte2, Cheng Zhang3,Kraig Frederickson1 and Igor V Adamovich1,4

1Department of Mechanical and Aerospace Engineering, Ohio State University, Columbus, OH, UnitedStates of America2 CentraleSupélec, Gif-sur-Yvette, France3 Institute of Electrical Engineering, Chinese Academy of Sciences, Beijing, People’s Republic of China

E-mail: [email protected]

Received 27 June 2017, revised 23 November 2017Accepted for publication 8 December 2017Published 10 January 2018

AbstractElectric field in nanosecond pulse discharges in ambient air is measured by picosecond four-wavemixing, with absolute calibration by a known electrostatic field. The measurements are done in twogeometries, (a) the discharge between two parallel cylinder electrodes placed inside quartz tubes, and(b) the discharge between a razor edge electrode and distilled water surface. In the first case,breakdown field exceeds DC breakdown threshold by approximately a factor of four,140±10 kV cm−1. In the second case, electric field is measured for both positive and negative pulsepolarities, with pulse durations of ∼10 ns and ∼100 ns, respectively. In the short duration, positivepolarity pulse, breakdown occurs at 85 kV cm−1, after which the electric field decreases over severalns due to charge separation in the plasma, with no field reversal detected when the applied voltage isreduced. In a long duration, negative polarity pulse, breakdown occurs at a lower electric field,30 kV cm−1, after which the field decays over several tens of ns and reverses direction when theapplied voltage is reduced at the end of the pulse. For both pulse polarities, electric field after thepulse decays on a microsecond time scale, due to residual surface charge neutralization by transportof opposite polarity charges from the plasma. Measurements 1 mm away from the discharge centerplane, ∼100μm from the water surface, show that during the voltage rise, horizontal fieldcomponent (Ex) lags in time behind the vertical component (Ey). After breakdown, Ey is reduced tonear zero and reverses direction. Further away from the water surface (≈0.9 mm), Ex is much highercompared to Ey during the entire voltage pulse. The results provide insight into air plasma kineticsand charge transport processes near plasma-liquid interface, over a wide range of time scales.

Keywords: electric field, ns pulse discharge, four-wave mixing, air plasma, water

1. Introduction

Over the last decade, numerous experimental and kineticmodeling studies have been conducted in transient high-pressure plasmas generated by ns pulse discharges, whichexhibit highly complex behavior on short time scales. Thiswork is stimulated by a wide range of potential applications,including plasma flow control, plasma-assisted combustion,

plasma-induced liquid phase chemistry, and plasma medicine.Some of the recent advances in this rapidly expanding fieldare presented in a special issue on fast pulsed dischargespublished by Plasma Sources Science and Technology (see[1] and references therein). One of the difficulties in quanti-tative understanding of kinetics of these plasmas is theiraccurate characterization, using nonintrusive diagnosticmethods with high temporal and spatial resolution. It is wellknown that rapidly varying distribution of the electric field inthese plasmas controls the electron energy distribution

Plasma Sources Science and Technology

Plasma Sources Sci. Technol. 27 (2018) 015011 (14pp) https://doi.org/10.1088/1361-6595/aaa06e

4 Author to whom any correspondence should be addressed.

0963-0252/18/015011+14$33.00 © 2018 IOP Publishing Ltd1

https://orcid.org/0000-0001-6311-3940https://orcid.org/0000-0001-6311-3940mailto:[email protected]://doi.org/10.1088/1361-6595/aaa06ehttp://crossmark.crossref.org/dialog/?doi=10.1088/1361-6595/aaa06e&domain=pdf&date_stamp=2018-01-10http://crossmark.crossref.org/dialog/?doi=10.1088/1361-6595/aaa06e&domain=pdf&date_stamp=2018-01-10

function (EEDF), as well as energy partition among internalenergy modes of molecules and atoms in electron impactprocesses. This is one of the dominant factors controlling thenumber densities of excited species and radicals, kinetics ofmolecular energy transfer, and energy thermalization rate. Atthese highly transient conditions and in the presence of strongspatial gradients (such as in plasmas sustained by streamersand ionization waves), the EEDF may also be non-local,significantly complicating plasma kinetics. Quantitativeinsight into these processes critically depends on spatio-temporally resolved measurements of the electric field.

In pulsed plasmas used for applications in biology andmedicine, such as electroporation and intracellular manipulation[2, 3], bacterial deactivation, sterilization, and wound healing[4], and cancer cell eradication [5], the electric field needs to beknown in close proximity of a dielectric surface covered byliquid (water or aqueous solution). In the present work, we useelectric field four-wave mixing, similar to coherent anti-StokesRaman spectroscopy (CARS), to measure electric field in nspulse discharges sustained in ambient air near liquid water sur-face. Recently, the four-wave mixing diagnostics, first suggestedin [6], has been used increasingly extensively for measurementsof electric fields in pulsed electric discharges in hydrogen,nitrogen, and air [7–17]. In our approach, the Stokes beam forthe four-wave mixing is generated in a high-pressure stimulatedRaman shifting (SRS) cell, collinear to the pump beam producedby a ps Nd:YAG laser. Time resolution of this method is limitedby coherence dephasing time of molecules excited by the pumpand Stokes beams, on the order of a few hundred picoseconds.Unless four-wave mixing and CARS signals are saved for everylaser shot, and subsequently post-processed, temporal resolutionis controlled by the high-voltage pulse generator jitter,approximately 2 ns in the present work. Since the pump and theStokes beam exiting the SRS cell are collinear, spatial resolutionof this method in the direction of the laser beam is limited by thesignal coherence length, several cm at the present conditions[17]. Spatial resolution in directions perpendicular to the focusedlaser beam is a few tens of μm. Absolute calibration is providedby measuring a known electrostatic electric field. The presentapproach is especially effective for non-intrusive measurementsof the electric field in pulsed discharges operating at highpressures, such as in quasi-one-dimensional volumetric plasmasbetween plane electrodes [10] and quasi-two-dimensional plas-mas in near surface dielectric barrier discharges [14].

The objective of the present work is to measure timeevolution and spatial distribution of the electric field in a nspulse discharge plasma in ambient air near the liquid watersurface, in a relatively simple quasi-two-dimensional geometry.This is critical for insight into fundamental kinetics of ioniz-ation, electron attachment, electron and ion transport, andcharge accumulation/neutralization on the liquid surface.Propagation of electric discharges over the surface of liquidwater has been studied previously, using microsecond pulsecorona/spark discharges in ambient air [18–21] and nano-second pulse surface ionization wave discharges in nitrogenand helium [22]. In both cases, charge transport and surfacecharge accumulation are among the dominant factors control-ling surface discharge self-organization [19] and ionization

wave propagation [22]. In these experiments, surface dischargepropagation speed, as well as electric potential and axialelectric field distributions over the water surface have beenmeasured using electrostatic and capacitive probes [18, 22]. In[18], it was concluded that propagation of the ionization wave(leader) and plasma formation in the wave front are affected bythe axial electric field ahead of the wave, as well as by thetransverse electric field in the front. However, non-intrusivemeasurements of electric field in these transient discharges,with high temporal and spatial resolution, are not available.

Typically, pulsed discharges in ambient air over liquidwater surface in pin-to-plane geometry develop as a manifoldof surface leaders [19], ‘sliding sparks’ [20], or filaments [21].In the present work, the use of quasi-two-dimensional geo-metry, as well as short discharge pulse duration, is intended toreduce this effect, such that the results could be used forkinetic modeling and assessment of predictive capability ofthe models, while comparison with the modeling predictionswould provide quantitative understanding of these processes.

2. Experimental

Experimental apparatus used for picosecond four-wave mixingis described in detail in our previous work [17]. Briefly, in thepresent work a picosecond Nd:YAG laser (Ekspla SL333)generating output pulses with spectral linewidth of 0.1 cm−1

and ∼150 ps long, with energy of up to 100 mJ per pulse at532 nm, is used to pump a high-pressure SRS cell in order togenerate a collinear Stokes beam at 607 nm. The collinearpump and Stokes beams are focused in the test region betweenthe discharge electrodes or between the electrostatic fieldcalibration electrodes of the same length, using a 500mm focaldistance lens. The laser beam diameter at the focal point itapproximately 50μm, such that the Rayleigh range forλ=0.6 μm is approximately 3.3 mm. After passing throughthe test region, the N2 four-wave mixing signal beam (at4.3 μm) is separated from the pump, Stokes, and anti-Stokes(N2 vibrational CARS) beams using dichroic mirrors. Inaddition, the IR four-wave mixing signal beam is spectrallyseparated from residual visible beams using a CaF2 equilateraldispersion prism, and its intensity is measured by a liquidnitrogen cooled InSb detector with matching preamplifier. Formeasurements of individual electric field vector components, apolarizer is placed in front of the IR detector. The intensity ofthe CARS beam is measured by a photodiode. IR, N2 CARS,and pump laser signals, as well as voltage and current wave-forms, are monitored and averaged by a LeCroy WaverunnerMXi-A digital oscilloscope. The absolute value of the electricfield is related to the time-integrated IR, pump, and CARSsignal intensities as follows: a=∣ ∣E I I I ,IR pump CARS where αis the calibration constant. The calibration is done by mea-suring the known electrostatic electric field between two par-allel plate copper electrodes. Separate calibrations are done forindividual components of electric field vector.

In the present experiments, a dielectric barrier dischargein room air is sustained in two different configurations,

2

Plasma Sources Sci. Technol. 27 (2018) 015011 M S Simeni et al

(i) between two parallel stainless steel cylinder electrodeswith outside diameter of 9.6 and 55 mm long, placed insidequartz tubes with inside diameter of 9.8 mm and wall thick-ness of 1.7 mm, set 1 mm apart (as shown in figure 1), and (ii)between a plane high-voltage electrode with a sharp edge (astainless steel, double-edge razor blade 37 mm long, and0.1 mm thick) and a rectangular 40 mm×75 mm groundedelectrode made of adhesive copper foil, attached to the outsidebottom surface of a quartz reservoir 3 mm deep filled withdistilled water (see figure 2). The first discharge configurationis chosen because electrostatic field between two parallelconducting cylinders covered by dielectric sleeves can beevaluated analytically, which provides additional verificationof the Laplacian electric field before breakdown. In thisgeometry, it is also straightforward to compare peak electricfield measured during ns pulse breakdown with the valuepredicted by Paschen law for DC breakdown.

In the first discharge configuration, the distance betweenthe two cylindrical electrode surfaces is 4.5 mm and the gapbetween two quartz tubes is 1 mm (see figure 1). In the secondconfiguration, the gap between the high-voltage electrode edgeand the liquid surface, 1–2mm, is set up by varying the waterlevel using a syringe partially filled with water and connected

to a small diameter plastic tube submerged into the reservoir.The water depth under the high-voltage electrode can be up to4.5 mm, which is achieved by overfilling the reservoir such thatthe water level becomes higher than the borders. This makespossible aligning the laser beam close to the water surface(within∼100 μm), without clipping on the reservoir walls. Thelaser beam is directed parallel to the cylindrical electrode axes(see figure 1) or parallel to the high-voltage electrode edge (seefigure 2).

The electrode assembly shown in figure 1 is alignedmanually, using spacers to set the distance between the quartztubes (i.e. the discharge gap). The electrode assembly shownin figure 2 is mounted on top of a 1D rotation state and a 3Dtranslation stage, allowing more precise control of the high-voltage electrode location and angle relative to the laser beam.During the experiment, the laser beam is aligned along theedge of the high-voltage electrode until scattering of the beamoff the electrode edge is no longer detected and the IR signalat peak pulse voltage is maximized, giving the reference beamposition. In both cases, the discharge is sustained by aMegaimpulse high-voltage pulse generator producing positivepolarity pulses with peak voltage of up to 26 kV and pulseduration of ∼10 ns FWHM, operated at pulse repetition rate

Figure 1. Schematic (approximately to scale) and a photograph of electrode assembly for the discharge between two parallel cylinderelectrodes placed inside quartz tubes. Electrode length L=55 mm, electrode diameter Di=9.6 mm, quartz tube outside diameterD0=13.2 mm, discharge gap d=0.9 mm.

Figure 2. (a) Schematic of electrode assembly for the discharge over liquid water surface. (b) Laser beam locations used during themeasurements are shown in the schematic. Water depth 3.0–4.5 mm, discharge gap varied from 1.0 to 2.0 mm.

3


of 10 Hz, controlled by an external delay generator. The samedelay generator also triggers the flash lamps of the laser, andthe second delay generator is used to trigger the Q-switch ofthe laser and thus control the delay time between the dis-charge pulse and the laser pulse, with the jitter of approxi-mately 2 ns. The time delay is varied from t=−15 ns (i.e.15 ns before breakdown between the discharge electrodes, att=0) to t=+25 ns. Although the Megaimpulse pulsegenerator is capable of producing pulses with peak voltage upto 40 kV, lower peak voltage was used to prevent coronaformation near high-voltage transmission line and to reduceEMI noise generated by the plasma.

In addition, the discharge near the liquid water surface isalso sustained using a custom-made pulse generator producing analternating polarity pulse train with peak voltage of up to 10 kVand significantly longer pulse duration of ∼100 ns FWHM [17],also operated at 20Hz. This discharge is used to measure electricfield during negative polarity pulses, from t=−150 ns tot=+3μs. The potentials on the high-voltage electrode and thegrounded electrode, as well as the current in the external circuit(on the ground side), are measured using Tektronix P-6015 highvoltage probe (bandwidth 75MHz) and Pearson 2877 currentprobe (bandwidth 200MHz). Plasma emission images weretaken by a Princeton Instruments PI-Max 3 ICCD camera.

3. Results and discussion

3.1. Discharge in air between two cylinder electrodes

Figure 3 plots pulse voltage waveforms (potentials on thehigh voltage electrode, ground electrode, and their difference)measured in the discharge between two parallel cylinderelectrodes placed inside quartz tubes placed 0.9 mm apart, asshown in figure 1. These waveforms are generated by theMegaimpulse pulse generator. From figure 3, it can be seen

that although the high-voltage electrode potential peaks at≈18 kV, peak voltage between the electrodes is significantlyhigher, ≈22 kV. The rate of voltage rise before breakdown,estimated as peak linear slope of the voltage waveform, is≈3 kV ns−1, with voltage rise time of ≈7 ns. From themaximum of the time-integrated four-wave mixing signal,discussed below, it is apparent that breakdown at these con-ditions occurs approximately 2 ns before the voltage peak, att≈0. This is consistent with plasma emission images takenwith 1 ns camera gate, shown in figure 4. It can be seen thatbreakdown occurs in the 1 mm gap between the quartz tubesas well as in narrow annular gaps between electrode surfacesand inner surfaces of quartz tubes. After breakdown, when theapplied voltage peaks and begins to decrease, the plasmaremains diffuse, without well-pronounced filaments, exceptfor the brighter spots near the ends of cylindrical electrodes(see figure 4). It is clear, however, that plasma emissionintensity distribution is not uniform. The images in figure 4demonstrate that before breakdown (at t< 0), the electric fieldbetween the electrodes, E(x), can be assumed to be electro-static (i.e. unperturbed by the plasma), and can be calculatedfrom the following analytic expression,

e

»

+ + + + +

⋅ ⋅-

++

e e eD D

+D+ +D

⎛⎝⎜

⎞⎠⎟

( ) ( ) ( )( )

( )( )

E xU

x W x W x

ln 1 ln 1 ln 1

1 1 1,

1

R R d

d

R

2 2 2

1 1 0

where x is the coordinate along the line connecting the cylinderaxes (x=0 corresponds to the midpoint of the gap), U is thepotential difference between the electrodes, R=4.8 mm is theradius of the electrodes, Δ=1.7 mm is the wall thickness ofquartz tubes, d=0.9 mm is the gap between the quartz tubes,W=R+Δ+d/2=6.95mm is half the distance betweenthe electrode centers, ε0=1, and ε1=3.8 is the dielectricconstant of quartz (fused silica). Equation (1) is obtained fromthe electric field generated by a single conducting cylindercovered by a dielectric layer, using a superposition principle.Mutual influence of two cylinders may distort the surfacecharge distribution, which would affect the net field distribu-tion. However, at the present conditions (effective gap distance,d*≈d+2Δ/ε1=1.8 mm, much smaller compared to thecylinder diameter of 9.6 mm), this effect on the field on the axisof symmetry is small. This has been verified by comparing theelectric field halfway between two metal cylinders (withoutdielectrics) calculated for the same cylinders diameter andeffective gap, (a) from the superposition principle and (b) fromexact solution of the Laplace equation [23]. The differencebetween the two results is approximately 2%. The uncertaintyof the electric field predicted by equation (1), ±10%, isdetermined primarily by the uncertainty of the discharge gapmeasurement. The Laplacian electric field halfway between thecylinder electrodes (at x=0), predicted by equation (1) basedon the potential difference between the electrodes plotted infigure 3, peaks at E=120±12 kV cm−1, at t≈2 ns.

Figure 3. Pulse voltage waveforms (potentials on the high-voltageand grounded electrodes, and their difference) in the dischargebetween two parallel cylinder electrodes placed inside quartz tubesshown in figure 1.

4


Figure 5 plots IR signal waveforms for several values ofelectrostatic electric field between two parallel plate electro-des of the same length (55 mm) used for calibration, setd=5 mm apart (a), and IR waveforms taken at differentmoments of time between the cylindrical electrodes beforebreakdown, for the discharge gap of d=0.9 mm (b). The‘near-DC’ electrostatic field used for calibration was gener-ated using the custom-made high-voltage pulse generator,with pulse duration of ∼100 ns. All waveforms in figure 5 areaveraged over 300 laser shots. As discussed in our previouswork [17], the time duration of the IR waveforms,∼10–40 μs, is controlled by the time constant of the InSb IR

detector preamplifier, and is much longer than the actualduration of the four-wave mixing signal, controlled by thecoherence decay time of N2 molecules. For calibration, the IRwaveforms are integrated over t=0–40 μs to obtain theintegrated IR intensity. The maximum calibration electricfield is limited by ‘near-DC’ breakdown between the edges ofparallel plate electrodes, which occurs approximately at25 kV cm−1. The lower bound sensitivity limit of electric fieldmeasurements at the present conditions is about 3–4 kV cm−1,limited by the background noise of the IR detector.

From figure 5, it is readily apparent that the electric fieldbetween the discharge electrodes is significantly higher

Figure 4. Discharge between two parallel cylinders (see schematic in figure 1) at the conditions of figure 3: (a) photograph of plasmaemission, (b) single-shot broadband emission image (camera gate 20 ns), and (c) collage of 100-pulse accumulation emission images (cameragate 1 ns, with time stamps indicated).

Figure 5. (a) IR signal waveforms for several values of electrostatic electric field between two parallel plate calibration electrodes separatedby 5 mm, (b) IR signal waveforms at different time moments between the parallel cylinders (before breakdown), for discharge gap ofd=0.9 mm. Dashed line in figure 5(b) indicates the signal taken between the two parallel plate calibration electrodes for the electric field of19 kV cm−1 (also plotted in figure 5(a)).

5


compared to the electrostatic field between the parallel platecalibration electrodes. Therefore extrapolation of the calibra-tion line obtained from the electrostatic field measurementstoward significantly higher electric fields, illustrated in figure 6,may result in considerable uncertainty. Peak electric field in thedischarge, inferred using the calibration line plotted in figure 6,is 140±10 kV cm−1 at t= 0 ns, with the uncertainty deter-mined by the calibration line slope uncertainty, ±7%. This ismuch higher compared to the electric field measured by nsfour-wave mixing in a double dielectric barrier ns pulse dis-charge between two parallel plate electrodes in ambient air atsimilar conditions, ≈20 kV cm−1 [10]. This is most likely dueto the slower rate of voltage rise in [10], ≈1 kV ns−1, whichreduced breakdown field, and longer laser pulse duration,3–5 ns, which limited the time resolution of the measurements.

Time-resolved electric field inferred from the four-wavemixing measurements using the calibration line plotted infigure 6 is plotted in figure 7. As expected, breakdown field atthe present conditions considerably exceeds DC breakdownthreshold in air,≈32 kV cm−1 for a 1 cm gap [24], due to rapidvoltage increase rate, ≈3 kV ns−1. Figure 7 shows that electricfield in the plasma after breakdown is reduced rapidly, to≈20 kV cm−1 over ≈5 ns, while plasma emission decays.After this, the electric field remains significantly belowbreakdown level during the rest of the voltage pulse, althoughwell above the sensitivity limit of the diagnostics, until thevoltage is reduced to zero at t=25 ns (see figure 3). Detailedelectric field measurements at longer time delays after break-down have not been made at these conditions. No evidence ofelectric field reversal during the voltage reduction, observed inprevious electric field measurements in a double dielectricbarrier ns pulse discharge in ambient air [10], is detected.Electric field reversal, as well as residual electric field after thedischarge pulse, is caused by charge accumulation on the di-electric surfaces. However, this effect can be reduced

significantly by surface charge neutralization by transport ofopposite polarity charges from the plasma [13]. Absence of thiseffect at the present conditions may be due to shorter voltagerise time (∼5 ns versus ∼10 ns), significantly higher peakelectric field (140 kV cm−1 versus ∼20 kV cm−1), and conse-quently higher electron density in the plasma compared to thatin [10]. This enhances electron transport to the surface and thusreduces positive surface charge accumulation, such that fieldreversal does not occur. Long duration discharge pulses in air,on the other hand, result in more significant electron decay bythree-body attachment to oxygen, with a characteristic time of∼10 ns at atmospheric pressure [24], which inhibits electrontransport and thus leads to field reversal.

To provide additional verification of the absolute electricfield inferred from the four-wave mixing data, it is compared tothe Laplacian field predicted by equation (1) before break-down, when the electric field remains electrostatic, for the sameelectrode gap of d=0.9 mm. Peak electric field predicted byequation (1), based on the voltage difference between theelectrodes plotted in figure 3, is E=120±12 kV cm−1 att=2 ns (E=110±11 kV cm−1 at t=0). This is sig-nificantly lower than the peak field obtained by the extra-polation of the calibration line, E=140±10 kV cm−1 att=0 ns (see figure 6). The reason for this difference is notfully understood. The most likely explanation is the error thatmay be caused by the extrapolation of the calibration line overnearly a factor of 7 (see figure 6). It is also possible that theLaplacian electric field inferred from the voltage waveformsmeasured by the Tektronix high-voltage probe may be affectedby the limited bandwidth of the probe (75MHz), in which casethe pulse shape during the voltage rise before the breakdownwould not be fully resolved. However, comparison of pulsevoltage waveforms between two parallel cylinder electrodes atthe conditions of figures 3 and 7, measured by the Tektronixprobe and by a custom-made high-bandwidth (estimatedbandwidth ∼1 GHz) voltage probe [25], showed that bothpulse peak voltage and FWHM agree within several per cent,i.e. the Tektronix probe remains quite accurate even for pulse

Figure 6. Extrapolation of the calibration line obtained from theelectrostatic field measurements between two parallel plate electro-des at the conditions of figure 5(a) (circles) toward higher electricfields observed between parallel cylinder discharge electrodes at theconditions of figure 5(b) (squares).

Figure 7. Voltage between the electrodes and electric field measuredin the discharge between two parallel cylinder electrodes, for thedischarge gap of d=0.9 mm, put on the absolute scale usingcalibration line shown in figure 6.

6


durations comparable with the nominal bandwidth, 1/75MHz≈13 ns. Finally, high-voltage pulse generator jitterrelative to the laser pulse, ≈2 ns, may also result in the dist-ortion of the voltage pulse used to calculate the Laplacian fieldat t22 ns is notvery well pronounced is the strong filamentary structure of thedischarge, such that the surface charge accumulation occursmainly in the filaments (visible in figure 8), and the plasmabetween the filaments is not fully shielded. Since the presentmethod measures the electric field value averaged along the

Figure 8. Broadband plasma emission images of the positive polarityns pulse discharge sustained above a distilled water surface (seefigure 2, front view): (a) single shot, camera gate 20 ns; (b) 100-pulse accumulations, camera gate 1 ns. Discharge gap 1.5 mm, waterdepth approximately 4 mm.

Figure 9. Electric field measured on the center plane of the positivepolarity ns pulse discharge sustained above a distilled water surfaceat the conditions of figure 8, ∼100 μm from the high voltageelectrode edge, plotted together with the voltage on the high-voltageelectrode.

7


high-voltage electrode, this effect results in an apparentelectric field offset, such as observed in figure 9. Finally,several data points taken at longer delays after the dischargepulse show that electric field decays on microsecond timescale.

In the remainder of this work, electric field is measured ina negative polarity ns pulse discharge sustained by the cus-tom-made pulse generator used in our previous paper [17]. Inthese measurements, only negative polarity discharge pulsesare used for electric field measurements since they generatemuch more diffuse plasma compared to positive polaritypulses, as illustrated in figure 10. In the images in figure 10,similar to the ones in figure 8, the discharge gap 0.9 mm, andthe water depth approximately 3 mm. It can be seen that,although multiple streamers are detected for both pulsepolarities, the positive polarity discharge generates an array ofwell-defined, localized discharge filaments (comparefigures 10(a) and (c)). This is consistent with the morphologyof the discharge sustained by the same pulse generator over aquartz plate [17].

Figure 11 shows a collage of additional broadbandplasma emission images at the conditions of figure 10, takenwith a shorter camera gate of 1 ns and accumulated over 50pulses. From these images, it is apparent that the dischargefirst propagates down toward the water surface, and thenspreads to both sides as a nearly symmetric surface ionizationwave with a well-defined luminous front, creating a char-acteristic ‘Eiffel tower’ emission pattern (as indicated byarrows in in figure 11(b)). In addition, diffuse corona dis-charge emission is observed between the high-voltage elec-trode edge and the water surface. Finally, it can be seen that afew streamers are reproducible pulse-to-pulse, contributing toplasma non-uniformity along the high-voltage electrode.Again, the electric field measured at these conditions repre-sents the value averaged along the high-voltage electrode,

with the four-wave mixing signal distribution along the line ofsight in the discharge being approximately uniform [17].

Figure 12 plots time-resolved electric field measured onthe center plane of the negative polarity ns pulse dischargesustained above a distilled water surface at the conditions offigures 10–11, plotted together with the voltage and currentpulse waveforms. An initial offset electric field, approxi-mately 5 kV cm−1, is detected before the discharge pulse, dueto the positive polarity charge accumulation on the watersurface during the lower peak voltage, positive polarity ‘pre-pulse’ generated approximately 6 μs before the main dis-charge pulse. The electric field increases following theapplied voltage, until breakdown occurs between the elec-trodes (at t≈0, Ubr≈5 kV, and Ebr≈28 kV cm

−1), pro-ducing a ‘kink’ in the applied voltage and resulting in currentrise (see figure 12). The peak field during breakdown onset isconsistent with the value measured in the positive polaritydischarge sustained by the Megaimpulse generator, Ebr≈85 kV cm−1 at Ubr≈16 kV (see figure 9). After breakdown,the electric field decays over ≈70 ns, due to charge separationin the plasma and negative polarity charge accumulation onthe dielectric plate (see figure 12). At t≈70 ns, the electricfield is reduced to near zero, indicating nearly completeplasma self-shielding. After this, the electric field reversesdirection and starts increasing again, as the applied voltage isreduced, peaking at ≈22 kV cm−1 at t≈170 ns (rememberthat the present technique measures the absolute value of thefield). As discussed above, field reversal during the voltagereduction occurs because after breakdown, the field in theplasma is controlled mainly by residual surface chargesaccumulated on the dielectric surfaces, in the directionopposite to the applied field to cancel it. Therefore reducingthe applied voltage results in the net field pointing in theopposite direction, and thus increases the absolute value ofthe field in the plasma, as evident in figure 12. A transientsecondary minimum in the field detected near t≈200 nsindicates a ‘reverse’ breakdown in several isolated filaments,observed in plasma images at this moment. After peaking at≈23 kV cm−1 at t≈300 ns, when the applied voltage isreduced to near zero, the electric field gradually decays to≈8 kV cm−1 at t=3 μs (see figure 12), indicating surfacecharge neutralization by transport of opposite polarity chargesfrom the plasma to the liquid surface. These results are qua-litatively similar to the electric field in the negative polarity nspulse discharge sustained over a quartz plate, using similarelectrode geometry and the same pulse generator [17].

Comparing measurement results in positive and negativepolarity discharges over liquid water (figures 9 and 12),several differences can be pointed out. First, peak electricfield in the positive polarity discharge exceeds that in thenegative polarity discharge by approximately a factor of 3.This is consistent with the difference in voltage at whichbreakdown occurs, ≈16 kV for the positive polarity pulseversus ≈5 kV for the negative polarity pulse. This suggeststhat the measured value of the field during breakdown is notaffected by plasma morphology after breakdown. The dif-ference in breakdown voltage and peak breakdown field isalmost certainly due to considerably shorter positive polarity

Figure 10. Plasma emission images of the negative polarity ns pulsedischarge above a distilled water surface (front view): (a) single-shot, negative polarity, (b) 100-pulse accumulations, negativepolarity, (c) single-shot, positive polarity, (d) 100-pulse accumula-tions, positive polarity. Discharge gap 0.9 mm, water depthapproximately 3 mm, camera gate 100 ns.

8


pulse voltage rise (≈5 ns for the Megaimpulse generatorversus ≈50 ns for the custom-designed pulse generator),which results in significant overvoltage before breakdown isinitiated. Indeed, breakdown voltages for positive and nega-tive polarity pulses produced by the custom-designed pulserare very close.

Second, well-defined electric field reversal during appliedvoltage reduction has been detected only during the negativepolarity pulses. As discussed above, field reversal occurs dueto surface charge accumulated on the dielectric (i.e. water)surface during the voltage rise, and is significantly reduced bytransport of the opposite polarity charges from the plasma.Therefore positive surface charge (accumulated on the watersurface during the voltage rise in the positive polarity pulse) isneutralized rapidly during the pulse, predominantly bytransport of electrons from the plasma, while negative surfacecharge (accumulated during the voltage rise in the negativepolarity pulse) is neutralized by much slower positive iontransport. Note that field reversal in the negative polarity

pulse discharge (at the conditions of figure 12) is unlikely tobe affected by electron attachment, since surface chargeneutralization is controlled by transport of positive ions to thedielectric surface. On the other hand, the absence of fieldreversal in the positive polarity discharge (at the conditions offigure 9) is apparent on the time scale comparable to that ofthree-body electron attachment to O2 (t∼10 ns in ambientair), such that surface charge neutralization is still controlledby transport of electrons to the surface.

Finally, residual electric field after the discharge pulsedecays on microsecond time scale, both for negative polarity(see figure 12) and for positive polarity (data not shown). Thisindicates that on this time scale, after most electrons decay toform negative ions, charge transport in the plasma is domi-nated by the positive ions and negative ions, respectively.Indeed, if transport in the positive polarity discharge plasmaon this time scale were dominated by electrons, residual fieldwould decay much faster, due to significant disparity betweenelectron mobility and positive ion mobility. Quantitative

Figure 11. Collage of plasma emission images at the conditions of figure 10, (a) front view, (b) side view. Camera gate 1 ns, 50-pulseaccumulation, time stamps are indicated in the images. Arrows indicate two apparent surface ionization wave fronts.

9


insight into the effect of charge transport and surface chargeneutralization on the electric field in the plasma can beobtained from detailed kinetic modeling including kinetics ofionization, electron attachment, electron-ion recombination,and ion-ion neutralization.

Additional electric field measurements have been done atthe same discharge gap of 1.0 mm but with water depth in thereservoir increased to 4.5 mm, by partially overfilling thereservoir to provide laser beam access to close proximity ofthe water surface, as discussed in section 2. Figure 13 showsplasma emission images of the negative polarity ns pulsedischarge at these conditions. These images are similar to theone shown in figures 10, 11 (including the ‘Eiffel tower’ sideview emission intensity distribution), with the exception ofthe corona discharge emission surrounding the high voltageelectrode (see figure 11), which was not detected at theseconditions. Time-resolved electric field was measured at twolocations on the discharge center plane, (1) ∼100 μm belowthe high voltage electrode and (2) ∼100 μm above the watersurface, as well as at two locations 1 mm away from thecenter plane, (3) ∼100 μm above the water surface and (4)∼100 μm below the high voltage electrode (≈1 mm above thewater surface), as shown schematically in figure 2(b). Toestimate the distance from the laser beam to the water surface,the transmitted beam at a lower laser power was reflected offa paper card, which also reflected light scattered off the sur-face. For measurements at locations (2) and (3), the laserbeam was set up 100 μm above the location when the entirebeam was scattered off the surface. At this location, no laserbeam scattering was detected on the card. At locations (3) and(4), both vertical and horizontal electric field components, Exand Ey, have been measured, using a polarizer placed in frontof the IR detector and separate calibration data sets, as dis-cussed in section 2.

The measurement results at location #1 (near the high-voltage electrode) and location #2 (near the water surface) aresimilar to the ones obtained for water depth of 3 mm andplotted in figure 12, including initial field offset (≈5 kV cm−1),peak field during breakdown (Ebr≈30 kV cm

−1), and residualfield decay time after the discharge pulse (≈2 μs). The onlysignificant difference is that in the measurements with waterdepth of 4.5 mm, electric field reduction to near zero was notdetected, most likely due to slight misalignment between thehigh-voltage electrode and the laser beam, such that the line-of-sight averaged horizontal electric field component was notzero [17].

Time-resolved horizontal and vertical electric fieldcomponents measured at location #3 (1 mm off the centerplate, 100 μm above water surface), plotted in figure 14,indicate that during the voltage rise, the vertical field exceedsthe horizontal field, peaking during breakdown at t≈0, suchthat Ex lags in time behind Ey by about 40 ns. After break-down, when both field components decay, the vertical field isreduced to near zero and reverses direction, indicating sig-nificant surface charge accumulation, while the horizontalfield direction remains the same, suggesting that residualsurface charge density on the centerline is higher. Note thatconclusive detection of electric field reversal requires obser-ving electric field reduction to near zero, which is limited bythe sensitivity limit of the diagnostics. Finally, measurementresults at location #4 (1 mm off the center plate and∼100 μm below high-voltage electrode, i.e. ≈0.9 mm abovewater surface), shown in figure 15, indicate that, contrary tothe measurements near the surface, the horizontal electric fielddominates the vertical field during the entire voltage pulse.Clearly, the proximity of the water surface in the surfaceionization wave discharge considerably enhances the verticalelectric field.

Figure 12. Electric field measured on the center plane of the negative polarity ns pulse discharge sustained above a distilled water surface(laser beam position #1 in figure 2(b)) at the conditions of figures 10–11, plotted together with the voltage and current pulse waveforms.Discharge gap 1 mm, water depth approximately 3 mm.

10


4. Summary

This work presents the results of electric field measurementsin nanosecond pulse discharges in ambient air by picosecondfour-wave mixing in a collinear geometry, using a ps Nd:YAG laser pump beam and a high-pressure Raman cell togenerate the Stokes beam. The electric field is determined bymeasuring the intensities of the pump, anti-Stokes, and IR(four-wave mixing) signal beams generated in the plasma.

Individual components of the electric field vector are mea-sured using a polarizer placed in front of the IR detector.Absolute calibration of the electric field, up to 20 kV cm−1, isdone by measuring a known electrostatic electric fieldbetween two parallel plate electrodes. Time resolution ofthese measurements, approximately 2 ns, is controlled by thejitter between the discharge pulse and the laser pulse. Spatialresolution, several cm at the present conditions, is limited bythe four-wave coherence length.

Figure 13. Plasma emission images of the negative polarity ns pulse discharge above a distilled water surface, at a greater water depth of4.5 mm and discharge gap of 1.0 mm: (a) front view, single-shot, 100 ns gate, (b) front view, 100-pulse accumulations, 100 ns gate, (c) topview, 100-pulse accumulation, 3 ns gate, (d) side view, 10-pulse accumulation, 3 ns gate. Time stamps are shown in the images.

Figure 14. Electric field vector components in the negative polarity ns pulse discharge sustained above a distilled water surface at theconditions of figure 13 (water depth 4.5 mm), ∼100 μm above water surface and 1 mm away from the center plane (laser beam position #3in figure 2(b)). Discharge gap 1.0 mm.

11


Time-resolved electric field is measured in two config-urations, (a) quasi-two-dimensional discharge sustainedbetween two parallel cylinder electrodes covered by dielectrictubes, and (b) discharge sustained between a razor edge high-voltage electrode and distilled water surface. The parallelcylinder geometry, for which electrostatic electric field can bedetermined analytically, is used for additional electric fieldverification until the breakdown moment. The plasma gen-erated in this configuration is diffuse, with no sign of indi-vidual streamers, which greatly reduces the uncertainty inlocal electric field measurements due to relatively low spatialresolution of the present method. The results show thatnanosecond pulse breakdown in this discharge occurs at ahigh peak electric field, 140±10 kV cm−1, which exceedsDC breakdown field by approximately a factor of 4. Afterbreakdown, the field in the plasma decays to 20 kV cm−1 over5 ns, but remains above the detection limit during theremainder of the discharge pulse.

In the discharge above the water surface, electric field ismeasured both for positive and negative voltage polarities,produced by two different pulse generators. In this case, thedischarge exhibits multiple streamers, in addition to diffuseplasma, such that the electric field represents a line-of-sightaveraged value along the discharge. The positive polaritypulse duration, ∼20 ns, is much shorter compared to that ofthe negative polarity pulse, ∼100 ns, resulting in a muchfaster voltage rise rate in the positive polarity discharge, up to≈3 kV ns−1, compared to ∼0.1 kV cm−1 in the negativepolarity discharge. For positive pulse polarity, the results ofelectric field measurements ∼100 μm from the edge high-

voltage electrode are qualitatively similar to those in thedischarge between two parallel electrodes. No initial offsetelectric field is detected before the discharge pulse, indicatingno residual charge accumulation on the water surface.Breakdown occurs slightly before the voltage peaks at≈16 kV, at the electric field of ≈85 kV cm−1. After break-down, the electric field decreases over ≈7 ns, due to chargeseparation in the plasma. Again, the minimum electric fieldduring the discharge pulse, approximately 8 kV cm−1, isabove the sensitivity limit of the present diagnostics. Noelectric field reversal such as observed in our previous work[17] is detected, except at the very end of the voltage pulse.

For negative pulse polarity, the results are similar to ourprevious measurements in the discharge above a quartz sur-face, for the same electrode geometry [17]. Electric fieldmeasured ∼100 μm from the edge high-voltage electrodefollows the applied voltage as it increases, until breakdownoccurs, after which the electric field in the plasma is reducedto near zero due to charge separation. Peak electric fieldmeasured during breakdown, ≈30 kV cm−1, is approximatelya factor of 3 lower compared to that in the positive polaritydischarge, which is consistent with the ratio of breakdownvoltages, 5 kV versus 16 kV. When the applied voltage isreduced, the electric field in the plasma clearly reversesdirection and increases again, until much weaker ‘reverse’breakdown occurs in several isolated discharge filaments,producing a secondary transient reduction in the electric field.After the pulse, the electric field is gradually reduced on amicrosecond time scale, due to residual surface charge neu-tralization by transport of opposite polarity charges from the

Figure 15. Electric field vector components in the negative polarity ns pulse discharge sustained above a distilled water surface at theconditions of figure 13 (water depth 4.5 mm), ∼100 μm below high-voltage electrode and and 1 mm away from the center plane (laser beamposition #4 in figure 2(b)). Discharge gap 1.0 mm.

12


plasma. Additional measurements are made at four differentlocations in the discharge, including two locations 100 μmfrom the water surface. Horizontal and vertical electric fieldcomponents measured near the surface 1 mm off the centerplane show that during the voltage rise, Ex lags in time behindEy, which peaks at breakdown moment ≈40 ns earlier. Afterbreakdown, when both field components decay, the verticalfield is reduced near zero and reverses direction. Away fromthe water surface (1 mm off the center plane, 1 mm above thesurface), Ex remains much higher than Ey during the entirevoltage pulse.

The present work shows significant qualitative similaritywith the results obtained in a ns pulse surface discharge over asolid dielectric surface (quartz) [17], including the breakdownfield value (∼30 kV cm−1), time scale for electric fieldreduction after breakdown (several tens of ns), evidence ofreverse breakdown, and time scale for electric field decayafter the discharge pulse (several μs). On the other hand, themaximum distance of ionization wave propagation over thewater surface is much shorter compared to surface ionizationwave over quartz (∼1 mm versus ∼5 mm), which is likelydue to more rapid electron attachment near the water surface.The results demonstrate considerable potential of the presenttechnique for electric field measurements in transient dis-charges in ambient air near liquid surfaces, within ≈100 μmfrom the surface and potentially with sub-nanosecond timeresolution, providing quantitative insight into charge transportand plasma kinetics near plasma-liquid interface. This diag-nostics is readily amenable to use for characterization of airplasmas near liquid surfaces in different geometries, forapplications in biology and medicine. Kinetic modeling ofpositive and negative polarity ns pulse discharges in air nearwater surface are expected to yield additional quantitativeinsight into plasma kinetics and transport, surface chargeaccumulation and decay, and spatial and temporal distributionof the electric field vector. The present experimental resultscan be used for detailed validation of the modelingpredictions.

Acknowledgments

The support of US Department of Energy Plasma ScienceCenter ‘Predictive Control of Plasma Kinetics: Multi-Phaseand Bounded Systems’, and National Science Foundationgrant ‘Nanosecond Pulse Discharges at a Liquid-VaporInterface and in Liquids: Discharge Dynamics and PlasmaChemistry’, is gratefully acknowledged. We would also likeDr David Burnette from Ohio University for the extendedloan of the Megaimpulse pulse generator, and Altos Photo-nics for their extensive and generous help with ps lasertune-up.

ORCID iDs

Igor V Adamovich https://orcid.org/0000-0001-6311-3940

References

[1] Brandenburg R, Bruggeman P J and Starikovskaia S M 2017 Fastpulsed discharges Plasma Sources Sci. Technol. 26 020201

[2] Joshi R P and Schoenbach K H 2010 Bioelectric effects ofintense ultrashort pulses Crit. Rev. Biomed. Eng. 38 255

[3] Batista Napotnik T, Reberšek M, Vernier P T, Mali B andMiklavčič D 2016 Effects of high voltage nanosecondelectric pulses on eukaryotic cells (in vitro): a systematicreview Bioelectrochemistry 110 1

[4] von Woedtke T, Reuter S, Masura K and Weltmann K-D 2013Plasmas for medicine Phys. Rep. 530 291

[5] Keidar M 2015 Plasma for cancer treatment Plasma SourcesSci. Technol. 24 033001

[6] Gavrilenko V P, Kupriyanova E B, Okolokulak D P,Ochkin V N, Savinov S Y, Tskhai S N and Yarashev A N1992 Generation of coherent IR Light on a dipole-forbiddenmolecular transition with biharmonic pumping in a staticelectric field JETP Lett. 56 1

[7] Ito T, Kobayashi K, Müller S, Luggenhölscher D,Czarnetzki U and Hamaguchi S 2009 Electric fieldmeasurement in an atmospheric or higher pressure gas bycoherent Raman scattering of nitrogen J. Phys. D: Appl.Phys. 42 092003

[8] Ito T, Kazunobu K, Czarnetzki U and Hamaguchi S 2010Rapid formation of electric field profiles in repetitivelypulsed high-voltage high-pressure nanosecond dischargesJ. Phys. D: Appl. Phys. 43 062001

[9] Müller S, Ito T, Kobayashi K, Luggenhölscher D,Czarnetzki U and Hamaguchi S 2010 Electric fieldmeasurements in near-atmospheric pressure nitrogen and airbased on a four-wave mixing scheme J. Phys.: Conf. Ser.227 012040

[10] Ito T, Kanazawa T and Hamaguchi S 2011 Rapid breakdownmechanisms of open air nanosecond dielectric barrierdischarges Phys. Rev. Lett. 107 065002

[11] Lempert W, Kearney S and Barnat E 2011 Diagnostic study offour wave mixing based electric field measurements in highpressure nitrogen plasmas Appl. Opt. 50 5688

[12] Yatom S, Tskhai S and Krasik Y 2013 Electric field in aplasma channel in a high-pressure nanosecond discharge inhydrogen: a coherent anti-stokes Raman scattering studyPhys. Rev. Lett. 111 255001

[13] Goldberg B M, Shkurenkov I, O’Byrne S, Adamovich I V andLempert W R 2015 Electric field measurements in a dielectricbarrier nanosecond pulse discharge with sub-nanosecond timeresolution Plasma Sources Sci. Technol. 24 035010

[14] Goldberg B, Böhm P, Czarnetzki U, Adamovich I andLempert W 2015 Electric field vector measurements in asurface ionization wave discharge Plasma Sources Sci.Technol. 24 055017

[15] Goldberg B, Shkurenkov I, Adamovich I V and Lempert W R2016 Electric field vector measurements in an AC dielectricbarrier discharge overlapped with a nanosecond pulsedischarge Plasma Sources Sci. Technol. 25045008

[16] Böhm P, Kettlitz M, Brandenburg R, Höft H and Czarnetzki U2016 Determination of the electric field strength offilamentary DBDs by CARS-based four-wave mixingPlasma Sources Sci. Technol. 25 054002

[17] Simeni Simeni M, Goldberg B M, Zhang C, Frederickson K,Lempert W R and Adamovich I V 2017 Electric fieldmeasurements in a nanosecond pulse discharge inatmospheric air J. Phys. D: Appl. Phys. 50184002

[18] Belosheev V P 1998 Study of the leader of a spark dischargeover a water surface Tech. Phys. 43 783–9

13


https://orcid.org/0000-0001-6311-3940https://orcid.org/0000-0001-6311-3940https://orcid.org/0000-0001-6311-3940https://orcid.org/0000-0001-6311-3940https://doi.org/10.1088/1361-6595/aa5205https://doi.org/10.1615/CritRevBiomedEng.v38.i3.20https://doi.org/10.1016/j.bioelechem.2016.02.011https://doi.org/10.1016/j.physrep.2013.05.005https://doi.org/10.1088/0963-0252/24/3/033001https://doi.org/10.1088/0022-3727/42/9/092003https://doi.org/10.1088/0022-3727/43/6/062001https://doi.org/10.1088/1742-6596/227/1/012040https://doi.org/10.1103/PhysRevLett.107.065002https://doi.org/10.1364/AO.50.005688https://doi.org/10.1103/PhysRevLett.111.255001https://doi.org/10.1088/0963-0252/24/3/035010https://doi.org/10.1088/0963-0252/24/5/055017https://doi.org/10.1088/0963-0252/25/4/045008https://doi.org/10.1088/0963-0252/25/4/045008https://doi.org/10.1088/0963-0252/25/5/054002https://doi.org/10.1088/1361-6463/aa6668https://doi.org/10.1088/1361-6463/aa6668https://doi.org/10.1134/1.1259074https://doi.org/10.1134/1.1259074https://doi.org/10.1134/1.1259074

[19] Belosheev V P 2000 Discharge leader self-organization on thewater surface Tech. Phys. 45 922–7

[20] Anpilov A M, Barkhudarov E M, Kop’ev V A, Kossyi I A andSilakov V P 2006 Atmospheric electric discharge into waterPlasma Phys. Rep. 32 968–72

[21] Lukes P, Clupek M and Babicky V 2011 Discharge filamentarypatterns produced by pulsed corona discharge at theinterface between a water surface and air IEEE Trans.Plasma Sci. 39 2644–5

[22] Petrishchev V, Leonov S and Adamovich I V 2014 Studies ofnanosecond pulse surface ionization wave discharges over

solid and liquid dielectric surfaces Plasma Sources Sci.Technol. 23 065022

[23] Engelen K, Jacqmaera P and Driesen J 2013 Electric and magneticfields of two infinitely long parallel cylindrical conductorscarrying a DC current Eur. Phys. J. Appl. Phys. 6424515

[24] Raizer Y P 1991 Gas Discharge Physics (Berlin: Springer)[25] Takashima K, Yin Z and Adamovich I V 2013 Measurements

and kinetic modeling of energy coupling in volume andsurface nanosecond pulse discharges Plasma Sources Sci.Technol. 22 015013

14


https://doi.org/10.1134/1.1259749https://doi.org/10.1134/1.1259749https://doi.org/10.1134/1.1259749https://doi.org/10.1134/S1063780X06110122https://doi.org/10.1134/S1063780X06110122https://doi.org/10.1134/S1063780X06110122https://doi.org/10.1109/TPS.2011.2158611https://doi.org/10.1109/TPS.2011.2158611https://doi.org/10.1109/TPS.2011.2158611https://doi.org/10.1088/0963-0252/23/6/065022https://doi.org/10.1051/epjap/2013120441https://doi.org/10.1051/epjap/2013120441https://doi.org/10.1088/0963-0252/22/1/015013

1. Introduction2. Experimental3. Results and discussion3.1. Discharge in air between two cylinder electrodes3.2. Discharge over liquid water

4. SummaryAcknowledgmentsReferences

electric field measurements in nanosecond pulse discharges ......electric ﬁeld measurements in...

Documents