Plasma Physics and Controlled Fusion

Plasma Phys. Control. Fusion 57 (2015) 014001 (8pp) doi:10.1088/0741-3335/57/1/014001

Challenges in understanding andpredictive model development ofplasma-assisted combustionIgor V Adamovich and Walter R Lempert

Nonequilibrium Thermodynamics Laboratory, Department of Mechanical and Aerospace Engineering,The Ohio State University, Columbus, OH 43210, USA

E-mail: adamovich.1@osu.edu

Received 20 June 2014, revised 22 July 2014Accepted for publication 28 July 2014Published 28 November 2014

AbstractThe key challenges to quantitative insight into fuel–air plasma kinetics, as well asplasma-assisted ignition and flameholding, are identified and assessed based on the results ofrecent experimental and kinetic modeling studies. Experimental and modeling approaches toaddress these critical issues are discussed. The results have major implications for thefundamental understanding of pulsed electric discharge dynamics, molecular energy transfer inreacting flows, plasma chemical reactions, and development of low-temperatureplasma-assisted combustion technologies.

Keywords: plasma-assisted combustion, optical diagnostics, kinetic modeling

(Some figures may appear in colour only in the online journal)

1. Introduction

Over the last 10–15 years, there has been significant progressin demonstrating the utility of non-equilibrium plasmas foraugmentation of combustion phenomena, such as reductionof ignition delay time and ignition temperature, as well asincrease in flame stability and flammability limits [1, 2]. Highpeak voltage, nanosecond pulse duration discharges are ofparticular interest for plasma-assisted combustion, since theycan generate diffuse non-equilibrium plasmas at high pressures(∼1 bar) and high pulse repetition rates (up to ∼100 kHz), andare characterized by high peak reduced electric fields, E/N ,of several hundred Townsends (1 Td = 10−17 V cm2). Atthese high E/N values, a significant fraction of dischargeinput energy goes into the population of excited states ofmolecules (vibrational and electronic), as well as moleculardissociation and ionization by electron impact. Collisionalquenching of the excited states (including reactive quenching),and reactions of radical species generated in the dischargeconsiderably expand the variety of chemical reactions in low-temperature fuel–air mixtures, resulting in fuel oxidation andignition. From a fundamental kinetics perspective, however,the dominant energy transfer and chemical reaction processes

in these plasmas are still not fully understood, especially whenfairly complex hydrocarbon fuels are involved. Without suchunderstanding, predictive kinetic modeling and analysis ofplasma-assisted combustion phenomena remains problematic.The main objective of the present work is to identify and assessthe most significant challenges to quantitative insight into fuel–air plasma kinetics, as well as plasma-assisted ignition andflameholding, based on the results of recent experimental andmodeling studies.

2. Electric field and electron density: measurementsand prediction

It is well known that energy partition among different energymodes in non-equilibrium, low-temperature, high-pressureplasmas (rotational, vibrational, electronic, dissociation, andionization) is controlled primarily by the electron energydistribution function. Over a wide range of conditions,this partition can be characterized in terms of the reducedelectric field, E/N . Electron density, ne, is the otherparameter that controls energy coupling to the plasma, suchthat specific power loading is proportional to the product ofE/N and ne. Thus, knowledge of these two key parameters

0741-3335/15/014001+08$33.00 1 © 2015 IOP Publishing Ltd Printed in the UK

Plasma Phys. Control. Fusion 57 (2015) 014001 I V Adamovich and W R Lempert

Figure 1. Experimental [6] and predicted electric field and appliedvoltage/electrode gap in a ns pulse, plane-to-plane discharge in N2 at0.25 bar, electrode gap of 1.2 mm.

is critical for insights into the kinetics of energy transferand chemical reactions in these plasmas. Estimating E/N

and ne in the plasma from discharge voltages and currentwaveforms, especially in transient discharges that develop asionization waves, may well result in significant uncertaintiesin predicting other plasma parameters, such as vibrationaland electronic excitation, e.g. N2(X1�, v = 0) + e →N2(X1�, v > 0, A3�, B3�, C3�) + e, as well as moleculardissociation by electron impact, e.g. O2(X1�) + e →O(3P,1 D) + O(3P) + e. Thus, measurements of electric field,electron density, and electron temperature in transient plasmassustained by nanosecond pulse duration discharges, and theircomparison with modeling predictions, are critical for fuel–air plasma characterization and for the validation of kineticmodels. Truly predictive models should incorporate, at thevery least, realistic electrode geometry, the Poisson equationfor the electric field, and equations for number densities ofcharge species (see [3, 4]), such that electric field and electrondensity distributions in the plasma could be predicted basedon the experimental voltage waveform, and compared with theexperimental results.

Four-wave mixing, and coherent anti-Stokes Ramanscattering (CARS) and Thomson scattering are two powerfullaser diagnostics that have been used recently for time-resolvedelectric field measurements [5–7], as well as time-resolvedelectron density and electron temperature measurements [8–10] in high-pressure, high specific energy loading molecularplasmas. Electric field measurements have been performedusing both H2 and N2 as the active species, which areamong the dominant species in plasma-assisted combustionenvironments. Sub-nanosecond time resolution electric fielddiagnostics are under development [7]. Figure 1 compares theexperimental [6] and predicted electric field in a nanosecondpulse discharge in nitrogen at 0.25 bar, using the kinetic

Figure 2. Thomson/Raman scattering spectrum in a 10% O2–Hemixture at 100 Torr, 100 ns after a ns pulse discharge current rise.Distance between spherical electrodes 1 cm, ne = 6 × 1013 cm−3,Te = 1.7 eV [10].

model developed in [11]. Rapid electric field reductionduring breakdown and the effect of cathode voltage fallon the field in the plasma after breakdown (such that thefield is significantly lower compared to the ratio of voltageover distance) are readily apparent. Thomson scatteringelectron density measurements in molecular plasmas are quitechallenging because of strong Rayleigh and Raman scattering[8]; however, Rayleigh scattering can be filtered out usinga mask in a triple-grating spectrometer, and the Ramanspectrum can be subtracted if the signal-to-noise and spectralresolution are sufficiently high [8]. Figure 2 plots a typicalThomson/Raman scattering spectrum in a 10% O2–He mixtureat 100 Torr, 100 ns after the beginning of the discharge pulse.The gap in the center of the spectrum, caused by the blockingof Rayleigh scattering, and the Raman lines of O2 are apparent.The inferred electron density and electron temperature forthese conditions are ne = 6 × 1013 cm−3 and Te = 1.7 eV.

3. Temperature and vibrational level populations:measurements and prediction

The parameter that has the most significant effect on chemicalprocesses in fuel–air plasmas, in particular on the rates ofchain branching fuel oxidation reactions, is gas temperature.Therefore time-resolved and spatially resolved temperaturemeasurements in these plasmas, as well as comparison withkinetic modeling predictions, are extremely important forunderstanding the mechanism of plasma-assisted ignition.Since in molecular plasmas a significant fraction of dischargeenergy is loaded into the vibrational mode of nitrogen, withsubsequent vibrational relaxation, time-resolved and spatiallyresolved measurements of N2 vibrational temperature andvibrational level populations are also critical for predictingthe rate of energy thermalization in the plasma. Recently,these measurements have become available due to progress inthe development of picosecond CARS [12] and spontaneousRaman scattering laser diagnostics [13]. As expected, O2

vibrational excitation is much less pronounced compared tothat of N2 [13].

2

Plasma Phys. Control. Fusion 57 (2015) 014001 I V Adamovich and W R Lempert

Figure 3. Experimental and predicted temperature and N2

vibrational temperature during and after a ns pulse discharge in airbetween two spherical electrodes 1 cm apart at 100 Torr [12].

Figure 3 plots experimental picosecond CARS andpredicted, time-resolved rotational/translational temperatureand ‘first level’ N2 vibrational temperature, Tv = ωe(1 −2 × e)/ln(Nv=0/Nv=1), during and after ∼100 ns durationdischarge pulse in air, at 100 Torr [12]. A ‘rapid’ temperaturerise, on a short time scale, t∼10−7–10−6 s, up to T ∼500 K(somewhat difficult to see in figure 3), is primarily due to anenergy release during quenching of excited electronic states ofnitrogen, N2(A3�, B3�, C3�) + O2 → N2 (X1�) + O + O.This has been predicted by kinetic modeling [14] and has alsobeen studied by optical emission spectroscopy in nanosecondpulse discharges at atmospheric pressure [15]. From figure 3, itcan be seen that the discharge produces significant vibrationalnon-equilibrium, Tvmax ∼ 2900 K, Tmax ∼ 850 K. In theseexperiments, N2(v = 0−9) vibrational level populations havealso been measured, indicating a strongly non-Boltzmann,‘bimodal’ vibrational distribution [12]. Kinetic modelingcalculations [12] demonstrated that a significant transient riseof N2 vibrational temperature after the discharge pulse (fromTvmax ∼ 1500 to ∼ 2900 K at t∼10−7–10−4 s, see figure 3) isdue to a ‘downward’ vibration−−vibration energy exchange,N2(v = 2) + N2(v = 0) → N2(v = 1) + N2(v = 1). Tv

reduction at t∼10−4–10−3 s, which coincides with a significant‘slow’ temperature rise in the same time scale, from T ∼500 Kto ∼ 850 K, is due to vibration–translation relaxation of N2 byO atoms, N2(v)+O → N2(v−1)+O. Obviously, both ‘rapid’and ‘slow’ temperature increases, caused by the quenchingof excited electronic states and the vibrational relaxation ofnitrogen, respectively, have a significant effect on the energybalance and thus are critically important for the rates of fueloxidation in nanosecond pulse fuel–air plasmas.

The effect of ‘rapid’ heating in nanosecond pulse fuel–air plasmas is illustrated in greater detail in figure 4,which compares experimental (ps pure rotational CARS)

N cm-3,2*

Tv(N2)T

v(N

2), K

N2*

T, K

T

Figure 4. Experimental and predicted temperature during and aftera ns discharge pulse in an H2–air mixture (φ = 0.14) between twospherical electrodes 0.9 cm apart at 40 Torr, plotted together withpredicted number density of electronically excited N2 molecules andT v(N2).

and predicted (using the kinetic model developed in [11])temperatures during and after a nanosecond pulse discharge ina H2–air mixture at 40 Torr. It can be seen that the time scalesfor the ‘rapid’ temperature rise and the decay of electronicallyexcited N2 molecules are the same (the ‘slow’ rise is dueto N2 vibrational relaxation by O atoms). ‘Rapid’ heatingbecomes considerably more pronounced at higher pressures[15], due to both higher specific energy loading in a localizeddischarge filament and faster quenching of excited electronicstates of N2. This raises the following questions: can plasmachemical reactions among excited electronic species andradicals generated in the low-temperature plasma compete with‘conventional’ fuel–air chemical reactions, the rates of whichrapidly increase with temperature? At what pressures does‘rapid’ heating become the dominant effect in transient fuel–air plasmas, compared to low-temperature excited species andradical species chemistry? Answering these questions wouldhave a considerable impact on the development of plasma-assisted combustion applications. This requires additionalsystematic studies in well-characterized fuel–air plasmas, overa wide range of pressures, as well as comparison with kineticmodeling calculations to evaluate the contributions of differentmechanisms.

4. Effect of excited species and radicals on plasmachemistry

Reactive quenching of excited electronic states and vibrationalrelaxation in air and fuel–air mixtures may have a significanteffect on plasma chemistry, generating radicals such as O, H,and OH, e.g. O(1D) + H2 = H + OH, and dissociating oxygenand fuel species, e.g. N2(A3�, B3�, C3�)+H2 → N2(X1�)+H + H, N2(A3�, B3�, C3�) + O2 → N2(X1�) + O + O.Plasma-generated radicals can initiate fuel oxidation chainreactions, such as H + O2 → OH + O, O + H2 → OH + H,and generate nitric oxide, N∗

2 + O → NO + N (here theasterisk stands for N2 excited electronic or vibrational states),

3

Plasma Phys. Control. Fusion 57 (2015) 014001 I V Adamovich and W R Lempert

Figure 5. Radial distributions of absolute H (left) and OH (right) across a point-to-point discharge filament during and after a ns dischargeburst (50 pulses) in Ar: O2: H2 = 80 : 20 : 2 mixtures at P = 40 Torr and T0 = 300 K [19]. Pulse numbers are indicated in the figure.

Figure 6. Experimental and predicted absolute O, N, and NOnumber densities after a ns pulse discharge in air between twospherical electrodes 1 cm apart at 100 Torr (low pulse energy, nearroom temperature conditions).

which may strongly affect radical concentrations. For thisreason, absolute measurements of electronically excited (bycavity ring-down spectroscopy and calibrated optical emissionspectroscopy) [16] and vibrationally excited N2 molecules (byCARS [12] or spontaneous Raman scattering [13]) as wellas those of N, O, NO, H, and OH radicals (by single-photonand two-photon absorption laser induced fluorescence (LIF)[16–18]) in transient fuel–air plasmas yield critical insightsinto the kinetics of chemical reactions of plasma-generatedradicals. Spatially resolved LIF, two-photon absorptionlaser induced fluorescence (TALIF), and Rayleigh scatteringmeasurements in transient plasmas with strong gradients [19]are especially desirable (see figure 5). Figure 6 comparesexperimental (LIF and TALIF) [18] and predicted [11] time-resolved absolute number densities of O, N, and NO aftera ns pulse discharge in air at 100 Torr, at low pulse energy,

and near room temperature conditions. In these experiments,N2(v = 0–4) populations have also been measured, byps CARS. The results of these measurements demonstratedthat (i) vibrationally excited N2(X1�,v) molecules do nothave a detectable effect on NO formation in ns pulse low-temperature air plasmas, and (ii) NO under these conditions isformed primarily by reactive quenching of a number of excitedelectronic states of N2, including multiple singlet and tripletstates.

The effects of vibrationally excited nitrogen molecules,N2(X1�,v), on plasma chemical reactions have been discussedin the literature for quite some time. In fact, one ofthe objectives of simultaneous ps CARS, LIF, and TALIFmeasurements [18] was to quantify the effect of theN2(X1�, v) + O → NO + N reaction on the rate of NOformation in low-temperature transient air plasmas. However,the contribution of this reaction, compared to the reactivequenching of excited electronic states of N2 appears to be small.This result does not imply that N2(X1�, v) molecules cannotaffect air and fuel–air plasma chemical reactions in general.Recent time-resolved measurements of OH in the nanosecondpulse discharge afterglow in preheated hydrocarbon–airmixtures [20], which demonstrated an anomalously long OHlifetime, led the authors to suggest that vibrationally excitednitrogen may delay OH recombination due to a near-resonanceenergy transfer to the HO2 radical, N2(X1�, v = 1)+ HO2 →N2(X1�, ν = 0) + HO2(ν2+ν3) → N2+H+O2 [21]. However,gas temperature and N2 vibrational temperature in theseexperiments have not been measured, and a more direct effectof heating the flow by the discharge on OH lifetime cannotbe ruled out. Direct vibrational temperature measurementsare critical due to a significant uncertainty in the state-specific rates of N2 vibrational relaxation in reacting fuel–airmixtures. It may well be possible that reactions of vibrationallyexcited N2 molecules in fuel–air plasmas may affectlow-temperature fuel oxidation chemistry, and additionalexperimental studies (such as simultaneous N2 vibrationalCARS and OH LIF measurements) are needed to verify thishypothesis.

4

Plasma Phys. Control. Fusion 57 (2015) 014001 I V Adamovich and W R Lempert

A closely related issue is the effect of long-livedmetastable species generated in molecular plasmas, such asO2(a

1�), on fuel-oxidizer plasma chemistry, compared toradicals participating in rapid reactions of fuel oxidation, suchas O, H, and OH. Recent experiments show that metastableO2(a

1�) molecules, which have a long lifetime and mayaccumulate in O2-containing plasmas up to mole fractions ofseveral per cent, have a significant effect on H2–O2 ignitiondelay length [22] and on C2H4–O2–Ar flame speed [23]. In[23, 24], the effect of O2(a

1�) was isolated from that of Oatoms and ozone (which serves as a source of O and H radicalsin the flame preheat zone). However, the predictive capabilityof kinetic models is limited by the lack of data on collisionalquenching of O2(a

1�) by radicals and hydrocarbon species[24, 25]. On the other hand, the effect of O atoms on CH4 andC2H4 oxidation in low-temperature plasmas [26–28], as wellas on the extinction limit of low-pressure CH4 diffusion flames[28, 29], appears to be relatively well understood. In particular,at temperatures far below the ignition threshold, radical speciesgenerated in the plasma predominantly recombine and do notresult in ignition, due to the near absence of chain branchingprocesses.

Finally, the effect of low-temperature plasma-generatedradicals on low octane number fuels exhibiting cool flamechemistry versus high octane number fuels for which coolflames are not observed, is not understood and requires furtherexperimental and modeling studies. At temperatures below theself-ignition temperature, plasma-generated O atoms (as wellas other radical or excited species) enhance alkyl radical (R)production (e.g. RH + O → R + OH). The subsequent fateof the alkyl radical determines whether cool flame chemistrywill occur or not [30, 31]. At high temperatures, the alkylradical thermally decomposes to form smaller hydrocarbons,thus delaying significant radical branching and heat releaseuntil later in the reaction sequence and, more importantly,higher temperatures. At low temperatures, the alkyl radicalmay form the alkylperoxy radical by reacting with molecularoxygen (R + O2 → RO2). The stabilization of RO2, followedby internal isomerization to form the hydroperoxyalkyl radical(QOOH), and the addition of a second O2 molecule to formO2QOOH, results in significant radical branching during thedecomposition and further reactions of O2QOOH. Enhancingthis low-temperature cool flame sequence, and enabling it overa broader range of conditions, is desirable for shortening theignition delays of large hydrocarbon fuels. The molecularstructure of the fuel, which affects the octane number,influences the rates of R and QOOH formation, and furtherstudy of how these processes are altered by the plasmachemistry is required [32]. The role of O2(a

1�) is of particularuncertainty in this low-temperature reaction sequence.

5. Development of predictive plasma-assistedignition/combustion chemistry mechanism

Quantitative predictions of ignition by low-temperatureplasmas requires knowledge of the rates of the molecularenergy transfer processes, such as electron impact excitationand dissociation of air and fuel species, vibrational relaxation,

rates and products of reactive quenching of excited electronicstates, reactions among charged species, as well as amechanism of fuel oxidation via ‘conventional’ chemicalreactions at low temperatures. Kinetic mechanisms of‘conventional’ combustion [33, 34] have been developed andvalidated for high temperature conditions and may well beinapplicable at low temperatures typical for many plasma-assisted combustion environments. An illustration of thisis found in a recent study [17] which compares absolutetime-resolved OH and temperature measurements by LIFin lean H2–air and hydrocarbon-air mixtures excited by aburst of nanosecond discharge pulses with kinetic modelingcalculations, using conventional hydrocarbon–air combustionmechanisms. Although modeling predictions for H2–air, CH4–air, and C2H4–air agree with the data fairly well [17], theagreement between the model and the data in C3H8–air isquite poor. This demonstrates the need for the development ofan accurate, predictive plasma-assisted combustion chemistrymechanism. The predictive capability of a plasma chemistrymodel involving such a large number of energy transferprocesses and reactions requires extensive, preferably built-in, kinetic sensitivity analysis to identify the dominantprocesses. Identifying the reduced reaction mechanism isextremely critical, especially for hydrocarbons, since thekinetic modeling of coupled ns pulse discharge dynamics,energy transfer in the plasma, and plasma-assisted combustionrequires the incorporation of a wide range of time scales,∼10−12–10−2 s, and is very computationally intensive evenin 1D geometry [35].

The experimental validation of such a mechanism wouldrequire comparison with the results of plasma-assisted ignitionexperiments with well-characterized conditions, using metricsthat would be sufficiently sensitive to the kinetics used. Themetrics used in conventional combustion kinetics includelaminar flame speed, as well as ignition delay time (e.g.measured in shock tubes or a rapid compression machine).In low-temperature plasma-assisted ignition, these metricsare clearly insufficient since they also need to be correlatedwith measurements of parameters controlling both plasmachemistry and conventional chemistry reactions, such as thenumber densities of key radical species and temperature inthe plasma. Since species number density measurementsusing laser diagnostics require significant signal accumulation,this requires the use of well-reproduced, repetitively pulsedplasma ignition cycles [36]. Time-resolved temperature, N2

vibrational temperature (in case of significant vibrational non-equilibrium), key radical species (such as OH, O, or H),and at least one key dominant species (such as fuel or O2)

measurements during plasma-induced ignition (as illustratedin figure 7), appear to provide sufficient amounts of data forkinetic mechanism validation. Systematic validation wouldrequire these measurements to be made over a range orequivalence ratios, fuels, pressures, and energy coupled to theplasma (e.g. number of discharge pulses). Note that single-shot ignition delay time or ignition temperature measurements,such as those obtained in a shock tube or a rapid compressionmachine, do not lend themselves to time-resolved temperatureand species number density measurements.

5

Plasma Phys. Control. Fusion 57 (2015) 014001 I V Adamovich and W R Lempert

Figure 7. Experimental and predicted temperature and OH duringignition after a ns discharge burst in plane-to-plane geometry.H2–air, T0 = 500 K, P = 80 Torr, φ = 0.4, 115 pulses [36].Temperature and Tv (N2) are measured by ps CARS at P = 92 Torr,120 pulses [17].

An approach that is complementary to time-resolvedmeasurements during repetitively pulsed ignition cycles couldbe spatially resolved measurements of temperature and speciesnumber densities across a premixed or diffusion laminarflame front, at conditions where radical species are generatedby the plasma. These experiments would not require theachievement of well-reproduced repetitive ignition, whichrequires operation at very low discharge burst repetition rate,∼1 Hz, to remove combustion products from the test cell [36].An example of these measurements (by OH LIF) [37], alongwith modeling calculations, is illustrated in figure 8 [38]. Notethat the effect of the plasma on flame speed and reaction zonelocation appears to be relatively modest [23, 24, 37], and assuch is unlikely to be a good metric for kinetic mechanismvalidation. On the other hand, measurement of radical speciesand temperature distributions in the preheat zone of the flame,especially in fuel-lean conditions when the effect of plasma ismost pronounced [37], may well be used for this purpose.

6. Effect of plasmas on non-premixed turbulentflames and reacting compressible flows

The effect of radical generation in nanosecond pulse dischargeplasmas on blow-off velocity and lean flammability limitsin premixed and jet diffusion flames has been studied fairlyextensively [39, 40], demonstrating stable combustion andflameholding at significantly higher velocities and lowerequivalence ratios compared to baseline. Recent experimentsalso provided insight into the temporal response of a lean,premixed, weakly turbulent flame to nanosecond pulsedischarge excitation [41], suggesting that the plasma may helpto control combustion instabilities, such as coupling betweenheat release and acoustic oscillations. However, the effect ofplasmas on non-premixed turbulent flames at high Reynolds

Figure 8. Experimental [37] and predicted [38] temperature and OHdistributions across a burner stabilized H2–O2–N2 flat flame(φ = 0.5, P = 25 Torr), with (blue) and without (red) ns dischargeburst (200 pulses) in plane-to-plane geometry. Squares and solidlines: OH; triangles and dashed lines: temperature.

numbers is still to be explored. In these flames, the presenceof plasma may help prevent local extinction and/or acceleratere-ignition by producing a pool of radicals at the locationswhere the local temperature and reaction rates become toolow to sustain combustion, due to flow field turbulence andits interaction with finite-rate chemistry. Additional work inthis field, using quantitative laser diagnostics to measure flowfield velocity and scalars in turbulent flames in the presenceof non-equilibrium plasma is critical. In particular, high-frame rate (∼10 kHz) turbulent flame imaging of velocity,temperature and species concentrations (e.g. particle imagingvelocimetry [42], OH planar laser induced fluorescence (PLIF)[42], Rayleigh scattering [43], Raman scattering [44], CHPLIF [45], and CH2O PLIF [46]), is necessary in order tobegin to understand the coupling between plasma kinetics andturbulent combustion dynamics.

One of the most critical issues of plasma-assisted ignition,combustion, and flameholding in non-premixed high-speedflows is whether the plasma needs to be located in the airflow,in the fuel injection flow, or in a partially premixed regionnear the fuel injector. Since mole fractions of hydrocarbonfuels in the mixture are typically fairly low (a few percenteven at stoichiometric conditions), it appears that plasmaexcitation of airflow and generation of sufficient amountsof excited electronic species and radicals, N∗

2 and O atoms,would be the obvious approach. However, the excitation ofthe airflow upstream of the fuel injection port may resultin a significant loss of excited species and radicals, bothdue to quenching and recombination, and due to mixingwith unexcited air, thus diluting the radical pool. A similarproblem exists with plasma excitation of a partially premixedfuel–air region in the vicinity of a fuel injector [47, 48].Since a rapid mixing of fuel and airflow is critical for stablecombustion, the radicals generated in the discharge may alsobe significantly diluted in the flow, thus reducing the impactof plasma excitation. A recently suggested approach [49]is to sustain the discharge partially in the fuel injection lineand terminate the current path in the combustor, such thatthe discharge filament would penetrate into the flow with the

6

Plasma Phys. Control. Fusion 57 (2015) 014001 I V Adamovich and W R Lempert

fuel jet. The advantages of this approach include (i) radicalgeneration in high-speed fuel injection flow and their rapidtransport into the airflow; (ii) discharge following a high-speed fuel jet and thus penetrating deep into the airflow; and(iii) fuel–air mixing enhancement. Simple analysis of highelectron density plasmas in a high-speed flow [49], such thatthe local ionization/recombination balance is not affected byconvection, demonstrates that a steady-state discharge filamentmay exist only if the current vector is parallel to the flowvelocity vector in the filament,

⇀j ‖ �u. Recently, it has also

been demonstrated that fuel–air mixing can be enhancedsignificantly by unsteady oscillating discharge filaments [50].Additional studies, both experimental and coupled electricdischarge/reacting compressible flow modeling, are necessaryto obtain quantitative insight into discharge behavior in high-speed compressible flows, and its effect on ignition andflameholding.

Finally, one of the least explored problems in non-premixed, high-speed compressible flows is the effect ofplasma-assisted ignition on combustion instability develop-ment. Evidence of such instability, caused by strong couplingbetween pressure in the combustor and discharge parameters(voltage and current), and resulting in high-amplitude oscil-lations of pressure and flame front, was detected in recentsupersonic combustion experiments [48]. A systematic ex-perimental study of this coupling is necessary to obtain insightinto the instability development mechanism, and possibly tocontrol and suppress it.

7. Summary

An overview of recent experimental and kinetic modelingstudies of low-temperature plasma-assisted combustionexhibits the following key challenges:

(i) Measurements and modeling predictions of electric field,electron density, gas temperature, and N2 vibrational levelpopulations in high-pressure pulsed fuel–air plasmas arenecessary for predictive insight into discharge energypartition.

(ii) Measurements and modeling predictions of excitedelectronics states of N∗

2 and key radicals (O, H, OH, andNO) are critical for quantifying their effect on pulsed fuel–air plasma chemistry, including fuels exhibiting cool flamechemistry.

(iii) Development of a predictive plasma-assisted combustionchemistry mechanism requires a set of temperature, N2

vibrational temperature, and radical concentration dataduring a repetitively pulsed plasma-enhanced ignitionprocess, or across a pulsed plasma-enhanced flame.The results would help quantify the effect of ‘rapid’heating, and the possible reactions of vibrationally excitedmolecules, compared to reactions of plasma-generatedradicals.

(iv) Kinetic sensitivity analysis is necessary to identifythe reduced reaction mechanism, which can beincorporated into a practical discharge dynamics/energytransfer/plasma chemistry kinetic model incorporating awide range of time scales and a realistic geometry.

(v) The effect of plasmas on non-premixed turbulent flamesneeds to be studied using high frame rate imaging oftemperature and radical species concentration fields.

(vi) Electric discharge dynamics and combustion instabilitydevelopment in reacting compressible flows need to bestudied experimentally, and by using coupled electricdischarge/reacting compressible flow modeling.

Acknowledgments

This work has been supported by the US Air Force Office ofScientific Research MURI ‘Fundamental Aspects of PlasmaAssisted Combustion’, Technical Monitor Dr Chiping Li.The authors would also like to thank R Yetter, J Sutton,S Leonov, N Popov, and N Aleksandrov for extensive technicaldiscussions.

References

[1] Starikovskaia S M 2006 J. Phys. D: Appl. Phys. 39 R265[2] Starikovskiy A and Aleksandrov N 2013 Prog. Energy

Combust. Sci. 39 61[3] Tholin F, Lacoste D A and Bourdon A 2014 Combust. Flame

161 1235[4] Takana H and Nishiyama H 2014 Plasma Sources Sci.

Technol. 23 034001[5] Ito T, Kobayashi K, Czarnetzki U and Hamaguchi S 2010

J. Phys. D: Appl. Phys. 43 062001[6] Boehm P, Luggenhoelscher D and Czarnetzki U 2013

Laser-spectroscopic electric field measurements in ans-pulsed microplasma in nitrogen 66th Annual GaseousElectronics Conf. (Princeton, NJ, October 2013)

[7] Goldberg B, O’Byrne S and Lempert W R 2012 AIAA Paper2012-0240 50th AIAA Aerospace Sciences Meeting(Nashville, TN, January 2012)

[8] van Gessel A F H, Carbone E A D, Bruggeman P J and van derMullen J J A M 2012 Plasma Sources Sci. Technol.21 015003

[9] van Gessel B, Brandenburg R and Bruggeman P 2013 Appl.Phys. Lett. 103 064103

[10] Roettgen A, Adamovich I V and Lempert W R 2014 AIAAPaper 2014–2113 45th AIAA Plasmadynamics and LasersConf. (Atlanta, GA, June 2014)

[11] Shkurenkov I, Burnette D, Lempert W R and Adamovich I V2014 Kinetics of excited states and radicals in a nanosecondpulse discharge and afterglow in nitrogen and air PlasmaSources Sci. Technol. 23 065003

[12] Montello A, Yin Z, Burnette D, Adamovich I V andLempert W R 2013 J. Phys. D: Appl. Phys. 46 464002

[13] Lo A, Cleon G, Vervisch P and Cessou A 2012 Appl. Phys. B:Laser Opt. 107 229

[14] Popov N A 2001 Plasma Phys. Rep. 27 886[15] Rusterholtz D L, Lacoste D A, Stancu G D, Pai D Z and

Laux C O 2013 J. Phys. D: Appl. Phys. 46 464010[16] Stancu G D, Kaddouri F, Lacoste D A and Laux C O 2010

J. Phys. D: Appl. Phys. 43 124002[17] Yin Z, Montello A, Carter C D, Lempert W R and

Adamovich I V 2013 Combust. Flame 160 1594[18] Burnette D, Montello A, Adamovich I V and Lempert W R

2014 Nitric oxide kinetics in the afterglow of a diffuseplasma filament Plasma Sources Sci. Technol. 23 045007

[19] Yin Z, Eckert Z, Adamovich I V and Lempert W R 2014Time-resolved radical species and temperature distributionsin an Ar–O2–H2 mixture excited by a nanosecond pulsedischarge Proc. Combust. Inst. at press

7

Plasma Phys. Control. Fusion 57 (2015) 014001 I V Adamovich and W R Lempert

[20] Wu L, Lane J, Cernansky N P, Miller D L, Fridman A A andStarikovskiy A Yu 2011 Proc. Combust. Inst. 33 3219

[21] Starikovskiy A 2012 AIAA Paper 2012-0244 50th AIAAAerospace Sciences Meeting (Nashville, TN, January 2012)

[22] Smirnov V V, Stelmakh O M, Fabelinsky V I, Kozlov D N,Starik A M and Titova N S 2008 J. Phys. D: Appl. Phys. 41192001

[23] Ombrello T, Won S H, Ju Y and Williams S 2010 Combust.Flame 157 1906

[24] Ombrello T, Won S H, Ju Y and Williams S 2010 Combust.Flame 157 1916

[25] Popov N A 2011 Plasma Sources Sci. Technol. 20 045002[26] Uddi M, Jiang N, Mintusov E, Adamovich I V and

Lempert W R 2009 Proc. Combust. Inst. 32 929[27] Zuzeek Y, Choi I, Uddi M, Adamovich I V and Lempert W R

2010 J. Phys. D: Appl. Phys. 43 124001[28] Sun W, Uddi M, Won S H, Ombrello T, Carter C and Ju Y

2012 Combust. Flame 159 221[29] Sun W, Uddi M, Ombrello T, Won S H, Carter C and Ju Y

2012 Proc. Combust. Inst. 33 3211[30] Westbrook C K 2013 Annu. Rev. Phys. Chem. 64 201[31] Zador J, Taatjes C A and Fernandes R X 2011 Prog. Energy

Combust. Sci. 37 271[32] Tsolas N, Togai K and Yetter R A 2015 Kinetics of plasma

assisted hydrocarbon oxidation in flow reactors 53rd AIAAAerospace Sciences Meeting (SciTech 2015) (Kissimmee,FL, 5–9 January 2015)

[33] Wang H, You X, Joshi A V, Davis S G, Laskin A,Egolfopoulos F and Law C K 2007 High-TemperatureCombustion Reaction Model of H2/CO/C1-C4 CompoundsUSC Mech Version II(http://ignis.usc.edu/USC Mech II.htm May 2007)

[34] Konnov A 2000 Detailed Reaction Mechanism for SmallHydrocarbons Combustion, Release 0.5.http://homepages.vub.ac.be/∼akonnov/

[35] Nagaraja S, Yang V and Adamovich I 2013 J. Phys. D: Appl.Phys. 46 155205

[36] Yin Z, Adamovich I V and Lempert W R 2013 Proc. Combust.Inst. 34 3249

[37] Li T, Adamovich I V and Sutton J A 2014 AIAA Paper2014-0664 52nd AIAA Aerospace Sciences Meeting(SciTech 2014) (National Harbor, MD, 13–17 January2014)

[38] Nagaraja S, Li T, Sutton J, Adamovich I and Yang V 2014Nanosecond plasma enhanced H2/O2/N2 premixed flatflames Proc. Combust. Inst. at press

[39] Pilla G, Galley D, Lacoste D A, Lacas F, Veynante D andLaux C O 2006 IEEE Trans. Plasma Sci. 34 2471

[40] Bak M S, Im S-K, Mungal M G and Cappelli M A 2013Combust. Flame 160 2396

[41] Lacoste D A, Xu D A, Moeck J P and Laux C O 2013 Proc.Combust. Inst. 34 3259

[42] Bohm B, Heeger C, Gordon R L and Dreizler A 2011 Newperspectives on turbulent combustion: multi-parameterhigh-speed planar laser diagnostics Flow Turbul. Combust.86 313–41

[43] Patton R A, Gabet K N, Jiang N, Lempert W R and Sutton J A2012 Appl. Phys. B 108 377

[44] Gabet K N, Jiang N, Lempert W R and Sutton J A 2010Appl. Phys. B 101 1

[45] Gabet K N, Patton R A, Jiang N, Lempert W R and Sutton J A2012 Appl. Phys. B 106 569

[46] Jiang N, Patton R A, Lempert W R and Sutton J A 2011Proc. Combust. Inst. 33 767

[47] Do H, Cappelli M A and Mungal M G 2010 Combust. Flame157 1783

[48] Do H, Im S-K, Cappelli M and Mungal M G 2010 Combust.Flame 157 2298

[49] Savelkin K V, Yarantsev D A, Adamovich I V and Leonov S B2014 Ignition and flameholding in a supersonic combustorby an electrical discharge combined with a fuel injectorCombust. Flame accepted

[50] Leonov S B and Yarantsev D A 2007 Plasma Sources Sci.Technol. 16 132

8