Nitrogen Vibrational Population Measurements in the Plenumof a Hypersonic Wind Tunnel

A. Montello,∗ M. Nishihara,† J. W. Rich,‡ I. V. Adamovich,§ and W. R. Lempert§

Ohio State University, Columbus, Ohio 43210

DOI: 10.2514/1.J051514

Picosecond coherent anti-Stokes Raman scattering is used for measurement of nitrogen vibrational distribution

function in the plenum of a highly nonequilibrium Mach 5 wind-tunnel incorporating a high-pressure pulser–

sustainer discharge. First-level vibrational temperatures of the order of 2000K are achieved in the 300 torr non-self-

sustained plasma discharge generated by a high E=n (�300 Td) nanosecond-pulsed discharge, which providesionization in combination with an orthogonal low E=n (�10 Td) dc sustainer discharge, which efficiently loads thenitrogen vibrational mode. It is also shown that operation with the nanosecond-pulsed plasma alone results in

significant vibrational energy loading, withTv�N2� of the order of 1100K.Downstream injection ofCO2, NO, andH2results in vibrational relaxation, demonstrating the ability to further tailor the vibrational energy content of the flow.

N2-NO vibration–vibration andN2-H2 vibration–translation rates inferred from these data agreewell with previousliterature results to within the uncertainty in rotational-translational temperature.

I. Introduction

T HE ability to tailor nonequilibrium hypersonic flows by controlof the loading of internal degrees of freedom (vibrational andelectronic states), as well as by control of dissociation and ionizationfraction, is a topic of much current interest. For example, turbulenttransition delay in aMach 5 flow over a 50 cone bymeans of injectionof carbon dioxide into nitrogen or airflow has been recentlydemonstrated [1]. Kinetic modeling calculations [2] suggest thattransition delay is caused by absorption of acoustic perturbations inthe boundary layer by vibrational energy modes of CO2, which mayalso result in CO2 dissociation. Another example is relaxation ofenergy stored in internal degrees of freedom of molecules behind abow shock, which may significantly increase shock standoffdistance. The presence of vibrationally and electronically excitedspecies in a hypersonic flow may strongly affect the emissionsignature from the shock layer. Finally, short-pulse electric dis-charges efficiently loading electronic energy levels of nitrogen andoxygen in air are currently being explored as means of hypersonicflow control by producing repetitive localized pressure perturbationsin the flow [3].

Since the cost of obtaining full-scale hypersonic flight-test data oroperating large-scale ground test facilities is extremely high,alternativemethods of data production are necessary. This provides asignificant incentive for the development and use of small-scale testfacilities that are capable of recreating environments seen in the flowconditions of interest, and which lend themselves to development ofoptical diagnostics of nonequilibrium hypersonic flows. Our

approach to generating a nonequilibrium hypersonic flow is to use ahigh-pressure low-temperature diffuse electric discharge sustained inthe plenum of a small-scale Mach 5 wind tunnel to load internalenergy modes of nitrogen and oxygen molecules [4,5]. A similarapproach has been used previously to develop electrically excited gasdynamic lasers [6,7].

Based on the results of our previous work, target nonequilibriumairflow parameters downstream of the discharge are as follows:runtime at steady-state conditions 5–10 s, plenum pressureP0 � 0:3–1:0 atm, translational/rotational temperature T0 � 300–400 K, and vibrational temperature Tv0 � 2000 K. Internal energymode disequilibrium in the flow excited in the discharge can becontrolled by injecting rapid relaxer species (nitric oxide, hydrogen,or carbon dioxide) into the subsonic nonequilibrium flow betweenthe discharge section and the nozzle throat. Quantitative analysis ofthese flows requires characterization of energy partitioning amongthe internal modes (vibrational and electronic). Kinetic modelinganalysis of strongly vibrationally and electronically nonequilibriumN2–CO–NO flows in supersonic nozzles has been presented inprevious work [8,9]. However, the predictive capability of non-equilibrium flow codes used for kinetic modeling is controlledprimarily by the accuracy of rate models for the vibrational energytransfer they are incorporating. Although state-specific nonempiricalvibrational energy transfer models among air species are available[10,11], and have been incorporated into flow codes [12,13],experimental data for validation of code predictions are scarce.

In this paper, we provide new experimental data characterizingvibrational energy loading of molecular nitrogen in a hypersonicnonequilibrium flow vibrationally excited by a non-self-sustainedelectric discharge in the plenum of a Mach 5 wind tunnel. Volumeionization, at plenum pressures in the range ofP0 � 200–370 torr, isaccomplished using a high peak reduced electric field (E=n�300 Td), 5 ns duration pulse discharge operating at a pulse repetitionrate of 100 kHz. An orthogonal dc sustainer discharge (E=n�10–30 Td), which accounts for approximately 80% of the totalpower loading into the flow, efficiently excites the N2 vibrationalenergy mode with as much as 80–90% of its input power going intovibrational excitation by electron impact. Temporally and spatiallyresolved vibrational level populations of nitrogen, N2�X1�g;v� 0–3�, as well as first-level vibrational temperature in the pulser–sustainer discharge, up to Tv�N2� � 2000 K, are measuredusing picosecond coherent anti-Stokes Raman scattering (CARS)spectroscopy [14], which is described in detail. It is also shown thatinjection of gases such as CO2, NO, and H2 downstream of thedischarge section, upstream of the CARS measurements location,results in partial vibrational relaxation of nitrogen excited in the

Received 20 July 2011; revision received 28 November 2011; accepted forpublication 30 November 2011. Copyright © 2012 by Aaron Montello.Published by the American Institute of Aeronautics and Astronautics, Inc.,with permission. Copies of this paper may be made for personal or internaluse, on condition that the copier pay the $10.00 per-copy fee to the CopyrightClearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923; includethe code 0001-1452/12 and $10.00 in correspondence with the CCC.

∗Graduate Research Assistant, Michael A. Chaszeyka NonequilibriumThermodynamics Laboratories, Department of Mechanical and AerospaceEngineering, 201 West 19th Avenue. Student Member AIAA.

†Postdoctoral Researcher, Michael A. Chaszeyka NonequilibriumThermodynamics Laboratories, Department of Mechanical and AerospaceEngineering, 201 West 19th Avenue. Member AIAA.

‡Professor Emeritus, Michael A. Chaszeyka Nonequilibrium Thermody-namics Laboratories, Department ofMechanical andAerospace Engineering,201 West 19th Avenue. Fellow AIAA.

§Professor, Michael A. Chaszeyka Nonequilibrium ThermodynamicsLaboratories, Department of Mechanical and Aerospace Engineering, 201West 19th Avenue. Associate Fellow AIAA.

AIAA JOURNALVol. 50, No. 6, June 2012

1367

http://dx.doi.org/10.2514/1.J051514

discharge. This demonstrates quantitatively the ability to control andtailor the vibrational energy content of the flow before its expansionthrough the Mach 5 nozzle of the wind tunnel. On the other hand,adding oxygen to the vibrationally excited nitrogen flow did notresult in a detectable vibrational temperature change, due to slowvibration–vibration (V–V) energy transfer from N2 to O2 and slowvibration–translation (V–T) relaxation at near room temperature.

II. Experimental

A. Mach 5 Wind Tunnel

The experiments conducted for this study were performed in theplenum of a Mach 5 nonequilibrium flow tunnel, previouslydeveloped and described in detail [4,5]. Briefly, the laboratory-scalewind tunnel, a schematic of which is given in Fig. 1, operates atplenum pressures from 0.25 to 1 atm with nitrogen or air suppliedfrom high-pressure cylinders. The tunnel is constructed fromtransparent acrylic plastic and is capable of producing steady-statenonequilibrium supersonic flow using a high-pressure diffuse,nanosecond pulser/dc sustainer electric discharge operating in theplenum section. The discharge can be tailored, as will be discussed,to load internal (vibrational and electronic) modes of nitrogen whilemaintaining low translational–rotational temperatures of T�300–400 K. About 8 cm downstream of the discharge section, thegas flows through a choked flow injector with 20 injection ports,1 mm in diameter, in both the top and bottom channel walls, alsoshown in the figure. Gases inducing vibrational relaxation ofnitrogen, such as oxygen, hydrogen, nitric oxide, CO2, etc., can beinjected into the flow, which allows further control of the energydistribution within the internal molecular modes in the flow. Themeasurement location is 9 cm downstream from the gas injection.The supersonic nozzle throat is downstream of the measurementlocation, and it is therefore unimportant for thework conducted here.The outlet of the flow tunnel leads to a 110 ft3 vacuum tank and200 ft3=min pump. The generation of nonresonant background(NRB) from the windows, which limits the sensitivity of the CARSdiagnostic, was reduced by using optical extension arms, as shown inFig. 1. It would be possible to remove the NRB contributed from thewindows completely via utilization of an alternate alignmentscheme; however, this was not done for the sake of experimentalsimplicity in this case.

The pulser–sustainer electric discharge, a schematic of which isgiven in Fig. 2, comprises two fully overlapping discharges:1) repetitively pulsed high-voltage nanosecond discharge and2) transverse dc sustainer discharge. As shown in Fig. 2, the pulserelectrodes (4 � 3 cm copper plates) are flush-mounted in the top andbottom walls of the discharge section (nozzle plenum), separated by0.5 cm, and are each covered by alumina ceramic dielectric plates116

in: thick. The bare sustainer electrodes (4 � 0:5 cm copper plates)are placed along the side walls of the discharge section and areremovable. In the present experiments, the sustainer electrodesare separated by 3.0 cm. Both the pulser and sustainer electrodes arerounded along the edges to reduce electric field nonuniformity. The

acrylic plastic walls of the channel downstream of the dischargesection are covered with alumina ceramic plates attached withsilicone rubber adhesive, which protects the walls from overheatingin the possibility of instability development or arcing in thedischarge. The main benefit from using the pulser–sustainerdischarge is the ability to generate stable nonequilibrium plasmas athigh pressures and high discharge energy loading.

The repetitive nanosecond-pulsed discharge operates using a highpeak voltage (up to 30 kV), short-pulse duration (5 ns), high pulserepetition rate (up to 100 kHz) voltagewaveformproduced by a high-voltage nanosecond pulse generator (FID GmbH). Figure 3 showstypical pulse voltage and current traces (top), aswell as instantaneouspower and coupled pulse energy traces (bottom), measured in arepetitively pulsed discharge in nitrogen at P0 � 350 torr and ��100 kHz [5], without dc electrodes present in the test section. Thewaveforms shown are for a pulse generated 0.1 s after the beginningof the pulse burst (i.e., for pulse number 10,000). The pulser isoperated using an external trigger/function generator. Each high-voltage pulse generates volume ionization in the discharge section,and the high voltage is then turned off before ionization/heatinginstability [15] has time to develop. The high frequency of pulserepetition prevents the plasma from fully decaying before the nextionizing pulse occurs, thus providing pulse-periodic spatiallyuniform ionization in the discharge section [16,17].

Between the ionizing pulses, energy is coupled to the flow by thedc discharge. The dc electrodes are connected to a Glassman 5 kV,2Apower supply operated in a voltage-stabilizedmode in serieswitha 1:5 k� ballast resistor. The dc voltage is deliberately kept belowbreakdown threshold, typically below 4–5 kV, to precludedevelopment of self-sustained (i.e., independent of pulsed ioniz-ation) dc discharge in a high-pressure flow, which would result ininstability development and arcing. The power coupled to the flowbythe dc discharge is significantly higher than the repetitively pulseddischarge power. Previously, this approach has been used in ourworkto sustain high-power discharges in a Mach 3–4 magneto-hydrodynamics wind tunnel [16,17] and in an electrically excited gasdynamic oxygen-iodine laser [7]. In the present experiments, therepetitively pulsed discharge is operated for up to several seconds andthe dc discharge for 0.5–1.0 s. The dc power supply is triggered by anexternally generated rectangular shaped trigger pulse. The rising

Fig. 1 Schematic of top view of the pulser–sustainer discharge showing optical extension arms and the location of optical measurement.

Fig. 2 Schematic of pulser–sustainer electric discharge geometry.

1368 MONTELLO ETAL.

edge of the rectangular pulse also triggers the function generator,which produces a burst of trigger pulses for the high-voltagenanosecond pulse generator, at preset pulse repetition rate andnumber of pulses in the burst. The estimated reduced electric fields inthe two discharges are significantly different: �E=N�peak � 300 Td inthe nanosecond-pulsed discharge, and E=N � 10 Td in the dcdischarge (1 Td� 10�17 V � cm2). At these conditions, a significantfraction of input power in the pulsed discharge is spent on electronicexcitation, dissociation, and ionization of nitrogen, while nearly allinput power in the dc discharge (up to�80–90% [15]) is stored in thevibrational energy mode of nitrogen, with little power going totranslational/rotational modes, i.e., to heat. Because of a very longN2vibrational relaxation time at near room temperature, �1 atm � s[18], this approach can create essentially vibrationally frozennitrogen and airflows in the supersonic test section, with vibrationaltemperature greatly exceeding the translational/rotational modetemperature.

During the experiment, both the main flow through the dischargeand the injection flows are controlled using solenoid valves. Themain flow rate is calculated using a choked flow equation based onthe plenum pressure and the nozzle throat area. The injection flow

rate has been bothmeasured directly using amass flow controller andcalculated from the choked flow equation, with both methods givingsimilar results. At the baseline conditions, nitrogen at P0 � 0:5–1:0 atm, the mass flow rate through the tunnel is 7:5–15 g=s and thesteady-state runtime at constant static pressure in the supersonic testsection is 5–10 s; runs can be repeated every few minutes.

B. Picosecond Coherent Anti-Stokes Raman Scattering

Diagnostic System

For direct measurement of vibrational energy loading provided bythe pulser–sustainer discharge, the CARS spectroscopic techniquehas been employed. CARS is a four-wave mixing technique that hasbeen used extensively for thermometry of combustion and other gasphase reacting and nonreacting flows [14,19]. Two of the primarybenefits of CARS are the coherence of the generated signal and theability to take localized measurements with high spatial resolution,resolved in either two or three dimensions depending on the phase-matching scheme employed. For this work, collinear, rovibrational,picosecond CARS has been used. The collinear geometry offersrelative ease of alignment over other arrangements while introducingonly minor complication for the data interpretation for thisexperiment, and the rovibration is chosen since the main objective ofthe present work ismeasurement of vibrational distribution functions(VDFs) of nitrogen. The choice to use a picosecond system ismotivated by several reasons. The use of picosecond CARS allowsthe possibility of complete suppression of the NRB; enables time-resolved measurements at nanosecond and subnanosecond time-scales; lowers necessary pulse energies, reducing the risk of windowdamage; and allows measurements in harsh environments via fibercoupling [14,20,21].

Figure 4 shows a schematic diagram of the CARS instrument andits interface to the test section. An Ekspla SL-333 Nd:YAG laserserves as the pump/probe source. The oscillator of the relativelycompact laser, which operates at 10 Hz, outputs pulses of approxi-mately 2 ns that are then pulse compressed via stimulated Brillouinscattering to approximately 150 ps. After two double pass amplifi-cation stages and a second harmonic crystal, the laser outputs pulsesof approximately 150 ps, with variable energy up to 120 mJ=pulse(at 532 nm). In addition to serving as the pump/probe beams for theCARS mixing, the Nd:YAG laser output also pumps the dye laser.Through the use of a broadband laser source for the Stokes beam,multiple nitrogen vibrational levels can be interrogated simulta-neously;with sufficient laser intensity, a single laser shot can producean entire spectrum from which the VDF can be obtained.

Fig. 3 Pulse voltage and current waveforms (top) and instantaneous

power and coupled energy (bottom) in nitrogen at P0 � 350 torr and�� 100 kHz (pulse number 10,000 in a burst) [5].

Fig. 4 Schematic diagram of picosecond CARS system.

MONTELLO ETAL. 1369

The broadband dye laser, patterned after that of Roy et al. [22],uses side-pumped oscillator and preamplifier cells, followed by anend-pumped final amplifier cell. With no output coupler, there is nodevelopment of mode structure in the dye laser beam, terming this“modeless”, and the lack of a grating provides broadband operation.A half-wave plate and thin film polarizer allow for adjustment of theratio between dye pump energy versus CARS pump/probe beamenergies. Dye laser energy efficiency is typically near 7%. For thiswork, amixture of rhodamine 640 (R640) and kiton red 620 (KR620)was used to achieve the desired spectral output. The oscillator/preamplifier cell were filled with amixture of 34mgR640 and 18mgKR620 in 750 mL of methanol, while the amplifier cell solution was20 mg R640 and 18 mg KR620 in 750 mLmethanol. Figure 5 showsthe broadband dye laser output, which is seen to span the first severalvibrational transitions of nitrogen.

In addition to the wavelength and phase-matching requirements,generation of maximum CARS signal requires the beams to arrive atthe interrogation region coincidentally in time [23]. An optical delaypath in the pump/probe beams is required to match optical pathlengths, accounting for the time lag of the dye laser. A dichroicmirrorreflecting the 532 nm pump/probe and transmitting the 607 nmStokes allows for the collinear beam combination, with the combinedbeams then focused into the test section with a 250 mm focal lengthlens. After the focal point, located in themiddle of the test section, thebeams are recollimated using another 250 mm focal length lens.Beyond this, a series of long-wavelength passing dichroic mirrorsreflect the 473 nm anti-Stokes signal and dump the pump/probe andStokes light. Finally, the CARS signal passes through a bandpassfilter, centered at 476 nmwith a full width at half maximum of 10 nmand a 100 mm lens focusing onto the spectrometer slit. The 0.75 mspectrometer is an Andor Shamrock 750, with a 600-lines-per-millimeter grating. The charge-coupled device camera is an AndorNewton. The spectrometer and camera are interfaced to a laboratorycomputer for data recording. The entire picosecond CARS system isplaced on a custom-built cart, designed and constructed from t-slotaluminum, allowing the entire setup to be easily transported betweenexperimental facilities.

III. Results and Discussion

In the present work, the nanosecond pulser was operated, asdescribed in Sec. I, at 100 kHz for approximately 0.6 s. During theexperiments with the dc sustainer electrodes present in the testsection, they were powered for approximately 0.5 s. While bothdischarges are initiated by the same triggering pulse, the nanosecondpulser responds essentially instantaneously while the dc sustainervoltage has�100–200 ms rampup time, due to capacitor charging inthe dc power supply. The pulser remains on for approximately100 ms after the dc voltage is switched off to reduce residual voltagebetween the dc electrodes. Figure 6 displays the dc discharge currentprofile in time, with the 10 Hz laser shots superimposed. As can beseen, each run allows for four to five laser shots during the discharge.For the data presented here, the Stokes laser beam was set to1:2 mJ=pulsewhile the pump/probe beam had 5:5 mJ=pulse. While

laser breakdown and stimulated Raman pumping can both becomesignificant as laser intensity increases, care was taken to watch forthese effects, and neither of themwere encountered for the laser pulseenergies used in this work.

Figure 7 shows a typical single laser shot CARS spectrum, takenwith the nanosecond pulser operating alone (i.e., without dc sustainervoltage) in a nitrogen flow through the wind tunnel at stagnationpressure ofP0 � 300 torr. As the figure clearly indicates, the v00 � 0and v00 � 1 peaks are both visible, indicating some vibrationalexcitation in this regime. Figure 8 shows a typical single-shotspectrum for the pulser–sustainer discharge in operation, with the dcsustainer power supply voltage VPS set to 4.5 kV. As can be seen, thevibrational excitation for this case is much more significant than forthe pulser alone, with vibrational levels v00 � 0; 1; 2; 3 distinguish-able and higher levels possibly populated. Figures 9 and 10 show thesame data, plottedwith a logarithmic scale on the vertical axis, whichallows much more clear visualization of the NRB and gives a betterillustration of the signal-to-noise ratio for this arrangement. Note thatthese spectra were each captured several hundred milliseconds afterthe beginning of the nanosecond pulser operation in the steady-stateregion.

Once the evaluation of the spectral peak intensity for all detectedvibrational peaks has been done, the first step in the data reductionprocess is the removal of the NRB baseline, performed byinterpolating spectra minimum values within successive troughs andsubtracting the interpolation envelope. Since the NRB signal occursat the same wavelengths and can constructively and destructivelyinterfere with the resonant nitrogen signal, this method mayintroduce some uncertainty; however, due to the significant differ-ence in magnitude between the resonant and nonresonant signals,this effect should be relatively small. The second step, adjusting theCARS spectra to account for the dye laser spectral profile, must beperformed since the CARS signal intensity is proportional to the

Fig. 5 Broadband dye laser spectral output, with the Stokes

frequencies for the first five vibrational transitions of nitrogen

superimposed (a.u. denotes arbitrary units).

Fig. 6 Timing sequence between dc discharge current and 10 Hz laser

pulses (arb. denotes arbitrary intensity units).

Fig. 7 Typical single-shot CARS spectrum, nanosecond pulser alone,

300 torr N2.

1370 MONTELLO ETAL.

Stokes laser intensity. The third step accounts for the variation in theCARS interaction cross section. Ignoring anharmonic effects, whichare very small for N2 [24], the cross section scales quadratically as�v00 � 1�2. Finally, the square root of these data must be taken, since

the CARS signal is proportional to number density squared, yieldingrelative number densities as a function of v00. Note that, for this work,the data processing described above was performed manually.

The number densities inferred from the raw experimental datausing this procedure are then normalized, and a Boltzmann plot suchas that shown in Fig. 11 is produced. This figure contains thenormalized number densities from a sequence of single-shot spectrataken from seven consecutive laser shots during onewind-tunnel run.As can be seen from thefigure, the first and last samples showweakervibrational excitation than the others, indicating the beginning andend of the dc sustainer voltage application. The v00 � 0 and v00 � 1number density ratio is used to determine the first-level nitrogenvibrational temperature Tv�N2� as follows:

Tv ��v

ln N0=N1

(1)

where �v � 3353 K is the energy difference between vibrationallevels v00 � 1 and v00 � 0 in the units of temperature.

The first-level temperature is chosen instead of attempting a fit ofall vibrational levels present due to the relatively high uncertainty towhich the higher levels are known, owing to the complicating effectsof the NRB and the quadratic dependence of the peaks on numberdensity. As Figs. 7–10 indicate, in the case of pulser operating alone,a vibrational temperature of 1100� 100 K is inferred, whereas forthe pulser–sustainer discharge with VPS � 4:5 kV, the vibrationaltemperature is 2150� 150 K.

To monitor the temporal characteristics of the vibrational loadingof the flow in the discharge, the laser and pulser timing weresynchronized, with the delay between them varied as desired. Thesemeasurements were conducted at a plenum pressure of 300 torr, withthe dc sustainer power supply voltage VPS set to 3.5 kV. The initialrun of the discharge resulted in six single-shot spectra, correspondingto acquisitions at time t� 100; 200; � � �, and 600ms after the pulser–sustainer discharge initiation. This procedure was repeated fourtimes, resulting in four single-shot spectra for each delay time. Therelative delay between pulser and laser was then decreased by 10 msand another four runs were taken, resulting in data collection att� 90; 190; . . . ; 590 ms, etc. This process was repeated until thedelay between the discharge initiation and the first laser pulse wasreduced to 10 ms. Mean and standard deviation statistics werecomputed for the four-point data set at each time position, the resultof which is seen in Fig. 12 alongwith a sample dc sustainer dischargecurrent profile superimposed.

The results demonstrate that the vibrational temperature behaviorvery closely matches the time-resolved sustainer current profile, asexpected. It should be noted that the characteristic time for electronimpact vibrational excitation in the discharge ismuch shorter than thetimescale in Fig. 12 (see discussion below). On the other hand, theestimated characteristic vibrational population decay time at theseconditions, due to extremely slow V–T relaxation in nitrogen (of theorder of a second), is much longer compared with the flow residencetime between the discharge section andCARSmeasurement location(a few milliseconds). Therefore, the results plotted in Fig. 12represent a quasi-steady-state time-resolved vibrational temperaturereached in the pulser–sustainer discharge in nitrogen. Consideringthe interval from 200 through 590 ms as approximately constantsustainer loading, an average Tv�N2� � 1671 K is observed, with a95% confidence interval of 25 K.

The lowest vibrational temperature values seen at the beginningand end of the measurement sequence in Fig. 12 correspond to theapproximate detection threshold for vibrational temperaturemeasurement using the present CARS system, Tv�N2� � 800–1000 K. There are a few factors that may result in reduction of thisthreshold. Most significantly, reduction of the NRB would enablemore sensitive detection of weak transitions. To accomplish this,future work will use what is known as unstable-resonator spatiallyenhanced detection CARS [19,25]. Additionally, increasing laserpower would also result in increased detection sensitivity to lowervibrational populations; however, care must be taken to prevent

Fig. 8 Single-shot spectrum, pulser–sustainer discharge, VPS�4:5 kV, 300 torr N2.

Fig. 9 Typical single-shot CARS spectrum, nanosecond pulser alone,

300 torr N2 (log scale).

Fig. 10 Single-shot spectrum, pulser–sustainer discharge, VPS�4:5 kV, 300 torr N2 (log scale).

MONTELLO ETAL. 1371

saturation effects, both of the CARS process itself and the detectionof the v00 � 0 transition. Note that the timescale for the steep decreasein the vibrational temperature near themoment when the dc sustainervoltage is turned off is controlled by the flow residence time betweenthe discharge section and the CARS measurement region of severalmilliseconds. Furthermore, it should be noted that a few vibrationaltemperature values at the end of the sequence demonstrate the samelevel of excitation measured with the pulser operating alone (e.g. seeFig. 7), corresponding to the time after the sustainer discharge turnsoff but while the pulser is still running. Spectra taken after the pulseris turned off, beyond the last data point in this plot, show nodetectable population of v00 � 1.

The set of data displayed in the plot in Fig. 13 shows the results ofvarying the discharge pressure as well as the sustainer discharge dcvoltage. Each data point results from averaging the vibrationaltemperature obtained from four to five single-shot spectra taken in asinglewind-tunnel run. The trend line of themean Tv�N2� alongwiththe corresponding standard deviations for each pressure conditionare plotted versus the dc power supply voltage setting VPS. Asexpected, increasing the dc sustainer voltage results in highervibrational excitation of nitrogen in the flow. In Fig. 14, Tv�N2� isplotted versus the energy loading per molecule "load (for VPS rangingfrom 2 to 4 kV) for each pressure condition. To calculate "load, thevoltage between the dc electrodes V is first calculated from VPS, thedc discharge current I, and the resistance of the ballast resistorRbal � 1:5 k�:

V � VPS � I � Rbal (2)

This voltage is then reduced by the voltage drop across the cathodesheath V0 estimated from previous measurements of the currentvoltage characteristic of the pulser–sustainer discharge [4],V0 � 290 V, and dc power loading P is calculated as

P� I � �V � V0� (3)

Finally, the energy loading per molecule "load is

"load �P

_n(4)

where _n is themolecular flow rate, calculated from themass flow ratethrough the discharge section/nozzle plenum. As the plotdemonstrates, the two higher pressure discharge cases display aconsistent, direct relationship between Tv�N2� and "load, while thelower pressure 200 torr case indicates somewhat different behavior(discussed in greater detail below).

Figure 15 shows vibrational temperature plotted versus theestimated reduced electric fieldE=n of the dc sustainer discharge. Asis well known [15], the reduced electric field controls the averageelectron energy, and therefore the input energy partitioning among

Fig. 11 Normalized Boltzmann plot for one sequence of single-shot CARS spectra in the pulser–sustainer discharge, VPS � 4 kV, 300 torr N2.

Fig. 12 Temporal evolution of Tv�N2� during sustainer dischargepulse, VPS � 3:5 kV, 300 torr N2.

Fig. 13 Effects of pressure and dc voltage on vibrational temperature

of nitrogen.

Fig. 14 Effects of sustainer dc discharge energy loading per molecule

on Tv�N2�.

1372 MONTELLO ETAL.

different electron impact excitation channels (vibrational andelectronic excitation, dissociation, ionization, etc.). In the presentwork, the reduced electric field was estimated simply as the reducedvoltage between the dc electrodes V-V0 divided by the electrode gapand the number density, assuming 375 K gas temperature in thedischarge. This is justified by our previous translational–rotationaltemperature measurements in the pulser–sustainer discharge at theseconditions [4,5] from N2 emission UV/visible spectroscopy, whichyields temperature in the range T � 350–400 K. Figure 15 showsessentially identical vibrational temperature behavior for all threepressures for E=n �7 Td than that observed for theother two pressure cases, which remain in good agreement. It wasnoted that, at lower plenum pressures, the discharge spatialuniformity decreases, with visible emission being most intense nearthe sidewalls of the discharge section, as illustrated in Fig. 16. In thisevent, a significant fraction of the energy deposition would likelyalso occur near the side walls, leading to a lower relative vibrationalloading at the center of the channel, near the CARS interrogationlocation.

The vibrational temperatures are used to calculate averagevibrational energy per molecule "avg in the discharge and the percentof discharge energy loaded into molecular vibrations �:

��"avg"load

(5)

According to theory [15], this fraction should be nearly 90% at10 Td while dropping off somewhat to 50% near 3.5 Td. Figure 17shows � plotted against E=n for the 370 and 300 torr cases (for therange of VPS � 2–4 kV) along with the theoretically predictedvalues. While the experimentally measured values are somewhatlower than those theoretically predicted, the trend observed is similarto the expected result, and there are a few factors that explain theoffset. Because of the simple collinear geometry of the CARSmeasurements, there is a small amount of signal that is collected fromoutside the test section, i.e., in the vibrationally cold room air, whichacts to reduce the apparent vibrational temperature. Because of the

extreme difference in laser intensity between the focus and thecollimated beam, this effect is slight and is estimated to introduce amaximum of approximately 3% uncertainty into the inferredvibrational temperature. This estimate is achieved by comparing thesignal generated with the tunnel evacuated (i.e., from nitrogenoutside the desired measurement volume) with the v� 0 level peakintensity collected during test runs. Additionally, as mentioned, thecalculation for reduced electric field was performed with theassumption of spatial uniformity, and it is possible that the actualE=nat the measurement location is somewhat lower than the estimatedvalue. A third factor, which was previously mentioned, is thedischarge spatial nonuniformity with the visual intensity, and likelythe energy deposition, being greater near the edges of the dischargethan in the center [4,5] (see Fig. 16). This effect would lead to anoverestimation of the "load near the center of the discharge, thusreducing the apparent � value at the measurement location. Theresults from the lowest pressure 200 torr case are not included here, asthey were not in good agreement with the expected trend, which isthought to be due to the rather poor discharge spatial uniformity atthis condition.

The final set of experimental data was obtained from injection ofgases inducing vibrational relaxation (of nitrogen excited in thedischarge) between the discharge section and the CARS measure-ment location, as shown in Fig. 1. In the present work, four differentinjection gases were used: oxygen, hydrogen, nitric oxide, andcarbon dioxide. These species have been chosen because the rates ofnitrogen vibrational relaxation in these four mixtures vary by severalorders of magnitude. Figure 18 plots nitrogen vibrationaltemperatures measured in these mixtures versus the partial pressureof the injected species, all performed at the same total pressure in theflow of 300 torr. Mole fractions of injection species can be easilycalculated from their partial pressures.

As expected, adding oxygen to the vibrationally excited nitrogenflow, up to 17% (nearly synthetic air), results in negligible reductionofN2 vibrational temperature, due to an extremely slow V–Venergytransfer rate coefficient:

N 2�v� 1� � O2�v� 0� ! N2�v� 0� � O2�v� 1�

kVV � 7 � 10�17 cm3=s at room temperature [26]. The characteristictime for nitrogen V–V relaxation at these conditions �VV �1=kVVnO2 � 10 ms at PO2 � 50 torr is longer than the flowresidence time between the injection location and the CARSmeasurement region, estimated to be �res � 2:0 ms. Characteristictime for V–T relaxation for N2-O2 is even longer. Carbon dioxideinjection, which results in significant reduction of nitrogenvibrational temperature, even at low CO2 mole fractions in the flow,illustrates the other extreme. The room temperature rate coefficientfor V–Venergy transfer

Fig. 15 Vibrational temperature vs estimated reduced electric field in

the sustainer discharge (1 Td� 10�17 V � cm2).

Fig. 16 Photograph of a pulser–sustainer discharge in nitrogen.

P0 � 350 torr, �� 100 kHz, VPS � 2:0 kV, flow into the page.

Fig. 17 Percentage of dc sustainer discharge energy loaded into

nitrogen molecular vibrations vs estimated reduced electric field

(1 Td� 10�17 V � cm2).

MONTELLO ETAL. 1373

N 2�v� 1� � CO2�000� ! N2�v� 0� � CO2�001�

is very rapid, kVV � 6 � 10�13 cm3=s [26] (�VV � 1=kVVnCO210 �sfor CO2 partial pressure of only 5 torr). CO2 vibrationally excited incollisions with nitrogen relaxes via rapid intramolecular V–Venergytransfers to �2 (010) and �1 (100) modes, with subsequent rapid V–Trelaxation. Indeed, the present experimental results demonstrateTv�N2� reduction to near threshold of the CARS diagnostic detectionwith only a few torr of CO2 injected into the flow.

Nitric oxide and hydrogen represent two intermediate cases. Therate coefficient of V–Venergy transfer from nitrogen to nitric oxide

N 2�v� 1� � NO�v� 0� ! N2�v� 0� � NO�v� 1�

is kVV � 1:4 � 10�15 cm3=s at T � 300 K and kVV�2:7 � 10�15 cm3=s at 400 K [27] (�VV � 1=kVVnNO � 4 ms and�VV � 3 ms for NO partial pressure of 5 torr, which is comparablewith theflow residence time). The extrapolated value atT � 500 K iskVV � 3:2 � 10�15 cm3=s [27] (�VV � 3 ms). The room temperatureV–T relaxation rate coefficient for N2-H2, N2�v� 1� � H2 !N2�v� 0� � H2, is kVT � 1:4 � 10�16 cm3=s [28] (�VT � 1=kVTnH2�40 ms for H2 partial pressure of 5 torr). Using the temper-ature dependence suggested in [29], this rate increases to kVT �3:9 � 10�16 cm3=s at T � 400 K��VT � 20 ms� and kVT�9:2 � 10�16 cm3=s at T � 500 K��VT � 10 ms�. Note that, similarto N2-O2, V–Venergy transfer for N2-H2 is extremely slow, due to alarge difference in vibrational quanta [30].

The vibrational temperatures Tv from different relaxant injectionpartial pressuresPadd can be used to infer theV–V relaxation rate kVV(for the NO injection, the V–T rate is calculated in the same mannerfor the case of H2 injection) via the following equation:

dEvdt� Ev��res� � Ev�0�

�res� PaddkBT

kVV�1 � e��v=T�Eeq�T� � Ev�0�

(6)

where dEv=dt is the time derivative of vibrational energy. Thevibrational energies of nitrogen at the CARS measurement locationwith and without relaxant gas injection Ev��res� and Ev�0�,respectively, are defined by

Ev ��v

e�v=Tv � 1 (7)

and the vibrational energy content due to purely thermal effectsEeq�T� is defined as

Eeq�T� ��v

e�v=T � 1 (8)

where T � 400 K is the translational/rotational gas temperature atthis condition. The transit time from the injector to the measurementlocation was estimated simply from the mass flow rate and thechannel cross section to be �res � 2:0 ms, and kB is the Boltzmann’sconstant. The results of this analysis are shown in Table 1, whichindicates an apparent value of kVV�N2 � NO�, inferred in the presentwork fairly close to the literature value [27] at T � 400 K, within thebounds of the uncertainty in this estimation. The apparent ratekVT�N2-H2� inferred here is more than a factor of two faster than theliterature value at T � 400 K [28,29]. Note, however, thatvibrational relaxation of nitrogen from Tv�N2� � 1700 K toTv�N2� � 1200 K in the presence of NO andH2 (see Fig. 18) resultsin reduction of the average vibrational energy per molecule by�Ev � 28 meV=molecule, with the resultant flow temperature riseby approximately �T � 90 K. At T � 500 K, the literature valuesare kVV�N2 � NO� � 3:2 � 10�15 cm3=s [27] and kVT�N2 � H2� �9:2 � 10�16 cm3=s [28,29], i.e., close to the rates inferred from thepresent measurements. More accurate measurements of these ratesusing the present approach would require simultaneous measure-ment of the flow rotational–translational temperature. We arecurrently implementing a higher spectral resolution instrument thatwill enable such measurements in future work.

While it is not shown in Fig. 18, injection of nonexcited N2 wasalso performed but showed little to no effect on the vibrationaltemperature. Note that, while the room temperature V–V energytransfer rate coefficient for nitrogen

N 2�v� 1� � N2�v� 0� ! N2�v� 0� � N2�v� 1�

is relatively high, kVV � 1:5 � 10�14 cm3=s [31] (�VV � 1=kVVnN�0:5 ms for injection N2 partial pressure of 5 torr), the resonantenergy transfer process simply results in redistribution of N2vibrational energy among larger number of molecules, with energystill being locked in the nitrogenvibrationalmode. Furthermore, witha constant mixture total pressure, the addition of vibrationally coldN2 beyond the discharge should have the effect of slowing the mainflow slightly as it traverses the discharge, allowing for somewhathigher specific energy loading in the main flow, with this additionalenergy then transferred into the vibrational mode of the coldmolecules but no loss of vibrational quanta.

The measurements summarized in Fig. 18 demonstrate feasibilityof generating low-temperature high-pressure vibrationally excitedflows of nitrogen and synthetic air at steady state, as well asfeasibility of tailoring the vibrational temperature of thesenonequilibrium flows using injection of efficient V–T and V–Vrelaxer species downstream of the discharge.

IV. Conclusions

This paper has described the implementation of picosecondCARSspectroscopy for measurements of the VDF in a nitrogen flowvibrationally excited in a high-pressure electric discharge in theplenum of a nonequilibrium hypersonic wind tunnel. The high-pressure discharge used for vibrational excitation of the flow is acombination of two fully overlapping transverse discharges, arepetitive nanosecond pulse discharge producing volume ionization,and a dc discharge used for energy loading. The main objective ofloading energy into vibrational and electronic energy modes of theflow, with its subsequent control by adding efficient relaxer speciesdownstream of the discharge, is to study the effect of internal energydisequilibrium on the hypersonic flowfield.

Fig. 18 Effects of gas injection on Tv�N2�, VPS � 5 kV, 300 torr totalmixture pressure.

Table 1 Estimated nitrogen vibration relaxation rate coefficients compared with literature values

Energy transfer mechanism Rate from literature at T � 400 K Rate from this workN2�v� 1� � NO�v� 0� ! N2�v� 0� � NO�v� 1� kVV � 2:7 � 10�15 cm3=s [27] kVV � 3:1 � 10�15 cm3=sN2�v� 1� � H2 ! N2�v� 0� � H2 kVT � 3:9 � 10�16 cm3=s [28,29] kVT � 1:1 � 10�15 cm3=s

1374 MONTELLO ETAL.

The present work has demonstrated that the CARS system, using apicosecond Nd:YAG laser and a picosecond modeless broadbanddye laser, generates sufficient peak power to acquire single-shotspectra with high signal to noise. The single-shot CARS spectra areused to infer instantaneous vibrational level populations of nitrogenN2�v� and the first-level vibrational temperature Tv�N2�.

The present results demonstrate the feasibility of sustaining highvibrational temperatures in a pulser–sustainer discharge in nitrogen,up to Tv�N2� � 2000 K, at nearly half the atmospheric pressure andat steady state. Previous results using nitrogen UV/visible emissionspectroscopy have indicated low translational–rotational temper-atures at these conditions, T � 350–400 K. This demonstrateshighly nonequilibrium conditions created by the discharge in thenozzle plenum. Time-resolved vibrational level populations ofnitrogen N2�v� 0–3� and vibrational temperature have beenmeasured using single-shot CARS spectra. The results show that thevibrational temperature in the pulser–sustainer discharge follows thedc discharge current, i.e., energy loading by the dc sustainerdischarge. In addition, the present results show that the repetitivenanosecond-pulse discharge operating alone produces vibrationalexcitation of nitrogen, with vibrational temperatures of up toTv�N2� � 1100 K.

As expected, vibrational temperature increases with the dcvoltage, i.e., at higher dc discharge energy loadings. Analysis ofexperiments conducted at plenum (discharge) pressures of P0 �300 torr andP0 � 370 torr have also shown that nitrogenvibrationaltemperature in the discharge increases with the reduced electric fieldE=n. At E=n� 13 Td, vibrational temperatures near Tv�N2� �2000 K are achieved at both discharge pressures. At a lower pressure,P0 � 200 torr, vibrational temperature levels off when the reducedelectric field is increased, which may be due to discharge non-uniformity development at these conditions. However, additionalwork is needed to determine the cause of this deviation.

Nitrogen vibrational temperature has been measured whiledifferent relaxer species were injected into the flow between thedischarge section and the CARS measurement location, with theseresults plotted vs partial pressure of injection flow. The resultsdemonstrate that injection of oxygen, up to 17%mole fraction in theflow, does not result in detectable change of nitrogen vibrationaltemperature, indicating extremely slowV–Venergy transfer fromN2toO2, aswell as slowN2-O2 V–Trelaxation at the present conditions.On the other hand, injection of even small amounts ofCO2 (less than1% mole fraction) results in dramatic N2 vibrational temperaturereduction, due to rapid V–V energy transfer from nitrogen to theasymmetric stretch vibrational mode of CO2, with subsequent V–Trelaxation of CO2. Injection of NO or H2 represents an intermediatecasewhen nitrogen vibrational temperature can be reduced graduallyby varying the gas injection amount. N2-NO V–V and N2-H2 V–Trates inferred from these data agree well with previous literatureresults to within the uncertainty in rotational–translationaltemperature.

The present results demonstrate the feasibility of generating low-temperature high-pressure vibrationally excited flows of nitrogenand synthetic air at steady state, as well as the feasibility of tailoringthe vibrational temperature of these nonequilibrium flows usinginjection of efficient V–TandV–V relaxer species downstream of thedischarge. Finally, the use of a picosecond CARS system, as anefficient diagnostic for spatially and temporally resolved vibrationallevel populations in high-pressure discharges, from single-shotCARS spectra, has been demonstrated.

Acknowledgments

This research is supported by the Department of Energy Office ofFusion Energy Science, contract DE-SC0001939, and the U.S. AirForce Office of Scientific Research Hypersonics program (TechnicalManager: John Schmisseur). The authors would like to thank SukeshRoy for much help and many useful discussions regardingfabrication of the picosecond broadband dye laser as well aspicosecond coherent anti-Stokes Raman scattering in general.

References

[1] Leyva, I. A., Laurence, S., War-Kei Beierholm, A., Hornung, H. G.,Wagnild, R., and Candler, G., “Transition Delay in HypervelocityBoundary Layers by Means of CO2=Acoustic Instability Interactions,”AIAA Paper 2009-1287, 2009.

[2] Wagnild, R., Candler, G. V., Leyva, I. A., Jewell, J. S., and Hornung, H.G., “Carbon Dioxide Injection for Hypervelocity Boundary LayerStability,” AIAA Paper 2010-1244, 2010.

[3] Nishihara, M., Takashima, K., Rich, J. W., and Adamovich, I. V.,“Mach 5 BowShock Control by a Nanosecond Pulse Surface DielectricBarrier Discharge,” Physics of Fluids, Vol. 23, 2011, pp. 66–101.doi:10.1063/1.3599697

[4] Nishihara,M., Takashima, K., Jiang, N., Lempert,W.R., Adamovich, I.V., and Rich, J. W., “Development of a Mach 5 Nonequilibrium WindTunnel,” AIAA Paper 2010-1567, Jan. 2010.

[5] Nishihara,M., Takashima, K., Jiang, N., Lempert,W.R., Adamovich, I.V., and Rich, J.W., “Nonequilibrium FlowCharacterization in aMach 5Wind Tunnel,” AIAA Paper 2010-4515, June–July 2010.

[6] Rich, J. W., Bergman, R. C., and Lordi, J. A., “Electrically Excited,Supersonic Flow Carbon Monoxide Laser,” AIAA Journal, Vol. 13,1975, pp. 95–101.doi:10.2514/3.49636

[7] Bruzzese, J. R., Hicks, A., Erofeev, A., Cole, A. C., Nishihara, M., andAdamovich, I. V., “Gain and Output Power Measurements in anElectrically Excited Oxygen-Iodine Laser with a Scaled Discharge,”Journal of Physics D: Applied Physics, Vol. 43, 2010, pp. 15–201.doi:10.1088/0022-3727/43/1/015201

[8] Chiroux de Gavelle de Roany, A., Flament, C., Rich, J. W.,Subramaniam, V. V., and Warren, W. R., “Strong VibrationalNonequilibrium in Supersonic Nozzle Flows,”AIAA Journal, Vol. 31,No. 1, 1993, pp. 119–128.doi:10.2514/3.49006

[9] Babu, V., and Subramaniam, V. V., “Numerical Solutions to NozzleFlows with Vibrational Nonequilibrium,” Journal of Thermophysicsand Heat Transfer, Vol. 9, No. 2, 1995, pp. 227–232.doi:10.2514/3.650

[10] Adamovich, I. V., Macheret, S. O., Rich, J. W., and Treanor, C. E.,“Vibrational Relaxation and Dissociation Behind Strong ShockWaves:Kinetic, I. RateModels,”AIAA Journal, Vol. 33, No. 6, 1995, pp. 1064–1069.doi:10.2514/3.12528

[11] Adamovich, I. V., Macheret, S. O., Rich, J. W., and Treanor, C. E.,“Vibrational Energy Transfer Rates Using a Forced HarmonicOscillator Model,” Journal of Thermophysics and Heat Transfer,Vol. 12, No. 1, 1998, pp. 57–65.doi:10.2514/2.6302

[12] Adamovich, I. V., Macheret, S. O., Rich, J. W., and Treanor, C. E.,“Vibrational Relaxation and Dissociation Behind Strong ShockWaves,II. Master Equation Modeling,” AIAA Journal, Vol. 33, No. 6, 1995,pp. 1070–1075.doi:10.2514/3.48339

[13] Candler, G. V., Olejniczak, J., and Harrold, B., “Detailed Simulation ofNitrogenDissociation in StagnationRegions,”Physics of Fluids, Vol. 9,1997, pp. 2108–2117.doi:10.1063/1.869330

[14] Roy, S., Gord, J. R., and Patnaik, A. K., “Recent Advances in CoherentAnti-Stokes Raman Scattering Spectroscopy: Fundamental Develop-ments and Applications in Reacting Flows,” Progress in Energy andCombustion Science, Vol. 36, 2010, pp. 280–306.doi:10.1016/j.pecs.2009.11.001

[15] Raizer, Y. P., Gas Discharge Physics, Springer-Verlag, Berlin, 1991,p. 100.

[16] Nishihara, M., Jiang, N., Rich, J.W., Lempert, W. R., Adamovich, I. V.,and Gogineni, S., “Low-Temperature Supersonic Boundary LayerControl Using Repetitively Pulsed MHD Forcing,” Physics of Fluids,Vol. 17, No. 10, 2005, Paper 106102.doi:10.1063/1.2084227

[17] Nishihara, M., Rich, J. W., Lempert, W. R., Adamovich, I. V., andGogineni, S., “Low-TemperatureM � 3 Flow Deceleration by LorentzForce,” Physics of Fluids, Vol. 18, No. 8, 2006, pp. 86–101.doi:10.1063/1.2265011

[18] Gordiets, B. F., Osipov, V. A., and Shelepin, L. A.,Kinetic Processes inGases and Molecular Lasers, Gordon and Breach, London, 1988,p. 53ff.

[19] Eckbreth, A. C., Laser Diagnostics for Combustion Temperature andSpecies, Gordon and Breach, Amsterdam, 1996, pp. 281–365.

[20] Kulatilaka, W. D., Hsu, P. S., Stauffer, H. U., Gord, J. R., and Roy, S.,“Direct Measurement of Rotationally Resolved H2 Q-Branch Raman

MONTELLO ETAL. 1375

http://dx.doi.org/10.1063/1.3599697http://dx.doi.org/10.2514/3.49636http://dx.doi.org/10.1088/0022-3727/43/1/015201http://dx.doi.org/10.2514/3.49006http://dx.doi.org/10.2514/3.650http://dx.doi.org/10.2514/3.12528http://dx.doi.org/10.2514/2.6302http://dx.doi.org/10.2514/3.48339http://dx.doi.org/10.1063/1.869330http://dx.doi.org/10.1016/j.pecs.2009.11.001http://dx.doi.org/10.1063/1.2084227http://dx.doi.org/10.1063/1.2265011

Coherence Lifetimes Using Time-Resolved Picosecond Coherent Anti-Stokes Raman Scattering,” Applied Physics Letters, Vol. 97, 2010,Paper 081112.doi:10.1063/1.3483871

[21] Hsu, P. S., Patnaik, A. K., Gord, J. R., Meyer, T. R., Kulatilaka, W. D.,and Roy, S., “Investigation of Optical Fibers for Coherent Anti-StokesRaman Scattering (CARS) Spectroscopy in Reacting Flows,”Experiments in Fluids, Vol. 49, 2010, pp. 969–984.doi:10.1007/s00348-010-0961-6

[22] Roy, S., Meyer, T. R., and Gord, J. R., “Broadband Coherent Anti-StokesRamanScattering Spectroscopy ofNitrogenUsing a PicosecondModeless Dye Laser,” Optics Letters, Vol. 30, 2005, Paper 3222.doi:10.1364/OL.30.003222

[23] Roy, S., Meyer, T. R., and Gord, J. R., “Time-Resolved Dynamics ofResonant and Nonresonant Broadband Picosecond Coherent Anti-Stokes Raman Scattering Signals,” Applied Physics Letters, Vol. 87,2005, Paper 264103.doi:10.1063/1.2159576

[24] Lempert, W. R., “Diagnostics,” Non-Equilibrium Air Plasmas atAtmospheric Pressure, edited by K., Becker, U., Kogelschatz, K. H.,Schoenbach, and R., Barker, IOP Publ., Bristol, England, U.K., 2005,p. 456.

[25] Green, S. M., Rubas, P. J., Paul, M. A., Peters, J. E., and Lucht, R. P.,“Annular Phase-Matched Dual-Pump Coherent Anti-Stokes RamanSpectroscopy System for the Simultaneous Detection of Nitrogen andMethane,” Applied Optics, Vol. 37, 1998, pp. 1690–1701.doi:10.1364/AO.37.001690

[26] Taylor, R. L., andBitterman, S., “Survey ofVibrational RelaxationDatafor Processes Important in the CO2-N2 Laser System,” Reviews ofModern Physics, Vol. 41, 1969, pp. 26–47.doi:10.1103/RevModPhys.41.26

[27] Doyennette, L., and Margottin-Maclou, M., “Vibrational Relaxation ofNO�v� 1� by NO, N2, CO, HCl, CO2, and N2O from 300 to 600 K,”Journal of Chemical Physics, Vol. 84, No. 12, 1986, pp. 6668–6678.doi:10.1063/1.450720

[28] Allen, D. C., Chandler, E. T., Gregory, E. A., Siddles, R. M., andSimpson, C. J. S. M., “Vibrational Deactivation of N2�v� 1� by n-H2and by p-H2,” Chemical Physics Letters, Vol. 76, 1980, pp. 347–353.doi:10.1016/0009-2614(80)87039-4

[29] Capitelli, M., Ferreira, C. M., Gordiets, B. F., and Osipov, A. I., PlasmaKinetics in Atmospheric Gases, Springer-Verlag, Berlin, 2001, p. 107,chap. 7.doi:10.1088/0741-3335/43/3/702

[30] Bott, J. F., “Vibrational Energy Exchange Between H2�v� 1� and D2,N2, HCl, and CO2,” Journal of Chemical Physics, Vol. 65, 1976,pp. 3921–3928.doi:10.1063/1.432884

[31] Ahn, T., Adamovich, I. V., and Lempert, W. R., “Determination ofNitrogen V–V Transfer Rates by Stimulated Raman Pumping,”Chemical Physics, Vol. 298, 2004, pp. 233–240.doi:10.1016/j.chemphys.2003.11.029

R. LuchtAssociate Editor

1376 MONTELLO ETAL.

http://dx.doi.org/10.1063/1.3483871http://dx.doi.org/10.1007/s00348-010-0961-6http://dx.doi.org/10.1364/OL.30.003222http://dx.doi.org/10.1063/1.2159576http://dx.doi.org/10.1364/AO.37.001690http://dx.doi.org/10.1103/RevModPhys.41.26http://dx.doi.org/10.1063/1.450720http://dx.doi.org/10.1016/0009-2614(80)87039-4http://dx.doi.org/10.1088/0741-3335/43/3/702http://dx.doi.org/10.1063/1.432884http://dx.doi.org/10.1016/j.chemphys.2003.11.029