RESEARCH ARTICLE

Picosecond CARS measurements of nitrogen rotational/translational and vibrational temperature in a nonequilibriumMach 5 flow

A. Montello • M. Nishihara • J. W. Rich •

I. V. Adamovich • W. R. Lempert

Received: 12 June 2012 / Revised: 20 November 2012 / Accepted: 7 December 2012

� Springer-Verlag Berlin Heidelberg 2012

Abstract Picosecond Unstable-resonator Spatially Enhanced

Detection Coherent Anti-Stokes Raman Scattering (USED-

CARS) is used for the measurement of nitrogen Q-branch

(DJ = 0) spectra in the subsonic plenum and supersonic

flow of a highly nonequilibrium Mach 5 wind tunnel.

Spectra are processed to infer rotational/translational (Trot)

and first-level vibrational (Tvib) temperatures in the

200–370 torr plenum simultaneously. Operation of the

nominally high reduced electric field (E/npeak * 500 Td),

nsec pulsed discharge alone results in fairly significant

vibrational loading, Tvib * 720 K/Trot * 380 K; addition

of an orthogonal low E/n (*10 Td) DC sustainer discharge

produces substantial vibrational loading, Tvib * 2,000 K/

Trot * 450 K. Effects of injection of CO2, NO, and H2

downstream of the pulser–sustainer discharge are examined,

which result in vibrational relaxation accompanied

by simultaneous gas heating, Tvib * 800–1,000 K/Trot *600 K. CARSk measurements within very low-density

flows in the Mach 5 expansion nozzle are also performed,

with Tvib measured in both the supersonic free-stream and

downstream of a bow shock created by a 5-mm-diameter

cylindrical test object in the Mach 5 flow. Measurements

within 300 lm of the cylinder leading edge show that for

pure N2, or N2 with 0.25 torr CO2 injection, no vibrational

relaxation is observed behind the bow shock.

1 Introduction

The ability to tailor nonequilibrium hypersonic flows by

control of the loading of internal degrees of freedom

(vibrational and electronic states), as well as by control of

dissociation and ionization fraction, is a topic of much

current interest. For example, turbulent transition delay in a

Mach 5 flow over a 5� cone by means of injection of carbon

dioxide into nitrogen or air flow has been recently dem-

onstrated (Leyva et al. 2009). Kinetic modeling calcula-

tions (Wagnild et al. 2010) suggest that transition delay is

caused by absorption of acoustic perturbations in the

boundary layer by vibrational energy modes of CO2, which

may also result in CO2 dissociation. Another example is

relaxation of energy stored in internal degrees of freedom

of molecules behind a bow shock, which may significantly

increase shock stand-off distance. Additionally, the pres-

ence of vibrationally and electronically excited species in a

hypersonic flow may strongly affect the emission signature

from the shock layer. Finally, short-pulse electric dis-

charges efficiently loading electronic energy levels of

nitrogen and oxygen in air are currently being explored as a

means of hypersonic flow control, by producing repetitive

localized pressure perturbations in the flow (Nishihara et al.

2011).

Since the cost of obtaining full-scale hypersonic flight

test data or operating large-scale ground test facilities is

extremely high, alternative methods of data production are

necessary. This provides a significant incentive for the

development and use of small-scale test facilities which are

capable of recreating environments seen in the flow con-

ditions of interest, and which lend themselves to the

development of optical diagnostics of nonequilibrium

hypersonic flows. Our approach to generating a nonequi-

librium hypersonic flow is to use a high-pressure, low-

A. Montello � M. Nishihara � J. W. Rich �I. V. Adamovich � W. R. Lempert (&)

Michael A. Chaszeyka Nonequilibrium Thermodynamics

Laboratories, Department of Mechanical and Aerospace

Engineering, The Ohio State University,

Columbus, OH 43210, USA

e-mail: Lempert.1@osu.edu

123

Exp Fluids (2013) 54:1422

DOI 10.1007/s00348-012-1422-1

temperature, diffuse electric discharge sustained in the

plenum of a small-scale Mach 5 wind tunnel to load

internal energy modes of nitrogen and oxygen molecules

(Nishihara et al. 2012).

Similar to our previous work (Nishihara et al. 2012),

target parameters for nonequilibrium gas flow include run

time for steady-state conditions of 5–10 s, with plenum

pressures in the range of P0 * 200–370 torr. As our pre-

vious results have shown, the pulser–sustainer electric

discharge present in the wind tunnel plenum is capable of

producing highly nonequilibrium flow conditions, with

translational/rotational temperature inferred by nitrogen

UV emission spectroscopy near Trot * 350–400 K, and

vibrational temperature up to Tvib * 2,000 K (Montello

et al. 2012). Additionally, as these results demonstrated,

the flow internal energy mode nonequilibrium can be fur-

ther controlled by injection of ‘‘rapid relaxer’’ species

(carbon dioxide, nitric oxide, or hydrogen) into the sub-

sonic nonequilibrium flow, after the discharge section but

upstream of the nozzle throat. Qualitatively, the results of

the study demonstrated effects similar to what vibrational

energy transfer rates predict; however, these rates are

highly sensitive to the gas (rotational/translational) tem-

perature. To compare experimental vibrational energy

relaxation effects with energy exchange rate predictions,

simultaneous direct measurement of the gas temperature is

necessary.

This paper provides new experimental data character-

izing both the vibrational energy loading and gas heating of

molecular nitrogen in a nonequilibrium flow excited by a

non-self-sustained electric discharge in the plenum of a

Mach 5 wind tunnel. Volume ionization, at plenum pres-

sures in the range of P0 = 200–370 torr, is accomplished

using a high peak reduced electric field (E/n)peak *500 Td, where (E/n)peak is the peak reduced electric field

inferred from the estimated peak electric field, 5 nsec

duration pulse discharge, operating at a pulse repetition

rate of 100 kHz. An orthogonal DC sustainer discharge

(E/n * 10–30 Td), which accounts for approximately

80 % of the total power loading into the flow, efficiently

excites the N2 vibrational energy mode. Theoretical pre-

dictions suggest that as much as 80–90 % of the DC sus-

tainer input power goes into nitrogen vibrational excitation

by electron impact at these reduced electric field values.

Temporally and spatially resolved ground- and first-level

vibrational populations of nitrogen, N2 (X1Rg, v = 0–1),

yielding first-level vibrational temperature in the pulser–

sustainer discharge, up to Tvib (N2) = 2,000 K, are mea-

sured using picosecond Coherent Anti-Stokes Raman

Scattering (CARS) spectroscopy (Eckbreth 1996; Roy et al.

2010).

Measurements using the nsec pulsed discharge alone

(i.e., without the DC sustainer) are also presented, the

results of which demonstrate that significant vibrational

loading occurs, Tvib * 720 K at 300 torr of pure N2, even

at the relatively high nominal reduced electric field,

E/npeak * 500 Td. Our quasi-one-dimensional nanosecond

pulse discharge model (Adamovich et al. 2009) is used to

compare predicted versus measured energy coupling to the

flow, as well as post-breakdown energy loading to nitrogen

vibrational states, with good agreement to the values

measured by CARS. Modifications to the CARS system

allow for the capture of high-resolution Q-branch spectra,

with partial resolution of the rotational structure. These are

fit with the Sandia CARSFT least squares fitting code

(Palmer 1989), yielding best fit rotational temperatures of

Trot * 380 K for the pulser operating alone and

Trot * 450 K for the pulser–sustainer discharge (sustainer

voltage VPS = 4.5 kV).

Additionally, the effect on nitrogen Tvib and Trot of

injection of CO2, NO, and H2, between the discharge

section and the CARS measurement location, which results

in partial vibrational relaxation of nitrogen accompanied by

gas heating, is also examined. This demonstrates quanti-

tatively the ability to control and tailor the vibrational

energy content of the flow prior to its expansion through

the Mach 5 nozzle of the wind tunnel. Conversely, addition

of oxygen or nitrogen (not excited in the discharge) to the

vibrationally excited nitrogen flow did not result in sig-

nificant vibrational temperature change, due to slow

vibration–vibration energy transfer from N2 to O2 for the

case of oxygen addition and extremely slow vibration-

translation relaxation at these temperatures for either O2 or

N2 injection.

Finally, also presented are the results of measurement of

nitrogen Q-branch CARS spectra captured within the

extremely low-density Mach 5 supersonic flow. Measure-

ments are made within the free-stream as well as at mul-

tiple locations behind the bow shock produced by a 5-mm-

diameter quartz cylinder mounted across the flow. As will

be discussed, the utilization of micrometer adjustment

translation stages along with pitch/catch optical arms

allows for measurement location adjustment in the vertical

and streamwise directions within the flow. CARS spectra

are captured as near as 300 lm from the cylinder leading

edge, for the nonequilibrium flow produced by the pulser–

sustainer discharge, for no relaxer injection, as well as the

slightly equilibrated case of 0.25 torr injection of CO2. In

both cases, the flow remains highly nonequilibrium, with

vibrational temperatures found similar to those measured in

the subsonic plenum for the same discharge conditions.

The integrated spectra are also used to estimate local

number density, leading to an estimated shock stand-off

distance measured to 1.0 mm, very close to the 1.2 mm

measured previously by Schlieren imaging and NO PLIF

(Nishihara et al. 2012).

Page 2 of 13 Exp Fluids (2013) 54:1422

123

2 Experimental

2.1 Mach 5 wind tunnel

The experiments conducted for this study were performed

in the Mach 5 nonequilibrium flow tunnel previously

developed and described in detail (Nishihara et al. 2012).

Briefly, the laboratory scale wind tunnel, a schematic of

which is given in Fig. 1, operates at plenum pressures from

0.25 to 1 atm with nitrogen or air supplied from high-

pressure cylinders. The tunnel, constructed from transpar-

ent acrylic plastic, is capable of producing steady-state

nonequilibrium supersonic flow by utilizing a high-pres-

sure, diffuse, nanosecond pulser/DC sustainer electric dis-

charge operating in the plenum section. The discharge can

be tailored, to load internal (vibrational and electronic)

modes of nitrogen, while maintaining low translational–

rotational temperatures. Previous measurements by nitro-

gen UV emission spectroscopy in a similar discharge

arrangement yielded Trot * 350–400 K (Nishihara et al.

2012). Downstream of the discharge section, the gas flows

through a choked flow injector with 20 injector holes 1 mm

in diameter, arranged in a single row in both the top and

bottom channel walls, also shown in Fig. 1. Gases inducing

vibrational relaxation of nitrogen, such as carbon dioxide,

nitric oxide, hydrogen, can be injected into the flow, which

allows further control of the energy distribution among the

internal molecular modes in the flow. After the injector, the

flow traverses the nozzle throat, with a height of 1.6 mm,

and passes through the aerodynamically contoured Mach 5

expansion section to the nozzle exit. There is a diffuser

section just upstream of the 8-inch-diameter exit pipe

leading to a 110-ft3 vacuum tank and 200-cfm pump.

Throughout the length of the flow tunnel, several quartz

windows are flush mounted in the side and top/bottom

walls, providing ample optical access to the flow.

The pulser–sustainer electric discharge, a cross-sectional

schematic of which is given in Fig. 2, is comprised of two

fully overlapping discharges, (a) a repetitively pulsed,

high-voltage nanosecond discharge and (b) a transverse DC

sustainer discharge. As shown in Fig. 2, the pulser elec-

trodes (3 cm 9 4 cm copper plates) are flush mounted in

the top and bottom walls of the discharge section (nozzle

plenum), covered by alumina ceramic dielectric plates

1/16 inch thick, with a gap of 0.5 cm in between. The

sustainer electrodes (4.0 cm 9 0.5 cm copper plates) are

placed along the side walls of the discharge section and are

removable. In the present experiments, the sustainer elec-

trodes are separated by 3.0 cm. Both the pulser and sus-

tainer electrodes are rounded along the edges to reduce the

electric field nonuniformity. The acrylic plastic walls of the

channel downstream of the discharge section are covered

with alumina ceramic plates, attached with silicone rubber

adhesive. These protect the plastic walls from overheating

in the possibility of instability development or arcing in the

discharge. The main benefit from using the pulser–sus-

tainer discharge is the ability to generate stable nonequi-

librium plasmas at high pressures and high discharge

energy loading.

The repetitive nanosecond pulse discharge operates

using a high peak voltage (up to 30 kV), short-pulse dura-

tion (5 nsec), high-pulse repetition rate (up to 100 kHz)

voltage waveform, produced by a high-voltage nanosecond

pulse generator (FID GmbH 60-100MC4). Figure 3 shows

typical pulse voltage and current traces (top), as well as

instantaneous power and coupled pulse energy (i.e., tem-

porally integrated power) traces (bottom), measured in a

repetitively pulsed discharge in nitrogen at P0 = 300 torr

and m = 100 kHz, with the DC electrodes present in the test

section, but not powered (Yin et al. 2012). The waveforms

shown are for a pulse generated 0.1 s after the beginning of

the pulse burst (i.e., for pulse #10,000). The pulser is

operated using an external trigger/function generator. Each

high-voltage pulse generates volume ionization in the dis-

charge section, and the high voltage is then turned off

before any ionization/heating instability (Raizer 1991) has

time to develop. The high frequency of pulse repetition

prevents the plasma from fully decaying before the next

ionizing pulse occurs, thus providing pulse-periodic, spa-

tially uniform ionization in the discharge section (Nishihara

et al. 2005, 2006).

Between the ionizing pulses, energy is coupled to the

flow by the DC discharge. The DC electrodes are con-

nected to a Glassman 5 kV, 2 A power supply, operated in

Model

Injector(NO, H2,CO2, etc.)

Pulserelectrodes

DC electrodes

Main flow(N2)

Nozzle throat Nozzle exit

Diffuser

To vacuum

Optical access(CARS measurement locations)

8 cm 9 cm 10 cm

Fig. 1 Schematic side view of

Mach 5 tunnel

Exp Fluids (2013) 54:1422 Page 3 of 13

123

a voltage stabilized mode, in series with a 1.5 kX ballast

resistor. The DC voltage is deliberately kept below

breakdown threshold, that is, typically below 4–5 kV, to

preclude the development of self-sustained DC discharge

(i.e., independent of pulsed ionization) in the high pressure

flow, which would result in instability development and

arcing. The power coupled to the flow by the DC discharge

is significantly higher than the repetitively pulsed discharge

power. Previously, this approach has been used in our work

to sustain high-power discharges in a Mach 3 MHD wind

tunnel (Nishihara et al. 2005, 2006), and in an electrically

excited gas dynamic oxygen–iodine laser (Bruzzese et al.

2010). In the present experiments, the repetitively pulsed

discharge is operated for 0.6 s, and the DC discharge for

0.55 s. The DC power supply is triggered by an externally

generated rectangular-shaped trigger pulse, the rising edge

of which also triggers the function generator that produces

a burst of 60,000 trigger pulses for the high-voltage

nanosecond pulse generator at m = 100 kHz pulse repeti-

tion rate. The estimated reduced electric fields in the two

discharges are significantly different, (E/n)peak * 500 Td

in the nanosecond pulsed discharge and E/n * 10 Td in

the DC discharge (1 Td = 10-17 V cm2). At these condi-

tions, a significant fraction of input power in the pulsed

discharge is spent on electronic excitation, dissociation,

and ionization of nitrogen, while nearly all input power in

the DC discharge (up to *80–90 % according to theory

(Raizer 1991)) is stored in the vibrational energy mode of

nitrogen, with little power going to translational/rotational

modes, that is, to heat. Due to a very long N2 vibrational

self-relaxation time at near room temperature, *1 atm sec

(Gordiets et al. 1988), this approach can create essentially

vibrationally frozen nitrogen and air flows in the super-

sonic test section, with vibrational temperature greatly

exceeding the translational/rotational mode temperature.

During the experiment, both the main flow through the

discharge and the injection flows are controlled using

solenoid valves. The main flow rate is calculated using a

1-D choked flow equation, based on the plenum pressure

and the nozzle throat area. The injection flow rate has been

both measured directly using a mass flow controller and

calculated from the choked flow equation, with both

methods giving similar results. At the baseline conditions,

nitrogen at P0 = 200–370 torr, the mass flow rate through

the tunnel is 3.9–7.2 g/sec and the steady-state run time at

constant static pressure in the supersonic test section is

5–10 s; runs can be repeated every few minutes.

2.2 Picosecond CARS diagnostic system

Picosecond CARS spectroscopy has been used for the

direct measurement of the ro-vibrational distribution

functions of nitrogen. CARS is a four-wave mixing spec-

troscopic technique which has been used extensively for

thermometry and species concentration measurements of

combustion and other gas phase reacting and nonreacting

flows (Eckbreth 1996; Roy et al. 2010). Simply stated,

CARS involves the interaction of three input photons,

termed pump, Stokes, and probe, with a molecule, resulting

in the generation of a signal photon. If the pump/Stokes

photons have a frequency difference corresponding to an

internal degree of freedom (generally rotational or vibra-

tional), a strong coherent (laser-like), ‘‘resonant’’ signal is

produced in a direction determined by what is known as the

‘‘phase-matching’’ criteria. All data presented in this work

used vibrational Q-branch (DJ = 0) transitions.

The CARS signal intensity, ICARS, has a linear depen-

dence on the Stokes laser intensity, IStokes, while scaling

quadratically with the magnitude of the CARS suscepti-

bility (proportional to number density, vCARSj j / N), that

is, ICARS / IStokes � N2. These effects are accounted for in

the data processing. Normalizing the CARS intensity by a

nonresonant background (NRB) spectrum captured in

argon accounts for the variation in the intensity profile of

the Stokes laser. It was experimentally determined that the

average NRB profile was very stable over the course of a

⊗Flow into the page

Pulsed electrodes

Sustainer DC electrodes (removable)

Alumina ceramic plates

0.5 cm3.0 cm

Fig. 2 Schematic diagram of pulser–sustainer discharge section

0 50 100 150 200 250 300 350

-10

0

10

20

Vo

ltag

e [k

V]

-40

0

40

80

Cu

rren

t [A

]

0 50 100 150 200 250 300 350

-0.50

0.51

1.5

Time [nsec]

Po

wer

[M

W]

01.534.56

En

erg

y [m

J]

Fig. 3 Nsec pulse voltage, current (upper), power, and coupled pulse

energy (lower) measurements. Nitrogen, P0 = 300 torr, m = 100 kHz

Page 4 of 13 Exp Fluids (2013) 54:1422

123

day, so one 200-shot average NRB spectrum was used to

normalize the data captured each day. After the data nor-

malization, the square root of the intensity is taken to

account for the quadratic dependence on number density;

all plots of spectral data below, are of the square root of

normalized intensity. For measurement of Tv, the CARS

cross section must also be accounted for, which for low

vibrational levels of nitrogen scales approximately as

(v’’ ? 1)2 (Lempert 2005) Since the square root of the data

has already been taken, dividing the integral of each peak

by (v’’ ? 1) yields the relative vibrational population in

each level.

For this work, the Unstable-resonator Spatially Enhanced

Detection (USED) phase-matching geometry (Marko and

Rimai 1979; Klick et al. 1981) has been employed. For this

geometry, the single 532 nm pump/probe beam is enlarged

with a telescope and the center portion of the beam is

removed, creating an annulus. This beam is then combined

coaxially with the Stokes beam by a dichroic mirror,

similar to a collinear alignment except that there is no

spatial overlap because the Stokes beam occupies the void

created within the pump/probe beam. After the beams pass

the focusing lens, the two regions of the annular pattern

remain spatially separated until they arrive very near to the

focal region; it is within this small overlap volume that

CARS signal generation occurs. According to the phase-

matching criteria, the generated CARS signal appears as a

ring outside of the pump/probe beam. This phase-matching

scheme is primarily chosen due to concerns about potential

beam steering that would occur for measurements per-

formed in the vicinity of the characteristic bow shock

produced by test objects in the Mach 5 section of the flow,

as well as a desire to perform measurements as close as

possible to the test object surface. There is a small sacrifice

in spatial resolution, relative to the more common folded

BoxCARS geometry, but this is not significant for the

measurements reported here. The transverse spatial

resolution for our system is on the order of 50 lm; a

measurement of the longitudinal resolution, performed by

scanning a glass flat in the vicinity of the CARS

measurement volume and observing the nonresonant

background signal, indicates that [95 % of the signal

generated comes within a 3 mm length at the beam focus,

as compared to *1 mm or less for BoxCARS, and 1–2 cm

for collinear CARS (determined in the same manner).

The broadband Stokes beam is generated with a home-

built dye laser, patterned after Roy et al. (2005), which

employs side-pumped oscillator and pre-amplifier cells,

followed by an end-pumped final amplifier cell (note there

is no output coupler). The combination of a half-wave plate

and thin film polarizer allows for adjustment of the ratio

between energy pumping the dye versus the energy in the

CARS pump/probe beam. Dye laser energy efficiency is as

high as 10 %. For this work, mixtures of Rhodamine 640

(R640) and Kiton Red 620 (KR620) were used. The

oscillator/pre-amplifier cells used a mixture of 34 mg R640

and 18 mg KR620 in 750 mL of methanol, while the

amplifier cell solution was 20 mg R640 and 18 mg KR620

in 750 mL methanol. The dye laser output is centered near

604 nm, with a full width at half maximum (FWHM) of

approximately 5–6 nm (Montello et al. 2012).

The dye laser is pumped by an Ekspla SL-333 Nd:YAG

laser, as can be seen in Fig. 4, which shows a schematic of

the CARS experimental arrangement. The nearly transform

limited YAG outputs pulses of approximately 150 pico-

seconds in duration, with variable energy output of up to

120 mJ/pulse at 532 nm. The choice to use a picosecond

(psec) system is motivated by several reasons, including

enabling time-resolved measurements at nsec and sub-nsec

time scales, as well as lowering necessary pulse energies,

reducing the risk of window damage. After the dichroic

mirror, which creates the 532/607 nm annulus, the beams

are focused into the test section with a 250-mm focal length

lens. After the focal point, located in the center of the wind

tunnel, the beams are re-collimated using a 100-mm focal

length lens.

After the collimating lens, a series of long-wavelength

passing dichroic mirrors reflect the *473 nm CARS beam

while dumping the pump/probe and Stokes beams. Finally,

the CARS signal passes through a short-pass filter, with a

cutoff wavelength of 503 nm, and is focused by a 100-mm

lens onto the slit of a 0.75 m And or Shamrock 750 spec-

trometer. At the exit plane of the spectrometer, a relay lens

magnification system, comprised of two Nikon F-mount

lenses, a 35-mm lens attached to the spectrometer and an

80- to 200-mm telephoto lens attached to the camera, pro-

vides variable spectral resolution magnification. For the

0.75 m Spectrometer

Broadband Dye Laser

Short-Pass Filter100mm Lens

Beam Dump

Dichroic Mirrors

Nd:YAG

Camera

Relay Lens Magnification

System

Flo

w

Wind Tunnel

Fig. 4 Experimental schematic diagram

Exp Fluids (2013) 54:1422 Page 5 of 13

123

work presented, the lens was set to give a *2.39 magni-

fication, which resulted in a spectral resolution of

*0.4 cm-1 when used with a 3,600 lpmm grating. This is

sufficient to partially resolve the rotational structure in room

temperature Q-branch spectra of nitrogen, as will be dem-

onstrated in the next section. The camera used is an Andor

Newton EM-CCD; the 1,600 by 400 pixel sensor array is

cooled to -90 �C, with the EM gain set to 150. The camera

and spectrometer are interfaced to a laboratory computer for

data recording. The entire picosecond CARS system is

placed on a custom-built cart, allowing the entire setup to be

easily transported between experimental facilities.

2.3 Mach 5 flow measurement setup

In the supersonic nozzle exit, as previously described, a

5-mm-diameter quartz cylinder model is mounted across

the hypersonic flow, which creates a bow shock, as

depicted in the schematic in Fig. 5a. The position of the

model relative to the entire tunnel can be seen in Fig. 1

above. For this work, a 7.5-mm-long section of 5-mm-

diameter quartz tube was secured to the leading edge of a

2-mm-diameter stainless steel rod, with the rod mounted to

the tunnel walls, as shown in Fig. 5b. The use of the short

rod section allows for closer access by the laser beams,

which propagate in the same direction as the rod axis, than

a full-length 5-mm-diameter rod would allow. Previous

Schlieren imaging (Nishihara et al. 2012) has shown that

the bow shock stand-off distance for this arrangement,

*1.2 mm for a plenum pressure of 300 torr, is uniform in

the spanwise direction over a distance of *6 mm, nearly

double the longitudinal spatial resolution of the system,

which as previously mentioned is *3 mm.

An experimental arrangement using optical pitch/catch

arms, along with two manual translation stages, allows for

simple adjustment of the CARS measurement location. A

3 inch by 18 inch optical breadboard section, which holds

the necessary turning prisms (for the vertical pitch/catch)

and CARS focusing and collimating lenses, is attached to

the vertical adjustment stage by a right-angle adapter plate.

The base of this stage is mounted to a 6 inch by 18 inch

breadboard, which also holds a second set of turning

prisms, comprising the horizontal pitch/catch. This larger

breadboard is mounted onto a horizontally positioned

translation stage, providing measurement location adjust-

ment down the length of the tunnel. Both stages are fitted

with fine adjustment micrometers, providing spatial reso-

lution of 10 lm per division in each dimension.

3 Results and discussions

3.1 Pure nitrogen flow, no discharge

This work focuses on both translational/rotational temper-

ature, Trot, as well as the ‘‘first-level’’ vibrational temper-

ature, Tvib, defined as:

Tvib ¼hvib

ln½n0=n1�ð1Þ

where hvib = 3,353 K is the energy difference between

vibrational levels v = 0 and v = 1 in temperature units. A

sample single-shot spectrum of 370 torr pure N2 flowing

through the tunnel plenum with no discharge is shown

~1.1 mm

5 mm

Bow Shock

(a) (b)

Flow

7.5 mm

Fig. 5 Schematic of a bow shock created by a quartz cylinder present

in supersonic flow, along with b a perspective view highlighting the

2-mm stainless steel support rod

2320 2324 2328 23320

20

40

60

80

100

-1]

Sq

rt(I

nt.

) [a

.u.]

2320 2324 2328 2332

Raman Shift [cm Raman Shift [cm-1]

DataCarsfit

(a) (b)

Fig. 6 Pure N2 CARS spectra

at 370 torr with no discharge,

a single-shot and b 20-shot

average, along with Carsfit best

synthetic spectra; Tfit = 322 K

Page 6 of 13 Exp Fluids (2013) 54:1422

123

below in Fig. 6a. While under low resolution the spectral

output of the Stokes beam from the modeless dye laser

appears very similar from one shot to the next, higher

resolution reveals significant shot-to-shot spectral profile

variation. To partially mitigate this, 20 single-shot spectra

are averaged together, an example of which can be seen

below in Fig. 6b, taken at P0 = 370 torr and nominal room

temperature, T * 300 K. For rotational temperature

inference, after the averaging is performed, the spectrum is

fitted with the Sandia CARSFT least squares fitting code

(Palmer 1989). The resulting nitrogen best fit rotational

temperature for the spectrum shown in Fig. 6b is Trot

(N2) = 322 K, with precision equal to ±20 K, determined

as discussed below. For vibrational temperature inference,

the v = 0 and v = 1 bands, which as can be seen in

Fig. 12, are well isolated spectrally and are numerically

integrated, which upon division by v’’ ? 1 to correct for

the cross section (for assumed harmonic potential), yield

the v = 1 and 1 level populations. Note that the baseline

observed in Fig. 12, which is due to a combination of

nonresonant background and data acquisition offset, is

subtracted prior to the numerical integration.

To quantify measurement precision and accuracy, 60

20-shot (for rotation) and 100-shot (for vibration) average

spectra are obtained, identically to that of Figs. 6b and 8a,

respectively, for rotational and vibrational temperature. For

rotational temperature, the set of spectra were individually

fit with the CARSFT code. The results are shown in the

histogram plot in Fig. 7, along with a Gaussian curve with

the same mean temperature, Trot = 316.6 K, and standard

deviation, r = 10.0 K. The system precision is demon-

strated to be quite good, with a 95 % confidence interval of

±20 K, despite the difficulty for this low temperature

condition (due to the small number of rotational levels

populated and the narrow spectral spacing of the low-lying

levels). The system accuracy is comparable, as the true gas

temperature for these measurements is *300 K. Note that

while not measured explicitly for this work, the absolute

accuracy of Q-branch CARS thermometry is typically

found to improve with increasing temperature due to the

population of higher rotational levels, which are more

readily resolved in the CARS spectra.

3.2 Pure nitrogen flow, nsec pulser alone

The first discharge condition for consideration is the case

of 300 torr pure nitrogen flowing through the discharge

with the nsec pulser in operation alone, that is, no DC

sustainer discharge. A typical spectrum is shown in Fig. 8a,

from which a value of Tvib = 761 ± 74 K is inferred,

significantly exceeding the gas temperature, Trot = 379 ±

15 K. Uncertainties reported are for 95 % confidence

interval, based on collection of 60 individual spectra, as

discussed in Sect. 3.1 above. (Note that no measureable

signal for the v = 1 transition is observed when the dis-

charge is not operated, indicating that the appearance of the

280 300 320 3400

2

4

6

8

10

12

14

Fit Temperature [K]

# C

ou

nts

Fig. 7 Histogram of CARS best fit rotational/translational tempera-

ture obtained from Carsfit; nominally room temperature N2,

P0 = 370 torr

2290 2300 2310 2320 23300

10

20

30

40

v = 0 →

← v = 1

Tvib

= 761 +/- 74 K

Trot

(v = 0) = 379 +/- 15 K

Sq

rt(In

t.) [a

.u.]

(a) Raman Shift [cm-1]

2320 2324 2328 2332

(b) Raman Shift [cm-1]

CarsfitData

Fig. 8 Sample 20-shot average

spectra, pulser alone,

300 torr N2 showing

a Nonequilibrium v = 0 and

v = 1 peaks, and b v = 0 peak

with Carsfit synthetic spectra

Exp Fluids (2013) 54:1422 Page 7 of 13

123

vibrational ‘‘hot’’ band is completely due to electron

impact excitation from the plasma, rather than potential

optical pumping due to stimulated Raman scattering).

Figure 8b plots the same data, zoomed on the ground

vibrational band, along with the CARSFT synthetic spec-

trum. Ten of these 20-shot averaged spectra have been

collected and processed for this condition, as well as for the

conditions of 200 and 370 torr nitrogen flowing through the

nsec pulser discharge. The results of these runs have been

plotted and are shown below in Fig. 9, along with error

bars showing the 95 % confidence bounds.

Since the nanosecond pulse peak voltage exceeds 20 kV

(see Fig. 3), the estimated peak reduced electric field in the

5-mm gap between the pulser electrodes is very high,

E/npeak * 300–400 Td at P0 = 200–400 torr. At these

high reduced electric fields, the discharge energy fraction

going to vibrational excitation of nitrogen by electron

impact is predicted to be very low, less than 3 % (Raizer

1991). However, kinetics of energy coupling to the plasma

in a nanosecond pulse discharge is more complex. Although

the electric field before breakdown is very high, field in the

plasma after breakdown is reduced considerably due to

charge separation in the sheath and plasma shielding. Since

the applied voltage continues to vary after breakdown, this

generates displacement current in the external circuit, equal

to the conduction current in the plasma between the elec-

trodes (Adamovich et al. 2009). This causes additional

energy coupling to the plasma at relatively low reduced

electric fields, that is, at the conditions when a significant

fraction of input energy may be loaded directly into the

vibrational energy mode of nitrogen by electron impact.

To evaluate the discharge pulse energy fraction stored in

nitrogen vibrations, we used an analytic model of nano-

second pulse discharge in quasi-one-dimensional plane

geometry, with both electrodes covered by dielectric layers

(Adamovich et al. 2009). Briefly, the model incorporates

key effects of pulsed breakdown, plasma shielding, and

sheath development on nanosecond time scale. The model

predicts pulse energy coupled to the plasma during and

after breakdown versus discharge geometry, pressure,

temperature, and pulse voltage waveform parameters. In

the model, the experimental pulse voltage waveform,

shown in Fig. 3 (Yin et al. 2012), was approximated by a

superposition of Gaussian shape pulses, as shown in

Fig. 10. Multiple voltage peaks in the experimental

waveform are due to pulse reflections off the load (the

discharge) and the pulse generator. The reflected pulses

couple additional energy to the plasma generated during

the incident pulse. It can be shown (Adamovich et al. 2009)

that after breakdown and sheath formation, the field in the

quasi-neutral plasma, Ep(t), which controls the energy

loading kinetics, is given as:

EpðtÞ ¼1

1þ els2l

ee0

2lenele

dVappðtÞdt

� ee0

2lenele

dVappðtÞdt

ð2Þ

where Vapp(t) is the applied voltage, le and ne are electron

mobility and quasi-steady-state electron density reached

after breakdown, l and e are the thickness of the dielectric

layers and the dielectric constant, and ls is the sheath

thickness. The analytic model is in very good agreement

with numerical modeling of nanosecond pulse discharge in

nitrogen (Adamovich et al. 2009).

The calculations have been done for nitrogen at T =

300 K and P0 = 200–370 torr, for the discharge gap of

L = 5 mm, electrode surface area of A = 3.0 x 3.6 cm =

10.8 cm2, and alumina ceramic dielectric thickness of

l = 1.5 mm (e = 9.1). The electron mobility was assumed

to be constant, le = 400 [760/P(torr)] cm2/V s, and the

Townsend ionization coefficient in nitrogen was taken to

be the same as in our previous work (Adamovich et al.

150 200 250 300 350 400

300

400

500

600

700

800

900

1000

Discharge Pressure [Torr]

Tem

per

atu

re [

K]

Tvib

Trot

Fig. 9 Rotational and first-level vibrational temperature for nsec

pulsed N2 discharge alone versus discharge (plenum) pressure

Fig. 10 Applied electric field (experimental and Gaussian fit), field

in the plasma, and electron density in the plasma; nitrogen,

P0 = 300 torr

Page 8 of 13 Exp Fluids (2013) 54:1422

123

2009). The discharge energy fraction going directly into

vibrational excitation by electron impact versus reduced

electric field was taken from the numerical solution of the

Boltzmann equation for plasma electrons (Raizer 1991).

The initial electron density before the pulse was assumed to

be very low, ne0 = 108/cm3. It was shown previously that

pre-ionization starts affecting the coupled pulse energy at

initial electron densities above ne0 = 1011/cm3 (Adamo-

vich et al. 2009).

Figure 10 plots the electron density and the field in the

plasma during and after breakdown at P0 = 300 torr. It can

be seen that after breakdown, the field in the plasma falls

rapidly and becomes about an order of magnitude lower

than the applied field, as discussed previously. During

breakdown, electron density increases by several orders of

magnitude and reaches steady state when the field in the

plasma becomes too low to produce additional ionization.

Figure 11 plots the reduced electric field in the plasma

during and after breakdown at the conditions of Fig. 10, as

well as energy coupled to the plasma, energy loaded into

nitrogen vibrational mode, and total energy in the load

(which includes energy coupled to the plasma and elec-

trostatic energy stored in the capacitive load) versus time.

These results are compared to the total energy in the load

measured in the experiment (Yin et al. 2012).

From Fig. 11, it can be seen that energy coupled to the

plasma by the incident pulse and two reflected pulses is

close to the experimentally measured value (in the exper-

iment, the coupled energy increases by an additional

*10 % due to additional pulse reflections not shown in

Figs. 10 and 11). It is also apparent that vibrational energy

loading occurs primarily after breakdown (during the latter

portion of the incident pulse and during reflected pulses),

when the reduced electric field in the plasma does not

exceed 30 Td (see Fig. 11). The net energy fraction loaded

into nitrogen vibrations, predicted by the model, is fairly

significant, approximately 33 %. This is in fairly good

agreement with the experimental nonequilibrium vibra-

tional and rotational temperatures inferred from the CARS

spectra, *15–30 % at P0 = 200–300 torr (see Fig. 9).

Thus, comparison of the results of CARS measurements

with kinetic model demonstrates that fairly significant

vibrational excitation can be achieved in repetitive nano-

second pulse discharges, due to energy coupling to the

plasma after breakdown.

3.3 Pulser–sustainer discharge, with relaxer injection

Figure 12 shows a pair of N2 CARS spectra obtained from

operation of the pulser–sustainer discharge, with a DC

power supply voltage (VPS) of 4.5 kV. The dash-dotted

black curve corresponds to a baseline case of pure nitrogen

flow at 300 torr, and the solid blue curve shows the spec-

trum when 1 torr partial pressure of CO2 is injected

downstream of the discharge, approximately 9 cm (*2 ms)

upstream of the CARS measurement location, as shown in

Fig. 1. It is clear that there is a significant difference

between the two conditions. For no relaxer injection, the

flow is extremely nonequilibrium, with a vibrational tem-

perature of nearly 2,000 K, while the inferred gas rota-

tional/translation temperature is Trot * 450 K. With only

1 torr partial pressure of carbon dioxide injected, nearly all

the vibrational energy has been removed from the nitrogen,

evidenced by the nearly equilibrated Tvib = 815 K and

Trot = 630 K.

Fig. 11 Reduced electric field, coupled energy, energy loaded into

vibrational mode of nitrogen, and total energy in the load (experi-

mental and predicted); nitrogen, P0 = 300 torr

2290 2300 2310 2320 23300

20

40

60

80

100

Raman Shift [cm-1]

Sq

rt(I

nt.

) [a

.u.]

no inj: Tvib

= 1980K, Trot

= 450K

1 Torr CO2: T

vib = 815K, T

rot = 630K

Fig. 12 CARS spectra for pulser–sustainer discharge at 300 torr total

pressure, in pure nitrogen (i.e., no relaxer injection, black dash-dot)and 1 torr CO2 partial pressure injection (blue solid)

Exp Fluids (2013) 54:1422 Page 9 of 13

123

Five different species (carbon dioxide, nitric oxide,

hydrogen, oxygen, and nitrogen) were injected over a range

of partial pressures. These gases were chosen because their

rates for nitrogen vibrational relaxation vary by several

orders of magnitude. Figure 13 plots both the nitrogen

vibrational and rotational/translational temperatures mea-

sured in these mixtures versus the partial pressure of the

injected species. It can be seen that the addition of oxygen

to the discharge-excited nitrogen (solid red curve), up to

20 % mole fraction (nearly ‘‘synthetic air’’), does not cause

a significant change in either the vibrational or rotational

gas temperature, due to the low vibration–vibration (V–V)

energy transfer rate coefficient, N2 (v = 1) ? O2 (v = 0)

? N2 (v = 0) ? O2 (v = 1), kVV = 3 9 10-17 cm3/s at

Trot = 450 K (Taylor and Bitterman 1969). For injection of

60 torr oxygen, the characteristic time for nitrogen V–V

relaxation, sVV * 1/kVVnO2*17 ms, is significantly

longer than the *2 ms flow residence time between the

injection location and the CARS measurement region. The

characteristic time for V–T relaxation for N2–O2 is even

longer.

As can also be seen in Fig. 13, carbon dioxide (magenta

dash-dot curve), which is known to be an extremely rapid

relaxer of nitrogen vibrations, exhibits the most rapid rate

of relaxation. Even at the lowest injection partial pressure,

PCO2= 0.25 torr, a substantial reduction in nitrogen

vibrational temperature, by more than 450 K, is observed,

as well as a corresponding increase in nitrogen rotational

temperature by 50 K. The rate coefficient for V–V energy

transfer, N2 (v = 1) ? CO2 (000) ? N2 (v = 0) ? CO2

(001), at Trot = 450 K, is very high, kVV = 4.5 9 10-13

cm3/sec (Taylor and Bitterman 1969) (sVV * 1/kVVnO2

*70 lsec for CO2 partial pressure of only 1 torr). CO2

vibrationally excited in collisions with nitrogen relaxes via

rapid intramolecular V–V energy transfer to m2 (010) and m1

(100) modes, with subsequent rapid V–T relaxation.

Injection of nitric oxide and hydrogen exhibits similar

behavior to carbon dioxide, although on a slower time

scale. The rate coefficient of V–V energy transfer from

nitrogen to nitric oxide, N2 (v = 1) ? NO (v = 0) ? N2

(v = 0) ? NO (v = 1), is kVV = 3.4 9 10-15 cm3/sec

at 450 K (Doyennette and Margottin-Maclou 1986)

(sVV * 1/kVVnNO *2 ms for NO partial pressure of 5 torr,

comparable to the flow residence time). The V–T relaxa-

tion rate coefficient for N2–H2, N2 (v = 1) ? H2 ? N2

(v = 0) ? H2, is kVT = 6.1 9 10-16 cm3/sec at 450 K

(Allen et al. 1980) (sVT * 1/kVTnH2*6 ms for H2 partial

pressure of 10 torr). Using the temperature dependence

suggested in the work by Capitelli et al. (2001), this rate

increases to kVT = 1.3 9 10-15 cm3/s at T = 550 K

(sVT * 3 ms). Similar to N2–O2, the V–V energy transfer

for N2–H2 is extremely slow, due to a significant difference

in vibrational quanta (Bott 1976).

Injection of nonexcited nitrogen was also performed;

although the V–V energy transfer rate coefficient for

nitrogen, N2 (v = 1) ? N2 (v = 0) ? N2 (v = 0) ? N2

(v = 1), is relatively high, kVV = 1.5 9 10-14 cm3/sec at

room temperature (Ahn et al. 2004) (sVV * 1/kVVnN2

*0.5 ms for injection N2 partial pressure of 5 torr), the

resonant energy transfer process simply results in the

redistribution of N2 vibrational energy among the ‘‘dis-

charge-loaded’’ and ‘‘cold-injected’’ molecules, but the

energy remains ‘‘locked’’ in the nitrogen vibrational mode.

As the results of Fig. 13 indicate, energy extraction from

the nitrogen vibrational mode results in an increase in the

gas temperature, that is, thermalization of the vibrational

energy as the discharge-loaded flow is equilibrated.

Figure 14 plots the average total nitrogen translational/

rotational ? vibrational energy per molecule as a function

of partial pressure of the injected species for all of the

conditions shown previously in Fig. 13. As the figure

demonstrates, within the experimental uncertainty, the total

.2 .3 .5 .7 1 2 3 5 7 10 20 30 50 70 100

400

600

800

1000

1200

1400

1600

1800

2000

2200

Injection Partial Pressure [Torr]

Tem

per

atu

re [

K]

Tv no inj

Tr no inj

CO2

NOH

2O

2"cold" N

2

Fig. 13 Pulser–sustainer discharge, VPS = 4.5 kV, various injected

species, 300 torr total mixture pressure

0 .2 .3 .5 .7 1 2 3 5 7 10 20 30 50 700

20

40

60

80

100

120

140

Injection Partial Pressure [Torr]

Avg

. En

erg

y L

oad

ing

/mo

lec

[meV

]

no injCO

2NOH

2O

2cold N

2

Fig. 14 Sum of nitrogen translational ? rotational ? vibrational

energy as a function of relaxant species partial pressure

Page 10 of 13 Exp Fluids (2013) 54:1422

123

nitrogen rotational, translational, and vibrational energy is

conserved. This indicates that any inter-species V–V

transfer is followed by rapid V–T relaxation, with a result

that there is negligible energy storage in the vibrational

modes of injected species.

3.4 Measurements in a supersonic flow

Finally, measurements of vibrational temperature in the

supersonic section have also been performed. These mea-

surements are especially challenging due to very low free-

stream static pressure, P0 = 1.2 torr, measured using a

wall pressure tap at the end of the nozzle. As mentioned, a

5-mm-diameter quartz cylinder is positioned in the super-

sonic flow, which creates a bow shock as shown previously

in Fig. 5a. The CARS spectra in Fig. 15a are eight-shot

averages, collected both in the supersonic free-stream and

behind the bow shock for 300 torr pure N2 in the plenum

with no discharge. The difference in signal-level results

from the different number densities present in the two

measurement locations; the spectra also demonstrate typi-

cal signal-to-noise levels for this arrangement. The plots in

Fig. 15b, c show ten-shot average spectra, collected in the

supersonic free-stream and behind the bow shock, respec-

tively, for 300 torr N2 in the plenum, with the pulser–

sustainer discharge in operation (sustainer VPS = 4.5 kV).

Electromagnetic interference (EMI) caused by the nsec

pulser can be clearly seen in both spectra and becomes

increasingly problematic as signal levels decrease. The

significant reduction in the ground vibrational level signal

strengths for both of these spectra (compared to the ‘‘no-

discharge’’ data acquired at each location) can be attributed

to both a reduction in total number density, due to gas

heating from the discharge, and significant population loss

to excited vibrational states. The ratio of the integrated

square root of peak intensities is used to determine the

first-level nitrogen vibrational temperatures, as previously

described.

Due to the quadratic scaling of CARS signal intensity to

the density, as previously mentioned, the integrated square

root of the CARS signal gives an estimate of local number

density. This can be used to determine whether the mea-

surement volume is upstream or downstream of the shock

location, due to the significant rise in density as the flow

traverses the shock. Spectra captured without the discharge

in operation, such as those seen in Fig. 15a, are collected at

various locations upstream of the cylinder, and the square

root of the spectral peaks is integrated to give an estimate

of local number density. The results are shown in Fig. 16a,

where the black x’s are for pulser–sustainer discharge

operating in 300 torr pure N2 flow, and the red ?’s are

recorded with 0.25 torr CO2 injection, as previously

described. As the plot indicates, a sharp rise in density

occurs 1.0 mm upstream of the cylinder surface. The

cluster of data points captured at x = -1.00 mm (number

density *10 in arbitrary units) represent the free-stream

position directly adjacent to the shock front. Moving

downstream, measurements captured at x = -0.99, -0.98,

and -0.95 mm, shown as the next three black x’s on the

plot, demonstrate much higher densities, indicating the

measurement region has moved beyond the shock location.

As mentioned above, shock stand-off distance detected

in the present work, 1.0 mm, is somewhat smaller than

0

2

4(b)

Sq

rt(I

nt)

[a.

u.]

2300 2310 2320 23300

5

10(c)

Sq

rt(I

nt)

[a.

u.]

Raman Shift [cm-1]

0

30

60(a)

Sq

rt(I

nt)

[a.

u.]

Free-streamBehind Shock

Fig. 15 Sample CARS spectra collected in the supersonic flow;

a eight-shot average spectra collected in the free-stream and behind

the shock layer with no discharge present; ten-shot average spectra

with the pulser–sustainer discharge operating, collected in the

b supersonic free-stream and c behind the bow shock

-2 -1.8 -1.6 -1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0300

600

900

1200

1500

1800

2100

Tvi

b [K

]

Distance from cylinder [mm]

Cyl. Surface→

(b)

-2 -1.8 -1.6 -1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0

10

20

30

40

~ n

um

ber

den

sity

[a.

u.]

Bow Shock→(a)

pure N2

CO2 inj.

Fig. 16 Inferred a estimated number density, and b vibrational

temperature in the supersonic flow free-stream and behind the bow

shock

Exp Fluids (2013) 54:1422 Page 11 of 13

123

measured in our previous experiments using a 4-cm-long,

5-mm-diameter cylinder, supported at the tunnel sidewalls,

1.2 mm (Nishihara et al. 2012). This may be explained by

the effect of finite cylinder length, which has been studied

in a Mach 4 flow in a shock tube experiment (Kim 1956).

In the work by Kim (1956), it was shown that the ratio of

stand-off distance to the cylinder diameter, d/D, tends to

decrease as the cylinder length-to-diameter ratio, l/D, is

reduced below l/D = 4. This effect is primarily due to flow

three-dimensionality, that is, spanwise flow behind the bow

shock. At the present conditions, l/D = 7.5/5 mm = 1.25.

The number densities measured behind the shock are

nominally 3–4 times higher than those measured in the

free-stream, somewhat less than the bow shock density

jump of 5.1, predicted by a 3-D compressible Navier–

Stokes flow code for a cylinder model extending wall-to-

wall (Nishihara et al. 2012). The density ratio across the

shock may be also affected by the three-dimensionality of

the flow over the short cylinder model used in the present

work, as well as possible beam steering due to the index

gradient in the vicinity of the shock. Since density mea-

surement was not a focus of this work, beam steering was

not examined in detail, other than confirmation that the

apparent density increased smoothly through the shock with

no large increase in shot-to-shot fluctuations in the signal.

As mentioned previously, the Unstable-resonator Spatially

Enhanced Detection Coherent Anti-Stokes Raman Scatter-

ing geometry was specifically used in order to minimize

potential beam steering (compared to folded BOXCARS),

although the effective of this strategy was not examined in

any detail.

The plot in Fig. 16b shows the inferred Tvib (N2) for

several locations both in the free-stream and behind the

shock. These data are taken with the pulser–sustainer dis-

charge operating in 300 torr N2 in plenum, with and

without 0.25 torr CO2 injection. Both with CO2 injection

and without injection, vibrational temperatures inferred

behind the shock are very similar to those observed in the

free-stream. As the recovery pressure behind the shock is

significantly lower than the plenum pressure, vibrational

relaxation beyond that present in the relatively high density

subsonic flow does not occur. The vibrational temperature

inferred for the case of 0.25 torr CO2 injection is close to

the value inferred in plenum at the same conditions,

Tvib * 1,450 K. The vibrational temperature for pure N2

without injection is *150 K less than the value measured

in plenum at the same conditions, Tvib = 1,900 K, a dif-

ference of less than 10 %. Significant spread in the data is

observed, primarily due to the rather low signal-to-noise

levels present due to the EMI effects from the nsec pulser

previously mentioned, as well as the general difficulty of

very low CARS signal levels in these extremely low-den-

sity flows.

4 Conclusions

Picosecond Unstable-resonator Spatially Enhanced Detec-

tion Coherent Anti-Stokes Raman Scattering Spectroscopy

(USED-CARS) has been used to determine N2 rotational/

translational and first-level vibrational temperature simul-

taneously, in a nonequilibrium flow excited by a nsec pulser/

DC sustainer electric discharge in the plenum of a Mach 5

wind tunnel at pressures in the range 200–370 torr, and ini-

tial temperature of *300 K. It is shown that operation of the

nominally high reduced electric field, (E/n)peak * 500 Td,

nsec pulsed discharge alone results in fairly significant

vibrational loading of pure N2, with Tvib * 720 K at

300 torr. The accompanying rise in translational/rotational

temperature is much smaller, with Trot * 380 ± 20 K.

Modeling predictions based on a recently published 1-D

pulsed nsec discharge model suggest that energy coupling

after discharge breakdown results in significant coupled

energy to the plasma at relatively low E/n (\30 Td), due to

sheath formation and plasma shielding. The addition of an

orthogonal DC sustainer discharge (E/n * 10–30 Td),

which accounts for approximately 80 % of the total power

loading into the flow, is shown to efficiently excite the N2

vibrational energy mode, leading to first-level vibrational

temperatures up to Tvib (N2) = 2,000 K and corresponding

rotational temperatures of Trot * 450 ± 20 K.

Measurements are also performed quantifying the effect

of relaxant gas species on nitrogen rotational/translational

and vibrational temperatures. Specifically, injection of

CO2, NO, and H2 results in partial vibrational relaxation of

nitrogen accompanied by gas heating, demonstrating the

ability to control and tailor the vibrational energy content

of the flow prior to its expansion through the Mach 5

nozzle of the wind tunnel. Determination of average

total nitrogen translational/rotational ? vibrational energy

demonstrates, to within the experimental uncertainty, that

total nitrogen rotational, translational, and vibrational

energy is conserved, indicating that any inter-species V–V

energy transfer occurring is followed by rapid V–T relax-

ation, such that energy storage in molecules other than

nitrogen is negligible.

Finally, the ability to capture vibrational CARS spectra

within very low-density flows present in a Mach 5

expansion nozzle has been demonstrated. Measurements of

first-level vibrational temperature in the supersonic free-

stream and downstream of the bow shock created by

insertion of a cylinder model in the Mach 5 flow have been

performed. Vibrational nonequilibrium persists through the

nozzle throat and expansion, with highly nonequilibrium

gas arriving at the bow shock. Measurements within

300 lm of the cylinder surface reveal that in pure N2

without relaxant species injection, as well as in N2

with 0.25 torr of CO2 injected, no detectable vibrational

Page 12 of 13 Exp Fluids (2013) 54:1422

123

relaxation occurs behind the bow shock, as the recovery

pressure and the corresponding number density are too low.

Acknowledgments The authors would like to thank Dr. Sukesh Roy

for much help and many useful discussions regarding fabrication of

the picosecond broadband dye laser as well as picosecond CARS in

general. Additionally, many thanks are given to Mr. Zhiyao Yin and

Dr. Keisuke Takashima for their work measuring the nsec pulse

waveforms. This research is supported by the AFOSR Hypersonics

program (Technical Manager Dr. John Schmisseur), and the DOE

Low Temperature Plasma Physics Plasma Science Center, Depart-

ment of Energy, Office of Fusion Energy Science Contract DE-

SC0001939.

References

Adamovich IV, Nishihara M, Choi I, Uddi M, Lempert WR (2009)

Energy coupling to the plasma in repetitive nanosecond pulse

discharges. Phys of Plasmas 16:113505

Ahn T, Adamovich IV, Lempert WR (2004) Determination of

nitrogen V–V transfer rates by stimulated Raman pumping.

Chem Phys 298:233–240

Allen DC, Chandler ET, Gregory EA, Siddles RM, Simpson CJSM

(1980) Vibrational deactivation of N2 (v = 1) by n-H2 and by

p-H2. Chem Phys Lett 76:347–353

Bott JF (1976) Vibrational energy exchange between H2 (v = 1) and

D2, N2, HCl, and CO2. J Chem Phys 65:3921–3928

Bruzzese JR, Hicks A, Erofeev A, Cole AC, Nishihara M, Adamovich

IV (2010) Gain and output power measurements in an electri-

cally excited oxygen-iodine laser with a scaled discharge. J Phys

D Appl Phys 43:015201

Capitelli M, Ferreira CM, Gordiets BF, Osipov AI (2001) Plasma

kinetics in atmospheric gases, Chap. 7. Springer, Berlin, p 107

Doyennette L, Margottin-Maclou M (1986) Vibrational relaxation of

NO(v = 1) by NO, N2, CO., HCl, CO2, and N2O from 300 to

600 K. J Chem Phys 84(12):6668–6678

Eckbreth AC (1996) Laser diagnostics for combustion temperature

and species. Gordon and Breach, Amsterdam

Gordiets BF, Osipov VA, Shelepin LA (1988) Kinetic processes in

gases and molecular lasers. Gordon and Breach, London

Kim C-S (1956) Experimental studies of supersonic flow past a

circular cylinder. J Phys Soc Japan 11(4):439–445

Klick D, Marko KA, Rimai L (1981) Broadband single-pulse CARS

spectra in a fired internal combustion engine. Appl Opt 20(7):1178

Lempert WR (2005) Diagnostics. In: Becker K, Kogelschatz U,

Schoenbach K, Barker R (eds) Non-equilibrium air plasmas at

atmospheric pressure. IOP Publishing, Bristol, UK

Leyva IA, Laurence S, War-Kei Beierholm A, Hornung HG, Wagnild

R, Candler G (2009) Transition delay in hypervelocity boundary

layers by means of CO2/acoustic instability interactions. AIAA

Paper 2009-1287

Marko KA, Rimai L (1979) Space- and time-resolved coherent anti-

stokes Raman spectroscopy for combustion diagnostics. Opt Lett

4(7):211

Montello A, Nishihara M, Rich JW, Adamovich IV, Lempert WR

(2012) Nitrogen vibrational population measurements in the

plenum of a hypersonic wind tunnel. AIAA J 50(6):1367

Nishihara M, Jiang N, Rich JW, Lempert WR, Adamovich IV,

Gogineni S (2005) Low-temperature supersonic boundary layer

control using repetitively pulsed MHD forcing. Phys Fluids

17(10):106102

Nishihara M, Rich JW, Lempert WR, Adamovich IV, Gogineni S

(2006) Low-temperature M = 3 flow deceleration by Lorentz

Force. Phys Fluids 18(8):086101

Nishihara M, Takashima K, Rich JW, Adamovich IV (2011) Mach 5

bow shock control by a nanosecond pulse surface dielectric

barrier discharge. Phys Fluids 23:066101

Nishihara M, Takashima K, Jiang N, Lempert WR, Adamovich IV,

Rich JW, Doraiswamy S, Candler GV (2012) Development of a

mach 5 nonequilibrium flow wind tunnel. AIAA J 50:2255–2267

Palmer RE (1989) The CARSFT computer code for calculating

coherent anti-stokes Raman spectra: user and programmer

information. Sandia National Labs., Rept SAND89-8206,

Livermore, CA

Raizer YuP (1991) Gas discharge physics. Springer, Berlin

Roy S, Meyer TR, Gord JR (2005) Broadband coherent anti-stokes

Raman scattering spectroscopy of nitrogen using a picosecond

modeless dye laser. Opt Lett 3:3222

Roy S, Gord JR, Patnaik AK (2010) Recent advances in coherent anti-

stokes Raman scattering spectroscopy: fundamental develop-

ments and applications in reacting flows. Prog Energ Combust

Sci 36:280–306

Taylor RL, Bitterman S (1969) Survey of vibrational relaxation data

for processes important in the CO2-N2 laser system. Rev Mod

Phys 41:26–47

Wagnild R, Candler GV, Leyva IA, Jewell JS, Hornung HG (2010)

Carbon dioxide injection for hypervelocity boundary layer

stability. AIAA Paper 2010-1244

Yin Z, Takashima K, Adamovich IV (2012) Measurements and

kinetic modeling analysis of energy coupling in nanosecond

pulse dielectric barrier discharges. AIAA Paper 2012-3093, 43rd

AIAA plasmadynamics and lasers conference, New Orleans, LA,

25–28 June 2012; accepted for publication in Plasma Sources

Science and Technology, 2012

Exp Fluids (2013) 54:1422 Page 13 of 13

123