Christopher J. Kliewer,1,* Yi Gao,2 Thomas Seeger,3 Brian D. Patterson,1
Roger L. Farrow,1 and Thomas B. Settersten1
1Combustion Research Facility, Sandia National Laboratories, 7011 East Avenue, Livermore, California 94551, USA2Lehrstuhl für Technische Thermodynamik and Erlangen Graduate School in Advanced Optical Technologies (SAOT),
Friedrich-Alexander-Universität Erlangen-Nürnberg, Am Weichselgarten 8, D-91058, Erlangen, Germany3Lehrstuhl Für Technische Thermodynamik, Universität Siegen, Paul-Bonatz-Strasse 9-11 57076 Siegen, Germany
Coherent anti-Stokes Raman spectroscopy (CARS)has been extensively developed over the past 30years as a nonintrusive probe of temperature andmajor species concentrations in combustion environ-ments [1]. CARS is diagnostic of a high spatial andtemporal resolution that generates a coherent signal.By fitting the signal spectrum, the gas-phase tem-perature can be calculated, as both the relative lineintensities and the Raman linewidths are highlytemperature dependent.
Two main variants of CARS have been used forcombustion diagnostics: vibrational CARS (VCARS)and pure-rotational CARS (RCARS). In VCARS, thedifference frequency between the pump and Stokesphotons excites vibrational transitions in moleculesof interest. Most often, this is the ν ¼ 0 → ν ¼ 1 vi-brational transition of N2, as N2 is present in signif-icant concentration in all air-breathing combustionenvironments. The Stokes laser is typically a broad-band source so as to excite the entire rovibrationalmanifold of the ground electronic state. A second nar-rowband pump pulse, often termed the probe pulse,is scattered from this excited rovibrational coher-ence, and the resultant signal is spectrally dispersedand detected. As the temperature of the probed
species increases, higher energy rovibrational levelsare thermally populated in a Boltzmann distribu-tion. In N2, for example, there is a significant mole-cular population in the ν ¼ 1 first vibrational levelfor temperatures exceeding 1200K.
In RCARS, the difference frequency between thepump and Stokes pulses excites rotational transi-tions. In the dual-broadband approach [2,3], a singlebroadband laser pulse provides both the pump andStokes photons. Higher precision and accuracy areachieved as the photon pairs exciting the transitionoriginate from numerous combinations of frequen-cies from within the bandwidth of a single pulse,minimizing the effect of laser mode fluctuations[2–4]. Again, a narrowband probe is scattered fromthese rotational coherences, and the resultant signalis dispersed by a grating and detected. Because rota-tional transition frequencies for most molecules, in-cluding important combustion species such as N2,O2, CO, and CO2, fall within a narrow spectral win-dow, multiple species may be excited and detectedwithin a single laser pulse if present in sufficient con-centration, as was recently demonstrated in a CO/airdiffusion flame [5].
The majority of experiments using RCARS havebeen at a low temperature [6], as VCARS is more sen-sitive at combustion temperatures (≥1200K), yet sev-eral studies have shown RCARS to be a viable andaccurate flame-temperature combustion diagnostic[7,8]. The wide spacing of rotational lines in anRCARS spectrum reduces the effect of coherent linemixing, which complicates VCARS at high pressures[5]. The use of RCARS is also beneficial in soot-ing combustion to avoid interference from laser-produced C2 Swan-band emission, which interfereswith standard nitrogen VCARS detection unlessthree-color schemes are employed [9,10].
The most common CARS experimental geometryuses a small angle phase-matching geometry inwhich all three laser beams are focused by the samespherical lens to a point [11]. The CARS-probed vol-ume in this geometry is shaped like a cylinder withtapered ends [12]. The long axis of the cylinder isoriented parallel to the laser beam propagation di-rection (z axis of Fig. 1) and is often termed the inter-
action length. If, for instance, a spherical lens with a300mm focal length is used, the interaction lengthwill be approximately 2–3mm, depending on the se-paration of the beams on the lens. The diameter ofthe probe-volume cylinder depends on how tightlythe laser beams are focused, but 150 μm is common.If higher spatial resolution is desired, large-anglephase-matching schemes may be employed [13]. Toacquire temperature or concentration profiles, eitherthe target or optics table is translated, and spectraare taken at each location.
As experimental combustion diagnostics are beingemployed to validate combustion simulation data, itis advantageous to simultaneously compare multipleresolved locations within a flame or reacting flow.Thus, extending the diagnostic capability of CARSbeyond single-point measurement is desirable, andthere have been a number of attempts with limitedsuccess for both VCARS and RCARS.
In the first VCARS imaging attempt, Murphy et al.[13] used a large-angle phase-matching configura-tion to detect the CH4 VCARS intensity along a line5mm in length with the goal of obtaining a concen-tration profile. Room-temperature CH4 was used be-cause of its very large vibrational cross section (8×that of N2 [14]). The major limitation was the inabil-ity to account for the spatially fluctuating laser beamprofiles, making absolute concentration measure-ments impossible. Snow et al. [15] later reported abroadband RCARS one-dimensional (1D) imagingexperiment in which RCARS spectra were recordedfor 17–20 individual spatially resolved volumeelements along a line using a large-angle phase-matching configuration in N2 and O2 flows. Becauseof signal-level limitations, the single-shot image waslimited to 1:8mm and cool temperatures. Stuffle-beam and Eckbreth [16] first demonstrated the useof a cylindrical lens in the planar BOXCARS phase-matching geometry to focus the three laser beams toa width of 150 μm (y axis of Fig. 1) and an interactionlength of 3mm (z axis of Fig. 1) to form a 1D VCARSsignal image. This technique was used to image N2,O2, and H2 at high pressures in the fizz zone of a solidpropellant flame. However, the useful line–CARS im-age was only 400 μm in spatial extent due to signallimitations, restricting its utility as an imaging diag-nostic. Sufficient signal was only obtained over400 μm, as it results from the more intense portionsof the laser beams. In a similar set of experimentsusing a high-energy nanosecond laser (>1 J at532nm), Jonuscheit et al. [17,18] quantified the accu-racy and precision of single-shot 1D VCARS imagingfor measuring the N2 temperature. The line–CARSimage formed was 6mm in spatial extent (x axis ofFig. 1) at room temperature, but the useful part ofthe image reduced to 2mm for measurements at1500K due to the reduction in signal level. The ac-curacy and precision of the technique for N2 tempera-ture evaluation were found to be similar to thatreported for single-point VCARS. The techniquewas demonstrated in a lean premixed laminar
Fig. 1. (Color online) Phase-matching arrangement for the 1DRCARS imaging presented in this work. The figure demonstratesphase matching for a vertical 1D line; both vertical and horizontalline-imaging configurations are presented. The z axis is nominallyin the laser propagation direction. The y axis is the focusing dimen-sion of the cylindrical lens, and the x axis corresponds to the lasersheet height.
CH4–air flame with a maximum evaluated tempera-ture of 1400K.
The development of an RCARS imaging techniquehas the added benefit of obtaining relative majorspecies concentration/gradients simultaneously withthe gas-phase temperature gradients. RCARS at ele-vated temperatures, however, suffers from an addi-tional loss in signal intensity, as the rotationalspectrum stretches out over a much larger Raman-shift range as compared with its VCARS counter-part. Bood et al. [19] developed a novel technique toaddress the signal-level issue by using a series of cy-lindrical lenses to focus the CARS beams to threespatially resolved locations. Temperatures of lessthan 450K were evaluated from spectra accumu-lated at three spatially resolved points above aheated aluminum plate, and it was concluded thatwith higher laser pulse energies, the technique maybe applicable to increased temperatures or to moremeasurement locations.
Here we demonstrate a significant advance by ex-ploiting the use of picosecond laser pulses, extendingrecent work [20–25] on the development of picose-cond RCARS to imaging. Using picosecond pulses,as opposed to nanosecond pulses, to excite and probethe rotational coherences benefits from the increasedpeak pulse power, which can more efficiently drivethe Raman coherence. Thus, high CARS signal levelsmay be obtained with relatively lower overall pulseenergies. This advantage of using picosecond laserpulses in CARS lends itself to adaptation to an imag-ing configuration, where pulse irradiances are neces-sarily decreased because of larger beam areas thanused in a focused geometry. Furthermore, the useof delayed probing suppresses interference to the sig-nal from the nonresonant susceptibility and smearedVCARS [23,24], especially problematic in fuel-richmeasurements, and this has been demonstrated toallow in situ acquisition of Raman linewidths directlyfrom the time domain [26,27]. In this work, a 1DRCARS image of up to 12mm in length was obtainedand demonstrated as a combustion diagnostic in arich premixed laminar propane flame. Both tempera-ture and O2:N2 concentration ratio gradients wereevaluated from the collected image. Both the accu-racy and precision of this technique for single-shotevaluations were determined in a heated flow of N2.
2. Experiment
Figure 1 depicts the phase-matching configurationused in this work. The three laser beams were fo-cused to the probe volume using a cylindrical lens offocal length 300mm.
The flame measurements were performed in a richlaminar premixed flame using propane as the fuel.Propane, N2, and O2 were premixed and emittedfrom a stainless steel tube of 5:17mm inner diam-eter. The flow rates were 1.4 standard liters per min-ute (slm) propane, 2:53 slm O2, and 6:5 slm N2 for anequivalence ratio of Φ ¼ 2:8. The flame nozzle waseither mounted horizontally or vertically with re-
spect to the laser table to orient the 1D line–CARSimage vertically or horizontally, respectively. Ineither arrangement, the 1D probe volume was re-solved across the flame front radially. Measurementswere taken at axial heights of 5 and 15mm, as shownin Fig. 2.
A regeneratively amplified Nd:YAG laser operat-ing at 20Hz was frequency-doubled to produce lightat 532nm. The 532nm output pumped a broadbanddye laser operating at 633nm using 4-dicyanomethy-lene-2-methyl-6-p-dimethylaminostyryl-4H-pyrandye. The pulse width of the Nd:YAG laser at 532nmwas approximately 65ps, as measured with a streakcamera. The dye laser consisted of a side-pumpeddye cell emitting amplified spontaneous emission,followed by three side-pumped dye-amplificationstages. Typically, 110mJ=pulse was used to pumpthe dye laser to produce 23mJ=pulse of 633nm light.A half-wave plate and polarizing beam splitter wereused to separate the red beam into two equal beamsfor use as the RCARS pump and Stokes beams. A sec-ond identical regeneratively amplified Nd:YAG laserwas used to supply the narrowband 532nm probepulse for the experiment with a typical pulse energyof 35mJ=pulse. Both Nd:YAG lasers were locked toan RF source, allowing for precise time synchroniza-tion of the picosecond pulses.
A 50mm square cylindrical lens with a 300mmfocal length was used to form sheets from all threelaser beams with a width of approximately 150 μm(FWHM) (y axis of Fig. 1) at their crossing and aheight of up to 1:2 cm (x axis of Fig. 1). The heightsof the sheets were adjusted by using telescopes to setthe beam diameters prior to the focusing lens. Carewas taken to ensure all beams were well collimated.To account for astigmatism in the dye laser beam,two cylindrical telescopes were used to collimate thered pulse in the vertical and horizontal dimensions.The interaction length of approximately 1:9mm inlength (FWHM) was measured by translating a thinglass cover slip (0:14mm thickness) through thebeam-crossing region and monitoring the four-wavemixing signal. Translation stages were used in thebeam paths of one of the broadband beams and in
Fig. 2. (Color online) Rich laminar premixed propane flame usedin this study (Φ ¼ 2:8). The green line segments crossing the flamefront denote the 1D probe volumes studied.
the narrowband probe beam to precisely adjust thetemporal overlap of all three pulses. The timing be-tween the broadband laser and the probe laser wascontrolled electronically to set delays between thecoherence driving pulses and the probe pulse from0ps to over 1ns with 20ps time steps. A planarBOXCARS phase-matching geometry was used.Following the beam crossing, the green probe beamand the nearly collinear pump beam were directed toa beam dump using a right-angle prism. Because thesignal was emitted nearly collinearly with one of thered pump beams, three dichroic mirrors were used toseparate the signal from the 633nm beam. No spec-tral filtering of the signal was required to suppressstray 532nm light, which lies very close spectrallyto the RCARS signal, because of the high signal-to-stray-light ratio afforded by using picosecond-lengthpulses [23]. A 75mm focal-length cylindrical lens wasused to focus the signal into the spectrometer slit. An1800 groove=mm diffraction grating was used in a1m spectrometer (SPEX-1000M) to disperse the sig-nal onto a back-illuminated air-cooled CCD camerawith 24 μm pixel spacing, providing a dispersion of0:44 cm−1 per pixel, as determined by fitting aRCARS spectrum taken in room air.
Special care was taken to relay image the x dimen-sion of the probe volume (see Fig. 1) onto the spectro-meter slit. Two different optical setups were used forvertical or horizontal imaging sheets. For the hori-zontal image, a turning periscope was used to directthe image through the vertical spectrometer slit. TheCCD was oriented such that the vertical dimension,consisting of 330 pixels, corresponded to the volumeelements of the 1D image (x axis; see Fig. 1), and thehorizontal chip dimension, consisting of 1200 pixels,corresponded to the Raman shift. Typically, 2 × 2 on-chip CCD binning was used. Thus, each pixel row inthe CCD readout corresponded to the RCARS spec-trum of an imaged volume element, and each imagedvolume element then had spatial dimensions of150 μm × 72 μm× 1:9mm (y × x × z in Fig. 1). The to-tal 1D region imaged to the CCD camera was 12mm.A resolution target was used to maximize and verifythe spatial resolution of the imaging optics by placingit at the beam crossing and illuminating it with whitelight. The CCD camera was attached to the spectro-meter on a mount that allowed translation of thecamera in and out of the spectrometer, allowing max-imization of the resolution. The final imaging lenswas placed on a translation stage and used to mini-mize the line-spread function measured using theresolution target. The line-spread function was eval-uated to have a width of 62 μm (FWHM) from the re-solution target tests. A razor blade was translatedthrough the crossing while illuminated with whitelight to verify the spatial calibration of the imagingoptics (x axis; see Fig. 1), and the line-spread functionmeasured with the razor blade was similar to thatmeasured with the resolution target.
To characterize the precision and accuracy ofsingle-shot images taken with this technique, 100
single-shot images were acquired in a heated flowof N2 at temperatures of up to 1200K 1 cm abovethe glass flow tube nozzle exit. The evaluated tem-peratures were compared to thermocouple readingsfor an assessment of accuracy and standard devia-tions calculated for an assessment of precision.Radial temperature profiles were taken with afine-wire (50 μm) type-R thermocouple in the heatedflow at 1 cm above the nozzle exit to characterize thetemperature gradient. Deviations of less than 2K inthe heated flow were observed over the interactionlength of 2mm at 1000K, as shown in Fig. 3.
Spectra were fitted using a time-independent spec-tral-fitting code that has been used extensively fornanosecond-laser-based RCARS [28]. The calculationof the CARS spectra and the temperature evaluationwere based on an algorithm that compares experi-mental spectra with spectra from a precalculatedlibrary by the use of a nonlinear least-squares proce-dure [29]. The experimental spectra were correctedfor the spectral profile of the broadband dye laserby dividing the measured spectra by a spectrum pro-duced by pure argon. Because argon has no rota-tional transitions, only the nonresonant four-wavemixing signal contributes to the spectrum. The pa-rameters for the experimental slit function aredescribed by a Voigt function and were determinedby fitting experimental spectra taken in air underambient conditions. These values were used as fixedinput parameters for the calculation of the theoreti-cal spectra. For the calculation of the Raman line-width, we used the modified exponential gap law[30]. The parameter sets used can be found in [31].
3. Results
A. RCARS Imaging in a Heated Flow of N2
We obtained 1D images in both horizontal and verti-cal sheet orientations. Figure 4 demonstrates a 1Dimage taken with horizontal sheets in room-temperature air. The collected image spans approxi-mately 12mm in space, with each volume elementcorresponding to 36 μm. A vertical chip binning oftwo was used such that each resolved spectrum cor-responds to 72 μm in the x direction of Fig. 1. TheO2:N2 ratio for the spectrum in Fig. 4(c) is evaluated
Fig. 3. Transverse temperature profile taken by thermocouple(symbols) in the N2 heated flow used in the single-shot precisionand accuracy measurements. The RCARS probed volume is indi-cated, and z ¼ 0 corresponds to the center of the nozzle.
to be 27% at an evaluated temperature of 295K,consistent with the composition of air.
A heated flow of nitrogen was used to characterizethe single-shot accuracy and precision of this tech-nique in addition to revealing information aboutthe single-shot detection limit with the current appa-ratus and image size. A series of 100 single lasershots was taken in the N2 flow at thermocouple tem-peratures of 408K, 697K, 985K, and 1205K. A sin-gle pixel row spectrum corresponding to the same72 μm element was isolated and evaluated from all100 images. These evaluated spectra were compared
to both a 100-shot-averaged spectrum and to thermo-couple readings from the probe volume location. Thecomparison between the averaged spectrum and thethermocouple readings gives an assessment of theaccuracy of the technique, while the standard devia-tion of the 100 evaluated single-shot spectra yields aprecision measurement. In the current experimentalsetup, single-shot spectra were reliable up to ap-proximately 1200K, but signal levels decreased toa level unsuitable for spectral fitting (less than100 counts on the strongest Raman line) for highertemperatures.
At all measured temperatures, the deviation be-tween the thermocouple reading and the evaluatedtemperature of the averaged spectra was always lessthan30K.This comparesquite favorablywithevalua-tions of the accuracy of nanosecond1DVCARS [18]. Inthis work, at 1230K (RCARS temperature), the stan-dard deviation of single-laser-pulse measurements is5:6%, as can be seen in Fig. 5(d). This precision is com-parable to thereported5%single-shot standarddevia-tion for nanosecond 1D VCARS at 1200K [18]. Thestandard deviation of single-shot single-point nanose-cond RCARS at 1200K was reported to be 50K [4], or4.2%. Although the usable region of a 300Kmeasure-ment in this work was 10mm, at a temperature of1200K in the heated flow, the portion of the 1DRCARS image with more than 100 camera countson the most intense Raman line had reduced by a fac-tor of 2 (5mm) because of the reduction in signalintensity with the elevated temperature.
B. RCARS Imaging in a Rich Propane Flame
In order to demonstrate the capability of 1D picose-cond RCARS as a flame diagnostic, we used the tech-nique to probe across the flame front of a richlaminar premixed propane flame with an equiva-lence ratio of Φ ¼ 2:8, as shown in Fig. 2. Analyseswere performed at y ¼ 5mm and y ¼ 15mm axialheights above the nozzle exit. Measurements wereacquired close to the nozzle exit to minimize theeffect of a possibly fluctuating flame front in space.Figure 6 displays the 1D RCARS image taken in thisflame. The radial distance from the center of the ax-isymmetric flame is on the vertical axis and theRaman-shift is given on the horizontal axis. As canbe seen, at long distances from the center of the flame(x > 8mm), the signal results from room air. Theroom air heats up with proximity to the flame front,located around x ¼ 3mm in Fig. 6, resulting inRCARS spectra that extend to larger Raman shifts(higher J levels). In Fig. 6(a), a horizontal streakcan be seen across the entire image. This interfer-ence results from both the nonresonant backgroundand smeared-VCARS contributions. A moderateprobe delay of just 100ps is adequate to effectivelysuppress this interference signal, as can be seen inFig. 6(b). As low-J rotational coherences dephase ona faster time scale than high-J coherences, the low-Jlines in an RCARS spectrum lose intensity fasterupon a probe delay than the high-J lines, leaving a
Fig. 4. (Color online) Raw RCARS 1D image taken in room airwith 50 laser shots. (a) CCD image. (b) Single-column intensity dis-tribution along a N2 Raman line, demonstrating the effectivelength of the imaged region. (c) Single pixel row spectrum.
spectrum that appears hotter [23]. It has been shown[23,24] that probe delays of up to 160ps (when usinga 100ps pulse width) haveminimal to no effect on theevaluated temperature and relative O2:N2 ratio intime-resolved picosecond RCARS experiments, evenat room temperature where the effect of the probe de-lay is most dramatic. Spectral heating becomes a con-cern at longer delays, for which differences in the J-dependent and species-specific coherence dephasingrate begin to impact the fitted temperature and re-lative O2:N2 concentration. We assessed the effectof a 100ps probe delay with the current pulse widthsin a temperature-controlled oven from room tem-perature to 1420K in N2 and found delay-inducedspectral heating of 25–45K, respectively.
Figure 7 displays a fitted spectrum taken from asingle pixel row in the image acquired at an axial
height of y ¼ 5mm near the flame front of the pre-mixed laminar propane flame, evaluated to be1930K. Figure 8 displays the evaluated profiles ob-tained from the CCD images at axial heights of y ¼ 5and y ¼ 15mm in the flame. The maximum fittedtemperature seen in Fig. 8(a) at a height of 5mmis 1985K, whereas in Fig. 8(b) at a height of15mm, the maximum flame temperature is evalu-ated to be 2090K. The difference of 105K in Tmaxfor the two profiles is possibly due to heat loss withproximity to the burner surface. In previous VCARSimaging experiments, temperatures 200K below theadiabatic limit were found 2mm from the burner exitin a CH4=air flame [18] over a Bunsen-style burner.As an estimate, the maximum flame temperaturewas calculated using the OppDiff code of the Chem-kin package [32] utilizing the propane combustion
Fig. 5. Evaluated temperature distributions for a single pixel row spectrum taken from 100 single shots in a heated flow of N2. Evalua-tions were performed at (a) 408, (b) 697, (c) 985, and (d) 1205K. The thermocouple temperature, average RCARS-evaluated temperature,and standard deviation for single-shot measurements are shown.
Fig. 6. (Color online) CCD images taken in the laminar propane flame. The probe delay in these experiments was set to (a) 0 and(b) 100ps.
mechanism found in [33]. The simulated tempera-ture profile resulted in a maximum temperature of2130K, which is 40K higher than the maximumtemperature evaluated from the RCARS image, a de-viation of 1.9%. As can be seen in Fig. 2, the richflame exhibits both an inner flame front and a faintouter flame due to the incomplete combustion causedby excess fuel. The region between these two zonesincreases with increased axial height, and the profilein Fig. 8(b) exhibits a wider high-temperature com-bustion zone than does Fig. 8(a). The error bars inFig. 8 are obtained by evaluating separate 1DRCARS images taken at the same location, each with1000 laser shots average, shown as ð�Þ2σ and do nottake into account any systematic error to the ps-RCARS technique, which is analyzed later in theheated flow measurements. The signal in the flamestudies at the given laser pulse energies was not ade-quate for single-shot evaluations at the high flametemperature (≈2100K) observed in this flame.
By averaging multiple laser shots, RCARS 1Dimages were obtained in a flame with evaluated tem-peratures as high as 2100K. On a single-shot basis,usable measurements were limited to approximately1200K for adequate signal counts for probing at
1 atm pressure with the current optical setup. Giventhat the signal intensity in this experiment scales as∝ 1=L3, where L (Δx in Fig. 1) is the spatial length ofthe 1D CARS image, reducing the beam sizes so thatthe 1D image length is reduced by a factor of 2 resultsin a factor of 8 increase in the signal. Therefore, theRCARS imaging setup used in this work may be sui-table for single-shot measurements at adiabaticflame temperatures over 2000K if the laser sheetsare reduced to 3–5mm when at room temperature,or under higher pressure conditions.
Up to now, such imaging capability using RCARShas been elusive because of low signal levels. RCARSsuffers greater signal loss at elevated temperaturesthan does VCARS due to the increased spectralspreading and decreasing population differences be-tween rotational levels. Further, in generating thedual-broadband pulses, more available laser energyis lost to the inefficiency of the dye laser than inVCARS, where just one broadband pulse is needed.The use of picosecond pulses allows for higher irra-diance, and thus higher signal levels, in the imagingconfiguration when compared to typical nanosecondsystems. The use of the time-resolved capability ofpicosecond RCARS further allows for the suppres-sion of interference from the nonresonant back-ground and smeared VCARS. In principle, the sameadvantage in signal level gained by using picosecondpulses to drive the rotational coherence could be in-creased further by driving the coherence with femto-second pulses. If the coherence is then probed with apicosecond pulse, an adequate spectral resolution isstill available to gain the temperature and relativespecies concentration gradients [34]. The broadbandnature of the transform-limited femtosecond pulseswould also obviate the need for a broadbanddye laser.
In principle, this technique could also be used togenerate absolute species concentrations, provideda suitable technique for accounting for the spatialfluctuations within the laser pulses is employed.For application to the imaging of turbulent flameenvironments, a careful analysis of beam-steering
Fig. 7. (Color online) Single pixel row from the image in Fig. 6(b).The experimental data are in black, and the theoretical evaluationis in red, evaluated to be 1930K at this location.
Fig. 8. (Color online) Temperature (black curve) and relative O2-to-N2 ratio profiles (red dotted curve) evaluated from 1000 laser shots atan axial height of (a) 5 and (b) 15mm in the premixed laminar propane flame. Images were taken with a 100ps probe delay. Error bars(�2σ) are shown, as calculated by analyzing repeated experiments.
effects is needed because of the fluctuating refractiveindex gradients.
4. Summary
The presented picosecond RCARS imaging techniqueholds promise for single-shot imaging of combustion,revealing both temperature and relative major spe-cies concentrations with the high precision and accu-racy of the CARS technique. The technique wasdemonstrated as a combustion diagnostic in a richlaminar premixed propane flame with evaluatedtemperatures near 2100K. The technique providedboth temperature and relative O2:N2 concentrationprofiles along a line of 330 volume elements withina single image. Imaged regions of up to 12mm werepresented at 300K on a single-shot basis, and thisreduced to 5mm at 1200K. The optical setup was ar-ranged to probe along either a vertical or horizontalline, allowing flexibility in the particular application.The interference-suppression advantages of probe-delayed picosecond CARS were exploited in the ima-ging geometry to avoid signal contributions from thenonresonant susceptibility and smeared VCARS in arich propane flame. The precision and accuracy of pi-cosecond 1D RCARS were assessed by acquiring 100single-shot images in a heated flow of N2, and theywere found to be comparable with reported valuesfor nanosecond-laser-based VCARS at temperaturesup to 1200K.
References1. A. Eckbreth, Laser Diagnostics for Combustion Temperature
and Species (Abacus, 1988).2. M. Aldén, P.-E. Bengtsson, and H. Edner, “Rotational CARS
generation through a multiple four-color interaction,” Appl.Opt. 25, 4493–4500 (1986).
3. A. C. Eckbreth and T. J. Anderson, “Simultaneous rotationalcoherent anti-Stokes Raman spectroscopy and coherentStokes Raman spectroscopy with arbitrary pump-Stokes spec-tral separation,” Opt. Lett. 11, 496–498 (1986).
4. T. Seeger and A. Leipertz, “Experimental comparison of sin-gle-shot broadband vibrational and dual-broadband pure rota-tional coherent anti-Stokes Raman scattering in hot air,”Appl.Opt. 35, 2665–2671 (1996).
5. F. Vestin, M. Afzelius, and P.-E. Bengtsson, “Rotational CARSfor simultaneous measurements of temperature and concen-trations of N2, O2, CO, and CO2 demonstrated in a CO/air dif-fusion flame,” Proc. Combust. Inst. 32, 847–854 (2009).
6. M. C. Weikl, F. Beyrau, and A. Leipertz, “Simultaneous tem-perature and exhaust-gas recirculation-measurements in ahomogeneous charge-compression ignition engine by use ofpure rotational coherent anti-Stokes Raman spectroscopy,”Appl. Opt. 45, 3646–3651 (2006).
7. J. Shirley, R. Hall, J. Verdieck, and A. Eckbreth, “New direc-tions in CARS diagnostics for combustion,” in Proceedings ofthe Fifteenth American Institute of Aeronautics and Astronau-tics Thermophysics Conference (1979), paper 80–1542,pp. 1–13.
8. J. Zheng, J. B. Snow, D. V. Murphy, A. Leipertz, R. K. Chang,and R. L. Farrow, “Experimental comparison of broadbandrotational coherent anti-Stokes Raman scattering (CARS)and broadband vibrational CARS in a flame,” Opt. Lett. 9,341–343 (1984).
9. R. P. Lucht, “Three-laser coherent anti-Stokes Raman-scattering measurements of two species,” Opt. Lett. 12, 78–80(1987).
10. S. P. Kearney, K. Frederickson, and T. W. Grasser, “Dual-pumpcoherent anti-Stokes Raman scattering thermometry in asooting turbulent pool fire,” Proc. Combust. Inst. 32, 871–878(2009).
11. A. C. Eckbreth, “BOXCARS—crossed-beam phase-matchedCARS generation in gases,” Appl. Phys. Lett. 32, 421–423(1978).
12. S. A. Tedder, M. C. Weikl, T. Seeger, and A. Leipertz, “Deter-mination of probe volume dimensions in coherent measure-ment techniques,” Appl. Opt. 47, 6601–6605 (2008).
13. D. V. Murphy, M. B. Long, R. K. Chang, and A. C. Eckbreth,“Spatially resolved coherent anti-stokes Raman spectro-scopy from a line across a CH4 jet,” Opt. Lett. 4, 167–169(1979).
14. H. W. Schrötter and H. W. Klöckner, “Raman scattering crosssections in gases and liquids,” in Raman Spectroscopy ofGases and Liquids A. Weber, ed. (Springer-Verlag, 1979),pp. 123–166.
15. J. B. Snow, J. Zheng, and R. K. Chang, “Spatially andspectrally resolved multipoint coherent anti-Stokes Ramanscattering from N2 and O2 flows,” Opt. Lett. 8, 599–601(1983).
16. J. Stufflebeam and A. C. Eckbreth, “CARS temperature andspecies measurements in propellant flames,” in NonintrusiveCombustion Diagnostics, K. Kuo and T. Parr, eds. (BegellHouse, 1994), pp. 115–131.
17. J. Jonuscheit, A. Thumann, M. Schenk, T. Seeger, and A.Leipertz, “One-dimensional vibrational coherent anti-StokesRaman-scattering thermometry,” Opt. Lett. 21, 1532–1534(1996).
18. J. Jonuscheit, A. Thumann, M. Schenk, T. Seeger, and A.Leipertz, “Accuracy and precision of single-pulse one-dimensional vibrational coherent anti-Stokes Raman-scattering temperature measurements,” Appl. Opt. 36,3253–3259 (1997).
19. J. Bood, C. Brackmann, P.-E. Bengtsson, andM. Aldén, “Multi-point temperature and oxygen-concentration measurementsusing rotational coherent anti-Stokes Raman spectroscopy,”Opt. Lett. 25, 1535–1537 (2000).
20. S. Roy, T. R. Meyer, and J. R. Gord, “Broadband coherentanti-Stokes Raman scattering spectroscopy of nitrogen usinga picosecond modeless dye laser,” Opt. Lett. 30, 3222–3224(2005).
21. S. Roy, T. R. Meyer, and J. R. Gord, “Time-resolved dynamics ofresonant and nonresonant broadband picosecond coherentanti-Stokes Raman scattering signals,” Appl. Phys. Lett. 87,264103 (2005).
22. T. R. Meyer, S. Roy, and J. R. Gord, “Improving signal-to-interference ratio in rich hydrocarbon-air flames usingpicosecond coherent anti-Stokes Raman scattering,” Appl.Spectrosc. 61, 1135–1140 (2007).
23. T. Seeger, J. Kiefer, A. Leipertz, B. D. Patterson, C. J. Kliewer,and T. B. Settersten, “Picosecond time-resolved pure-rotational coherent anti-Stokes Raman spectroscopy for N2
thermometry,” Opt. Lett. 34, 3755–3757 (2009).24. T. Seeger, J. Kiefer, Y. Gao, B. D. Patterson, C. J. Kliewer, and
T. B. Settersten, “Suppression of Raman-resonant interfer-ences in rotational coherent anti-Stokes Raman spectroscopyusing time-delayed picosecond probe pulses,” Opt. Lett. 35,2040–2042 (2010).
25. S. Roy, J. R. Gord, and A. K. Patnaik, “Recent advances in co-herent anti-Stokes Raman scattering spectroscopy: funda-mental developments and applications in reacting flows,”Prog. Energy Combust. Sci. 36, 280–306 (2010).
26. C. J. Kliewer, Y. Gao, T. Seeger, J. Kiefer, B. D. Patterson, andT. B. Settersten, “Picosecond time-resolved pure-rotationalcoherent anti-Stokes Raman spectroscopy in sooting flames,”Proc. Combust. Inst. 33, 831–838 (2011).
27. W. D. Kulatilaka, P. S. Hsu, H. U. Stauffer, J. R. Gord, and S.Roy, “Direct measurement of rotationally resolved H2
28. T. Seeger, F. Beyrau, A. Bräuer, and A. Leipertz, “High-pres-sure pure rotational CARS: comparison of temperature mea-surements with O2, N2 and synthetic air,” J. Raman Spectrosc.34, 932–939 (2003).
29. E. Magens, “Nutzung von Rotations-CARS zur Temperatur-und Konzentrations-messung in Flammen,” Ph.D. thesis(Universität Erlangen-Nürnberg, 1992).
30. M. L. Koszykowski, L. A. Rahn, R. E. Palmer, and M. E.Coltrin, “Theoretical and experimental studies of high-
resolution inverse Raman spectra of N2 at 1–10atm,” J. Chem.Phys. 91, 41–46 (1987).
31. L. A. Rahn and R. Palmer, “Studies of nitrogen self-broadeningat high temperature with inverse Raman spectroscopy,” J.Opt. Soc. Am. B 3, 1164–1169 (1986).
32. R. Kee, F. Rupley, E. Meeks, and J. Miller, “CHEMKIN-III: aFortran chemical kinetics package for the analysis of gas-phase chemical and plasma kinetics,” Tech. Rep. SAND96-8216 (Sandia National Laboratories, 1996).
33. “Chemical–kinetic mechanisms for combustion applications,” ,Center forEnergyResearch (CombustionDivision),Universityof California at San Diego, http://maeweb.ucsd.edu/combustion/.
34. J. D. Miller, M. N. Slipchenko, T. R. Meyer, H. U. Stauffer,and J. R. Gord, “Hybrid femtosecond/picosecondcoherent anti-Stokes Raman scattering for high-speedgas-phase thermometry,” Opt. Lett. 35, 2430–2432(2010).
