Intracavity laser absorption spectroscopy of flames: concentration measurements of intermediate species

Proceedings

Proceedings of the Combustion Institute 30 (2005) 1575–1582

www.elsevier.com/locate/proci

of the

CombustionInstitute

Intracavity laser absorption spectroscopy of NH2

in methane/air flames doped with N2O, NO, and NH3

I. Rahinov, N. Ditzian, A. Goldman, S. Cheskis*

School of Chemistry, Sackler Faculty of Exact Sciences, Tel Aviv University, Ramat Aviv, Tel Aviv 69978, Israel

Abstract

Intracavity laser absorption spectroscopy (ICLAS) was used to measure concentration profiles of NH2

in low pressure (30 Torr) methane/oxygen/nitrogen flames doped with small amounts of N2O, NO, andNH3. The effective optical length of ICLAS, which dictates the sensitivity of the method, is controlledby the generation time of the quasi-cw-laser. The effective optical length of 0.87 km, reached at generationtime of 75 ls, provides very high sensitivity: from 5 · 1010 molecules/cm3 at 500 K in the vicinity of burnerup to 2 · 1011 molecules/cm3 in the burned gas zone (T � 1800 K). The radial profile of NH2, measuredusing a tomographic technique, indicates that the radical is located mainly inside a cylinder with diameterequal to the burner diameter. For the first time, the absolute mole fraction profiles of NH2 were measuredin hydrocarbon flames with different dopants and compared with one-dimensional calculations based onthe GRI-3.0 mechanism. The dependence of the NH2 concentrations on the equivalence ratio is in excellentagreement with the model prediction. The absolute NH2 concentration values for different dopants are pre-dicted very well by the calculations apart from the NO-doped flame where GRI-Mech 3.0 overpredicts theNH2 concentration by a factor of 2. In the ammonia-doped flame, the calculations predict an additionalconcentration maximum located close to the burner. This maximum is not observable in the experiment.All the experimental profiles exhibit 1–2 mm shift further from the burner surface in comparison with thepredicted ones. The reasons of those discrepancies are discussed.� 2004 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Laser diagnostics; Spectroscopy; Laminar flames; Nitrogen chemistry

1. Introduction

The NH2 radical, amidogen, plays an impor-tant role in nitrogen chemistry of flames and inthe post-combustion environment. NH2 is postu-lated to be a key intermediate in determining thefate of fuel nitrogen in combustion systems [1].The NH2 reactions dictate whether fuel nitrogenis converted to nitric oxide or to molecular nitro-

1540-7489/$ - see front matter � 2004 The Combustion Institdoi:10.1016/j.proci.2004.07.027

* Corresponding author. Fax: +972 3 640 9293.E-mail address: [email protected] (S. Cheskis).

gen. A fast reaction of amidogen with NO is thekey reaction in the processes of selective non-cat-alytic reduction (SNCR) methods of NO removalfrom exhaust gases [2]. This radical also plays animportant role in metal organic chemical vapordeposition (MOCVD) of gallium nitride (GaN)[3,4]. The measurements of NH2 in flames and inother reactive environments are very importantfor elucidation of the details of the processes men-tioned above. NH2 has an intensive spectrum inthe visible region [5,6], which can be used for con-centration measurements. NH2 has been moni-tored in flames by conventional laser absorption

ute. Published by Elsevier Inc. All rights reserved.

Fig. 1. Schematics of the experimental arrangement.ICLAS: The pumping argon-ion laser beam is focusedon the dye jet by mirror M3. The homemade dye lasercavity is formed by M1, M2, and OC (Output Coupler)mirrors. AOM1 and AOM2 are the Acouso-OpticModulators. PCCD is the Photodiode Charge CoupledDevice. LIF: SHG, second harmonic generation crystal;F, Schott Glass U-340 filter; D, diaphragm; and PD,post-flame photodiode.

1576 I. Rahinov et al. / Proceedings of the Combustion Institute 30 (2005) 1575–1582

techniques [1,7], and by laser-induced fluorescence(LIF) [8–11]. The amidogen radical has also beenmonitored at high temperatures in shock tubes bylaser absorption spectroscopy and frequencymodulation (FM) spectroscopy [12–16]. However,the sensitivity of the conventional absorptionspectroscopy is insufficient for measurements ofthe trace amounts of NH2. The sensitivity ofLIF, which is one of the most sensitive techniquesof laser spectroscopy, is lower for NH2 than formany other species due to very effective quenchingof the upper electronic state of amidogen in colli-sions with other molecules. Even for rare gases thequenching rate constant is close to the gas kineticvalue [17]. Williams and Fleming [11] used LIF tostudy CH4/O2/Ar flames doped with NO, CH3CN,NH3, and CH3NH2. They measured the concen-tration profiles of CN, NH, NCO, NH2, andNO in these flames. Due to a weak NH2 fluores-cence signal, the amidogen spectra were recordedonly in the ammonia- and methylamine-dopedflames. The authors also needed to add these do-pants in three times their normal concentration(1.05%).

In this work, we applied intracavity laserabsorption spectroscopy (ICLAS) to determinethe absolute concentration profiles of amidogenin low pressure CH4/O2/N2 flames doped withnitrogen oxides and ammonia. ICLAS, whichwas demonstrated to be especially suitable forNH2 detection [18–20], does not suffer from thedrawbacks in sensitivity, characteristic for LIF.Additional advantage of the ICLAS technique isa possibility of direct measurement of absoluteconcentrations, which are crucial for validationof the chemical mechanisms of the combustionprocesses. In our previous studies [18,19], we ob-served presence of NH2 at large distances abovethe burner, where according to theoretical predic-tions its concentration should be negligible. It wasproposed that this discrepancy related to theabsorption of NH2, which is located outside thecentral flame zone. In the present work, we useda different vacuum chamber that allows us tostudy a radial dependence of the NH2 concentra-tion, and for the first time we obtained absolutemole fraction profiles of amidogen, which areapplicable for comparison with predictions ofone-dimensional model.

2. Experimental

Figure 1 shows the schematics of the experi-mental apparatus used in this work.

2.1. Flame apparatus

Low pressure flames were supported on the 6-cm diameter stainless steel porous plug watercooled McKenna burner located in the evacuated

chamber (40 cm inner diameter). The burner wasmounted on a PC-controlled motorized position-ing unit (two LM 60, Owis-Staufen), which allowsthe burner movement along the vertical and hori-zontal axis with accuracy of 1/64 mm. The gasflows were regulated by calibrated mass-flow con-trollers (1179A and 1259C, MKS Instruments). Anitrogen co-flow was separately connected to thesliding outer jacket of the burner. A feedbackvalve controller (Model 252/253, MKS Instru-ments) with an exhaust throttle valve regulatedthe gas pressure in the chamber. Flames withequivalence ratios of 0.8, 1.0, and 1.17 were inves-tigated. All experiments were performed at30 Torr pressure in the combustion chamber.

2.2. Intracavity apparatus

The optical arrangement of the ICLAS is basedon a quasi-cw homemade dye laser. The flame islocated inside the laser cavity and isolated fromthe other parts of the cavity by two glass windowsat Brewster angle. The windows are 12.6 mm thickwith 1� wedge to minimize interference fringes inthe laser spectrum. The dye jet (Rhodamine 590,Exciton Laser Dyes), pumped by an argon-ionlaser (Coherent Model Innova 90-6), is the lasingmedium located inside the astigmatically compen-sated folded cavity formed by three mirrors.

Fig. 2. Absorbance measured at the pQ1,N (7) line ofNH2 (marked with an asterisk on the spectra shown onthe inset) versus the generation time of the broadbandlaser source. On the inset: the NH2 radical spectrarecorded in the rich (u = 1.17) methane/air flame dopedwith 2.25% N2O at different generation times.

I. Rahinov et al. / Proceedings of the Combustion Institute 30 (2005) 1575–1582 1577

Mirrors M1 and M2 have 75 mm radii of curva-ture and �99% reflectivity, the output couplermirror (OC) is planar with �98% reflectivity andwedged to minimize interference effects, the lengthof the laser cavity is 156 cm. The tuning of thecentral wavelength of the broadband dye lasersource was accomplished by varying the dye con-centration, alignment of the laser cavity, and par-tially eclipsing the laser beam inside the cavity bydiaphragm that introduced broadband losses,shifting the laser generation profile to a shorterwavelength. When a narrowband absorber is lo-cated inside the laser cavity, the spectrum of abroadband laser source is changed, according tothe following equation (see, for example, the re-view of Baev et al. [21] and references therein):

Iðm; tgÞ ¼ Ieðm; tgÞ exp½�nrðmÞLeqðtgÞ�; ð1Þwhere Ie (m, tg), ‘‘the spectral envelope,’’ is a rela-tively slow-varying function of the wavenumber;r (m) is the narrowband absorption cross-section;and n is the concentration of the absorbing spe-cies. The ‘‘envelope function,’’ Ie (m, tg), is anequivalent of the ‘‘blank’’ spectrum in the classicalabsorption spectroscopy. The absorber species inthe flame occupy only a fraction of the totallength of the cavity, therefore the equivalent path-length is given by

LeqðtgÞ ¼lLctg; ð2Þ

where l is the optical length of the absorbing com-pound (the flame diameter, which is supposed tobe equal to the diameter of the burner, i.e.,6 cm), L is the length of laser cavity, tg is the gen-eration time, and c is the velocity of light. Thegeneration time was controlled with the aid oftwo acousto-optic modulators (AOM) (AFM-40IntraAction). The modulators operation is syn-chronized by pulse/delay generator (Model 555Berkeley Nucleonics). The first modulator(AOM1) chopped the argon-ion laser and pro-vided dye laser pulses with typical duration of20–75 ls and delay of 10 ls between the consecu-tive pulses (i.e., with repetition rate of 12–33 kHz).The second modulator (AOM2) sent the laserbeam during the sampling time of about the last5–10 ls of the AOM1 pulse to the entrance slitof the high resolution 1 m spectrograph (SPEX1000 M). Figure 2 demonstrates the measuredabsorbance of the pQ1,N (7) line of NH2 versusgeneration time. A linear dependence of theabsorbance on the generation time was observed,as predicted in Eq. (2), at least until 200 ls. Thelaser radiation is dispersed by 100 grooves/mmechelle grating working in the 29th order. Theuse of echelle grating at high order and high inci-dence angle allows to reach high resolution anddispersion (�0.05 cm�1/pixel). The spectrum is de-tected on a 2048-element photodiode charge cou-

pled device (PCCD) (ALTON-Model LS-2000).The spectral resolution is about 0.003 nm as mea-sured by observing He–Ne laser line. The infor-mation from the PCCD is sent to a computerfor data processing and storing. The exposuretime of the PCCD was typically 300 ms, whichpermitted averaging of 3500–10,000 spectra. Anadditional computer averaging of 50 PCCD sig-nals was used. The background signal of thePCCD (less than 1–2% of the total signal) wassubtracted ‘‘on the fly’’ for each exposure time.The ICLAS intensity spectrum contained besidesthe NH2 lines, spectral features from the atmo-spheric absorption (mainly water overtone spec-tra), as well as the spectral modulation of thelaser losses, and the interference on the flat win-dow of the PCCD. Figure 3 illustrates the proce-dure used for separation of the amidogen radicalspectrum from those features. The experimentalspectrum, I (m, tg), which contains amidogen spec-tral features, was divided by the reference spec-trum, Ie (m, tg) (both spectra appear on the upperpanel of Fig. 3). The resulting NH2 spectrum,I (m, tg)/Ie (m, tg), obtained in this manner is shownon the lower panel of Fig. 3. The reference spec-trum, Ie (m, tg), was recorded at high altitudesabove the burner (typically �70 mm), where noradical absorbing species were present, or on eachcorresponding height above the burner, when thesupply of N2O, NO, or NH3 was switched off. We

Fig. 3. Evaluation of NH2 spectrum. On the upper panel: (a) Ie (m, tg): reference spectrum recorded at 77 mm above theburner surface, where no absorbing radicals are present. This spectrum contains overtone water lines (marked by ticks inthe uppermost part of the upper panel according to HITRAN database [22]). (b) I (m, tg): spectrum recorded at 5 mmabove the burner surface, which contains NH2 lines along with water overtone lines. On the lower panel: (c)

Iðm;tgÞIeðm;tgÞ:

spectrum containing amidogen spectral features marked by ticks with dropped dashed lines according to [5]. The arrowmarks the position of the pQ1,N (7) line of NH2, that was used for the absolute concentration determination. The spectragiven here were measured in a stoichiometric (u = 1.00) CH4/O2/N2 flame doped with 0.2% ammonia. The lasergeneration time was 20 ls.

1578 I. Rahinov et al. / Proceedings of the Combustion Institute 30 (2005) 1575–1582

observed no difference between the amidogenspectra obtained in those two manners. The spec-trograph was calibrated with the aid of knownpositions of the atmospheric water absorptionlines tabulated in HITRAN database [22]. Thediameter of the laser beam in ICLAS, which dic-tates the spatial resolution of the method, wasmeasured using 2D-CCD camera (SensiCamSVGA, PCO CCD Imaging) aligned to view thelaser beam, emerging from the homemade dye la-ser cavity through the output coupler [19]. Thediameter was found to be 6600 lm (FWHM).

2.3. Temperature measurements

The temperature profiles are necessary both formodel calculations and the evaluation of absoluteconcentrations from the recorded experimentalspectra. The excitation spectra of OH were usedfor temperature determination. We used a tunabledye laser (Continuum, ND60), operating withDCM dye, pumped by the second harmonics ofthe Nd:YAG laser (Surelite I, Continuum). Thedye laser fundamental was doubled by KDP crys-tal and applied to scan the spectral range of306.7–307.07 nm, containing the OH radical tran-sitions related to the R-branch of (0,0) band of theA2R+–X2Pi system. The spectral transitions lo-cated in this range cover rotational quantum num-bers from 2 to 15 and are very convenient fortemperature determination [23,24]. The laserbeam was apertured by diaphragm and passedthe flame unfocused. The energy of the laser pulsewas typically 0.2–0.6 lJ, to avoid saturation

effects. The typical diameter of the laser beam inthe position corresponding to the center of theburner was determined to be 6500 lm (FWHM).The fluorescence was detected at right angle to thelaser beam by a gated intensified CCD (ICCD)camera. We used a narrow (typically 20 ns) detec-tor gate, which started immediately after the laserpulse. No spectral filtering of the fluorescence wasused. The signal of the post-flame photodiode(PD) was digitized by oscilloscope (LeCroy Wave-runner, LT 262, 350 MHz, 1 Gs/s) and used fornormalization of the obtained OH spectra, the lat-ter intensity was found to be linearly dependenton the incident laser intensity. The typical statisti-cal error of the obtained Boltzmann plots did notexceed 50 K.

3. Results and discussion

3.1. Absolute concentrations

The lower panel of Fig. 3 shows a fragment ofthe NH2 spectrum recorded in the stoichiometric(u = 1.00) flame doped with 0.2% NH3. We usedthe pQ1,N (7) line of the (090)–(000) band

belonging to the eA2A1–eX2

B1 electronic transitionfor determination of the absolute concentrations.This line, located at 16,739.90 cm�1, is ratherintensive and not overlapping with lines of differ-ent rotational quantum numbers and water over-tone lines (see Fig. 3). Several measurements ofthe absorption coefficient have been done for thisline [7,12,16]. The correlation between the total

Fig. 4. Experimental mole fraction profiles (openedcircles) along with model prediction results (GRI 3.0-Mech) (solid line) in stoichiometric (u = 1.00) CH4/O2/N2 flame doped with: (A) 0.14% NH3; (B) 0.34% NO;and (C) 2.25% N2O. Dashed line on (A) represents theprediction of the GRI 3.0-Mech modified with additionof the NH2 + HO2reaction.

I. Rahinov et al. / Proceedings of the Combustion Institute 30 (2005) 1575–1582 1579

number density of the NH2 radical and theabsorption value obtained from the experimentalspectrum in the ICLAS experiment is given by:

½NH2�total ¼A

ðl=LÞctgrðT ;DmÞ; ð3Þ

where A is the absorbance value, obtained exper-imentally, l is an optical length of the absorbingcompound, L is the length of the homemade dyelaser cavity, c is the light velocity, and r (T,Dm)is a temperature dependent absorption coefficient,which was calculated using a known oscillatorstrength of the pQ1,N (7) line [12]. The spectral lineis fitted well by a Gaussian profile, so that

r T ;Dmð Þ ¼ pe2

mc2f

4 ln 2

p

� �1=21

DmF B Tð Þ; ð4Þ

where f = 6.35·10�5 is the oscillator strength of thetransition [12], e is the electron charge,m is the elec-tron mass, Dm is the FWHMof the spectral line ob-tained from the Gaussian best fit to experimentallineshape, and FB(T) is the Boltzmann factor.

3.2. Influence of different dopants

The mass-flow rates of methane, oxygen, nitro-gen, and the dopants in the gas mixture suppliedto the burner and producing the flames studiedin this work are given in Table 1. Note that wewere able to observe NH2 in ammonia-dopedflame with a signal-to-noise ratio of about 100,while the mole fraction of ammonia was 0.14%.The typical noise level in the recorded ICLASspectra allows us to estimate the minimum detect-able number density of the NH2 radical: from�5 · 1010 molecules/cm3 in the vicinity of the bur-ner (T � 500 K) to �2 · 1011 molecules/cm3 in theburned gases zone (T � 1800 K).

The experimental profiles of NH2 are shown inFig. 4 along with the calculated ones. The calcula-tions were performed with the PREMIX code [25]and the GRI 3.0 combustion mechanism [26].There are no substantial changes in the calculatedprofiles when the binary and thermal diffusionare taken into account. We did not try to includepossible heterogeneous processes on the burnersurface in our calculations. The measured temper-ature profiles were used as inputs for the model.We used the nominal diameter of the McKennaburner (i.e., 6 cm) as the length of absorbing com-

Table 1The parameters of the studied flames: the fuel equivalence rati

Equivalence ratio Dopant u = 1.17

CH4

sccmO2

sccmN2

sccm

NH3 6 sccmNO 15 sccm 780 1330 3420N2O 100 sccm

pound in evaluation of the absolute concentra-tions from the experimental absorbance values.The tomographic measurements show that�90% of the NH2 absorbance is due to the radicallocated within the 6 cm of the nominal burnerdiameter.

All experimental profiles exhibit 1–2 mm shifttowards higher position above the burner in com-parison with model prediction. This shift is spe-cific for NH2, and is not common to otherradicals present in the flame. This shift is higherthan usually observed in line-of-sight measure-ments due to flame unflatness (�0.65 mm forCH as observed by Luque et al. [27]). The modelprediction agrees well with the measured positionof HCO profile maximum as illustrated in Fig. 5,

o u = 2[CH4]/[O2]; the mass-flow rates are given in sccm

u = 1.00 u = 0.80

CH4

sccmO2

sccmN2

sccmCH4

sccmO2

sccmN2

sccm

450 900 2990 360 900 2990

Fig. 5. Measured HCO mole fraction (opened circles)along with calculated (GRI 3.0-Mech) mole fractionprofile (solid line) and the measured temperature profile(dashed line).

Fig. 6. Measured mole fraction of NH2 (opened circles)compared to calculation by GRI 3.0-Mech (solid line).The plots are given for N2O-doped flames with differentequivalence ratios: (A) u = 1.17; (B) u = 1.00; and(C) / = 0.80.

1580 I. Rahinov et al. / Proceedings of the Combustion Institute 30 (2005) 1575–1582

even though the maximum HCO concentration islocated in the zone of the steep temperature gradi-ent. A similar shift of the NH2 profile in theammonia-doped flame was also observed by Wil-liams and Fleming [11].

In the ammonia-doped flame, the calculationspredict two peaks for the NH2 profile. The firstpeak is located very close to the burner and causedmainly by the reaction OH + NH3 fi NH2 +H2O. We do not observe this peak in the experi-ment. The appearance of the first peak in the pre-dicted NH2 profile exhibits high sensitivity to thetemperature. If the temperature in the first fewmillimeters above the burner surface in the inputtemperature profile is raised by about �200 K,the first peak decreases by about factor of �3.Having that in mind, we performed a careful tem-perature profile evaluation in the pre-heat zone byOH-LIF as well as by NO-LIF (doping the flamewith less than 0.05% NO). The temperature pro-files measured by NO-LIF and OH–LIF agreewithin 20–30 K. Thus, it seems to be not an arti-fact and the discrepancy can be caused by hetero-geneous processes at very small distance to theburner surface as well as by the deficiency in thegas-phase chemical mechanism. The GRI-Mech3.0 does not take into account the reactionbetween HO2 and NH2. The mole fraction ofHO2 in the pre-flame zone is rather high, and cal-culation with experimentally measured rate con-stant of this reaction results in the effectiveelimination of the first peak of NH2. Figure 4shows that calculations with the reactionNH2 + HO2 (k = 2.5 · 1013 cm3 mol�1 s�1 [2])added to GRI-Mech 3.0 substantially reducesthe first NH2 peak, improving the agreement withthe experiment. The absolute concentration valuesare predicted very well by GRI-Mech 3.0 apartfrom the NO-doped flame, where the calculationsoverpredict the experimental mole fraction byabout a factor of 2. The Miller–Bowman mecha-nism (MB) [28] strongly underpredicts (by more

than an order of magnitude) the NH2 absoluteconcentration in NO- and N2O-doped flames.According to GRI-Mech, NH2 is formed at themaximum of the concentration profile by the reac-tion HNCO + H@NH2 + CO and is consumed bythe reactions with H, OH, and O. It is possible tofit the experimental results with the calculationsby varying these constants within their uncer-tainty limits. A deeper analysis is needed to findthe most appropriate way to optimize the GRI-Mech. This analysis must take into account therequirement of simultaneous fitting the NH2 pro-files in the flames doped with both N2O and NO,where the chemical mechanism seems to be rathersimilar. The calculations for the NH3-doped flameproduce very similar results both for MB andGRI-Mech 3.0. Taking into account differentamounts of the added dopants (see Table 1) therelative efficiency of amidogen production for dif-ferent dopants is in reasonable agreement with theGRI-Mech 3.0 calculation results.

3.3. Influence of the equivalence ratio

Figure 6 presents the mole fraction profiles ofNH2 recorded in the N2O-doped flames with dif-ferent equivalence ratios. The calculated valuesof the NH2 mole fraction agree very well with

I. Rahinov et al. / Proceedings of the Combustion Institute 30 (2005) 1575–1582 1581

the measured ones. Note that we observe �4-foldincrease of the amidogen mole fraction when thepremixed gases composition is changed from lean(u = 0.80) to rich (u = 1.17). This phenomenon isin good quantitative agreement with the predic-tion of GRI 3.0 mechanism.

3.4. Effect of chamber geometry

In our previous measurements [18,19], we ob-served relatively high concentrations of NH2 mea-sured at the large distances above the burner (upto 50 mm). Such behavior is not predicted byany one-dimensional model. We proposed thatthis effect can be explained by the presence ofNH2 in the zone surrounding the flame. In thepresent work, we used a vacuum chamber whichis much wider than that was used previously(40 cm diameter instead of 16 cm). In the presentexperimental configuration, it was also possibleto measure the radial profile of NH2 using a tomo-graphic technique by moving the burner in thedirection perpendicular to the laser beam. Wefound that NH2 is located mainly inside of the cyl-inder with a diameter equal to the burner diame-ter. We do not observe NH2 at distances above�25 mm above the burner in agreement with cal-culations (see Figs. 4 and 6). Thus, the NH2 ob-served previously at high distances was indeedrelated to the flame confinement by the vacuumchamber. To check it we placed a metal disk di-rectly against the burner center and at �30 mmdistance above it. We observed a strong increaseof NH2 concentration in the vicinity of this diskin the NO-doped flame. Possible influence of dif-ferent surfaces and confinements on NH2 concen-tration is the subject of the ongoing research,nevertheless, the tomographic measurements,and a good agreement with the calculations pointto the absence of such influence in the presentexperimental system.

4. Conclusions

Absolute concentrations of the NH2 radicalhave been determined in low-pressure CH4/O2/N2 flames doped with ammonia and nitrogen oxi-des using ICLAS. The NH2 spectra were obtainedwith very good signal-to-noise ratio with theamount of dopant as small as 0.14% for NH3

and 0.34% for NO. The GRI-Mech 3.0 predictsvery well the observed maximum absolute concen-trations for all dopants except NO where the NH2

concentration is overpredicted by the factor of 2by the model. The results obtained in this workcan be used as a ‘‘target’’ for modifications ofthe GRI-Mech. These modifications will probablyinclude inserting the reaction between NH2 andHO2 into the mechanism to allow elimination ofthe non-observable peak located very close to

the burner surface in the calculation. The smallshift of the observed profiles 1–2 mm further fromthe burner in comparison with the calculationsalso demands explanation. The observed influenceof the flame confinement on NH2 concentrationalso deserves a further study, because of theimportance of this effect for understanding ofresults of different experiments both basic andapplied, where NH2 plays a significant role.

Acknowledgments

This research was supported by the Israel Sci-ence Foundation (Grant No. 574/00) and by theJames Franck, German-Israeli Binational Pro-gram in Laser Matter Interaction.

References

[1] R.M. Green, J.A. Miller, J. Quant. Spectrosc.Radiat. Transfer 26 (1981) 313–327.

[2] A.M. Dean, J.W. Bozzelli, Combustion chemistry ofnitrogen, in: W.C. Gardiner (Ed.), Gas-Phase Com-bustion Chemistry. Springer, New York, Berlin,1999, pp. 125–341.

[3] S.P. Denbaars, Proc. IEEE 85 (1997) 1740–1749.[4] S.N. Mohammad, H. Morkoc, Prog. Quantum

Electron. 20 (1996) 361–525.[5] S.C. Ross, F.W. Birss, M. Vervloet, D.A. Ramsay,

J. Mol. Spectrosc. 129 (1988) 436–470.[6] K. Dressler, D.A. Ramsay, Philos. Trans. R. Soc. A

251 (1959) 553–602.[7] M.S. Chou, A.M. Dean, D. Stern, J. Chem. Phys.

76 (1982) 5334–5340.[8] K.N. Wong, W.R. Anderson, J.A. Vanderhoff, A.J.

Kotlar, J. Chem. Phys. 86 (1987) 93–101.[9] M.C. Branch, M.E. Sadeqi, A.A. Alfarayedhi, P.J.

Van Tiggelen, Combust. Flame 83 (1991) 228–239.[10] R.A. Copeland, D.R. Crosley, G.P. Smith, Proc.

Combust. Inst. 20 (1984) 1195–1203.[11] B.A. Williams, J.W. Fleming, Combust. Flame 110

(1997) 1–13.[12] K. Kohse-Hoinghaus, D.F. Davidson, A.Y. Chang,

R.K. Hanson, J. Quant. Spectrosc. Radiat. Transfer42 (1989) 1–17.

[13] M. Votsmeier, S. Song, D.F. Davidson, R.K.Hanson, Int. J. Chem. Kinet. 31 (1999) 445–453.

[14] S.H. Song, R.K. Hanson, C.T. Bowman, D.M.Golden, J. Phys. Chem. A 106 (2002) 9233–9235.

[15] J. Deppe, G. Friedrichs, H.J. Romming, H.G.Wagner, Phys. Chem. Chem. Phys. (1999) 427–435.

[16] M. Votsmeier, S. Song, D.F. Davidson, R.K.Hanson, Int. J. Chem. Kinet. 31 (1999) 323–330.

[17] J.B. Halpern, G. Hancock, M. Lenzi, K.H. Welge,J. Chem. Phys. 63 (1975) 4808.

[18] I. Derzy, V.A. Lozovsky, N. Ditzian, I. Rahinov, S.Cheskis, Proc. Combust. Inst. 28 (2000) 1741–1748.

[19] V.A. Lozovsky, I. Rahinov, N. Ditzian, S. Cheskis,Faraday Discuss. Chem. Soc. 119 (2001).

[20] I. Rahinov, N. Ditzian, V.A. Lozovsky, S. Cheskis,Chem. Phys. Lett. 352 (2002) 169–175.

1582 I. Rahinov et al. / Proceedings of the Combustion Institute 30 (2005) 1575–1582

[21] V.M. Baev, T. Latz, P.E. Toschek, Appl. Phys. B-Lasers Opt. 69 (1999) 171–202.

[22] L.S. Rothman, C.P. Rinsland, A. Goldman, S.T.Massie, D.P. Edwards, J.M. Flaud, A. Perrin, C.Camy-Peyret, V. Dana, J.Y. Mandin, J. Schroeder,A. McCann, R.R. Gamache, R.B. Wattson, K.Yoshino, K.V. Chance, K.W. Jucks, L.R. Brown,V. Nemtchinov, P. Varanasi, J. Quant. Spectrosc.Radiat. Transfer 60 (1998) 665–710.

[23] K.J. Rensberger, J.B. Jeffries, R.A. Copeland, K.Kohse-Hoinghaus, M.L. Wise, D.R. Crosley, Appl.Opt. 28 (1989) 3556–3566.

[24] S. Cheskis, I. Derzy, V.A. Lozovsky, A. Kachanov,D. Romanini, Appl. Phys. B 66 (1998) 377–381.

[25] R.J. Kee, J.F. Grcar, M.D. Smooke, J.A. Miller, AFortran Program for Modelling Steady LaminarOne-Dimensional Premixed Flames, Rep. SAND85-8240, Sandia National Laboratories, Livermore,CA, 1991.

[26] C.T. Bowman, R.K. Hanson, D.F. Davidson, W.C.Gardiner, V. Lissianski, G.P. Smith, D.M. Golden,M. Frenklach, H. Wang, M. Goldenberg. Availablefrom: <http://www.me.berkeley.edu/gri_mech>.

[27] J. Luque, P.A. Berg, J.B. Jeffries, G.P. Smith, D.R.Crosley, J.J. Scherer, Appl. Phys. B-Lasers Opt. 78(2004) 93–102.

[28] J.A. Miller, C.T. Bowman, Prog. Energ. Comb. Sci.15 (1989) 287–338.