Infrared Chemiluminescence of the Reaction N + 02-> NO + 0

F. Hushfar, J. W. Rogers, and A. T. Stair, Jr.

The chemiluminescent nature of the reaction N + 02 - NO + 0 followed by N + NO -> N2 + 0 hasbeen investigated in the steady state through detection of characteristic infrared radiation emitted byNO molecules. Both Av = 1 radiation at 5.4 u and Av = 2 at 2.7 ,u exhibit linear dependence on 02pressure. In the case of the overtone, low-resolution spectra are obtained at several 02 pressures, andit is shown that under the experimental conditions reported here the spectra should closely reflect theinitial vibrational distribution of the NO molecules as they are formed. From the spectral results weare able to estimate an overtone quantum efficiency for the reaction N + 02 - NOT + 0. Dependingupon the experimentally reported value of the rate for this reaction, the quantum efficiency is between0.2 and 0.5.

1. Introduction

This is a report on a quantitative study of infraredchemiluminescence from the gas-phase exchange reac-tion N + 02 resulting in NO and 0. The experiment tobe described below was designed to detect infraredradiation from the fundamental and first overtone ofNOT (a vibrationally excited NO) and establish whetheror not the reaction is infrared chemiluminescent. Wehave observed infrared radiation from this reaction andhave obtained low-resolution spectra for the first over-tone of NO and are able to deduce a quantum efficiencyfor the reaction. The energy released per N 02 NO + 0 reaction (1.4 eV) is sufficient to populatevibrational levels up to v = 6. When oxygen moleculesand nitrogen atoms are mixed homogeneously in a suit-able reaction vessel, two predominant and competingreactions with characteristic rate constants k1 and k2 takeplace:

kiN(4S) + O2(X32) - NO + 0,

k2N(4S) + NO -* N2 + 0,

(1)

those of Clyne and Thrush' and Clark and Wayne5

agree closely, giving a value of -10-16 cm3 /sec. Theaccepted value of k2 is 3 X 10-11 cm 3 /sec at 300 K.6

Rate constants for reverse reactions of (1) and (2) andfor the various three-body recombinations amongspecies on both sides of (1) and (2) are of no consequencehere. The concentration of NO molecules as a functionof time can be shown to be closely approximated by

[NO] = (k1/k 2)[02] [1 - exp(-k2 [Nlt)]. (3)

In the steady state this concentration is independent ofN-atom concentration and dependent only on 02 con-centration and k and k2:

[NO18 = K[02, (4)

where brackets indicate number of molecules per cm3

and K = k/k2. If NO molecules produced in thismanner are formed in higher vibrational levels, thenpower emitted per cm3 for Av = 2 transitions in the ab-sence of other radiation loss mechanisms is given by

W = 2 [Nvl(hv)A 2(2)

where N and 02 on the left-hand sides are ground-statespecies. NO production rate constant k and depletionrate constant k2 have been studied extensively. 1-6There is some variation in experimentally determinedvalues of k. Measurement by Wilson4 indicates avalue of 4 X 10-17 cm3 /sec for k at 300 K, whereas

The authors are with the Air Force Cambridge Research Labo-ratories, Bedford, Massachusetts 01730.

Received 9 March 1971.

(5)

where [Nt] is number density in vibrational level v, h isphoton energy, and A's are respective transition prob-abilities. Approximating constant photon energy overthe band, Eq. (5) can be rewritten in terms of photondensity per second:

photons/sec-cm3 = - [NV]Av_2 . (6)

The purpose of this experiment, in addition to establish-ing chemiluminescence nature of reaction (1), is tomeasure the quantity represented in Eq. (6) and exploreits dependence on 02 density to see whether it exhibits alinearity similar to Eq. (4) in which total production ofNO molecules is predicted.

August 1971 / Vol. 10, No. 8 / APPLIED OPTICS 1843

MICROWAVECAVITY

INTEGRATING -SPHERE

PUMP

Fig. 1. Schematic of the apparatus used in investigation ofN + 02 reaction.

11. Experimental Apparatus

The schematics of the apparatus is shown in Fig. 1.The major components of this system are (a) N-atomsource, (b) reaction vessel, (c) infrared filters, and (d)infrared detection unit. These are discussed sepa-rately.

(a) N-Atom Source

Nitrogen atoms are produced in a microwave cavityoperated at 2450 \IHz and flow into the reaction vessel.Radiation from the discharge is minimized to tolerablelevels in a Wood's trap. The discharged gas travels adistance of 15 cm before entering the reaction cellthrough a 1-mm-diam opening in the cell. No ap-preciable backflow of oxygen into the discharge regionoccurs. A particle undergoes approximately 105 colli-sions as it traverses the distance between the dischargeregion and the reaction cell. This is more than ade-quate for quenching any metastable nitrogen atoms pro-duced in the discharge.7 No attempt was made to verifythis experimentally.

The discharge was modulated at several frequenciesbetween 3 Hz and S Hz, and phase-sensitive detectionwas employed for discrimination against the 300 Kbackground. Quantity 1/k2 [N ] in Eq. (3), which is thetime constant of the system of reaction (1) and (2), ismuch smaller than the period of modulation. This en-sures that steady-state condition is attained well withinthe observation time.

(b) Reaction Vessel

This is a 2-liter gold-plate Pyrex integrating sphere,where N atoms from the discharge and oxygen mole-cules flowing in from a side port are mixed. Thissphere is dimpled before being plated, to randomlyscatter the emission from a source at any positionwithin the sphere. This was independently confirmedby measuring the power output of the cell as a small lightbulb was moved into various points within the sphere.With the flow of nitrogen through the discharge tube set

for maximum signal, the nitrogen atom concentration isroutinely determined by NO titration method andfound generally to be about 1-1.57 of the total nitrogenpressure in the sphere. The pressure-measuring deviceis a 1-mm AIKS Baratron Head whose sensitive elementis directly exposed to the gas for accurate pressuremeasurements. The sphere is pumped on by a Kinneymechanical pump whose pumping speed at the throat ofthe pump remained constant over the pressure rangesused in this experiment. The pumping speed at thereaction cell is approximately 4 liters/sec. Thus themean residence time of the molecules in the cell is cal-culated to be 0.5 sec as compared to the modulationperiod of 0.16 sec. Emitted photons from excited NOmolecules eventually exit through a circular CaF2window 2.5 cm in diameter.

(c) Radiation Filters

NO fundamental (v = 1) and first overtone (v = 2)vibration/rotation bands occur at 5.4 ,u and 2.7 , re-spectively. Integrated intensities of these bands wereexamined using band filters. The fundamental filterhad maximum transmission at 5.5 M, and no measurabletransmission below 5.1 and above 5.9 A. The over-tone filter for Av = 2 sequence had maximum transmis-sion at 2.85 ju and 50% transmission points at 2.60 , and3 .10 p. For spectral investigation of Av = 2 sequence, acircular variable filter (noted as CVF in Fig. 1 andhereafter) was employed. This CVF, a disk filter 10cm in diameter, covers the wavelength range of 1.6-3.14ui over 0-180 rotation and retraces 3.14-1.6 , through180-360' rotation. The 1.6-pt position corresponding to0° rotation of CVF was masked off for reference pur-poses. The filter is driven by a synchronous motor ca-pable of running at several fixed rotational speeds(1, 5, 15, and 30 min). Wavelength vs angle for CVFwas obtained independently and coincided with thecalibration chart supplied by the manufacturer.

(d) Infrared Detection System

The NO fundamental band at 5.4 )u is obtained byusing a 3 mm X 3 mm PbSe detector cooled to liquidnitrogen temperature. The overtone band at 2.7 isinvestigated by a 20 mm X 20 mm PbS detector at dry-ice temperature. For use with the CVF to obtain low-resolution spectra, a long, narrow PbS detector (2 mmX 10 mm) oriented radially with the CVF was used.This resulted in a resolution element of 0.13 ,, too largefor resolving vibrational levels of NO molecules. Agermanium filter was placed immediately in front of thedetectors for the purpose of eliminating all extraneousradiation below 1.8 u.

The entire optical system was calibrated for radiationloss factors, transmission factors, and detector spectralresponse. This calibration against wavelength was ob-tained by admitting radiation from a blackbody ofknown temperature into the sphere and recording thedetector response as the CVF scanned the wavelengthregion from 2 u to 3.14 ,. The resulting experimentalfunction is thus used to convert detector response intopower output and will be referred to as R(X).

1844 APPLIED OPTICS / Vol. 10, No. 8 / August 1971

U)

Ž3.0 -j-j

20-jt 0 -OXYGEN(0,, A - NITROGEN

10 X - ARGON

0 100 300 500 700PRESSURE IN MILLITORR

Fig. 2. NO overtone radiation at 2.7 (v = 2) obtainedthrough the overtone filter vs mixing gas pressure.

Ill. Results and Discussion

The first set of data obtained in this experiment wasthe integrated intensity of the first overtone band ofNO at 2.7 A. Figure 2 shows the detector signal as afunction of 02 gas pressure. Nitrogen pressure in thesphere was kept constant throughout the run at 250mTorr. The results for argon and nitrogen gases react-ing with N atoms (instead of 02) are also shown. In thesecases no signal beyond the discharge noise was observed,a strong indication that signal observed with 02 was notdue to electronically excited N2 formed in three-bodyrecombination of nitrogen atoms. Another source ofradiation can be due to vibrational exchange betweenNO (formed in N + 02) and N2 , which exists abun-dantly in the sphere. To investigate this possibility,nitric oxide was introduced into the sphere to mix withdischarged nitrogen. A few millitorr of NO was neededto titrate out all the nitrogen atoms. Additionalamounts of NO produced no signal until the NO partialpressure reached well into the torr range.

Further evidence for formation of vibrationallyexcited NO is measurement of the integrated intensityof infrared radiation obtained by using the NO funda-mental filter (5.4 ). This is shown in Fig. 3. Inaddition, the results depicted in Figs. 2 and 3 exhibitlinear dependence of signal on 02 concentration, whichwas postulated in Eq. (6). Data shown in Figs. 2 and3 are corrected for loss of signal due to imperfect modula-tion of N-atom density in the sphere as follows. Atthe modulation frequency used, the residence time ofparticles in the integrating sphere is higher than theperiod of modulation and consequently [N] does notundergo perfect modulation. This condition of imper-fect modulation causes loss of signal. To correct forthis loss, integrated intensities at fixed pressures in thesphere were obtained as a function of modulation fre-quency. The resulting signal vs modulation frequencycurve is used to correct the results. The linearity ofthe signals shown in Figs. 2 and 3 can be employed todraw certain conclusions regarding the gas kinetictemperature in the reaction cell. Oxygen was always

admitted to the cell at z300 K. If the gas mixture inthe cell were greater than 300 K due either to the dis-charged gas or to relaxed chemical energy of the reac-tions occurring, the gas mixture would be at a highertemperature at low 02 pressure than at high 02 pressure.Because of the strong temperature dependence of k1,this effect should result in considerable deviation fromlinearity of signal vs pressure at high 02 pressure. Thisis clearly not occurring, therefore we conclude that thekinetic temperature of the gas mixture was essentiallyroom temperature. In the experiment the dischargetube was always kept cool by air flow from a compressor.

Figure 4 shows the actual low-resolution spectrum ofNO overtone as obtained on a Varian chart recorderwith the use of the CVF at a single 02 pressure, and istypical of spectra obtained over the 02 pressure range.The wavelength scale as obtained from the CVF cali-bration chart is also shown. The dip at 1.6 u is due tothe masked-off reference point at 0° on the CVF. Theknee at 1.8 A is the cutoff of the germanium filter. Thesecondary peak at 2.2-2.3 4u remains unknown, and itsidentification will require further investigation. Thelarge peak is due to the Av = 2 sequence of the NOmolecule. The actual signal obtained here was cor-rected for the total response function R(X) at each X, andthe resulting spectrum in terms of microwatts/micron

C,)

0

2

-j

z0co

0 200 400 600 800PRESSURE IN MILLITORR

Fig. 3. NO fundamental radiation at 5.4 u (Av = 1) obtainedthrough the fundamental filter vs 02 pressure.

N+O2-N~oO

;26 f 02.0 as 3j4 225 220 I;6

Fig. 4. Low-resolution spectrum of NO overtone radiationobtained through the circular variable filter. Wavelength

scale in microns is also shown.

August 1971 / Vol. 10, No. 8 / APPLIED OPTICS 1845

z 10.00

` 8.0

g 6.00a:

2 40z_t 2.0zCDV)

0

_

) _

I _r . ' - I _,

C X

I

/ C..I/ \j \x

2 2.5 3.0X IN MICRONS

3.5

Fig. 5. A sample of the NO overtone spectrum, corrected forlosses and system response.

was replotted after the residual noise was removed.Such a graph is shown in Fig. 5. It is seen that theAv = 2 band extends beyond the wavelength range ofthe CVF at 3.14 A. The spectrum was obtained in theimperfect modulated condition and is a valid descrip-tion due to the fact that the reference (Off) signal hasthe identical spectral distribution as the signal On. Thespectra obtained over the 02 pressure range of the ex-periment were all identical. This indicates that colli-sional deexcitation is not taking place, a fact which isalso borne out in the following discussions.

Since the fast reaction (2) always follows reaction (1),N-atom density decays at a rate twice that of (1) andtakes place with a time constant of (2k[02])-l. For50 mTorr and 750 mTorr of 02 (lowest and highest 02pressures in this experiment), this time constant is 3see and 0.3 sec, respectively. Initial N-atom density is2 X 1014 cm-3 . After one residence time of 0.5 seeN-atom density is 1.7 X 1014 cm-3 at 50 mTorr of 02and 1.7 X 10's cm-3 at 750 mTorr of 02. Using thesevalues of [N], we calculate (k2 [N])-I (time constant fordestruction of NO by N) to be 0.2 msec and 0.7 msec,respectively. This means that NO molecules existless than a millisecond before removal by N atoms, overthe 02 pressure range of the experiment. This time ismuch shorter than the radiation time for NOt. TheNO-NO collision frequency is approximately 20 sec'at 750 mTorr of 02 and is much smaller at 50 mTorr of02. This results in 2 X 10-2 collisions among NOmolecules before removal by N atoms and 2 X 10-1collisions before radiation, hardly sufficient to establishvibrational equilibrium' or vibrational deexcitationamong NO molecules.9 Collisions of vibrationallyexcited NO molecules with other gases in the reactioncell also are not important in deexcitation, since (foran over-all collision frequency of 2 X 106 sec') an NOmolecule undergoes only 2 X 103 encounters beforeremoval by N atoms or 2 X 104 encounters before radia-tion, whereas at least 106 collisions are required for de-excitation.8 On the other hand, even at lowest 02pressure of 50 mTorr the number of collisions beforedestruction by N atoms or before radiation providesadequate time for rotational relaxation. Our conclu-

sion, based upon these calculations, is that the spectralinformation shown in Fig. 5 reflects the initial vibrationaldistribution of NO molecules as they are formed inreaction (1). Unfortunately the poor spectral resolu-tion of the CVF system used here does not permit defini-tive information to be inferred about the relative inten-sities in various A = 2 transitions, although apparentlythe 4 -*- 2 vibrational transition is the dominant feature.

In order to calculate number of photons emitted fromthe curve in Fig. 5, a technique must be devised to esti-mate the behavior of the spectrum beyond the wave-length range of the CVF at 3 . In the absence of apriori knowledge of the initial vibrational distribution,we undertook to construct theoretical NO spectra withthe assumption of separate vibrational and rotationalstate Boltzmannization and have compared the experi-mental results to these calculated spectra. Figure 6shows the comparison of the results with the syntheticspectra obtained with various combinations of vibra-tional and rotational temperatures. The best matchoccurs for T = (5000 ±t 500) K and T = (300 50)K. This sort of comparison may eventually provefortuitous. We cannot overemphasize the fact thatthis kind of curve-fitting is purely empirical and is doneonly for the purpose of reducing the error in the calcula-tion of the power emitted represented by the area underthe curves in Fig. 5. Due to the low resolution, noconclusions should necessarily be drawn from this con-

Tv('K) TR('K)

>_ A) 300 300B) 5000 300

z 1.0 A C) 7500 500XXX EXPERIMENTALPOINTS

-j50X

0

2.5o3.0 3.A IN MICRONS

Fig. 6. Comparison of the experimentally obtained spectrumwith theoretically constructed NO spectra at different combina-

tions of rotational and vibrational temperatures. ;

U)

I-0

0)

0

z

ti

z

or

wr

30h

2.0

10

0100 200 300 400 500 600 70002 PRESSURE IN MILLITORRS

Fig. 7. NO overtone radiation emitted per cm3

vs 02 pressure.

1846 APPLIED OPTICS / Vol. 10, No. 8 / August 1971

I . I I* . II

cerning the initial vibrational distribution of NO mole-cules. When we extend our experimental spectra inthis manner (dotted portion of the curve in Fig. 5), thedata are further reduced as follows.

The spectra were integrated individually to obtaintotal power emitted at respective oxygen pressures.The results were then divided by the emitting volume(2 X 103 cm 3) and plotted against pressure (Fig. 7). Asseen, the NO overtone radiation exhibits linear depen-dence on 02 pressure similar to Eq. (4), in which theproduction of NO molecules is predicted. Calculationof the slope of the line in Fig. 7 after appropriate con-version shows that 1.5 X 10-6 photons (per second per02 molecule) are emitted. To calculate a quantumefficiency figure aq, it is necessary to arrive at the numberof overtone vibrational quanta produced per 02 mole-cule for comparison with K of Eq. (4). This could bedone in the case of a well-resolved spectrum by remov-ing Einstein transition probability A-2 from eachvibrational transition in the power spectrum. Tocalculate a useful number from our unresolved spectra,we define an average A as A = NAV-2/N,. As-suming harmonic oscillator relationship between suc-cessive A's and using the experimentally estimatedvibrational temperature of 5000 K and A0

2 = 0.81sec-' from the measurement of NO overtone intensityby Schurin and Ellis,' 0 we obtain a value of 2.28 sec-for A. Dividing 1.5 X 10-6 photons per second per 02

molecule by A, we obtain a value of 0.66 X 10-6 with anestimated error of ±+S % 30 for the number of NO vibra-tional quanta per 02 molecule. In terms of this numberand ki and k2 the quantum yield X is 0.66 X 10-6 k 2/k,.Using previously indicated values of k2 and kj, we cal-culate 7 0.2 if k = 10-16 cm3 /sec3 '5 and j ~ 0.5 ifWilson's value of k, is used.4

Very accurate measurement of -q must of course awaitfurther refined spectrometric or interferometric mea-surement to resolve the vibrational/rotational transi-tions and to alleviate the necessity of empirical fittingwith theoretical models of the spectra. Efforts toinvestigate further both the fundamental and overtoneregions are currently in progress.

We express our gratitude to Marshall H. Bruce andFrank P. DelGreco of this laboratory for very helpfuldiscussions and contributions. We are also indebted toThomas C. Degges of Visidyne, Inc., for suggestionsregarding the initial vibrational distribution. Thetechnical assistance of Charles P. Dolan and Edwin E.Eckberg in design and construction of the apparatus isalso gratefully acknowledged.

Referencesl. G. B. Kistiakowsky and G. G. Volpi, J. Chem. Phys. 27, 1141

(1957).2. F. Kaufman and J. R. Kelso, in Seventh Symp. Combustion.

(Univ. Press, London, 1959), p. 53.3. M. A. A. Clyne and B. A. Thrush, Proc. Roy. Soc. (London)

A261, 259 (1961).4. W. E. Wilson, J. Chem. Phys. 46, 2018 (1967).

. I. D. Clark and R. P. Wayne, Proc. Roy. Soc. (London) A316,539 (1970).

6. M. A. A. Clyne and B. A. Thrush, Proc. Roy. Soc. (London)39, 1772 (1963).

7. G. Black, T. G. Slanger, G. A. St. John, and R. A. Young, J.Chem. Phys. 51, 116 (1969).

S. N. Basco, A. B. Callear, and R. G. W. Norrish, Proc. Roy.Soc. (London) A260, 459 (1960)..

9. H. J. Bauer and K. F. Sahn, J. Chem. Phys. 42, 3400 (1965).10. B. Schurin and R. E. Ellis, J. Chem. Phys. 45, 2528 (1966).

Frederic Weigl of Collins Radio Company, a member of the Applied Optics Patents Panel.

August 1971 / Vol. 10, No. 8 / APPLIED OPTICS 1847