Spatially resolved saturated absorption measurements of OH inmethane-air flames
Giorgio Zizak, Francesco Cignoli, and Sergio Benecchi
A cross-beam saturated absorption spectroscopy technique, utilizing a single pulsed dye laser, has beendeveloped for local concentration measurements in flames. With a differential detection of the probe and thereference laser beam intensities, a significant improvement of the technique has been achieved. In this work
the basic theory of the method is discussed. Its use in combustion studies is demonstrated by presenting OH
concentration profiles in two premixed laminar methane-air flames.
1. Introduction
Minor molecular species determination in combus-tion gases is usually performed by laser-induced fluo-rescence (LIF) techniques. Due to the high sensitivityof these techniques, spatially and temporally resolvedmeasurements of temperature and species concentra-tion have been performed in a variety of combustionsystems. However, fluorescence signals originatefrom upper excited states and, therefore, are affectedby collision dynamics, such as electronic quenchingand rotational and vibrational energy transfer. Hencefor absolute total number density measurements,knowledge of the collisional energy transfer in theupper and in the ground state is needed. This hasbeen achieved only in a few cases, whereas most of theLIF concentration measurements have been carriedout introducing significant approximations.
Absorption techniques directly probe the groundstate population and are not affected by such prob-lems. But they are line-of-sight techniques and can-not be used for information on variations of propertiesalong the absorption path.
A cross-beam saturated absorption spectroscopy(XBSAS) technique was introduced by Goldsmith 1 forspatially resolved absorption measurements in flames.Saturated absorption spectroscopy (SAS), with coun-terpropagating laser beams, is a well-known techniquefor Doppler-free high-resolution spectroscopy of mole-cules in the gas phase. 2' 3
The XBSAS technique is conceptually similar toStark modulation spectroscopy developed for local ab-sorption measurements in flames.4-7
The authors are with CNPM-CNR, 69 via Baracca, 20068 Pes-chiera Borromeo, Milano, Italy.
In Goldsmith's original work, a narrowband cw dyelaser was employed to detect, with high accuracy, Naatoms seeded in a hydrogen-air flame. The use of abroadband pulsed dye laser for analytical atomic spec-troscopy in flames was also attempted. 8 9 It was evi-dent that a laser with high intensity stability was need-ed to detect weak signals (small probe volume and lowspecies concentration) on top of the large laser beamintensity.
Recently two optical arrangements, based on inter-ference and polarization effects, were proposed to re-duce the background intensity in cross-beam pulseddye laser experiments.10 11 These techniques havebeen used only for atomic species detection. More-over as pointed out in Refs. 10 and 11, quantitativemeasurements in flames are, at the moment, preventedby theoretical (polarization) and experimental (inter-ference) problems. To reach saturation in spatiallyresolved saturated absorption measurements of mo-lecular species, high power pulsed dye lasers must beemployed. The strong intensity fluctuations of com-mercially available pulsed dye lasers have precludeduse of the XBSAS technique for molecular detectionsin combustion gases. Some attempts to indirectlyevaluate the local absorption coefficient of OH havebeen performed by a simultaneous measurement oftotal absorption and 1-D imaging of the linear fluores-cence. By an iterative computer procedure the OHlocal absorption was obtained. 2 13
The only direct measurement of OH concentrationin flames performed with the XBSAS technique wascarried out by Kychakoff et al.'4 with a complex ex-perimental arrangement (an argon-ion pumped dyelaser and a Nd:YAG pumped dye laser tuned to thesame absorption transition). As mentioned by theauthors,14 the major drawbacks of the technique werethe critical optical alignment and the low sensitivity(measurement of small variations of large signals).
In this work we employed an optical arrangement topartially overcome such problems, allowing the use of asingle pulsed dye laser. The constant background issuppressed with a reference beam and a differentialdetection system. In this way OH concentration mea-surements were obtained. Reasonable system stabil-ity and measurement precision were achieved as shownby the OH concentration radial profiles presented fortwo different methane-air flames.
II. Theory
Basically in the XBSAS technique the output of adye laser, tuned to an absorption transition, is splitinto a weak probe and a strong pump beam.1 The twobeams are crossed at the probe volume so that thepump beam saturates the absorption transition andthe intensity of the probe beam is increased. Thisvariation in the probe signal is associated with thechange in absorption and, hence, with the ground statenumber density of the absorbing species within theprobe volume.
With reference to Fig. 1 we call 1(X) the monochro-matic spectral irradiance of the probe beam incidenton the flame, I'(X) is the transmitted irradiancethrough the flame without the saturating beam, andI"(X) is the transmitted irradiance with the saturatingbeam. We divide the absorption path length intothree parts: the length before the interaction region,L1, the interaction length, Al, and the length after theinteraction region, L 2. We designate
modified by the presence of the pump beam. Thetransmitted spectral irradiance without and with thepump beam can be written as
I'(X) = 0 (X) exp(-K 1 - AK'- K2),
I"(A) = 10 (X) exp(-K 1 - AK" - K2 );
(2a)
(2b)
the contribution of AK" for a saturating pump beam isnegligible. The measurable transmitted laser intensi-ties are
4L = J I(X)dX,
rL = | IP(X)dX.
(3a)
(3b)
For narrow line excitation (Aabsorpt >> AXaser), Eqs.(2) and (3) can be combined to give
ln(IL/'L) = (4a)
where k is the average absorption coefficient in theinteraction length over the source linewidth AXiaser.Alternatively, by series expansion of the logarithmicfunction, we use the expression
= r_____ = kAXAl.L L
(4b)
In general the absorption coefficient kx is a complexfunction of the wavelength and is described by thewell-known relationships. 1 51 6
K1 = lLl = J. k(x)dl,
_ r~+Al+L,
K2 = k2L2 = I k(x)dl,fL1 +A1
AK' = kAAl = fL+ k\(x)dl,
(la)f re2X,2fN
mc2
k = kOb(a,v),
k =4r ln2e2 fNmc
2AXI
(lb)
(lc)
I(av) = a exp(-y 2)dy7r f- a2+ (v-y) 2
AK" = kiAl= J, k(x)dl, (ld)
where kx(x) is the monochromatic absorption coeffi-cient function of the position in the flame, and k'(x) isthe absorption coefficient within the probe length,
1 L
p ~ (A r' o b ' I (A)
Fig. 1. Scheme of the probe and saturating beam intersectiongeometry.
In Eqs. (5), c is the speed of light, e and m are thecharge and mass of the electron, Xo and AXD are, respec-tively, the center-line wavelength and the Dopplerhalfwidth of the absorption line, N is the concentrationof the absorbing species, f is the oscillator strength, and3(a,v) is the Voigt function with the dumping parame-ter a. The function (a,v) describes the variation of kxover the entire absorption line. However kiA must beevaluated over the laser linewidth, AXiaser. This re-stricts the range of v to an upper limit of AXiaser/AXabsorpt,'6 so that kh, = k(a,v), where is an averageover the range from zero to AXIlaser/AXabsorpt. Difficul-ties arise in evaluating b(a,i), especially when AXiaser iscomparable with AXabs. With the conversion factorsfor the intensities' parameters,17 the number densityNnb [#/cm 3 ], of the absorbing species, for narrowbandexcitation, can be evaluated by
AlL /r c AXDnb = 4 1n2 h A1X0,(a,o)Bl2
where h is the Planck constant, B12 is the Einstein
transition probability for absorption, and all the othersymbols have been previously defined.
For broadband laser excitation (AWabsorpt << AXIaser)
similar expressions can be derived. Integrating Eqs.(2a) and (2b) and by using series expansions for theexponentials, we obtain for the difference of the laserintensities transmitted through the flame, IL - 1L, theexpression
- = I: I(X)AK'[1 - (K, + K2 + AK)]dX, (7)
where the terms containing AK" have been neglectedbecause of saturation. Neglecting also AK'/2 in com-parison with (K, + K2), the laser intensities' differencecan be approximated by
4L - 1t-L I-(X)AK'dX. (8)
If the spectral profile of I"(X) is a continuum over theabsorption bandwidth, we have
L - L i"(x 0 ) AK'dX, (9)
and with Eq. (1) the total absorption factor (IL- fL)ILcan be related to the total absorption coefficient by
rL - fL Al f kdX. (10)
Remembering relation (5a), the number density of theabsorbing species for broadband excitation is givenby16
Nb=AIL c AXIaserNbb = y4 h A1X0B1 ' (11)
which is similar to relationship (6).Comparison of Eqs. (6) and (11) shows that, for low
concentration measurements, the decisive factor con-cerning the sensitivity of a line or broadband source isthe ratio' 6
4 ln2 AXIasert(ax)(R= AX (12)
7r AXDJ
For OH detection in a flame, AXD , 2.5 X 10-2 A and ais within the 0.5 range.'8 The Voigt function evaluat-ed for a very narrow bandwidth laser is 5(0.5,0) =
Fig. 2. Experimental apparatus.
0.61.16 Therefore the use of a broadband laser, withtypical bandwidth of the order of 0.1 A, gives a sensitiv-ity reduction of -50% with respect to a narrowbandexcitation source.
Ill. Experimental Apparatus
As previously stated, the major difficulty in per-forming spatially resolved absorption measurements isto detect small intensity variations, AIL, on top of thelarge background of the laser intensity. This difficul-ty is particularly severe in using pulsed dye lasers, oflow repetition rates (10-20 Hz), which present pulse-to-pulse variations of the intensity up to 10%. Toovercome this difficulty we have used a reference beamtechnique. As shown in Fig. 2, the UV output of a dyelaser (Lambda Physik FL 2002), pumped by an ex-cimer laser (Lambda Physik EMG 201), is mildly fo-cused by a quartz lens with long focal length. It is thendivided by a quartz plate into three beams: pumpbeam (transmitted), probe and reference beams (frontand back reflections). The two weak beams, separat-ed by -2 mm, are rotated by a Dove prism and sent tothe flame one over the other. The pump beam passesthrough an aperture (-0 = 1 mm, not shown for clar-ity) and crosses the lower weak beam (probe), slightlyout of focus, with an angle of 8°. This configurationwas utilized to reduce edge effects of the saturatingbeam. The probe and reference beams are then for-warded by quartz prisms to two photodiodes (EG&GFND100) connected in such a way that the outputcurrent depends on Iref - Iprobe. When the pump isblocked, the probe and the reference beams experienceapproximately the same absorption through the flameand, with the use of appropriate neutral density filters,the output signal from the photodiodes is nearly zero.In the presence of the pump beam the absorption of theprobe is slightly reduced giving a variation in the pho-todiode's current. This signal is processed by an opti-cally triggered boxcar integrator (PAR 162 + 165) anda strip chart recorder (Yokogawa F-3052). Atmo-spheric pressure methane-air premixed flames wereproduced by a Mecker type circular burner of 30-mmdiameter (Alkemade burner). The burner was mount-ed on a three-axis translation stage for radial investiga-tion of the flames.
IV. Experimental Procedure
The UV output of the dye laser, running at 10 Hz,was tuned to the Q,6 line of the A22+ - X2 7r (0,0) bandof OH. This was verified by fluorescence excitationscans, measured with a small monochromator (JYH200) and a photomultiplier (RCA 1P28), not shownin Fig. 2. From a fluorescence scan the UV spectralbandwidth of the dye laser was estimated to be 0.16 A.This result was also confirmed by scanning a high-resolution monochromator (Jarrell Ash 25-100 withO.Q3-A spectral resolution in the second order) acrossthe laser line. This laser spectral bandwidth is consid-erably larger than the absorption line 25 X 10-2 A).18,19
From an estimate of the longitudinal mode spacing inour dye laser, more than ten modes are present within
Fig. 3. (a) Laser intensity incident, I, and transmitted through theflame, I'. (b) XBSAS signal due to the saturation of the probe
volume.
the absorption line. Crossing of the probe and thesaturating beams was set at 15 mm from the burnertop. To obtain the probe volume length, an intensityprofile was measured by scanning a 50-,gm slit acrossthe saturating beam. Assuming a top-hat profile witha diameter equal to the full width at half-intensity, theprobe volume length was approximated to 4.3 mm. Acrude verification of the saturation of the OH transi-tion was obtained by reducing the pump intensity (-1mJ) with neutral density filters (Melles-Griot) andmeasuring the total band fluorescence signal. No sig-nificant fluorescence variation was observed for a 50%reduction in the laser intensity. After a careful align-ment of all the laser beams, the pump and the refer-ence beams were cut by black paper and the probebeam intensity, IL, incident on the flame was mea-sured, as shown in Fig.3(a). The flame was lit and theprobe intensity decreased to the L value, due to theabsorption along the flame. Then the reference beamwas allowed to pass through the flame and the intensi-ty was adjusted, with neutral density filters, in such away that the output current from the photodiodes wasnearly zero. The sensitivity scale of the boxcar wasincreased and some noise became evident at the zerolevel, as shown in Fig. 3(b). The pump was thenallowed to saturate the transition and a signal AILcould be easily observed. The settings of the boxcaraverager were 15-ns gate and 10-As time constant, giv-ing an "apparent time constant" of twenty triggers.20
V. Results and Discussion
Signals, AL, of the order of a few percent of theincident laser intensity, IL, could be measured. Thisapproximately corresponds to the saturation of one-tenth of the total absorption path, in agreement withgeometrical estimates. To test the ability of the tech-nique in performing spatially resolved OH concentra-tion measurements, two premixed methane-air flameswere investigated by translating the burner. For eachlateral position the transmitted probe intensity wasmeasured and the zeroing between the probe and refer-ence beams was restored. Figures 4 and 5 show theradial profiles of a near-stoichiometric flame and a richflame. The two vertical scales, drawn in the figures,correspond to the measured signals and to the absolutetotal number density, as calculated from expression
10 -
5.
N
H
H
5-
II
CH - AIR
STOICHIOMETRIC FI.AME
- 15
.* 00 0 0
0
0
0
£ 5 10
Fig. 4. Radial profile of OH in the stoichiometric CH 4 -air flame.
A,-
~010-
-
N I
0
'H
CH - AIR
RICH FLAME
h1 - 15
0 0
S.
I
0 f
0 0
£ ~ 5 10 mn
Fig. 5. Radial profile of OH in the rich CH4-air flame.
(11) and by applying a partition function calculated forthe estimated flame temperature of 2000 K. As is wellknown the population of the N = 6 rotational level isinsensitive to temperature variations, and no calcula-tion of the exact partition function for the actual flametemperature has been performed. From Fig. 4 weobserve that the peak value of the stoichiometric flameoccurs at the center line and the OH concentrationslowly decreases toward the borders of the flame, asexpected. In the rich flame, Fig. 5, the peak concen-tration is at the border of the flame where secondarycombustion of the methane with the surrounding airoccurs. The maxima of the two flames match well andcorrespond to the stoichiometric combustion of amethane-air mixture.
In Figs. 4 and 5 a few typical error bars have beenreported showing that the technique can give reliableand precise measurements. The absolute values of the
OH concentration, found in this work, agree well withprevious measurements in similar flames.' 4"19 Howev-er, for more accurate measurements an effective probevolume length must be carefully evaluated.14 Theprobe beam crosses the saturating beam in regions ofdifferent intensity. At the edges of the focused pumpbeam, saturation cannot occur and AK" in Eq. (2b)cannot be neglected. Beam-edge effects, however, arecommon to other laser-based diagnostic techniques,for example laser-induced saturated fluorescence.2 1-23
Several papers have recently been published dealingwith probe volume determination of the saturated flu-orescence signals.24-27 These results can provide indi-cations in evaluating the interaction region of crossedbeam measurements.
VI. Conclusions
We have proved that spatially resolved saturatedabsorption measurements of OH can be performed inatmospheric pressure flames with a single pulsed laserand a relatively simple apparatus. Although tested onpremixed laminar flames, the XBSAS technique witha reference beam is expected to give results even inturbulent stationary flames. The precision of thetechnique can be improved by ratioing, pulse by pulsein a dual channel boxcar averager, the XBSAS signal tothe intensity of the transmitted probe pulses. In thelatter arrangement the XBSAS technique is potential-ly a single pulse method. Modified optical setups areunder investigation to extend the applicability of theXBSAS technique to nonstationary turbulent flames.Furthermore, temperature measurement can be ob-tained by scanning the dye laser through the absorp-tion band. Combination of the fluorescence and theXBSAS measurements will allow direct evaluation ofthe quantum efficiency of the molecular transitionsand/or the absolute calibration of the fluorescence sig-nals.
The authors wish to thank G. B. Daminelli for mak-ing the electronic circuit. This work was financed inpart by the Commission of the European Communitiesunder contract EN3E-0087-I.
References1. J. E. M. Goldsmith, "Spatially Resolved Saturated Absorption
Spectroscopy in Flames," Opt. Lett. 6, 525 (1981).2. V. S. Letokhov, "Saturation Spectroscopy" in High-Resolution
Laser Spectroscopy, K. Shimoda, Ed. (Springer-Verlag, Berlin,1976).
3. W. Demtroder, Laser Spectroscopy (Springer-Verlag, Berlin,1982).
4. K. Knapp and R. K. Hanson, "Spatially Resolved Tunable Di-ode-Laser Absorption Measurements of CO Using Optical StarkShifting," Appl. Opt. 22,1980 (1983).
5. R. L. Farrow and L. A. Rahn, "Spatially Resolved InfraredAbsorption Measurements: Application of an Optical StarkEffect," Opt. Lett. 6, 108 (1981).
6. J. E. M. Goldsmith and R. L. Farrow, "Spatially Resolved Opti-cal Stark-Modulation Spectroscopy in Flames," Opt. Lett. 7,215(1982).
7. R. L. Farrow, "Spatially Resolved IR Absorption Spectroscopyby Optical Stark Modulation," Appl. Opt. 21, 4183 (1982).
8. P. E. Walters, G. L. Long, and J. D. Winefordner, "SpatiallyResolved Concentration Studies of Ground State Atoms andIons in an ICP: Saturated Absorption Spectroscopic Method,"Spectrochim. Acta Part B 39, 69 (1984).
9. P. E. Walters, J. Lanauze, and J. D. Winefordner, "SpatiallyResolved Concentration Studies of Ground State Atoms in aFlame: Saturated Absorption Spectroscopic Method," Spec-trochim. Acta Part B 39, 125 (1984).
10. G. Zizak, J. Lanauze, and J. D. Winefordner, "Cross-Beam Satu-rated Interference Spectroscopy in Flames," Appl. Opt. 24,3319(1985).
11. G. Zizak, J. Lanauze, and J. D. Winefordner, "Cross-Beam Po-larization in Flames with a Pulsed Dye Laser," Appl. Opt. 25,3242 (1986).
12. D. Stepowski and A. Garo, "Local Absolute OH ConcentrationMeasurement in a Diffusion Flame by Laser Induced Fluores-cence," Appl. Opt. 24, 2478 (1985).
13. H. M. Hertz and M. Alden, "Calibration of Imaging Laser-Induced Fluorescence Measurements in Highly Absorbing Fla-mes," Appl. Phys. B 42, 97 (1987).
14. G. Kychakoff, R. D. Howe, and R. K. Hanson, "Spatially Re-solved Combustion Measurements Using Cross-Beam Saturat-ed Absorption Spectroscopy," Appl. Opt. 23, 1303 (1984).
15. E. H. Piepmeier, "Atomic Absorption Spectroscopy with LaserPrimary Sources," in Analytical Laser Spectroscopy, N. Omen-etto, Ed. (Wiley, New York, 1979), Chap. 3.
16. L. De Galan, W. W. McGee, and J. D. Winefordner, "Compari-son of Line and Continuous Sources in Atomic Absorption Spec-trophotometry," Anal. Chim. Acta 37, 436 (1967).
17. A. P. Thorne, Spectrophysics (Chapman & Hall, London, 1974).18. K. C. Luck and F. J. Muller, "Simultaneous Determination of
Temperature and OH-Concentration in Flames Using High-Resolution Laser-Absorption Spectroscopy," J. Quant. Spec-trosc. Radiat. Transfer 17, 403 (1977).
19. R. J. Cattolica, "OH Radical Nonequilibrium in Methane-AirFlat Flame," Combust. Flame 44, 43 (1982).
20. EG&G Princeton Applied Research model 165 gated integrator,Operating and Service Manual (1980).
21. J. W. Daily, "Saturation of Fluorescence in Flames with a Gauss-ian Laser Beam," Appl. Opt. 17, 225 (1978).
22. J. L. Bowen and A. P. Thorne, "Time-Resolved Fluorescenceand Population Measurements in Laser-Pumped Barium Va-pour," J. Phys. B 18, 35 (1985).
23. C. Th. J. Alkemade, "Anomalous Saturation Curves in Laser-Induced Fluorescence," Spectrochim. Acta Part B 40, 1331(1985).
24. K. Kohse-Hoinghaus, W. Perc, and T. Just, "Laser InducedSaturated Fluorescence as a Method for Determination of Radi-cal Concentrations in Flames," Ber. Bunsenges. Phys. Chem. 87,1052 (1983).
25. J. T. Salmon and N. M. Laurendeau, "Calibration of Laser-Saturated Fluorescence Measurements Using Rayleigh Scatter-ing," Appl. Opt. 24, 65 (1985).
26. J. T. Salmon and N. M. Laurendeau, "Analysis of Probe VolumeEffects Associated with Laser-Saturated Fluorescence Measur-ements," Appl. Opt. 24, 1313 (1985).
27. M. J. Cottereau, "Single-Shot Laser-Saturated FluorescenceMeasurements: a New Method," Appl. Opt. 25, 744 (1986).
