Combined Vibrational and Rotational CARS for Simultaneous Measurements of Temperature and Concentrations of Fuel, Oxygen, and Nitrogen
PER-ERIK BENGTSSON,* LARS MARTINSSON, and MARCUS ALDEN Department of Combustion Physics, Lund Institute of Technology, P.O. Box 118, S-221 O0 Lund, Sweden
Simultaneous measurements of temperature and relative concentrations of fuel, oxygen, and nitrogen using combined vibrational coherent anti-Stokes Raman Spectroscopy (CARS) and dual-broad-band rotational CARS have been demonstrated with the use of a Nd:YAG laser and a single dye laser. With the use of a double-folded BOXCARS phase-matching scheme, both the vibrational and the rotational CARS signals were generated in such a way that the signals were superimposed at the spectrograph. With an additional mirror arrangement inside the spectrograph, both signals were recorded simultaneously on a single diode-array detector. The accuracy of single-shot fuel concentration measurements has been investigated, and measurements in a methane/air diffusion flame have been demonstrated. The influence of systematic errors on measured concentrations is discussed.
Index Headings: Rotational CARS; Vibrational CARS; Concentration measurements; Temperature measurements.
INTRODUCTION
Coherent anti-Stokes Raman spectroscopy (CARS) is an established technique for nonintrusive, temporally and spatially resolved measurements of temperature and ma- jor species concentrations in combustion processes? -3 In concentration measurements using vibrational CARS, a frequent restriction is the problem that only a single con- stituent can be monitored with a pump laser and a single dye laser, and simultaneous probing of several species increases the complexity of the experimental setup. For example, adding a second dye laser [running either in a narrow-band mode (dual pump) 4 or in a broad-band mode (dual Stokes)] to the setup allows simultaneous probing of two species. 5 By adjustment of the frequency difference between the two broad-band spectral dye profiles in the dual Stokes approach to match the vibrational Raman frequency of a species (dual broad-band), yet another spe- cies can be probed. 6 In some cases several species can be monitored with the use of the same dye if the dye spectral profile is broad enough to cover several resonances, for example, CO at 2145 cm -l and N2 at 2330 cm-t. 7
For pure rotational CARS, most molecules of interest in combustion have Raman resonances in the region up to several hundred wavenumbers, and they can be si- multaneously excited with the use of the frequencies of a single dye. An exception is H2, which has its first tran- sition at 587 cm -1 (J = 1 ----~ J = 3). The peak intensity of the rotational CARS signal from a single species de- pends, to a first approximation, on the square of its ro- tational Raman cross section and the square of its mole fraction. In air-fed combustion, when measurements are
Received 1 March 1994; accepted 29 August 1994. *Author to whom correspondence should be sent.
performed in the unburned fuel/air mixture, it is found that nitrogen with a high mole fraction, and oxygen with a high rotational Raman cross section (~2.2 times the cross section of nitrogen), 8 will almost always dominate the rotational CARS spectrum. In measurements on the post-combustion gases, the nitrogen spectrum will be the dominating signature of the total rotational CARS spec- trum.
The most advantageous approach of pure rotational CARS is dual-broad-band rotational CARS (DB- REARS). 9'1° In this approach an arbitrary narrow-band laser is used together with an arbitrary broad-band dye laser. The dye laser beam is separated into two equally strong beams, and the three beams are arranged either in a planar I1 or a folded 12,13 BOXCARS phase-matching scheme. In dual-broad-band rotational CARS, each ro- tational Raman resonance is driven by multiple pairs of photons, each pair consisting of one photon from each of the dye laser beams with a frequency difference equal to the Raman frequency. These frequencies are then coupled to the frequencies of the Nd:YAG laser to give rise to the rotational CARS frequencies. To cover a large spectral range, one selects a spectrally broad dye such as DCM so that a large part of the spectrum can be detected even at flame temperatures. The dual-broad-band rotational CARS approach using a narrow-band Nd:YAG laser and a DCM dye laser has been used at our laboratory in sev- eral previous investigations to study the potential of this approach for measurements of, for example, temperature 14,L5 and oxygen concentration. 16 Also, DB- RCARS has been applied for temperature measurements in sooting flames 17 and temperature and oxygen concen- tration measurements in an internal combustion (IC) en- gine. 1 s
Rotational CARS has the potential of measuring oxy- gen concentrations in situ relative to the nitrogen con- centration. 16 There is no general applicability of rotational CARS to hydrocarbon fuel measurements, although the technique can be useful in certain situations, as discussed below. Vibrational CARS can be used to probe the C-H stretch resonances in hydrocarbons, which can generally be found in the region 2850 to 3100 cm -1. A suitable combination of rotational and vibrational CARS mea- surements should therefore allow simultaneous detection of fuel, oxygen, and nitrogen. Lucht et al. demonstrated simultaneous rotational and vibrational CARS on nitro- gen in experiments using a Nd:YAG laser, a narrow-band dye laser, and a broad-band dye laser (three-laser CARS).19 In the present study, another technique for simultaneous rotational and vibrational CARS, first demonstrated in Ref. 10, is investigated in more detail. Vibrational CARS
FIo. 1. (A) Double-folded BOXCARS phase-matching scheme for the generation of simultaneous vibrational CARS and dual-broad-band ro- tational CARS signals in the same direction. R 1 and R2 originate from a broad-band dye laser, and G 1 and G2 from a narrow-band Nd:YAG laser. GI, G2, and R2 give the vibrational CARS signal, and RI, R2, and G 1 give the rotational CARS signal. (B) View of the spectrograph depicting the extra mirror arrangement used to obtain the vibrational CARS signal simultaneously with the rotational CARS signal on the diode-array detector.
is combined with dual-broad-band rotational CARS em- ploying a single dye laser. With the use of a double-folded BOXCARS phase-matching scheme/° both signals can be arranged so that they propagate in the same direction. The diagnostic potential of the technique is discussed, and simultaneous detection of methane and nitrogen is demonstrated in a methane/air diffusion flame.
EXPERIMENTAL
The experimental equipment is described elsewhere ~5'17 and is only briefly presented here. A Nd:YAG laser pro- ducing 550 mJ/pulse and a single dye laser with DCM as the dye were used. Approximately 90% of the Nd:YAG radiation at 532 nm [At g 10 ns, full width at half-max- imum (FWHM) = 0.7 cm -~] was used to pump the dye laser, while the remaining 10% was used directly in the CARS process. With the use of DCM, the dye laser pro- duced ~80 mJ in each pulse with a FWHM of ~280 cm -l around a center wavelength of ~630 nm. Both the Nd:YAG laser beam and the DCM dye laser beam were split into two beams, and the four beams were arranged in a double-folded BOXCARS phase-matching scheme, l° as shown in Fig. 1A. The focusing lens had a focal length of 300 mm, the distance between G2 and the superim- posed beams G 1 and R 1 was 14 mm, and the beam R2 was propagating through the lens at a distance of 10 mm from the center. These values gave an intersection region with an approximate diameter of 40 tim and approximate length of 1 mm. The rotational CARS signal (produced by R1, R2, and G2), and the vibrational CARS signal
1
:~0.8
4 0.6
• 0.4
.~. 0.2
0
Rotat ional CARS Vibrat ional
. . . . . . Nitrogen CARS , , Oxygen
50 100 150
Raman shift / cm q
I 2914
Fio. 2. Accumulated 100-shot spectrum of a methane/air mixture at ambient temperature and pressure. Methane is probed with the use of vibrational CARS, and nitrogen and oxygen with dual-broad-band ro- tational CARS, No compensation has been made for the finite spectral width of the broad-band dye laser.
(produced by G 1, R2, and G2) propagated in such a way that they were superimposed on each other from the probe volume, and they were focused onto the slit of a 1-m spectrograph. In Fig. 1B the arrangement inside the spec- trograph is shown. The fourth order of the rotational CARS signal was directed to the output of the spectro- graph with the use of the ordinary mirror arrangement, and the fifth order of the vibrational CARS signal was directed to the spectrograph output with an additional mirror arrangement. The detector was a diode-array de- tector, consisting of 1024 pixels, and the data were pro- cessed in an optical multichannel analyzer, OMA III (EG&G PARC).
The phase matching, the detection path, and the extra mirror arrangement inside the spectrograph were opti- mized on the vibrational CARS signal from pure meth- ane, which was generated with the dye laser in the narrow- band mode. This signal was clearly visible with the naked eye and easily traced through the setup. The fourth beam R 1 was then introduced into the setup, and the rotational CARS signal was optimized with the use of the signal from nitrogen.
M E A S U R E M E N T S AND RESULTS
To test the technique for the generation of simultaneous rotational CARS spectra of nitrogen (and oxygen), and vibrational CARS spectra of the hydrocarbon C-H stretch vibration, we selected the strong methane resonance at 2914 cm -~ . In Fig. 2, a spectrum recorded in a methane/ air mixture at ambient temperature and pressure is shown, demonstrating simultaneous detection of methane, oxy- gen, and nitrogen. The methane signal and the nitrogen and oxygen signals were directed to different parts of the diode-array detector. In Fig. 3, four spectra, each con- sisting of 100 accumulated single shots, are shown from measurements at ambient pressure and temperature with methane concentrations of 2, 4, 6, and 8% in nitrogen. Normalization has then been made on the highest peak in the nitrogen spectrum. The ratio between the integrated methane and nitrogen signals was found to be propor- tional to the square of the concentration ratio between methane and nitrogen in the mixture. The spectra in Fig. 3 were recorded with the following average pulse energies in the different beams: G1, 20 mJ; G2, 29 mJ; R1, 30
APPLIED SPECTROSCOPY 189
1.4
1.2
"~ 0.8
:~0.6
,~0.4 c
0.2
o 20 60
Rotat ional CARS
Nitrogen
100 140 Rarnan shift / cm -1
Vibrational CARS 8 % CH4 ~-->
6 % CH4---~
4 % C H 4 ~ ^2 °/° CH4~ L
2914
Fie. 3. Four accumulated 100-shot spectra with different concentra- tions of methane in nitrogen at ambient temperature and pressure. The spectra are normalized at the highest nitrogen peak, and they are not compensated for the finite spectral width of the broad-band dye laser.
m J; R2, 34 mJ. With the use of these laser pulse energies, the peak signal strength of a single shot was about l0 s counts. The beams R2 and G2 are involved in the gen- eration of both vibrational and rotational CARS signals; however, since the third beam in the CARS process for vibrational CARS is G 1 and for rotational CARS is R 1, it is possible to influence the relationship between the vibrational and rotational CARS signals by varying the incident laser pulse energies of R1 and G1.
Single-shot measurements were performed in a cell un- der constant conditions to establish the uncertainty in the evaluated concentrations attributable to the CARS in- strument. For 50 single-shot spectra from a mixture of 12% methane in nitrogen, the relative standard deviation of the integrated methane signal was 15.9%, and that of the integrated nitrogen signal was 13.4%. The calculated value of o- for the ratio between the signals was 19.0%, and the correlation coefficient between the signals was as low as ~ 0.1. Since the relative concentration of a species depends on the square root of the ratio between the CARS signals, a mole fraction of methane of 0.12 was found to be measured with a standard deviation of ~0.011. The low correlation between the vibrational and rotational CARS signal is mainly due to the difference in the noise in the two signals. While the noise in the vibrational CARS signal depends on amplitude-mode fluctuations in the dye laser, the noise in the rotational CARS signal is mainly due to phase-mode fluctuations. 2°
The technique was demonstrated in a methane/air dif- fusion flame on a Wolfhard Parker burner. In Fig. 4, two accumulated 100-shot spectra are shown from the same measurement position in the flame. Spectrum A shows a combined vibrational CARS spectrum from methane and rotational CARS spectrum from nitrogen. From the ro- tational CARS spectrum of nitrogen (obtained by block- ing the beam G1 in Fig. 1A), the temperature was eval- uated to be 1365 K. Spectrum B was obtained by blocking R 1. If a CCD detector had been used instead of a diode- array detector, the vibrational and rotational CARS sig- nals could have been recorded at different heights on the CCD chip, and the overlap of the vibrational and rota- tional CARS spectra, as seen in spectrum A, would have been avoided. This technique offers a simple way of ex- perimentally generating the vibrational CARS spectra of
1 0.9
,~ 0.8 "~ 0.7
0.6 0.5
.~ 0.4
• 0.3 0.2 0.1
0 80
A---->
B---~
• , . , . , . ,
120 160 200 I 2914 Raman shift / cm -1
FiG. 4. Spectra A and B are accumulated 100-shot spectra generated at the same point in a methane/air diffusion flame. Spectrum A shows simultaneous detection of nitrogen with rotational CARS and methane with vibrational CARS. Spectrum B (reduced in height in comparison with spectrum A) shows the vibrational CARS spectrum of methane, obtained by blocking beam R1, at a temperature of 1365 K. The tem- perature was evaluated from the rotational CARS spectrum of nitrogen obtained by blocking beam G 1. The spectra have not been compensated for the finite spectral width of the broad-band dye laser.
hydrocarbon fuels over a large range of temperatures from ambient to flame temperatures, and of simultaneously probing the temperature using dual-broad-band rotation- al CARS.
Although the C-H stretch vibration in methane was chosen in these experiments, the C-H stretch vibrations in other hydrocarbon fuels are easily studied with only minor changes in the additional mirror arrangement in- side the spectrograph. In Fig. 5, vibrational CARS spectra are shown for five hydrocarbon fuel molecules-propane, methane, ethane, ethene, and benzene--which all have vibrational Raman shifts between 2850 and 3100 cm -1. All spectra are normalized to the same height; however, the ethene signal was a factor of ~ 5, and the ethane and propane signals a factor of ~ 30, weaker than the methane signal.
Hydrocarbon fuels have a higher nonresonant suscep- tibility than oxygen and nitrogen7 ] However, in premixed combustion situations with stoichiometfic mixtures of a hydrocarbon fuel and air, the fuel concentration is low,
Propane Methane Ethane 1 i
0.8 !
• 0.6 ii
~'o.4 i:, .~
¢, i': o 2850 2900
Ethene
i
i :
ii ii !i
. . . . . . . . ~." .\.~
2950 3000
Raman shift / cm 1
Benzene
J • , . - - . . ,
3050 3100
FIG. 5. Vibrational CARS spectra for five hydrocarbon fuels at 293 K: propane, methane, ethane, ethene, and benzene. The spectra are normalized to the same height, and they have been compensated for the finite spectral width of the broad-band dye laser.
190 Volume 49, Number 2, 1995
i
1
0.8
o.6
0 ,4
0.2
0
i i i i i
i i i
i i i
C2H2/Air' ~ =2 "-->'i~ C2H2/Air • = l ~ i i i
. ~', .~.'ji i !: ~ . . l 20 30 40 50 60 70 80
Raman shift / cm "1
I I
I
Acetylene Oxygen Nitrogen
• , , , , , ,
90 100 110 120
Fro. 6. Dual-broad-band rotational CARS spectra recorded at 293 K for three conditions: air; stoichiometric mixture (4, = 1) of C2H~/air; and 4, = 2 C2HJair. All spectra are normalized to the same height and compensated for the finite spectral width of the broad-band dye laser.
and the increase in nonresonant CARS susceptibility of the mixture is small in comparison with air. The impact on the resulting spectrum is small, and there is a negligible effect on the evaluation of oxygen concentrations and nitrogen temperature since the nonresonant susceptibility of the gas is also a parameter that is fitted in the rotational CARS code. Methane has no rotational Raman reso- nances due to the symmetry of the molecule, and gives no resonant contribution to a rotational CARS spectrum of a methane/air mixture. Although ethane, propane, and ethene have resonances in their rotational CARS spectra, the only hydrocarbon fuel that gives a large spectral con- tribution to the total spectrum for a stoichiometric mix- ture of fuel and air is acetylene. In Fig. 6, rotational CARS spectra are shown for air; a stoichiometric mixture of acetylene and air with 4, = 1; and a mixture of acetylene and air with 4, = 2. The addition of acetylene to our rotational CARS code, which at present contains oxygen and nitrogen, will afford us the possibility of obtaining accurate, in situ calibrated acetylene concentrations to- gether with relative oxygen/nitrogen concentrations.
DISCUSSION
Several potential systematic errors of the described technique can be identified which emphasize the need for calibration measurements in connection with the mea- surement itself. The pulse energy from the Nd:YAG laser is used both in the CARS process and to pump the dye laser, and it was found that a 10% decrease in the pulse energy from the Nd:YAG laser (at 532 nm after the fre- quency-doubling crystal) resulted in a decrease in the ratio between the integrated CARS signals of methane and nitrogen of ~ 12%. Also, changes in the detection ar- rangement may lead to a change in the signal ratio. In an effort to prevent stray light (at the laser frequency in the spectral vicinity of the signal) from reaching the detector, several apertures are often used in the detection path. If, for example, the diameter of an aperture is adjusted, this step can affect the ratio between the methane and nitrogen signal if the diameters of the vibrational and rotational
CARS signal beams are different, thus making it necessary to recalibrate the setup.
Another effect that may limit the potential of accurate concentration measurements is the dependence of each of the CARS signals on temperature, which must be known when studies are made at different temperatures. 22 The limited dynamic range of the detector must also be taken into consideration when one is using CARS for concen- tration measurements. Since CARS signals have a qua- dratic dependence on the mole fraction and a strong de- pendence on temperature, the ability to evaluate concentrations from two CARS signals will be highly de- pendent on the dynamic range of the detector. Other ef- fects which must be taken into account are interfering nonlinear processes leading to population perturbations, 23 which may occur at laser intensities that are too high. In this study, no signs of population perturbations were ob- served, since no spectral perturbations were seen when the laser intensities were varied, and a quadratic signal dependence on methane concentration was established.
For dual-broad-band CARS approaches, four-wave mixing processes can occur which may interfere with the desired signals?.1° In the present phase-matching scheme, the C-H stretch vibration can be excited with the use of G2 and R2, and the coherence can be scattered off the broad-band beam R1 to give a broad-band "smeared" vibrational CARS spectrum from methane at the fre- quencies where the rotational CARS spectrum appears. The influence of this interference was negligible in our measurements but could have been worse if the vibra- tional CARS signal from methane had been higher rela- tive to the rotational CARS signal.
To summarize, combined vibrational CARS and dual- broad-band rotational CARS were performed with a Nd: YAG laser and a single dye laser, a double-folded BOX- CARS phase-matching arrangement, and a single spec- trograph and detector. The technique offers simultaneous measurements of temperature and concentrations of hy- drocarbon fuel, oxygen, and nitrogen, and it can thus be a valuable diagnostic tool in different combustion appli- cations especially at the lower temperatures in the mixing and precombustion regions. The uncertainty in the eval- uated concentrations, attributable to the CARS instru- ment on a shot-to-shot basis, was studied by measure- ments on methane/nitrogen mixtures in a cell. Systematic errors in the technique have been identified and dis- cussed. The technique was also demonstrated in a meth- ane/air flame.
ACKNOWLEDGMENTS
This work was financially supported by the Swedish Board for In- dustrial and Technical Development (NUTEK), the Swedish Research Council for Engineering Sciences (TFR), and AB Sydkraft.
1. A.C. Eckbreth, Laser Diagnostics for Combustion Temperature and Species (Abacus Press, Cambridge, Massachusetts, 1987).
2. D. A. Greenhalgh, "Quantitative CARS Spectroscopy", in Advances in Non-linear Spectroscopy, R. J. H. Clark and R. E. Hester, Eds. (Wiley and Sons, New York, 1988), pp. 193-251.
3. W. Stricker and W. Meier, "The Use of CARS for Temperature Measurements in Practical Flames", in Trends in Applied Spec- troscopy (Council of Scientific Research Integration Education, Re- search Trends, Trivandrum, India, 1993).
APPLIED SPECTROSCOPY 191
4. R. P. Lucht, Opt. Lett.12, 78 (1987). 5. A. C. Eckbreth, T. J. Anderson, and G. M. Dobbs, Appl. Phys. B
45, 215 (1988). 6. A. C. Eckbreth and T. J. Anderson, Appl. Opt. 24, 2731 (1985). 7. K. Aron, L. E. Harris, and J. Fendell, Appl. Opt. 22, 3604 (1983). 8. W. R. Fenner, H. A. Hyatt, J. M. Kellam, and S. P. S. Porto, JOSA
63, 73 (1973). 9. A. C. Eckbreth and T. J. Anderson, Opt. Lett. 11,496 (1986).
10. M. Ald6n, P.-E. Bengtsson, and H. Edner, Appl. Opt. 25, 4493 (1986).
11. A. C. Eckbreth, Appl. Phys. Lett. 32, 421 (1978). 12. Y. Prior, Appl. Opt. 19, 1741 (1980). 13. J. A. Shirley, R. J. Hall, and A. C. Eckbreth, Opt. Lett. 5, 380
(1980). 14. L. Martinsson, P.-E. Bengtsson, M. Ald6n, and S. Kr/511, in Tem-
perature: Its Measurements and Control in Science and Industry, Vol. 6 (American Institute of Physics, New York, 1992), pp. 679- 684.
15. L. Martinsson, P.-E. Bengtsson, M. Ald6n, S. Kr611, and J. Bonamy, J. Chem. Phys. 99, 2466 (1993).
16. L. Martinsson, P.-E. Bengtsson, and M. Ald6n, "Oxygen Concen-
tration and Temperature Measurements in N2-02 Mixtures Using Rotational Coherent Anti-Stokes Raman Spectroscopy", paper in press.
17. P.-E. Bengtsson, L. Martinsson, M. Ald6n, and S. Kr/511, Combust. Sci. Techn. 81, 129 (1992).
18. P.-E. Bengtsson, L. Martinsson, M. Ald6n, B. Johansson, B. Lasses- son, K. Marforio, and G. Lundholm, "Dual Broadband Rotational CARS Measurements in an IC Engine", in Proceedings from the 25th International Symposium on Combustion (Combustion Insti- tute, Pittsburg, Pennsylvania, 1994), in press.
19. R. P. Lucht, R. E. Palmer, and M. A. Maris, Opt. Lett. 12, 386 (1987).
20. M. Ald6n, P.-E. Bengtsson, H. Edner, S. Kr611, and D. Nilsson, Appl. Opt. 28, 3206 (1989); Erratum, Appl. Opt. 29, 4434 (1990).
21. T. Lundeen, S.-Y. Hou, and J. Nibler, J. Chem. Phys. 79, 6301 (1983).
22. T. Dreier, B. Lange, J. Wolfrum, and M. Zahn, Appl. Phys. B 45, 183 (1988).
23. A. Gierulski, M. Noda, T. Yamamoto, G. Marowsky, and A. Slen- czka, Opt. Lett. 12, 608 (1987).
