MEASUREMENT AND MODELING OF OH, NO, AND CO 2 INFRARED RADIATION IN A LOW TEMPERATURE AIR PLASMA

Transcription

1 MEASUREMENT AND MODELING OF OH, NO, AND CO 2 INFRARED RADIATION IN A LOW TEMPERATURE AIR PLASMA Denis M. Packan, * Richard J. Gessman, * Laurent Pierrot, Christophe O. Laux, and Charles H. Kruger Department of Mechanical Engineering Stanford University, CA Abstract Measurements were made of the infrared emission spectrum of a low temperature (~34 K air plasma containing small quantities of CO 2 and O. A 5 kw radio-frequency inductively-coupled plasma torch was used to produce the air plasma in local thermodynamic equilibrium. The absolute intensity spectra obtained were compared to numerical simulations with the NEQAIR2-IR code. The spectroscopic models tested include the infrared radiation of OH and NO, and the absorption of plasma emission by room-air CO 2 and O. The emission from CO 2 (ν 1 +ν 3 and (ν 3 bands was also modeled, using a correlated-k model. Comparisons between experiment and calculations are presented. Introduction The infrared spectral emission and absorption of air is an important issue for in-flight and ground-based infrared signature detection. While atmospheric transmission windows are well documented, the radiation of air plasmas in the infrared has been the object of few experimental investigations. Accurate radiation models are also important for optical diagnostics to determine temperatures and concentrations in both equilibrium and nonequilibrium plasmas. The measurements presented here represent the second phase of an effort conducted in our research facilities to model and understand the infrared radiation of air. The 5kW plasma torch facility can produce LTE air plasmas in a controlled manner at both high and low temperatures, allowing measurements under reproducible and well-known conditions. The first phase of our effort included high temperature air (~8K infrared radiation experiments and modeling. These measurements led to the development of the radiation code NEQAIR2-IR, 1 as an extension of NASA code NEQAIR. 2 The second phase, presented here, focuses on low temperature (~34K LTE air plasma, where molecular radiation is predominant and atomic line radiation negligible. The current measurements allowed us to test new spectroscopic models for OH * Graduate Research Assistant, AIAA Student Member. Postdoctoral Fellow, AIAA Member. Research Associate, AIAA Member. Professor, Vice-Provost, Dean of Research and Graduate Policy, AIAA Member. Copyright 1999 by the authors. Published by the, Inc. with permission 1

2 AIAA infrared emission and room air (CO 2 and O absorption. CO 2 emission was also taken into account using the correlated-k model implemented by Pierrot et al. 3 model SDD-2E1-S1 indium-antimonide cryogenically-cooled (liquid N 2 infrared detector with integrated pre-amplifier. A 3-grooves/mm grating blazed at 4. µm was used at first order, and higher order spectra were rejected by using two long-pass filters with cut-offs at 1.5 µm and 3. µm, respectively. The entrance slit was 1 mm and the exit slit 2.8 mm. al Setup and Conditions The experimental work was conducted with a 5 kw TAFA model 66 RF induction plasma torch powered by a LEPEL model T-5-3 power supply operating at a frequency of 4 MHz. As shown in Figure 1, a copper nozzle of 1.-cm exit diameter was utilized at the top of the upward firing torch, and the plasma was cooled while flowing through a watercooled brass test section 6 cm in length. All measurements were made 1 cm downstream of the exit of the test section, in a region where the flow is laminar. Figure 1. Torch head schematic, presented here with a ~2 cm-long test section instead of the 6 cmlong used in the experiment. The set-up for spectral measurements includes a SPEX model 75M 3/4 meter scanning monochromator fitted with a Cincinnati Electronics Figure 2. Setup for optical diagnostics. The air injected into the torch contained ~33ppm of CO 2 and ~ mole fraction of water vapor. The latter number was determined a posteriori (see page 7. The model for OH was thus tested in relative intensity only. The temperature and humidity in room air were recorded during the experiment and calibration. In both cases, the measured ambient temperature was 25 C and the relative humidity was 42% (1% uncertainty, which correspond to a mole fraction of O of.13±.1 (Reynolds and Perkins, 4 p The spectral range of the measurements covered µm so as to capture NO first overtone and OH fundamental rovibrational bands. The measured temperature profile was obtained from Abel-inverted profiles of absolute NO (1- bandhead intensities. The temperature profile had a maximum value of 34 K at the center of the plasma plume. The plasma produced by the torch under the present conditions was found to be close to Local Thermodynamic Equilibrium in previous work. 5 The concentrations of the different species used in the simulations were therefore 2

3 AIAA calculated by assuming chemical equilibrium at atmospheric pressure and at the measured temperatures. The mole fractions of the main species are plotted in Figure 3, along with the measured temperature profile. Temperature [K] CO 2 (x1 NO (/1 T OH 22 O Radius [cm] Figure 3. Radial temperature and mole fraction profiles at the exit of the test section. Spectral Calibration Absolute intensity calibrations were made using an Optronics Laboratories calibrated tungsten lamp (model OL55 traceable to NIST standards. Water vapor and carbon dioxide present over the 6- meter long optical path that separates the plasma source or calibration lamp from the detector can absorb a significant fraction of emission between 1.5 and 2.9 µm and at wavelengths 4. µm. Both the calibration signal and the experimental spectrum were affected by room-air absorption. It could be thought possible to recover the calibrated plasma torch emission spectrum by simply dividing the measured plasma torch signal by the calibration signal. Such a procedure would be flawed as absorption lines of room air may occur at different wavelengths than plasma emission lines. In other words, the absorption lines of room-air water vapor and carbon dioxide always contribute to absorption features in the calibration spectrum, because of the continuum nature of the emission of the calibration lamp, but do Mole fraction not necessarily absorb plasma emission. This is illustrated in Figure 4. Figure 4 shows a small portion of the emission spectrum of NO fundamental before convolution (bottom curve, along with a plot of the spectral transmittance of 6 meters of room air (top curve. The O lines at 3449 Å and 346 Å, for example, fall in between NO emission lines and thus do not absorb plasma emission, but absorb the continuous emission of the calibration lamp at these wavelengths. Therefore the corresponding absorption feature in the calibration spectrum does not appear and should not be corrected for in the plasma spectrum. Intensity [mw/(cm 2.sr. µm] O transmittance NO emission lines Wavelength [A] Figure 4. NO fundamental emission lines and room-air transmittance. The correct procedure to calibrate the spectrum is thus to deconvolve room air absorption from the calibration signal, and then to divide the plasma emission signal by the corrected (i.e. room-air absorption-corrected calibration signal. With this procedure, the calibrated plasma spectrum represents the spectrum emitted by the plasma and absorbed by room air, and the absorption features of water vapor and carbon dioxide remain part of the experimental emission spectrum. In order to model this spectrum, it is then necessary to determine the spectral emission of the plasma and then solve the transport equation over a path of room air. The deconvolution procedure is presented next Transmittance 3

4 AIAA Deconvolution of room-air absorption features in the calibration signal: The raw calibration signal (detector voltage is shown in Figure 5..1 Transmittance Detector signal [V] 1E-3 O O + CO 2 CO 2 O Figure 6. Calculated O contribution to the spectral transmittance of 6 meters of room air. 1E Figure 5. Raw calibration spectrum obtained with the OL55 calibration standard. The absorption features of room air species (CO 2 and O are indicated. Absorption by CO 2 at 4.3 µm can be removed by a straight line interpolation. This approximation is certainly valid because the OL55 tungsten filament emission and the grating response are smoothly varying in that spectral range. The other absorption features occur over wider wavelength ranges and cannot be deconvolved using such a simple procedure. For these features, our deconvolution procedure consists in dividing the calibration signal by the spectral transmission spectrum (convolved with the instrumental slit function of room air. The transmission spectrum of room air was obtained with the HITRAN96 database 6 with the following parameters: room temperature of 298 K, optical pathlength through room air equal to 6 meters, O mole fraction of.13 and CO 2 mole fraction of 33 ppm. Figure 6 and Figure 7 show the calculated contributions of water vapor and carbon dioxide, respectively, to the transmittance of a 6 meter long room-air path. Transmittance Figure 7. Calculated CO 2 contribution to the spectral transmittance of 6 meters of room air. Results of the deconvolution procedure around 1.8 and 2.7 µm are presented in Figure 8. 4

5 AIAA Detector Signal [V].1 Raw Calibration ( O+CO 2 -recovered Calibration Polynomial fit Detector Signal [V] 1E-3 1E-4 Raw Calibration O-recovered Calibration Averaged Calibration Figure 8. Absorption-deconvolved calibration spectrum in range µm (The absorption features are due to room air O and CO 2. The red curve represents the corrected calibration using the spectral absorption coefficient determined with HITRAN. As can be seen in Figure 8, the deconvolution procedure at 1.8 µm produces a smoothly-varying trace. At 2.7 µm, the recovered spectrum still shows residual O absorption lines which cannot be removed by varying the O mole fraction. We believe that these residual features could be due to inaccuracies in the HITRAN database or to uncertainties on the instrumental slit-function, which was deduced from the entrance and exit slit widths and the theoretical reciprocal linear dispersion of the monochromator. 1 In future work, the slit functions will be measured experimentally with a monochromatic light source to increase the accuracy of the deconvolution procedure. The recovered calibration spectrum between 2.5 µm and 3. µm shows a cusp and an S-shape variation that is likely due to the spectral response of the ruled grating. Since the grating spectral response should be relatively smooth, we replaced the part of the red curve between 2.5 and 3. µm with a fourth order polynomial fit (blue curve. Results of the deconvolution procedure between 4.5 µm and 5.5 µm are presented in Figure Figure 9. Absorption-deconvolved calibration spectrum in range µm (The absorption features are due to room air O. In this range, room air absorption is mostly due to O. As can be seen from the figure, the deconvolved calibration signal exhibits smooth variations in this spectral range. The small oscillations on the red curve are due to etalonning interference by the quartz window of the tungsten calibration lamp. These oscillations were finally smoothed out using adjacent averaging (blue curve. Figure 1 shows the summary comparison between the raw calibration spectrum and the deconvolved calibration spectrum. Detector Signal [V].1 1E-3 Raw Calibration Recovered calibration 1E Figure 1. Calibration spectrum corrected for water vapor and carbon dioxide absorption. 5

6 AIAA Radiation Model The measured emission spectrum, corrected for the spectral response of the detection system and calibrated in absolute intensity using the procedure discussed in the foregoing section, is presented in Figure 11. Intensity [ µ W/(cm 2.sr] OH Fundamental NO First Overtone CO 2 (ν 1 +ν 3 NO Fundamental CO 2 (ν Figure 11. Measured IR emission spectrum (present work: P=1 atm, maximum plasma temperature = ~34K. The spectrum shows the fundamental bands of NO at ~5 µm, the ν 3 band of CO 2 at ~4.3 µm, and the lines of OH fundamental and of the NO first overtone ( v=2 along with the (ν 1 +ν 3 band of CO 2 between 2.5 µm and 4.15 µm. We summarize next our modeling efforts for these radiative transitions. NO rovibrational bands The fundamental ( v = 1 and first overtone ( v = 2 rovibrational bands of NO (X 2 Π are clearly seen in the experimental spectrum shown in Figure 11. A detailed, accurate model of fundamental and overtone bands of NO (X 2 Π was implemented in NEQAIR2-IR, as previously described in Ref. 1 The code determines rotational line positions by diagonalizing the Hamiltonian of Amiot. 7 For each vibrational band, vibrational dipole moments M v'v" defined as: ( M 2 ( 2 v'v" = Ψv' rde( r Ψv' ( rdr, where D e stands for the electric dipole moment function (EDMF, were determined using the accurate ab initio EDMF of Langhoff et al. 8 The Hönl-London factor expressions (corresponding to Hund s case a recommended by Spencer et al. 9 were employed. The model provides accurate line intensities and spectral positions, which are of particular importance for highresolution spectroscopic diagnostics and for the accurate simulation of absorption of fundamental and overtone NO bands by atmospheric water vapor. CO 2 bands: ν 3 antisymmetric stretch at 4.3 µm and (ν 1+ν 3 band at 2.7 µm Computations of the CO 2 band spectrum at 2.7 and 4.3 µm were obtained with a correlated-k model 1 for CO 2 implemented by Pierrot et al. 3 The correlated-k model is a narrow-band model in which the actual spectrum is replaced on each narrow band by the reordered spectrum, so that the spectral integration is carried out using typically 1 points instead of several thousands. In the case of CO 2 infrared radiation, this model affords radiative intensity predictions within a few percent accuracy. 11 The parameters used for the simulations presented in this paper are based on a 16- point Gaussian quadrature and a spectral decomposition over intervals of width 25 cm -1. OH rovibrational bands The fundamental ( v = 1 bands of OH (X 2 Π are clearly seen between 2.5 and 4 µm in the experimental spectrum shown in Figure 12. As for the NO infrared bands, an accurate line-by-line model of fundamental (and overtone bands of OH (X 2 Π was incorporated in NEQAIR2-IR. Rovibrational term energies and line positions for the 1- and 2-1 bands of this transition are determined by diagonalizing the Hamiltonian of Stark et al. 12 (with corrections of Levin et al. 13. As for the case of NO infrared bands, this model provides the highly accurate spectral positions required for reliable simulations of absorption by atmospheric water vapor. In future work, we also intend to incorporate additional bands originating in higher vibrational levels, with line 6

7 AIAA positions for these levels determined using the parameters of Coxon. 14 Transition probabilities of the OH infrared lines have a very strong dependence on centrifugal distortions of the vibrational potentials. As a result, the P and R branches show an anomalous distribution with intense P branches (2.6 to 4. µm and very weak R branches (2.4 to 2.6 µm. In the NEQAIR2-IR model, we utilized the P- and R-branch transition probabilities determined by Holtzclaw et al. 15 Room air absorption In this work, the HITRAN96 database 6 was used to determine the transmittance spectrum of room air over a 6-meter optical path. CO 2 concentration was taken equal to 3 ppm. An O mole fraction of.14 was used. The procedure for taking into account absorption is as follows: first, the emission spectrum of the plasma is computed at high spectral resolution (1 points per angström, or approximately 1 points per line. Then, attenuation of this spectrum as a result of water vapor and carbon dioxide absorption is determined with Beer s law, using the line strengths and spectral broadening coefficients of O and CO 2 (HITRAN96 database. Finally, the resulting spectrum is convolved with the instrumental slit function. various radiating systems of importance in this spectral range are shown separately in Figure 13. It should be noted here that the mole fraction of OH was determined by matching the measured absolute intensities of rotational lines of the P-branch of OH. The OH concentration determined in this manner was then used to infer, using chemical equilibrium relations, the mole fraction of water in the air injected inside the plasma torch. As already mentioned in the introduction, the mole fraction of water injected in the torch was thus found to be approximately This value is significantly lower than the mole fraction of water vapor in room air (.13. This difference is not surprising as the air injected in the torch was prepared by compressing atmospheric air at an earlier time when the relative humidity may have been lower. Significant amount of water vapor may also have been removed by the water trap mounted at the exit of the compressed air tank. Work is in progress to accurately monitor the amount of water vapor injected in the plasma torch during experiments. Absorption by water vapor in the 6-meter optical path is particularly significant between 2.5 and 2.95 µm, as can be seen in Figure 14 where spectral simulations obtained with and without water absorption are compared with the measured spectrum. Comparison of measured and computed spectra Range µm The spectrum measured over the range µm (black curve is compared with the spectral predictions of NEQAIR2-IR (red curve in Figure 12. As can be seen, all spectral features are well reproduced by the simulations, except between 2.8 and 3. µm where the model underpredicts the measurements (possibly because we have not yet incorporated the (3-2 band of OH into NEQAIR2-IR. The contributions of the 7

8 AIAA Intensity [ µ W/(cm 2.sr] NEQAIR2-IR: OH fund., NO first overtone, CO 2 (ν 1 +ν 3 Intensity [ µw/(cm 2.sr] NEQAIR2-IR, without water vapor absoprtion NEQAIR2-IR, with water vapor absoprtion Figure 12. Comparison between measurements and models of infrared emission by air at ~34 K and 1 atm. The simulations incorporate the effect of water vapor and carbon dioxide absorption over the 6-meter length of room air between the plasma and the detector. Intensity [ µ W/(cm 2.sr] OH (1- R-branch bandhead NO (2- bandhead CO 2 (ν 1 +ν 3 NEQAIR2-IR: NO first overtone NEQAIR2-IR: OH fundamental Correlated-k model: CO 2 (ν 1 +ν 3 OH (1- P-Branch... P 1,2 (N'=13 P 1,2 ( CO 2 (ν 3 Figure 13. Contributions of NO first overtone, OH fundamental, and CO 2 (ν 1 +ν 3 bands to the total emission spectrum in the range µm. Note the abnormally weak OH R branch. These simulations incorporate room air absorption over a 6-meter pathlength. This comparison underscores the importance of computing highly accurate positions for all emission and absorption spectral lines, in the present case those of the NO overtone, OH fundamental, and O absorption bands Figure 14. Comparison between the measured emission spectrum and NEQAIR2-IR simulations with and without absorption by water vapor in the optical path. Range µm The measured CO 2 ν 3 band spectrum can be compared in Figure 15 with the predictions of the correlated-k model presented earlier. The model also accounts for absorption by room air CO 2 over the optical path separating the plasma from the detector. Intensity [µw/(cm 2.sr] 1 5 Simulated CO 2 (ν 3 Spectrum (Correlated-k Model NO 1- Bandhead Figure 15. CO 2 spectrum computed with the c-k model parameters of the EM2C Laboratory, 11,16 and comparison with the measured spectrum. Note the effects of absorption by room air CO 2 in the range µm. This absorption is clearly responsible for near extinction in the range µm of the emission from low-lying rotational levels of CO 2. The lines appearing at both edges of the absorbed region correspond to hot CO 2 rotational lines. The model appears to 8

9 AIAA overestimate the measured CO 2 band intensity by approximately 3%. This discrepancy may be due to the fact that the parameters used in the c-k model are only valid at temperatures below 29 K, which is a temperature slightly lower than the centerline temperature of the plasma considered here (~34 K. Another possible explanation for the discrepancy may be that the air injected in the torch contained a lower concentration of CO 2 than the typical 33-ppm. However, for the simulation to agree with the measurements, one would have to assume an unreasonably low CO 2 concentration of ~24 ppm. Thus we believe that the intensity differences between the measured and predicted spectra are more likely due to the use of the c-k model beyond its range of validity. Range µm Figure 16 and Figure 17 compare measurements and modeling results for NO fundamental bands. Two simulated spectra are presented, with and without incorporating the effect of water vapor absorption over the 6-meter path of room air between the plasma and the detector. By matching the depth of the water absorption features in Figure 17, we determined the mole fraction of water vapor in the room to be approximately.14. This value is close to the.13 mole fraction recorded during the experiment, and within the uncertainties of the measurement. As can be seen in Figure 17, the predictions of the model (with O absorption are in excellent agreement with the measured spectrum. It should be noted again that both the measurements and computations are on absolute intensity scales. Intensity [ µw/(cm 2.sr] NEQAIR2-IR: NO fundamental without water absorption Figure 16. NO fundamental band spectrum computed with NEQAIR2-IR without water vapor absorption, and comparison with experimental spectrum. Note that a (small constant value of.8 µw/(cm 2.sr was added to the simulated spectra shown in the figure in order to match the offset, possibly due to underlying continuum radiation, that is apparent at 4.92 µm. Intensity [ µ W/(cm 2.sr] NEQAIR2-IR: NO fundamental with water absorption 5-4 Figure 17. NO fundamental band spectrum computed with NEQAIR2-IR with water vapor absorption, and comparison with experimental spectrum. Conclusion Detailed absolute intensity infrared measurements and modeling of the spectral emission of atmospheric pressure air plasmas at temperatures up to 34 K have been made between 2.4 µm and 5.5 µm. The cold gas injected in the plasma torch contained an estimated mole fraction of water vapor of ~ and an estimated carbon dioxide mole fraction of ~

10 AIAA The main emitting systems are the fundamental and overtone bands of NO and OH, and the (ν 3 and (ν 1 +ν 3 bands of CO 2. Special attention was paid to the effects of ambient air absorption in the optical path between the plasma and the detector. Excellent absolute intensity agreement is obtained between the measured and simulated spectra of NO emission, CO 2 emission, and room-air CO 2 and O absorption. Very good relative intensity agreement was obtained for OH. Our experimental facility is in the process of being upgraded to allow better monitoring of the humidity of air injected in the plasma torch. This will eventually allow us to test the absolute intensity of the OH model. Acknowledgments This work has been supported by the Ballistic Missile Defense Organization and the National Aeronautics and Space Administration by the mean of grant NAG under the cognizance of Drs. David Mann and Winifred Huo. References 1. Laux, C.O., Gessman, R.J., Hilbert, B., and Kruger, C.H., "al Study and Modeling of Infrared Air Plasma Radiation," Proc. 3th AIAA Thermophysics Conference, AIAA , San Diego, CA, Park, C., Nonequilibrium Air Radiation (NEQAIR Program: User's Manual, NASA-Ames Research Center, Report No. NASA-TM8677, Pierrot, L., "A fictitious-gas-based absorption distribution function global model for radiative transfer in hot gases," JQSRT, 62, , Reynolds, W.C. and Perkins, H.C., Engineering Thermodynamics, Mc Graw Hill, New York, Gessman, R.J., Laux, C.O., and Kruger, C.H., "al study of kinetic mechanisms of recombining atmospheric pressure air plasmas," Proc. 28th AIAA Plasmadynamics and Lasers Conference, AIAA , Atlanta, GA, Rothman, L.S., Gamache, R.R., Tipping, R.H., Rinsland, C.P., Smith, M.A.H., Chris Benner, D., Malathy Devi, V., Flaud, J.-M., Camy-Peyret, C., Perrin, A., Goldman, A., Massie, S., Brown, L.R., and Toth, R.A., "The HITRAN Molecular Database: Editions of 1991 and 1992," JQSRT, 48, , Amiot, C., "The Infrared Emission Spectrum of NO: Analysis of the v=3 sequence up to v=22," Journal of Molecular Spectroscopy, 94, 15, Langhoff, S.R. and Bauschlicher, C.W., Jr, "Theoretical Dipole Moment for the X 2Π state of NO," Chem. Phys. Letters, 223, , Spencer, M.N., Chackerian, C., Jr., and Giver, L.P., "The Nitric Oxide Fundamental Band: Frequency and Shape Parameters for Rovibrational Lines," J. Molec. Spectrosc., 165, , Goody, R. and Young, Y., Atmospheric Radiation,, 2nd ed. Oxford, Pierrot, L., "Developpement, Etude Critique et validation de modeles de proprietes radiatives infrarouges de CO 2 et O a haute temperature. Application au calcul des transferts dans des chambres aeronautiques et a la teledetection," Thesis, Ecole Centrale Paris, Stark, G., Brault, J.W., and Abrams, M.C., "Fourier-Transform Spectra of the A 2 Σ + -X 2 Π v = Bands of OH and OD," J. Opt. Soc. Am. B, 11, 3-32, Levin, D.A., Laux, C.O., and Kruger, C.H., "A General Model for the Spectral Radiation Calculation of OH in the Ultraviolet," JQSRT,, 61, , Coxon, J.A., "Optimum Molecular Constants and Term Values for the X 2 Π (v 5 and A 2 Σ + (v 3 States of OH," Can. J. Phys., 58, , Holtzclaw, K.W., Person, J.C., and Green, B.D., "Einstein Coefficients for Emission from High 1

11 AIAA Rotational States of the OH (X 2 Π Radical," JQSRT, 49, , Soufiani, A. and Taine, J., "High Temperature Gas Radiative Property Parameters of Statistical Narrow-band Models for O, CO 2, and CO and correlated-k Model for O and CO 2," Int. Journal of Heat and Mass Transfer, 4, ,