Demonstration of high-speed 1D Raman scattering line imaging

Transcription

1 Appl Phys B (2010) 101: 1 5 DOI /s RAPID COMMUNICATION Demonstration of high-speed 1D Raman scattering line imaging K.N. Gabet N. Jiang W.R. Lempert J.A. Sutton Received: 7 July 2010 / Revised version: 26 July 2010 / Published online: 11 September 2010 Springer-Verlag 2010 Abstract In this manuscript, we demonstrate high-speed (10-kHz acquisition rate) 1D Raman scattering line imaging under turbulent flow conditions. Because of the inherently weak scattering process, Raman scattering measurements typically require high laser pulse energies that are not available from commercial high-repetition-rate laser systems. Using a custom pulse-burst laser system at OSU, we demonstrate the ability to generate a series of high-energy laser pulses with 100 µs pulse separations, which can be used for high-speed Raman scattering imaging. The ultimate goal is a capability of quantitatively measuring all major combustion species and deducing time-varying mixture fraction profiles in turbulent combustion environments. Toward this goal, we report initial results depicting the Raman scattering imaging of O 2,N 2,CH 4, and H 2 in a turbulent non-reacting CH 4 /H 2 jet issuing into air. To the authors knowledge, this represents the first reported temporally-sequential 1D image sequences of major species measured via high-speed Raman scattering. 1 Introduction Spontaneous Raman scattering (SRS) is a well-established technique for the simultaneous measurement of major species concentrations, which, for example, when combined with Rayleigh scattering (for temperature measurements) allows the deduction of the mixture fraction (ξ) in nonpremixed combustion environments. Single point and line K.N. Gabet N. Jiang W.R. Lempert J.A. Sutton ( ) Department of Mechanical and Aerospace Engineering, The Ohio State University, E447 Scott Laboratory, 201 West 19th Avenue, Columbus, OH 43210, USA sutton.235@osu.edu Fax: imaging measurements have been paramount in furthering the understanding of species mixing and turbulent-chemistry interactions under combusting conditions (see, e.g., [1 9]). Because turbulent combustion processes are highly transient, species concentration measurements that are resolved in both space and time are highly desired. However, this requirement dictates that the data acquisition of scalar fields should be at a much faster sampling rate than typical characteristic time-scales of turbulent processes ( 1kHz). Recent advances in solid-state, diode-pumped lasers and high-speed camera (e.g., CMOS) technology have allowed laser diagnostic approaches such as particle-imaging velocimetry (PIV) and planar laser-induced fluorescence (PLIF) to be extended into the high-speed (khz acquisition rate)domain (see,e.g., [10 23]). While high-speed PIV measurements require low pulse energies due to high signal levels from the Mie scattering of particles, scalar measurements are much more complicated; thus, high-speed imaging has been restricted to tracer-lif measurements under non-reacting conditions (see, e.g., [10 13, 16, 18, 23]) and OH PLIF under reacting conditions (see, e.g., [14, 15, 17, 19 22]). Considering only reacting flows for the moment, high-speed OH PLIF imaging has been very useful in characterizing transient events such as ignition and extinction; however, the access to important fundamental scalars such as temperature and mixture fraction has not been available. Mixture fraction (ξ)imaging has been an important focus of combustion diagnostics for more than two decades, with several approaches using various combinations of laser diagnostics including Raman scattering, Rayleigh scattering, and laser-induced fluorescence (or variants thereof) to deduce ξ (see [24] and references within). However, the measurement of all of the major species via Raman scattering should yield the best accuracy in mixture fraction over a broad range of conditions. While Raman (and/or Rayleigh)

2 2 K.N. Gabet et al. scattering diagnostics are common at low-repetition rates, the low pulse energies from commercially-available highrepetition-rate laser systems are prohibitive to Raman (and Rayleigh) scattering diagnostics. The application of these imaging techniques in a high-speed or temporally-sequential manner would represent a significant step forward toward understanding the physical processes governing turbulent flow and flame dynamics. In this paper, we present the recent success in our laboratory using pulse-burst laser technology (see, e.g., [25 28]) to produce a series of high-energy laser pulses that are used for high-speed 1D Raman scattering in a turbulent non-reacting CH 4 /H 2 jet issuing into air. 2 Experiment In this paper, we consider a turbulent isothermal jet issuing from a circular tube into an annular, low-speed (0.3 m/s), co-flowing stream of air. The jet fluid, which is a mixture of 40% CH 4 and 60% H 2 by volume, issues from the fuel tube at 25 m/s corresponding to a Reynolds number of 9,000 based on tube diameter. The jet-fluid mixture was chosen so that the mixture-averaged Rayleigh scattering cross-section was constant across pure jet fluid, pure air, and any mixture of the two streams. In this way, a simultaneous Rayleigh scattering measurement 1 allows for a simple correction for pulse-to-pulse energy fluctuations under the assumption of negligible differential diffusion effects. However, it should be noted that differential diffusion effects that may arise from hydrocarbon/h 2 mixtures are explicitly accounted for under combusting conditions since the temperature and major species are solved simultaneously using an iterative process (see, e.g., [1 9]). Measurements are performed at x/d = 10 and r/d = 1, where x is the axial position, r is the radial position, and D is the tube diameter. 2.1 Pulse-burst laser system The pulse-burst laser system at Ohio State, shown schematically in Fig. 1a, has been described in detail previously in Ref. [26] and thus will only be described briefly here. The laser system is a master oscillator, power amplifier (MOPA) design, which consists of a single-frequency (<10 3 cm 1 ) cw diode-pumped ring laser operating at 1064 nm serving as the primary oscillator, an electro-optic dual Pockels cell 1 Simultaneous Rayleigh/Raman scattering measurement capabilities are integrated into the collection system as discussed in Sect Under the non-reacting conditions currently reported, the Rayleigh scattering signal is simply used as a pulse-energy monitor since the number density and Rayleigh scattering cross-sections are constant throughout the flowfield; while under combustion conditions, the Rayleigh scattering channel provides temperature information and an additional external pulse energy monitor is added to the system. Fig. 1 (a) Schematic diagram and optical layout of pulse-burst Nd:YAG laser at OSU. (b) A typical ten-pulse laser burst trace at 532 nm with 100 µs inter-pulse spacing (10-kHz acquisition rate). Pulse-to-pulse intensity variations are less than 8% pulse slicer, and a series of flashlamp-pumped Nd:YAG amplifiers. The cw laser is initially pre-amplified in a doublepass variable pulse width ( ms) flashlamp-pumped amplifier and then formed into a burst of laser pulses by rapidly rotating the polarization of the pre-amplified pulse by one of two Pockels cells as described by Wu et al. [25]. In the present experiment, the slicing process creates a train of 15-ns wide pulses, which are separated by 100 µs, corresponding to a repetition rate of 10 khz. The pulse train is then further amplified by a series of four additional flashlamp-pumped Nd:YAG amplifiers, resulting in a system gain of To reduce amplified spontaneous emission (ASE) buildup in the system, a phase conjugate mirror (PCM) is placed between amplification stages 3 and 4 [26]. The PCM is an optical cell filled with a high index-of refraction liquid (e.g., FC-75 or CS 2 ) that uses the principle of stimulated Brillouin scattering (SBS) to act as an intensity filter and break the unwanted ASE growth. In addition, the SBS PCM eliminates the low-intensity pedestal which is superimposed on

3 Demonstration of high-speed 1D Raman scattering line imaging 3 the desired pulse-burst sequence due to the finite on/off contrast ratio of the Pockels cell pulse slicer. If the pump beam intensity is above a minimum threshold, a coherent beam is backscattered at 180 and the desired high-intensity laser burst is backscattered toward the final amplifier stages, while the sources of low-intensity background (e.g., ASE) do not exceed minimum threshold and pass through the PCM cell. Finally, the series of 1064-nm laser pulses are frequencydoubled to 532 nm with a KD*P crystal. In the current work, the individual energies of the 532-nm pulses are limited to 150 mj to avoid dielectric breakdown of the gases in the probe volume. Future work will include a three-leg pulse stretcher so that higher laser pulse energies can be used. An example series of 532-nm pulses are shown in Fig. 1(b), where the intensity difference between the pulses is less than 8%. 2.2 Raman (and Rayleigh) scattering line imaging system We have assembled a Raman line imaging system similar to that at Sandia National Laboratories (see, e.g., [1 5, 7, 8]) and the German Aerospace Center (DLR) (see, e.g., [3, 9]), with a few notable exceptions, that combines Rayleigh and Raman scattering to obtain high-speed measurements of temperature and major species in turbulent reacting flows. As a first step, we demonstrate the ability to acquire 10- khz Raman scattering image sequences in an isothermal turbulent jet. Referring to Fig. 2(a) as a reference, the 532- nm output from the pulse-burst laser is focused through a 750-mm focal length plano-convex lens to a spot size of approximately 180 µm. A pair of 150-mm diameter achromats (Qioptic Linos Photonics) collects both Raman and Rayleigh scattered light, which is focused through an adjustable slit of a custom-built high-throughput transmission imaging spectrometer, similar to that described in Ref. [29]. One high-speed CMOS camera (Vision Research, Phantom 710) for the Rayleigh scattering and one intensified (LaVision HS-IRO) CMOS camera (Vision Research, Phantom 710) are integrated into the imaging spectrometer. As shown in Fig. 2(b), the optical system consists of five commercially-available camera lenses, wavelength filtering optics, and one custom transmission holographic grating (Kaiser Optical) to disperse the species-specific Raman-shifted light. Upon entering into the spectrometer, the Rayleigh- scattered (532-nm) portion of the collected light is split off using a dichroic beamsplitter and focused onto one of the CMOS cameras for 1D Rayleigh imaging with a magnification of 1.2. Again, in the present study, the Rayleigh signal is simply used to correct for laser pulseto-pulse variations. The remaining light ( nm) is transmitted through the optical train of the Raman spectrometer. The four camera lenses are specifically matched to the desired magnification ratio and spectral dispersion of Fig. 2 (a) Overview of experimental setup for high-speed 1D Raman/Rayleigh scattering imaging. (b) Detailed schematic and optical layout of the high-throughput, imaging spectrometer, which integrates the cameras for both Raman and Rayleigh scattering into one unit the system. The four lenses used in conjunction with the Raman-portion of the optical system are all Nikon camera lenses consisting of 85-mm f/1.4, 50-mm f/1.2, 85-mm f/1.4, and 105-mm f/1.8 lenses. This combination of lenses (when coupled to the 150-mm diameter f/4 and f/2 collection achromats) results in an overall system magnification of 1.5 for the Raman scattering. The Raman spectrum is dispersed through the custom transmission grating (1200 lines/mm, 22 incident/refracted angles, >85% efficiency from 550 to 680 nm) and onto one intensified-cmos camera. 3 Sample results and discussion The demonstration of high-speed 1D Raman scattering imaging is conducted under very favorable conditions, that

4 4 K.N. Gabet et al. Fig kHz sequence of 1D Raman scattering images in a turbulent CH 4 /H 2 jet issuing into a co-flowing stream of air. Images are centered at x/d = 10 and r/d = 1 is, under isothermal, non-reacting flow conditions. These test conditions represent those of higher signal strength and will serve as a benchmark test case for future experiments with signal collection enhancements. Figure 3 shows a tenimage sequence obtained at a 10-kHz acquisition rate of the major species profiles (O 2,N 2,CH 4, and H 2 ) in the turbulent CH 4 /H 2 jet issuing into air. Image processing consists of darkfield and background image subtraction, corrections for camera/intensifier spectral response, and normalization by laser energy. Since quantitative species measurements are not obtained in the present demonstration, corrections such as non-uniform sensor (pixel) response were not performed. Each image corresponds to a field-of-view of 7 mm, which currently is limited by the width of the entrance slit to the spectrometer. However, the smaller field-of-view is beneficial when using a two-stage image intensifier (as in the present study) as the use of image intensifiers can result in artificial gradient broadening in regions of large image gradients due to local MTF and blurring effects. For small field-of-views, the physical gradients of the species concentrations are spread out over larger portions of the sensor and subsequently are not as adversely affected as would be the case of larger field-of-views with larger gradients with respect to finite pixel spacing. In order to increase the signalto-noise ratio (SNR), the images were smoothed with a 13 pixel median filter along the spatial dimension of the CMOS sensor. This results in an in-plane spatial resolution of approximately 170 µm, which is matched to the spatial resolution defined by the focused laser beam. It should be noted that by simple median filtering, we can reduce the shot noise imposed by the intensifier, but we cannot reduce readout noise, as the current CMOS sensors cannot be binned on chip. Future work will include investigating alternative noise-suppression methodologies that are appropriate for the high-speed imaging. The images in Fig. 3 were broken into seven columns corresponding to the rotational-vibrational Raman bands of the major species (O 2,N 2,CH 4, and H 2 ) and potential background interferences at selected spectral regions (negligible under non-reacting conditions). Each column (nominally 15 pixels wide) was integrated in the spectral dimension of the sensor; the integrated columns corresponding to background Fig. 4 Example radial profiles (from the images of Fig. 3) of the Raman scattering of CH 4 as a function of time. Radial distance (in mm) is measured from the jet centerline. Every other radial profile is plotted in red for viewing clarity interference was subtracted from those corresponding to major species profiles; and the resulting radial profiles of O 2, N 2,CH 4, and H 2 were determined as a function of time. An example of one of these sets of profiles is plotted for CH 4 (in arbitrary units) in Fig. 4, where only the first 600 µs have been shown for clarity. To estimate the SNR of the current measurements, we consider regions of nearly constant air for O 2 and N 2 and regions in the fuel core for CH 4 and H 2. From these image sequences, we calculate the SNR of O 2, N 2,CH 4, and H 2 as 6, 11, 30, and 5, respectively. The SNR values for O 2 and N 2 agree with those made in still air under nearly identical laser and signal collection conditions. However, it is clear that the SNR does not scale as the square root of the signal (as would be expected under shot-noise - limited systems), indicating that other sources such as the read noise of the HS-IRO/CMOS system is significant for this lower level of laser energy. While the SNR of the presented images are modest at this point, it should be pointed out that the images were acquired

5 Demonstration of high-speed 1D Raman scattering line imaging 5 with only 150 mj/pulse. As noted within the literature, conventional low repetition-rate Raman scattering imaging is performed with >1 Joule/pulse (see, e.g., [1 9]). We are currently building a next-generation burst-mode laser system that is designed for both increases in the laser output pulse energy (>1 J/pulse) and burst duration ( 250 pulses/sequential images). Combining this increase in laser pulse energy with a new red-enhanced high-speed intensifier now available within our laboratory, we expect increases of signal levels by a factor of and SNR by a factor of >4. Acknowledgements The support of Air Force Office of Scientific Research grant FA (Julian Tishkoff Technical Monitor) is greatly appreciated. The authors acknowledge previous financial support for the development of the pulse-burst laser system from NASA (Paul Danehy Technical Monitor), the U.S. Air Force Research Laboratory Propulsion Directorate (James Gord Technical Monitor), and the Air Force Office of Scientific Research (J. Schmisseur program monitor). The authors also wish to thank R.S. Barlow from Sandia National Laboratories for very useful discussions concerning the Raman scattering imaging system. References 1. J.H. Frank, R.S. Barlow, Proc. Combust. Inst. 27, 759 (1998) 2. R.S. Barlow, J.H. Frank, Proc. Combust. Inst. 27, 1087 (1998) 3. W. Meier, R.S. Barlow, Y.L. Chen, J.Y. Chen, Combust. Flame 123, 326 (2000) 4. A.N.Karpetis,R.S.Barlow,Proc.Combust. Inst.29, 1929 (2002) 5. R.S. Barlow, A.N. Karpetis, Proc. Combust. Inst. 30, 673 (2005) 6. D. Geyer, A. Kempf, A. Dreizler, J. Janicka, Proc. Combust. Inst. 30, 681 (2005) 7. R.S. Barlow, A.N. Karpetis, Flow Turbul. Combust. 72, 427 (2005) 8. G. Wang, A.N. Karpetis, R.S. Barlow, Combust. Flame 148, 62 (2007) 9. L. Wehr, W. Meier, P. Kutne, C. Hassa, Proc. Combust. Inst. 31, 3099 (2007) 10. J.D. Smith, V. Sick, Appl. Phys. B 81, 579 (2005) 11. J.D. Smith, V. Sick, Appl. Opt. 44, 6682 (2005) 12. C.M. Fajardo, J.D. Smith, V. Sick, Appl. Phys. B 85, 25 (2006) 13. C.M. Fajardo, V. Sick, Proc. Combust. Inst. 31, 3023 (2007) 14. C. Kittler, A. Dreizler, Appl. Phys. B 89, 163 (2007) 15. W. Paa, W. Muller, M. Stafast, W. Treibel, Appl. Phys. B 86, 1 (2007) 16. C.M. Fajardo, V. Sick, Exp. Fluids 46, 43 (2009) 17. M.E. Cundy, V. Sick, Appl. Phys. B 96, 241 (2009) 18. B. Peterson, V. Sick, Appl. Phys. B 97, 887 (2009) 19. I. Boxx, C. Heeger, R. Gordon, B. Bohm, A. Dreizler, W. Meier, Combust. Flame 156, 269 (2009) 20. I. Boxx, M. Stohr, C. Carter, W. Meier, Appl. Phys. B 95, 23 (2009) 21. B. Bohm, C. Heeger, I. Boxx, W. Meier, A. Dreizler, Proc. Combust. Inst. 32, 1647 (2009) 22. I. Boxx, C. Heeger, R. Gordon, B. Bohm, M. Aigner, A. Dreizler, W. Meier, Proc. Combust. Inst. 32, 905 (2009) 23. R.L. Gordon, C. Heeger, A. Dreizler, Appl. Phys. B 96, 745 (2009) 24. R.S. Barlow, Proc. Combust. Inst. 31, 49 (2007) 25. P. Wu, W.R. Lempert, R.B. Miles, AIAA J. 38, 672 (2000) 26. B.S. Thurow, N.B. Jiang, M. Samimy, W.R. Lempert, Appl. Opt. 43, 5064 (2005) 27. J.D. Miller, M. Slipchenko, T.R. Meyer, N.B. Jiang, W.R. Lempert, J.R. Gord, Opt. Lett. 34, 1309 (2009) 28. N. Jiang, M. Webster, W.R. Lempert, Appl. Opt. 48, B23 (2009) 29. R.S. Barlow, G.-H. Wang, P. Anselmo-Filho, M.S. Sweeney, S. Hochgreb, Proc. Combust. Inst. 32, 945 (2009)