Diagnostic Imaging in Flames with Instantaneous Planar Coherent

Mar 20, 2014 - Coherent anti-Stokes Raman spectroscopy (CARS) was pioneered in measurements of such processes almost 40 years ago and is authoritative...
1 downloads 8 Views 3MB Size
Download PDF
Letter pubs.acs.org/JPCL

Diagnostic Imaging in Flames with Instantaneous Planar Coherent Raman Spectroscopy A. Bohlin and C. J. Kliewer* Sandia National Laboratories, Livermore, California 94551, United States S Supporting Information *

ABSTRACT: Spatial mapping of temperature and molecular species concentrations is vitally important in studies of gaseous chemically reacting flows. Temperature marks the evolution of heat release and energy transfer, while species concentration gradients provide critical information on mixing and chemical reaction. Coherent anti-Stokes Raman spectroscopy (CARS) was pioneered in measurements of such processes almost 40 years ago and is authoritative in terms of the accuracy and precision it may provide. While a reacting flow is fully characterized in three-dimensional space, a limitation of CARS has been its applicability as a point-wise measurement technique, motivating advancement toward CARS imaging, and attempts have been made considering onedimensional probing. Here, we report development of two-dimensional CARS, with the first diagnostics of a planar field in a combusting flow within a single laser pulse, resulting in measured isotherms ranging from 450 K up to typical hydrocarbon flame temperatures of about 2000 K with chemical mapping of O2 and N2. SECTION: Spectroscopy, Photochemistry, and Excited States energies ωpump, ωStokes, and ωprobe are mixed with the internal energy levels of the probed molecules to generate a fourth photon at energy ωCARS according to ωCARS = ωpump − ωStokes + ωprobe. Another condition is phase-matching (momentum conservation), which constrains how the wave vectors of the incident beams must be arranged to effectively generate the coherently scattered light. In experimental situations where high spatial resolution is required, the phase-matching condition for CARS signal generation has traditionally been fulfilled through a three-beam BOXCARS23 configuration, generating signal only where the laser beams are crossed. The current work utilizes near-transform-limited femtosecond (fs) pulses for the pump−Stokes interaction and narrow-band picosecond (ps) pulses for the probe process, termed hybrid fs/ps CARS.24−27 The extreme efficiency of fs impulsive excitation28 in combination with a high-peak-power ps probe pulse results in very strong CARS signals. With the extensive history of CARS, a natural question is why the capability for two-dimensional imaging was not developed earlier. While work on one-dimensional CARS line probing has been demonstrated,29−32 establishing gas-phase CARS for instantaneous planar imaging, however, requires designing a novel scheme for both the signal generation and the signal detection. Because the physical scalars quantifying an event in a reactive flow exist only for an instantaneous moment of time, laser scanning or sample rastering procedures may not be employed. Also, in generating a spatially correlated twodimensional signal, there exists no feasible, spatially well-

C

ombustion of fossil fuels remains the largest source of energy production in the world. Global concerns regarding energy security, environmental pollution, and anthropogenic climate change have motivated a large body of research devoted to the experimental measurement1 and numerical simulation2,3 of combustion processes. To extract precise scalar information of temperature and species concentrations in most realistic flame conditions is a demanding task for any optical instrument. In order to effectively freeze the time scales of a reactive flow process, data needs to be acquired instantaneously, that is, on the basis of a single laser pulse. This criterion is fulfilled inherently by the Raman scattering process, where the temporal resolution is limited only by the time duration of the probe pulse. The excellent chemical selectivity and the coherent nature of the scattered signal beam make CARS a strong candidate for this challenging environment. The CARS spectrum has been shown to be highly sensitive to both species concentration4−6 as well as the rovibrational temperature7 of molecules in the probe volume. CARS is today well-established, and the variants of setups reported over the years are manifold.8−10 Strategies have been developed for probing pure rotational, rovibrational, and electronic spectral regions of molecules, and applications span a wide variety of fields, such as combustion,11−14 cell biology,15,16 plasma physics,17,18 nanocomposite materials,19 and the standoff detection of explosives.20 The variant implemented in the current work is dual broad-band rotational CARS,21,22 first established almost 3 decades ago. It belongs to the family of nonlinear optical four-wave-mixing techniques, which must satisfy two physical conditions. One condition is that of energy conservation, where three incident photons with © 2014 American Chemical Society

Received: February 21, 2014 Accepted: March 20, 2014 Published: March 20, 2014 1243

dx.doi.org/10.1021/jz500384y | J. Phys. Chem. Lett. 2014, 5, 1243−1248

The Journal of Physical Chemistry Letters

Letter

Figure 1. Experimental setup for instantaneous planar coherent Raman spectroscopy. The signal generation follows the two-beam CARS phasematching scheme, intersecting the probe and the pump/Stokes beams in a plane geometry with approximate dimensions of ∼2 × 9 mm. The generated CARS light may be isolated from the probe beam either by utilizing a short-pass frequency filter or by polarization gating. The signal is dispersed with a 3600 l/mm diffraction grating at Littrow’s angle and imaged onto the CCD via double-passing a single effective lens. A line spread function of less than 60 μm has been measured, quantifying the imaging quality.

Figure 2. Spatial maps of specific S-transitions and the derived temperature contour. (a) A spatially resolved planar CARS signal as it is dispersed and detected on the CCD. The folded spectrum extends two horizontal rows and consists of many fully isolated N2 and O2 S-branch transitions. The pixels in each of the individual transitions are mapping a specific spatial location in the probed field. (b) Two specific N2 S-transitions, belonging to a low (J = 7) and high (J = 28) initial rotational quantum state. The intensity depends on the thermal population of the specific states involved, governed by the different zones of the reaction. (c) Derived temperature map ranging from 450 up to ∼2000 K, extending a view of 2.4 mm × 9.1 mm in the flame (see Figure 3a). (d) Examples of rotational N2 CARS spectra originating from a specific spatial location, extracted as the pixel-topixel values of each respective transition.

two-beam CARS setup for gas-phase studies is similar to that of crossed-beam polarization spectroscopy,36−39 also demonstrated in a hybrid fs/ps Raman-induced Kerr effect spectroscopy scheme40 but without the constraint of detecting only the induced birefringence. Indeed, the polarization analyzer may be removed altogether and the probe suppressed solely through the use of a spectral filter when desired. In this work, the two-

resolved arrangement in three overlapping beams (pump, Stokes and probe) using the standard BOXCARS scheme. Recently, a non-phase-matched two-beam CARS scheme was developed for gas-phase measurements,33−35 combining the pump and the Stokes beams into a single coherence excitation beam. Here, the generated CARS light may be isolated from the probe beam utilizing polarization gating. The simplicity of the 1244

dx.doi.org/10.1021/jz500384y | J. Phys. Chem. Lett. 2014, 5, 1243−1248

The Journal of Physical Chemistry Letters

Letter

Figure 3. Interrogated fields, chemical imaging, and transformation to lab view. (a) Photograph of the slightly rich premixed laminar CH4/air flame, indicated with frames for the interrogated fields. The imaged fields are sectioning the measurement object at separate positions, all covering both unreacted and reacted mixtures. (b) Example of chemically specific imaging, where the species N2 and O2 exist in different regions of the probed field. (c) Transformation formula for mapping the detection view onto the lab coordinates. Demonstrated is the N2 S(28)-transition having an elliptic shape in the detection view and being stretched to a circle in the lab coordinates.

figures show how the molecular population of these levels is distributed in space, as a result of being differently thermally populated by the progress and energy release of the combustion reaction. In the unreacted region, located in the inner part of the flame cone, the temperature is modest, and a substantial signal intensity is observed in this region of the low-energy S(7)-transition image, while leaving a “hollow” looking structure in the image of the high-energy S(28)-transition. The distributed signal from the N2 S(28)-transition borders the signal from the N2 S(7)-transition and can be used as a reasonable estimate for marking the instantaneous position of the flame front. The temperature was evaluated fitting the extracted spectra to a theoretical library of precalculated spectra for the prevailing conditions, and the results are shown in Figure 2c. The derived temperature map consists of expected isotherms ranging from 450 K up to typical hydrocarbon flame temperatures of about 2000 K. In Figure 2d, three examples of extracted spectra with different evaluated temperatures are indicated with their originating spatial location. Figure 3a displays a photograph of the flame, and the planar interrogated fields are marked with white boxes. These fields are transverse sections of the flame orthogonal to the flow, analyzed in Figures 2 and 3b and c, respectively. The capacity for multiplexed spectroscopy, with chemically specific imaging for species N2 and O2, is demonstrated in Figure 3b. At the tip of the flame, O2 is confined to the unburned region and correlates well spatially with the low-energy N2 S(5)-transition. The substantial intensity on the high N2 S(28)-transition in

dimensional coherent Raman signal generation plane is formed through a thin sheet (focused in one direction only using a cylindrical lens) of the pump/Stokes excitation pulse, intersected with an unfocused collimated probe pulse, as depicted in Figure 1. In detecting the two-dimensional signal, the information needs to be resolved both spatially and spectrally, incorporating a spectrometer capable of space−frequency division. For this purpose, we built a novel device,41 as illustrated in Figure 1. The design is conceptually similar to the tomographic hyperspectral imaging spectrometer,42 with the exception that here we are probing narrow, well-isolated spectral lines distributed in a more confined spectral range (0−350 cm−1), which substantially simplifies the data extraction. For instance, when probing rotational N2 S-branch transitions (from rotational quantum number J to J + 2), the adjacent spectral lines are separated by ∼8 cm−1 and are less than 0.1 cm−1/atm in spectral width. Further details of the spectrometer are available in the Supporting Information. Three examples of planar fields, interrogated in a slightly rich premixed laminar CH4/air flame, are depicted in Figure 2a. Here, 25 individual S-branch transitions from N2 and 14 from O2 are detected. Many of the J-specific transitions are fully isolated on the CCD, allowing for simple pixel-to-pixel extraction of spectra mapping of a specific spatial location in the probed field. In Figure 2b, two separate N2 transitions, the S(7) and the S(28), belonging to a low and to a high initial rotational quantum number (J = 7 and 28) are depicted. The 1245

dx.doi.org/10.1021/jz500384y | J. Phys. Chem. Lett. 2014, 5, 1243−1248

The Journal of Physical Chemistry Letters

Letter

matching the acquisition rate of the CCD. In order to increase the probe pulse irradiance and, accordingly, the 2D-CARS signal level, demagnification of the probe beam while still remaining below the damage threshold of the optics is required. Current work in our laboratory is addressing this issue with modifications to the 2D spectrometer. In conclusion, we have reported a major advancement for coherent Raman spectroscopy, with the first instantaneous twodimensional mapping of chemical species and rovibrational distribution in a flame. The achievement was enabled by two recent advances by the current authors. The first is the development of two-beam CARS, a generic non-phase-matched scheme for simplified signal generation, and a method for generating a spatially correlated 2D-CARS signal. The second deals with a spectrally resolved detection of a planar signal, accomplished with a novel spectrometer design for simultaneous planar imaging and multiplex spectroscopy provided within a single laser pulse. The present work demonstrates an important capability in instantaneous mapping of spatially distributed temperatures and species in combustion, with opportunities to leverage cutting-edge diagnostics for many fast dynamical gas-phase systems. We see an interesting potential in performing joint studies between this and other techniques, for example, that of simultaneous thermal- and flow-field measurements using combined planar CARS and particle imaging velocimetry (PIV). By utilizing ultra-broad-band two-beam CARS,35 extension of 2D-CARS to all Raman-active species may become possible. With continued progress, we strongly believe that the capacity will add significantly to the information provided for more rapid development of future energy systems that are more environmentally friendly than those currently employed.

Figure 2b indicates elevated temperature and extends through most of the detection view because the interrogated plane is, for the most part, intersecting the product gases of the reacted region immediately following the flame front. In Figure 3c, transformation of coordinates to physical space is shown for the N2 S(28)-transition, mapping the detection view onto the lab view. The two-dimensional CARS plane is aligned orthogonally to the burner nozzle, which is a tilted plane with respect to the face of the CCD and dependent on the two-beam crossing angle (see the Supporting Information for more detail). The measurements were optimized for flame thermometry and detecting N2 and O2 with small interference. The size of the mask filtering the probe beam was selected to allow for many N2 and O2 lines to be fully isolated, yielding a total of approximately 2100 spatially correlated spectra (in a 74 × 28 pixel frame) to be collected within a single laser shot. This can be improved with larger dispersion, obtained by increasing the focal length of the single effective lens in the spectrometer and implementing a large-format CCD for the detection. To ensure robust signal-to-noise for flame thermometry, the temperature map in Figure 2c was acquired by analyzing 100 accumulated shots, which is a valid procedure in a laminar flame. A cutoff is employed at a signal-to-noise threshold where the fits within the detected spectral window did not converge reliably, which corresponded to a maximum temperature of ∼2200 K, and areas of the 2D-CARS image where the signal counts fell below this cutoff are colored black in all figures of this Letter. For turbulent or transient combustion, only single laser shot analysis may be performed. Information regarding the precision, accuracy, and spatial resolution of the thermometry measurements is discussed in the Supporting Information. Two derived temperature maps are displayed in Figure 4, which are based on the analysis from a single shot (a) and from

■

EXPERIMENTAL METHODS The experiment was performed with time-synchronized femtosecond (fs) and picosecond (ps) laser systems configured in a hybrid fs pump ps probe setup. The fs laser system consisted of an oscillator (KM Laboratories, Halcyon) synced to an external 100 MHz radio frequency (RF) source acting as a master clock. The oscillator pumped a fs regenerative amplifier (KM Laboratories, Wyvern 1000), producing output pulses of 45 fs full width at half-maximum (fwhm) centered at a wavelength of ∼800 nm and operated at about 3 mJ/pulse with a rate of 1 kHz. Pulse characteristics were recorded with a Grenouille autocorrelator (Swamp Optics) to ensure a neartransform-limited pulse at the experiment. The ps laser system consisted of a 20 Hz regenerative amplified mode-locked Nd:YAG laser, with a seed laser phase-locked to the same external 100 MHz RF source, allowing for precise electronic timing between the fs and ps pulses at the experiment with subps jitter. The output of the frequency-doubled Nd:YAG at 532 nm was ∼40 mJ/pulse with a pulse duration of approximately 90 ps fwhm. The complete detection scheme consists of two carefully matched object-to-image planes. A mask is introduced to shape the probe beam into a distinct spatial structure, which is an important feature to assign a coordinate system to the probed field. The position of the mask is relay-imaged to the signal generation plane with demagnification (×2.5) to increase the irradiance, and from here, the spectrometer relay images the signal ∼1:1 onto the CCD. The spectrometer consists of a single effective lens, propagating the signal through a 3600 l/mm diffraction grating, where the dispersed light is collected

Figure 4. Temperature mapping acquired with a single laser shot versus 100 accumulated shots. (a) Two-dimensional CARS thermometry analyzed from a single laser shot, imaging a view of 2.4 mm × 9.1 mm in a laminar premixed CH4/air flame. (b) The same condition as that presented in (a), but instead, the analysis is based on 100 accumulated shots. The result presented is the same as that in Figure 2c, except that the detection view has been transformed into physical space coordinates.

100 accumulated shots (b). The result presented in panel (b) is identical to the temperature map in Figure 2c, except that the detection view has been transformed into physical space coordinates. The single-laser-pulse 2D-CARS data in Figure 4a contain more pixels where the signal-to-noise ratio fell below our cutoff for reliable fit convergence than that in the average case. A movie of 50 raw single-shot spectra is presented in the Supporting Information, with an employed frame rate of 5 Hz, 1246

dx.doi.org/10.1021/jz500384y | J. Phys. Chem. Lett. 2014, 5, 1243−1248

The Journal of Physical Chemistry Letters

Letter

(3) Pitsch, H. Large-Eddy Simulation of Turbulent Combustion. In Annual Review of Fluid Mechanics; Annual Reviews: Palo Alto, CA, 2006; Vol. 38, pp 453−482. (4) Regnier, P. R.; Taran, J. P. E. Possibility of Mesaureing Gas Concentrations by Stimulated Anti-Stokes Scattering. Appl. Phys. Lett. 1973, 23, 240−242. (5) Lucht, R. P. 3-Laser Coherent Anti-Stokes Raman-Scattering Measurements of 2 Species. Opt. Lett. 1987, 12, 78−80. (6) Vestin, F.; Bengtsson, P. E. Rotational CARS for Simultaneous Measurements of Temperature and Concentrations of N2, O2, Co, and CO2 Demonstrated in a CO/Air Diffusion Flame. Proc. Combust. Inst. 2009, 32, 847−854. (7) Seeger, T.; Leipertz, A. Experimental Comparison of Single-Shot Broadband Vibrational and Dual-Broadband Pure Rotational Coherent Anti-Stokes Raman Scattering in Hot Air. Appl. Opt. 1996, 35, 2665− 2671. (8) Druet, S. a. J.; Taran, J. P. E. CARS Spectroscopy. Prog. Quantum Electron. 1981, 7, 1−72. (9) Eckbreth, A. C. Laser Diagnostics for Combustion Temperature and Species; Gordon and Breach Publishers: Amsterdam, The Netherlands, 1996. (10) Roy, S.; Gord, J. R.; Patnaik, A. K. Recent Advances in Coherent Anti-Stokes Raman Scattering Spectroscopy: Fundamental Developments and Applications in Reacting Flows. Prog. Energy Combust. Sci. 2010, 36, 280−306. (11) Grisch, F.; Attal-Tretout, B.; Bresson, A.; Bouchardy, P.; Katta, V. R.; Roquemore, W. M. Investigation of a Dynamic Diffusion Flame of H2 in Air with Laser Diagnostics and Numerical Modeling. Combust. Flame 2004, 139, 28−38. (12) Kearney, S. P.; Frederickson, K.; Grasser, T. W. Dual-Pump Coherent Anti-Stokes Raman Scattering Thermometry in a Sooting Turbulent Pool Fire. Proc. Combust. Inst. 2009, 32, 871−878. (13) Bohlin, A.; Nordström, E.; Carlsson, H.; Bai, X.-S.; Bengtsson, P.-E. Pure Rotational Cars Measurements of Temperature and Relative O2-Concentration in a Low Swirl Turbulent Premixed Flame. Proc. Combust. Inst. 2013, 34, 3629−3636. (14) Magnotti, G.; Cutler, A. D.; Danehy, P. M. Development of a Dual-Pump Coherent Anti-Stokes Raman Spectroscopy System for Measurements in Supersonic Combustion. Appl. Opt. 2013, 52, 4779− 4791. (15) Duncan, M. D.; Reintjes, J.; Manuccia, T. J. Scanning Coherent Anti-Stokes Raman Microscope. Opt. Lett. 1982, 7, 350−352. (16) Zumbusch, A.; Holtom, G. R.; Xie, X. S. Three-Dimensional Vibrational Imaging by Coherent Anti-Stokes Raman Scattering. Phys. Rev. Lett. 1999, 82, 4142−4145. (17) Pealat, M.; Taran, J. P. E.; Taillet, J.; Bacal, M.; Bruneteau, A. M. Measurement of Vibrational Populations in Low-Pressure Hydrogen Plasma by Coherent Anti-Stokes Raman Scattering. J. Appl. Phys. 1981, 52, 2687−2691. (18) Yin, Z. Y.; Montello, A.; Carter, C. D.; Lempert, W. R.; Adamovich, I. V. Measurements of Temperature and Hydroxyl Radical Generation/Decay in Lean Fuel−Air Mixtures Excited by a Repetitively Pulsed Nanosecond Discharge. Combust. Flame 2013, 160, 1594−1608. (19) Konorov, S. O.; Mitrokhin, V. P.; Smirnova, I. V.; Fedotov, A. B.; Sidorov-Biryukov, D. A.; Zheltikov, A. M. Gas- and CondensedPhase Sensing by Coherent Anti-Stokes Raman Scattering in a Mesoporous Silica Aerogel Host. Chem. Phys. Lett. 2004, 394, 1−4. (20) Bremer, M. T.; Wrzesinski, P. J.; Butcher, N.; Lozovoy, V. V.; Dantus, M. Highly Selective Standoff Detection and Imaging of Trace Chemicals in a Complex Background Using Single-Beam Coherent Anti-Stokes Raman Scattering. Appl. Phys. Lett. 2011, 99, 101109. (21) Eckbreth, A. C.; Anderson, T. J. Simultaneous Rotational Coherent Anti-Stokes Raman-Spectroscopy and Coherent Stokes Raman-Spectroscopy with Arbitrary Pump Stokes Spectral Separation. Opt. Lett. 1986, 11, 496−498. (22) Alden, M.; Bengtsson, P. E.; Edner, H. Rotational CARS Generation through a Multiple 4-Color Interaction. Appl. Opt. 1986, 25, 4493−4500.

at Littrow’s angle. The individual transitions are reflected with a split mirror folding them onto separate horizontal rows of the CCD detection window. For presentation, the spectral images are processed with histogram equalization to stretch the global contrast in the data displayed; however, the temperature contour is calculated based on raw data. As can be seen in Figure 3a, the probed region spans both unreacted and reacted mixtures, and there is a large dynamic range in signal intensities between low- and hottemperature zones. A time delay of 160 ps between the pump/ Stokes and the probe pulses was employed to fully suppress the nonresonant scattered signal and to equalize the intensity between low- and hot-temperature signals as the colder molecules undergo much faster collisional dephasing during the probe delay. The spectral libraries used in the fit were generated using a time domain CARS code for accounting for collisional effects during the probe delay. The two-beam CARS approach ensures an automatic overlap of the pump/Stokes fields both temporally and spatially, acting as a coherent spike in terms of autocorrelation, and is a clear advantage for the constructive signal build up. The implementation introduces a slight deviation from the perfect phase-matching condition, however, and the dependence is as an increasing function of the Raman shift as well as the twobeam crossing angle.33 In this aspect, and to obtain longer interaction length, which is beneficial for the signal strength, it is realistic to employ somewhat shallow crossing angles typical of most CARS experiments. In the current setup, the laser beams were set up in a crossed two-beam configuration with a small enough incident crossing angle (∼6°) to achieve a nearperfect phase-matching condition for all of the involved transitions (