Two-dimensional gas-phase coherent anti-Stokes Raman scattering (2D-CARS) thermometry is demonstrated at 1 kHz in a heated jet. A hybrid femtosecond/picosecond CARS configuration is used in a two-beam phase-matching arrangement with a 100-femtosecond pump/Stokes pulse and a 107-picosecond probe pulse. The femtosecond pulse is generated using a mode-locked oscillator and regenerative amplifier that is synchronized to a separate picosecond oscillator and burst-mode amplifier. The CARS signal is spectrally dispersed in a custom imaging spectrometer and detected using a high-speed camera with image intensifier. 1-kHz, single-shot planar measurements at room temperature exhibit error of 2.6% and shot-to-shot variations of 2.6%. The spatial variation in measured temperature is 9.4%. 2D-CARS temperature measurements are demonstrated in a heated O2 jet to capture the spatiotemporal evolution of the temperature field.
© 2016 Optical Society of America
The detailed, fundamental study of turbulent combustion requires accurate, time-dependent measurements of temperature and species concentrations in addition to their spatial distributions . Accurate measurements of temperature, which plays a leading role in chemical kinetics, are of particular interest because they are used to validate combustion models and identify regions of heat release in combusting flows. While many temperature-measurement techniques have been applied successfully in combusting flows, coherent anti-Stokes Raman scattering (CARS) thermometry has shown superior accuracy and precision over an extremely wide range of temperatures and experimental conditions [1,2]. The fidelity of ultrafast CARS temperature measurements results from background-free signal generation, insensitivity or reduced sensitivity to pressure, large signal-collection efficiency, and calibration-free analysis . Even so, CARS thermometry has been limited to point (zero-dimensional) or line (one-dimensional) measurements because of the high peak powers required for efficient nonlinear signal generation [2,3].
Recently, two-dimensional CARS imaging and spectroscopy were demonstrated at 20 Hz  by combining two novel CARS developments, hybrid femtosecond/picosecond (fs/ps) pulse sequencing [5–7] and simplified two-beam phase matching [8–10]. A single fs laser pulse from a 1-kHz regenerative amplifier comprised both the degenerate pump and Stokes pulses, and a ps laser pulse from a 20-Hz regenerative amplifier was used as the probe [4,9]. Precise synchronization of the fs and ps oscillators ensured accurate CARS measurements while the high-energy ps probe and two-beam phase-matching configuration enabled two-dimensional CARS imaging . Although the fs amplifier produced pulses at 1 kHz, the measurement repetition rate was limited by the ps amplifier (20 Hz) and framing rate of the imaging spectrometer camera (several Hz). Operation of ps regenerative amplifiers with high pulse energy is challenging at high repetition rates, because of large thermal loading on the lasing medium, making extension of 2D-CARS to repetition rates above tens of Hz impractical with current approaches.
Recent advances in burst-mode laser technology have extended burst periods (10–100 ms), increased beam quality (M2 < 2), increased pulse energy (>1 J), and reduced pulse width (<100 ps) for bursts of high energy laser pulses with repetition rates up to 1 MHz . Notably, these improvements have made burst-mode lasers suitable for pumping nonlinear spectroscopic processes, enabling ps CARS point temperature measurements at 100 kHz . This work seeks to extend the repetition rate of 2D-CARS thermometry to 1 kHz by using high-energy, burst-mode amplification of a pulse-width-flexible ps oscillator with precise synchronization to a mode-locked fs oscillator. Even so, most practical combustion devices require measurement repetition rates on the order of 10–100 kHz for fully resolving transient heat release and flame propagation. However, the burst-mode-laser-based methodology demonstrated here could be extended to 10 kHz in a straightforward manner by synchronization to a commercially available 10-kHz fs regenerative amplifier. Measurement repetition rates of 10 kHz are commonly used in subsonic turbulent combustion experiments to track transient fluid–flame interactions. Increasing the repetition rate of 2D-CARS beyond 10 kHz will require advances in ultrafast burst-mode laser technology or high-power, high-speed regenerative amplifier technology.
2. Experimental details
The 2D-CARS experimental configuration in the current work follows the approach set forth by Bohlin and Kliewer . As shown in Fig. 1, the degenerate pump and Stokes pulses were derived from a single ~100-fs laser pulse at 800 nm with pulse energy of 2.6 mJ, generated from a 1-kHz regenerative amplifier (Legend Duo, Coherent Inc.). The 10-mm-diameter pump/Stokes beam was formed into a sheet using a single + 1000-mm cylindrical lens and focused through the center of a round jet with diameter of 2 mm. The probe pulse was generated using a separate burst-mode laser amplifier coupled to a highly flexible pulsed oscillator. Unlike our previous work that used a fixed-frequency, fixed-pulse-width, mode-locked ps oscillator with burst-mode amplification [13,14], the pulse width of the oscillator used in the current work is continuously variable from 60 ps to 1 ns, and the oscillator can be operated at frequencies exceeding 100 MHz. As such, the ps probe pulse was synchronized to the fs oscillator on a shot-to-shot basis and amplified at 1 kHz over a 10 ms period. The resulting ~100 ps pulses were frequency doubled to 532 nm. The collimated, 7-mm-diameter probe beam was passed through a 4 × 4-mm square mask and then reduced to 2 × 2-mm using a 2 × 1 telescope. The positive–positive telescope ( + 500 mm and + 250 mm lenses) was arranged to relay-image the beam profile at the plane of the mask to the center of the round jet. The probe energy at the crossing was 5 mJ/pulse. The pump/Stokes and probe pulses were arranged in a two-beam-phase-matching configuration with a crossing angle of 12 degrees. This configuration provided sufficient phase-matching bandwidth for excitation of all significant O2 S-branch transitions at elevated temperature  and yielded a planar overlap region of 2 mm by 9.6 mm between the pump/Stokes sheet and the probe square profile. The polarization of the probe beam was rotated by 45° relative to the pump/Stokes beam and set using a Glan-laser polarizer, and the time delay between the pump/Stokes and probe interactions was controlled by a high-resolution motorized delay stage in the fs pump/Stokes beam line.
The blue-shifted CARS signal was separated from the probe pulse by polarization filtering using a Glan-laser polarizer with ~105 suppression as an analyzer and detected using a custom imaging spectrometer. In the imaging spectrometer, a + 750-mm lens in combination with a 3600 mm−1 grating at the Littrow angle was used to image the CARS measurement plane to a detector. The backwards-propagating dispersed light from the grating was angled downward for spatial separation using a square mirror. The CARS signal was detected with a high-speed two-stage intensifier (LaVision HS-IRO) coupled to a high-speed complementary metal-oxide semiconductor (CMOS) camera (Photron SA-Z) with resolution of 1024 × 1024 pixels at 1 kHz. The dispersion grating and collimating lens were chosen to generate a dispersion of 0.1 cm−1 per pixel. This is larger than the typical bandwidth of a single O2 rotational transition (<0.1 cm−1) and enables each pixel to be evaluated as an independent sample. The imaging intensifier, however, imparts additional spatial distortion on the system, and a total instrument function of 0.46 cm−1 was measured, corresponding to 4.6 pixels, by imaging a point source of light and measuring the spectral dispersion. The image intensifier was operated at a gain of 35 out of 100, and a linear relationship between the intensified signal and incident light was verified by changing the pulse energy of the probe pulse and recording the generated CARS signal.
Precise synchronization of the pump/Stokes and probe pulses was accomplished using low-jitter triggering of the custom ps oscillator as shown schematically in Fig. 2. The Ti:Sapphire regenerative amplifier system comprises a mode-locked oscillator (Vitara, Coherent Inc.), which generates a continuous 80-MHz train of 100-fs pulses, and a high-power regenerative amplifier (Legend Duo, Coherent Inc.), where the repetition rate is reduced to 1 kHz before amplification. The fixed, 80-MHz repetition rate of the mode-locked fs oscillator was used as the master trigger for all electronics. Approximately 0.1% of the fs oscillator energy was split from the fs oscillator before entering the pulse picker and regenerative amplifier and coupled to a 25-GHz photodiode (1414, Newport). To minimize electrical jitter, the rising edge of the photodiode signal was used to trigger a picosecond pulse generator in the custom ps oscillator at 80 MHz. The ps pulse train was produced by repetitively forming individual pulses from a 1064.4-nm (vacuum), 30-mW continuous-wave narrowband diode laser. The ps pulse slicer is based on a fiber-coupled electro-optic modulator (EOM) with >10 GHz bandwidth and >40 dB extinction ratio. The EOM bias is automatically controlled to minimize the output power when in the “closed” position by splitting a portion of the EOM output into the bias control circuit. The EOM itself is modulated by a ps pulse generator (T240, Highland Technology Inc.) capable of generating pulses with a 60-ps rise/fall time and adjustable pulse duration. An RF amplifier with 12-GHz bandwidth is used to increase the modulated voltage to the required 4V. The 80-MHz, 1064.4-nm pulse train is amplified to a peak power of 10 W using an ytterbium-based fiber amplifier. The output of the fiber amplifier was coupled to an acousto-optic-modulator- based (AOM) pulse picker with 10 ns rise/fall time and >50 dB extinction ratio. The AOM was triggered from the pulse picker in the fs amplifier and a separate pulse generator was used to select the ps pulse with minimum delay between the fs and ps systems. The flexible oscillator, including EOM-based pulse slicer and AOM-based pulse picker, was entirely fiber coupled and required no alignment. The 1-kHz output of the fiber- coupled AOM was amplified further in a series of free-space, flashlamp-pumped, Nd:YAG burst-mode power amplifiers, similar to that reported previously by the authors . The total jitter at the CARS measurement plane was measured to be ± 8 ps using a 20-GHz oscilloscope. This is significantly less than the pulse duration (122 ps FWHM at 1064 nm and 107 ps FWHM at 532 nm) and probe delay (100 ps) and would result in a maximum temperature fit error of ± 14 K under the conditions reported here.
3. Results and discussion
The spectrally dispersed 2D-CARS signal of O2 from a single laser shot at room temperature is shown in Fig. 3. Two unique masks with different geometric features and spatial dimensions are employed to highlight the spatial resolution of the 2D-CARS images. Each square is an exact image of the pump/Stokes and probe overlap plane and represents a single rotational transition with an initial rotational quantum number as indicated above the image. The CARS image was not corrected explicitly for beam profile variations; therefore, the intensity variation within a single square is due to the non-uniformity of the beam profile, not variation in O2 concentration. The vertical dimension is set by the half-height of the mask, 2 mm, while the horizontal dimension represents the overlap region of the pump/Stokes and probe pulses, 9.6 mm. Because of the two-beam phase-matching configuration used in this work, the 9.6-mm overlap region is imaged onto a 2-mm horizontal region and absolute pixel resolution is reduced. Each square O2 transition image is ~75 × 75 pixels so that the absolute pixel resolution is 27 µm in the vertical dimension and 128 µm in the horizontal dimension. Because the spectral dispersion is aligned with the horizontal dimension, making the horizontal resolution a function of both the spatial and spectral instrument functions, the spatial resolution was estimated in the vertical dimension, using the spatial frequency at 20% response, as 79 µm which corresponds to 3 pixels. In the horizontal dimension, the effective resolution is 589 µm (total instrument function of 4.6 pixels) and is a function of both the spatial and spectral instrument functions. If we assume that the total instrument function is a convolution of the spatial and spectral instrument functions and that the spatial instrument function, 3 pixels, is identical in the vertical and horizontal dimensions, the spectral instrument function in the horizontal dimension is estimated to be 3.5 pixels or 0.35 cm−1. In this case, the spectral resolution is the limiting resolution in the horizontal direction, while the spatial resolution is the limiting resolution in the vertical direction.
The O2 spectrum was obtained at a single spatial location by mapping each 2D-CARS image of an individual rotational transition to a common grid and selecting the integrated intensity of each corresponding point, IN(x,y). The mapping function was obtained by replacing the square mask with a 50-µm pinhole to produce a 25-µm point source, after the 2 × 1 telescope, which is less than the optical resolution of the imaging system. The location of each transition in the point-source-derived O2 spectrum was used to map the individual CARS images to a common grid. Once the transition-specific CARS images were mapped to a common 75 × 75 pixel grid, the spectrum at each pixel was generated. An intensity threshold was applied to remove low-intensity, noisy data points from each spectrum. Temperature was only calculated for pixels where more than five transition-specific intensities were above the threshold value. A time- and frequency-dependent hybrid fs/ps CARS model was used to generate a library of stick spectra, representing the integrated intensity of each transition, as a function of temperature at a fixed probe delay of 100 ps . The residual was minimized between the experimental intensity values at each transition and the integrated model values, and the best-fit temperature was extracted from the library. This process was repeated for every point in each single-shot 2D-CARS image to generate the temperature maps shown in Fig. 4. To quantify the accuracy and precision of the 1-kHz 2D-CARS thermometry technique, ten single-shot temperature images were acquired at 1 kHz in pure O2 at a room temperature of 295 K and used to compute mean temperature, shot-to-shot temperature variation, and spatial temperature variation. At 295 K, the mean measured temperature averaged across all ten images and entire spatial domain was 287.2 K corresponding to 2.6% error. This is within the maximum jitter-dependent uncertainty of ± 14 K as described previously. The average of the ten spatial RMS measurements was 27.1 K, corresponding to 9.4% variation, and the average temporal RMS of all points in the 2D-CARS signal was 7.4 K, corresponding to 2.6% variation. The mean error and shot-to-shot RMS values are in good agreement with previous hybrid fs/ps RCARS temperature measurements at a single point at 1 kHz [7,15] and that reported for single-shot 2D-CARS thermometry at 20 Hz . While the beam profile variations observed in Fig. 3 are not evident in the computed temperature maps in Fig. 4, the large spatial RMS value may occur because of increased noise and reduced spatial resolution associated with the high-speed, intensified CMOS detector. Additionally, coherent structures appear in each image, angled at 45°, which can be attributed to interference in the Glan-laser polarizer used for polarization suppression.
A proof-of-principle demonstration of the 1-kHz 2D-CARS thermometry technique was performed in a jet of O2 with diameter, D, of 2 mm issuing into room-temperature air. The jet was operated with room-temperature O2 at 295 K and heated O2 at ~700 K using an inline, electric resistance heater. The gas temperature was measured in the gas line using a type K thermocouple. The mean jet exit velocity was estimated as 104 m/s using the known mass flow rate of the O2. The volumetric flow rate was modulated prior to entering the heater using an oscillating piston in a cylinder driven with a sinusoid at 100 Hz by an external function generator. The 2D-CARS measurement volume was centered 6.4 mm above the jet exit (Y/D = 3.2). Single-shot transition-specific images of the jet under cold (295 K) and hot (~700 K) conditions are given in Fig. 5. The large-scale jet structure is clearly visible in both images, and a systematic shift in population towards higher rotational-quantum-number states is observed. Although the N2 signal is present and overlaps with the O2 images, the N2 signal falls below the noise threshold and is not detectable because the Raman cross section of N2 is weak relative to O2 and the intensifier reduces the dynamic range of the measurement. While an air jet would have allowed simultaneous imaging of both N2 and O2 signal, the overlapping signals can be difficult to interpret, which is beyond the scope of the current demonstration. It should be noted that the images are not normalized for beam-profile variations because the size and pixel resolution of the detector limited the number of individual transitions that could be imaged simultaneously. The zeroth order image, typically used for beam normalization, could be imaged in future work by using a larger detector, reducing the spatial resolving power, or using multiple cameras simultaneously. However, in this work the beam uniformity is high in the horizontal dimension, as shown in Fig. 3, although significant variations are observed in the vertical direction. As such, quantitative O2 concentration cannot be obtained in the vertical dimension, but reasonable estimates may be possible in the horizontal dimension although caution is urged. A single slice through the 2D-CARS images at Y/D = 3.35 is given in Fig. 5 (bottom) for the cold and hot jets. The O2 CARS signal levels are normalized to the peak value at N = 11. For both cases, the jets exhibit a Gaussian profile across the width with full-width at half-maximum (FWHM) of 2.4 mm for the cold jet and 2.7 mm for the hot jet. Quantitative comparisons would require beam profile normalization; however, the measured jet widths are reasonable given the downstream position, and the variation falls within the spatial resolution of the imaging technique in the horizontal dimension (0.59 mm).
The jet temperature was computed for each 1-kHz single-shot image using the process described previously and shown in Fig. 6. The spatial variations in the size and shape of the jet are a function of variation in the O2 concentration due to large-scale entrainment of room air coupled to the jet forcing at 100 Hz and the threshold applied while evaluating temperature, which removes pixels with insufficient information. While the average temperature along the centerline, 589 K, is reasonable given the nominal temperature in the gas line, 700 K, a variation in temperature is observed along the jet width in each image. The cause for this variation can be seen in the hot-jet image in Fig. 5. At N = 7, the “hot” O2 signal is slightly weighted to the left side of the jet, while at N = 15 the O2 signal appears to be weighted to the right side of the jet. Because preferential population of low N states corresponds to low temperature, and preferential population of high N states corresponds to high temperature, the apparent shift in O2 concentration causes the variation in temperature observed in Fig. 6. Although this systematic temperature variation does not appear to be physical in nature, it may result from non-ideal imaging in the spectrometer. However, there does not appear to be any systematic bias observed at room temperature in the O2 images presented in Figs. 3 and 5 or the computed temperature in Fig. 4.
Time series of temperature along the centerline are given for axial distances of Y/D = 2.95, 3.08, 3.22, and 3.35 in Fig. 6 (bottom). Missing points correspond to regions where either the jet scalar (O2) was below the threshold value or the noise was too large for temperature evaluation. Qualitatively, the temperature decreases with increased axial displacement as expected and decreases with time, although noise and systematic concentration bias are present as described above. While only a proof-of-concept demonstration, this indicates that high-speed 2D-CARS can be applied successfully to obtain spatially and temporally evolving measurements of temperature and concentration in a turbulent forced jet. Further detailed quantitative analyses will require technical solutions for minimizing detector noise, maximizing detector dynamic range and resolution, and increasing the imaging region. Increasing laser energy may eliminate the need for an image intensifier and, therefore, would significantly reduce the noise and enhance the dynamic range and spatial resolution. However, high-speed CMOS cameras still suffer from reduced dynamic range (12 bit), increased readout noise, and low resolution (~1 megapixel) as compared to traditional CCD cameras used for spectroscopy.
Single-shot two-dimensional gas-phase thermometry was demonstrated at 1 kHz using fs/ps CARS. The measurement repetition rate was increased 50 × over previous state-of-the-art measurements by utilizing a high-speed burst-mode laser as the probe source, seeded by a highly flexible ps oscillator locked to the fs pump/Stokes source. The custom ps oscillator enabled high precision synchronization with a fs mode-locked oscillator. Measured synchronization jitter was less than 8 ps for arbitrary frequencies and pulse durations. The spatial resolution of the single-shot 2D-CARS images, acquired with a high-speed intensified CMOS camera, was estimated as 79 µm in the non-dispersive dimension and 588 µm in the dispersive dimension. The resulting temperature measurements at room temperature show an error of 2.6% and shot-to-shot RMS of 2.6%. The spatial RMS under uniform temperature conditions was 9.4% and attributed to increased noise from the high-speed detector. Finally, the temporal evolution of a forced jet was captured at 1 kHz over 10 ms using the high-speed 2D-CARS thermometry system.
Air Force Research Laboratory (Contract No. FA8650-15-D-2518); Air Force Office of Scientific Research (LRIR: 15RQCOR202, Dr. Enrique Parra, Program Manager; LRIR: 14RQ06COR, Dr. Chiping Li, Program Manager).
The authors acknowledge technical assistance from Dr. Paul Hsu and Mr. Ethan Legge and technical discussions with Dr. Hans Stauffer.
References and links
1. A. C. Eckbreth, Laser Diagnostics for Combustion Temperature and Species, Second Edition (Gordon and Breach Publishers, 1996).
2. S. Roy, J. R. Gord, and A. K. Patnaik, “Recent advances in coherent anti-Stokes Raman scattering spectroscopy: Fundamental developments and applications in reacting flows,” Pror. Energy Combust. Sci. 36(2), 280–306 (2010). [CrossRef]
3. S. Roy, W. D. Kulatilaka, D. R. Richardson, R. P. Lucht, and J. R. Gord, “Gas-phase single-shot thermometry at 1 kHz using femtosecond coherent anti-Stokes Raman scattering (fs-CARS) spectroscopy,” Opt. Lett. 34, 3857–3859 (2009). [CrossRef] [PubMed]
4. A. Bohlin and C. J. Kliewer, “Communication: Two-dimensional gas-phase coherent anti-Stokes Raman spectroscopy (2D-CARS): Simultaneous planar imaging and multiplex spectroscopy in a single laser shot,” J. Chem. Phys. 138(22), 221101 (2013). [CrossRef] [PubMed]
5. B. D. Prince, A. Chakraborty, B. M. Prince, and H. U. Stauffer, “Development of simultaneous frequency- and time-resolved coherent anti-Stokes Raman scattering for ultrafast detection of molecular Raman spectra,” J. Chem. Phys. 125(4), 044502 (2006). [CrossRef] [PubMed]
6. J. D. Miller, M. N. Slipchenko, T. R. Meyer, H. U. Stauffer, and J. R. Gord, “Hybrid femtosecond/picosecond coherent anti-Stokes Raman scattering for high-speed gas-phase thermometry,” Opt. Lett. 35(14), 2430–2432 (2010). [CrossRef] [PubMed]
7. J. D. Miller, S. Roy, M. N. Slipchenko, J. R. Gord, and T. R. Meyer, “Single-shot gas-phase thermometry using pure-rotational hybrid femtosecond/picosecond coherent anti-Stokes Raman scattering,” Opt. Express 19(16), 15627–15640 (2011). [CrossRef] [PubMed]
8. M. N. Slipchenko, B. D. Prince, S. C. Ducatman, and H. U. Stauffer, “Development of a simultaneously frequency- and time-resolved Raman-induced Kerr effect probe,” J. Phys. Chem. A 113(1), 135–140 (2009). [CrossRef] [PubMed]
11. N. Jiang, M. Webster, W. R. Lempert, J. D. Miller, T. R. Meyer, C. B. Ivey, and P. M. Danehy, “MHz-rate nitric oxide planar laser-induced fluorescence imaging in a Mach 10 hypersonic wind tunnel,” Appl. Opt. 50(4), A20–A28 (2011). [CrossRef] [PubMed]
12. M. N. Slipchenko, J. D. Miller, S. Roy, T. R. Meyer, J. G. Mance, and J. R. Gord, “100 kHz, 100 ms, 400 J burst-mode laser with dual-wavelength diode-pumped amplifiers,” Opt. Lett. 39(16), 4735–4738 (2014). [CrossRef] [PubMed]
13. S. Roy, J. D. Miller, M. N. Slipchenko, P. S. Hsu, J. G. Mance, T. R. Meyer, and J. R. Gord, “100-ps-pulse-duration, 100-J burst-mode laser for kHz-MHz flow diagnostics,” Opt. Lett. 39(22), 6462–6465 (2014). [CrossRef] [PubMed]
14. S. Roy, P. S. Hsu, N. Jiang, M. N. Slipchenko, and J. R. Gord, “100-kHz-rate gas-phase thermometry using 100-ps pulses from a burst-mode laser,” Opt. Lett. 40(21), 5125–5128 (2015). [CrossRef] [PubMed]
15. S. P. Kearney, “Hybrid fs/ps rotational CARS temperature and oxygen measurements in the product gases of canonical flat flames,” Combust. Flame 162(5), 1748–1758 (2015). [CrossRef]