Technique developments and performance analysis of chirped-probe-pulse femtosecond coherent anti-Stokes Raman scattering combustion thermometry.

This work characterizes the state of the art in the analysis of high-repetition-rate, ultrafast combustion thermometry using chirped-probe-pulse femtosecond coherent anti-Stokes Raman scattering (CPP fs-CARS). Several key aspects of the CARS spectroscopy system are described, including: (1) the ultrafast laser source, (2) use of the frequency-doubled idler versus signal from the optical parametric amplifier, (3) the geometry constraints for phase matching, and (4) spectral fitting for single-shot temperature measurements. A frequency-dependent instrument response function (IRF) for the detection system was modeled as a variable-width Gaussian and implemented through a frequency convolution of synthetic spectra. Proper accounting of the IRF increased spectral fitting performance in the high-frequency region where signal oscillations are weaker and narrower. Aggregated data from 25 system performance assessments taken over four months yielded accuracy and precision of 2.7% and ±3.5% for flame temperatures, and 9.9% and ±6.1% at room temperature, using the commonly reported method. A new processing technique, based on the statistical method of maximum likelihood, was implemented for turbulent flames where strong fluctuations in expected temperatures necessitate use of multiple temperature calibrations. Results from multiple sets of laser parameters are combined to generate an error-weighted temperature from the top-performing calibrations. A testing procedure was designed to characterize system performance when the range of expected temperatures is unknown, simulating the random temperature field of a highly turbulent flame. Accuracy error of the CPP fs-CARS system increased in this more-stressing test at all temperatures, but precision was significantly affected only at room temperature. System stability is characterized, and the contribution from shot-to-shot laser fluctuations on measurement precision is quantified. Finally, the near-adiabatic and steady assumptions for the Hencken burner calibration flame are examined in an axial scan; significant deviations from ideal behavior were observed only at heights of more than four diameters above the burner surface.