1. The molecular iodine cells used to monitor the laser frequency and to filter the Rayleigh signal are of the starved-cell design described by Elliott and Beutner [17]. Each cell is a 76-mm diameter glass cylinder with optically flat windows on each end. In the starved-cell arrangement, each cell contains essentially pure I2vapor that is superheated to 100oCbykaptonresistance heaters on the cell sidewalls. Superheating of the cell walls ensures that none of the vapor inside the cell condenses back to the solid phase so that a constant I2number density is kept in the cell. The reference cell is 127 mm in length and permanently sealed for enhanced longtime stability with a nominal I2saturation temperature of 35oC. The molecular-filter cell is 254 mm in length and has valved ports to a vacuum line and a temperature-controlled I2crystal reservoir so that the amount of I2inthecellcanbevariedifdesired.Forthe experiments reported in this paper, the I2saturation temperature for the molecular filter is 45oC. The molecular filter has a longer length and higher I2concentration relative to the reference I2cellsothatthe maximum level of background rejection is provided in the Rayleigh images. Calibration data for both the reference and molecular filter cells are shown in Fig. 4. The long-time stability of the molecular-filter cell, whichhasnotbeenpermanentlysealed,isnoteworthy. DataProcessing
2. Time-meanandsingle-laser-pulse FRS temperature data from the stable calibration jet are shown in Fig. 5. The in-plane spatial resolution in these temperature images is 160 µm × 160 µm, with an out-of-plane resolution dictated by the local laser-sheet thickness of approximately 200-500 µm. The time-mean results in Fig. 5a were obtained by averaging 200 single-pulse temperature images, and this mean field shows the extent of the uniform-temperature jet core. The singlepulse temperature field in Fig. 5b exhibits more noise (primarily MCP and shot noise) than the averaged temperature field and shows the impact of instantaneous jet structure caused by a sinuous-wave instability that is clearly evident at z= 50-60 mm or 3.3 to 4.0 jet diameters.
3. Using the above-mentioned procedure for estimation of local product composition, the FRS results for φ< 1.3 deviated by no more than 50 K from the adiabatic flame temperature. For comparison, the average of several sets of CARS measurements performed in the same burner are also shown in Fig. 8a. The CARS temperatures are nominally 25-50 K less than the adiabatic flame temperature for all φ, which is generally a result of low-levels of heat loss from the flame. FRS temperatures in Fig. 8a for φ< 1.3 are generally within 50 K of the CARS results as well. For φ > 1.3, the FRS temperatures are nominally 50 K higher than the adiabatic flame temperature and 75-100 K higher than the CARS measurements for all three FRS data sets. The reason for this apparent hightemperature bias in the FRS results at high φ is not understood at thistime.
4. The large variations in local scattering cross section in this diffusion flame are clearly much less conducive to FRS thermometry than the smaller variation in the premixed combustion case discussed above. To make the interpretation of the FRS data feasible, additional quantitative information on the local chemical composition is required. To provide this additional information, we have imaged the methane vibrational Raman scattering signal from near 15,847 cm-1(629 nm) and 15,780 cm-1(633 nm) to obtain the local fuel mole fraction. Raman imaging is a challenging experimental approach because Raman scattering cross sections are generally 2-3 orders of magnitude less than Rayleigh cross sections, resulting in a prohibitively weak Raman signal. Similar combined linear Rayleigh/fuel Raman measurements have been previously performed in turbulent nonpremixed flames by a collaboration of researchers from Yale and Sydney [19-22], using a multiple-laserpass intracavity technique for single-pulse Raman measurements. Since the slot flame of the present investigation was laminar and periodic, averaging the weak Raman signal over a large ensemble of laser shots for improved SNR was permissible and the intracavity technique was not required here. The large laser pulse energies (1100 mJ/pulse) and high (~60%) ICCD photocathode quantum efficiencies near 633 nm further aided in making Raman imaging feasible here.