Temperature estimation of carbon-dioxide by Fourier transform infrared spectrometry

An infrared spectrometric method for remote sensing of hot gases is described in this dissertation. Fourier transform infrared (FT-IR) spectrometers are potentially very useful for monitoring the gases emitted from power plant smokestacks. The emission spectrum of the smokestack effluent gas can be collected using a telescope and an FT-IR spectrometer. A spectroscopic means of estimating the temperature of the stack gases is needed to measure the concentration of pollutants from such a spectrum. No field studies were conducted, but the temperature sensing method was demonstrated for the CO{dollar}sb2{dollar} laser bands near 10-{dollar}mu{dollar}m.; Absorption spectra of pure CO{dollar}sb2{dollar} gas were obtained using a tunable diode laser and a high resolution FT-IR spectrometer. The CO{dollar}sb2{dollar} was maintained at elevated temperature (100 to 200{dollar}spcirc{dollar}C) and constant pressure in a one meter gas sample cell. The extremely high resolution of these spectrometers permitted measurements to be made on the collision broadening parameters for this gas, in the absence of significant broadening due to the instrument. Measurements of the broadening parameters, and the study of the variation in linewidth with temperature, showed that the temperature broadening exponent is larger than 0.51, as reported by one research group, and was found to lie in the range of 0.6 to 0.8 depending on the line studied.; This information was used to calculate the true line intensities of CO{dollar}sb2{dollar} spectra measured at moderate resolution (0.2 to 0.4 cm{dollar}sp{lcub}-1{rcub}{dollar}). The correction for the response of the spectrometer and the broadening effect which it has on narrow lines was calculated quantitatively. For absorption spectra measured with this resolution, extremely good agreement was found between the measured temperature and the temperature estimated spectroscopically. For emission spectra, the agreement was much poorer, with a standard deviation of the estimate error of 22{dollar}spcirc{dollar}C, and an average error of 25{dollar}spcirc{dollar}C in excess of the actual temperature. The signal-to-noise ratio of the emission spectra was lower than that of the absorption spectra, and so the effect of noise on the temperature determination method was studied. These results have important applications in several problems of remote sensing of the atmosphere.