A NOVEL FOURIER TRANSFORM TECHNIQUE TO SEPARATE FINE AND DIFFUSE FEATURES IN AN INFRARED SPECTRUM

This research was made possible by a grant from the National Science Foundation (NSF ATM93-03058).Author Institution: Indiana University, Bloomington, IN 47405.One of the major obstacles facing atmospheric scientists who use IR to characterize condensed phase materials is that there are several IR active gases $(CO, CO_{2}, H_{2}O)$ which can overlap peaks of interest. In a typical IR spectrum, condensed features are often as much as $200 cm^{-1}$ wide while the overlapping gas features are resolution limited. Subtraction and/or ratioing against appropriate reference spectra are often unable to separate the two types of features because minor temperature fluctuations $(\pm 0.1 K at 300 K)$ are often sufficient to significantly change the relative populations (and hence absorbance) of the high rotational lines while leaving the low rotational lines virtually unchanged. In addition, spectral subtraction is notorious for altering the baseline of a spectrum, which would greatly change the shape of a broad feature. The absorbance at each frequency, $A^(\nu)$, is a function of resolution, and this dependence can be expanded as a power series: $A(\delta)^(\nu)={a_{0}}^(\nu)+{a_{1}}^(\nu)\delta+{a_{2}}^(\nu)\delta^{2}+\ldots$ By transforming sample and background high resolution interferograms to various lower resolutions with the addition of an appropriate amount of zero-filling, it is possible to use a curve-fitting routine to determine the values of the constants for each frequency. The $a_{0}$ constants contain the absorbance spectrum for all features which are much broader than the lowest resolution used, while all of the resolution-limited features are contained in the remaining resolution-dependent constants