Download files
Abstract
In this thesis, a digital signal processing (DSP) strategy for the Fourier transform infrared (FT-IR) spectrometer called Super Clip Mathematics is presented. This work is motivated by the issue of collecting a reliable FT-IR background spectrum while collecting data in a dynamic environment. Super Clip Mathematics addresses this problem by extracting data contained within a single interferogram in lieu of collecting a background spectrum. This regime utilizes the pre-established technique of super clip apodization (SCA) and defines a new DSP strategy called complementary super clip apodization (CSCA), where both methods are generally defined by the umbrella term of Super Clip Mathematics. SCA is a DSP strategy that calculates a background spectrum by isolating and Fourier transforming the central burst of an interferogram. A secondary strategy is described herein as CSCA, where all of the interferogram except the center-burst is Fourier transformed for the calculation of an absorbance spectrum. Upon isolation of these regions of the time-domain waveform, each method may subsequently calculate a quantitative absorbance spectrum.
Interferogram manipulation in this fashion has been previously discounted by its community because there is a lack of systematic investigation into its potential. The outcomes of this work suggest that the limitation of Super Clip Mathematics is not the signal processing strategy itself, but the way in which it is implemented. An implementation algorithm that systematically optimizes the number of points truncated or included during Super Clip Mathematics analysis is presented. With this protocol, carbon dioxide, nitromethane, ethanol and acetone interferograms are analyzed to calculate quantitative absorbance spectra. This is significant because the latter three substances have spectral widths that are considered too wide by convention, yet their successful analysis in univariate and multivariate studies is shown. For comparison, the linear regression models built with Super Clip Mathematics spectra are compared to those built with spectra calculated by taking the logarithmic ratio of a background and sample spectra; and the results indicate that there is almost no statistical difference between them. The issue of collecting a characteristic background may be circumvented by utilizing data already present in the sample interferogram with Super Clip Mathematics.