A Window Fourier-Domain Infrared Search System

Abstract

A search system is presented that utilizes Fourier-domain infrared data. This system removes or nulls features from infrared absorbance spectra that are found to be dominant and similar in the spectra of interest. This method enhances spectral differences. An example of one such region that is commonly nulled is the C-H stretch region of the spectra. These bands occur in organic compounds and, while generally being quite strong, contribute little to differentiation between spectra. More than one region may be nulled with this system, and the regions are chosen by visual inspection of the spectra of interest. The nulled spectra are then transformed into the Fourier domain, where data in the region that is 55–155 points displaced from the centerburst are extracted and normalized. Finally, a complete library search is performed on the normalized Fourier-domain data. The actual matching algorithm uses a simple dot product calculation, and excellent search results are obtained. This method can be used to distinguish similar compounds that are not easily distinguished by other infrared search methods. In the most difficult cases, a minimum of two-orders-of-magnitude improvement in the difference in dot products is observed when the results of full bandpass spectra are compared with the results of spectra that had one or more regions nulled.