Two conventional minimization criteria such as entropy minimization and path minimization and newly suggested exponential minimization criterion have been tested in extracting each pure component spectrum from two binary mixture infrared absorption spectra via spectrum subtraction. Entropy minimization criterion is based on the fact that the entropy reaches minimum value when a correct scale factor is obtained during subtraction procedure, and path minimization criterion is based on the fact that the path reaches minimum value when a correct scale factor is obtained. Exponential minimization criterion is based on the fact that the minimum value of the total summation of absolute valuta of residuals is obtained for a correct scale factor when two types of exponential curves are fitted to the difference spectrum in a specific range obtained during interactive subtraction procedure. In the case of very closely overlapped characteristic peaks present in spectra, satisfactory results have not been obtained in extracting each pure spectrum on the basis of conventional minimization criteria. Exponential minimization chiterion, however, newly suggested in this study was proved to be a good objective candidate on peak subtraction, even though very closely overlapped characteristic peaks exist in spectra.
