A general-purpose method for eliminating one or more interferences from spectra in a completely objective way is described. The method, called least-squares subtraction, depends only on the linearity of the spectra and the linear independence of the analyte's spectrum from those of the interferences. The spectrum that results after correction is modified from the spectrum of the pure analyte, but it can still be used for quantitation and identification. Because of this modification, the method is best applied to problems where the presence of the analyte is obscured by the interferences, or when fast, automatic removal of one or more time-varying interferences from a series of spectra is required. The method is illustrated by the removal of a strong interference from the ATR measurement of the infrared spectrum of an enzyme-substrate complex and the removal of a changing chromatographic baseline from the output of a liquid chromatograph's diode array detector.
Richard F. Lacey, "Elimination of Interferences in Spectral Data," Appl. Spectrosc. 43, 1135-1139 (1989)