Quantitative resolution of overlapping tailing peaks obtained by diode-array detector high performance liquid chromatography in the absence of pure standards using simple chemical knowledge

The diode-array high performance liquid chromatograms of 10 samples of various mixtures of the polyaromatic hydrocarbons 1,2-benzanthracene and chrysene were recorded in different proportions ranging from 1:1 to 1:0.001. A new approach to deconvolution based upon determining purity curves, correct selection of points in time and wavelengths, and multiple linear regression is proposed. The key to the algorithm is to select those statistically acceptable solutions for which the recovered spectra are positive and the chromatographic peaks are unimodal. It is shown that these two constraints reduce the range of solutions significantly and in eight of the samples result also in good estimates of relative chromatographic peak area, demonstrating that simple chemical information when combined with statistical methods can result in quantitative resolution in the absence of external standards.