Novel techniques for peak recognition and deconvolution by computer-aided photodiode array detection in high-performance liquid chromatography

Digital algorithms for manipulating spectrochromatographic data, obtained by means of linear photodiode array detectors, have been examined as tests for peak purity in high-performance liquid chromatography and for their quantitative performance. Two model systems, noscapine and papaverine, and the red pigments R112 and R3, served as examples of closely overlapping and slightly overlapping systems, respectively. The absorbance ratio plot as a function of time performed less well in recognising peak overlap than the second derivative of the elution profile (d²A/dt²). A novel approach has been developed for optimising the spectral suppression technique, where a difference absorbance function deconvolutes a component of known spectral properties to permit detection of 1% noscapine in papaverine, for which the spectra differ substantially. Spectral suppression was less successful in deconvoluting trace amounts of R112 and R3, the spectra of which are closely similar. Spectral deconvolution with a "dummy" linear absorbance standard, and a non-absorbing reference wavelength, permitted the elution profiles of both model systems to be completely resolved. This method revealed the elution profile of an unknown, closely overlapping peak in a forensic sample containing papaverine and noscapine. © 1983.