A procedure based on fitting the experimentally computed autocovariance function (ACVF) of multicomponent chromatograms to theoretical models is introduced by which both the single component interdistance model (IM) of the retention times is tested and the statistical attributes of the multicomponent chromatogram (I.e. number m of single components, peak width, and parameters of the IM) are determined. Four different IM—exponential, uniform, normal, and gamma—are considered. In essence, when fitted to these theoretical models, the experimental ACVF—expressing the chromatographic response correlation on the time distance—provides the information necessary to establish both the type of retention pattern and gives the necessary parameter estimation. The procedure is tested by using computer-generated chromatograms with different IMs and uncorrelated peak heights, in which density and m are varied. It is shown that the chromatographic attributes m and peak width derived from the best fitting IM are unbiased. Moreover, even if the best fitting IMs do not always coincide with the true model, because of their flexibility and approximating properties they always give a correct description of the retention pattern provided that the results are correctly interpreted. © 1992, American Chemical Society. All rights reserved.
Fourier Analysis of Multicomponent Chromatograms. Recognition of Retention Patterns
PASTI, Luisa;DONDI, Francesco
1992
Abstract
A procedure based on fitting the experimentally computed autocovariance function (ACVF) of multicomponent chromatograms to theoretical models is introduced by which both the single component interdistance model (IM) of the retention times is tested and the statistical attributes of the multicomponent chromatogram (I.e. number m of single components, peak width, and parameters of the IM) are determined. Four different IM—exponential, uniform, normal, and gamma—are considered. In essence, when fitted to these theoretical models, the experimental ACVF—expressing the chromatographic response correlation on the time distance—provides the information necessary to establish both the type of retention pattern and gives the necessary parameter estimation. The procedure is tested by using computer-generated chromatograms with different IMs and uncorrelated peak heights, in which density and m are varied. It is shown that the chromatographic attributes m and peak width derived from the best fitting IM are unbiased. Moreover, even if the best fitting IMs do not always coincide with the true model, because of their flexibility and approximating properties they always give a correct description of the retention pattern provided that the results are correctly interpreted. © 1992, American Chemical Society. All rights reserved.I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.