INFLUENCE OF NOISE, PEAK POSITION AND SPECTRAL SIMILARITIES ON RESOLVABILITY OF DIODE-ARRAY HIGH-PERFORMANCE LIQUID-CHROMATOGRAPHY BY EVOLUTIONARY FACTOR-ANALYSIS

The resolvability of peaks using diode-array detection high-performance liquid chromatography is influenced by factors including peak position, noise, spectral similarities and relative peak intensities. Extensive simulations are performed of pairs of peaks at increasing separations, and different spectral similarities. Graphs of PC2 versus PCI are presented that demonstrate that shape is dependent on peak separation and width on spectral similarities as indicated by correlation coefficients. Noise distorts these graphs. Approaches for determining the number of significant components in the mixture and where each compound elutes are discussed, and a numerical approach of looking at the size of eigenvalues in different windows is adopted. Both fixed window (EFF) and expanding window (EFE) approaches are compared using 16 selected simulations. Noise is found to be by far the most serious factor influencing resolvability. EFE methods are more influenced by spectral similarities than EFF methods.