DIAGNOSIS AND RESOLUTION OF MULTIWAVELENGTH CHROMATOGRAMS BY RANK MAP, ORTHOGONAL PROJECTIONS AND SEQUENTIAL RANK ANALYSIS

Cases with unresolved chromatographic peaks, where the diagnosis and subsequent resolution using common procedures from evolutionary factor analysis fails, are investigated and discussed in some detail. A new procedure is then developed. Local principal component analysis is first performed using a window procedure called eigenstructure tracking analysis. This evolving and dynamic procedure provides the best possible sensitivity for assessment of the number of coeluting analytes at a particular retention time as the method adapts the window size to the number of coeluting analytes. The result is a rank map in retention title direction, showing, in any retention time interval, the number of eluting analytes. The information in the rank map is subsequently used to define a set of n orthogonal projection matrices (n is the total number of detected analytes in the chromatogram). Each projection matrix is constructed from n - 1 loading vectors obtained, e.g., from principal component analysis of the zero-concentration window for a particular analyte. This procedure enables unambiguous identification of the peak pattern in an unresolved region. Resolution of the chromatographic peaks can then be accomplished by the combined use of orthogonal projections and a procedure called sequential rank analysis to solve the problem of embedded peaks. Sequential rank analysis assumes local symmetry around peak: maxima and uses first-order differentiation for unique resolution of embedded peaks. The approach is demonstrated on some simulated chromatographic systems.