Chemical variance, a useful tool for the interpretation and analysis of two-dimensional chromatograms

A new method is presented for the analysis of comprehensive two-dimensional data. The approach complements existing chemometric methods for analysing these kinds of data (e.g. multivariate-curve-resolution methods, MCR), and is especially suited for the analysis of families of compounds. The method is based on finding a new pair of axes (replacing the retention time axes) to represent the data. A drop in the matrix rank is observed when one of the axes collects all the variance for a certain family of compounds. This is achieved by introducing retention models (in both chromatographic dimensions) that describe the behaviour of a family of compounds as a function of the property that varies between the members. After this transformation, a single family of compounds gives rise to a single source of variance. This suggests the application of MCR techniques in a second step to separate the contributions of different (overlapped) families of compounds. For the latter application, more research has to be performed. The approach is illustrated using the separation of families of functional poly(methyl-methacrylate) polymers with different numbers of hydroxyl end groups and degrees of polymerisation or, equivalently, with the molecular weight as the key descriptive property within one family.