SELF-MODELING MIXTURE ANALYSIS OF SPECTRAL DATA WITH CONTINUOUS CONCENTRATION PROFILES

Self-modeling mixture analysis techniques can be divided in two classes. The first class consists of algorithms where special properties of spectra are used as a first step. One of these techniques determines pure variables, i.e. variables that have intensity contributions from only one component. Another technique determines the purest spectra in a mixture data set. The second class consists of techniques where special properties of the concentration profiles are used in the algorithm, such as a unimodal shape of the concentration profile. This tutorial focuses on the second class of techniques. Applications to gas chromatography-mass spectrometry, time-resolved mass spectra and liquid chromatography-UV detection will be given.