P4.4 Analysis and implementation of regularization techniques for pure component decompositions
Supervisors:
Topic and overall goal:
Multivariate curve resolution (MCR) methods aim at extracting the concentration profiles of the pure components and their spectra from mixed spectroscopic data. MCR methods require that the measured spectral data comply with the bilinear Lambert-Beer law. A first step is to determine the number of chemical species which is followed by the recovery of the pure component profiles. MCR methods can be applied to (time-)series of UV-vis, IR or Raman spectra and others as, in the case of spectroelectrochemistry, to series of spectra depending on the electric potential. For this purpose, we develop improved mathematical methods and numerical algorithms based on our previous work and apply the developed methods in the project areas of the RTG. The work focuses on reducing factor ambiguities by the application of constraint-based optimization techniques with constraints as unimodal concentration profiles or their consistency with a kinetic model and many others. These methods have the advantage to produce reliable pure component information. However, they suffer from their complexity which sometimes makes them difficult to use for non-experienced users.