Multivariate curve resolution of step-scan FTIR spectral data

摘要

Step-scan FTIR is a time-resolved spectroscopic technique.It can successfully probe transient photochemical systems through two-way data sets:spectra are recorded as the system is evolving with time.This work focuses on chemometrics for time-resolved data sets and in particular on the potential of the multivariate curve resolution(MCR)based on alternating least squares.The usefulness of this approach is demonstrated in a case study,which involves several intermediate states with time constant ranging from the microsecond to the millisecond.The chemometrics methodology is detailed in two points.MCR relies on the bilinear decomposition of an evolving multi-component data set into the product of two matrices related respectively to the time and to the spectral domains of the extracted components.The first aspect concerns the determination of the number of components observable in the spectral data set and the calculation of initial estimates of either the time or spectral profiles.Principal component analysis based methods such as evolving factor analysis are performed and finally allow a first insight into the contributions of the four intermediates extracted.The second point concerns the alternating least squares optimization under natural constraints to recover pure intensity profiles and pure difference spectra profiles.The characterization of each photo-intermediate unaccessible individually except in non-physiological conditions is thus assessed without any assumption about the kinetic of the system.