Reliable multivariate curve resolution of femtosecond transient absorption spectra

In this work, we discuss the reliability of soft-modelling multivariate curve resolution of transient absorption spectra of benzophenone which is chosen as a case study. The main issue concerns the recovery of short-time intermediate excited state that is probed in the UV–Visible on pico-second time scale. The corresponding spectrum and kinetic profile are provided. As such information has never been reported elsewhere, we propose to evaluate the pertinence and the robustness of the resolution through different data analysis strategies. It is shown how overfit situation can be detected analyzing the 2-way matrix of residuals. Another approach concerns the analysis of augmented data matrix. It enables to provide global solutions that are thus less affected by rotational ambiguities. The application of hard-modelling constraint on the kinetic profiles in MCR-ALS is also evaluated as a way to validate the results.