Revealing hidden structure in time-resolved spectral matrices using multivariate analysis of the streak camera data

Abstract

Two new multi-stage smoothing LRA and PARAFAC-based methods were suggested and applied to study a 3D array composed of a set of the time-resolved spectral matrices (TRSM) containing instant fluorescence spectra of the fluorescein solution in methanol measured by the streak camera with picosecond resolution. Numerical modeling showed that both methods combined with Fourier processing allowed identifying the hidden structure of the pattern obscured entirely by noise. The PARAFAC algorithm was shown to have some advantages over LRA since the averaging of matrices in the last method causes information losses.Mathematical analysis of 20 TRSM of the fluorescein solution in methanol allowed us to expand the time range of instantaneous spectra while studying fast processes.The time dependence of the statistical moments of the spectra and their second-order derivatives was associated with the molecular reorganization processes of fluorescein solvates after absorption of excitation quanta.