Quantitative resolution of spectroscopic systems using computer-assisted target factor analysis (CAT)

Abstract

Factor analysis (FA) is widely applicable in analytical chemistry. For suitable data structures, the generality of the approach is a strong point of FA. Using target testing procedures, a wealth of specific informations on the components in a chemical system can be extracted without stating a specific model. Target testing, however, is time-consuming and often heuristic. In the following, application of a computer-assisted target factor analysis (CAT) algorithm is described. CAT estimates a rotation matrix that transforms abstract factors into physically meaningful informations by using general constraints, i.e. non-negativity of absorptions and/or concentrations. Thus, the rotation step in factor analysis is automated and allows application of target factor analysis to situations where none or only very limited information on the real factors and their respective contributions is available. Thus, a target testing procedure of guessing physically meaningful factors and iteratively adapting these factors is performed automatically. Nevertheless, CAT is not a black box procedure. The relative importance of different optimization constraints is balanced interactively. CAT can be applied to all situations where factor analysis can be used. CAT is demonstrated using UV-Vis absorption spectra of a Nd(III) polyoxometalate cryptate system and the system U(VI)-CO₂-H₂O as examples.