Multivariable factor analysis for time-resolved laser fluorescence spectroscopy of the uranyl adsorption on gibbsite

Sao, Hirokazu*; Ishida, Keisuke; Saito, Takumi*; Aoyagi, Noboru   ; Nagasaki, Shinya*; Tanaka, Satoru*

Structures of the Ln/An in solution and on mineral surfaces have been assessed by various spectroscopic techniques. In a relatively simple system containing a few species in solution this can be done unequivocally. However, on mineral surfaces at relatively high pH or in the presence of ligand molecules measured spectra are hard to interpret because of the co-existence of many species and/or the heterogeneity of the surfaces. In addition spectroscopic measurements involve several parameters, not only spectroscopic parameters like wavelength and delay time, but also chemical parameters like pH and concentrations of reactants. Therefore, the information on the number and relative amounts of different chemical species formed should be pursued in the entire multi-dimensional parameter space and one needs a systematic approach based on statistical argument for the purpose. The goal of this study is to construct a multivariable analysis method for TRLFS, which can extract the number of components, their pure spectra and the variation of their quantities from a series of spectra measured as a function of wavelength, time and chemical conditions. Uranyl sorption on gibbsite is studied by TRLFS as a test system for this purpose. It has been shown that the spectral shapes of uranyl on gibbsite varies as the delay time of the detector increases and as pH of the solution is changed, suggesting the presence of multiple sorption complexes of uranyl on gibbsite. The obtained TRLFS spectra are processed by the fixed-size window evolving factor analysis to extract the number of statistically meaningful components, followed by nonlinear least square regression for the deconvolution of temporally varying spectra.