Five-way excitation-emission-kinetic-pH data processed by a new algorithm: Unfolded partial least-squares with residual quadrilinearization.

MAGGIO, R.; OLIVIERI, A.C.; MUÑOZ DE LA PEÑA, A.

Budapest

Congreso; XIII Conferencia en Quimiometría en Química Analítica; 2012

Hungarian Chemical Society

Only in a few cases have data higher than second-order been recorded and used to construct quantitative calibration models and to develop practical analytical methodologies [1]. Unfolded partial least-squares in combination with residual quadrilinearization (U-PLS/RQL), is developed and presented as a new latent structured algorithm for the processing of five-way instrumental data. In order to check its analytical predictive ability, fluorescence excitation-emission-kinetic-pH data were measured and processed. Five-way data, in addition to the ”second-order advantage”, would display the obvious advantage of providing richer analytical information, implying more stable methods towards interferences and matrix effects. The concentration of the fluorescent pesticide carbaryl was determined in the presence of two uncalibrated interferents, the pesticides fuberidazole and thiabendazole, which are usually found in environmental samples. The hydrolysis of the analyte was followed at different pH values using a fast-scanning spectrofluorimeter, recording the excitation-emission fluorescence matrices (Scheme 1), during its evolution to produce 1-naphthol, which is also fluorescent [2]. The new model implies two stpes: (1) calibration with PLS after unfolding the arrays into vectors, without including data for the unknown sample, and (2) modeling the interferent contribution with the RQL procedure, analogous to that already described for four-way data, extended one dimension further [3]. The estimation of the analytical figures of merit of the new multi-way data processing algorithm will be presented and discussed, following a new approach based on uncertainty propagation theory. The newly developed algorithm is simple to implement, and seems to achieve superior performance than the classical parallel factor analysis (PARAFAC) towards the studied five-way data set in terms of predictive ability, because of its inherent latent-structured flexibility, which allows the handling of data that are not quadrilinear, as in the present case. The use of higher-order data opens new strategies for resolving analytical situations in complex samples.  Acknowledgement: Universidad  Nacional de Rosario, CONICET (Project PIP 1950), ANPCyT (PICT 2010-0084), MCI of Spain (Project CTQ2011-25388) and Junta de Extremadura (Project GR10033-FQM003).  References: [1] A.C. Olivieri, G.M. Escandar, A. Muñoz de la Peña, Trends Anal. Chem., 30, 607 (2011). [2] R.M. Maggio, P.C. Damiani, A.C. Olivieri, Anal. Chim. Acta, 677, 97 (2010). [3] J.A. Arancibia, A.C. Olivieri, D. Bohoyo Gil, A. Espinosa Mansilla, I. Durán Merás, A. Muñoz de la Peña, Chem. Int. Lab. System. 80, 77 (2006).