Evaluation of multiple linear regression applied to impedimetric sensing

Abstract

AbstractIn this work, the application of multiple linear regression (MLR) on the development of electroanalytical methods based on electrochemical impedance spectroscopy (EIS) data was investigated. This approach was proposed considering that more than one element of the equivalent electrical circuit fitted to impedance spectra could keep linear relationship with the analyte concentration and be used in a multivariate calibration model, rather than using only one element of the circuit in a univariate regression, the benchmark procedure in the context of impedimetric (bio)sensing. First, MLR was evaluated in the individual determination of the redox probe ferrocyanide in aqueous solutions, followed by the determination of catechol and hydroquinone in tap water samples. MLR produced better predictions than univariate regression when more complex electrochemical systems (catechol and hydroquinone) and more complex samples (tap water) were analyzed. The determinations of hydroquinone and catechol in the direct analysis of spiked tap water samples were successful, with apparent recoveries ranging from 94% to 136%.