analytica chimica acta 614 ( 2 0 0 8 ) 213–222 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/aca Amperometric sensors based on poly (3,4-ethylenedioxythiophene)-modified electrodes: Discrimination of white wines L. Pigani a , G. Foca b , K. Ionescu a , V. Martina a , A. Ulrici b , F. Terzi a , M. Vignali c , C. Zanardi a , R. Seeber a,* a Dipartimento di Chimica, Universit ` a di Modena e Reggio Emilia, via G.Campi 183, 41100 Modena, Italy b Dipartimento di Scienze Agrarie e degli Alimenti, Universit ` a degli Studi di Modena e Reggio Emilia, Padiglione Besta, via Amendola 2, 42100 Reggio Emilia, Italy c Vinicola San Nazaro, Via Gonzaga 12, 46020 Pegognaga (MN), Italy article info Article history: Received 22 January 2008 Received in revised form 4 March 2008 Accepted 12 March 2008 Published on line 20 March 2008 Keywords: Amperometric sensor Electronic tongue Conducting polymer White wines Principal component analysis Partial least squares-discriminant analysis abstract The voltammetric responses on selected white wines of different vintages and origins have been systematically collected by three different modified electrodes, in order to check their effectiveness in performing blind analysis of similar matrices. The electrode modifiers consist of a conducting polymer, namely poly(3,4-ethylenedioxythiophene) (PEDOT) and of compos- ite materials of Au and Pt nanoparticles embedded in a PEDOT layer. Wine samples have been tested, without any prior treatments, with differential pulse voltammetry technique. The subsequent chemometric analysis has been carried out both separately on the signals of each sensor, and on the signals of two or even three sensors as a unique set of data, in order to check the possible complementarity of the information brought by the different electrodes. After a preliminary inspection by principal component analysis, classification models have been built and validated by partial least squares-discriminant analysis. The discriminant capability has been evaluated in terms of sensitivity and specificity of classification; in all cases quite good results have been obtained. © 2008 Elsevier B.V. All rights reserved. 1. Introduction In recent years, efforts have been made to develop arrays of non-specific sensors, coupled with chemometric tools (the so-called “electronic tongues”), devoted to recognition and dis- crimination of complex liquid mixtures [1–5]. The feeling is that similar hardware plus software systems may constitute in such cases effective alternative approaches to quite difficult, if even possible, systematic compositional analyses. ∗ Corresponding author. Tel.: +39 059 2055027; fax: +39 059 373543. E-mail address: renato.seeber@unimore.it (R. Seeber). The nature of the sensors identifies different approaches followed in order to build up an electronic tongue. In par- ticular, arrays of amperometric sensors have proven to be highly efficient in discriminating among different kinds of liq- uid nourishments [6–10]. The response signal to submit to the so-called “blind analysis”, consisting in this case of a series of voltammetric curves obtained by the different electrodes, is taken as a sort of “fingerprint” of the sample analysed. From this point of view, increasing the number of sensors 0003-2670/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2008.03.029