El reconocimiento de la cerveza con el flujo a través de matriz de sensores basados ​​en

Talanta xxx (2006) xxx–xxx

The recognition of beer with flow-through sensor array based on miniaturized solid-state electrodes
Patrycja Ciosek ∗ , Wojciech Wr´ blewski o
Department of Analytical Chemistry, Warsaw University of Technology, Noakowskiego 3, 00-664 Warsaw, Poland Received 18 August 2005; received in revised form 7 December 2005; accepted 14 December 2005

Abstract Flow-throughelectronic tongue based on miniaturized solid-state potentiometric sensors has been developed. A simple technique, i.e. membrane solution casting on the surface of the planar Au transducers was applied for the preparation of classical ion-selective and partially selective microelectrodes, introduced in the flow-through sensor array. The performance of the designed electronic tongue was tested inthe qualitative analysis of various brands of beer. Samples of the same brand of beer but with different manufacture dates, originating from different manufacture lots, have been applied in the studies. The combination of PLS and ANN techniques allowed the discrimination between different brands of beer with 83% of correct classifications. © 2005 Elsevier B.V. All rights reserved.
Keywords:Electronic tongue; Sensor array; Solid-state electrode; Flow-through analysis; Food classification

1. Introduction Food quality, safety, and uniformity of products are emerging tasks in food industry. Characterization of complex samples in food industry is conducted with the application of various analytical techniques based mainly on efficient separation techniques, e.g. liquid chromatography coupledwith spectroscopic detection systems. These techniques are time-consuming, expensive, and require specialized equipment and pre-treatment of sample. Chemical sensors can be used in modern analytical systems eliminating the necessity of a special sample preparation. Systems based on sensor arrays are especially powerful tools for the discrimination of characteristic properties of food samples,distinguishing among various types of them, and providing the recognition of their taste [1–5]. However, sophisticated chemometric methods are necessary to analyse the responses of the sensor array systems. Application of various pattern recognition tools, such as Artificial Neural Networks (ANN), Partial Least Squares (PLS), Soft Independent Modelling of Class Analogy (SIMCA), K-Nearest Neighbours (KNN)enables the analysis of

∗

Corresponding author. Tel.: +48 22 6607873; fax: +48 22 6605631/7873. E-mail address: pciosek@ch.pw.edu.pl (P. Ciosek).

measured data and the classification of multidimensional pattern spaces [6]. Brewery industry is a growing branch of foodstuff production. Nowadays, the consumption of the beer is growing at the cost of beverages with bigger alcohol content,such as wine. This is due to greater health consciousness and stricter law concerning driving and drinking. The characterization of beer flavour is complicated due to its complex composition and traditionally is performed with the combination of gas chromatography and organoleptic profiling panels or using HPLC methods [7]. These analytical procedures are usually expensive, time-consuming and demandthe participation of specialized panellists. For these reasons there is a strong need for the development of fast methods and portable systems for product control and classification in brewery industry. First attempts to construct sensor array systems for beer recognition were already presented. Pearce et al. [8] described the multisensor system (so-called electronic nose) based on 12 gas sensors,which could discriminate between various kinds of beer (Lager and Ale) according to their odour. High concentration of alcohol caused problems in such approach, so Heberle et al. proposed the combination of electronic nose with a pre-separation step, realized with gas chromatography column [9]. However, the analysis of the ionic and non-volatile

0039-9140/$ – see front matter © 2005 Elsevier...