The electronic tongue is an electrochemical sensor that is composed of a sensor array of metal working electrodes, one Ag/ AgCl reference electrode, and an auxiliary electrode made of stainless steel. The working electrodes are non-selective with partly overlapping selectivity. The measurement techniqueis based on voltammetry, and pulses of potentials of different magnitude are applied to the electrodes and the current is registered. The current arises when potentials are sufficient to induce redox reactions at the surface of the working electrodes. Chemometric methods are used to extract information from the measurements made with the electronic tongue. A prerequisite of the sechemometric procedures is that the relevant information in a data matrix is found in the variable variance. Principal component analysis (PCA) is a technique used to find patterns such as clusters of samples and outliers in a data matrix X. Soft independent modeling of class analogy (SIMCA) is a suitable technique for classifying samples, and it is based on using PCA to construct the class models. New samples are subsequently compared with all the models and are or are not assigned to one or more of the models. Partial least squares regression (PLS-R) is also a chemometric technique that uses PCA, in this case to model two different matrices, X and Y. The electronic tongue is a robust method that allows fast and easy measurements and does not destroy the samples during measurements. These features make the device highly interesting in industrial applications, such as performing measurements in harsh environments or on-line in a production process.

For understandable reasons, microorganisms thrive in the foods we eat, and itis important to be able to detect and monitor their activity in food. To study the capacity of the electronic tongue in this context, the performance of the device in measuring microbial activity was evaluated. The results show that the electronic tongue could efficiently survey the growth of mold in liquid media, and it could also be used to predict the ergosterol content in mold samples (paper I). Moreover, the technique was successful at distinguishing between different species of microorganisms (papers III and IV). The electronic tongue used in the present studies appeared to be better suited for such differentiations than another such device based on potentiometry (paper IV). Theoretically, the electronic tongue measures redox active compounds, and this was supported by characterization of samples using both the electronic tongue and a high performance liquid chromatography instrument equipped with an electrochemical detector (paper II).

The electronic tongue shows promise as a method for detecting and measuring the activity of microbes. However, before this technique can be implemented in industrial applications, it must be further investigated regarding the detection limits for microbial activity, repeatability/reproducibility, and the possibility of characterizing the measured compounds.