HIGH ELECTRONIC SELECTIVITY OBTAINABLE WITH NONSELECTIVE CHEMOSENSORS

This paper gives a short review of the different methods of pattern recognition for chemosensors, showing that the identification of substances is essential as a first step. After a definition of measures for the electronic selectivity, two examples are presented showing realistic conditions. Usually the real number of distinguishable substances (patterns) is much lower than the theoretical limit. Two methods to improve this situation are sketched; one by tailoring the cross-sensitivities of the sensor elements and the other by reducing the number of sensor elements.