A Comparison of Multi-Transducer Arrays and Single-Transducer Arrays for the Determination of Multi-Vapor Mixtures

This study provides a comprehensive analysis of multi-transducer (MT) array performance, illustrating quantitatively the advantages of such arrays over their single-transducer (ST) counterparts. Calibrations were performed for 11 vapors with five cantilevers, five capacitive, and five calorimetric sensors coated with five different polymers. Using these data in Monte Carlo simulations coupled with a disjoint principal component regression pattern recognition algorithm we examine the predicted rates of recognition achievable for analyses of individual vapors and their binary and ternary mixtures. Optimal MT arrays consistently outperform optimal ST arrays of similar dimension, and a number of ternary mixtures could be reliably analyzed with MT arrays that could not be analyzed with any ST arrays. Recognition rates did not increase significantly by including > 5 or 6 sensors in the array.