Designing Optimal Surface Acoustic Wave Electronic Nose for Body Odor Discrimination

Abstract

The paper suggests a method for selection of polymer coatings for vapor sensor array based electronic nose systems. It is proposed that prior to actual experimental sensor array fabrication one can (i) consider all available potential polymers whose interaction chemistry with target vapors is documented, (ii) apply statistical clustering methods for grouping polymers of similar chemistry into clusters, and then (iii) make an optimal selection of only one polymer from each cluster for defining the sensors array. The process at stage (iv) would be based on simulation results for vapor discrimination by using best known sensor response models. In this work we have applied this strategy for designing a surface acoustic wave sensor array for discrimination of odorants from living and dead human body. After making optimal selection of polymers we generate a model based sensor array data for 33 volatile organics from human body. In brief, principal component analysis and hierarchical cluster analysis methods have been used for making polymer selection by analyzing vapor-polymer partition coefficient data matrix. The latter is generated by using the available solvation parameters and the linear-solvation-energy relationship for 22 potential polymers. Of these, only 3 polymers—SXPYR (alkyl-amino-pyridyl substituted polysiloxane), P4V (poly 4-hexafluoro isopropanol-styrene) and PEI (poly ethylenimine) were selected to make an optimal set on basis of principal components load analysis. The simulation data were generated by using a 3-element surface acoustic wave sensor array model functionalized with the selected polymers. Finally, the principal component analysis is shown to yield good separability in feature space between the living and the dead body odors, and also between various components in the living body odor.