Selective detection and discrimination of nitro explosive vapors using an array of three luminescent sensory solid organic and hybrid polymer membranes

Abstract

tThree fluorogenic sensory polymer membranes, film shaped (one organic and two hybrids containingcomplexes of Tb(III) and Sm(III)) based on a dicoumarol motif were designed and prepared for use assensory matrices to selectively detect nitro explosives (TNT, RDX and PETN) in the environment. Thus,the exposure of these membranes to these explosives and to 6 explosive mimics gave rise to variations inthe fluorescence spectra of the materials as a function of the type of vapor and exposure time. Therefore,the fluorescence intensity recorded over the wavelength range from 381 to 575 nm and at times of5, 10, 15 and 30 min for each of the 3 membranes had a highly defined internal structure. Thus, theprincipal component analysis (PCA) showed that only 3 principal components (PCs) were needed for eachmembrane for its description (9 PCs in total compared with the 195 wavelengths). The PCA allows forboth discrimination between explosives and their mimics and for distinction between the 3 explosives.A SIMCA (soft independent modeling of class analogy) model was built using the scores from the PCsto a confidence level of 95%, which is completely specific (100%) for the vapors of each explosive andexplosive mimic. Furthermore, the 3 explosive models and the model built using the 6 explosive mimicshave a sensitivity of 100% and 95.8%, respectively. The modeling power and discriminant capability fromthe 9 PCs show that no membrane could be removed to build these models. Using a multilinear regressionwith the 9 PCs as predictor variables and binary responses, it was possible to model the fitted responseusing a normal distribution for each class (explosives and explosive mimics). There is a probability of1.5 × 10−9that a sample is not an explosive when actually it is and a probability of 0.05 that a sample isan explosive when it is an explosive mimic.