Recognition of 2,4-DNT, RDX and TATP in various matrices by FTIR-partial least squares-discriminant analysis and kinetics of surface sublimation.

摘要

A methodology useful for processing spectroscopic information using pattern recognition was designed, developed and implemented in detection of energetic materials at trace level in air, sand and surfaces. Partial Least Squares-1 (PLS) was used to generate vectors to be used with pattern recognition. These vectors were then coupled to Discriminant Analysis by adjusting to a discriminating function. Fourier Transform Infrared (FTIR) spectra of traces of 2,4-dinitro-toluene (2,4-DNT) in air, triacetone triperoxide (TATP) in air and air free of those explosives were recorded and used for generated the vectors. Short wave infrared (Near IR) and Long wave infrared (Mid IR) regions were studied and used for the model. Two vectors were necessary for good discrimination for TATP and four vectors for 2,4-DNT; but, when the regions were weighted from the response of detector, the model was improved and less vectors were needed for all discrimination in the prediction of new samples. (Abstract shortened by UMI.)