Trace explosives detection using photo-thermal infrared imaging spectroscopy (PT-IRIS): theory, modeling, and detection algorithms

Abstract

We are developing a technology for stand-off detection based on photo-thermal infrared imaging spectroscopy (PT-IRIS). In this approach, one or more infrared (IR) quantum cascade lasers are tuned to strong absorption bands in the analytes and directed at the sample while an IR focal plane array is used to image the subsequent thermal emissions. In this paper we present recent advances in the theory and numerical modeling of photo-thermal imaging and spectroscopy of particulates on flat substrates. We compare the theoretical models with experimental data taken on our mobile cartbased PT-IRIS system. Synthetic data of the photo-thermal response was calculated for a wide range of analytes, substrates, particle sizes and analyte mass loadings using their known thermo-physical and optical properties. These synthetic data sets can now be generated quickly and were used to accelerate the development of detection algorithms. The performance of detection algorithms will also be discussed.

Detection of trace explosives on relevant substrates using a mobile platform for photothermal infrared imaging spectroscopy (PT-IRIS)

“…In contrast, the z-transform at 1000Hz dramatically enhances the signal from the surface particles while drastically reducing the signal from the bulk substrate. Similar particle/substrate relative enhancement was observed for the other analyte and substrate combinations 12 . This z-transform process makes it much more likely to correctly identify the surface analytes, as we show in the following section on algorithms and classifiers.…”

“…A more thorough analysis is presented elsewhere 12 . Each of the sample coupons was measured using the pulse profile described in the previous section at each of the 88 QCL illumination wavelengths.…”

“…Along with the differential processing, which inherently measures the slow "quasi-static" response of the sample, an "impulse" processing method analogous to a Laplace transform was also performed at several frequencies for both the long pulse and for the sequence of 16 fast pulses. A full description of the Laplace transform (more appropriately described for digitized data as a "z-transform") method is presented elsewhere 12 . The purpose of extracting the z-transform response at several frequencies is to characterize the thermal time constant of the PT-IRIS signal.…”

“…The sample responses to each of the 88 QCL wavelengths measured constitute the "features". Feature vectors of this form are typically used as the input data for classifier algorithm analysis routines 8,12 . Each pixel within the laser-illuminated area was treated as an independent "sample".…”

“…The details of the analysis and classifier algorithm methods are described elsewhere 12 . Here we summarize the basic steps taken and highlight a few results.…”