Investigation of statistics strategies for improving the discriminating power of laser-induced breakdown spectroscopy for chemical and biological warfare agent simulants

Laser-induced breakdown spectroscopy spectra of bacterial spores, molds, pollens and nerve agent simulants have been acquired. The performance of several statistical methodologies-linear correlation, principal components analysis, and soft independent model of class analogy-has been evaluated for their ability to differentiate between the various samples. The effect of data selection (total spectra, peak intensities, and intensity ratios) and pre-treatments (e.g., averaging) on the statistical models have also been studied. Results indicate the use of spectral averaging and weighting schemes may be used to significantly improve sample differentiation.