Multivariate statistical classification of surface enhanced Raman spectra of chemical and biological warfare agent simulants

Abstract

Initial results which demonstrate the ability to classify surface enhanced Raman (SERS) spectra of chemical and biological warfare agent simulants are presented. The spectra of 2 endospores (<i>B. subtilis, B. atrophaeus</i>); 2 chemical agent simulants (Dimethyl methylphosphonate (DMMP), Diethyl methylphosphonate (DEMP)); and 2 toxin simulants (Ovalbumin, Horseradish peroxidase) were collected on multiple substrates fabricated from colloidal gold adsorbed onto a silanized quartz surface. The use of principle component analysis (PCA) and Hierarchical Clustering was used as a method of determining the reproducibility of the individual spectra collected from a single substrate. Additionally, the use of partial least squares-discriminate analysis (PLS-DA) and soft independent modeling of class analogies (SIMCA) on a compilation of data from separate substrates, fabricated under identical conditions, demonstrates the feasibility of this technique for the identification of known but previously unclassified spectra.