Implementation of statistical methods on LIBS data for classification of residues of energetic materials (nitro compounds)

Shikha Rai1, A.K. Rai1*, I.M.L. Das2, K.C. Tripathi3

1Laser Spectroscopy Research Laboratory, Department of Physics, University of Allahabad, Allahabad 211002, India

2Department of Physics and K. Banerjee Center of Atmospheric & Ocean Studies, University of Allahabad, Allahabad 211002, India

3K. Banerjee Center of Atmospheric & Ocean Studies, University of Allahabad, Allahabad 211002, India

Adv. Mater. Lett., 2012, 2 (1), pp 32-37

DOI: 10.5185/amlett.2010.11184

Publication Date (Web): Apr 08, 2012

E-mail: awadheshkrai@rediffmail.com

Abstract


Our key aim is to validate the use of statistical methods for analysis of Laser-Induced Breakdown Spectroscopy (LIBS) datasets of pure nitro compounds (4-nitroaniline and 4-nitrotoluene) and of test samples formed in Cu matrix. Laser-Induced Breakdown Spectroscopy (LIBS) provides the spectral lines of the constituent elements. The interest behind this study is to establish the essence behind the supplementation of LIBS analysis with statistical methods. When the energetic materials were doped with the interferents, such as Cu metal powder it leads to the alteration of the spectral profile of both the target samples, which have similar constituent elements such as C, H, N and O. So, for this situation, it is difficult to classify the test samples from their pure samples only on the basis of its spectral signatures. Hence, in order to classify these sets, we have applied sophisticated chemometric techniques such as linear correlation and Principal Components Analysis (PCA) to familiar LIBS datasets and found that 50% test samples of 4-nitroaniline and 70% test samples of 4-nitrotoluene were successfully discriminated. The causes for partial classification for both the samples have also been discussed in detail.

Keywords

LIBS, nitro compounds, linear correlation, principal component analysis

Upcoming Congress

Knowledge Experience at Sea TM