Despite its intrinsic advantages, translation of laser induced breakdown spectroscopy for material identification has been often impeded by the lack of robustness of developed classification models, often due to the presence of spurious correlations. While a number of classifiers exhibiting high discriminatory power have been reported, efforts in establishing the subset of relevant spectral features that enable a fundamental interpretation of the segmentation capability and avoid the ‘curse of dimensionality’ have been lacking. Using LIBS data acquired from a set of secondary explosives, we investigate judicious feature selection approaches and architect two different chemometrics classifiers –based on feature selection through prerequisite knowledge of the sample composition and genetic algorithm, respectively. While the full spectral input results in classification rate of ca.92%, selection of only carbon to hydrogen spectral window results in near identical performance. Importantly, the genetic algorithm-derived classifier shows a statistically significant improvement to ca. 94% accuracy for prospective classification, even though the number of features used is an order of magnitude smaller. Our findings demonstrate the impact of rigorous feature selection in LIBS and also hint at the feasibility of using a discrete filter based detector thereby enabling a cheaper and compact system more amenable to field operations.

Laser induced breakdown spectroscopy (LIBS) has surfaced as an attractive alternative to mass spectrometry and wet chemistry methods for chemical identification, driven by its real-time, label-free nature. Rapid analysis needs, especially in high-energy materials and pharmaceutical compounds, have further fueled an increasing number of refinements in LIBS. Yet, isomers are seldom identifiable by LIBS as they generate nearly identical spectra. Here we employ a suite of chemometric approaches to exploit the subtle, but reproducible, differences in LIBS spectra acquired from structural isomers, a set of pyrazoles, to develop a sensitive and reliable segmentation method. We also investigate the possible mechanistic principles (causation) behind such spectral variations and confirm their statistically significant nature that empowers the excellent classification performance.

An experimental design based on spectral construction, which has potential to minimise the sample consumption, the number of laser shots and time required to collect the data from laser induced breakdown spectroscopy for identification of the explosive materials is reported in the study. This approach is an ideal solution in the field of hazardous material detection, where the availability of the sample can be a serious limiting factor. The experimental data recorded on a set of five high energy materials has been considered to test the performance of the proposed methodology. Multiple spectra are constructed by assuming a normal distribution at each wavelength of the spectrum, where random numbers are generated using the mean and standard deviations obtained from arbitrarily chosen five experimental spectra from each class. The newly generated spectra are called as synthetic spectra. The correct classification obtained from -K -nearest neighbour combined with principal component analysis and partial least square -discriminant analysis demonstrated very promising results. The correct classification rates differed by only 4 per cent -7 per cent as compared to conventional approach where experimental spectra alone are considered for the analysis. Further, when RDX is excluded, the obtained results are almost identical with conventional approach.