19 August 2015 in Scientific Reports
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 ofca.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 toca.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.
1 November 2020 in Waste Management
30 May 2019 in Journal of Analytical Atomic Spectrometry
We report the identification of five extensively used post-consumer plastics using the femtosecond laser-induced breakdown spectroscopy (fs-LIBS) technique. All the samples were obtained from a local recycling unit. An ultra-fast amplifier system delivering pulses of ∼50 fs duration was used for the investigation. Initially, a 2D scatter plot approach was employed in which a pair of selected spectral features was utilized to discriminate the samples. H–CN and C2–CN plots have shown promising discrimination. Apart from bivariate studies, principal component analysis (PCA) has been employed for the exploratory analysis where excellent segregation was achieved. Further, data were analysed using a supervised algorithm, artificial neural network (ANN). For efficient identification, an ANN model was constructed and evaluated by exploiting five different spectral windows/regions. The average identification rates obtained for all samples are in the range of 97 to 99% depending on the spectral window. The correct prediction rates with 100% accuracy were achieved when 10 prominent spectral features were employed. These findings demonstrate that the fs-LIBS combined with multivariate techniques can work as an efficient spectroscopic tool for the identification of post-consumer plastics.
6 November 2017 in Defence Science Journal
We report the ‘standoff detection’ of explosives at 1 m in laboratory conditions, for the first time in India, using Laser Induced Breakdown Spectroscopy combined with multivariate analysis. The spectra of a set of five secondary explosives were recorded at a distance of 1 m from the focusing as well as collection optics. The plasma characteristics viz., plasma temperature and electron density were estimated from Boltzmann statistics and Stark broadening respectively. Plasma temperature was estimated to be of the order of (10.9 ± 2.1) .103 K and electron density of (3.9 ± 0.5) .1016 cm-3. Using a ratiometric approach, C/H and H/O ratios showed a good correlation with the actual stoichiometric ratios and a partial identification success could be achieved. Finally employing principle component analysis, an excellent classification could be attained.
31 March 2016 in The Analyst
Laser induced breakdown spectroscopy provides a novel route for the rapid identification of structural isomers.