Linking ammonium nitrate-aluminum (AN-AL) post-blast residues to pre-blast explosive materials using isotope ratio and trace elemental analysis for source attribution

Forensic science practitioners are often called upon to attribute crimes using trace evidence, such as explosive remnants, with the ultimate goal of associating a crime with a suspect or suspects in order to prevent further attacks. The explosive charge is an attractive component for attribution in crimes involving explosives as there are limited pathways for acquisition. However, there is currently no capability to link an explosive charge to its source via post-blast trace residues using isotope ratios or trace elements. Here, we sought to determine if pre-blast attribution signatures are preserved after detonation and can be subsequently recovered and detected. A field study was conducted to recover samples of post-blast explosives from controlled detonations of ammonium nitrate-aluminum (AN-Al), which were then analyzed via isotope ratio mass spectrometry (IRMS) and inductively coupled plasma-mass spectrometry (ICP-MS) for quantitation and profiling of isotopes ratio and trace element signatures, respectively. Oxygen and nitrogen isotope ratios from AN-Al yielded some of the most promising results with considerable overlap within one standard deviation of the reference between the spreads of pre- and post-blast data. Trace element results from AN-Al support the findings in the isotope ratio data, with 26 elements detected in both pre- and post-blast samples, and several elements including B, Cd, Cr, Ni, Sn, V, and Zn showing considerable overlap. These preliminary results provide a proof-of-concept for the development of forensic examinations that can attribute signatures from post-blast debris to signatures in pre-blast explosive materials for use in future investigations.

Composite Machine Learning Algorithm for Material Sourcing

This study developed a composite machine learning algorithm for attribution of materials of forensic interest (like ammonium nitrate) to original sources. k-nearest neighbor and random forest models were used for source elimination and classification, respectively, in a two-step, composite algorithm based on particle color, size/shape, and trace element concentration features. Novel approaches for simulation to supplement within-source reference features based on empirically measured multi-lot analyses, an improved hold-one-lot-out method for cross-validation, an assessment of the likelihood of the presence of a reference sample, fusion of the source probabilities from the respective classification models, and the calculation of metrics for assessing ensemble sourcing performance are described. Excellent sourcing predictions were obtained; the sourcing algorithm identified the correct source as the top choice 89% of the time, and the correct source was identified to be an average of 2.7 times more likely than the most likely incorrect source.