LIBS development methodology for forensic nuclear materials analysis

Abstract

Nuclear forensics (NF) is an important tool for the analysis of intercepted nuclear and radiological materials (NRM) to establish the relationship between NRM and their attribution in support of nuclear security. The critical challenge in NF at present is the lack of direct, rapid and non-invasive analytical techniques, especially for concealed and limited size NRM. Laser induced breakdown spectroscopy (LIBS) coupled with chemometrics has the potential to rapidly detect, quantify and attribute concealed NRM of limited sample size in air at atmospheric pressure. The goal of this study was to explore LIBS in air at atmospheric pressure in conjunction with chemometrics towards direct and rapid NF analysis. In this work, the limit of detection of uranium in cellulose was obtained at 76 ppm. The uranium lines at 385.464 nm, 385.957 nm and 386.592 nm were identified in the LIBS spectra of uranium trioxide (bound in cellulose) and uranium bearing mineral ores. The uranium lines were classified into weak and resonant lines based on the signal-to-background ratio. Multivariate calibration models utilizing weak and resonant uranium lines were developed using an artificial neural network (back-propagation algorithm). The model using weak lines predicted the uranium concentration in the certified reference material with relative error prediction (REP) = 4.32% while that using resonant lines predicted with REP = 9.75%, thus demonstrating the robustness of chemometrics enabled LIBS. The calibration model using weak uranium lines predicted uranium concentrations in Kenya's uranium-bearing mineral ores between 103 and 837 ppm. Principal component analysis based on spectral feature selection successfully grouped the samples into their mineral mines (origin). Thus, LIBS in air under atmospheric pressure combined with chemometrics not only realizes direct, rapid, minimally invasive detection and quantification of trace levels of uranium in typical uranium bearing mineral ores but also aids in source attribution of these ores.