Laser Raman Microspectrometric Assessment Of Uranium Forensics Signature In Aerosols Over A Model Nuclear Atmosphere

Abstract

Nuclear forensics relies on environmental sampling to detect undeclared nuclear activities. In the nuclear fuel cycle, mining and processing of uranium is accompanied by immobilization and translocation of uranium to the environment. Capability for direct and rapid analysis of micro-size particles such as aerosols in the environment is therefore a powerful tool to monitor undeclared nuclear activity. The strength of particle analysis is that it allows analyzing very small amounts of material exhibiting an undeclared nuclear forensic signature. Uranium bearing aerosol micro-particles sampled from e.g. a uranium mine or reactor atmosphere have unique inherent signatures that can be used to study the dynamics of detecting undeclared nuclear activities. For this work, typical uranium-specific Raman scatter bands were characterized using 532 nm and 785 nm lasers-based Raman microspectroscopy for different uranium molecules with different embedding anions (uranium trioxide, uranyl chloride, uranyl sulphate and uranyl nitrate). The Raman scatter bands varied in the range of (810 to 870) ± 15 cm-1 for both laser excitations. The uranium forensic signatures were used to characterize uranium bearing aerosol particulate matter. A multivariate calibration strategy using artificial neural network (both new feedforward and cascade correlation algorithms) was developed for quantification of trace uranium in aerosol particles sampled around Mrima hills which is a quasi-uranium mine. The validation of the analytical method was done using a synthetic membrane standard based on the IAEA-RGU-1 certified reference material and the relative error of prediction was found to be ≤ 10%. Depending on the sampling location, the concentration of uranium in the aerosols was found to range from (200-800 ppb) being more enriched in PM4.5 as compared to PM2.5 size fraction. This shows that most of the uranium was from the immediate environment as opposed to long range transport to the study area. In addition, principal component analysis was employed on both particulate sizes to explore the variability of their intrinsic chemical components. Using both the scores and loadings plots, the samples classified based on their sampling source fields. Whereas all PM2.5 grouped together with some PM4.5 that were sampled away from road sides, all other PM4.5 particles were grouped differently indicating that their source is different. The PCA also showed that most of PM2.5 were as a result of sea spray as shown by presence of chloride based uranium bands in the loadings plot. The heterogeneity of individual aerosol particles was characterized by overlap of spectral information (completely buried in the background) that could not be resolved by conventional Raman analysis. iv To yield information about the specific uranium species present in the particles as well as the uranium distribution within the particles, MCR-ALS was employed. The MCR-ALS gave results at an explained variance of 80.883 % and lack of fit of 9.56. Three distinct uranium scatter bands specific for uranium (814, 854 and 868 cm-1) were resolved as typical forensic signatures specific for uranium molecule in aerosol particles. The findings of this work are important in developing a spectral library that can be used by nuclear security authorities for environmentally sampled particulate matter.