There is an increasing need in forensics as a whole for quantitative methodologies capable of driving legally defensible data interpretations. In our efforts, initial studies on morphology analysis of nuclear materials, using the morphological analysis for materials (MAMA) software package developed at LANL, has provided preliminary quantitation of particle features measured in scanning electron microscopy images and is here, for the first time, combined with other parameters from material characterization (e.g. trace elements, isotopics, etc.) to facilitate multi-variate statistical analysis/differentiation. Five uranium oxides from two international round robin studies were analyzed for full chemistry and three of those samples were further subjected to morphological analysis via MAMA processing. Statistical processing, using Eigenvector Research Solo Software (commercial, off the shelf software), was performed, using (1) trace element and isotopic measurements, (2) independent MAMA morphology values, and (3) a combined approach incorporating both of these former data sets simultaneously. Principal components analysis indicated that samples from the two round robin studies could be distinguished by chemistry data alone, however the multiple samples within each study overlapped. Single particle analysis from MAMA was less effective at providing separation between samples, but averaged values were able to resolve the round robin studies. When combined with the MAMA morphology data, the three individual samples could be resolved from one another within a 95% confidence level. The initial success of the statistical analysis processing of uranium oxides indicates that using principal components analysis, along with other techniques such as hierarchical clustering analysis and partial least squares discriminant analysis can distinguish nuclear materials based on their chemical and morphological signatures and may provide context for manufacturer or origin of the material in question.<br/>This abstract is LA-UR-17-23272.

Multi-variate statistical analysis enhancing preliminary morphological signature development strategies for nuclear forensic-related materials

254th National Meeting (2017)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.