Overview of Algorithms for Using Particle Morphology in Pre-Detonation Nuclear Forensics

Tom Burr,Ian Schwerdt,Kari Sentz,Marianne Wilkerson,Luther Mcdonald

doi:10.3390/a14120340

Abstract

A major goal in pre-detonation nuclear forensics is to infer the processing conditions and/or facility type that produced radiological material. This review paper focuses on analyses of particle size, shape, texture (“morphology”) signatures that could provide information on the provenance of interdicted materials. For example, uranium ore concentrates (UOC or yellowcake) include ammonium diuranate (ADU), ammonium uranyl carbonate (AUC), sodium diuranate (SDU), magnesium diuranate (MDU), and others, each prepared using different salts to precipitate U from solution. Once precipitated, UOCs are often dried and calcined to remove adsorbed water. The products can be allowed to react further, forming uranium oxides UO3, U3O8, or UO2 powders, whose surface morphology can be indicative of precipitation and/or calcination conditions used in their production. This review paper describes statistical issues and approaches in using quantitative analyses of measurements such as particle size and shape to infer production conditions. Statistical topics include multivariate t tests (Hotelling’s T2), design of experiments, and several machine learning (ML) options including decision trees, learning vector quantization neural networks, mixture discriminant analysis, and approximate Bayesian computation (ABC). ABC is emphasized as an attractive option to include the effects of model uncertainty in the selected and fitted forward model used for inferring processing conditions.

Highlights

References [1,2,3,4,5,6,7] describe pre-detonation nuclear forensics goals
Results based on the 22 Morphological Analysis of Materials (MAMA) measurements [2] are given in Table 1 for nine machine learning (ML) options including decision trees, flexible discriminant analysis (FDA), mixture DA (MDA), linear DA (LDA), k-nearest neighbor, approximate Bayesian computation (ABC), learning vector quantization (LVQ), multivariate adaptive regression with splines (MARS), and support vector machines (SVM)
When all three pathways are in the data, the CCRs for decision trees, MDA, LDA, FDA, k-nearest neighbor, LVQ, SVM, ABC, and MARS, are 0.83, 0.76, 0.74, 0.74, 0.81, 0.75, 0.76, 0.87, and 0.87, respectively

Summary

Introduction

References [1,2,3,4,5,6,7] describe pre-detonation nuclear forensics goals. Baseline morphology and discussion of the technical production details for four precipitation conditions (ADU, AUC, SDU, and MDU) are provided in [1,2]. One related inference goal is to determine factors that impact particle morphology, such as calcination temperature, production pathway involving UO3, U3O8, or UO2, at 400 ◦C, 800 ◦C, and 510 ◦C (in H2), respectively, and impurities for a given precipitation condition. The analysis of surface morphology for particulate samples has been of great interest as a possible indicator of the synthetic conditions used to produce nuclear materials, such as uranium ore concentrates (UOCs). The technique of quantitative morphological analysis as applied to nuclear forensics is this paper’s focus

Methods

Results

Conclusion