Reconstruction of Material Elemental Composition Using Fast Neutron Resonance Radiography

Abstract Fast neutron resonance radiography (FNRR) is an imaging method that exploits characteristic cross-section structures (peaks and troughs) of certainelements in the energy-range of 1-10 MeV to identify materials in a large volume object. In FNRR, the neutron energy spectrum transmitted through an object carries information about the elemental composition of thatobject. The principal elements present in most explosives are: carbon, oxygen, nitrogen andhydrogen. Explosives are characterized by high fractions of nitrogen and oxygen as well as low fractions of carbon and hydrogencompared to benign materials. Detection of explosives in cargo employing FNRRis based on determination of the local areal densities of these four elements and their ratios. In our measurements, the transmission spectrum is usually divided in 100 - 500 energy bins, representing 100 - 500 linear equations containing four unknown areal densities of HCNO. This is an overdetermined problem, which allows us to derive not only the fourexpectation values of their areal densitiesbut theirprobability distribution as well. For this purpose, a model was formulated and implemented within a software package which performs Bayesian analysis of complex statistical models using Markov chain Monte-Carlo (MCMC). This model was tested successfully both on simulated and experimental data. This work will describe the model and the outcome of elemental ratios reconstruction for several materials from experimental data.