When a sample is analysed with neutron activation analysis (NAA) neutron self-shielding and gamma self-absorption affect the accuracy. Both effects become even more important when the mass of a sample analysed is changed from small (say, 1 g) to large (say 30 kg). Therefore, corrections have to be carried out. In this article only the correction method for neutron self-shielding is considered for a thermal neutron beam irradiating large homogeneous samples for prompt-gamma NAA (PGNAA). The correction method depends on the macroscopic scattering and absorption cross sections of the sample. To avoid doing experiments with samples with different macroscopic scattering and absorption cross sections, the Monte Carlo model MCNP is applied in the development of the correction method. The computational development of the method to determine these cross sections through flux monitoring outside the sample is described.
In order to correct for neutron self-shielding in large-sample prompt gamma NAA, a method has been developed to determine
the macroscopic scattering and absorption cross sections, i.e., Σa and Σs, using four Cu flux monitors placed around the sample. With Monte Carlo computations, the neutron densities throughout the
sample and the resulting and the corresponding self-shielding factor as calculated from the Σa and Σs as obtained through the Cu monitors were compared to the true values. The derived Σa and Σs were found to be sufficiently accurate as long as Σt = Σa + Σs was less than 0.6 cm−1 and Σs/Σt was greater than 0.1.
A Monte Carlo study was carried out to determine the influence of the effective scattering mass (Me) of the atoms on the neutron density profile inside and outside the sample illuminated by a thermal neutron beam as in large-sample
prompt-gamma neutron activation analysis (LS-PGNAA). From theory it is known that the spatial neutron density distribution
(n(r)) inside a large sample is not the same for atoms with the same macroscopic scattering and absorption cross-section (Σs and Σa) but different Me, due to anisotropic scattering at low Me. The probability of neutron absorption in the sample was found to be the same for materials with equal Σs and Σa but different Me, even though the neutron density distribution in the sample was found to change slightly. In view of typical sample, collimator
and detector dimensions, it is concluded that Me does not need to be taken into account in a correction method for neutron self-shielding in LS-PGNAA.
Authors:I. Degenaar, M. Blaauw, P. Bode, and J. de Goeij
A benchmark study was carried out to verify whether MCNP is useful in the design stage of a PGNAA facility for large samples
up to 1 m length and 0.15 m diameter, using a 2.54 cm diameter thermal neutron beam. For this facility neutron self-shielding
and gamma-attenuation correction methods have to be developed. The relative spatial neutron-density distributions within three
samples with different macroscopic scattering and absorption cross sections were studied in a comparison between an MCNP simulation
and an irradiation experiment using copper wires as neutron monitors. The neutron density in the sample was within statistical
agreement between experiment and simulation. Typically the relative spatial neutron-density distributions agreed to within
1%. Therefore, MCNP can be used in design studies for the development of a large sample PGNAA facility as specified.
Authors:M. Blaauw, I. Degenaar, C. Yonezawa, H. Matsue, and J. de Goeij
A method is proposed for the implementation of large-sample prompt-gamma neutron activation analysis (LS-PGNAA). The method
was tested with four different sample materials at the thermal PGNAA facility at JAERI, Japan. The macroscopic scattering
cross section (Σs) and absorption cross section (Σa) of the samples were determined by monitoring the neutron flux in four positions just outside the sample container. With
the Σs and Σa determined, the spatial neutron density distribution [n(r)] inside the sample material was derived. Taking n(r) and the gamma-ray self-absorption into account simultaneously, the effective geometric gamma-ray detection efficiency for
large samples as a function of gamma-ray energy was calculated. Taking silicon as test element, the concentrations found agreed
to within 7% with the known concentrations in the four sample materials examined, both when using relative standardization
and with absolute standardization.