Sensitivity and uncertainty analysis of the simulation of 123I and 54Mn decay in liquid scintillation vials.

General purpose, widely distributed Monte Carlo simulations for radiation transport have found many applications in the fields of radionuclide metrology, detector research and nuclear medicine. A full evaluation of the uncertainties due to the simulation of physical quantities and correction factors is required for model validation, uncertainty analysis and quality assurance, yet this evaluation is rarely explicitly carried out. The decay of the most important gamma and x-ray emissions of 123I and 54Mn in the ANSTO Triple-to-Double Coincidence Ratio (TDCR) liquid scintillation detector have been simulated using the radiation transport code Geant4. The energy deposition spectrum in the scintillant and the interaction probability for each emission has been determined. Both of these calculated quantities are required for an absolute activity measurement of these nuclides using the TDCR method. A sensitivity and uncertainty analysis of the simulation model with respect to these output parameters has been carried out and is presented here. The uncertainty in the simulation results due to the input parameter uncertainties was found to be several times larger than the statistical uncertainty component for a typical number of simulated decay events. The scintillant volume was the most sensitive input parameter to produce changes in the output quantities studied, indicating that the minimization of the uncertainty of this parameter would be most beneficial for reducing the uncertainty of the simulation outputs. The scintillant density and composition were also important, which is significant as the uncertainties associated with these input parameters are generally not well known. Estimates of the relative uncertainty associated with the simulation outputs due to the combined stochastic and input uncertainties are provided.Slovak Institute of Metrolog