SUMMARY

This dissertation investigates the counting efficiency calibration of a gas flow proportional counter with beta-particle emitters in order to 1) estimate the uncertainty and sources of variability in the calibration process, 2) construct a Monte Carlo model supplementing empirical measurements, and 3) evaluate parameters and processes of the simulation model that affect the representation of beta-particle interactions within the detector system. Monte Carlo simulation results by the MCNP5 code were compared with measured counting efficiencies as a function of sample thickness for 14C, 89Sr, 90Sr, and 90Y. The samples were precipitated as SrCO3 with areal thicknesses from 3 to 33 mg cm-2 , mounted on membrane filters, and counted on a low background gas flow proportional counter. The estimated fractional standard deviation was 2–4% (except 6% for 14C) for efficiency measurements of the radionuclides, and 1–2% for simulations. The curves of simulated counting efficiency vs. sample mass thickness agreed within 3% of the curves of best fit drawn through the 25 - 49 measured points for each of the four radionuclides. Contributions from this research include development of uncertainty budgets for the analytical processes; evaluation of alternative methods for determining chemical yield critical to the measurement process; correcting a bias found in the MCNP normalization scheme for histogram distributions and affecting the sampling of beta-particle spectra; clarifying the interpretation of the commonly used ICRU beta-particle spectra for use by MCNP; and evaluation of instrument parameters as applied to the simulation model to obtain estimates of the counting efficiency from simulated pulse height tallies.