Accuracy of neutron self-activation method with iodine-containing scintillators for quantifying ¹²⁸I generation using decay-fitting technique

We evaluated the accuracy of a self-activation method with iodine-containing scintillators in quantifying ¹²⁸I generation in an activation detector; the self-activation method was recently proposed for photo-neutron on-line measurements around X-ray radiotherapy machines. Here, we consider the accuracy of determining the initial count rate R₀, observed just after termination of neutron irradiation of the activation detector. The value R₀ is directly related to the amount of activity generated by incident neutrons; the detection efficiency of radiation emitted from the activity should be taken into account for such an evaluation. Decay curves of ¹²⁸I activity were numerically simulated by a computer program for various conditions including different initial count rates (R₀) and background rates (R_B), as well as counting statistical fluctuations. The data points sampled at minute intervals and integrated over the same period were fit by a non-linear least-squares fitting routine to obtain the value R₀ as a fitting parameter with an associated uncertainty. The corresponding background rate R_B was simultaneously calculated in the same fitting routine. Identical data sets were also evaluated by a well-known integration algorithm used for conventional activation methods and the results were compared with those of the proposed fitting method. When we fixed R_B = 500 cpm, the relative uncertainty σ_R0/R₀ ≤ 0.02 was achieved for R₀/R_B ≥ 20 with 20 data points from 1 min to 20 min following the termination of neutron irradiation used in the fitting; σ_R0/R₀ ≤ 0.01 was achieved for R₀/R_B ≥ 50 with the same data points. Reasonable relative uncertainties to evaluate initial count rates were reached by the decay-fitting method using practically realistic sampling numbers. These results clarified the theoretical limits of the fitting method. The integration method was found to be potentially vulnerable to short-term variations in background levels, especially instantaneous contaminations by spike-like noise. The fitting method easily detects and removes such spike-like noise.