Statistics of the LS-detector in the case of low counting efficiency

In the case of high-quenched 3H and 63Ni sources a distinct incompatibility of theoretical and experimental detection efficiency in the triple liquid scintillation (LS) detector was observed (Appl. Radiat. Isot. 52 (2000) 643). The authors concluded, that the Poisson distribution does not properly describe the detection process, when less than one photoelectron is expected in one of the photomultipliers. Application of other distributions of photoelectrons, e.g. binomial, does not solve the observable problem of incompatibility. Measurements of a set of 55Fe sources have been performed with phototubes defocusing and grey filters for decreasing the counting efficiency of the TDCR detector. Differences between counting results of the 55Fe source and the light emitting diode (LED), simulating the scintillation source, excited by a pulse generator have been observed. Various distributions (Poisson, binomial and Polya) were used for the determination of the theoretical counting efficiency in both cases. The Poisson law gave a good result in the case of the LED but the Polya law had to be applied in the case of 55Fe. The results were independent of the scintillator volume. It seems that the validity of the Polya law in the case of LS-sources is related to the scintillator itself. Measurements of the 3H solution confirmed that conclusion.