STATISTICAL PROPERTIES OF A PEAK-SEARCH ALGORITHM FOR GAMMA-RAY SPECTROMETRY AS RELATED TO CURRIE DETECTION LIMITS

Most peak-search algorithms for gamma-ray spectrometry are based on digital filters, and most of them resemble the Routti-Prussin filter. The statistical properties of this filter must be known to determine detection limits correctly. Its properties were therefore studied as a function of peak width, background level and differential nonlinearity of the ADC using simulated spectra, and they were found not to be as expected from simple statistical considerations. These deviations were modelled, resulting in formulas yielding Currie's critical level L(C) and detection limit L(D) as a function of peak width and background level.