Improving the control of systematic uncertainties in precision measurements of radionuclide half-life

Many experiments designed to precisely determine the half-life of a radionuclide employ a long lived reference source to help determine the impact on the data of any systematic variation in the detector and associated electronics. The half-life of the radionuclide of interest is determined from the ratio of its decay rate data to the decay rate data from the reference source. This correction procedure assumes that any underlying systematic affects the data and reference measurements in exactly the same way. In this paper we show that when some systematic effects affect the two differently, the ratio procedure can leave artifacts in the corrected data that can compromise an unbiased and precise assessment of the radionuclide half-life. We describe two methods that can help overcome this problem. We also describe several statistical tests that help determine which effects may underlie systematic variations in the data. We discuss an illustrative example based on previously published (32)Si and (36)Cl data recorded by an experiment at Brookhaven National Laboratory. We correct the data for systematic variation related to climate variation and estimate the (32)Si half-life to be T1/2=171.8±1.8. The reduction in uncertainty in the (32)Si half-life, relative to the previous estimate based upon this data, is equivalent to that which would be achieved through increasing the size of the data set by almost 3.5 times.