Analysis of a small sample geometry for concurrent identification and quantification of mixed-nuclide samples

Abstract

An easier way to perform a complete radionuclide analysis was investigated for this thesis. Complex chemical separations are not always possible or practical in preparation for counting a sample, especially a very small sample, on several different systems. Carrying out complex procedures may offer many opportunities for errors to be introduced into the analysis. Using the same sample and the same geometry, a technique was devised to identify and quantify several different nuclides in a single, small sample, each emitting a different type of radiation. High purity germanium detectors (HPGe) and liquid scintillation counters (LSC) can be used to quantify activities of several known sources. This can be done without buying more or expensive equipment and these machines are already available in most radiation counting laboratories. Several radionuclides (Co-60, Cs-137, Sr-90, and Fe-55) were added to a single, small (1- ml) sample. This sample was analyzed on the HPGe and LSC and the information gathered with both instruments was used to quantify the radionuclides. Efficiencies of each machine were calculated and used to convert counts in a region of interest on the acquired LSC spectrum to sample activity and activity into counts. Results had varied accuracy, however, were found to be useful in certain situations. Small errors were associated with the gamma emitting radionuclides while larger errors were found with the other radionuclides. The large uncertainty of the Cs-137 activity probably led to a large error in determining activities of other nuclides. It was found though several follow up experiments that when the activities are similar the analysis technique works better. This technique was found to be best utilized in a lab setting when the sample has several types of radiation, the radionuclide composition is known, and just the activities are needed. Such a technique may have limited use with unknown samples: statistical uncertainty is the limitation.