Creation of a database of radium distribution coefficients in soils

Abstract

Naturally occurring radionuclide materials (NORMs) are geological materials rich in indigenous radioactive elements. Some of these naturally occurring radionuclides (NOR) are present in many geological resources (e.g., mining ores and fossil fuels), whose exploitation might involve extracting NORs and exposing them into the surface. Then, they are more exposed to weathering and environmental factors that may increase their mobility. Consequently, understanding their mobility is important for risk assessment models involving radionuclides. A possible approach to study the mobility of a radionuclide is using the distribution coefficient (Kd). It describes the relation between the concentration of the element in the solid phase and in the soil solution at equilibrium. So, a high Kd implies that the radionuclide is mostly sorbed into the solid fraction of a soil, i.e., it has a lower mobility. Ideally, parametric models are to be proposed to predict the Kd as a function of soil properties. However, sorption involves many and complex mechanisms, which make it difficult to build such models. Alternatively, some variables may be used to propose Kd best estimates with a quantified variability for soil groups (which are to be made of intervals from those same variables). This was the case of this project, where a prospective analysis for radium (a NOR) was to be carried out. In order to do so, a Kd (Ra) database of different soils was expanded and reviewed with literature research, both to add new entries and to improve the soil and experimental characterizations. The inclusion criteria and data organization had to be rearranged to fulfill this goal. Once the dataset had been expanded, uni- and multivariant correlation analyses had to be carried out by creating new partial datasets for each variable, mainly to discern which soil properties were significant in reducing Kd (Ra) variability. Both pH and carbonate concentration (CO32-) in the soil solution proved significant in the univariant correlation analyses. Additionally, the multivariant analyses showed many more significant correlations, but most importantly two of them: CEC/(Ca+Mg)ss and (Caexch+Mgexch)/(Ca+Mg)ss. These las two combinations were so relevant to observe because they had been proven significant for other earth alkaline metals. Interestingly, some of the correlations found in this project are exceptionally good, which means that some of these variables could end up being used in establishing parametric models. Finally, Kd (Ra) best estimates for grouped soils according to the value of certain variables were proposed. These variables were pH, CO32-, and CEC/(Ca+Mg)ss. All these proposed best estimates presented a smaller variability than the Kd (Ra) best estimated of the overall database. Thus, grouping soils according to their own properties helped to reduce the variability of the Kd (Ra) best estimate