Prediction Models for Radiological Characterization of Natural Aggregates Based on Chemical Composition and Mineralogy.

Abstract

The radiological characterization of aggregates used in construction materials is essential to determine their suitability from a radiological protection perspective and to ensure their safety for health and the environment. While the activity concentrations of radionuclides present in construction materials are typically determined using gamma spectrometry, an alternative approach involves the development of statistical methods and predictive models derived from the chemical composition of the material. A total of 39 aggregates used in construction of various types (siliceous, carbonatic, volcanic, and granitic) have been analyzed, correlating their chemical compositions obtained through X-ray fluorescence (XRF) with the activity concentrations of natural radionuclides measured via gamma spectrometry using principal component analysis (PCA). The results obtained allowed for the observation of an inversely proportional relationship between the chemical composition of the grouping of siliceous and carbonatic aggregates and the content of radionuclides. However, the set of granitic aggregates showed a strong correlation with the natural radioactive series of uranium, thorium, and ⁴⁰K. Conversely, the radionuclide content of volcanic aggregates was independent of their chemical composition. The results obtained from the PCA facilitated the development of different models using multiple regression analysis. The chemical parameters obtained in the proposed models were related to the typical mineralogy in each grouping, ranging from primary minerals such as feldspars to accessory minerals such as anatase, apatite, and pyrolusite. Finally, the models were validated using independent samples from those used to determine the models, achieving RSD (%) values ≤ 30% in 50% of the activity concentrations of ²²⁶Ra, ²³²Th(²¹²Pb), and ⁴⁰K, as well as the estimated ACI.