Univariate and multivariate analysis to elucidate the soil properties governing americium sorption in soils

• Americium sorption–desorption behaviour was studied in contrasting soils.
• Americium was strongly and irreversibly sorbed in all soils.
• Organic matter, pH and the soil specific surface area controlled americium interaction.
• Models for predicting the Kd (Am) values in soils using soil properties were developed.

Americium (Am) solid–liquid distribution coefficients (Kd) and desorption percentages were quantified in 20 soils with contrasting edaphic properties by means of batch tests aimed at elucidating the mechanisms and soil properties governing Am sorption. The Kd (Am) values ranged from 103 to 105 L kg− 1, whereas the desorption percentages were always < 1.2%, thereby demonstrating a strong and irreversible sorption of Am. Univariate and multivariate data analyses were performed with the obtained sorption parameters to identify the soil properties controlling Am sorption behaviour and to propose Kd (Am) prediction models. Univariate Linear Regressions (ULR) and Multiple Linear Regressions (MLR) based on the use of original soil variables and a Principal Component Analysis (PCA) and Partial Least Squares Regression (PLS) based on latent variables revealed that Am sorption was controlled by organic matter, pH, the carbonate and amorphous iron content and the soil specific surface area. The MLR and PLS models explained up to 73% and 70% of the Kd (Am) variability, respectively; thus, they can be used to estimate Kd (Am) values directly from soil properties.