The Use of Laboratory Adsorption Data and Models to Predict Radionuclide Releases from a Geological Repository: a Brief History

Abstract

ABSTRACTRadionuclide (RN) adsorption has long been recognized as important to assure the isolation of nuclear wastes in a geological repository [1]. Laboratory measured RN adsorption data have generally been expressed as distribution coefficient (Kd) values or adsorption isotherms. The proper application of these models is to site conditions nearly identical to those used in the laboratory adsorption experiments. This has required that multiple Kd's and isotherms be determined in a wide range of experiments designed to bracket expected repository conditions.The surface complexation (SC) adsorption models were introduced in the late 1970's. The best known of these models incorporate electrical double layer (EDL) theory [2]. Their use requires that the water chemistry and surface properties of adsorbing rocks and minerals be fully characterized. Adsorption is then studied as reactions involving specific aqueous RN species (often complexes) and specific surface sites. Because the SC models are relatively mechanistic, they may allow extrapolation of adsorption results to repository conditions that lie outside the limited experimental range used to parameterize a given model. Turner [3] has shown that the diffuse layer model (the simplest SC model) fits a wide range of RN adsorption data as well as the more complex models. Others have suggested ways to generalize and estimate SC model parameters for a variety of minerals, rocks and engineered materials (cf. [4,5,6,7,8,9,10,11,12]. Degueldre and Werlni [12] and Degueldre et al. [13] have proposed a simplified SC model for RN adsorption that avoids EDL theory, in which the adsorption of RN species is estimated from linear free energy relationships.It is appropriate to ask how accurately RN adsorption behavior must be known or understood for total system performance analysis (TSPA). In most geological settings now being considered for repository development globally, it may suffice to select bounding Kdvalues for the different rock types (cf. [14,15]). Use of the SC models to describe RN adsorption can provide us with increased confidence that minimum Kd's and the distribution of Kdvalues we might propose for TSPA are in fact conservative.