Sorption of Cs, Ni, Ac(III), Tc(IV), Zr, U(IV) and Se(IV) on MX 80 bentonite: an experimental approach to assess model uncertainty

Full text of publication follows: Various models are proposed in the literature to describe radionuclide sorption on bentonite in a mechanistic, thermodynamic manner: multi site surface complexation with or without electrical double layer representation and single or multi site ion exchange. Uncertainties exist as well in the parametrisation strategies. In order to provide a coherent modelling approach and to assess the impact of model uncertainty, a series of sorption experiments were initiated for Cs, Ni, Ac, Tc(IV), Zr, U(IV), and Se(IV) varying m/V, pH, pCO2 and sulfate content within the concentration range of groundwater compositions expected in the deep underground research laboratory of ANDRA located in the communes of Bure and Saudron. Prior to the sorption tests, the site occupancy of the bentonite was equilibrated with the groundwater by water exchange cycles, until water composition was not anymore changed by the bentonite. Time dependent potentiometric titrations allowed assessing calcite and montmorillonite dissolution rates. Radionuclide oxidation states were analysed spectroscopically. Oxidation state stability was assured by performing the experiments in an inert gas box. Equilibrium times of radionuclides in the bentonite/water system were about 2 days. Potential radionuclide precipitation and colloid formation were assessed by ultrafiltration. The Kd values of tetravalent elements Tc, Zr and U were in the range of 10000 mL/g. Maximum cation adsorption is typically observed close to the zero point of charge. Two modelling approaches were used: a surface complexation model coupled to a simple ion exchange and a full-scale multi-site ion exchange model, both models being coupled to the same radionuclide solution chemistry model. The fit to the data did not allow assessing model validity, since both models described equally well the series of experimental observations. The models provide a coherent description as long as they are used to predict bentonite performance inside the parameter range used for model calibration. An attempt was made to limit the quantity of sites because the models become contingent if the number of sites and competitive reactions is increased. If the effect of the electrical double layer on the surface complexation constants is ignored these constants and the corresponding site densities can directly be used to derive corresponding exchange constants. However, much stronger effects are expected if the inclusion of the electrical double layer cannot be avoided as for example in the application to transport calculations. (authors)