Sorption Modeling of Strontium, Plutonium, Uranium, and Neptunium Adsorption on Monosodium Titanate

Abstract:

We examined the ability of various equilibrium isotherms to replicate the available data for the adsorption of strontium (Sr), plutonium (Pu), uranium (U), and neptunium (Np) on monosodium titanate (MST) during the treatment of simulated and actual Savannah River Site high-level waste. The data come from numerous experimental studies conducted between 1999 and 2002. The analysis considered 29 isotherm models from the literature. As part of this study, we developed a general method for selecting the best isotherm equation. The selection criteria for rating the isotherm equations considered the relative error in predicting the experimental data, the complexity of the mathematical expressions, the thermodynamic validity of the expressions, and statistical significance for the expressions.

The Fowler Guggenheim-Jovanovic Freundlich (FG-JF), the Fowler Guggenheim-Langmuir Freundlich (FG-LF) and the Dubinin-Astashov (DA) isotherms each reliably predicted the actinide and Sr adsorption on MST. The first two models describe the adsorption process by single layer formation and lateral interactions between adsorbed sorbates, while the DA model assumes volume filling of micropores (by osmotic pressure difference). These two mechanisms include mutually exclusive assumptions. However, we cannot determine which model best represents the various adsorption mechanisms on MST. Based on our analysis, the DA model predicted the data well. The DA model assumes that an initial sorption layer forms after which networking begins in the pore spaces, filling the volume by a second mechanism. If this mechanism occurs in MST, as the experimental data suggest, then we expect all the empty and closed spaces of MST to contain actinides and Sr when saturated. Prior microstructure analyses determined that the MST surface is best described as heterogeneous (i.e., a semicrystalline outer layer on an amorphous core) or composite material for adsorption. Therefore, we expect the empty spaces (of nanometer size) between the crystalline units in the fibrous material to provide sorption area for the actinides and Sr. Additional conclusions from this study follow.

Since each of the three models work reliably, we recommend use of the computationally simplest model as the primary tool until future work can differentiate between the two mechanisms. The DA model possesses a simpler mathematical form with fewer parameters and operations.

The experimental data for actual and simulated wastes generally showed consistent agreement. However, the data sets do include considerable variance from a number of causes including the following:

•	The Pu sorption data appear most consistent (e.g., between actual and simulated waste) and most easily predicted. Since Pu removal efficiency proves most important for the process design efforts, this consistency of the data proves especially beneficial.
•	Extremely high mass loadings of U on MST result in multilayer sorption behavior and divergence from classical single monolayer isotherm forms. Prior X-ray studies demonstrate that U begins to network, or form dimers, which agrees with this interpretation. This U behavior also shows a complex interaction, and a direct correlation, with sorption data for the other radionuclides. We believe these data suggest nucléation (e.g., precipitation) of the actinides in the micropore space for both Np and Pu. For Sr, the high U loadings appear to inhibit the sorption of Sr.
•	Nearly all the solutions contained U as the radionuclide with the highest mass concentration. These data show the widest variance.
•	The composite data set indicates a notable variance in sorption for different batches of MST. The sorption of Sr with different batches of MST shows the largest variance among the four radionuclides for different batches of MST. This variance remains a relatively unexplored aspect of the process design.
•	Similarly, the experimental data included a wide variety of solution compositions. As such, the mathematical expressions implicitly account for variances in solution chemistry typical of that anticipated within the Salt Waste Processing Facility and Actinide Removal Process. The reader must consider the ranges of these concentrations when applying the expressions.
•	Increasing temperature decreases Sr, Pu, and, to a lesser extent, U sorption on MST. The opposite effect occurs with Np. This temperature variance further suggests a nucléation behavior for Np.
•	Nearly all the data used in developing the sorption models came from experiments using solutions with all the principle radionuclides of interest present simultaneously. We modeled the data without invoking competition between the actinides and Sr despite the large concentrations of both U and Np. Since the model does not explicitly invoke competition, the optimized parameters implicitly carry the impact of interaction within the concentration ranges of the original data. Hence, extrapolation of the models to concentrations markedly outside those ranges may result in poorer predictive ability.