Application of Surface Complexation Modeling to Describe Uranium (VI) Adsorption and Retardation at the Uranium Mill Tailings Site at Naturita, Colorado (NUREG/CR-6820)
On this page:
Download complete document
Manuscript Completed: April 2003
Date Published: December 2003
James A. Davis and Gary P. Curtis
U.S. Geological Survey
345 Middlefield Road, Mail Stop 465
Menlo Park, California 94025
J.D. Randall, NRC Project Manager
Division of Systems Analysis and Regulatory Effectiveness
Office of Nuclear Regulatory Research
U.S. Nuclear Regulatory Commission
Washington, DC 20555-0001
NRC Job Code W6813
The objective of this study was to demonstrate a surface complexation modeling approach at the field scale for estimating Kd values and the retardation of a sorbing radionuclide with complex aqueous chemistry. The Uranium Mill Tailings Remediation Act (UMTRA) site near Naturita, Colorado, was chosen for study, because it had a well-developed and definable uranium (VI) plume in a shallow alluvial aquifer and had spatially variant chemical conditions that we believed would be important in influencing U(VI) transport and retardation. It was shown in laboratory batch and column experiments with Naturita sediments that the adsorption and retardation of U(VI) by the Naturita sediments was strongly influenced by the dissolved carbonate concentration (alkalinity). A generalized composite surface complexation model (GC-SCM) was developed for the Naturita aquifer background sediments (NABS) based on fitting batch U(VI) adsorption data. With only two surface reactions (four surface species), the GC-SCM without electrical double-layer terms was able to accurately simulate Kd values for U(VI) adsorption on the Naturita aquifer sediments over the observed range of pH and dissolved carbonate and U(VI) concentrations. For the range of Naturita aquifer chemical conditions, alkalinity was more important than either variable pH or U(VI) concentration in influencing U(VI) mobility. Kd values ranged from 0.29 to 22 mL/g when calculated for all Naturita groundwater analyses using the SCM. Low Kd values were associated with portions of the U(VI) groundwater plume containing high concentrations of dissolved U(VI) and alkalinity. Higher Kd values were associated with low concentrations of dissolved U(VI) and alkalinity.
In addition to the common experimental technique of batch adsorption studies, methods were investigated to estimate U(VI) Kd values in the field. Such methods are needed for: 1) validation of SCM model parameters for transport simulations within performance assessment (PA) models, and 2) estimation of initial conditions for adsorbed radionuclides for transport simulations describing previously contaminated sites. It was shown that isotopic exchange and desorption extraction methods can be an important part of a field characterization and modeling program. GC-SCM-predicted U(VI) Kd values generally agreed to within a factor of 2 to 3 with experimental estimates of the Kd values of the U-contaminated sediments. This agreement with the experimental determinations of sorbed U(VI) in the contaminated portion of the Naturita alluvial aquifer provides confidence in the predictive capability of the GC-SCM, which was developed from data with uncontaminated Naturita sediments. Another approach used to validate the GC-SCM was the determination of in-situ Kd values by suspending NABS samples in wells with U-contaminated groundwater for periods of time ranging from 3–15 months. In-situ (field) Kd values were calculated from groundwater measurements of dissolved U(VI) and U extracted from the suspended sediment samples. The in-situ Kd values in 17 wells ranged from 0.5 to 12 mL/g, with the Kd values decreasing with increasing alkalinity. There was close agreement between these measured in-situ Kd values and model-predicted Kd values using the GC-SCM.
Transport simulations conducted for the field scale demonstrated the importance of using the SCM to describe U(VI) adsorption rather than a constant-Kd modeling approach. A major conclusion from the transport simulations was that PA modelers must recognize not only that variable chemical conditions can cause a range of K Kd values to be observed, but also that the spatial distribution of Kd values within that range is not likely to be a random function or a normal distribution. In plumes with chemical gradients, the spatial distribution of Kd values can be quite complex and be characterized by significant spatial character. The linkage of traditional contaminant-transport models and reactive-contaminant-transport models to models for dose assessment in PA was investigated. It was shown that a constant-Kd modeling approach is not always conservative compared to using an SCM to described radionuclide retardation in PA. Transport simulations with a rate-controlled U(VI) adsorption model agreed well with those that used the local chemical equilibrium approximation. The simulations also showed that predicted U(VI) transport was nearly identical whether or not surface charge was explicitly considered within the GC-SCM.
The challenge in applying the surface complexation concept in the environment is to simplify the SCM, such that predicted adsorption is still calculated with mass laws that are coupled with aqueous speciation, while lumping parameters that are difficult to characterize in the environment in with other parameters. In order to be applied by solute transport modelers and within PA applications, the complexity of the adsorption model needs to be balanced with the goal of using the simplest model possible that is consistent with observed data. This can be achieved with the semi-empirical, site-binding GC modeling approach used in this report and previously demonstrated for modeling Zn retardation in a sand and gravel aquifer (Kent et al., 2000). The GC-SCM is a compromise between the simple constant-Kd approach and more complex SCM that are difficult to apply to the environment at present. Historically, solute transport modelers have lacked the necessary expertise to apply the SCM modeling approach and many have believed that the SCM approach is too complex to be applied. While it is true that the most complex SCM are too difficult to apply at present, it is demonstrated in this study that the GC modeling approach can be easily applied to simulations of radionuclide transport at the field scale and included with PA modeling. The GC modeling approach is preferable to completely empirical approaches, such as the constant-Kd model or adsorption isotherms, because the important linkage between surface and aqueous species (and associated thermodynamic data) is retained in the modeling through the coupling of mass action equations. This linkage also provides a framework for conducting uncertainty analyses that is based on process level parameters rather than on ranges of Kd values that result from lumping together multiple processes. In the authors' opinion, the current operational paradigm that employs constant-Kd values to describe the retardation of radionuclides at the field scale introduces more uncertainty than is necessary. This uncertainty could be reduced and more completely understood in the future with the use of the GC-SCM approach.