Trapping radionuclides in geologic media3 July 2003
The Nuclear Energy Agency (NEA) has held five workshops on radionuclide migration in geological media. It discussed approaches to acquiring field data and testing and modelling flow and transport of radionuclides in geological formations.
GEOTRAP is the OECD Nuclear Agency Agency's (NEA's) project on radionuclide migration in geologic heterogeneous media. It provides a forum to exchange information on field data and on testing and modelling the transport of radionuclides in actual geological formations and is a basis for the site characterisation, evaluation and safety assessment of deep repository systems. The project has the following themes:
• Field tracer transport experiments; design, modelling, interpretation and role in the prediction of radionuclide migration.
• The basis for modelling the effects of spatial variability on radionuclide migration.
• The characterisation of water-conducting features and their representation in models of radionuclide migration.
• Confidence in models of radionuclide transport for site-specific performance assessment.
• Geological evidence and theoretical bases for radionuclide retention processes in heterogeneous media.
Deep geologic repositories are sited and designed such that most or all radionuclides remain within the repository and the surrounding geosphere for prolonged periods. Thus, for most radionuclides, substantial radioactive decay is expected to occur before migration processes can convey them to the biosphere. Processes that prevent or delay the migration of radionuclides are called retention processes. Retention processes that operate in the geosphere were the focus of the fifth GEOTRAP workshop.
The migration of radionuclides in solution through the geosphere may occur by advection and by diffusion. Off-diagonal Onsager processes are in general negligible. Radio-nuclides may also be sorbed or incorporated into the structure of colloids, which themselves migrate by advection and diffusion. The scope of the workshop encompassed a number of processes that may prevent or delay radionuclide migration and thus fall under the category of retention processes.
Retention processes can be divided into those that result in retardation and those that result in immobilisation. These processes are affected both by the properties of the radionuclides themselves and by the properties of the geological medium, and are favoured by:
• The presence of stagnant, or very slowly moving groundwater in the host rock around the repository, including zones of stagnant water adjacent to zones containing flowing water (such as fractures) into which matrix diffusion can occur.
• Groundwater and mineralogical compositions that favour geochemical interactions of radionuclides at mineral surfaces.
• The accessibility of sorbing surfaces (such as by diffusion from fractures).
• A porous structure that favours the filtration of radionuclide-bearing colloids.
The scope of the fifth GEOTRAP workshop explicitly excluded repository-induced effects on retention processes in the geosphere (for example, effects related to excavation-disturbed zones, high pH or organic plumes, thermal impact of the emplaced wastes). Long-term natural changes in the retention-relevant characteristics of the geosphere were, however, included in the scope.
Importance of retention processes
The importance of retention processes is recognised in other areas, such as mining and the management of non-radioactive hazardous wastes. There are different factors that affect retention processes at different types of site. For example, there will be a higher concentration of organic materials within and around landfill sites. The fact that the same processes are recognised to operate in systems or applications unconnected to radioactive waste disposal, however, means that there is a large body of expertise that radioactive waste disposal projects might benefit from.
The relative importance of different retention processes is site-specific. For example, colloid-related retention processes are of limited importance for many rocks considered as host formations because of, for example, low observed concentrations of groundwater colloids or the low mobility of colloids in some dense clays.
A radionuclide that is sorbed is reversibly attached to a solid surface by one of a range of mechanisms, including adsorption (an accumulation of matter at the interface between a solution and a solid adsorbent) and ion exchange (in which counter-ions accumulate at the interface of a solid adsorbent with permanent structural change).
These processes have been widely observed in laboratory and field experiments, as well as in natural systems. The nature of surface species and the stoichiometry of surface reactions in complex geological environments are becoming better understood by high-resolution surface analytical techniques.
At least three types of models have been developed for describing radionuclide sorption at mineral surfaces at which equilibrium exists between aqueous and sorbed phases:
• Empirical partitioning relationships, which are widely used in performance assessment, the most simple of which use the concept of a distribution coefficient, Kd.
• Mechanistic models, which describe the details of chemical species formation at mineral surfaces using thermodynamic formalisms and the surface complexation concept to describe absorption.
• Semi-empirical site-binding models, which use concepts from the first two model types.
Models of sorption kinetics are less well developed and take the form of empirical relationships.
Mechanistic models have been well developed for simple mineral/water interfaces, and are being improved for more complex geological situations. Models are subject to uncertainties associated with mineral/organic coatings and electric double-layer properties. Models are mainly used to develop an understanding of the effects of varying groundwater conditions. Semi-empirical site-binding models have not been widely used to date, but may represent a good compromise in terms of data collection needs and the need to predict radionuclide distributions over a range of chemical conditions.
The use of empirical partitioning relationships to represent sorption continues to be widespread in performance assessments and there seems little desire among performance assessors or regulators to discontinue their use. Some performance assessments have employed non-linear sorption isotherms. Most, however, emply the Kd approach, which implicitly assumes rapid, reversible, and linear sorption.
Another approach is to employ semi-empirical functions to represent dependence of Kd on geochemical conditions such as pH. Mechanistic models are regarded as too complex and subject to uncertainty for direct inclusion in performance assessment models, but are valuable for justifying the selection of Kd values.
Matrix diffusion the diffusion of radionuclides between regions of flowing water, such as fractures, and connected regions of stagnant water, such as pore spaces in the fracture wallrock may be increased by surface diffusion or restricted by anion exclusion.
Numerous field observations offer mainly generic evidence that matrix diffusion occurs in fractured rock systems. Several methods exist, or have been suggested, to characterise matrix pores. Field tracer transport experiments also provide site-specific evidence of matrix diffusion, and can be used to test conceptual models for the distribution and accessibility of pore spaces. Interpretation may be made more complicated by matrix heterogeneity. The evidence provided by field tracer transport experiments may relate to only a part of the total pore space that is of interest to performance assessment.
The key issue to be addressed in modelling matrix diffusion in geologic media is matrix heterogeneity. Approaches include the use of probability distribution of matrix residence times and the explicit modelling of different rock matrix zones, assigning different characteristics to each. Further development of process-level models is likely to require better understanding of the geometry and connectivity of matrix pores. The degree to which matrix diffusion retards radionuclide transport also depends on flow conditions and may be strongly influenced by heterogeneous groundwater flow.
Matrix heterogeneity is often treated in a simplified manner in performance assessment models, due to uncertainties in the characterisation of the matrix. Some programmes have developed and applied performance assessment models that allow explicit modelling of different rock matrix zones. Surface diffusion is generally omitted from models. Anion exclusion is sometimes included via parameter selection.
Colloid-facilitated radionuclide transport is the association of radionuclides with either natural groundwater colloids or colloids arising from the presence of a repository, and the subsequent transport of radionuclide-bearing colloids through the geosphere. The distances over which transport occurs depend, in part, on colloid stability and the rate of filtration.
Colloid-facilitated radionuclide transport is generally of concern in performance assessment only for more radiotoxic nuclides. In addition, colloids must also be able to compete against immobile-phase solids in the sorption of radionuclides, and must be transported over considerable distances to be of concern. Radionuclide-bearing colloids arising from processes associated with a repository may be of most concern in performance assessments, since natural colloid concentrations are often low. Colloid generation, deposition, radionuclide uptake and transport are understood in general terms, but are subject to numerous uncertainties. With respect to transport, it has not, in general, been possible to demonstrate that radionuclide-bearing colloids will be immobile in natural systems, other than some dense clays. The role of natural organic matter colloids in radionuclide transport has been relatively little studied, but may, in some instances, enhance the transport of radionuclides through subsurface systems.
A specific class of colloids dominates many systems, and process-level models have to be tailored for application to specific systems. Many process-level models have been developed, but are subject to numerous uncertainties. Recent advances in understanding the role of geochemical heterogeneity in colloid transport and release suggest promising avenues for understanding and modelling colloid transport in real systems. There is a need for more high-quality datasets from relevant systems in order to test, and therefore further refine, colloid-facilitated radionuclide transport models and a need to develop better methods to upscale process models developed for laboratory systems to the field scale.
Some performance assessments have explicitly included colloid-facilitated radionuclide transport in geosphere transport models, although such models are usually highly simplified due to the absence of relevant site-specific data, and the complexity and consequent uncertainties associated with the relevant processes. Other assessments have used a variety of lines of reasoning to argue that the process is unimportant for the particular system under consideration.
Immobilisation of radionuclides in the bulk of a solid may, depending on the nature and rates of any remobilisation processes, be a more effective retention mechanism than sorption at surfaces. Highly effective immobilisation of ions in mineral phases in diverse environmental settings is widely observed in nature, although conditions under which immobilisation occur often differ significantly from those of interest to performance assessment. Laboratory and modelling studies, often integrated with analogue research, also support the development of the basic understanding of immobilisation processes. Immobilisation by precipitation and by co-precipitation in carbonates and Fe- and Mn-oxyhydroxides are of particular importance in many natural analogues, and are also likely to be important in many repository systems. Short-term laboratory measurements on bulk systems often do not allow slow immobilisation processes to be resolved and fundamental understanding is, in many cases, poor. High-resolution techniques offer the prospect of improved understanding in this area. The incorporation of radionuclides in a solid, can, under some circumstances and over long timescales, be reversible. The dissolution or erosion of solids incorporating immobilised radionuclides may occur as physicochemical condition change over the long timescales associated with performance assessment.
The difficulties encountered in the mechanistic modelling of sorption also apply to the application of these models to immobilisation processes. In particular, the lack of high-resolution techniques has been an obstacle to process-model development. Nevertheless, current knowledge is adequate to perform simplified calculations of the likely magnitude of immobilisation processes in some systems, so that such processes can be compared in terms of their possible effectiveness with reversible retardation processes.
Performance assessment modelling
Immobilisation processes have not so far been taken into account in performance assessments in a quantitative manner. This omission is generally the result of the absence of relevant site-specific data and uncertainties associated with the slowness and complexity of the processes and justified on the grounds of conservatism. The uncertainties are such that the prospects of including immobilisation processes in performance assessment models used to demonstrate compliance with regulatory guidelines in the near future appear low. There are good prospects for some improvement in the understanding of these processes, which will at least improve qualitative arguments for the long-term retention of many safety-relevant radionuclides in the geosphere.
Based on the workshop discussions and presentations, a number of recommendations emerged.
Database of observations
It will be useful to compile a database of geological phenomena and laboratory observations that support or provide information on immobilisation processes in geological environments, and also information on processes that can lead to remobilisation. Such a database could be used to provide qualitative arguments to support the effectiveness of the geosphere transport barrier in performance assessments.
Scaling up of laboratory data
Data on sorption for use in performance assessments is often principally derived from batch sorption experiments in the laboratory. Data relevant to colloid-facilitated radionuclide transport is also often obtained in the laboratory. There can, however, be unavoidable differences between the conditions under which studies are conducted in the laboratory and those that prevail in the field. Laboratory sorption experiments are often conducted on samples that have been mechanically disaggregated or ground up, thus potentially creating new sorbing surfaces. Furthermore, the chemical environment in the laboratory may be perturbed by the presence of air. Mechanistic sorption models can be useful in assessing the effects of differences in the chemical environment. In situ mass transport experiments are recommended, however, as possibly the only way to test the assumptions associated with the upscaling of laboratory experiments to field conditions and, more generally, to test models of transport and retention processes under conditions relevant to performance assessment.
Characterisation of fractured media
To support the modelling of matrix diffusion in performance assessments, there is a need to develop and refine methods to characterise the degree of channelling within fractured rocks and the distance into the wallrock over which matrix diffusion can be expected to occur. Currently, significant uncertainties can arise in geosphere transport modelling due to the lack of quantitative information on these characteristics.
In order to provide convincing site-specific evidence that matrix diffusion can be relied upon as a retention mechanism, it is advantageous to concentrate efforts on characterising any higher porosity altered zones adjacent to fractures, where the existence of diffusion-accessible porosity may be relatively easy to demonstrate. It is more difficult to demonstrate that matrix diffusion can occur in more distant and less porous, unaltered wallrock. Furthermore, it may be unnecessary to invoke matrix diffusion in order to demonstrate the effectiveness of the geosphere transport barrier.
When field tracer experiments are carried out, especially if they are to be used to derive parameters for performance assessment models, care must be taken to collect high-accuracy data from the tails of breakthrough curves, the form of which may reflect matrix heterogeneity. Where possible, derived parameters should be supported independently by non-tracer-test data.
Heterogeneity and variability
Several examples were presented in the workshop which indicated the importance of carefully considering the possible implications of heterogeneity over a wide range of scales. This also included variable timescales, when modelling retention processes in performance assessment. Averaging may have non-conservative consequences; it may not, for example, be appropriate to represent the whole of a heterogeneous geological medium by a single Kd or single diffusion coefficient or by probability distribution functions for these parameters. The structure of relevant features over a range of scales should, therefore, be considered when developing and applying models in performance assessments, as should the possible consequences of variations of structure, mineralogy, and groundwater composition with time.
In this context, the use of semi-empirical functions to represent the dependence of Kd on geochemical conditions can be useful in representing the effects of spatial heterogeneity and temporal variability of geochemical conditions on radionuclide transport. The use of such functions may be seen as a compromise approach between, on the one hand, the use of fixed Kd values, which may be over-simplistic in some circumstances and, on the other hand, the incorporation of mechanistic equilibrium sorption models in radionuclide transport codes, which may lead to excessively complicated and non-transparent assessment models. The use of semi-empirical site-binding models in performance assessment calculations could also be considered, although this would require a greater emphasis to be placed on the site-specific characterisation of the mineral assemblages and variable groundwater composition along radionuclide transport paths.
Understanding colloid-facilitated transport
It was concluded that performance assessment methods for colloid-facilitated radionuclide transport exist, or can readily be developed, that are adequate, given the current level of scientific understanding and availability of data, provided equilibrium reversible sorption can be assumed. It is recommended, however, that the validity of the assumption of equilibrium reversible sorption should be more thoroughly investigated. In particular, the kinetics of desorption from colloids have not been addressed in most investigations to date. The mechanisms for inorganic colloid generation by erosion of near-field materials and by changes in hydrogeological and geochemical conditions also deserve more attention in future studies.
Dialogue and cross-fertilisation
Retention processes are represented in a simplified manner in performance assessment models. Dialogue is essential between regulators, performance assessors, and technical specialists in various scientific disciplines, in order to identify the features and processes that are most relevant to performance, and to identify where improved understanding is, and is not, required for performance assessment purposes. Matrix diffusion in more distant and less porous, unaltered wallrock could be an example of a process which has an operation that is difficult to demonstrate, but for which feedback from performance assessment indicates that improved understanding may not be essential.
Dialogue between the groups is also required in order to ensure that model simplifications are well supported and acceptable to the regulators. In some countries, the public and other interested groups may be asked to comment on any concerns they have. Given that the importance of retention processes is recognised and has been studied outside the field of radioactive waste disposal, in areas such as mining and the management of non-radioactive hazardous wastes, there is scope for cross-fertilisation with these activities.