Facilities for vitrification of high-level nuclear waste in the United States are scheduled for startup in the next few years. It is, therefore, appropriate to examine the current scientific basis for understanding the corrosion of high-level waste borosilicate glass for the range of service conditions to which the glass products from these facilities may be exposed. To this end, a document has been prepared which compiles worldwide information on borosilicate waste glass corrosion. Based on the content of this document, the acceptability of canistered waste glass for geological disposal is addressed.

Waste glass corrosion in a geologic repository may be due to groundwater and/or water vapor contact. The important processes that determine the glass corrosion kinetics under these conditions are discussed. Testing data together with understanding of the long-term corrosion kinetics are used to estimate radionuclide release rates. These rates are discussed in terms of regulatory performance standards.

A summary of the current state of defense high-level radioactive waste in the U.S. is presented. Reasons for the selection of borosilicate glass as the matrix for the waste glass product over other glasses is enunciated. The product and processing requirements which the waste glass must meet are summarized. Finally the effects of crystallization and glass-in-glass phase separation on waste glass durability are discussed. Crystallization, under typical cooling conditions, has minimal effect on durability but glass-in-glass phase separation should be avoided.

A literature survey has been performed to assess the effects of the temperature, glass surface area/leachate volume ratio, leachant composition, leachant flow rate, and glass composition (actual radioactive vs. simulated glass) used in laboratory tests on the measured glass reaction rate. The effects of these parameters must be accounted for in mechanistic models used to project glass durability over long times. Test parameters can also be used to highlight particular processes in laboratory tests. Waste glass corrosion results as water diffusion, ion exchange, and hydrolysis reactions occur simultaneously to devitrify the glass and release soluble glass components into solution. The rates of these processes are interrelated by the effects of the solution chemistry and glass alteration phases on each process, and the dominant (fastest) process may change as the reaction progresses. Transport of components from the release sites into solution may also affect the observed corrosion rate. The reaction temperature will affect the rate of each process, while other parameters will affect the solution chemistry and the particular processes that are observed during the test. The early stages of corrosion will be observed under test conditions which maintain dilute leachates and the later stages will be observed under conditions that generate more concentrated leachate solutions. Typically, water diffusion and ion exchange reactions dominate the observed glass corrosion in dilute solutions, while hydrolysis reactions are dominant in more concentrated solutions. Which process controls the long-term glass corrosion is not fully understood, and the long-term corrosion rate may be either transport- or reaction-limited.

The weathering of glass is reviewed by examining processes that affect the reaction of commercial, historical, natural, and nuclear waste glass under conditions of contact with humid air and slowly dripping water, which may lead to immersion in nearly static solution. Radionuclide release data from weathered glass under conditions that may exist in an unsaturated environment are presented and compared to release under standard leaching conditions. While the comparison between the release under weathering and leaching conditions is not exact, due to variability of reaction in humid air, evidence is presented of radionuclide release under a variety of conditions. These results suggest that both the amount and form of radionuclide release can be affected by the weathering of glass.

Water contact subjects waste glass to chemical attack that results in the formation of surface alteration layers. Two principal hypotheses have been advanced concerning the effect of surface alteration layers on continued glass corrosion: (1) they act as a mass transport barrier and (2) they influence the chemical affinity of the glass reaction. In general, transport barrier effects have been found to be less important than affinity effects in the corrosion of most high-level nuclear waste glasses. However, they can be important under some circumstances, for example, in a very alkaline solution, in leachants containing Mg ions, or under conditions where the matrix dissolution rate is very low. The latter suggests that physical barrier effect may affect the long-term glass dissolution rate. Surface layers influence glass reaction affinity through the effects of the altered glass and secondary phases on the solution chemistry. The reaction affinity may be controlled by various precipitates and crystalline phases, amorphous silica phases, gel layer, or all the components of the glass. The surface alteration layers influence radionuclide release mainly through colloid formation, crystalline phase incorporation, and gel layer retention. This paper reviews current understanding and uncertainties.

Models for borosilicate glass dissolution must account for the processes of (1) kinetically-controlled network dissolution, (2) precipitation of secondary phases, (3) ion exchange, (4) rate-limiting diffusive transport of silica through a hydrous surface reaction layer, and (5) specific glass surface interactions with dissolved cations and anions. Current long-term corrosion models for borosilicate glass employ a rate equation consistent with transition state theory embodied in a geochemical reaction-path modeling program that calculates aqueous phase speciation and mineral precipitation/dissolution. These models are currently under development. Future experimental and modeling work to better quantify the rate-controlling processes and validate these models are necessary before the models can be used in repository performance assessment calculations.

The release of glass components into solution, including radionuclides, may be influenced by the presence of radiolytically produced nitric acid, carboxylic acid, and transient water dissociation products such as ·OH and O2-. Under batch test conditions, glass corrosion has been shown to increase up to a maximum of three-to five-fold in irradiated tests relative to nonirradiated tests, while in other studies the presence of radiolytic products has actually decreased glass corrosion rates. Bicarbonate groundwaters will buffer against pH decreases and changes in corrosion rates. Under high surface area-to-solution volume (S/V) conditions, the bicarbonate buffering reservoir may be quickly overwhelmed by radiolytic acids that are concentrated in the thin films of water contacting the samples. Glass reaction rates have been shown to increase up to 10-to-15-fold due to radiation exposure under high S/V conditions.

Radiation damage to solid glass materials results in bond damage and atomic displacements. This type of damage has been shown to increase the release rates of glass components up to four-fold during subsequent corrosion tests, although under actual disposal conditions, glass annealing processes may negate the solid radiation damage effects.

The design and analysis of waste package and engineered systems for geologic disposal of radioactive wastes are being conducted in most countries with so-called performance assessment models and associated computer codes that solve the governing equations. The models range in complexity from simple 1-D analytical solutions to multidimensional, multicomponent reactive transport models implemented in large computer codes. Borosilicate waste glass is certainly the most studied and probably best understood waste form that has been developed for encapsulating high-level waste. Reasonably good models are available that describe the reaction kinetics and transformation of the glass into a paragenetic sequence of more stable secondary phases. Given the broad range of computational tools and extensive database on glass/water reactions that is available, it is surprising that little progress has been made to link these models and data in a comprehensive performance assessment of waste glass in a repository setting. In this paper, we will explore reasons why this linkage has not been developed, give several examples illustrating the importance of doing so, and illustrate an approach to accomplish such a linkage.

This paper provides an overview of waste glass alteration. Emphasis is on the evolution of surface layers and factors affecting the alteration rate when glass is subjected to an aqueous environment. The extent, type and rate of alteration is determined by a variety of parameters including time, temperature, glass composition, pH, Eh, composition of leachate/geology, the presence of other engineered barriers, flow conditions and surface area/volume ratio. Models (based on kinetic and thermodynamic considerations) developed by other researchers that are consistent with the experimental observations from the laboratory and field experiments are discussed. The morphology of the surface layers as described by interpretation of secondary ion mass spectrometry (SIMS) analyses on waste glasses and the role the various layers play in the alteration process is presented as well.

Many types of procedures have been developed for testing of the chemical durability of nuclear waste forms. These procedures differ from each other in critical aspects, such as duration, replenishment or non-replenishment of the leachates, and S/V ratio. As a result, different answers to basic questions, such as how waste-form leachability depends on its chemical composition, are obtained when different test methods are used. Furthermore, the possibility that some glasses may exhibit a leach rate excursion within the test period causes the composition dependence to be an even more sensitive function of test duration and of leachant replenishment. These factors also complicate the use of test data for the prediction of long-term waste form behavior.

The actual status of an in-situ test programme exposing different waste glass samples directly to Boom clay is reviewed. Corrosion test tubes have been retrieved after residence for 5 years at 16° C, 2 years at 90° C, and 5 years at 170° C. The corrosion is interpreted in terms of mass loss, surface analysis by SEM and profiling by EPMA and SIMS. At 16° C, glasses dissolve about 0.02 – 0.08 µm per year. At higher temperature dissolution is more than two orders of magnitude larger. A good agreement is obtained between the mass losses and the surface analyses. The advantages and limitations of the Belgian in-situ tests are compared with the conclusions of an international expert group.

An extremely important question for the eventual disposal of glass in natural environments is the relevance of laboratory testing of glass durability to the long-term performance of glass in geologic environments. The purpose of this study was to attempt to provide an empirical answer to that question, by applying the hydration thermodynamics approach (which has successfully been applied to laboratory tests of glass durability) to the results of longer-term testing in natural environments.

The results show that hydration thermodynamics is a useful tool for explaining the effects of glass composition observed in in-situ tests, in several environments. Thus, it appears to provide a link between laboratory tests of glass durability and the results of in-situ tests in natural environments. Perhaps the most important conclusion of this effort is that the in-situ test results emphasize the importance of control of chemical composition during glass production as a means of achieving a durable glass.

The common observation of glasses persisting in natural environments for long periods of time (up to tens of millions of years) provides compelling evidence that these materials can be kinetically stable in a variety of subsurface environments. This paper reviews how natural and historical synthesized glasses can be employed as natural analogues for understanding and projecting the long-term alteration of high-level nuclear waste glasses. The corrosion of basaltic glass results in many of the same alteration features found in laboratory testing of the corrosion of high-level radioactive waste glasses. Evidence has also been found indicating similarities in the rate controlling processes, such as the effects of silica concentration on corrosion in groundwater and in laboratory leachates. Naturally altered rhyolitic glasses and tektites provide additional evidence that can be used to constrain estimates of long-term waste glass alteration. When reacted under conditions where water is plentiful, the corrosion for these glasses is dominated by network hydrolysis, while the corrosion is dominated by molecular water diffusion and secondary mineral formation under conditions where water contact is intermittent or where water is relatively scarce. Synthesized glasses that have been naturally altered result in alkali-depleted alteration features that are similar to those found for natural glasses and for nuclear waste glasses. The characteristics of these alteration features appear to be dependent on the alteration conditions which affect the dominant reaction processes during weathering. In all cases, care must be taken to ensure that the information being provided by natural analogues is related to nuclear waste glass corrosion in a clear and meaningful way.

The current knowledge on the glass dissolution mechanism and the representation of glass dissolution concepts within overall repository performance assessment models are briefly summarized and uncertainties related to mechanism, radionuclide chemistry and parameters are discussed. Understanding of the major glass dissolution processes has been significantly increased in recent years. Long-term glass stability is related to the long-term maintenance of silica saturated conditions. The behavior of individual radionuclides in the presence of a dissolving glass has not been sufficiently and results do not yet allow meaningful predictions. Conservative long-term predictions of glass matrix dissolution as upper limit for radionuclide release can be made with sufficient confidence, however these estimations generally result in a situation were the barrier function of the glass is masked by the efficiency of the geologic barrier. Realistic long-term predictions may show that the borosilicate waste glass contributes to overall repository safety to a much larger extent than indicated by overconservatism. Today realistic predictions remain highly uncertain and much more research work is necessary. In particular the long-term rate under silica saturated conditions needs to be understood and the behavior of individual radionuclides in the presence of a dissolving glass deserves more systematic investigations.

The best possibility for gaining an understanding of the likely future behavior of a high level nuclear waste disposal system is use of the scientific method. However, the scientific approach has inherent limitations when it comes to making long-term predictions with confidence. This paper examines these limiting factors as well as the criteria for admissibility of scientific evidence in the legal arena, and concludes that the prospects are doubtful for successful licensing of a potential repository under the regulations that were binding until recently. Suggestions are made for remedying this situation.

There is no technical basis for setting a time limit of 10,000 years on the regulated performance of a nuclear waste repository. First, accurate prediction of releases for such periods is not possible. Second, there is nothing unique about 10,000 years. Third, equally toxic materials, which never transform to non-toxic substances by radioactive decay, have less stringent requirements. And fourth, during a 10,000 year time frame, natural disasters will dwarf the worst possible outcomes of repository placement.

Analyses could be required to extend as long as doses above current radiation protection guidelines are possible (perhaps several million years), but these results should be recognized as qualitative information rather than evidence of quantitative compliance with exact numerical limits. Concern for what will happen over long times can be addressed for the next several hundred years by maintaining waste retrievability. At that time, uncertainty about future performance should have been reduced significantly.

The U. S. Department of Energy (DOE) is developing a nuclear waste management system that will accept high-level radioactive waste, transport it, store it, and ultimately emplace it in a deep geologic repository. The key activity now is determining whether Yucca Mountain, Nevada is suitable as a site for the repository. If so, the crucial technological advance will be the demonstration that disposal of nuclear waste will be safe for thousands of years after closure. This paper assesses the impact of regulatory developments, legal developments, and scientific developments on such a demonstration.

The Swedish Nuclear Power Inspectorate (SKI) has recently concluded its evaluation of the Swedish programme for the development of a system for the management of nuclear waste. The programme was compiled and issued by the Swedish Nuclear Fuel and Waste Management Company (SKB). In this process of programme formulation and review, considerable attention has been paid to the question of how scientific studies should be directed and performed in order to provide the support needed in the programme.

When the objectives and limitations of such scientific studies are to be analysed, it is vitally important that the implications of the long timescales and the lack of feedback of post-closure experience are fully understood and adequately dealt with. It must be realised, that simulations will have to be utilized for the systems development work as well as for the assessments of safety. Such performance simulations have to be based on a thorough understanding of the pertinent phenomena as well as on a comprehensive base of experimental data. They must also be based on a good overall perspective that can only be obtained through full performance assessments.

Thus, a sound scientific base is needed. The components of this base must be traceable. Moreover, the links must be clearly identified between the knowledge base and the development of the disposal system as well as the performance assessment of such a system. Performance simulations and performance assessments should be utilized, not only as tools to develop a system and to evalute its safety but also to identify areas where further research is required - or unnecessary.

The knowledge base will, however, always be limited, and a lack of understanding must also be recognised and adequately dealt with, e.g. by accounting for uncertainties in the performance simulations, by relying on bounding (conservative) assumptions or by robust repository design.

Thus, the success of a programme for the handling and disposal of nuclear waste is highly dependent on the strategy applied for the utilization of the scientific knowledge base. The requirements on such a strategy increase considerably when substantial commitments are to be made.

Radiation damage development and the corresponding energy storage in pure undeformed single crystals have frequently been studied in laboratory experiments, however little is known of irradiation experiments on natural rock salt (polycrystalline, deformed and impure) under geological conditions. The relevance of these parameters to the defect formation was revealed by a joint United States/Federal Republic of Germany in-situ test in the Asse Mine. Natural rock salt was heated and irradiated using Co-60 sources.

Calculations of the amount of halite expected to be decomposed by radiolysis during the experiment were performed using the 1985 version of the Jain-Lidiard model. Qualitative agreement between theory and analyses was found for all the performed analyses. Quantitative and qualitative deviations of the natural samples behaviour from that of single undeformed crystals were observed and attributed to the influence of sulfatic admixtures, ‘polycrystallinity’ and strain on radiation damage development and anneal.

Special chemical methods, such as iodometric titration and uv-visible spectroscopy were applied in order to measure the hypochlorite ion, which forms in irradiated salt. Composite samples located closest to the Co-60 source averaged 0,4 micromoles neutral chlorine atoms per gram salt, a factor of two more than from other positions. Uv-vis analyses revealed more than a factor of ten greater neutral chlorine concentrations in coloured halite. Similarly, optical absorption measurements indicated a factor of ten difference in sodium metal colloid concentrations.

The radiation stability of polyphase titanate ceramic waste forms was studied using analytical transmission electron microscopy, in combination with in situ irradiation by 30 keV He ions, followed by staged annealing. Two experiments were conducted. In the first, a reconnaissance investigation was made of the stabilities of the synthetic minerals hollandite, zirconolite, and perovskite when subjected to a total dose of 1.8 × 1017 He+ cm-2. It was found that all phases amorphized at approximately the same rate, but perovskite recovered its structure more rapidly and at lower temperatures than the other phases. In particular, annealing for 10 minutes at 1000°C was sufficient for perovskite to completely regain its crystallinity, while zirconolite and hollandite were only partially restored by these conditions. In the second experiment, the response of a thin hollandite crystal to irradiation was examined by selected area electron diffraction. At a dose of 1.5 × 1015 He+ cm-2its incommensurate superstructure was disrupted, but even at a dose of 3 × 1016 He cm-2 the hollandite subcell was largely intact. For this dose, total recovery was achieved by annealing for 1 minute at 1000°C.