We have measured the dissolution rate of a simple five-component borosilicate glass (Na2O, CaO, Al2O3, B2O3, SiO2) using a flow-through system. The experiments were designed to measure the dissolution rate constant over the interval pH 1 through pH 13 at 3 temperatures (25°, 50° and 70°C). Dilute buffers were used to maintain a constant pH. Analyses of solutions and solid surfaces provided information that is used to develop a kinetic model for glass dissolution.

Under all conditions we eventually observed linear dissolution kinetics. In strongly acidic solutions (pH 1 to pH 3) all components but Si were released in their stoichiometric proportions and a thick, Si-rich gel was formed. In mildly acidic to neutral solutions the gel was thinner and was both Si- and Al-rich, while the other components were released to solution in stoichiometric proportions. In mildly to strongly alkaline solutions all components were released to solution in stoichiometric proportions. By varying the flow rate at each pH we demonstrated a lack of transport control of the dissolution rate.

The dissolution rates were found to be lowest at near-neutral pH and to increase at both low and high pH. A rate equation based on transition-state theory (TST) was used to calculate dissolution rate constants and reaction order with respect to pH over two pH intervals at each temperature. At 250C between pH 1 and pH 7 based on the Si release rate the log rate constant for glass dissolution (g glass/m20d) was −0.77 and the order with respect to pH was −0.48. Between pH 7 and pH 13 the log rate constant for glass dissolution was −8.1 and the order with respect to pH was +0.51. The measured simple glass dissolution rate constants compare very well with constants estimated by fitting the same TST equation to experimental results obtained for SRL-165 glass and to dissolution rate estimates made for synthetic basaltic glasses.

Recent studies were carried out in France to improve ion exchange resin solidification in cement-based matrices. Work is conducted on different types of waste - cationic, anionic and mixed-bed resins - under regenerated and loaded states. The main objective is to increase the waste incorporation rate up to roughly 50% (in volume). Hydraulic cement formulas are developed taking into account waste chemistry, in particular the presence of ions which inhibit setting.

Enthalpies of mixing of major oxide and waste components of proposed DWPF glasses are estimated. Several proposed glass compositions have very similar molar proportions of major structure-related components. Heats of mixing of major components are predicted to be small (glassy reference states), but may be endothermic or exothermic. Lanthanides are likely to form regions in the glasses rich in R-O-Si bonds but relatively depleted in Si-O-Si bonds. Measured enthalpies of solution of simple lanthanide-bearing glasses resemble those of mechanical mixtures of these end-members, with near-zero enthalpies of mixing for fairly large variations in lanthanide concentrations. DWPF glasses have alkalis in excess of those needed to charge-balance all T3+ cations in tetrahedral coordination. Electropositive +4, +5, and +6 cations are expected to use these excess alkalis to stabilize their own coordination polyhedra, forming complexes with effectively constant macroscopic stoichiometric ratios which contribute to enthalpies of solution in direct proportion to their concentration. Therefore, variations in the concentrations of high-valence cations in DWPF waste containment glasses are also predicted to result in near-zero enthalpies of mixing.

A kinetic model for the dissolution of borosilicate glass, incorporated into the EQ3/6 geochemical modeling code, is used to predict the dissolution rate of a nuclear waste glass. In the model, the glass dissolution rate is controlled by the rate of dissolution of an alkalidepleted amorphous surface (gel) layer. Assuming that the gel layer dissolution affinity controls glass dissolution rates is similar to the silica saturation concept of Grambow [1] except that our model predicts that all components concentrated in the surface layer, not just silica, affect glass dissolution rates. The good agreement between predicted and observed elemental dissolution rates suggests that the dissolution rate of the gel layer limits the overall rate of glass dissolution. The model predicts that the long-term rate of glass dissolution will depend mainly on ion concentrations in solution, and therefore on the secondary phases which precipitate and control ion concentrations.

The effects of groundwater chemistry on glass durability were examined using the hydration thermodynamic model. The relative durabilities of SiO2, obsidian, basalt, nuclear waste glasses, medieval window glasses, and a frit glass were determined in tuffaceous (J–13) groundwater, basaltic (GR–4) groundwater, WIPP–A brine, and Permian Basin brine (PBB–3) using the monolithic MCC–I durability test. In the groundwater–dominated MCC–l experiments, the interaction of the glasses and the initial groundwater (leachant) caused the formation of unique assemblages of secondary phases. The secondary phase formation, in turn, controlled the final groundwater (leachate) pH and ionic strength, I[t].

Correlations of the final leachate pH and I[t] with the Si release from the glass indicated that it is the influence of the secondary phase formation on the leachate pH and I[t] that controls the final dissolution rate of the glass. Since I[t] and the pH of the leachates are functions of the precipitation reactions, inclusion of the experimentally determined solution pH in the free energy of hydration model provides for the functional dependence of the dissolution rate on the secondary precipitation. Therefore, superposition of the linear equation for the groundwater and deionized water experiments occurs and the hydration free energy model can be used to compare glass durability in deionized water and in repository groundwaters.

Experimental data were obtained for conventional pH values corrected for liquid junction, amorphous silica solubility and glass corrosion in concentrated salt brines. The data were interpreted with a geochemical model. The brine chemistry was described with the Pitzer formalism [1] using a data base which allows calculation of brine compositions in equilibrium with salt minerals at temperatures up to 200°C.

In MgCl2 dominated brines Mg silicates form and due to the consumption of Mg the pH decreases with proceeding reaction. A constant pH (about 4) and composition of alteration products is achieved, when the alkali release from the glass balances the Mg consumption. The low pH results in high release of rare earth elements REE and U from the glass. In the NaCl dominated brine MgCl2 becomes exhausted by Mg silicate formation. As long as there is still Mg left in solution the pH decreases. After exhaustion of Mg the pH rises with the alkali release from the glass and analcime is formed.

Short and long-term geochemical interactions of R7T7 nuclear glass with water at 100°C were simulated with the DISSOL thermokinetic computer code. Both the dissolved glass quantity and the resulting water composition, saturation states and mineral quantities produced were calculated as a function of time. The rate equation used in the simulation was first proposed by Aagaard and Hegelson: v = k+.S.a(H+)-n(l - e-(A/RT)). It simulates a gradually diminishing dissolution rate as the reaction affinity diminishes. The best agreement with 1-year experimental data was obtained with a reaction affinity calculated from silica activity (Grambow's hypothesis) rather than taking into account the activity of all the glass components as proposed by Jantzen and Plodinec. The concept of residual affinity was introduced by Grambow to express the fact that the glass dissolution rate does not cease. We prefer to replace the term “residual affinity” by “contextual affinity”, which expresses the influence on the dissolution rate of three factors: the solution chemistry, the metastability of SiO2(m), and the possible precipitation of certain aluminosilicates such as zeolites.

Perovskite (CaTiO3 )and hollandite ([BaxCsy,][(Ti,Al) 3+ Ti4 + ]016) 2x+y 8-2x-y are two of the three major phases of the Synroc titanate mineral assemblage. The chemical durabilities of these two phases have been investigated by solution analysis, surface analysis and electron microscopy. While a detailed model of the durability of perovskite has evolved from previous work, much less is known about the mechanisms which affect the durability of hollandite. The present investigations show that, for single-phase hollandite, selective extraction by ion exchange of Cs+ and Ba 2+ does not play a significant role. On the other hand, total dissolution of the network does seem to occur as a result of base-catalysed hydrolysis. Also, complexation and precipitation of Al and TI species in solution has the effect of driving the system to low pH. These results can be correlated with the more complete model for dissolution of perovskite in order to infer the overall effects of aqueous attack on the Synroc HLW solid.

Test samples of 131 type glass that have been reacted for extended time periods in water vapor atmospheres of different relative humidities and in static leaching solution have been examined to characterize the reaction products. Analytical electron microscopy (AEM) was used to characterize the leached samples, and a complicated layer structure was revealed, consisting of phases that precipitate from solution and also form within the residual glass layer. The precipitated phases include birnessite, saponite, and an iron species, while the intralayer phases include the U-TI containing phase brannerite distributed within a matrix consisting of bands of an Fe rich montmorillonite clay. Comparison is made between samples leached at 40°C for 4 years with those leached at 90°C for 3-1/2 years. The samples reacted in water vapor were examined with scanning electron microscopy and show increasing reaction as both the relative humidity and time of reaction increases. These samples also contain a layered structure with reaction products on the glass surface.

Laboratories from 11 countries participated in a ‘Round Robin’ comparative test excercise using the European Commission's Repository Systems Simulation Test [1,5]. The compulsory core part of this exercise involved testing an inactive simulant of the French high-level waste glass (SON68) in containers in which the conditions were similar to those that might occur after disposal of vitrified waste in a future repository. There were three options for the test, representing repositories in clay, granite and salt. Fifteen laboratories, mostly from within the European Community, but including laboratories from USA, Canada and Japan took part in the Round Robin. Most of the laboratories only covered one option: clay, 3, granite, 10, and salt 6. The test procedure was closely specified by the European Communities. This contribution and a companion paper [2] summarize the acquired information.

The HLW borosilicate glass SON68 (R7T7), to be returned to and eventually disposed of in various geological repositories by Cogema's reprocessing customers, was tested in a stainless steel autoclave developed and provided by CEA, Marcoule. The glass was prepared and delivered by CEA; together with crushed granite, sand, smectite and groundwater for the granite option. SCK-Mol provided the clay, smectite, sand and the recipe for claywater, HMI-Berlin the recipe for the rock salt composition. The analytical method was the only variable to be chosen by the participants. After termination of the experiments (duration 14-364d, temperature 90°C), the concentrations of the glass constituents Si, B, Mo, Li, in solution had to be determined. The data base was used to evaluate the test procedure and to identify analytical problems.

A ‘Round Robin’ test of the European Commission 's “Repository Systems Simulation Test” has been undertaken. This involved testing an inactive simulant of the French high-level waste glass SON68 (RTT7) in containers which also held backfill, geological materials and water. Ten laboratories undertook the granite option, six salt and three clay. The test procedure was closely specified and is given in detail in a companion paper [1]. Experiments with glass powder and (optionally) solid specimens in contact with aqueous solutions were run for periods of up to one year at 90 °C and the solutions were analysed for Si, B, Li and Mo (compulsory) and other constituents of the corroded glass (optional). The analytical results are used to compare the performance of the glass under conditions representing those that might occur in future repositories in the three media.

The first retrieval of an in-situ experiment on the interaction waste form - clay host in the underground laboratory under the Mol site has been finished successfully. The test consisted in a two years exposure of various candidate simulated waste glasses at 90°C to Boom clay. The retrieval was done by overcoring. The experimental data showed satisfactorily correspondence between in-situ and laboratory simulation tests both for mass loss and surface analytical data, supporting the validity of the in-situ test as it was performed. The thickness of waste form dissolved within two years varies between 40 and 325 μm (case of the high-level waste glasses), depending on the composition. Matrix dissolution is expected to be the major mechanism of interaction.

The interaction between simulated HLW glasses and clay has been investigated, as part of a larger programme to evaluate the performance of vitrified HLW in repository conditions. Experimental conditions were carefully chosen to simulate repository conditions as close as possible. The glass corrosion seems to be controlled by a diffusion process, but there are indications of a (small) final corrosion rate. The leaching of 134Cs, 90Sr, 239Pu and 241Am is characterized by extremely small mobile concentrations leached (for 239pu and 241Am, below detection limit). The presence of a gamma irradiation field does not increase the glass dissolution. The production of radiolythic hydrogen is strongly reduced by the presence of clay.

The long term corrosion of two reference Belgian high-level waste glasses (SAN60 and SM58) were investigated in pure water. The corrosion was studied using powdered glass at a high surface area to volume ratio and temperatures of 90 and 120°C. The experimental data at 90°C reveal a “final” leach rate which decreases with time. At 120°C this “final” state is transient and followed by an enhanced dissolution, which was correlated with extensive surface crystallization. Modelling using the PHREEQE and GLASSOL computer codes described the initial corrosion, but was unable to account for the enhanced dissolution at 120°C.

The results of hydrothermal leach tests are intended to be used to predict long-term low-temperature glass dissolution. It is often assumed that data can be extrapolated to other conditions using an Arrhenius-type equation. Hydrothermal leaching mechanisms and their temperature dependence in R7T7 glass were investigated in static experiments lasting from 7 days to 1 year at 150°C and 250°C. Leachates, surface layers and crystalline products were analyzed by ICP, TEM, SEM, EMP, XRD and cathodoluminescence. Unexpectedly, no actual saturation conditions in solutions were reached after one year leaching at 150°C nor at 250°C. The effect of precipitation of alteration products (a silica-enriched amorphous layer and aluminosilicates [smectite at 150°C, smectite and zeolites at 250°C]) is discussed. However, the formation of large cracks in the bulk glass results in a higher glass reacting surface and a higher dissolution rate at 250°C. Arrhenius calculations cannot be used to extrapolate our hydrothermal data to lower-temperatures.

R7T7 glass alteration was investigated in the presence of various moist clays. In contact with smectite 4a, selected in France as a potential engineered barrier material, the glass was significantly corroded: after 6 months the glass corrosion rate was practically the same as the initial alteration rate in double-distilled water. Substantially lower alteration was observed in contact with bentonite 6 activated by sodium carbonate.

Smectite 4a consumes silicon released by glass corrosion, and thus retards the rise to high silicon concentrations in solution at which the glass corrosion rate diminishes. Glass can therefore in some cases be altered as much in moist clay as in water with high renewal rates. Other experiments with smaller quantities of smectite 4a showed that the phenomenon becomes less important in time: its duration is proportional to the ratio between the clay mass (C) and the glass surface area (SA). Comparing the results of studies at different C/SA ratios indicates that low glass corrosion rates are obtained more slowly at higher C/SA ratios.

Tests with 239Pu-doped R7T7 glass also showed that the radionuclide retention factor in the alteration film at the glass surface is only 6, compared with a factor of nearly 50 in double-distilled water.

The Product Consistency Test (PCT) is a glass leach test developed at the Savannah River Site (SRS) to confirm the durability of radioactive nuclear waste glasses that will be produced in the Defense Waste Processing Facility. The PCT is a seven day, crushed glass leach test in deionized water at 90°C. Final leachates are filtered and acidified prior to analysis. To demonstrate the reproducibility of the PCT when performed remotely, SRS and Argonne National Laboratory have performed the PCT on samples of two radioactive glasses. The tests were also performed to compare the releases of the radionuclides with the major nonradioactive glass components and to determine if radiation from the glass was affecting the results of the PCT.

The test was performed in triplicate at each laboratory. For the major soluble elements, B, Li, Na, and Si, in the glass, each investigator obtained relative precisions in the range 2–5% in the triplicate tests. This range indicates good precision for the PCT when performed remotely with master slave manipulators in a shielded cell environment. When the results of the two laboratories were compared to each other, the agreement was within 20%. Normalized concentrations for the nonradioactive and radioactive elements in the PCT leachates measured at both facilities indicated that the radionuclides were released from the glass slower than the major soluble elements in the glass. For both laboratories, the normalized releases for both glasses were in the general order Li∼B∼Na>Si>Cs-137>Sb-125>Sr-90. The normalized releases for the major soluble elements and the final pH values in the tests with radioactive glass are consistent with those for nonradioactive glasses with similar compositions. This indicates that there was no significant effect of radiation on the results of the PCT.

Glass monoliths of the WVC4M4, WVCM50, SRL 165, and SRL 202 compositions were reacted in steam and in hydrothermal liquid at 200°C. The glass reaction resulted in the formation of leached surface layers in both environments. The reaction in steam proceeds at a very low rate until precipitates form, after which the glass reaction proceeds at a greater rate. Precipitates were formed on all glass types reacted in steam. The assemblage of phases formed was unique to each glass type, but several precipitates were common to all glasses, including analcime, gyrolite, and weeksite. Reaction in steam occurs in a thin layer of condensed water which becomes saturated with respect to the observed phases after only a few days of reaction. The reaction in steam is accelerated relative to reaction in hydrothermal liquid In the sense that secondary phases form after a shorter reaction time, that is, after less glass has reacted, because of the smaller effective leachant volume present in the steam environment. A knowledge of the secondary phases which form and their influence on the glass reaction rate is crucial to the modeling effort of the repository program.

Eventual liquid water contact of high-level waste glass stored under the unsaturated conditions anticipated at the Yucca Mountain site will be by slow intrusion of water into a breached container/canister assembly. The water flow patterns under these unsaturated conditions will vary, and the Unsaturated Test method has been developed by the YMP to study glass reaction. The results from seven different sets of tests done to investigate the effect of systematically varying parameters such as composition and degree of sensitization of 304L stainless steel, water input volume, and the interval of water contact are discussed. Glass reaction has been monitored over a period of five years, and the parametric effects can result in up to a ten-fold variance in the degree of glass reaction.

Materials Interface Interaction Test results for up to 2 years indicate that surface layers formed on synthetic basaltic glass in brine are predominantly Mg-chloride or oxychloride. Layers formed on HMI and R7T7 borosilicate glass are Mg-silicates. Layer thickness increases from 1 year to 2 years for basaltic glass and HMI borosilicate glass. Mg-silicate contains Ni and Zr in amounts that are generally related to the composition of associated glass, but layers formed on basaltic glass also contain Ni and Zr, elements not present in the glass. Ni and Zr are, in part externally derived, however, Ti appears to be completely immobile, and is not found in layers formed on Ti-free glass (R7T7). Some chemical components of the surface layers are derived, in part, from alteration of glass but major components of the layers are from the brine.