Glass dissolution takes place through metal leaching and hydration of the glass surface accompanied by development of alteration layers of varying crystallinity. The reaction which controls the long-term glass dissolution rate appears to be surface layer dissolution. This reaction is reversible because the buildup of dissolved species in solution slows the dissolution rate due to a decreased dissolution affinity. Glass dissolution rates are therefore highly dependent on silica concentrations in solution because silica is the major component of the alteration layer.

Chemical modeling of glass dissolution using reaction path computer codes has successfully been applied to short term experimental tests and used to predict long-term repository performance. Current problems and limitations of the models include a poorly defined long-term glass dissolution mechanism, the use of model parameters determined from the same experiments that the model is used to predict, and the lack of sufficient validation of key assumptions in the modeling approach. Work is in progress that addresses these issues.

Leaching cells were designed to test the alterability of nuclear waste glasses under conditions closely simulating an actual geological repository. This paper summarizes the results of twenty long-duration tests simulating a variety of storage conditions. The effects of the backfilling materials, the canister, glass cracking and crystallization, a activity and the nature of the host rock are discussed. Moreover, an experiment has been in progress for over seven years in a granite medium; after three years the corrosion rate dropped to about 2 × 10−3 g m−2 d−1, and is sustained only by the water renewal due to sampling at regular intervals. These results constitute a valuable data base on R7T7 glass alteration in geological media at 90°C. Glass alteration mechanisms described in pure water appear to be applicable to a hard rock medium. In clay and salt, however, considerable work remains to be done before environmental factors can be taken into account in an overall dissolution model.

Experimental evidence indicates that glass corrosion rates decrease proportionally with the increase of silicic acid concentration in the solution contacting the glass surface. A minimum corrosion rate (Rsat) is reached when the solution becomes saturated with respect to an unidentified amorphous siliceous compound. In a repository where the vitrified waste form is surrounded by compacted bentonite, the silica dissolved from the glass will diffuse into the pore solution and concentration gradients will be established throughout the backfill material. The silicic acid concentration at the glass-bentonite interface, and thus the glass corrosion rate, will then be diffusion controlled. Moreover, experimental work suggests that significant sorption of silica by clay minerals in bentonite may accelerate glass corrosion.

A model describing glass corrosion coupled with diffusive transport and sorption of silica in bentonite has been developed and incorporated in a FORTRAN computer code (GLADIS). The model assumes: (a) a linear isotherm for the sorption of silica (KD), (b) time and space invariant pH, temperature and ionic strength, (c) proportionality between the quantity of silica precipitated and the amount of glass dissolved and (d) cylindrical geometry. Preliminary calculations with a particular parameter set at 90°C, assuming no silica sorption on the bentonite, indicate for an unfractured glass block that a stationary state is rapidly reached in which the silica concentration at the glass-bentonite interface is lower than the saturation concentration. This implies that the glass corrodes at a more rapid rate than Rsat (RsS ∼ 8 Rsat) If moderate silica sorption is assumed (KD = 0.5 m3 kg−1), the attainment of stationary conditions is delayed by the removal of silicic acid from solution, and the average corrosion rate is further increased by a factor ∼ 2.

The corrosion of glasses in aqueous media is described by the general rate equation proposed by Grambow. In its simplest form, this equation may be written as: V = Vo [1-a(H4SiO4)/a*(H4SiO4)]. The purpose of this investigation was to measure the initial dissolution rate (Vo) of R7T7 nuclear waste glass at temperatures between 90 and 250°C, and to determine activation energies of reaction.

Leaching solutions were maintained far from equilibrium using a dynamic leaching apparatus: the High-Temperature Soxhlet device. The corrosion rates were defined by weight loss measurements and by silicon, boron, sodium and lithium analysis in the leachate. Throughout the test temperature range, the normalized mass losses were linear with time. The initial dissolution kinetics calculated from weight loss (before and after scraping off the surface layer) and from Si, B, Na and Li release, followed an Arrhenius relation. Activation energies range from 58 to 60 kJ·mol−1 depending on the element. This slight discrepancy is attributed to a more pronounced retention factor for certain elements In the alteration layer as the temperature rises. The mean activation energy of reaction (about 59 kJ·mol−1) Is typical of a surface-controlled reaction. This energy is unique in the temperature range investigated.

Isotopically labeled water has been utilized by several researchers to help elucidate the glass dissolution mechanism in aqueous media. However, most of the results have been obtained with simple silicate glasses and have utilized low surface-to-volume (S/V) ratios and relatively short time scales. In this paper we report the results of an extensive study of the leach behavior of the preliminary West Valley nuclear waste glass composition, WV205, in deuterium oxide using multiple S/V ratios at both short and long times. The WV205 glass was leached in parallel in both ordinary deionized water and in deuterated water at five S/V ratios (20, 100, 200, 2000 and 6000 m−1) at 90°C under PCT (the SRL-modified MCC3 procedure) conditions with sampling at 1, 3, 7, 28, 56, 120 days, and four times annually thereafter; the tests will continue for several years. Initial rates were determined by measurements at shorter times with polished monoliths. A significant isotope effect, as measured by the ratio of leach rates, RH20/RD20, was found at all S/V ratios and reached values as large as 360%. Our data yield different values for RH20/RD20 in the diffusion, matrix dissolution, and saturation dominated regimes with the largest effect in the middle stage. Large values of RH20/RD20 would be characteristic of a primary kinetic isotope effect suggestive of a bond breakage involving hydrogen in the rate determining step.

R7T7 nuclear waste glass dissolution in highly dilute aqueous media under static conditions at 90°C occurs according to two different mechanisms depending on the solution acidity. In acid media (pH 4.8 and 5.5), preferential extraction of glass network modifiers results in the formation of an alkali metal-depleted surface region on which amorphous and crystallized (phosphate) compounds rich in transition elements precipitate. Steady-state dissolution conditions are not reached, as attested by variable normalized SI, B and Na mass losses. Glass dissolution is stoichiometric in basic media (pH 7 to 10): the strong bonds of the silicated network are broken at a rate that increases with the pH: the glass dissolution rate increases by a factor of 15 between pH 7 and 10. Under these conditions, alteration products at the glass/solution interface do not constitute a short-term kinetic barrier against the release of the major glass components.

Leach tests on borosilicate glasses using a partial-replenishment technique show that in certain cases large excursions in leach rate may appear at the end of periods ranging between 1 and 20 months. The extent and time of these leach rate transients are very sensitive to glass and leachant composition and to S/V ratio, and they may be eliminated by slight increases in Si and Al levels in the glass or by using a neutrally buffered leachant. This phenomenon, which may reflect cracking of the initial surface layer, can have a considerable effect on the results and complicate the modeling of leach tests which strongly reflect the extent of leaching during the early stages of glass/leachant interaction. However, these effects appear to have relatively little significance in the long term.

The examination of reacted layers that form as glass reacts can provide insight into reaction mechanisms that control long-term glass reaction, and into processes by which radionuclides are released to solution. Examples are given for the natural glasses obsidian and tektite, and the nuclear waste glasses 131 and 165, where the layers form both in-situ and precipitated from solution, where they are both amorphous and crystal 1ine, where selective incorporation of actinide elements Into stable phases may offer a barrier to release, and where sloughing of the layers from the glass acts as an alternative radionuclide release process to solubility-limited control.Several reaction processes occur simultaneously as glass reacts, and the projection of glass performance to long time periods requires the identification of those processes that dominate the long-term reaction.

The analytical expression used to model glass reaction in computer simulations such as EQ6 is compared to the results of experiments used to support the simulations. The expression correctly predicts the acceleration observed in experiments performed at high glass surface area/1eachant volume ratios (SA/V) upon the formation of secondary phases. High resolution microscopic analysis of reacted glass samples suggests that the accelerated nature of the reaction after secondary phase formation is due to changes in the reaction affinity (i.e., is a solution effect) and not a change in the glass reaction mechanism. The composition of solutions in contact with reacted samples reflect the effects of the secondary phases predicted in the model. Experiments which lead to the generation of secondary phases within short reaction times can be used to identify important secondary phases which must be Included in the data base of computer simulations to correctly project long-term glass reaction behavior.

Radiation may effect the long-term performance of glass in an unsaturated repository site by interacting with air, water vapor, or liquid water. The present study examines (1) the effects of alpha or gamma irradiation in a water vapor environment, and (2) the influence of radiolytic products on glass reaction. Results Indicate that nitric and organic acids form in an irradiated water vapor environment and are dissolved in thin films of condensed water. Glass samples exposed to these conditions react faster and have a different assemblage of secondary phases than glasses exposed to nonirradiated water vapor environments.

Subglacial basaltic glasses from Iceland have been studied in order to investigate REE behaviour during low-temperature weathering. As heavy elements accumulate in the hydrated superficial corrosion layer of borosilicate glasses, REE are found to be enriched in the natural corrosion layer of basaltic glasses (palagonite). However, mass balance calculations performed for our natural samples evidenced that this enrichment is only relative: the corrosion process leads to absolute losses of elements including REE.

Mineralogical, textural and compositional data accompanying greenschist facies metamorphism (to 300°C) of basalts of the East Rift Zone (ERZ), Kilauea, Hawaii may be evaluated relative to published and experimental results for the surface corrosion of borosilicate glass. The ERZ alteration sequence is dominated by intermittent palagonite, interlayered smectite-chlorite, chlorite, and actinolite-epidote-anhydrite. Alteration is best developed in fractures and vesicles where surface reaction layers root on the glass matrix forming rinds in excess of 100 microns thick. Fractures control fluid circulation and the alteration sequence. Proximal to the glass surface, palagonite, Fe-Ti oxides and clays replace fresh glass as the surface reaction layer migrates inwards; away from the surface, amphibole, anhydrite, quartz and calcite crystallize from hydrothermal fluids in contact with the glass. The texture and composition of basaltic glass surfaces are similar to those of a SRL-165 glass leached statically for sixty days at 150°C. While the ERZ reservoir is a complex open system, conservative comparisons between the alteration of ERZ and synthetic borosilicate glass are warranted.

A series of experiments was conducted using SRL 165 synthetic waste glass to investigate the effects of surface preparation and leaching solution composition on the alteration of the glass. Samples of glass with as-cast surfaces produced smooth reaction layers and some evidence for precipitation of secondary phases from solution. Secondary phases were more abundant in samples reacted in deionized water than for those reacted in a silicate solution. Samples with saw-cut surfaces showed a large reduction in surface roughness after 7 days of reaction in either solution. Reaction in silicate solution for up to 91 days produced no further change in surface morphology, while reaction in DIW produced a spongy surface that formed the substrate for further surface layer development. The differences in the surface morphology of the samples may create microclimates that control the details of development of alteration layers on the glass; however, the concentrations of elements in leaching solutions show differences of 50% or less between samples prepared with different surface conditions for tests of a few months duration.

The partial-replenishment test can be an important tool in exploring the long-term leach behavior of glasses. In order to demonstrate the use of partial-replenishment test data in modeling, experiments were conducted which show that after a relatively short period test results become largely independent of exchange interval and volume exchange fraction. The results can be characterized by the ratio between these two parameters, which corresponds to the residence time. Sampling of the leachates during a given exchange interval shows that leachate concentrations quickly recover following the exchange, and that this recovery can be satisfactorily described using a simple model based on first-order approach to equilibrium at the silica saturation level.

Glasses suitable for high-level nuclear waste vitrification must meet a number of requirements including processability, durability, and phase stability. Our new data indicate that the West Valley Reference 5 composition meets these requirements: it is phase stable over the expected range of melter temperatures and residence times, and only minimal (<2 vol %) secondary phase formation, consisting predominantly of iron-group spinels, is expected in the canister cooled glass. Leach tests have shown that the durability of Reference 5 is significantly less sensitive to spinel formation than the earlier Reference 4. However, natural process variations during production will result in a range of glass compositions around the nominal composition. In this paper we report on the effects of composition changes around Reference 5 on the phase stability upon heat treatment and the consequent effects on chemical durability. Since variations in the waste-stream levels of most major components can be accommodated by adjusting the quantities of the glass-forming additives we focus here on a group of components for which this is generally not the case: Ca, Ce, Cr, Mn, Ni, S, and Fe. While many of these components are present at below 1 wt % they could have potentially significant indirect effects on product durability due to enhancement of secondary phase formation. The results provide a data base to support broad (≥300%) tolerance ranges to variation of levels of most of these components. Viscosity data are also reported.

Samples of a fully simulated nuclear waste glass under consideration for use in the West Valley Demonstration Project were isothermally heat treated and studied by scanning transmission electron microscopy (STEM). Both fully oxidized and partially reduced specimens were heat treated for 3 hours at 600°C, 700°C, 800°C and 900°C and for 45 days at 10°C above Tg (447°C -461°C) and at 10°C below TB (427°C -441°C). Microstructural features in untreated as well as the heat-treated glasses were studied using STEM. Analysis by energy dispersive spectroscopy (EDS) was used to identify the elemental compositions of the features observed.

The predominant crystalline phase present in the samples heat treated for 3 hours was an Fe3O4 type spinel with Ni and Cr substituting for some of the Fe. The spinels varied in size from 0.02 to 10.0 µm, the larger crystals present at higher heat-treatment temperatures. Long term heat treatment above Tg resulted in the formation of small iron-containing crystals ranging in size from 10 to 35 nm. Heat treatment below Tg produced no crystalline or amorphous phases.

The Defense Waste Prsocessing Facility (DWPF) at the Savannah River Site (SRS) will be the first production facility in the United States for the immobilization of high-level nuclear waste. Production of DWPF canistered wasteforms will begin prior to repository licensing, so decisions on facility startup will have to be made before the final decisions on repository design are made. The Department of Energy’s Office of Civilian Radioactive Waste Management (RW) has addressed this discrepancy by defining a Waste Acceptance Process. This process provides assurance that the borosilicate-glass wasteform, in a stainless-steel canister, produced by the DWPF will be acceptable for permanent storage in a federal repository. As part of this process, detailed technical specifications have been developed for the DWPF product.

SRS has developed detailed strategies for demonstrating compliance with each of the Waste Acceptance Process specifications. An important part of the compliance is the testing which will be carried out in the DWPF. In this paper, the bases for each of the tests to be performed in the DWPF to establish compliance with the specifications are described, and the tests are detailed. The results of initial tests relating to characterization of sealed canisters are reported.

Leaching tests were carried out on well-characterised Synroc C specimens containing 10 wt% simulated HLW. The effect of temperature and leachant composition on the mechanisms and kinetics of leaching were investigated by leaching Synroc specimens in deionised water at 70°C and 150°C, and in silicate and bicarbonate solutions at 70°C. SEM and TEM were used before and after leaching to characterise the primary and secondary phases in and on the Synroc samples, and to describe the morphological and chemical changes which occurred. Leachants were analysed using ICP/optical emission spectroscopy and ICP/mass spectrometry.

Titaniferous surface layers were observed on the surface after exposure to leachants. Analysis of the leaching data has established that these surface layers can reduce the release of HLW elements from Synroc and that variations in temperature and leachant composition can hinder or promote the formation of secondary phases.

The degradation and dissolution of spent nuclear fuel in groundwaters occurs by three partly concurrent processes: dissolution of mobile species which have migrated to the fuel rod void, grain boundary attack and matrix dissolution. Of the first two processes, both clearly dependent on the irradiation history of the fuel, the first is fairly well understood, but there is currently insufficient data on the migration of key nuclides to grain boundaries during reactor operation and their release kinetics during subsequent water corrosion to enable satisfactory modelling of this process to be performed.

This paper presents results from recent experiments included in the Swedish programme which are relevant to this issue.

The dissolution behaviour of Zircaloy clad used CANDU fuel in aqueous solutions has been examined under mildly reducing conditions at 95°C for a period of 16 months. The effects of various container components and groundwaters on the dissolution of the fuel have been investigated. The components studied were titanium (container material), carbon steel (fuel bundle support basket) and soda-lime glass beads (container infill material). Leaching solutions included deionized water (DIW) and synthetic saline groundwaters.

The presence of glass beads or carbon steel did not appear to significantly affect solution concentrations of 137Cs, 90Sr, 99Tc, or 238U. Radionuclide concentrations were one to two orders of magnitude higher in saline groundwaters than in DIW. After about 30 days leaching, continued radionuclide release from the fuel was strongly inhibited in the absence of oxygen, and radionuclide concentrations in solution remained virtually constant. X-Ray photoelectron spectroscopy (XPS) measurements of the oxidation state of fuel fragments leached in groundwaters, showed a surface composition of less than UO2.33, below the postulated threshold for oxidative dissolution.