Abstract
(Englisch)
|
Long-term corrosion tests have been conducted over a time span of two years to study the degradation of vitrified nuclear waste in the presence of humid clays. Experiments carried out in the past clearly indicated that the degradation of the French nuclear waste glass R7T7 is enhanced in the presence of clays, which will be present as a major component of the engineered barrier or of the host-rock formation in future radioactive waste repository systems. Current Swiss repository designs foresee the disposal of vitrified high-level radioactive waste (using the same glass type investigated in this study) in a clay-rich environment.
In a preliminary modelling study, the enhanced degradation induced by the presence of clay was interpreted to derive from sorption of silica on the clay mineral surfaces. Model calculations performed assuming instantaneous reversible sorption of silica suggested that such a process could be detrimental in typical repository situations, as it could reduce considerably the lifetime of vitrified radioactive waste. The results of these calculations, however, were found to depend strongly on the (highly uncertain) silica sorption parameters.
The main objective of this investigation was to identify the mechanism causing the observed enhancement of glass corrosion and to provide accurate parameters to describe this effect quantitatively. One major aim was to determine whether sorption or precipitation of silica controls the kinetic behaviour of the glass in contact with water-saturated clay.
The two clays used for this investigation, a smectite-rich clay ('smectite 4a') and an untreated argillaceous rock ('Boom clay'), were first conditioned with distilled water at 90oC for one month in order to reach a state of chemical equilibrium. Cylindrical samples of the French R7T7 glass were weighed accurately and then placed in the middle of steel vessels. The vessels were filled with the pre-conditioned clay paste, then closed tightly and placed in the oven at 90 oC. At regular intervals (after 1 month and subsequently each 3 months) a single test was stopped and the cleaned glass sample re-weighed to determine the mass of glass leached. The pore solution was separated from the clay for chemical analysis.
The data for the tests conducted with smectite 4a show that the leaching behaviour of the R7T7 glass is compatible with the assumption of moderate silica sorption on the clays. Fitted distribution coefficients (Kd-values) lie close to 0.02 m3 kg-1 and are similar to those determined in independent experiments carried out with the same glass and Boom clay by the Belgian partners (0.005 to 0.02 m3kg-1). The arrangement of the data points allows us to exclude precipitation of a pure silicium dioxide phase as a mechanism controlling the kinetics of glass dissolution.
The data concerning the tests carried out in the presence of Boom clay agree only partially with the model predictions. In this case, two clusters of data must be distinguished: one with glass mass losses exceeding ~ 90 g m-2 and the other with mass losses clearly below this limit. These two groups are found to correlate with the pH of the interstitial pore water: glass dissolution was clearly much weaker in all cases for which a pH of less than 7 was measured, compared to the tests where a consistently uniform pH of 7.7 was determined. The low pH values are probably due to oxidation of pyrite, which is present in variable amounts in the Boom clay (pyrite oxidation reactions produce sulphuric acid). The slower corrosion measured in connection with acidic interstitial solutions is consistent with the well-established decrease of glass dissolution rates with pH in the near-neutral region.
Due to the variation in pH, it was therefore not possible to fit all the data with a single distribution coefficient and the two clusters had to be modelled separately. The high pH data can be fitted reasonably well by assuming a 0.05 m3 kg-1 sorption isotherm, which is in the same order of magnitude as the Kd extrapolated for smectite 4a. The low pH data are best reproduced by a simple model corresponding to a constant glass corrosion rate of 0.13 g m-2d-1. This value exactly matches the theoretical initial rate obtained for Boom Clay at pH < 7 and implies that sorption coefficients for silica are very high (although the extent of corrosion is smaller than for the tests producing alkaline pore waters). This result agrees with the systematic increase of the Kd with the iron content in the clay, observed during the sorption experiments carried out at SCK-CEN and attributed to strongly sorbing ferric oxy-hydroxides formed through pyrite oxydation.
A serious inconsistency arises from the unexpectedly high silica concentrations measured in the interstitial waters for both smectite 4a and Boom clay experiments. Final concentrations exceeded the 10-2 M level, which is well above the concentrations predicted by our model (~ 2 x 10-3 M for smectite 4a). The high silica levels can be reconciled with model calculations if it is assumed that silica in solution forms complexes with organic ligands supplied by the clay. This hypothesis cannot be tested with the presently available data but seems reasonable considering that 1-3 weight % total organic carbon, mainly in the form of humic and fulvic acids, are present in the Boom clay. However, if complexation of Si is assumed as an explanation for the high silica concentrations, mass balance considerations require that the distribution coefficient be reduced by one order of magnitude (i.e. from 0.02 to ~ 0.002 m3/kg for smectite 4a).
In conclusion, this work allowed us to constrain distribution coefficient for silica on two representative clay buffer materials within the range 0.002- 0.05 m3/kg. Preliminary calculations for a generic repository environment using these values for Si sorption indicate that reductions in the lifetime of the vitrified waste forms would be minor and will have no significant influence on the repository performance.
|
