In Sweden and Finland the KBS-3 concept is planned to be used for disposal of spent nuclear fuel. A canister consisting of a cast iron insert for mechanical strength and a copper shell with a thickness of 50 mm as corrosion protection, is planned to be used. The canister will be embedded in a bentonite buffer backfill and deposited in granitic bedrock at depths greater than 400 m. Assessments of post-closure safety have to include the early corrosion by initially entrapped oxygen and by radiation, while sulfide is the major corrodant in the long term. Chloride is generally considered “beneficial” as chloride ions promote active dissolution rather than passivation (which might result in localised or “pit” corrosion), even if chloride ions will increase the concentration of copper in solution by the formation of copper chloride complexes. A higher concentration of copper in solution at the canister surface would be a driving force for corrosion because oxidised copper might be transported away from the canister. An assessment of such chloride-assisted corrosion is thus relevant in a post-closure safety assessment.
Earlier assessments have been performed by Posiva and SKB up to a chloride concentration of 1.7 mol/kg, at temperatures up to 100 °C. In this study, the aim has been to investigate chloride concentrations up to 5 mol/kg, even though the groundwater and bentonite porewater concentrations of chloride are not expected to exceed 1 mol/kg. In addition, the updated copper concentrations have been used in an assessment of canister corrosion. The new thermodynamic calculations were performed with updated software (PHREEQC) and databases, and a comparison with experimental data was performed before the database was chosen. The following database comparison steps were taken:
All copper species taken into account in different databases were listed, which revealed that the dominating complex CuCl32− was not included in all databases.
The activity coefficient models were compared, including the specific treatments in the different databases. It was noted that the uncertainties originating from the choice of activity coefficient expressions, may be as large as the uncertainties in the equilibrium constants used.
Experimental data on the solubility of CuCl(s) in varying chloride concentrations at 25 °C were compared to calculated values with the different databases.
From this exercise, it was decided to use the LLNL database provided with the PHREEQC software distribution, as it showed the best conformity to experimental values at high chloride concentrations. Values for the temperature dependencies (enthalpy change) for the complexes CuCl2− and CuCl32− were taken from the Mintex.v4 database, as these were lacking in the LLNL database.
For the assessment of corrosion on a KBS-3 canister, the assumption was made that transport of copper-chloride complexes away from the canister is determining the rate of corrosion. Three different mass-transport regimes were considered representing (i) intact bentonite and rock without fractures, (ii) with a rock fracture intersecting the deposition hole, and (iii) the case of partially eroded bentonite. A threshold corrosion rate of 10 nm/year was defined, simply based on the argument that even if high chloride concentrations were present throughout the one-million year assessment period, the resulting depth of corrosion should only amount to 10 mm or approximately 20% of the available corrosion allowance. There would, therefore, be a substantial allowance for corrosion by sulfide. The calculations show that, for groundwater chloride concentrations up to 5 mol/kg, the corrosion rate is less than 10 nm/yr, for all three mass-transport conditions regardless of the temperature (up to 100 °C). Corrosion from elevated groundwater chloride conditions is thus considered not to significantly impact the canister lifetime.
