When the canister temperature falls below 100C, water may evaporate from the hotter interior of the glass and condense on the inside canister surface. The amount of water in the glass (about 40 grams) is significant in this respect: the surface density of water necessary to initiate corrosion is only about 0.01 grams per square meter of surface, whereas the free internal surface of the container is about 0.3 square meters. Thus only about 3 mg of water, or less than 0.01% of the inventory, is necessary to wet the whole free surface. There also may be a risk of stress-corrosion cracking on the inside surface due to the presence of hydroxyl ions in the VHLW.
It is clear from the above discussion that the quality control measures for such procedures as welding and decontamination must be significantly more stringent for sensitized stainless steels than for unsensitized ones. This is reflected in the U.S. Nuclear Regulatory Commission Regulatory Guide 1.44, "Control of the Use of Sensitized Stainless Steel." However, there is no indication that either France or Japan has enacted stringent quality control measures addressed specifically at sensitization.
Other accident scenarios for which canister sensitization may have an impact on the outcomes are:
Canister drop accident.
There is evidence that the resistance of stainless steels to fracture upon impact can decrease substantially if it becomes sensitized. This would affect the outcome of the "canister drop" incident in which a canister is accidentally dropped to the bottom of the storage tube when it is being loaded or unloaded. Also, the canisters would be more susceptible to fracture in the event of an incident (such as a plane crash or an earthquake) that would damage the integrity of the storage facility.
Flooding of the storage facility.
In the event that the storage facility were flooded with sea water (following a tsunami, for example), sensitized canisters would corrode much more rapidly than unsensitized ones. If the flood waters receded slowly, widespread failure from stress-corrosion cracking would occur within a matter of weeks. This would greatly increase the costs and risks of the clean-up operation and could lead to a significant release of radioactivity into the environment.
The issue of VHLW canister sensitization took on greater urgency when it was revealed in August 1995 that one of the 28 canisters sent from France to Japan had failed one of the inspection tests performed by Japanese authorities.10 In this incident, an unusually high emission rate of a fission product, cesium-137 (Cs-137), was observed for a number of days during a confinement test, in which the canister was placed in a vacuum chamber and the radioactivity deposited on the outlet filters of the chamber was measured. This raised the troubling possibility that the canister indeed had developed a leak.
The contention that this event could be the signature of a leaky canister was quickly denied by JNFL, who claimed instead that it was due to "loose contamination" on the surface which was somehow dislodged during the test procedure. However, the arguments they originally offered to support their position have been shown to be faulty.
For instance, it was claimed that if Cs-137 were leaking from within the canister, then another volatile fission product, ruthenium-106 (Ru-106), would have been observed as well, but was not. However, the ratio of Ru-106 to Cs-137 is very low in the VHLW (0.002), and one can show from the published radionuclide inventories that the amount of Ru-106 that one would expect to observe would be very close to or even below the detection limit of the equipment. Thus the fact that Ru-106 was not detected does not mean that a leak can be excluded as the source.
In the Expert Advisory Committee Report, it is stated, without any supporting data, that the concentrations of radioactive cesium and ruthenium are expected to be the same in the plenum (air space) at the top of the VHLW canister. This statement is completely wrong and contradicts all available information about the relative volatilities of Cs and Ru from VHLW, including the results of a study by the Japan Atomic Energy Research Institute (JAERI).11 In this study, it was demonstrated that Ru is about 20 percent as volatile as Cs. Combining this with the relative inventories of Cs-137 and Ru-106 in the VHLW canister, the concentration of Ru-106 in the plenum should be about 0.04% of the Cs-137 concentration, or a factor of 1 / 2500 smaller than the Expert Advisory Committee figure.
JNFL also maintained that their observation that the Cs-137 emission eventually decreased to "normal levels" implies that the emission did not originate from a leak. This conclusion is inaccurate. Repeated application of the vacuum test could have caused a leak to become plugged with fines, the small glass particles that are adjacent to the inner surface of the VHLW canister.
In the Expert Advisory Committee Report, a new piece of information was offered to the public for the first time. It was claimed that following the release of the initial inspection report, JNFL took measurements of another radionuclide, europium- 154 (Eu-154). JNFL apparently found that the ratio of Eu-154 to Cs-137 found in the confinement test filter paper was the same as the ratio present in the surface contamination and the ratio in the original liquid waste. They then claimed that this result was proof that the anomalous confinement test result was due to surface contamination and not a leak, since Eu-154 is not volatile and would not be expected to leak from the interior.
This conclusion would follow if the only material that could leak from within a VHLW canister was the vapor present in the air space in the canister. However, minute particles of the glass matrix could be leaking as well. Also, there are some puzzling aspects of the JNFL result: for instance, one would not expect that the Eu-154/Cs-137 ratio would be the same in the surface contamination as in the original high-level waste, since Eu-154 and Cs-137 have different vapor pressures and would condense in different amounts. Thus it is far from clear that the Eu-154 result, while interesting, is conclusive.
Instead of trying to argue away the problem, JNFL and STA should agree to a broader, peer-reviewed investigation being launched into whether the cesium contamination incident is an ominous signal that accelerated corrosion of the VHLW canisters, resulting from sensitization, is taking place. This would entail taking apart a number of VHLW canisters and analyzing the interior surfaces for signs of corrosion. Unless this is fully resolved, one cannot exclude the possibility that more serious leaks, or the complete failure of a VHLW canister during a storage accident, will occur in the future.
The SUH 309 steel sensitization problem described above is only one indication of the problems that Japan may experience as it faces the need to manage high-level radioactive wastes which have been conditioned overseas with a process of which it has limited knowledge and over which it has limited control. JNFL's continuing refusal to address the sensitization issue in a constructive way is in direct contradiction with the safety agreement it concluded with Aomori Prefecture. The governor of Aomori should not allow any more VHLW to be imported until Cogema, BNFL and JNFL have agreed to the peer-reviewed investigation mentioned above, and have directly demonstrated the integrity of the canisters that must be relied on to contain this extremely hazardous material for the indefinite future.
The primary conclusion of this paper is that sensitization of VHLW canisters is a wholly undesirable and completely avoidable occurrence. It without a doubt increases the risks associated with VHLW storage, by rendering canister integrity more vulnerable to changes in environmental conditions that are difficult to accurately predict. Japanese authorities should not allow any more sensitized VHLW canisters to enter Japan. Furthermore, they should request that COGEMA switch to a material, such as one of the stabilized stainless steels, which is much more resistant to sensitization during VHLW processing.
11. H. Kamizono et. al., JAERI, "Volatization of 137 Cs and 106 Ru from Borosilicate Glass Containing Actual High-Level Waste," J. Am. Ceram. Soc., 22 (1989), 1438. Back to document