The Theory of Radiation Damage in Salt Crystals and Rocks, the Leading Questions and Underlying Research Issues
A. Garcia Celma, L. Vons and H. Donker
Excerpt from: EUR 16743 EN, Part I.3
2. PROBLEM DEFINITION: THE BASIC THESES
The general question leading our research is: Will radiation damage in rock salt endanger the containment of the waste in radioactive waste repositories?
To answer this question is not easy, because experiments for the length of time during which gamma radiation above the natural background will be present in a repository are impossible to carry out. Experimentators have to choose between reproducing the total dose to be expected in a repository by using irrealistically high dose rates or to reproduce the dose rate and never reach the total dose.
Moreover, since dose rate and total dose do not relate linearly with the obtained radiation damage we have to rely on rather complex computer simulations based on theoretical models to predict the behaviour of a repository. To develop the computer simulations, the processes implied in radiation damage development have to be well understood. One of the best known theoretical models is that of Jain and Lidiard [Jain and Lidiard,1977]. Modifications to the Jain-Lidiard model had been introduced before we started this research [Lidiard, 1979; Van Opbroek and den Hartog, 1985], and some other modifications were suggested and/or implemented during these years. Most modifications and theory development during our research are either the work of the Groningen University [Groote and Weerkamp, 1990; Seinen et al., 1992; and Seinen, 1994] or are the result of our work [Soppe, 1993; Soppe and Kotomin 1994; Soppe and Prij, 1994 a, and 1994b; Soppe et al., 1994 and Donker et al., in prep]
At the beginning of our research little was known about the factors and/or functions that relate the experimentally obtained results, at high dose rates, with those expected under repository conditions, at low dose rates. Our research was therefore aimed at controlling the validity, for the case of a repository, of the thesis underlying the safety problems which could be envisaged as a result of radiation damage in rock salt. An essential part of the experimental programme consisted of narrowing the gap between the experimental irradiations and the prospective repository irradiations of rock salt in what respects dose rates and total doses. Figure 1 shows that the conditions of irradiation in the HFR experiments are nearer to repository conditions than ever was the case in a long lasting laboratory experiment.
The theses considered can be resumed as follows: -
a) all other conditions being constant, low dose rates of gamma rays produce more damage than high dose rates until a maximum is reached and this efficiency decreases again. Repository conditions are thought to lie in the dose rate regime where the efficiency increases. This thesis stems from theoretical models [Van Opbroek and Den Hartog, 1985; Groote and Weerkamp, 1990].
b) crystal defects -either mechanical (dislocations), or chemical (impurities)- enhance the efficiency of radiation damage development. All other conditions constant, impure and br strained crystals ought to develop more damage than pure undeformed crystals. According to this, and to the fact that natural rock salts are always deformed and impure, natural rock salts .ought to contain more stored energy than the pure single crystals used in most experiments. This thesis derives from interpretations of experimental work performed by [Groote and Weerkamp, 1990; Compton ,1957; Ikeda and Yoshida, 1967; Levy et al.,1980, and 1981; Den Hartog ,1988 and Den Hartog et al., 1990].
c) radiation damage in a repository will not reach a saturation level but will grow until a percolation limit is reached and an instantaneous back reaction takes place. This thesis was poned at the Groningen University [Groote and Weerkamp, 1990; Semen et al., 1992; Den Hartog et al., 1990 and Weerkamp et al., 1994] and is based on the interpretation of experimental results on NaC1 doped with K and on the predictions of the original Jain-Lidiard model.
d) if the huge amounts of stored energy that were found in crystals irradiated at the Groningen University are taken into account "explosive back reactions" could threaten the integrity of a repository [Den Hartog et al., 1990; Prij, 1991 and number 2 and 23 in this volume]. This is based on rock mechanics calculations performed at the Solid State Physics Department of the Groningen University. e) neither fluid assisted recrystallization, nor other sorts of creep mechanisms will act in any significant way during the active life of a repository. Therefore the influence of these mechanisms in reducing stress (and consequently, stored energy), although recognized when discussing the rheological properties of a repository [Spiers et al., 1986 and Urai et al., 1986] is absolutely ignored when discussing radiation damage.
Theory development from which these basic theses stem, had been mostly based on experiments on pure (or containing controlled amounts of impurities) undeformed single crystals irradiated at atmospheric pressure (or in vacuum). These crystals and conditions are, however, very different from the impure, polycrystalline and deformed rock salt irradiated at the lithostatic pressure expected to occur in a repository. Although we aimed at reproducing repository conditions as nearly as possible, samples such as pure undeformed monocrystals and irradiation of samples at atmospheric pressure, were nonetheless included in the experimental plan to form a link with existing theories and to allow us to countercheck the theses. To this end a variety of samples was irradiated in two sorts of experiments: GIF A experiments, which were planned to prove or disprove theses c, d, and e, and GIF B experiments which were planned to prove or disprove theses a, b and e (GIF = Gamma Irradiation Facility, the compact used fuel element storage racks in the cooling pool at the High-Flux-Reactor in Petten, The Netherlands, Figs. 1a and 1b in A. Garcia Celma, A.J. Nolten, W.A. Feliks, H. Van Wees, "Gamma Irradiation Experiments in Natural and Synthetic Halite, p. 85 of A. Garcia Celma, H. Donker (eds.), The Effects of Gamma Radiation on Salt, European Commission Topical Report EU 16743 EN (1996).
6. DISCUSSION AND PARTIAL CONCLUSIONS
6. 1. Dose rate effects (thesis a)
The statement that low dose rates are more efficient than high dose rates in producing damage for the same total dose holds for given irradiation condition intervals, sample compositions, and microstructures. However, it does not hold for long times of experimentation in polycrystalline material, and the modified Jain-Lidiard model (1985) exaggerates the efficiency of low dose rates.
Moreover, to reach high doses at low or moderated dose rates, long periods of time are necessary and the subsequent increasing importance of crystal creep processes and its coupled anneal, increasingly hinder radiation damage development, at least when the experiments are performed at 100 C.
6.2. Plastic deformation and impurities effects (thesis b)
Plastic deformation (creep) of crystals takes place due to dislocation development and motion. Natural crystals are always deformed to a certain extent. Dislocations fixate colour centres, thus enhancing nucleation of Na-colloids during irradiation of the crystals that contain them. However, during irradiation (both in pressurized and non-pressurized samples) creep takes place and the motion of dislocations involves motion and anneal of colloids. Since NaCI creeps and adopts another microstructure quickly (it is said to have a fading fabric memory), the starting state of deformation of the samples becomes irrelevant for long experiments. Therefore, the enhancement of colloid nucleation caused by strain in natural NaCl crystals as compared to the pure undeformed single crystals of the theories is only valid for relatively short experimentation times, i.e. either low total dose or very high dose rate experiments. The enhancement of damage by deformation is probably irrelevant for long periods of time and low dose rates.
Nonetheless, the existence, and therefore the development of new grain boundaries affects radiation damage stabilization, at the microscopic level. The grain and subgrain boundaries are the most efficient diffusion paths in the samples, and the differences in concentration of mutually annihilating defects which arise from the different diffusivities towards these big diffusion channels where they all disappear, are at the origin of damage heterogeneity and consequent stabilization.
Regarding chemical crystal defects, such as lattice impurities, only the effect of brine as lattice impurity has been evidenced by the experiments. OH- ions as lattice impurity, ease Na- colloid nucleation, but also seem to enhance anneal at longer periods of time. This is probably due to enhanced dislocation mobility produced by OH- ions as lattice impurity.
Since huge amounts of time are involved to reach large total doses at low dose rates, the effect of both interstitial brine and dislocations is that they reduce the efficiency of irradiation in damaging the rock salt for increasing times (and total dose) at the same dose rate. This also implies that there is a natural limit to the enhancement of damage caused by low dose rates, since at a given dose rate the time needed to produce damage has to be longer than the time needed for creep in natural situations.
Intracrystalline creep by irradiation has been more intense in the natural rocks than in the pure single crystals of NaCl. This justifies that the most damaged parts of natural crystals do not contain as much stored energy as the pure undeformed single crystals, at least for long experiments. The difference in creep properties has been attributed by us to the different density of preferrent diffusion paths between these two sorts of samples. The diffusion path network is denser in natural rocks which contain many grain and subgrain boundaries.
6.3. Saturation of damage in natural rock salt (thesis c)
For the Sp-800 samples irradiated in GIF A, damage saturation has been reached at 140 J/g, or in other words, in the conditions of our experiments only less than 2 mol% of the natural salt can be decomposed by irradiation. This saturation of damage is also predicted by the last versions of our models for pure undeformed NaCI, however, our experiment shows that saturation also occurs in natural rock salts. Note that 12 mol% decomposition is the lower limit required for spontaneous sudden back reactions taking place.
Radiation damage saturation, as all saturations phenomena, has to be described as due to an enhancement of the back reaction with increasing total product of the forward reaction (damage). It is therefore dependent on the concentration of defect present and on the rate of production of new defects, and thus in our case it ought to be dependent on the dose rate. An important difference between our experiments and other experiments in which much higher damages were reached is that in these last experiments the dose rate was about a factor 20 higher than in our experiment. The processes which could bring about this enhanced back reaction are discussed in 6.4.
It is important to emphasize in our GIF A experiment the saturation of damage cannot be assumed to be due to fluid assisted recrystallization. The duration of the experiment was such that it has to be assumed that at the moment in which saturation of damage was reached the brine present at the start of the experiment had already been consumed. Note that in this experiment the rate of colloid development at the start of the experiment was higher than the rate of fluid assisted recrystallization. Fluid assisted recrystallization consequently took place on crystals which had already developed colloids and brine was decomposed, possibly even completely decomposed after various fluid assisted recrystallization episodes. This is confirmed by the fact that during the DTA measurements of the samples irradiated to high total doses no mass loss is observed, contrary to samples irradiated to low total doses where mass loss due to the evaporation of water has been observed. This would mean that even in absence of H20 damage saturation is reached.
6.4. Explosive vs. gradual back reactions (thesis d)
The saturation of radiation damage must be due to an increased rate of anneal mechanisms at high damage levels. Whether this increasing rate of the back reaction is only due to the shorter diffusion distances which have to be surpassed by the F- and H-centres in order to meet an annihilating defect, or is also due to an increasing number of creep mechanisms, is not yet clear.
The behaviour of rock salt creeping as a consequence of irradiation has never been extensively studied. However, the creep behaviour of salt when the damaging agent is stress is well known. Comparing the behaviour of creep with that of radiation-induced creep, which is the same at the microstructural level, it can be doubted whether the reached damage saturation stage will be stable or transitory. Indeed, this analogy suggests that damage (and its stored energy aspect) could increase again after a given total dose has been reached.
In 'normal' creep high strain rates produce brittle failure while low strain rates induce steady state at stress levels which are lower the lower the strain rate. Extrapolation of this analogy, on the basis of the microstructural observations performed [Garcia Celma and Donker, 1994b] would imply that:
a) high dose rates will not produce damage saturation but brittle failure of the samples, perhaps the "explosive back reactions" observed at the Groningen University [Den Hartog, 1988; Den Hartog et al., 1990; 1992; 1993a, and 1993b] and,
b) low dose rates, as those expected for a repository would induce damage saturation at lower damage levels.
Anyway, even assuming that the observed saturation of damage would be a transitory state, it has to be concluded that this saturation would also have taken place if the experiment had been performed with repository relevant dose rates. This last conclusion is based on the fact that the total dose reached in our GIF A experiment (1223 MGy) is higher than that of a radioactive waste repository (276.6 MGy) [De Haas and Hehnholdt, 1989], and therefore the damage can not increase more.
6. 5. Intra and intercrystalline creep mechanisms (thesis e)
Regarding interciystalline processes, fluid-assisted recrystallization was already known to eliminate radiation damage. In our work fluid-assisted recrystallization has been proven to be able to take place in irradiated polycrystalline salt without decomposing the brine, as long as Na colloids have not yet extensively developed. It has also been shown that fluid-assisted recrystallization can occur if the brine is present as vapour, and that recrystallization can continuously proceed even on already recrystallized areas. If, before colloids can develop into a crystal, the crystal is cleaned of defects (F- and H-centres) by recrystallization, brine will never be consumed and colloids will never develop.
In short, for lower dose rates, the ratios of anneal by fluid--assisted recrystallization /damage development, as well as the ratios of anneal by intracrystalline creep/damage development increase. Therefore, in spite of the enhancement of damage for low dose rates it follows that for each low dose rate (e.g. lower than 100 kGy/h) there must be a different and lower saturation level of radiation damage (lower than 2 mol%.). This could be proven by experiments as those performed in GIF A, but with lower dose rates, e.g. dose rates starting at 100 kGy/h.
address of this page