Modelling the leaching of fractured SON 68 glass block Laure CHOMAT, Frédéric BOUYER*, Stéphane Gin CEA, Marcoule DEN/VRH/DTCD/SECM/LCLT BP 17171 30207 Bagnols-sur-Cèze Cedex Stéphane ROUX LMT Cachan, 61 avenue du président Wilson 94235 Cachan Cedex Abstract : experimentations carried out on SON 68 glass model cracks in static condition, show a strong coupling between transport and leaching, depending to a form factor (aperture). Moreover, convective transport induced by gravity was observed in vertical model cracks, whereas only diffusion was detected in horizontal model cracks. Besides, an original device was developped to study the impact of gradient temperature, as a convective driving force, on alteration. These experimentations allow to build a model of the alteration in simplified cracks. First of all, the transport is deduced from Poiseuille’s law. The so evaluated velocity is introduced in a porous geochemical software (HYTEC1), which allows us to model the alteration. This model development has been applied to describe alteration within simple silicate glass cracks. It will then be extended to study all simple compositions glasses, as well as R7T7 glass model cracks, and more generally network of fractures. Context In France, vitrification is the process used for the immobilisation of HL radwastes. A large amount of glass with radioelement is cast in iron canister. During the cooling process in those canisters a cracks network appears in the glass block, increasing the developed surface. When considering the geological disposal concepts, this aspect may have a great influence on the long term behaviour, which must be ensured during several tens of thousand years. Indeed, exposed to water, the altered glass releases a total amount of elements proportional to the reactive surface. Moreover, the glass dissolution is a complicated coupling between incongruent leaching, secondary phases precipitation and transport. The study of the glass block appears to be too complex and shows the necessity to consider simplified object to determinate a model describing the elementary phenomenon involved in cracks. Thus, experiments were conducted on single ideal cracks in different configurations and in different alteration conditions. Besides, the use of modelling aims highlighting the different transport mechanisms and its coupling with chemical reactions in cracks, and finally will help to predict the behaviour of real cracked media. Experiments Static alteration condition on a single model crack The experiments were conducted on SON 68 glass and consisted in leaching simplified model cracks (two well polished glass pieces of 25 * 25 mm2 area, separated by a calibrated Teflon rubber or polyamide yarn) maintained in vertical or horizontal position. The alteration condition is NaOH 0.25 mol/l at 90 1 °C for all experiments. The use of basic * corresponding author (frederic.bouyer@cea.fr) pH permits to enhance chemical reactivity2, and to make observable the coupling between chemistry and transport. The chart below presents the range of experimentations carried out, indicating apertures and position of the studied model crack. Vertical : Top aperture (m) Bottom Horizontal : Right aperture (m) Left free 60 40 40 free 80 60 60 free 160 90 90 free 220 220 220 free 550 550 550 40 40 110 80 60 60 200 100 60 120 Table 1: apertures studied according to the position imposed to the model crack (until a < 160 m, uncertainty of measurement is 10 m, above it’s 20 m) When the upper aperture is indicated as free, only one clamping device was applied on the bottom of the vertical model crack so that the upper aperture was not ensured to be at the same bottom value. The first studies carried out on 60 m aperture model crack show a significant impact of the transport on alteration within cracks. This will be pointed out by an altered glass layer thickness deeper on the exit edges than in the middle of the crack (see figure 1). The new phase resulting from the incongruent glass leaching and silicon recondensation is regarded as the glass altered layer and was observed by SEM. This figure depicts different transport mechanisms, depending on the position vertical or horizontal of the fracture. 45 Top of the crack Bottom of the crack 40 altered glass thickness (m) 35 30 25 20 15 10 5 0 0 5 10 15 20 25 situation in crack (mm) horizontal model crack vertical model crack Figure 1: altered glass thickness according to the position of the crack and the situation into, measured by SEM (case of the vertical /horizontal ideal fractures, with an aperture of 60 µm) The dissymmetric altered glass layer thickness profile, observed in vertical model crack, indicates that a convective transport mechanism prevails, whereas in horizontal crack the Brownian diffusion mechanism induces a symmetric profile. Besides, the minimum alteration thickness, where local saturation is reached very rapidely, is close to the bottom of the cracks, indicating that a flux goes from the top to the bottom of the crack. The results of the whole range of experiments confirm theses assessments and the type of altered glass layer thickness profile for each case are depicted on the following synoptic figure. Horizontal crack : athreshold : 220-220 / 550-550 µm a 550-550 µm a 40-40, 60-60, 90-90 220-220 µm a 80-110, 100-200 µm symetric flat Vertical crack : athreshold : 160-free / 220-free µm a 60-free 90-free, 160-free µm 40-40, 60-60 µm a 60-120 µm flat dissymetric a 220-free, 550-free µm Figure 2: synoptic figure presenting different altered glass layer profile as a function of the aperture Those different experiments show also that the coupling between chemistry and transport appears until a threshold aperture is reached, which is between 220 and 550 m or 160 and 220 m for respectively horizontal or vertical model crack with a length of 25 mm. So, the significance of the coupling between transport and chemistry depends of a form factor. Though, model cracks presenting variable aperture according to crack length indicate the same shape profile than these expected for constant aperture crack in the same position. So, a variation of aperture until 100 m, inducing different local leaching conditions, can’t explain the dissymmetric profile. A convective transport appears to be the only way to understand that phenomenon. Besides, experiment conducted on a vertical model crack and analysed by ToF SIMS confirmed the dissymmetric profile in pure water alteration condition : the way of accelerating reactive processes (i.e. the use of NaOH for the alteration of SON68 glass) does not affect these conclusions. Therefore, the whole range of experiments described previously demonstrate in an original way that, in vertical configuration, convective transport induced by density gradient must be considered. Compared to gravity, temperature gradient seems to be an unlikely cause of convective transport, as experiments were performed in drying furnace. Nevertheless, it is expected into glass block. That’s the reason why its influence has been also studied on model fracture. Thermoconvection experiments To evaluate the impact of a temperature gradient, as a convective velocity motor on alteration within model crack, a leaching device was especially designed. It allows producing a convective transport between two cells filled of pure water and maintained at different controlled and regulated temperatures. The linkage between these two cells consists in two model cracks with a fixed aperture separated by around 9.4 cm high; consequently an equal but opposite velocity is produced into these two model fractures. The first range of parameter introduced in this experiment was a temperature gradient of 4-5 °C and an aperture of 60 m. It’s difficult to maintain the temperature gradient with an one degree accuracy, because of the variation of the room temperature. So, the temperature gradient imposed as well as the evolution of a KCl tracer introduced in the hotter vault were indexed. These data indicate that the temperature gradient is about 4 °C during the first ten days, then increase to achieve a constant value of about 5 °C after twenty days. The tracer analysis points out a velocity around 2-0.7 10-5 m.s-1. After one month of leaching, although the study of the altered glass layer leads to thickness measurement difficulties and interpretation, the analysis of the altered glass layer into the two cracks shows an undeniable presence of convective transport, as expected (a thermo-convective velocity is imposed). Another experiment with an aperture of 82 m and a temperature gradient of 5 °C was performed. The KCl tracer evolution indicates a velocity of 10-4 m.s-1 and the altered glass layer thickness will be soon analysed. Afterwards, an aperture of 200 m will be considered with the same temperature gradient. The results of these experimentations are particularly interesting for modelling, because the alteration condition is well defined. Modelling The interest of modelling comes from the possibility to describe what occurs within fractures (so at the fracture scale), that is chemical reactivity, flow and species transport. This model, validated on our experimentations, will allow us to simulate what happens at large time scale as well as on crack network at the glass block scale. Transport modelling When considering just a model crack, it’s not required to refer to Navier-Stokes equation, Poiseuille’s law equation is fully sufficient. The lubrification approximation and a voluminal laminar stationary flow of a Newtonian fluid can be considered as good approximation in our cases. In this approach, the velocity profile in the crack is parabolic and depends on the density variation, due either to temperature gradient or gravity. In the case of gravity, the application of the Poiseuille’s law is more complex, because we need to take into account the coupling between chemistry and transport to evaluate the accurate difference between the element concentrations in and out of the crack. Neverthless, simple assumptions can give good order of magnitude of the velocity induced. This model can notably predicts a significant velocity (of the order of 10-4 m.s-1) even for little density variation (concentration variation). On the other hand, in thermoconvection experiments we only consider the impact of temperature on Poiseuille’s law. The velocity obtained for an 60 m aperture is of the order of 3-4 10-5 m.s-1, which is a little higher than the measured velocity thanks to KCl tracer in the experiments. Moreover, this method gives also a velocity for 82 m aperture near to 8 10-5 m.s-1, closed to the experimental value. This simple Poiseuille’s law is in good agreement with the experimental values and is sufficient to evaluate the convective driving force. Geochemistry modelling The modelling of glass alteration is very complex and requires to consider incongruent leaching, secondary phases precipitation and transport. At the present time, no software can associate Navier-Stokes resolution with a complex chemistry solver, because of very different time scaling in both. As transport in model crack can be easily apprehended by Poiseuille’s law, the use of a porous geochemical software (HYTEC 1) to describe the complicated chemical reactions involved in glass leaching will be viewed as the best compromise. HYTEC is the association (coupling) of a chemical module and a transport module. Chemical reactions are based on thermodynamic equilibriums and an abundant database is also available (thermodynamic and kinetic reactions can be easily added). The transport can be represented by diffusion and by advection, but only by the mean of Darcy’s law. We overcome the problem of the use of a geochemical model on simple ideal fractures thanks to some tricks. Firstly, the surface interaction has been allowed by limiting diffusion transport into glass medium to solid diffusion and by imposing a high contrast of permeabilities between the glass and the species medium. Furthermore, the glass dissolution, for which the kinetic’s law is depending on an inhibitor term, induces a secondary phase (socalled gel) thermodynamically more stable than the glass itself. The velocity, calculated from the Poiseuille’s law, has been introduced as a constant value within the geochemical model. Kinetics describing the glass alteration as well as the gel appearance for a simple pure silicate glass have been investigated, in order to reproduce the gel phase formation, as well as alteration profile within a fracture. The introduction of a motion simulating a convective flow allows us to predict dissymmetric profiles. These first results convince us to pursue in that direction in order to model SON 68 glass alteration, for which only chemistry has to be refined. Conclusion The original experiments realized on SON 68 model cracks in static alteration conditions show up that transport has a strong impact of alteration in model crack, until a threshold form factor is reached. Besides, gravity induces a significant convective driving force responsible for a dissymmetric altered glass layer thickness profile in the vertical model fracture. Moreover, a special device permits us to study the effect of the flow motion induced by gradient temperature (assumed to be present in glass block) on leaching in model cracks. The experiments point out that a temperature gradient of some degrees produces a velocity high enough to impact on the model cracks leaching. All those results enforce the knowledge on transport mechanism and on the coupling between transport and chemistry, which is the basis of the model crack leaching modelling. The use of the Poiseuille’s law and a modified geochemical model allows us to simulate what occurs within simple ideal fractures. This methodology will soon be applied to glasses of industrial interest for describing and understanding phenomena involved in model crack, and afterwards in network glass block. Jan van der Lee, Laurent De Windt, Vincent Lagneau and Patrick Goblet, Computers & Geosciences 29, p. 265-275 (2003).. 1 2 Solange Ribet, Stéphane Gin, Journal of Nuclear Materials 324, p. 152-164 (2004).