X-ray microtomography study of pellet / powder bentonite mixture upon wetting

Among the techniques considered to seal radioactive waste in geological layers, pellet/powder bentonite mixture appears to be an interesting candidate due to its swelling properties and high radionuclide migration retardation properties. One of the key issues related to this material is its change in microstructure upon wetting. A special experiment was performed, aiming at evaluating such microstructure changes. A cylindrical column, filled with initially heterogeneous mixture, was submitted to wetting during several monthes. X-ray microtomography was used to follow the temporal three-dimensional evolution of the mixture. The pores between the pellets were found to be progressivelly filled, allowing a quasi-homogeneous material to be reached at the end.


Introduction
Radioactive waste is continuously generated in nuclear plants, and becomes a major concern for many governements, due to its long-duration activity, threatening the biomass and civilian populations.Sealing those by-products underground in the geological layers has been revealed to be an efficient technique nowadays.Among the different techniques under development, using non-compacted pellet/powder bentonite mixtures is especially interesting: easy transportation and installation, able to fill the pore initially present when submitted to wetting, due to its swelling and high radionuclide migration retardation properties.In this context, the Institute of Radioprotection and Nuclear Safety (IRSN, France) has designed and implemented the SEALEX (SEALing performance EXperiments) project (Figure 1), which the present work belongs to: in situ experiments are performed in IRSN's Underground Research Laboratory (URL -Tournemire, France) [1], while additional small-scale tests are carried out in the laboratory [2].The main goal is to quantify the sealing capacity of this method in long term.More info about this article: http://www.ndt.net/?id=23698More info about this article: http://www.ndt.net/?id=23698

Material and methods
The current work involves a pellet/powder mixture: MX80 bentonite, 80/20 proportion in dry mass, target dry density: 1.49 Mg/m 3 , provided by the Laviosa-MPC Company.Pellets geometry was cylindro-spherical (diameter and height: 7mm).Before wetting, initial void ratio e and water content were respectively e = 0.235 -0.278 and 3.17%.
A special set-up (see Figure 2), allowing experiments to be conducted into a microtomograph, was designed.It consisted of a 30 mm-thick transparent PMMA cell (inner diameter: 60 mm, height: 120 mm).The mixture was deposited inside the cell, between two porous stones, and enclosed between rigid platens, to ensure a quasi constant volume corresponding to the in-situ conditions.On upper and lower sides of the cylinder, tank waters allowed constant wetting, which resulted in mixture swelling.The volume being maintained, pressure evolution was monitored.This set-up was periodically placed into a X-ray microtomograph to evaluate the three-dimensional structural evolution of the mixture.Microtomograph was an Ultratom (RX Solutions, France).Reconstruction software was XAct (RX Solutions, France).The X-ray tube is a Hamamatsu L10801 and the imager a Paxscan Varian 2520V (1960 x 1536, pitch: 127 µm).X-ray source parameters were 160 kV and 120µA, and voxel size was 50 µm.Scan duration was 19h.To reduce beam hardening while maintaining a good signal to noise ratio, a 1.5mm thick copper filter was placed in front of the X-ray generator (note: much thicker copper filters have also been tested, which showed the necessity to greatly increase the power to maintain a good signal to noise ratio.But this decreased resolution while maintaining reasonable scanning duration.It was finally decided to use 1.5 mm to remove most of the beam hardening and use small glass beads to remove the residual beam hardening).Those glass beads were randomly placed within the mixture and used to normalize the data and remove beam hardening: radial evolution of the grey levels from the center to the border of the cell was flatened thanks to those beads of identical density and chemical composition.

Results
Experimental results are summarized in Figure 3 and 4. Initially, the structure was heterogeneous, characterised by an assembly of pellets surrounded by pores and powder.The pores progressively disappeared and a quasi-homogeneous material was obtained at the end.A residual heterogeneous attenuation field, between previously existing pellets, can be observed.As could be expected, wetting progressed from both sides to the center of the cylindar.Nevertheless, a contrario, at least at the beginning, wetting qualitatively seemed faster coming from the top due to more pores existing there.When a focus is performed to analyse the data, mechanisms internal to the pellets can be exhibited and it can also be seen (see Figure 5, shown as a typical example) that the hydro-mechanical process is a function of the distance to the center of the specimen and of interactions between the pellets.Internal mechanism: hydration leads to both pellets swelling and cracking, due to both internal hydro-mechanically induced stresses and pre-existing cracks, which can be observed inside free single pellets only submitted to hydration (see [3]).External mechanism and spatial variability: cracks first appear on the pellets along the external border of the cylindar, while internal pellets apparently remain still.Nevertheless, it can be inferred that those cracks have two origins: 1) as was already underlined, internal stresses and pre-existing cracks and 2) contact forces between the pellets.Then, the pellets progressively disappear, and the mixture progressively becomes homogeneous.A study of the density-hydration function was also conducted.Mechanically induced phenomenons could not be separated from the hygrometric phenomenons.The axial force exhibits a constant increase followed by a plateau after 100 days (see figure 6).At the same time, more and more water was removed from the two water tanks, due to succion.This water volume inside the specimen was considered a second monotonously increasing variable, timely correlated with the force increase.It was thus decided to consider a single global variable, which we will improperly call apparent density, based on the educated guess that the density increases with time, as the total volume is constant.So that so called apparent density is rather a increasing function of the density.As was said, most of the beam hardening was removed thanks to the copper filter (note that an additional interesting feature of filtering is to harden the beam, which makes the result less sensitive to chemical composition and more to density. Still, there were grey level radial variations in the glass beads, which has no physical meaning.Those profiles were flattened considering the glass beads which (considering their chemical composition and density were stable during time, and identical between beads) must have the same grey levels everywhere.A linear fitting of the grey levels was then conducted, using the PMMA of the cell and two bigger glass beads situated inside the specimen (diameter 10mm), to make grey level be a function of density (Figure 7).

Discussion and conclusion
The temporal evolution of a wetted bentonite mixture was studied.It can be shown that a quasi-homogeneous material could be obtained in 100 days, almost without any residual pore, which is in line with the practical aim of this study.It will now be interesting to compare different mixture compositions and average densities and study their influence on the final structure of the material.Concerning the evolution of the apparent density, interesting observations can be made.First, after only 100 days, which is short as compared to the time considered for waste disposal, the sample is already quasi-homogeneous.The question then arises whether such homogeneity can be reached in a full size experiment, where a tomograph cannot be used to verify it.Second, the apparent density in the "solid" part appears to be rather stable during the whole process -though even apparently slightly decreasing, which is consistent with [4]-and it appears that the method used to estimate the density slighty overestimates the density of the dry pellets.Finally, considering the aim of this study, it is important to note that even the biggest pores are fully and homogeneously filled when the experiment duration is reached.One additional step to acertain this hypothesis and improving the current methodology would be to separate hygrometric phenomenons from mechanical ones, inside the solid phase.For example, based on [5], grey levels could be used for density and digital image correlation (DIC) for deformation.But an important problem arises: the modification of the solid phase during the experiments, which behaves more and more like a fluid, which implies DIC can be less and less used, due to the ongoing disappearance of the texture of the solid phase.

Figure 2 :
Figure 2: Experimental set-up containing the mixture

Figure 3 :Figure 4 :
Figure 3: Vertical slice.Temporal evolution of the structure of the mixture

Figure 5 :
Figure 5 : focus on a typical slice exhibiting the successive mechanisms taking place

Figure 6 :
Figure 6: evolution of the axial swelling pressure of the mixture

Figure 7 :
Figure 7: evolution of the apparent density as a function of time