Transport phenomena through porous materials for radioactive waste storage

The primary objective of this study was to investigate the influence of the electrical double layer on the ionic flux in presence of an external electric potential. The present work included two parts performed at the same time. The experimental part consisted in characterizing the properties of the porous structure of the interested materials by means of 3 methods: water open porosity, Mercury Intrusion porosimetry and Thermoporometry. Three materials were used in this study: meta-kaolin-based geo-polymers, a porous glass served as a model material due to its well-known pore size reported in the literature, and a type-V cement paste according to the European classification. After the characterization test, the materials transport properties were measured: natural diffusion test and migration test at the LECD (CEA Cadarache), and EIS at the LMDC (INSA Toulouse). In addition, we developed a new procedure and a new interpretation of the thermoporometry test, which reveals additional information on the porous structure of the geo-polymer and the porous glass. It was demonstrated that geo-polymers are a complex material with a multi-level pore range. We showed that the sodium-based material possesses only 11% of meso-pore in comparison with the total pore volume, whereas the latter is up to 98% in the MIP results. The MIP cannot detect the very strong ink-bottle effect inside the geo-polymers. This hypothesis is then consolidated by the results of the transport test, which all show that the effective diffusion coefficient of the geo-polymer is approximately one order lower (∼10^-10 m²/s) than the diffusion coefficient in the bulk medium of the tracer (∼10^-9 m²/s). With the controlled porous glass, both the MIP and TMP data are concentrated in the meso-pore region, in the vicinity of 7 nm if the pore shape is assumed to be cylindrical, with more than 90% of the total volume. Again, the effective coefficient diffusion of the hardened cement pastes, calculated from the migration test and the diffusion test do not match like in the previous work [Wattez, 2013]. In the modeling section, we recognized that when the pore size is close to the Debye length, the effect of the EDL on the ionic concentration is non-negligible. A study of the electrochemical behavior of the ionic species at the pore scale is necessary to understand the difference at the material scale. As the majority of the available pore size characterization methods are based on the assumption that the pore has a cylinder form, a cylindrical pore model, in which the channel is filled by a solution similar to that of the interstitial solution of the cement based material, was recently developed [Nguyen et al., 2017]. Using the Debye-Huckel approximation, we demonstrated that the electro-neutrality is established in the whole cylindrical pore after a certain diameter. For example, with a zeta potential of 25 mV, the electro-neutrality condition is obtained when the pore diameter is bigger than 200 nm. The EDL also influences strongly the ionic flux when an external electrical field is applied. It increases the electrical conductivity of the pore in comparison with the value in the bulk medium, particularly when the pore size is decreased. Next, the pore model was scaled up to the material size by using a sophisticated virtual porous architecture. The modelled materials here were the porous glass and the cement paste. The author modified the model developed by Vu et al. [2009] and Boher et al. [2013] to generate an artificial pore network which pore size distribution matched well the experience. The coupled electric phenomena governed by the Poisson-Boltzmann equation and Nernst-Planck equation were solved in the pore network, constituted of several pore channels with variable cross section. The modeling shows that when the tortuosity equals 1, the predicted electrical conductivity differs from the one in migration tests. It is 6 times higher in case of SCPG and 100 times for HCP. The difference may be related to the topology of the porous materials and as the model accounts only the constrictivity of the porous material, this difference is linked to the tortuous nature of the materials. This assumption was confirmed when a new glass porous network was reconstructed by means of a tortuosity of 2. We also calculated the effective diffusion coefficient of each ionic species in the REV and compared it to its bulk diffusion coefficient. The results show that the diffusion ratio is not the same for cation and anion in 2 study cases. The diffusion ratio increased noticeably between HCP and SCPG, probably because the quantity of finer pores of SCPG (in the vicinity of 7 nm) is more important than that of HCP. This should be taken into account for high confinement materials, particularly in electrokinetic tests like migration or EIS. (author)