Diffusion of HTO, 36Cl and 22Na in the Mesozoic rocks of Northern Switzerland: III. Cross-lab comparison of diffusion measurements on argillaceous twin samples

In the past decades, there has been a high interest in evaluations of argillaceous rock formations to host repositories for spent fuel or radioactive waste material. One of the main reasons for this interest is that argillaceous rocks are dominated by clay minerals, resulting in a fine-grained sediment matrix with micropores, low pore connectivity and thus low permeability, and good self-sealing properties. This entails that diffusion is the prime mechanism for transport of gas, fission products and radionuclides in clay-rich rocks. The determination of diffusion parameters is therefore key to evaluate the quality of the rock from long-term safety perspectives. Crosslab comparison of diffusion data is however often challenging as various methods, concepts and models are utilised in the different laboratories across the globe. Here, a direct cross-lab comparison study of throughdiffusion experiments was performed to compare and assess the effect of experimental method (through-diffusion) and modelling uncertainties of the parameters by comparing results obtained by two independent laboratories. The R&D group ‘Disposal’ at SCK CEN (Belgium) and the PSI-LES group (Switzerland) both performed through-diffusion experiments on (nearly) the same sample material using their in-house experimental and modelling methodologies for through-diffusion experiments. Adjacent (twin) samples at five depths between 870 and 940 m in the Trüllikon1-1 borehole (Switzerland) were selected and each lab subjected these five clay rock samples to diffusion in synthetic pore water with three different tracers, HTO, 36Cl− and 22Na+, representative for neutral, anionic and cationic transport behaviour. The two labs used a similar design of diffusion cell and worked with similar experimental conditions, but there were small differences in the experimental set-up/conditions and in the modelling approach. The independently determined diffusion parameters from SCK CEN and PSI-LES for all three tracers confirmed previously observed uncertainties. For all three radionuclides, the variability of the effective diffusion coefficients estimated independently by both institutes was less than a factor 2 and in general much lower (deviations ranging between 0 and 73%). Besides, the parameter estimations of the capacity factor (or accessible porosity in case of HTO and 36Cl− ) agreed well. Moreover, the experimental datasets of HTO and 36Cl− were also cross-fitted. The evaluation revealed that the minor deviations can be attributed predominantly to variations in temperature (experimental conditions) and to a lesser extent to minor distinctions in the modelling approach. It is important to acknowledge that local heterogeneity might also contribute to these differences.