Long-term spatiotemporal monitoring of diffusion processes in Opalinus drill cores with GeoPET and parameterization with Comsol Multiphysics

Typically, the assessment of effective diffusion parameters of natural geologic media is conducted by repeated concentration measurements in diffusion cells. They contain small-sized samples which are regarded as 1D-homogeneous “black boxes”. Alternatively, 1D-tracer profiles can be obtained by means of abrasive pealing [1]. These methods have in common that 2D or higher spatial inhomogeneity of structure (and composition) and anisotropy cannot be considered in one and the same sample. In the course of the past decade we established the GeoPET-method allowing the direct, non-destructive, quantitative spatiotemporal visualization of (e.g.) diffusion processes in natural geological media on drill-core scale [2-4]. Here we couple it with the parameterization of heterogeneous and anisotropic effective diffusion parameters by means of inverse modeling with a finite element based numerical model (COMSOL Multiphysics, V4.2a) [5].
Out GeoPET-method is characterized by unrivalled sensitivity and selectivity to positron-emitting radio nuclides (here: PET tracers, Positron Emission Tomography, a nuclear medicine imaging method) in geological systems, without physical and chemical impact on the observed (reactive) transport process, and with adequate spatial and temporal resolution. Requirements for reaching the physical limit of image resolution of nearly 1 mm are a high-resolution PET-camera, like our ClearPET scanner (Raytest), and appropriate correction methods for scatter and attenuation of 511 keV-photons in the dense geological material. The latter are by far more significant than in human
and small animal body tissue (water). For long-term experiments, like diffusion monitoring, unconventional PET-nuclides, like 124I (T1/2 = 4.18 d), 58Co (T1/2 = 71.3 d) and 22Na (T1/2 = 4.602 a), are applied. The sensitivity is in the order of 0.1 kBq per voxel (1.5 mm3), which is limited by the background radiation.
Diffusion experiments were conducted on Opalinus clay drill cores with diameter of 10 cm and a length of 8 cm. These were cast into epoxy resin and a small axial drill hole was filled with 1 ml synthetic Opalinus pore water (OPW), labeled with a PET-nuclide. In first tests we applied 124Iodide.
Over 3 weeks we observed fast tracer propagation in distinct zones of the sample. This we interpret as due to suction into partially unsaturated zones. One sample was re-saturated over 3 months with OPW and then with OPW labeled with 22Na. The diffusion of the PET-tracer was observed for the following cause of 7 months. Then the specific activity fell below the detection threshold.
The quality of the obtained > 20 GeoPET-images is still improvable. Extensive Monte-Carlo simulations of our measurements, considering material dependent scattering cross sections of all occurring nuclear physical effects, have provided us with profound knowledge on the impact of scattering and attenuation on image quality [6]. Typical artifacts and blurring are identified as result of Compton scattering, which affects more than 70% of all recorded coincidences. Skillful selection of energy window and gantry diameter is capable of reducing this value to about 47%. This percentage will further drop significantly once the currently developed, appropriate scatter correction algorithms
are applied. Further, these Monte-Carlo simulations are a potential method for inversion-based image reconstruction [6].
We modeled this experiment with COMSOL Multiphysics ® 4.2a (3D convection-diffusion equation, PDE mode, PARDISO solver) for reproducing the observed spatiotemporal concentration distribution data with the differential equation for anisotropic diffusion and adsorption. By importing GeoPET images from various time steps and applying the Optimization Module (least square fit applying the Levenberg-Marquardt algorithm) to these images we efficiently determined best fit values e.g. of the diffusion tensor. Combined with the parameter sweep operation the sensitivity analysis is performed in parallel and covers the range of literature values for porosity and Kd values for 22Na+ sorption on Opalinus clay [5].
The experimental data could be reproduced quite well, but the obtained parameter values for diffusion parallel and normal to the bedding are slightly larger than reported in [7]. This is in accordance with our observations of an emerging gas bubble in the tracer reservoir. In spite of the long re-saturation period, suction tensions caused by unsaturated clay zones must have significantly influenced the transport regime by an additional advective component.
References
1. Van Loon, L.R. and J. Eikenberg, A high-resolution abrasive method for determining diffusion profiles of sorbing radionuclides in dense argillaceous rocks. Applied Radiation and Isotopes, 2005. 63: p. 11-21.
2. Richter, M., et al., Positron Emission Tomography for modelling of geochmical transport processes in clay. Radiochimica Acta, 2005. 93: p. 643-651.
3. Kulenkampff, J., et al., Evaluation of positron emission tomography for visualisation of migration processes in geomaterials. Physics and Chemistry of the Earth, 2008. 33: p. 937-942.
4. Gründig, M., et al., Tomographic radiotracer studies of the spatial distribution of heterogeneous geochemical transport processes. Applied Geochemistry, 2007. 22: p. 2334-2343.
5. Schikora, J., Simulation of diffusion-adsorption processes in natural geological media by means of COMSOL Multiphysics, in Faculty of mechanical Science and Engineering. 2012, Dresden Technical University: Dresden, Germany. p. 95.
6. Zakhnini, A., et al., Monte Carlo simulations of GeoPET experiments: 3D images of tracer distributions (18F, 124I and 58Co) in Opalinus Clay, anhydrite and quartz. 2012. 2012 (submitted).
7. Gimmi, T. and G. Kosakowski, How mobile are sorbed cations in clay and clay rocks? Environmental Science and Technology, 2011. 45: p. 1443-1449.