1          Introduction

1.1  Background

Generating energy through nuclear fission leads to the production of radioactive waste, which can contain uranium, plutonium, so-called “minor actinides” such as neptunium, americium and curium as well as fission and activation products. In particular, the safe disposal of high-level radioactive waste requires its isolation from the biosphere for several hundred thousand years. In line with German safety requirements for the disposal of heat-generating radioactive waste, a site-specific safety case covering an assessment period of one million years must be prepared to demonstrate the safety of the repository concept (EndlSiUntV). In connection with this, one important issue is radionuclide behaviour (the retardation, mobilisation, and potential release and transport of radionuclides) within the disposal zone of the deep geological repository. the primary aim of radioactive waste disposal is to immobilise and contain long-lived radionuclides over a geological timescale in the repository. The internationally agreed concept is to dispose of high-level radioactive waste in deep geological repositories using a multi-barrier system that fulfils different safety functions (Alonso et al. 2005, Yoshida et al. 2005, OECD-NEA 2013)

In the ongoing site selection process, different repository concepts in different host rock formations (rock salt, clay rock and crystalline rock) are currently being comparatively evaluated in Germany. One important part of the evaluation is the application of preliminary safety assess­ments for each site and the consideration of scenarios that can lead to the release of radionuclides. Repository concepts in crystalline rock, such as the KBS-3 concept in Sweden and Finland, which foresees using a disposal canister with very low corrosion rates in combination with a bentonite buffer as (geo)engineered barriers (SKB 2011), are widely applied.

One relevant scenario, which includes a possible breach of the disposal canister and ensuing radionuclide release, is the intrusion of low-mineralised glacial/meteoric waters into the disposal zone, which can cause erosion of the bentonite barrier (Liu & Neretnieks 2006). Aspects of such a scenario are addressed in the Colloid Formation and Migration (CFM) project, which is performed within the framework of an international consortium at the Grimsel Test Site (GTS), the underground research laboratory (URL) operated by Nagra, the National Cooperative for the Disposal of Radioactive Waste, in the Swiss Alps. The CFM project deals with the relevant processes of bentonite erosion and radionuclide migration in the near field of the geological and engineered barrier system of a deep geological repository constructed in crystalline rock. The focus is on clay-colloid-release mechanisms and the facilitated radionuclide transport investigated in the form of laboratory and in-situ experiments as well as within the context of model developments and applications. This programme is the only combined experimental and modelling research project worldwide performed under in-situ conditions over longer time scales in a URL within an international consortium (https://www.grimsel.com/gts-projects/). The CFM project is supported by German ministries (BMWi/BMUV) within the KOLLORADO-e³ [1] project, and partners from the Friedrich-Schiller University Jena, the company Gesellschaft für Anlagen- und Reaktorsicherheit (GRS) GmbH, and the Institute for Nuclear Waste Disposal of the Karlsruhe Institute for Technology (KIT-INE) are involved. The outcome of previous studies within the CFM project has been reported (e.g. in Noseck & Schäfer 2020; Huber et al. 2014 and 2016). The present report describes the results of the KOLLORADO-e³ project.

1.2                     Colloid Formation and Migration (CFM) Experiment

To assess the impact of bentonite colloids on the long-term, post-closure safety of a repository in crystalline host rock, a series of field experiments have been carried out in Nagra’s underground research laboratory (Grimsel Test Site, GTS) within the framework of the Colloid Formation and Migration project (CFM, Schlickenrieder et al. 2017). CFM is an international project currently joined by BGE (Germany), BMUV (Germany), KAERI (Republic of South Korea), Nagra (Switzerland), NUMO (Japan) and NWS (United Kingdom). Numerous organisations support the CFM project.

CFM further pursues previous GTS colloid studies such as the Grimsel Colloid Exercise (Degueldre et al. 1990) or the Colloid and Radionuclide Retardation Experiment (CRR, Möri 2004). CFM started in 2004 with a first phase dedicated to studying the in-situ boundary conditions, predictive modelling and supporting laboratory programmes in the controlled Area with a focus on the so-called Migration Shear zone (Möri & Blechschmidt 2006). CFM Phase 2 lasted from 2008 to 2013 and was used to enhance the field setup with three new monitoring boreholes and the corresponding instruments in the boreholes and tunnel and to test the robustness of the system while characterising the flow domain. To this end, multiple tracer tests were carried out. Phase 3 started in January 2014 and lasted until 2018. The Long-term In-situ Test (LIT) was initiated in May 2014 with the emplacement of a compacted bentonite ring source containing radionuclide tracers in a central borehole. LIT was overcored in 2019 after an in-situ experiment duration of approximately 4.5 years. CFM Phase 4 was started in 2019 with the implementation of the i-BET (in-situ Bentonite Erosion Test) consisting of roughly 50 kg of Na-dominated bentonite emplaced in a highly fractured and water -conducting interval. Since then, i-BET has been continuously monitored. As i-BET is implemented outside the radioprotection-controlled area, experiments within the CFM project are performed in two distinct areas of the GTS.

1.3                     Aims and starting point of the current Kollorado-e³ project

The KOLLORADO-e³ research programme is based on the outcome of the preceding KOLLORADO-e² project. The overall objective of both projects is to shed light on the multitude of different interaction forms and paths between radionuclides, colloids and the rock matrix. The following topics are of Special interest: (i) the erosion behaviour of compacted bentonite in fractured rock, (ii) obtaining a better understanding of the speciation of homologues and radio­nuclides under the field conditions at the GTS, (iii) the consequences of kinetically controlled adsorption and desorption processes, (iv) the effect of fracture heterogeneity on bentonite erosion and (v) the effect of fracture heterogeneity on flow and non-reactive tracer transport. Aside from the experiments performed in the laboratory as well as under in-situ conditions at the GTS and the application of sensitive analytical techniques, various modelling and simulation approaches were used:

• Benchmark calculations with thermodynamic speciation models and the respective databases;
• Macroscopic 1d/2d models using simplified geometry implementing sorption/desorption kinetics of colloids/radionuclides;
• extension of an available 2d model on bentonite erosion to include heterogeneous flow fields based on random aperture distributions; and
• Comsol multiphysics® to simulate flow and transport of non-reactive tracers in single fractures with complex geometry.

For the computer simulations of the radionuclide transport, two transport codes used for perfor­mance assessment were applied: (i) the 1D-transport COFRAME (Reiche et al. 2016), which considers colloid-facilitated radionuclide transport in fractured media using a double-porosity approach and (ii) the transport code CLAYPOS, which simulates one-dimensional radionuclide transport in low-permeable media either in cylindrical or in planar geometry (Rübel et al. 2007).

The Long-term In-situ Test (LIT) at the GTS ran from 2014 to 2018. Its objective was to investigate in-situ colloid formation and mobilisation of radionuclides from a radionuclide-labelled bentonite plug (Schlickenrieder et al. 2017). A radionuclide cocktail filled into glass

vials was embedded in a bentonite block, which in turn was emplaced inside a packer system in the GTS, where it was brought into contact with a water-bearing fracture. During the KOLLORADO‑e² project, samples from observation boreholes and a tunnel surface packer (Pinkel) were continuously taken, analysed and evaluated until the experiment was terminated and overcored at the end of 2018. A series of mock-up tests under inert gas conditions (argon atmosphere) performed in the laboratory using a plexiglass setup were performed within KOLLORADO-e² in order to study and quantify bentonite erosion processes. Natural Febex bentonite or a mixture of synthetic montmorillonite and Grimsel groundwater (GGW) were used for this type of experiments to keep it close to the expected LIT scenario. Besides investigating bentonite erosion and radionuclide mobilisation, an additional focus lay on a feasibility study aiming to find a suitable way to conserve the LIT sample during overcoring.

Very low erosion rates and no indication of a gravitational erosion were observed in the LIT (Noseck & Schäfer 2020), which contradicts the outcome of experimental work by SKB and POSIVA in the laboratory on artificial plexiglass fracture setups (Alonso et al. 2019) and the results of our own laboratory experiments with a similar horizontal test setup (Rinderknecht 2017). A key task of KOLLORADO-e³ was the investigation of possible reasons for the discrep­ancy of results and notably the influence of accessory mineral phases in the bentonite on the erosion behaviour in laboratory experiments.

During the lifetime of LIT, of the radionuclides ⁴⁵Ca, ⁷⁵Se, ⁹⁹Tc, ¹³⁷Cs, ²³³U, ²³⁷Np, ²⁴¹Am and ²⁴²Pu, only ⁹⁹Tc was observed at ultra-trace concentration levels in the observation boreholes (Noseck & Schäfer 2020). Radionuclide migration simulations described in the present report focused on investigating the diffusion of the radionuclides contained in the glass vials in the bentonite. In mock-up tests, release curves could be obtained for ⁹⁹Tc as well as for ²³⁷Np and ²³³U from the bentonite. Respective 1D model calculations for purely diffusive transport, based on the data and model approaches used for Nagra’s long-term safety analysis (van Loon 2014), were carried out with the CLAYPOS code (Rübel et al. 2007). Scoping calculations for diffusion profiles of all radionuclides used in the LIT and in the mock-up test were performed. For the trivalent and tetravalent actinides, the model calculations predicted that during the duration of the experiments, transport should only take place over a few mm (Noseck & Schäfer 2020). Steady-state profiles were expected for radionuclides in oxidised form of e.g. ⁹⁹Tc, ²³⁷Np and ²³³U. The LIT samples could only be delivered to the radioanalytical laboratories relevant for charac­terisation with a significant delay due to technical problems during over-coring and cutting, as well as due to the Corona pandemic. A comparison with experimental data that were made available recently will be described and discussed in the present report.

Radiotracer migration experiments under in-situ conditions, notably for elements with low solubilities, usually suffer from high dilution due to dispersion, resulting in concentrations that are difficult to analyse using classic instrumental techniques. By developing and applying appropriate methods on the basis of the extremely detection-sensitive accelerator mass spectrometry (AMS) (Quinto et al. 2017, Quinto et al. 2019), radionuclides could also be determined in the ultra-trace range in samples. Transport of bentonite colloids dispersed in GTS groundwater spiked with relevant radionuclides had been investigated in those studies. AMS analysis allowed for the detection of actinides in the long-term tailings of elution curves at very low concentrations. Those data allowed for the determination of consistent reaction rates, which served as the basis for a forecast of colloid transport at repository scale in the Kollorado-e³ project. In a series of tracer experiments on colloid- facilitated radionuclide transport at the Grimsel Test Site, the clear influence of kinetics on radionuclide sorption and desorption reactions as well as on colloid filtration was determined. Calculations were carried out with the FRAME-COFRAME programme package (Reiche et al. 2014).

Thermodynamic benchmark calculations were performed within KOLLORADO-e² in order to better understand the speciation and thus the transport and retardation effects of homologues and radionuclides under the evolving geochemical conditions in the field experiments at the GTS, particularly the LIT (Kunze et al. 2008). Since an earlier benchmark performed in 2000 (Bruno et al. 2000), updates related to radionuclide-speciation calculations, have been reported in the literature. These are mainly related to new thermodynamic data and more detailed knowledge on groundwater and porewater compositions as well as redox conditions for the scenario of interest. In particular, the conditions at the interface between bentonite and crystalline rock have become more important for the current experiments and have been included into the new benchmark. The speciation of the radionuclides, for example their redox state, is of crucial importance for their mobility, solubility and interaction with colloids in systems in the near- and far-fields of a repository constructed in granite rock. To check (i) the influence of the geochemical conditions on the radionuclide speciation and (ii) the applicability of the currently available databases, benchmark speciation calculations for the six radionuclides used in the LIT(⁷⁵Se(VI), ⁹⁹Tc(VII), ²³³U(VI), ²³⁷Np(V), ²⁴¹Am(III), ²⁴²Pu(IV) and ²³²Th(IV)) in bentonite porewater, Grimsel groundwater and mixed water that forms at the bentonite/crystalline rock interface have been performed within the current project. Assuming a reasonably solid phase, the solubility and the most relevant mobile species for each element in five different model waters have been determined. The initial results of two modelling groups have already been published (Montoya et al. 2022). Depending on the radionuclide, pH values, redox conditions, carbonate, silicate, iron and/or calcium concentrations are decisive for their speciation and maximum concentration, which are determined by the solubilities of the corresponding solid phases. In general, the results of both modelling groups show good agreement. Any differences result from the consideration of different species, for example the existence of silicate complexes for trivalent and pentavalent actinides or the consideration of poly-selenides. However, for both systems the data situation is still uncertain. These benchmark calculations showed, among other things, the possible influence of the silicate concentrations measured in the respective water and porewater as well as the uncertainties in the input data on the radionuclide speciation. Differences arose in the calculation of maximum concentrations for radionuclides, which are required for the planning of radionuclide tracer cocktails for in-situ experiments. These differences are mainly caused by different solubility-determining solid phases and their crystallinity, which the respective experts suggested as relevant for their calculations. Results of all six modelling groups are summarised in the present report.

The specific goals of KOLLORADO-e³ were to develop an improved mechanistic understanding of the integrity of the bentonite barrier and the colloid-mediated radionuclide transport using advanced spectroscopic and microscopic methods.

Key topics included:

• Laboratory and in-situ studies on bentonite erosion with a focus on the role of accessory mineral abundance in the bentonite;
• Post-mortem analysis of samples from the lit and respective mock-up experiments;
• Analysis of fallout-actinides at the gts in order to gain insight into actinide mobility under given environmental conditions.

Experimental studies were complemented by modelling and simulation studies:

• Benchmark exercise related to radionuclide speciation under experimental conditions relevant for CFM studies;
• Modelling radionuclide migration under the conditions of the LIT at the GTS and laboratory mock-up experiments;
• Upscaling of simulations to repository scales (space and time).

 

[1]     In-situ Experimente zur Bentonit Langzeit-Stabilität und der Radionuklidmobilität an der Grenzfläche Bentonit – Kristallin (FKZ: 02 E 11759A-C).