Spent fuel

Several natural analogues for SF have been extensively studied since the late 1960s to obtain more information and data on the resistance of such material to oxidation and on degradation under natural, repository-relevant conditions. The best SF analogues are sedimentary uranium ore deposits formed under reducing conditions, as indicated in Chapter 6 of NAB 23-10 (Johnson et al. 2023). Examples include a range of uranium ore bodies such as El Berrocal in Spain, Poços de Caldas in Brazil, Oklo in Gabon, Koongarra in Australia, Cigar Lake in Canada and the Wit­watersrand Au-U placer deposit in South Africa (Armstrong 1981). Researchers have explored these analogues to investigate in detail various different processes influencing the long-term behaviour of spent fuel (Smellie & Karlsson 1999, Smellie et al. 1997, West et al. 1992). According to Milodowski et al. (2015), naturally occurring uranium minerals such as uranite and, to a lesser extent, pitchblende, serve as excellent analogues for SF.

In the study of the Cigar Lake ore body in Canada, Smellie & Karlsson (1996) addressed impor­tant questions regarding radiolysis, fuel dissolution, and redox fronts (Section 6.4 in NAB 23-10, Johnson et al. 2023). They found that the ore body had remained essentially unchanged since its formation over 1,300 million years ago, thus supporting the long-term stability of SF. The most notable result is the evidence of UO₂ resistance against radiolytic oxidation. In particular, the Cigar Lake uranite ore maintained its predominately reduced state despite undergoing significant alpha radiolysis and experiencing prolonged hydrothermal activity spanning millions of years. Bruno & Spahiu (2014) concluded, based on a subsequent reanalysis of the original Cigar Lake data, that “the long-term stability of the UO₂ matrix is warranted by the recombination of the radiolytically generated oxidants and reductants”.

In the Oklo natural reactor in Gabon (Francewill Basin), uranium ore is mainly composed of UO₂ in direct contact with organic matter, filling the porosity of the sandstone (Section 6.5 in NAB 23-10, Johnson et al. 2023). Isotopic signatures indicated that criticality has occurred and was reached in regions of highest uranium enrichment and porosity. Within the reactor zones (which represent the fuel analogues) UO₂ remains the dominating phase (Gauthier-Lafaye et al. 1996). The environmental conditions remained reducing throughout the history of the reactor, due to the large amounts of associated hydrocarbons. In the analogy with the high-level waste near field, the organic matter takes the “chemical role” of the metallic canister. Despite its complex tectonic and hydrothermal history, UO₂ has been mostly preserved over 2,000 million years. This serves as evidence that extensive radiolytic oxidation of UO₂, resulting in massive UO₂ dissolu­tion and reprecipitation as U(VI), did not occur (Johnson et al. 2023). The identification of H₂‑rich fluid inclusions (along with O-rich fluid inclusions, Dubessy et al. 1988) could be interpreted as the product of water radiolysis. The coexistence of these fluid inclusions with the well-pre­served uraninite suggests that the radiolytically produced H₂ was chemically activated, providing protection against the oxidation of UO₂.

The study of Johnson et al. (2023) concluded in Section 7.2 that “The low fractional dissolution rates recom­mended to be used in safety assessment are based on multiple arguments, including natural analogue studies showing that uraninite remains stable over geologic time frames.”

Hence, natural analogues complement the data derived from laboratory experiments, offering insights into spent fuel dissolution over geological timescales. However, it is important to acknowledge the limitations of the analogies, given the inherent differences in environmental conditions between the geological environment of the analogies and that of a deep geological repository (Section 7.2 of NAB 23-10, Johnson et al. 2023).

Reprocessed high-level waste

For over six decades, starting with Marshall 1961, natural glasses, such as volcanic glass and tektites, as well as archaeological glasses, such as uranium-bearing drinking vessels, slag or glass blocks of Roman age, have been the subject of extensive studies as analogues for reprocessed (vitrified) high-level waste. The primary focus has been on determining dissolution processes and rates, along with identifying the characteristics of secondary alteration products (see Miller et al. 2000 for specific examples).

Among these glasses, basaltic glasses exhibit the highest resemblance to contemporary RP-HLW formulations and are generally acknowledged as the most suitable analogues for borosilicate HLW glasses (Havlova et al. 2008). Due to the absence of high B₂O₃ in these glasses and radionuclides in general (Posiva 2023) both natural and archaeological glasses are of limited use as analogues of RP-HLW. Despite this limitation, the absorption of dissolved uranium on the surface of degradation products of natural glass, as observed by MacDougall (1977), suggests that radionuclides released from RP-HLW could potentially be sequestered by analogous secondary degradation phases on the waste package. Additionally, based on the examination of 425 natural glasses, it is suggested that over half of those were 2 million years old (Ewing 1979). Such information provides confidence that rapid degradation of RP-HLW is unlikely to occur during the time period for assessment.

Studies on Roman glass blocks buried in seawater-saturated mud provide additional insights despite the limitations of this chemical analogy (Section 4.4.5.2 in NAB 23-09, Curti 2022). Peripheral regions (rims) exposed to seawater exhibit higher degradation rates compared with internal regions. The long-term behaviour of basaltic glasses is extrapolated from laboratory experiments and natural occurrences, albeit with uncertainties. Degradation rates derived from natural basalts can underestimate true rates due to unrealistic assumptions and temperature uncertainties as summarised in Section 4.4.5.2 in NAB 23-09 (Curti 2022). Even though there are undisputable large uncertainties in estimating nuclear waste glass corrosion rates from natural and archaeological analogues due to varying conditions and compositions, these studies improve qualitative understanding of glass dissolution processes.

HLW disposal canisters

Mechanical and corrosion-related degradation processes will eventually lead to the breaching of the disposal canisters, which will then no longer provide complete containment of radionuclides. Along with copper, carbon steel is one of the few candidate canister materials for which there are natural and/or archaeological analogues that can help support long-term predictions of the corrosion behaviour (King 2021, Neff et al. 2010, Alexander & Reijonen 2023).

Various types of iron analogues, including natural analogues such as native iron (though rare) and archaeological analogues such as iron artefacts, have been extensively studied in the literature. Analogue studies can be used to support canister lifetime predictions by the provision of corrosion damage data, typically in the form of either corrosion rates (e.g., Crossland 2005) or pit depths (e.g., JNC 2000), over timescales unavailable to experiments, and by supporting the development of conceptual models and the validation of mechanistically based corrosion models.

Corrosion-related analogue studies often form part of a well-structured canister development programme that combines the results from short-term laboratory experiments, simulations and modelling, large-scale in-situ tests, and analogue studies (Dillmann et al. 2014). In general, analogues are used to support predictions of the rate and extent of uniform and localised corrosion rather than environmentally assisted cracking mechanisms, for which there are few, if any, analogues (King et al. 2023, Nagra 2024d). In addition, analogue studies are not widely used to support predictions of the mechanical stability of the canister.

Bentonite buffer

In line with the current safety and repository concept, bentonite is commonly used as a buffer material in repositories for the disposal of radioactive waste (Karnland et al. 2008, Karnland et al. 2007, Smellie 2001) and thus its thermal and long-term stability is critical. Examples of the long-term chemical and mineralogical stability of bentonite include those from Saskatchewan, Canada (75 Ma Avonlea bentonite), Wyoming, USA (95 Ma MX-80 bentonite) (Keto 1999) and Kiruna, Sweden (probably in the order of hundreds of Ma to 1 Ga old bentonite, Alexander et al. 2024), all periods greatly excess those required for bentonite in engineered barriers.

Other aspects of bentonite longevity have been addressed in a range of natural analogue studies and quantitative data are available for corrosion product interaction (Alexander & Reijonen 2023), low-alkali cement leachate interaction (Milodowski et al. 2016), illitisation (Leupin et al. 2014), host rock interaction (Alexander & Reijonen 2023) and canister settling (Alexander et al. 2017). A dedicated study of the thermal alteration of bentonite has been initiated at an identified site, offering valuable data for future post-closure safety cases (Alexander & Reijonen 2023). Despite some shortcomings identified by Laine & Karttunen (2010) in natural analogues studies for bentonite applications in the framework of post-closure safety cases, there is confidence that bentonite can meet safety related requirements for at least the entire time period for assessment.

Cementitious materials

An understanding of cement durability and degradation processes is needed to assess the evolution of the cementitious near field of the L/ILW repository. Cementitious materials are considered metastable products from a thermodynamic standpoint, implying that they will undergo changes and remineralisations over time due to internal and external cement degradation processes. These processes include aggregate-concrete interaction, alkali-silica reaction, mineralogical alteration by carbonatisation, leaching, and interaction with fluids containing concentrated ions (see Section 2.2 in NTB 23-03, Kosakowski et al. 2023).

Most information on the longevity of cement phases as present in Ordinary Portland cement (OPC) has been gathered during the Maqarin Natural Analogue Project of which Nagra was a partner (see Pitty & Alexander 2011, Martin et al. 2016). The natural clinker-analogues at Maqarin, in northern Jordan, were formed by low-pressure and high thermal heating of the organic-rich biomicrite and chalk host rocks around 2 Ma²⁶ ago, generating calcium silicate and calcium aluminate-ferrite mineral assemblages, which have similarities to clinker phases in modern OPC as summarised in Martin et al. 2016. Subsequent groundwater flow through these naturally generated clinker phases resulted in hyperalkaline fluids percolating through fractures and cracks created by tectonic fracturing of the biomicrite. This led to precipitations of cement minerals such as C-A-S-H, ettringite, thaumasite, and carbonates. The precipitations occurred mostly along the fractures which were infilled with such natural cement phases and underwent secondary alterations (see Milodowski et al. 2015, Martin et al. 2016 for details), which eventually sealed the former open fracture interfaces (see Fig. ‎9‑2). Dating of the system suggests the fracture-biomicrite interfaces sealed within a very short period of only a few years.

In distal regions along fractures further away from the natural clinker, diffusive exchange over sufficient time resulted in silicate dissolution and the precipitation of zeolites (or zeolite precursors) (Martin et al. 2016). Such secondary reaction phases are comparable to those produced when concrete is leached in the laboratory and in Underground research laboratory (URL) experiments (e.g., Mäder et al. 2006), clearly indicating the relevance of the analogy.

Fig. ‎9‑2:Heavily fractured micrite host rock with abundant sealed fractures containing a comparable product to natural concretes at the Maqarin site, Jordan

For scale, see hammer in bottom left corner (Pitty & Alexander 2011).