The Boom Clay Formation in northeastern Belgium was deposited during the Lower Oligocene (Rupelian) in a shallow marine environment of the North Sea Basin, at estimated paleo-depths of 50–100 m (Vandenberghe and Mertens, 2013; Zeelmaekers et al., 2015). It comprises rhythmically alternating clay-rich and silt-rich beds (Dehandschutter et al., 2005), subdivided into three stratigraphic members: the Belsele-Waas Member (siltiest), the Terhagen Member (pale grey clay, lowest coarse fraction), and the Putte Member (organic-rich dark clays) (Aertsens et al., 2004). Mineralogically, the formation is considered homogeneous, containing ∼60 wt % clay minerals dominated by illite, with subordinate smectite, kaolinite, chlorite, ∼20 wt % quartz, and ∼10 wt % feldspar (Aertsens et al., 2004; Frederickx et al., 2018, 2021). Its burial history includes continuous subsidence, interrupted by a minor uplift phase during the Chattian, resulting in a modest over-consolidation ratio (OCR) of ∼2.4 (OECD and NEA, 2022).

Hydraulic conductivity data compiled for this study originate primarily from boreholes in northeastern Belgium (Essen-1, Zoersel, Weelde-1, Doel-2b, Mol-1) and the HADES Underground Research Laboratory (URL) in Mol. In these locations, present-day formation top depths range from ∼50 to 260 m, and the Boom Clay shows thicknesses between 80 and 100 m. Most laboratory measurements were obtained using constant-head permeameter tests and diffusion/migration experiments, while in situ data, exclusively from piezometers, were collected in URL-drilled boreholes. In total, 269 laboratory and 30 in situ measurements were evaluated, yielding a hydraulic conductivity range of 3.5×10-10 to 2.1×10-13 ms-1.

The Amaltheenton Formation, deposited under shallow-marine conditions during the Upper Pliensbachian in the Lower Saxony Basin (Germany), consists predominantly of organic-lean, grey claystones and calcareous claystones, frequently intersected by siderite concretions. Formation thicknesses between 100 and 200 m have been inferred from borehole profiles and outcrop data (Burnaz et al., 2024; Marten et al., 2024; Wijesinghe et al., 2025). In the Hils and Sack synclines of Lower Saxony, the formation experienced highly variable maximum burial depths, ranging from approximately 1300 to 3300 m over a SE–NW transect of ∼50 km (Castro-Vera et al., 2024; Burchartz et al., 2026). Cretaceous inversion tectonics resulted in substantial uplift to shallow depths (from few metres to several decametres below the surface with respect to the formation top) and pronounced over-consolidation (Stahl, 1993; Voigt et al., 2021; Gaus et al., 2022). As no directly measured OCR values are available for AMA, we estimate a conservative OCR>20 from the ratio of reconstructed maximum burial depth to present-day depth, assuming hydrostatic pore pressure conditions.

Within the MATURITY project, eight boreholes were drilled across five locations, intercepting the AMA at variable depths (formation tops between 2 and 74 metres below ground surface (m b.g.s.)) and formation thicknesses, capturing the full extent of the lateral burial variability. A comprehensive investigation campaign was carried out, involving both laboratory and in-situ testing. Laboratory hydraulic conductivity data is based on permeameter experiments while in situ data stems from hydraulic testing conducted using straddle-packer systems in both single- and double-packer configurations (Burchartz et al., 2026).

The mineralogical composition of the AMA is similar across the study area, with total clay mineral contents ranging from 58 wt % to 75 wt %, framework silicates between 7 wt % and 23 wt %, and carbonates from 1 wt % to 8 wt % (Burchartz et al., 2026). Laboratory-derived hydraulic conductivities fall between 1.8×10-12 and 1.3×10-13 ms-1, while in-situ measurements indicate significantly higher values, ranging from 3.4×10-5 to 1.1×10-7 ms-1, reflecting the influence of fractures, stress relief, or testing-related disturbances (Burchartz et al., 2026).

Toarcian–Domerian argillites have been extensively studied at the Tournemire URL, located in a Mesozoic basin on the southern margin of the Massif Central, France (Bonin, 1998). Research at this site, led primarily by the French Institute for Radioprotection and Nuclear Safety (IRSN), focuses on the suitability of these formations for hosting high-level radioactive waste. The sedimentary succession, composed of marlstones and argillites, reaches thicknesses of up to 250 m (Constantin et al., 2004) and was deposited under shallow-marine conditions at estimated water depths of approximately 50 m (Humbezi Desfeux et al., 2024).

Following deposition, compaction and cementation occurred as a result of loading by thick overlying limestone units (Bonin, 1998). 1D thermal history modelling suggests that the formation was buried up to 2000 m (Barbarand et al., 2001; Peyaud et al., 2005; Mazurek et al., 2006), while other references state depths between 900–1700 m (OECD and NEA, 2022), and 2500–3000 m during maximum burial (Barbarand et al., 2001; Charpentier et al., 2003). However, the consistently low porosity, typically below 10 % (Boisson et al., 2001; Humbezi Desfeux et al., 2024; Patriarche et al., 2004a), and predominantly brittle mechanical behaviour that results in the formation of fractures (Constantin et al., 2002; Su et al., 2017) indicates relatively deep burial. Considering a present day coverage of approximately 300 m (Aalenian to Bathonian lime- and dolostones) (De Windt et al., 1999), and in the absence of directly measured OCR values, the To-Do argillites are estimated to exhibit an OCR of >5, derived from the ratio of reconstructed maximum burial depth to present-day depth under assumed hydrostatic conditions.

Mineralogically, the formation consists of 40 wt %–50 wt % clay minerals, with quartz contributing 10 wt %–20 wt % of the bulk composition. Carbonates (primarily dolomite and calcite), along with pyrite, account for 3 wt %–5 wt % and ∼3 wt %, respectively (De Windt et al., 1999; Boisson et al., 2001). The clay fraction is dominated by illite and kaolinite, with subordinate mixed-layer illite/smectite (with >70 % illite), mica, and minor chlorite (Bonin, 1998; De Windt et al., 1999; Boisson et al., 2001; Constantin et al., 2004).

Hydraulic conductivity data from the Tournemire site are limited and largely derived from Boisson et al. (2001). Hydraulic conductivity values are in the range of 10-14 to 10-15 ms-1 from laboratory tests, and 10-11 to 10-14 ms-1 from in situ measurements.

The Callovo-Oxfordian argillite, deposited during the Middle Callovian to Middle Oxfordian in the eastern part of the Paris Basin, has been the focus of extensive research over the past three decades. These efforts, led by the French National Radioactive Waste Management Agency (ANDRA), have culminated in the selection of this formation as the host rock for a deep geological repository. Numerous deep boreholes and the Bure Underground URL are located within a ∼250 km2 area in the Meuse/Haute-Marne district of north-eastern France (Delay et al., 2007; Distinguin and Lavanchy, 2007; Cosenza et al., 2014).

The COx succession exhibits a consistent thickness exceeding 100 m across the study area and is subdivided into four lithostratigraphic sub-units. These units show a systematic upward trend of decreasing clay content and increasing proportions of silt and carbonates (Gaucher et al., 2004; Cosenza et al., 2014). Mineralogically, the rock is dominated by clay minerals, particularly illite and mixed-layer illite/smectite, with total clay contents reaching up to 60 wt % (Gaucher et al., 2004; OECD and NEA, 2022).

At present, the top of the formation lies between 400 and 600 m below the surface (Delay et al., 2007). A one-dimensional burial model by Blaise et al. (2014) suggests that the formation experienced maximum burial depths of approximately 850 m (±200 m) and associated temperatures of around 50 °C during the Late Cretaceous (OECD and NEA, 2022).

Both, laboratory and in situ tests consistently report low hydraulic conductivities, typically ranging between 10-12 and 10-14 ms-1 (Distinguin and Lavanchy, 2007), supporting the formation's classification as a very low-permeability barrier rock.

The Aalenian (Middle Jurassic) Opalinus Clay Formation, deposited in an epicontinental sea covering large parts of Central Europe, is the designated host rock for high-level radioactive waste disposal in Switzerland (Wetzel et al., 2003; Zimmerli et al., 2024). Over recent decades, extensive research has been conducted on the formation by numerous institutions and scholars. Under the guidance of the National Cooperative for the Disposal of Radioactive Waste (NAGRA), the Opalinus Clay has been investigated through multiple deep and shallow boreholes across several siting regions in northern Switzerland, as well as in the Mont Terri URL (Fig. 1).

Based on sedimentological features, the OPA is subdivided into three distinct facies zones, characterized by the presence of thin calcareous silty-sand layers or siderite nodules, and a shallowing-upward trend that is reflected in increasing silt and fine sand content toward the top of the formation (Nagra, 2002; Gautschi, 2017; Zimmerli et al., 2024). Mineralogically, the formation is dominated by clay minerals, which make up approximately 59 wt % ± 12 wt % of the bulk rock composition. Among these, illite is the most abundant (22 wt % ± 8 wt %), followed by kaolinite (19 wt % ± 6 wt %), mixed layer illite/smectite (14 wt % ± 7 wt %), and chlorite (7 wt % ± 3 wt %) (Gautschi, 2017).

The burial history of the OPA is spatially variable. In northern Switzerland, the formation reached maximum burial depths of 1650–1700 m, whereas in the Mont Terri area, maximum depths were more modest at around 1350 m (Mazurek et al., 2006). This burial history is reflected in present-day porosity values, which are generally lower in northern Switzerland (11 %–12 %) than at Mont Terri (12 %–18 %) (Gautschi, 2017). Correspondingly, estimated over-consolidation ratios (OCR) based on burial and present-day depth are between 1.5–3 in northern Switzerland and 2.5–5 at Mont Terri (OECD and NEA, 2022).

Hydraulic properties of the OPA have been thoroughly investigated through both laboratory and in situ measurements. Reported values of hydraulic conductivity span a broad range from 10-4 to 10-14 ms-1, depending on location, facies, and depth of the studied rocks (Nagra, 2002; Mazurek et al., 2006; Gautschi, 2017).

In addition to Switzerland, the OPA is also being considered as a potential host rock in Germany, where it is investigated at various locations. Hekel (1994) conducted hydraulic conductivity tests in several shallow wells, demonstrating a clear depth-dependent trend in permeability and influence of local topography. There, the OPA is present a few metres below the surface, representing strongly uplifted rock sections, influenced by decompaction and weathering.

The Boda Claystone Formation is the proposed host rock for long-term high-level radioactive waste (HLW) disposal in Hungary (Németh et al., 2016). This Upper Permian unit reaches a thickness of approximately 800–1000 m and forms part of a much larger, 2000–4000 m thick siliciclastic continental sedimentary sequence (Fedor et al., 2008, 2019; Konrád et al., 2010; Németh et al., 2016). Investigations of the BCF began in the 1990s, focusing primarily on several deep boreholes and an underground research laboratory in the Western Mecsek Mountains region. Structurally, the formation is divided into two main blocks: the Boda and Gorica blocks, with most research activities concentrated in the Boda block (OECD and NEA, 2022).

The BCF consists of relatively homogeneous lacustrine sediments deposited in a playa-basin environment, arranged in monotonous sedimentary cycles (Fedor et al., 2008). Lithologically, it can be subdivided into three main units: a lower transitional unit (100–150 m) of fine-grained sandstones, a middle unit (350–450 m) of albite-rich claystones and siltstones, and an upper unit (400–500 m) of interbedded claystone, albite-rich clayey siltstone, and silty claystone, locally containing dolomite and siltstone beds with desiccation cracks and septarian dolomite concretions (Konrád et al., 2010).

Mineralogically, the BCF is dominated by clay minerals (up to 55 wt %), primarily illite and chlorite, with minor smectite, kaolinite, and vermiculite. Other significant components include authigenic albite, detrital quartz, carbonate minerals (calcite and dolomite), and hematite (Árkai et al., 2000; Németh et al., 2016).

The BCF experienced substantial burial, reaching maximum depths of approximately 3500–4500 m which led to extreme consolidation, resulting in over-consolidated (OCR∼4.2) highly indurated, brittle rock characterized by very low porosity (mostly <3 %) and high bulk densities (∼ 2.6 gcm-3) (OECD and NEA, 2022). Publicly available hydrogeological data for the formation are limited. However, the few available in situ hydraulic test results indicate a wide range of hydraulic conductivities, spanning from 10-13 to 10-8 ms-1. In contrast, laboratory-based water permeability measurements show a much narrower range of K values, consistently between 10-13 and 10-14 ms-1. This large variability suggests a reduced self-sealing effectiveness of fractures, likely related to the high degree of induration and brittle behaviour of the rock (Kovács, 2001).

The dataset compiled for this study consists of hydraulic conductivity (K, in ms-1) entries, obtained through both laboratory-based and in situ testing across six argillaceous formations in Europe (Fig. 1). Primary data sources include peer-reviewed publications, technical reports, and theses, and were supplemented with data ranges from the Clay Club Catalogue (OECD and NEA, 2022). A complete reference list is given in Table A1.

Most data were collected from vertical boreholes, either drilled from the surface or accessed via URL's. In a few cases, in situ data were derived from horizontal boreholes. Where reported, borehole and sample orientation relative to bedding or stratification was considered, allowing for an assessment of hydraulic anisotropy.

To ensure the comparability and consistency of the dataset, a set of inclusion criteria was applied during data selection. These include: i.

A credible and well-documented testing methodology (laboratory or in situ).

In addition to hydraulic conductivity data, petrophysical parameters such as porosity, grain density, and bulk density were included to support the interpretation of burial- and diagenesis-related trends. Porosity values were derived from standard petrophysical methods, including gravimetric water-loss after drying at 105 °C, bulk-density/grain-density relationships, and helium pycnometry. Accordingly, these values represent matrix properties of intact samples and do not account for fracture porosity. Similarly, laboratory hydraulic conductivity measurements represent matrix hydraulic conductivity, whereas field-scale measurements were retained as reported in the original sources and may reflect bulk rock-mass behaviour where fractures or other heterogeneities are present.

The hydraulic conductivities compiled for this study were based on various test methods and types. The abundance of each test method across the dataset and individual formations is shown in Fig. 2. It is recognizable that the vast majority of laboratory-based hydraulic conductivities is derived by permeameter tests (e.g. flow-through tests), while in situ methods are dominated by packer tests. Laboratory experiments were predominantly conducted on intact core plugs and are therefore considered representative of matrix hydraulic properties. In contrast, in situ measurements may incorporate the hydraulic influence of fractures and other rock-mass heterogeneities and are therefore interpreted as bulk rock-mass hydraulic conductivity.

Here, only a concise overview is given of the different testing methods in Tables 2 and 3. For detailed information, the reader is referred to the references given in the tables.

The compiled dataset of 782 hydraulic conductivity (K) measurements span twelve orders of magnitude, from 10-16 to 10-4 ms-1 (Fig. 3) and covers a depth range of few metres below surface and approximately 1100 m (Fig. 4). In situ test results (n=370) display substantially greater scatter in hydraulic conductivity than laboratory data, and range between 10-15 to 10-4 ms-1 as opposed to laboratory data which range between 10-16 to 10-10 ms-1). Formation-specific K distributions and selected summary statistics are given in Table 5.

The Opalinus Clay Formation exhibits the broadest variability in hydraulic conductivity and depth range, with in situ values extending from 10-14 to 10-4 ms-1 (n=284), depending on test depth that covers tests from a shallow zone (0–65 m) and an intermediate to deep zone (250–1065 m). Laboratory results show a significantly lower spread and cluster between 10-14 and 10-12 ms-1. Effective stress conditions applied during test performance lie between 5 and 50 MPa. The Boom Clay dataset includes 269 laboratory measurements, primarily from permeameter and migration experiments, giving K values from 10-13 to 10-10 ms-1. The lab samples stem from five boreholes in northern Belgium and cover a depth range between 59 and 366 m. In situ data from piezometer pressure-recovery tests (n=30) fall within a narrower range between 10-12 and 10-11 ms-1, derived from boreholes drilled from the HADES URL in Mol with test depths between 223–258 m. The Callovo–Oxfordian argillite, represented by 81 hydraulic conductivity entries from the Bure URL and boreholes in close proximity, shows close agreement between laboratory measurements (10-16 to 10-12 ms-1) and in situ tests (10-15 to 10-12 ms-1). Sample depths for laboratory tests range between 353–543 m, while in situ tests were carried out between 355–630 m depth. The Toarcian–Domerian argillite (To-Do) is represented by a modest dataset derived from the Tournemire URL, with laboratory K values between 10-14 and 10-13 ms-1 (n=8; sample depths between 230–339 m) and in situ results spanning from 10-14 to 10-11 ms-1 (n=6; test depths between 275–380 m). The Amaltheenton Formation displays a pronounced lab–field contrast, with laboratory values of 10-14 to 10-12 ms-1 and in situ results several orders of magnitude higher, ranging from 10-11 to 10-5 ms-1. This formation was investigated based on several shallow boreholes (<100 m depth) in the eastern margin area of the Lower Saxony Basin, Germany (Gaus et al., 2022; Castro-Vera et al., 2024; Burchartz et al., 2026). The Boda Claystone Formation is represented by a small number of water permeability measurement results that lie between 1.1×10-14 and 1.3×10-13 ms-1. Samples originate from depths around 1000 m, taken from boreholes drilled from the Alpha shaft URL (Fedor et al., 2008). In situ data of the BCF is sparse. For this study the data range given in the CCC (OECD and NEA, 2022) was adapted. It shows hydraulic conductivities in a wide scatter between 10-13 and 10-8 ms-1 and a best estimate of 1.0×10-12 ms-1. The meaning of “best estimate” for the BCF is not further differentiated in the CCC but it serves as selected or calculated reference value based on expertise from the available data (OECD and NEA, 2022).

The compiled hydraulic conductivities span a depth range from near-surface (a few metres below ground) to ∼1100 m true vertical depth. Most in situ tests (n=242) are concentrated at depths shallower than 100 m, while laboratory samples predominantly originate from deeper intervals (Fig. 4). Overall, the in situ hydraulic conductivity shows a pronounced decrease within the first 100 m of present-day depth, spanning nine orders of magnitude (10-13 to 10-4 ms-1). In contrast, laboratory data from the AMA and OPA from the same depth lie in a distinctively narrower range between 10-14 to 10-12 ms-1. Between ∼200 and ∼1100 m, laboratory and field-derived K of over-consolidated, lithified formations (OPA, COx, To-Do) converge and predominantly show clustering within a range of 10-14 to 10-12 ms-1. An exception to this is the BCF, where in situ values show a significantly wider scatter with hydraulic conductivities up to 10-8 ms-1, while lab values fall between 10-14 and 10-13 ms-1. The slightly over-consolidated Boom Clay shows a minor monotonic depth trend and a generally low scatter over a depth range of 59 to 366 m. Field and laboratory data between 223 and 260 m overlap closely, spanning from 10-12 to 10-11 ms-1. Across the moderately to strongly over-consolidated formations (OPA, COx, To-Do, BCF), the central 50 % of data at present-day depths >100 m fall within an interquartile range of 3.9×10-14 to 2.4×10-13 ms-1 (in situ) and 3.3×10-14 to 2.3×10-13 ms-1 (laboratory), corresponding to a factor of ∼6–7 in both cases, i.e., less than one order of magnitude, and in close agreement between the two datasets.

Bulk mineralogical data provide quantitative information on clay minerals, carbonates, and framework silicates (quartz + feldspar + accessories). Overall, the dataset captures a wide range of mineralogical compositions (Fig. 9), with Boom Clay and To-Do representing the extremes of carbonate-poor and carbonate-rich endmembers, respectively. Clay mineral content is highest in OPA and AMA, lowest in To-Do, while quartz + feldspar + accessory contents are generally inversely related to carbonate abundance. Generally, the mineralogical ranges of the present dataset align with those reported in the Clay Club Catalogue (OECD and NEA, 2022).

The petrophysical properties of the investigated formations show pronounced inter-formational variability (Table 7 and Fig. 10). Mean porosity values in indurated and over-consolidated formations range from below 2 % in the BCF over ∼8 % in To-Do, 11 % in AMA, 12 % in OPA, and 15 % in COx. The Boom Clay displays distinctly higher porosities of up to 40 %, reflecting its relatively shallow burial depth and limited diagenetic overprint. Grain densities show little variability across formations, with means clustering around 2.66–2.72 gcm-3, consistent with typical argillaceous mineral assemblages (Totten et al., 2002). Bulk densities generally follow porosity trends. High-porosity formations such as the Boom Clay exhibit markedly lower bulk densities (∼1.66 gcm-3), whereas the more compacted and diagenetically altered units (BCF, COx, To-Do, OPA, AMA) show bulk densities in the range of 2.42–2.46 gcm-3. Again, the BCF represents a lower-end member with a bulk density around 2.65 gcm-3.

Across the investigated formations, mean porosity values show a broadly positive correlation with laboratory-measured hydraulic conductivity (Fig. 11a), reflecting pore space closure in response to mechanical compaction and cementation (Mazurek et al., 2011). Conversely, hydraulic conductivity exhibits a negative correlation with bulk density (Fig. 11b). However, COx shows slightly lower mean K compared to OPA and AMA despite a generally higher porosity of about 3 %–4 %, illustrating that K may additionally be influenced by factors such as pore size distribution, mineralogical composition and spatial mineral distribution (Gaus et al., 2022, and references therein). Similarly, data from the BCF show slightly higher hydraulic conductivity than the Toarcian-Domerian formation, despite the BCF's distinctly lower porosity and higher bulk density.

The formations considered in this study experienced markedly different burial and uplift histories, with maximum burial depths ranging from only several decametres below ground (Boom Clay) to several kilometres (Fig. 12). Four formations (OPA, To-Do, AMA, BCF) reached depths exceeding 1000 m during their geological evolution, followed by partial to significant uplift, resulting in pronounced over-consolidation. Among these, the AMA shows the largest internal burial depth variability, with reconstructed maximum burial depths between 1400 and 3300 m (Castro-Vera et al., 2024; Burchartz et al., 2026), while the OPA reached depths of 1350–1700 m (Mazurek et al., 2006). The BCF represents the deepest end-member, with burial depths of up to 4500 m.

The porosity distribution follows a clear burial-related trend with a good exponential fit (R2=0.80; Fig. 12b), representing systematic pore-space reduction as a response to compaction and cementation processes along gradual burial (Addis and Jones, 1985; Jones and Addis, 1985). A slight deviation from this trend is observed within the AMA dataset, where porosity increases slightly between burial depths of 2440 m (BO4.0) and 3300 m (BO5.0). Castro-Vera et al. (2024) attributed this to local overpressure development at the BO5 site, which likely reduced effective stress during burial. Porosities in AMA and OPA agree closely for sites that experienced similar burial depths (e.g., BO1 ≈ Jura Ost; BO2 ≈ Mont Terri).

Hydraulic conductivity exhibits a less pronounced, more complex relationship with maximum burial depth (Fig. 12a). A substantial decrease in laboratory-derived hydraulic conductivity is evident between the shallowly buried Boom Clay and the more deeply buried Callovo-Oxfordian argillite, which faced a maximum burial depth of approximately 850 m. This pronounced decline vanishes from burial depths beyond ∼1000 m. Instead, subtle intra-formational trends can be identified. In the OPA, hydraulic conductivity decreases progressively from the Mont Terri URL to the deeper Jura Ost, Zürich Nordost, and Nördlich Lägern sites. A similar, though gentler, decrease is observed in the AMA data, with a slight rise in hydraulic conductivity at the site which experienced the deepest maximum burial (BO5.0, 3300 m), mirroring the porosity pattern and suggesting the same influence of local overpressure (Gaus et al., 2022; Castro-Vera et al., 2024; Burchartz et al., 2026).

For formations that experienced maximum burial depths <2000 m, laboratory-derived and in situ hydraulic conductivity show a close agreement, indicating consistent matrix and bulk hydraulic properties at depth. Here, discrepancies are smaller than one order of magnitude and therefore lie well within the prior defined natural variability of K. In contrast, a divergence between laboratory- and field-scale hydraulic conductivity values emerges in formations with burial depths ≥2000 m, notably in the To-Do and BCF. The latter shows differences of several orders of magnitude between laboratory and field measurements. In situ AMA data were excluded here, as near-surface processes, particularly decompaction, produced elevated hydraulic conductivity.

The development of discontinuities such as fractures, joints, and bedding planes enhances secondary permeability and can significantly increase fluid flow through the rock mass (Neuzil, 1994; Ishii et al., 2011; Neuzil, 2015, 2019). Argillaceous formations, however, possess a remarkable capacity to self-seal such dilatant features over time, a key property that underpins their consideration as potential host rock for radioactive waste disposal (Bock et al., 2010). This self-sealing capacity arises from several mechanisms, including stress-state-dependent mechanical fracture closure (increased effective normal stress), swelling of clay minerals, mineral precipitation, creep deformation, and shearing or slaking processes (Bock et al., 2010; Di Donna et al., 2022; Berry et al., 2025). Bock et al. (2010) synthesized numerous studies on the self-sealing behaviour of most of the formations investigated here (excluding the AMA), showing that self-sealing occurs across all formations, but varies on both spatial and temporal scales. Differences in sealing effectiveness are strongly controlled by clay mineralogy, degree of induration, and porewater chemistry (Bock et al., 2010). A detailed and very recent overview of self-sealing processes in argillaceous media is provided by Berry et al. (2025).

In the soft Boom Clay, self-sealing is primarily governed by clay mineral swelling (discrete smectite fraction of the clay mineral components is at 7 wt %–24 wt % and 7 wt %–23 wt % illite-smectite interlayers; after: Zeelmaekers et al., 2015) and occurs over short time periods, leading to a near-complete recovery of the initial matrix hydraulic properties. Within the SELFRAC project, this process was shown to be efficient across multiple scales, although complete mechanical healing remains limited (Bastiaens et al., 2007; Van Geet et al., 2008; Gonzalez-Blanco et al., 2024). The close agreement between laboratory and in situ hydraulic conductivity measurements further underlines the strong self-sealing capacity of Boom Clay, indicating effective closure of potential fractures at the rock-mass scale. For the moderately indurated COx and OPA formations, numerous studies have examined self-sealing behaviour with a focus on swelling and mechanical consolidation processes (Van Geet et al., 2008; Zhang et al., 2010; Zhang, 2011; Di Donna et al., 2022; Wang et al., 2024, 2025). Swelling-induced sealing is efficient in both formations, and depending on mineralogical composition and confinement, hydraulic conductivity can be restored to values close to the initial matrix level (Voltolini and Ajo-Franklin, 2020; Wang et al., 2022a, 2025; Zhang and Talandier, 2023). Discrete swellable clay mineral phases (mostly smectite) occur in COx at ∼11 wt % to 26 wt % of the clay mineral fraction and as illite-smectite (I-S) mixed layers of >20 wt % (Hu et al., 2014; Wang et al., 2022a). While OPA is depleted in discrete smectite phases, it contains 10 wt % to 38 wt % I-S (Kneuker et al., 2023, 2026).

In the moderately indurated formations (COx, OPA), self-sealing typically occurs over time spans of months to a few years (Bock et al., 2010) and is dominated by swelling, consolidation, and creep mechanisms (Bastiaens et al., 2007). This is reflected in the presented dataset, which shows close agreement between laboratory and field-scale hydraulic conductivity values, indicating efficient sealing of natural or excavation-induced discontinuities, provided the absence of near surface perturbations e.g., due to decompaction and weathering. This interpretation is further corroborated by the findings of Gautschi (2001), who, based on observations from more than 6.6 km of tunnel sections excavated through the Opalinus Clay in northern Switzerland, reported that even in faulted and folded domains only negligible water inflows occurred. Notably, all recorded inflows were restricted to sections with an overburden of less than approximately 200 m. In contrast, self-sealing appears to be diminished in the shallow OPA zone, as evidenced by elevated and variable hydraulic conductivity values in shallow boreholes in the Swabian Alb (Hekel, 1994) and the Lausen borehole (Vogt et al., 2017; Crisci et al., 2019). This depth contrast highlights the role of confinement, which in combination with clay mineral swelling, results in effective self-sealing processes (Voltolini and Ajo-Franklin, 2020).

Toarcian-Domerian argillites exhibit overlapping in situ and laboratory hydraulic conductivity ranges, mostly within the the pre-defined natural variability. However, localized zones of higher conductivity (10-11 ms-1) identified in field experiments suggest the presence of unsealed, hydraulically active fractures (Boisson et al., 2001). Additional evidence of water inflow zones in boreholes (Beaucaire et al., 2008) and minor moist areas in the Tournemire URL (Mazurek et al., 2009) supports this interpretation. These observations indicate a lower self-sealing efficiency compared to the less indurated formations (Boom Clay, COx, and OPA). According to Bock et al. (2010), a conservative threshold of approximately 40 wt % clay mineral content marks a threshold below which self-sealing becomes ineffective. The mineralogical composition of To-Do lies near this threshold, suggesting that swelling-driven sealing is less pronounced. In addition, swelleable clay mineral phases lay at ∼9 wt %–15 wt % of the total clay mineral fraction, notably below those of Boom, COx, and OPA (Bock et al., 2010). Geng et al. (2021) observed substantial sealing during compaction-creep deformation in To-Do samples, but primarily under coupled THCM conditions corresponding to simulated depths around 3.8 km, i.e., conditions not relevant for repository environments.

In the case of the BCF, the highly indurated nature and general depletion of swellable clay minerals (smectite content near the detectable limit of >1 wt %) do not favour rapid self-sealing (Kovács, 2001). The pronounced discrepancy between small-scale (laboratory) and large-scale (rock mass) hydraulic conductivity is evident in the compiled dataset and likely reflects low self-sealing capacity. However, Kovács (2001) reported partial sealing due to degradation of non-swelling clay minerals and mineral precipitation within joints. Such processes may contribute to very long-term sealing and should be considered in detailed repository performance assessments although their timescales likely exceed the time-span over which a host-rock is required to prevent radionuclide migration to the environment (e.g., 1 Ma in Germany; StandAG, 2017) (Bock et al., 2010). The transition from sedimentary to metamorphic rock marks the ultimate threshold beyond which self-sealing becomes ineffective (Bock et al., 2010).

The self-sealing behaviour of the Amaltheenton Fm has not yet been characterized but is currently being investigated within the framework of the MATURITY project (Burchartz et al., 2026).

The general diagenetic changes occurring during progressive burial were outlined in the introductory chapter. The compiled data clearly demonstrate that maximum burial exerts a lasting influence on both the hydraulic and petrophysical characteristics of the investigated formations. This influence is particularly evident in the relationship between porosity (or inversely, bulk density) and maximum burial depth observed across the dataset. An initial porosity loss exceeding 20 % is observed between the Boom Clay and the COx, corresponding to a differential burial of approximately 600 m. This highlights the pronounced porosity reduction caused by mechanical compaction during early burial diagenesis (Athy, 1930; Addis and Jones, 1985; Cripps and Czerewko, 2017; Ewy et al., 2020). At this stage, the most substantial reduction in hydraulic conductivity is also observed, while both the porosity-depth and hydraulic conductivity-depth trends flatten with further burial. The initial pore closure associated with mechanical compaction thus represents an important control on the low hydraulic conductivity characteristic of argillaceous formations. Neuzil (1994) indicates the existence of a log-linear relationship between porosity and permeability/hydraulic conductivity across a broad range of argillaceous formations, a trend also reflected in the data compiled for this study. This relationship is directly resolved within OPA, where region-specific porosity and laboratory K show a strong log-linear correlation (R2=0.81, n=4; Fig. 13). A comparable, though less consistent, relationship is observed for AMA. Across the investigated formations, a broad porosity-K relation can be observed that is comparatively weaker (Fig. 11a), reflecting the additional influence of pore size distribution, mineralogical composition, and grain size distribution across formations of differing genesis. This is specifically evident from the COx data that shows a slightly lower geom. mean K (6.31×10-14 ms-1) while exhibiting higher mean porosity (14.72 %) compared to OPA (1.18×10-13 ms-1; 11.58 %). This might be as a result of early diagenetic precipitated non-biogenic carbonates that potentially results in the clogging of pore throats, reducing water accessible flow paths and associated K (Esteban et al., 2007; Yven et al., 2007; Menaceur et al., 2015; Robinet et al., 2015). Nevertheless, the general reduction in K with decreasing porosity across the investigated formations can be traced to maximum burial depths of around 2000 m, although the trend becomes less pronounced with depth. Beyond this threshold, no further systematic decrease in hydraulic conductivity is apparent. It can therefore be concluded that hydraulic conductivity remains positively correlated with porosity down to porosity values of roughly 10 %, or better burial depth <2000 m, which is in in agreement with previous findings by Mazurek et al. (2011).

At maximum burial depths exceeding 2000 m, a divergence emerges between laboratory-derived matrix hydraulic conductivity and field-derived rock mass values, with the latter increasing up to several orders of magnitude. This divergence likely reflects the growing hydraulic influence of fractures superimposed on an otherwise low-conductivity matrix (Laubach et al., 2009; Gale et al., 2014). As discussed in Sect. 5.1, burial depth also exerts a strong control on the self-sealing efficiency of argillaceous formations, which progressively lose their swelling capacity with increasing induration. This loss is closely linked to the time- and temperature-dependent transformation of smectite to the thermally more stable illite via an intermediate mixed-layer (I/S) phase (Pollastro, 1993; Peltonen et al., 2009), described by the reaction: 3smectite+K+⟶time and temperatureillite+quartz+nH2O This reaction results in the release of interlayer water, the crystallization of authigenic quartz, the formation of smaller and stiffer crystal structures (Carcione et al., 2022), and consequently, further porosity loss, increased brittleness, reduced ductility (Berthonneau et al., 2017), and diminished swelling potential. In contrast to moderately indurated formations such as the COx and OPA, where effective self-sealing can counteract long-term fracture permeability, highly indurated formations such as the BCF exhibit the combined effects of increased brittleness and strongly reduced swelling potential, preventing efficient fracture sealing. As noted by Mazurek et al. (2009), the To-Do formation represents a transitional case, where the indurated matrix allows for fracturing while retaining a limited, though not entirely lost, self-sealing capacity.

A pronounced present-day depth trend of hydraulic conductivity reduction is apparent from the indurated and over-consolidated formations, which is in agreement with global datasets published earlier (see e.g., Appel and Habler, 2001; Batlle-Aguilar et al., 2016; Asem and Gardoni, 2022). No pronounced depth trend could be observed on the soft Boom Clay. In the (moderately) indurated and over-consolidated formations OPA and AMA, elevated and variable in situ hydraulic conductivity is observed in the upper metres to decametres of depths. This trend diminishes with depth. At depths >250 m in situ and laboratory derived hydraulic conductivity overlaps in all formations that experienced maximum burial <2000 m. Stronger indurated formations (To-Do and BCF) showed scale dependency of the hydraulic conductivity with small scale (laboratory) values fitting the natural variability (10-14 to 10-12 ms-1), while large scale (in situ) hydraulic conductivity can be substantially higher. Based on these observations two distinct domains depending on the extent of decompaction can be differentiated.

The dataset compiled for this meta study integrates hydraulic, petrophysical, mineralogical, and burial history data from six argillaceous formations, spanning laboratory and field scales. While this data allows systematic links between burial evolution and hydraulic behaviour to be drawn across formations, it also introduces several sources of heterogeneity that should be considered when interpreting the results.

First, the compiled data originate from a wide range of studies conducted over several decades, employing different testing methods, test conditions, and sampling protocols. While we account for this variability through careful review of the included references, such heterogeneity introduces potential bias that cannot be fully excluded. For example, sample orientation with respect to bedding is not reported for all laboratory measurements. Although, the anisotropy observed in this study is comparatively low (Kp/Ks=∼1.1 to 2.7), studies suggest that the anisotropy ratio can yield values up to 5 or even exceeding these (Clennell et al., 1999; Bastiaens et al., 2007; Bock, 202; Neuzil, 2019). Such methodological differences represent an additional, largely unquantifiable source of scatter contributing to the natural variability envelope defined in Sect. 4.3.4.

Second, argillaceous formations are internally heterogeneous at the facies scale (e.g., variations in carbonate type and content, grain size distribution, or bedding structure within a single formation), which cannot be fully resolved in the present compilation. Moreover, the properties reported here were measured at specific boreholes or sampling locations and may not be representative of the same formation elsewhere. Formations of broadly similar lithological composition and depositional origin will exhibit exactly the same hydraulic behaviour. Thus, the trends identified in this study should not be extrapolated to other formations or occurrences without independent verification. The formation-averaged hydraulic behaviour discussed here may therefore mask internal variability that is relevant at the scale of a specific repository site.

Third, the reconstruction of maximum burial depth for each formation carries formation-specific uncertainty. Burial estimates were adopted from published reconstructions rather than independently determined, and rely on different methods depending on data availability (e.g., basin modelling, vitrinite reflectance, apatite fission track analysis), which are not equally well constrained across all six formations. For example, for To-Do, the Clay Club Catalogue reports 1300–1700 m of maximum burial, while other references (e.g., Mazurek et al., 2008, 2011), report approximately 2000 m. For the AMA in the Hils syncline, the maturity trend is very well established (Littke et al., 1988; Burchartz et al., 2026), but calculated maximum temperatures and maximum burial depths vary, depending on (i) basal heat flow, (ii) selected lithologies of eroded sediments, and (iii) the algorithm describing the kinetics of vitrinite reflectance increase (Bruns et al., 2016; Froidl et al., 2020).

Fourth, the number of available measurements differs substantially between formations and analyses. For formations that faced maximum burial >2000 m, little hydraulic data is publically available, constraining derived analysis to few measurements or data ranges. The trends identified for these categories should be regarded as indicative rather than conclusively established. Additionally, data from maximum burial depths <850 m is represented by a single formation (Boom Clay).

Addressing these gaps, through additional first-hand data from underrepresented depth ranges and formations, and more systematic reporting of pore structure and mineralogy across future studies, remains an important direction for future work.

The compiled and discussed data indicate that maximum burial depth, and the associated stress and temperature history, emerge as a primary and consistent control of hydraulic conductivity across the investigated formations. Together with mineralogical composition, and spatial mineral and pore-size distribution, these burial-related processes govern the spatial and temporal evolution of hydraulic conductivity in argillaceous formations.

Fluid flow through consolidated, stiff clay formations such as Boom Clay, and slightly to moderately indurated argillaceous rocks (COx, OPA) can be adequately described by means of classical continuum concepts, representing the material as porous media (Bock et al., 2010). In these cases, where formations remain unaffected by weathering and decompaction, the matrix hydraulic conductivity is representative of upscaled rock mass properties. At both scales, diffusion can therefore be considered the dominant solute transport mechanism. In contrast, highly and very highly indurated and brittle formations show increasing influence of fracture flow so that matrix and rock mass hydraulic conductivity diverge as advective processes dominate the hydraulic behaviour on a rock mass scale. In such cases, discrete fracture network approaches (such as done by Tóth et al., 2022, for BCF) provide a more meaningful framework for understanding fluid flow and solute transport processes within these formations (Bock et al., 2010).