Abstract

Bentonite is a potential material for use in the engineered barrier of radioactive waste repositories because of its low hydraulic permeability, self-sealing capability and retention capacity. It is expected that bentonite would react at the elevated temperatures accompanying the radioactive decay in the nuclear waste. The presented study was started in order to improve understanding of the coupled influence of temperature and (pore) water on the physicochemical and mineralogical properties of bentonite during thermal treatment under near-field relevant conditions. Granular Na-bentonite MX-80 was differently saturated (Sr = 1–0.05) and stored at different temperatures (50–150°C) in a closed system. Upon dismantling after different periods of time (3 to 18 months), mineralogical characteristics, cation exchange capacity and content of leachable cations, as well as physicochemical properties such as surface area and water adsorption were investigated.

The results showed a high mineralogical stability. A slight conversion from the sodium to an earth alkali form of the bentonite was observed. However, considerable changes in the physicochemical properties of the bentonite were observed, particularly by treatment above the critical temperature of 120°C. The cation exchange capacity decreased during heating at 150°C by approximately. 10%. The specific surface area dropped by more than 50%. The water uptake capacity under free swelling conditions showed a slight tendency to lower values especially for samples heated for more than 12 months. The water vapour adsorption ability in contrast drops by 25% already within three months at T = 120°C. These changes are mostly related to the variations in the interlayer cation composition and to smectite aggregation processes. The observed alterations are rather subtle. However, temperatures ⩾120°C had a remarkable negative influence on different properties of MX-80.

In several countries, a deep geological repository is foreseen as the most suitable solution for the issue of radioactive waste disposal (Russell & Facella, 2006). Such an underground facility should isolate radioactive waste from the biosphere and ensure the dissipation of the heat of decay (Kahr & Müller Vonmoos, 1982). Currently designed repositories for radioactive waste are multibarrier systems composed of the natural barrier (host rock) and a combination of several engineered barriers (Nagra, 2002). An important part of the engineered barrier system is the bentonite backfill. The use of bentonite as the engineered barrier is of crucial importance because of its low hydraulic permeability, self-sealing capability and retention capacity (Pusch, 2001). Bentonite is a natural material, predominantly consisting of smectite, and represents a product of the early diagenetic alteration of vitric fallout ash in subaqueous, mainly marine environments (Chamley, 1989). Bentonite shows a great stability against weathering (Laine & Karttunen, 2010). Nevertheless, its properties can alter under harsh conditions such as elevated temperature and pressure or wetting-drying cycles (Galán, 2006; Pusch, 2006). The exposure of bentonite to hot steam leads to a decrease in swelling capability and to an increase in hydraulic conductivity, all of which are also mirrored in the bentonite microstructure (Couture, 1985; Pusch, 2000). It is expected that bentonite would react at the elevated temperatures accompanying the radioactive decay in the nuclear waste (Sato et al., 1998). Temperatures below 50°C in the repository bentonite buffer and its full saturation are expected within 100–200 years (Senger & Ewing, 2008). The swelling behaviour and geochemical reactions of bentonite have been studied by saturation with pore water solutions with different chemical composition and ionic strength under confined conditions at elevated temperatures in short-term laboratory experiments (Montes-H & Geraud, 2004; Herbert et al., 2008; Kaufhold & Dohrmann, 2009, 2010, 2011; Gómez-Espina & Villar, 2010; Perdrial & Warr, 2011; Kasbohm et al., 2012). Only minor alteration of the bentonite was reported. The described initial transformation of smectite to interstratified illite-smectite, and the decrease in layer charge are rather insignificant. Significant chemical changes are related to cation exchange processes only. The sorption ability of MX-80 remained unchanged up to 250°C of heating for 24 h.

Several in situ experiments have been started in order to investigate the behaviour of bentonite under elevated temperature in the potential host rock in order to study the far-to-near-field conditions in detail. Some of these projects are the FEBEX in Grimsel (enresa, 2000) and the Heater Experiment in Mont Terri (Göbel et al., 2006) both in Switzerland as well as the Alternative Buffer Materials project (ABM; Svensson et al., 2011), and the Long Term Test of Buffer Material project (LOT; Olsson & Karnland, 2011), both situated in theÄspö Hard Rock Laboratory, Sweden. Generally, these experiments demonstrated a high chemical and mineralogical stability of the bentonite during thermal treatment. However, not all experiments showed the same reactions. The investigation of the thermo-hydro-mechanical (THM) properties of the bentonite after five years treatment at repository conditions in the FEBEX experiment revealed unchanged retention and swelling capacity (Villar & Lloret, 2007). The analysis of the treated bentonite from the HEATER experiment discovered slight alterations related to cementing processes (by precipitation of amorphous Si/Al compounds). The specific surface area (SSA) decreased by about 20%. Besides the interlayer cation exchange an insignificant decrease of the mean layer charge (MLC) was detected (Plötze et al., 2007). In the ABM project, besides slight tendencies of cation exchange, an increase in the cation exchange capacity (CEC) of the bentonite was measured in the direction of the heating source (Nagra, 2011; Svensson et al., 2011). However, the differences appear to border the line of analytical error. As in the HEATER experiment, a significant decrease of the SSA of the MX-80 bentonite was measured as well. Additionally, in the LOT project, carbonate dissolution and gypsum dehydration in MX-80 according to the temperature gradient were observed. Close to the heater in the A2 parcel test, calcite was dissolved and gypsum dehydrated within the almost six-years of heating whereas gypsum precipitated in the colder and more humid part near the host rock (Olsson & Karnland, 2011).

The present study was started in order to improve the understanding of the coupled influence of temperature and (pore) water on the physicochemical and mineralogical as well as on the hydraulic and mechanical properties of bentonite during thermal treatment. The temperature and hydration state of the bentonite were adapted to the expected conditions in the near-field of a planned repository for radioactive waste (Fig. 1). In the experiment the Na-bentonite MX-80 was used, which is one of the possible buffer materials. The bentonite was saturated with artificial pore water of the same chemical composition as the pore water of the Opalinus clay – a potential host rock for the radioactive waste repository in Switzerland. The thermal treatment was carried out under laboratory conditions in a small-scale closed system, providing full control of physical and chemical factors (mass and bulk density of the bentonite, content and chemical composition of the pore water, and temperature). The time scale of the experiment with more than three years of treatment will be in the same range as most of the in-situ long-term experiments.

The chosen set-up allows observation in short time steps of the initial changes in the important bentonite properties under near-field relevant conditions comparable with large-scale in-situ experiments but in a closed system under controlled environmental conditions.

Materials and Description of the Experiment

Bentonite

The Na-bentonite MX-80 (Wyoming, AMCOL International Corp.) was used in this experiment. The mineralogical composition of the bentonite is characterised by a high content of smectite and about 10 wt.% of mica-like minerals and tectosilicates each. In traces, cristobalite, calcite and gypsum were found but also pyrite (Table 1). The bentonite is a granular material from the ESDRED experiment (Engineering Studies and Demonstration of Repository Design), where different granulometric distributions were tested in emplacement studies (Plötze & Weber, 2007). The chosen granular mixture ESDRED “E” (uniformity index CU = 13.22, curvature index CC = 1.65, d50 = 1.61 mm, dry bulk density of bentonite granules ρd = 1.962 g/cm3) achieved the highest dry emplacement density (ρd = 1.513 g/cm3). Various portions of bentonite (26–46 g) were filled with a dry bulk density of ρd = 1.5 g/cm3 in tight containers (Fig. 2). The containers consisted of an ematal-coated aluminium shell with an inner PTFE casing. The PTFE casing prevented contact of the bentonite material with the aluminium shell to avoid the formation of corrosion products in order to exclude an uncontrolled chemical influence. The containers had an inner diameter of 40 mm and an inner height of 25 mm.

Artificial pore water

The bentonite was saturated with artificial pore water (APW) at five saturation degrees (Sr): 0.05, 0.25, 0.5, 0.75 and 1 (compare Fig. 1) corresponding to ratios of solution to dry clay weight of 0.02, 0.08, 0.19, 0.33 and 0.53, respectively. The chemical composition of the APW (pH 7.9, alkalinity 0.55, ionic strength 0.23, Na-Ca-Mg-K type, main cation Na) corresponded to the pore water of the Opalinus Clay formation in the Brown Dogger facies (Mäder, 2009). The precise chemical composition of the APW (Table 2) was calculated using Phreeqc (version 2.15.07) software and NAGRA-PSI database Version 01/01.

Thermal treatment

The variably saturated bentonite were stored in the tight containers at temperatures (50, 65, 105, 120, and 150°C) according to the expected temperature conditions at the different distances between the canister with the radioactive waste and the host rock (Fig. 1; Senger & Ewing, 2008). Accordingly, bentonite with Sr = 1 was stored at 50°C (vicinity of the host rock) and the Sr = 0.05 sample at 150°C (vicinity of the canister). One container was prepared for each temperature level and heating time. The heating time intervals were 3, 6, 12 and 18 months. The material was then taken out of the container, dried overnight at 50°C, crushed in an agate mortar and homogenized for further analysis.

Analytical Methods

Mineralogical characterization

X-ray diffraction analysis. The samples were characterized by X-ray powder diffraction using a Bruker AXS D8 Advance diffractometer (source Co-Kα radiation λ= 1.790 Å) equipped with primary and secondary soller and automatic theta compensating divergence and antiscatter slits. Data were collected from 2–80°2θ (2–40°2θ for textured specimens) with a step size of 0.02°2θ and a counting time of 4 s/step. The mineral composition of the samples was determined on randomly oriented powder and textured smear specimens. The samples were micronized in ethanol to <20 μm using a McCrone micronizing mill equipped with agate grinding elements. After drying, the samples were front-loaded into the sample holders by taping a razor blade onto the sample within the sample holder in order to, both, maximize the random orientation of the minerals and to achieve a flat surface of the sample (Zhang et al., 2003). The textured specimens were prepared to easier identify the clay minerals. The basal reflections of the oriented clay minerals are enhanced through this technique (Brindley & Brown, 1980). An amount of each sample was mixed with deionised water. The paste was smeared onto the flat glass, dried in the laboratory atmosphere and analysed. The textured specimens were then saturated with ethylene glycol (EG) to examine smectite swelling (MacEwan & Wilson, 1980). The (00l) diffraction peaks can be observed as a rational series of d values for a discrete clay mineral. The coefficient of variation (CV) for the (00l) diffractions of samples saturated with propane–1,3–diol (instead of EG) were determined in order to detect the possible formation of interstratified irregular illite/smectite with a CV⩾0.75 (Moore & Reynolds, 1997). The XRD was also used for the determination of the mean layer charge (MLC). The samples were intercalated with dodecyl ammonium (chain length nC = 12) and washed with ethanol (Lagaly, 1994). A few milligrams of talc were added to the intercalated samples as a line positioning standard for the XRD pattern. The prepared suspension was then dropped onto a glass slide, dried in vacuum at 40°C and immediately measured. The MLC was calculated as MLC = [d001 - 5.52]/32.98 (Olis et al., 1990). The determination of the basal spacing, as well as the qualitative phase analyses were carried out by processing the diffraction pattern in the DIFFRACplus EVA software (Bruker AXS). The quantitative composition was calculated by Rietveld analysis using the BGMN®/AutoQuan software (Bergmann & Kleeberg, 1998).

FTIR spectroscopy. The Fourier Transform Infrared Spectroscopy (FTIR) was used as an additional method for the mineral composition analysis. Particularly, the presence of amorphous silica and the ion allocation in the octahedral position in smectite were investigated (Madejová et al., 2011). The IR spectra were measured in the middle-infrared region (4000–380 cm–1) with a Perkin Elmer FTIR 2000 device. The measurements were carried out in transmission mode using KBr pellets with two different weights (0.5 mg and 3 mg in 200 mg of KBr) for each sample.

Thermal analysis. Thermogravimetric and mass spectrometric analysis (TG-MS) was used as additional diagnostic tool particularly for the investigation of the dehydration and dehydroxylation characteristics of the smectite, the latter particularly for the analysis of the cis-/transvacancy character of its octahedral sheet (Emmerich, 2011). The measurements were carried out with a thermobalance Mettler-Toledo TGA/SDTA851e. With the coupled mass spectrometer Balzers ThermoStar GSD 300T the evolved water, CO2 and SO2 were investigated as a part of the mineral composition analysis. 110 mg of each sample was thermally analysed in a platinum crucible over the range 25–1000°C (heating rate 10°C/min) in a dry and CO2-free airflow atmosphere (50 mL/min).

Cation exchange capacity, leachable cations and pore water chemistry. The CEC of the samples was determined using the complexes of copper (II) with triethylenetetramine (Meier & Kahr, 1999). The CEC should be comparable with the amount of exchangeable cations that entered the exchanging solution. However, this amount of exchangeable cations is not only represented by the interlayer cations, but also includes the dissolved cations in the pore water (including the added APW) and additionally, the cations from the soluble minerals dissolved during the exchange procedure. Therefore, the term “amount of exchangeable cations” was replaced here with the “amount of leachable cations” (ALC). The ALC was determined by the analysis of cation concentrations (Na+, K+, Mg2+, Ca2+) in the exchanging solutions after CEC measurement using atomic absorption spectrometry (AAS; device Perkin Elmer AAnalyst400). The chemistry of the pore water itself was determined by three times washing 2 g of each sample (in replicates) in 15 mL methanol. The suspension was centrifuged and the supernatant decanted and analysed by AAS.

Physical and physicochemical properties

Specific surface area determination by gas adsorption. The SSA represents ideally the external surfaces of crystals excluding the interlayers of sheet silicates but is for bentonites also influenced by the microporosity (Kaufhold et al., 2010b). The SSA of the samples (300–700 mg) was determined by N2 adsorption using an 11-point BET method (Brunauer et al., 1938). The adsorption isotherm measurements were carried out with a Quantachrome Autosorb-1MP surface area analyser (N2 relative pressure p/p0 0.05–0.3).

Water vapour adsorption and water uptake under free swelling conditions. The ability of MX-80 to adsorb water vapour was determined in an atmosphere with a relative humidity (RH) of 75% (Keeling et al., 1980), which was achieved in a closed desiccator above an oversaturated NaCl solution (Madsen & Kahr, 1992). The sample weight was checked until constant weight (after ∼6–9 weeks). The water content of such a treated sample (dried at 105°C until weight constancy) (DIN ISO 11465, 1996) represents the level of the water vapour adsorption ability (WVAA).

The uptake capacity for liquid water under free swelling conditions was determined with the Enslin-Neff technique (DIN 18132, 1995; Kaufhold et al., 2010a). About 85 mg of each sample (particles <400 μm) was used for the analysis. The water-uptake capacity (WUC) was related to the initial dry weight of the bentonite, taking into account the separately determined initial water content. The influence of evaporation during the measurements was eliminated by subtracting the evaporation determined in a separate Enslin-Neff apparatus filled with milled quartz.

To investigate the possible influence of heating on the gas adsorption behaviour of the bentonite (smectite aggregation/cementation, Couture effect), the SSA and WVAA were measured on selected samples again after remoulding. It is expected that the shearing in excess of water will break up the aggregates. The remoulding was carried out by mixing a certain amount of the differently treated bentonite (approximately 2 g) with deionized water and mechanically shearing with a pestle to form a homogeneous paste without lumps. After air-drying at ambient conditions, the samples were crushed in an agate mortar and homogenized.

Results

Mineralogical characterization

The mineral composition of the pristine and the thermally treated MX-80 is shown in Table 1. The bentonite underwent some slight changes in mineral composition during the thermal treatment. The gypsum content in the pristine material was 1.0±0.3 wt.% and did not change in the samples stored at temperatures below 105°C. No gypsum was found in the samples stored at T > 65°C due to its dehydration and simultaneous bassanite formation. The calcite content in the pristine and treated MX-80 varies between 0.5–1.0±0.3 wt.%. However, the ambiguous tendency of its decrease with temperature borders the line of the analytical error. No amorphous silica was detected by FT-IR analysis (Fig. 3a,b). The smectite content seems to be slightly reduced over time regardless of treatment temperature. The sum of mica and mica-type layers/collapsed smectite layers showed a weak opposite tendency (Fig. 4a,b). However, this increase is in the range of measurement error as well as additionally influenced by the heterogeneity of the pristine MX-80 and therefore rather ambiguous. The coefficient of variation (CV) in the basal spacing was determined for two specific samples to prove this increase of the content of mica-type layers as a tendency of illitization, which would be probably connected with the formation of interstratified illite/smectite (I/S). The CV of the pristine material was 0.2. For the sample with the highest content of mica and mica-type layers/collapsed smectite, which was the sample heated at 120°C for 12 months, a CV of 0.38 was calculated. This insignificant increase does not indicate illitization. The slight increase of the MLC from 0.27–0.28 for samples heated up to three months to values of 0.29–0.30 for longer heated samples seemed to show the transformation of the smectite into a higher-charged form (Table 3). However, considering the measurement error of 6–13% (Lagaly & Weiss, 1969) all measured changes of MLC lay within the measurement error. The thermal analyses revealed for bentonites with higher saturation degree Sr = 1 and 0.75 stored at 50 and 65°C, respectively, a shoulder at approximately 140°C in their dehydration curves (Fig. 5a). The dehydroxylation curves showed an unchanged cis-/trans-vacant character of the smectite in all samples (Fig. 5b). No change in the basal distances (17.0 Å) of the smectite saturated with EG was observed (Fig. 6). The small shift of the basal (00l) spacing of smectite from pristine 12.3 Å towards 12.5 Å would indicate a partial replacement of interlayer Na+ by bivalent cations. However, these measurements were not carried out under controlled relative humidity and are therefore not entirely unambiguous.

Cation exchange capacity, leachable cations and pore water chemistry

No significant changes in the CEC were observed for samples stored at temperatures up to 120°C. A slight reduction seemed to occur after six months of heating. On longer heating, the CEC increased and even exceeded the value of the pristine material. However, these variations lay within the measurement error and were also influenced by the heterogeneity of the MX-80. However, the CEC of the MX-80 samples stored at T = 150°C showed a distinct decrease in CEC of approximately 10% (Table 4, Fig. 7a).

No clear trends in the leachable Na+, K+, Mg2+ and Ca2+ concentrations with treatment time and temperature could be observed. The amount of leachable Na+ of the bentonite MX-80 is, as expected, the highest in comparison with the other cations (57±4 cmol(+)/kg in pristine MX-80; Table 4, Fig. 7b). However, considering the measurement error, a slight decrease could be observed for samples stored at 150°C and at 120°C for 18 months, which is similar to the trend in CEC.

Variation in the concentrations of leachable cations was influenced by the amount of added APW. In the samples with higher saturation degrees, the higher amount of added Mg2+ was mirrored in the higher amount of leachable Mg2+ (Table 4, Fig. 7c). The amount of leachable Ca2+, instead, seemed to be more dependent on the dissolved Ca2+ from the soluble Ca-phases. The dissolution of calcite and gypsum/bassanite is a known effect that occurs during CEC determination when using complexes of Cu-triethylenetetramine (Dohrmann & Kaufhold, 2010). The concentration of leachable Ca2+ of ∼30 cmol(+)/kg is similar for all samples and originates from the dissolution of calcite and gypsum in the pristine material during the CEC analysis.

The major cation in the pore water of the MX-80 Na-bentonite was, as expected, Na+. The content in the initial pore water of the pristine MX-80 was 1.8±0.3 cmol(+)/kg. The concentration of Na+ in the pore water decreased from approximately 9 to 1 cmol(+)/kg. The values of K+, Mg2+ and Ca2+ in the pore water of both heated and pristine bentonite, were below the limit of detection by AAS (Table 4; compare with Table 1).

Specific surface area

The specific surface area (SSA) of MX-80 decreased with increasing temperature over time (Table 5, Fig. 8a). The SSA of the pristine material (33 m2/g) dropped by more than 50% to 15 m2/g for the material stored at 150°C for 18 months. The samples stored at other temperatures followed a similar trend. Even after remoulding, none of the samples reached the SSA values of the pristine material (Table 5, Fig. 8b). Only the SSA of samples stored at 120°C for 12 and 18 months showed a clear trend of recovery reaching the values of samples heated for 3 and 6 months. Rather a tendency of a decreasing SSA was observed for samples stored at 50°C.

Water vapour adsorption

The water vapour adsorption ability (WVAA) of the pristine MX-80 stored at 75% RH atmosphere was 17±0.3 wt.%. This value did not change up to 105°C (Table 6, Fig. 9a). The WVAA decreased over time for MX-80 thermally treated at T > 105°C to values of 13–12 ± 0.3 wt.%. The measurements of the remoulded samples showed a clear recovery of the WVAA reaching the value of the pristine material (Fig. 9b).

Water uptake under free swelling conditions

The WUC of the heated samples did not change up to 105°C. Even taking into account that the variations of the WUC, values lay within the measurement error of about 25%, a tendency to decrease with time could be observed for samples heated at T ⩾ 105°C for more than 12 months (Table 7, Fig. 10).

Discussion

The analysis of the Na-bentonite MX-80 ESDRED “E” treated under near-field thermal and saturation conditions in a closed system showed an overall good stability of its mineral composition and physicochemical properties over the 18 month test period. Elevated temperatures (⩾105°C) causes some minor changes.

The smectite content seems to slightly decrease over time regardless of treatment temperature. However, this decrease is in the range of measurement error as well as additionally influenced by the heterogeneity of the pristine MX-80 and is, therefore, rather ambiguous. Interestingly, the sum of mica and mica-type layers/collapsed smectite layers showed a weak opposite tendency (Fig. 4a, b). These changes could possibly indicate an initial illitization, which would be probably connected with the formation of interstratified illite/smectite (I/S). However, the insignificant increase of the coefficient of variation (CV) in the basal spacing from 0.2 of the pristine material to 0.38 of the sample with the highest content of mica and mica-type layers/collapsed smectite does not reflect a formation of interstratified I/S. The slight increase of the MLC from 0.28 to 0.30 for longer heated samples reflects more likely the formation of more highly charged smectite, rather than the formation of illite. In contrast, a very weak decrease of the MLC of smectite from initial 0.35 to 0.33±0.01 was found in the HEATER experiment (Plötze et al., 2007). However, all these insignificant changes lie within the measurement error. Nevertheless, significant mineral changes in MX-80 were predicted by modelling its reaction with a representative geological fluid from Callovo-Oxfordian (COX) formation (Gaucher et al., 2006) under repository thermal conditions (T = 100°C) (Montes-H et al., 2005; Marty et al., 2010). Formation of saponite, vermiculite, illite, and chlorite by simultaneously decreasing content of smectite within 10,000 years was described.

The CEC is one of the most valuable properties for assessing the radionuclide retention of bentonite. Therefore, bentonite alteration connected with changes in CEC is an important process to monitor. A weak increase of the CEC towards the heating source was observed in the in-situ ABM and LOT experiments (Nagra, 2011; Svensson et al., 2011, Olsson & Karnland, 2011). The investigation of long-term heated samples in an open system showed, after an initial decrease of the CEC within one year, a slight increase on further heating at temperatures >120°C (Valter & Plötze, 2010). The tendency for CEC increase over time after an initial decrease was also observed in the present work. However, the measured CEC variations are close to the analytical error of the method in all these experiments. In contrast, in the present work, the CEC of the samples stored above the critical temperature of 120°C decreases non-ambiguously. The same observation was reported by Gómez-Espina & Villar (2010). This decrease in CEC cannot be explained with the measured changes in MLC. The lower CEC is most probably caused by a stronger fixation of the exchangeable cations and the higher content of mica-type layers/collapsed smectite.

Olsson & Karnland (2011) reported in the LOT project a redistribution of Ca-sulfate phases. Their results indicate the dissolution of gypsum in the colder part of the bentonite barrier at the contact to the host rock with ground water and the precipitation of anhydrite in the bentonite close to the heater within almost six years. These observations could be reproduced also by modelling (Itälä & Olin, 2011). In the closed systems of our experiment, such redistribution was excluded. Gypsum was dissolved in the samples with higher saturation degrees Sr > 0.5 at temperatures T < 105°C. The Ca2+ and Mg2+ from the dissolved gypsum and from the added APW exchanged the Na+ in the smectite interlayer. This is clearly visible in the dehydration curves of the bentonites in thermal analysis. The dehydration curves show a shoulder at approximately 140°C, which corresponds to the release of the second water layer attached to the bivalent cations (Emmerich, 2011). In the samples stored at temperatures > 65°C gypsum dehydrated to bassanite but not to anhydrite. For this thermal decomposition probably higher temperatures or longer time are necessary.

The type of adsorbed and exchangeable cations influences such mechanical and physicochemical properties like plasticity and shear strength, water adsorption and swelling, specific surface area, microstructure and hydraulic conductivity. The amount of leachable Na+ of the bentonite MX-80 is, as expected, the highest in comparison with the other cations (Table 4, Fig. 7b). The amount of leachable cations was influenced by the added amount of APW. The Na+ in the bentonite was exchanged by the K+, Mg2+ and Ca2+ particularly in the samples with higher saturation degree (Sr = 1–0.75). Additionally, the Na+ was partly exchanged by Ca2+ from dissolved gypsum and calcite. Consequently, the concentration of Na+ in the pore water in these samples is higher as the initial concentration (7–9 vs. 2 cmol(+)/kg, Tab. 4). The same cation exchange was observed in the LOT experiment (Olsson & Karnland, 2011; Itälä & Olin, 2011).

Generally, the specific surface area of Ca/Mg-bentonites is higher than that of Na-bentonites (Kaufhold et al., 2010b). This is in agreement with our findings. The samples with the higher Ca2+/Mg2+ content showed higher values of SSA. The SSA decreased by about 50% after thermal treatment with increasing temperature over time. A similar decrease of SSA was also observed in the ABM-project and the HEATER experiment (Nagra, 2011; Plötze, et al., 2007). The variations in SSA between the Ca/Mg- and Na-bentonite were explained with differences in their microporosity from the quasi-crystalline overlap region of smectite and the accessible area of the interlayer (Kaufhold et al., 2010b), i.e. the “roughness” of the smectite platelet edges. However, with increasing treatment temperature, the SSA decreases also for samples with higher amount of leachable Na+. Therefore, additional effects on the microporosity have to be considered, e.g. dissolution and precipitation processes leading to aggregation effects. The influence of such effects was investigated by measurement of the SSA of some samples after remoulding. However, the changes in SSA after remoulding bordered the line of analytical error and did not show non-ambiguous trends (Table 5, Fig. 8b). None of the samples reached the SSA value of the pristine material. Only for samples stored at 120°C for 12 and 18 months, a tendency of SSA recovery was observed reaching the values of samples heated over shorter time.

Swelling is a typical feature of bentonite when in contact with water in gaseous or liquid form. The amount of adsorbed water depends on the type of smectite, the layer charge density, and on the type of exchangeable cations in the interlayer (Laird, 1999; Kaufhold et al., 2010a). In addition, the water adsorption under free swelling conditions (WUC) is strongly affected by the microporosity of bentonite (Kaufhold et al., 2010a). The lower the amount of leachable Na+, the lower is the WUC (Kaufhold et al., 2010a). The WUC of the thermally treated MX-80 in the presented experiment decreased with increasing temperature over time and had a similar development as the SSA. This decrease could be observed especially for samples heated at T ⩾ 105°C for more than 12 months (Table 7, Fig. 10). However, these samples have a higher content of leachable Na+ and consequently should have a higher WUC. The sole influence of the type of the interlayer cation could be thus excluded. The WUC of the thermally treated bentonite is, as the SSA, rather dependent on the bentonite aggregation and microstructure.

A link between the SSA and the WVAA has been observed previously by Kaufhold et al. (2010b). Both methods are based on gas adsorption. However, the adsorption mechanisms are rather different. In contrast to N2, which will be adsorbed on external surfaces and micropores, the polar water molecules mainly hydrate the adsorbed cations. The WVAA is therefore strongly influenced by the type of these cations. However, at the applied relative humidity of 75% the difference in the water adsorption between Ca2+ and Na+ is not significant (Bérend et al., 1995; Cases et al., 1997; Plötze & Kahr, 2003). The presented results show in WVAA two plateaus independent from time with a significant drop at T > 105°C (Fig. 9). A similar decrease of the WVAA of MX-80 (by approximately 20%) after its long-term heating in an open system at temperatures up to 200°C was also described by Valter & Plötze (2010). The clear recovery of the WVAA reaching the value of the pristine material was observed for remoulded samples (Fig. 9b). This behaviour can be explained by cementing processes during thermal treatment, which “glue” the smectite platelets. The remoulding procedure breaks up the cemented platelets. The water molecules can enter the interlayer and hydrate the interlayer cations. The WVAA of the thermally treated bentonite is rather dependent on the bentonite aggregation and microstructure than caused by the so-called Couture effect, which describes an irreversibly reduced water uptake of bentonite after steam treatment at 250°C (Couture, 1985).

Conclusions

The analysis of the Na-bentonite MX-80 treated under near-field thermal and saturation conditions in a closed system showed a high mineralogical stability over the 18-months test period. Only minor changes occurred at elevated temperature (⩾105°C): (i) The smectite content seems to decrease slightly over time regardless of treatment temperature. However, this decrease is in the range of measurement error as well as being additionally influenced by the heterogeneity of the pristine MX-80. Furthermore, the insignificant increase of the coefficient of variation in the basal spacing and of the MLC indicated more likely the formation of higher charged smectite, rather than of illite. Amorphous silica could also not be detected. (ii) The gypsum content did not change in samples stored at temperatures below 105°C. At T ⩾ 105°C gypsum dehydrated to bassanite. (iii) The calcite content showed a subtle tendency of decrease with temperature.

Whereas the mineralogical alterations are close to insignificant, considerable changes in physicochemical properties were observed: (i) The cation exchange capacity is one of the most valuable properties for assessing the radionuclide retention of bentonite. A very weak increase towards the heating source was observed in several in-situ experiments. In contrast, in our experiment, the CEC decreased non-ambiguously for the samples stored above the critical temperature of 120°C. (ii) A slight conversion from the sodium to an earth alkali form of the bentonite was observed. The Ca2+ and Mg2+ from the dissolved Ca-phases (gypsum, calcite) and from the added artificial pore water exchanged the Na+ in the smectite interlayer. The type of exchangeable cations as well as aggregation processes connected with dissolution/precipitation influenced physicochemical properties like water adsorption and swelling, specific surface area and microstructure. (iii) The SSA decreased by ∼50% after thermal treatment with increasing temperature over time. (iv) The water uptake capacity under free swelling conditions decreased with increasing temperature over time especially for samples heated at T ⩾ 105°C for more than 12 months. (v) The water vapour adsorption ability showed a significant drop at T = 120°C. The ability to recover by mechanical remoulding can be explained with cementing processes during the thermal treatment.

The observed alterations are mostly subtle. However, temperatures ⩾120°C had a remarkable negative influence on different properties of MX-80.

The present work is part of the project “Mineralogical, geochemical and geotechnical characteristics of granular bentonite after storage at near-field relevant conditions (T, w)” within the ESDRED project and co-financed by the Swiss National Cooperative for the Disposal of Radioactive Waste (NAGRA) and the European Community under the SP6 Program (contract number FI6W-CT-2004-508851).

Hanspeter Weber from NAGRA is acknowledged for the coordination of the project. We acknowledge Urs Mäder (University Bern) for the calculation of the chemical composition and delivery of the recipe of the artificial pore water. Adrian Zweidler is thanked for designing the thermal treatment containers. The manuscript was kindly improved by Laurence Noel Warr and an anonymous reviewer.