- E. Schweizerbart'sche Verlagsbuchhandlung, D-70176 Stuttgart
The cement-bentonite interface is a highly dynamic region in engineered barrier systems (EBS) designed to isolate radioactive wastes. Two very different chemical environments are held together: a near-neutral clay mineral-calcite equilibrium-dominated system (hydrated bentonite system) and a thermodynamically unstable hyperalkaline one (cement). The interface is, therefore, in permanent disequilibrium due to the conditions of changing pH that are related to the different degradation steps of the cement. The formation of phillipsite-(Na, K), (Si/Al ratio of 1.8–2.4), has been found in a series of closed-system hydrothermal tests at 35–90°C when the resulting pH was within the 11.7–12.6 range. The chemistry of the equilibrium solutions, rather than the crystallization substrata, controls the Si/Al atomic ratio and the type of zeolite that has been formed. This is based on the decrease of Si/Al at higher pHs and the predominance of phillipsite-Na. The formation of phillipsite-Na in these experiments is in agreement with the available thermodynamic data on zeolites.
The safety of a deep geologic repository for highlevel radioactive waste is based on the multibarrier concept (Bilitewski et al., 1997; Savage, 1995), and is determined by the characteristics of the design and construction of the engineering barriers as well as their stability in time. It is necessary to evaluate changes that can affect the mechanical, hydraulic and geochemical properties of the barriers. This implies determining the long-term stability of the mineral products used as barrier constituents. It is also necessary to take into account the physicochemical conditions caused by the heat generated in the disintegration of the radionuclides and by the interaction with water coming from the host rock.
Compacted bentonite is the main constituent of the engineering barrier to be considered in the deep geologic repository to be constructed in crystalline host rock. The function of the bentonite barrier is the object of research projects that simulate realscale installation operations. These experiments serve to check its in situ behaviour. PRACLAY (Verstrich et al., 1998), CERBERUS (Noynaert et al., 1997) or FEBEX (Huertas & Santiago, 1998) are demonstration experiments in which the cement-bentonite interface has been found to be a chemically reactive zone. In this region, two very different geochemical environments are held in contact. One environment, under near-neutral conditions of pH, is characterized by the dissolution equilibrium of clay minerals and calcite. The latter mineral is frequently present as an accessory mineral. The other environment consists of a thermodynamically unstable and hyperalkaline system: the minerals of the mortar of hardened cement degrade continuously under the influence of the groundwater flow.
The pH of the leachates of ordinary Portland cement (OPC) oscillates between 14 and 13 during the initial more alkaline stage, and from 12.6 to 10 as the dissolution of portlandite, Ca(OH)2, or the hydrated calcium silicate gels (CSH-gel) proceeds (Adenot & Buil, 1992; Lovera et al., 1997; Faucon et al., 1997, 1998). The duration of the hyperalkaline stage will depend on the cement mass that is leached and on the rate of the transport processes. Therefore, it will be regulated by the permeability of the cement and by the diffusion of solutes in favuor of the concentration gradients in the system. Glasser & Atkins (1994) indicate that it would be necessary to replace the pore water of OPC with pure water more than 1000 times to reduce the pH to 11. This implies that the alkaline conditions will last for long periods of time in the cement-benton-ite interface.
The formation of zeolites is one of the reactive phenomena characteristic of the decomposition of silicates present in a cement mortar (quartz, quartz-feldspar aggregates) (Lagerblad & Trägårdh, 1994). On the other hand, zeolites are frequent reaction products in alteration processes of volcanic glasses. The interaction of rocks with fluids at basic pH has taken place in different geochemical environments which influence the chemical and structural nature of the zeolites formed (Hay, 1986; Boles, 1988; Hawkins, 1981; Barth-Wirsching & Höller, 1989; Kawano et al., 1993; Querol et al, 1997; Kawano & Tomita, 1997). The Spanish reference bentonite contains significant quantities of volcanic glass (Cobeña et al., 1999) so that zeolite formation as a consequence of their interaction with the solutions coming from the cement is highly probable.
The formation conditions of the different zeolite species depend on the composition of the reaction solution, the temperature and the time of treatment (Barrer 1982; Gottardi & Galli, 1985). The composition of the mineral substrata is also an important factor in zeolite crystallization, since the main formation mechanism is heterogeneous nucleation on the mineral surface (Hawkins, 1981). The influence of this factor is fundamental in relatively diluted reactive solutions (< 0.1M). In more concentrated media (> 0.1M), the nature of crystallized zeolite is independent of the nucleation substrata.
The thermodynamic approaches establish the conditions in which minerals can be formed from an aqueous solution of known composition. Chipera & Bish (1997, 1999) have developed a thermodynamic database that includes diverse zeolites of variable composition characterizing the alteration of volcanic tuffs in the Yucca Mountain region (Nevada, USA). The validation of these data should demonstrate a correspondence between the formation conditions observed in nature and those found in laboratory experiments. This will allow the formulation of predictions about the long-term stability of geologic materials implemented as engineering barriers. Hence, one of the objectives of this work is to check the available thermodynamic data for zeolites against the thermodynamic status of the solutions resulting in our experiments.
Formation of the zeolites phillipsite and analcime has been found to be an important alkaline hydrothermal alteration process occurring in the bentonite used in this work (Ramírez, 2000; Huertas et al., 2000). In this study, the formation conditions of these minerals are analyzed as a function of the temperature and chemical composition (activity) of the reaction waters obtained in the experiments.
The study of the alteration that developed in the contact between the cement and the bentonite was carried out by means of a series of experiments in which the cement was substituted by alkaline solutions. Hermetic batch reactors of PTFE were used, to temperatures of 35,60 and 90°C, during periods of between 1 and 12 months. In them, the bentonite is mixed with several solutions in the 10 to 13.5 pH range (Table 1). The ratio KOH/NaOH = 2 in solutions B and C was formulated in agreement with the interstitial waters of cements obtained by Andersson et al., (1989), and with two OPC cements referenced in Huertas et al. (2000). The interstitial waters contain potassium and sodium hydroxides mixed with small amounts of calcium (< 1 mM at portlandite saturation), chlorides, sulfates and carbonates (< 0.5 mM). The solutions at pH ≥ 12.6 were saturated in portlandite, Ca(OH)2. The water/bentonite weight ratio was 1/3 (0.241/80 g).
Pure montmorillonite (< 2 μm size fraction of the bentonite) and basic solutions with NaOH or KOH were also reacted (0.031/10 g) (Table 1). The objects were to check both the reactivity of the smectite and the effect of adding sodium or potassium to the reaction media. After completion of the tests, the solid and liquid phases were separated by centrifugation and analyzed by several techniques. Analyses of the resulting solution of the montmorillonite tests were not performed due to the impossibility of separating enough solution volume from the dispersed clay gels.
The bentonite used comes from the location of Serrata de Níjar (Almería, Spain) and is called FEBEX bentonite. It contains mainly montmorillonite (93 ± 3%). K-feldspar (2 ± 1%), plagioclase (1 ± 0.7%), quartz (2 ± 0.5%), calcite (1 ± 0.7%) and cristobalite (2 ± 0.2%) are the main accessroy minerals as determined by X-ray diffraction (Cobeña et al., 1999; Linares et al., 1993; Caballero et al., 1983). Observation of the > 50 μm fraction of the bentonite by means of Scanning Electron Microscopy and Energy Dispersive Analysis (SEMEDX) confirmed this composition qualitatively. However, the presence of volcanic glass was detected and its aspect is shown in Fig. 1.
The chemical composition of the volcanic glasses characterized by SEM-EDX, is given in Table 2. The glass contains tabular-shaped crystals on the surface (Fig. 1a). Analysis of these inclusions gives a composition of zeolites of the heulandite/clinoptilolite series, with the following average formula (5 analyses): (Na0.21±0.2 K0.58±0.5Ca0.57±0.2) (Al1.69±0.2 Si7.28±0.3)O18·6H2O (Si/Al ratio: 4.3 ± 0.4). The scarcity of this material with regard to the other bentonite components explains the impossibility of characterizing it by XRD. The quantities of glass and of fragments of unaltered volcanic rock have been determined by optical microscopy in samples of compacted bentonite to be 1.5 ± 0.1 and 1.2 ± 0.8%, respectively (Cobeña et al., 1999).
Characterization of the products of bentonite alteration was carried out by the random-powder XRD method, using a SIEMENS D5000 diffractometer (CuKα; 2 mm and 0.6 mm of divergence and reception slits, respectively). The XRD profiles were measured in 0.04 2𝛉 goniometer steps for 3 s. On the other hand, systematic observations were done using an SEM-EDX device (PHILIPS XL30, W-source, DX4i analyzer and Si/Li detector), both in the altered bentonite and in the > 50 μm size fraction. The analyzer was previously calibrated with a multimineral sample: the USGS standard AGV-1 (Govindaraju, 1994).
The chemical composition of the zeolites was obtained by means of numerous EDX analyses of the altered samples. These analysis were performed on clean surfaces to avoid as much as possible all sources of contamination. Many analyses were not taken into account in view of anomalous results regarding structural formulae of the XRD-detected phillipsite.
The chemical composition of the resulting altered solutions was analyzed using routine techniques of water analysis: visible spectrometry of dissolved SiO2 and Al, automatic titration (Ca2+, Mg2+ and alkalinity), flame emission (Na+ and K+) and ion-selective electrode methods (SO42-, Cl−). In this work, reference is only made to the activities of the species relevant to the dissolution equilibrium of zeolites.
To study the state of equilibrium of the system, the chemical speciation of the resulting aqueous solutions has been calculated by module EQ3NR of code EQ3/6 (Wolery, 1992). The reaction boundaries of phillipsite with respect to the other zeolite phases has been calculated by taking the ΔG°f,298 and ΔH°f,298 from Chipera & Bish (1999) for phillipsites with a chemical composition in the same range as those described in this work. These data were not available from the phillipsite thermodynamic variables summarized by Helgeson et al. (1978). Extrapolations to 35–90 °C have been performed by modifying the SUPCRT92 database (Johnson et al., 1992) and by introducing the aforementioned standard phillipsite properties. Therefore, the Maier-Kelly (C°p (T)) coefficients function, included in this software package, is used to calculate the required thermodynamic reaction properties at the different temperatures.
When the initial pH of the solutions is 13.5 and sodium is present in the reaction media, zeolite minerals are detected as alteration products of bentonite (solutions C and F). The phillipsite appears as zeolite product except when the initial solution is composed exclusively of NaOH 0.5M. In this case, the phillipsite is mixed with analcime. Fig. 2 shows the XRD patterns with the characteristic reflections of phillipsite and analcime.
Phillipsite is produced in a period of 30 days only at 90 °C and is detected by SEM in close association with the fragments of volcanic glass. In the long term, this mineral forms in the whole range of temperatures and the zeolite proportion increases in the mass of clay. In Table 3, the results of its semiquantification are shown in function of the test type and time of treatment. The radial polycrystalline aggregates correspond to the most frequent form of phillipsite crystallization (Fig. 3a). However, at 35 °C and 12 months it has developed as tabular crystals interbedded with clay (Fig. 3b). Analcime was only detected by means of XRD.
In Table 4, the structural formulae of phillipsite are shown for the different tests in which they have been observed. The formulae have been calculated from EDX analyses as an average of five samples at each temperature and time of treatment. The O32.12H2O oxygen and hydration basis is the same basis as used by Chipera & Bish (1999).
The atomic ratio Si/Al of phillipsite is between 2.33 and 1.92, and is independent of the corresponding mineral substratum, volcanic glass or clay (Fig. 4a). Similarly, the molar fraction of sodium in the zeolites formed stays at a constant value of 0.8, while in the substrata it oscillates between 0.4 and 0.9 (Fig. 4b). This implies that the driving force of its formation is the composition of the aqueous solutions that have intervened in the alteration of the bentonite.
In Table 5, the values of the pH and the activities of the aqueous species that intervene in the formation of phillipsite are presented. The activity of aluminium was not considered because its concentration in solution is below the detection limit for the technique used in the determinations (< 10−6 M).
The pH decreases with temperature at equal periods of time (Fig. 5a). This variable regulates the composition of the aqueous solution. At each temperature and starting from 90 days of reaction the composition of the solutions and the pH were very stable. The resulting solution becomes more sodic as the pH decreases (Fig. 5b), coinciding with the highest rate of phillipsite formation. On the other hand, the activity of the dissolved alkaline cations decreases with the temperature, coinciding with zeolite formation (Fig. 5c).
The formation of sodium zeolite agrees with the prevalence of sodium as dissolved species. Montmorillonite favours the incorporation of potassium, regarding sodium as the exchangeable cation in the smectite. Therefore, the relationship between potassium and sodium is inverted in the resulting solutions and oscillates from 1/3 to 1/5 compared to the initial solutions (K/Na = 2). The activity of silica in these experiments showed values that indicate conditions of subsaturation in relation to quartz and the other silica polymorphs. This fact would justify the absence of important variations of silica activity in function of temperature (Fig. 5c) and the dissolution prints that are frequently observed on volcanic glass.
Phillipsite crystallization takes place in an interval of final pH between 11.7 and 12.6. This variation controls the value of the Si/Al ratio of the zeolite. As the pH of the solution increases, the Si/Al ratio of the zeolite decreases (Fig. 5d). This pH increase produces a change in the distribution of the soluble species of aluminosilicates. The more polymerized species are richer in silicium and are stable to values of lower pH. In contrast, the species that are less polymerized are stable to values of higher pH, which brings about a decrease in the Si/Al ratio. The zeolite formed should, therefore, have a Si/Al ratio that is directly inherited from these precursory polymers (Donahoe & Liou, 1985; Barth-Wirsching et al., 1989; Kawano & Tomita, 1997).
Discussion of results
In the obtained results the composition of phillipsite does not depend on the composition of the growth substrata, but instead is conditioned by the properties of the solution.
Although the concentration of hydroxide in the reacting solution determines the Si/Al ratio of the zeolite formed, the type of alkaline cation favours a determined crystallization sequence. Therefore, in sodic environments, and in function of the pH increase (10–14), the obsidian (volcanic glass) alteration produces the sequential crystallization of smectite, phillipsite and rodosite. On the other hand, both phillipsite and analcime can crystallize in strongly alkaline sodic environments (Kawano & Tomita, 1997).
In order to evaluate and to relate the stability of the zeolites under the conditions of alkaline alteration of the bentonite, equilibrium diagrams have been calculated. They represent the boundary conditions of the zeolite species that have been observed: phillipsite, analcime and heulandite. The structural formulae and the thermodynamic data that have been used are shown in Table 6. These formulae are expressed in the oxygen and hydration basis O14.5H2O used by Helgeson et al. (1978) in their database. The thermodynamic standart properties given by Chipera & Bish (1999) were recalculated to this basis. Phillipsite-Na has been chosen to simplify the problem. The representation of the relative stability fields for phillipsite-Na and phillipsite-K in an activity/activity K+ versus Na+ space showed that the experimental solution activities were located in the phillipsite-Na field rather than in the equilibrium line between both species. In fact, the Na/K ratio is higher than 3 in the structural formulae presented in Table 4. Bearing in mind these variables, the following equilibria are represented:
2 analcime + 3SiO2, aq + Ca2+ + 4H2O = heulandite + 2Na+
K = a2Na+/(a3SiO2.aCa2+) phillipsite + Ca2+ + 2SiO2, aq + H2O = heulandite + 2Na+
K = a2Na+/(a2SiO2·aCa2+)
As deduced from the equilibrium constants, the activities of the cations and of the monomeric aqueous silica are the key parameters affecting the phase relationships.
In Fig. 6, the fields of stability of these zeolites are represented as a function of the silica activities and of the quotient between the activities of sodium and calcium in solution for each temperature. The chemical composition of the resulting long-term solutions are located clearly in the stability field of phillipsite. This is in agreement with the generation of this zeolite instead of heulandite-clinoptilolite type. Heulandite-clinoptilolite would characterize the alteration of the volcanic glass during the stage of formation of bentonite itself, which takes place at near-neutral or slightly acidic pH (Caballero et al., 1983). It has also been observed how the equilibrium boundary of analcime/phillipsite moves as a function of the increase in temperature, narrowing the field of phillipsite stability. Therefore, the increased temperature would favour analcime formation instead of phillipsite when the activity of sodium was slightly superior to that which characterizes the experimental alteration environment generated in these tests. This possibly takes place when NaOH 0.5M solution is used in the alteration tests of the pure montmorillonite (solution F). Unfortunately, the complete chemical analysis of the resulting solutions of these tests was not available.
The first porewaters resulting from the degradation of cement contain sodium and potassium hydroxides. As a consequence the formation of phillipsite crystals should occur on short term. The formation of such crystalline substances instead of amorphous gels prevents, however, a negative impact in the plasticity of the clay matrix. On the other hand, the cation-exchange capacity of the zeolites is an interesting addition to the properties of the barrier system.
In the crystallization of the zeolites that took place in the hydrothermal alkaline alteration of the bentonite of La Serrata de Níjar (Almería, Spain), the chemical composition of the reacting solution controls its chemical and structural nature. Phillipsite and analcime can only form at pH > 11.6 in a sodic environment.
The thermodynamic calculations confirm that phillipsite is the more stable zeolite under these experimental conditions. Analcime formation is favoured by a temperature rise and by an increase in sodium activity. These conditions can only be met in the experiments with pure montmorillonite and extremely sodic solutions.
It is not expected that the production of zeolites will affect negatively the performance of the engineered barrier system (EBS) constructed to isolate radioactive wastes. However, the spatial extent of the reaction should be studied on a real scale by considering open-system experiments.
This work was carried out under the CE contract F14W-CT96–0032, Effects of Cement on Clay Barrier Performance
- Received 4 September 2000.
- Modified version received 28 November 2000.
- Accepted 2 January 2001.