Effects of OH– activity and temperature on the dissolution rate of compacted montmorillonite under highly alkaline conditions

Abstract The highly alkaline environment induced by cementitious materials in a deep geological disposal system of high-level radioactive waste is likely to alter montmorillonite, the main constituent of bentonite buffer materials. Over long time periods, the alteration may cause the physical and/or chemical barrier functions of the buffer materials to deteriorate. In order to evaluate the long-term alteration behaviour, the dissolution rate, RA (kgm-3 s-1), of compacted pure montmorillonite (Kunipia-F) was investigated experimentally under conditions of hydroxide ion concentration of 0.10-1.0 mol dm-3 at temperatures of 50-90°C. The dissolution rate data, including those from a previous study at 130°C, were formulated as a function of the activity of hydroxide ions, aOH- (mol dm-3), and temperature, T (K), and expressed as RA = 104.5·(aOH-)1.3·e-55000/RT by multiple regression analysis, where R is the gas constant. The dissolution rate of montmorillonite was greater in the compacted montmorillonite than in the compacted sand-bentonite mixtures. The difference can be explained by considering the decrease in aOH- in the mixtures accompanied by dissolution of accessory minerals such as quartz and chalcedony. The dissolution rate model developed for pure montmorillonite is expected to be applied to bentonite mixtures if quantification of the decrease in aOH- is achieved somehow.

A B S T R AC T : The highly alkaline environment induced by cementitious materials in a deep geological disposal system of high-level radioactive waste is likely to alter montmorillonite, the main constituent of bentonite buffer materials. Over long time periods, the alteration may cause the physical and/or chemical barrier functions of the buffer materials to deteriorate. In order to evaluate the long-term alteration behaviour, the dissolution rate, R A (kg m -3 s -1 ), of compacted pure montmorillonite (Kunipia-F) was investigated experimentally under conditions of hydroxide ion concentration of 0.10-1.0 mol dm -3 at temperatures of 50-90°C. The dissolution rate data, including those from a previous study at 130°C, were formulated as a function of the activity of hydroxide ions, a OH -(mol dm -3 ), and temperature, T (K), and expressed as R A = 10 4.5 ·(a OH -) 1.3 ·e -55000/RT by multiple regression analysis, where R is the gas constant. The dissolution rate of montmorillonite was greater in the compacted montmorillonite than in the compacted sand-bentonite mixtures. The difference can be explained by considering the decrease in a OHin the mixtures accompanied by dissolution of accessory minerals such as quartz and chalcedony. The dissolution rate model developed for pure montmorillonite is expected to be applied to bentonite mixtures if quantification of the decrease in a OHis achieved somehow.
KEYWORDS: compacted montmorillonite, highly alkaline conditions, dissolution rate, activity of hydroxide ions, temperature, accessory minerals.
Highly alkaline environments induced by cementitious materials in a deep geological disposal system for highlevel radioactive waste are likely to alter montmorillonite, the main constituent of bentonite buffer materials, affecting adversely the physical and/or chemical barrier functions of the buffer materials. In order to evaluate the long-term alteration behaviour, the dissolution rate of montmorillonite should be determined.
Various studies on the dissolution rates of clays in alkaline solutions used different pulverized smectites (Savage et al., 1992;Bauer & Berger, 1998;Kubo et al., 1998;Sato et al., 2004;Rozalén et al., 2008Rozalén et al., , 2009, and determined the rates under much lower solid/solution ratios than those of the disposal sites (Table 1). These data need some modification in order to evaluate the dissolution of buffer materials under the confined disposal conditions. Savage & Liu (2015) reported that the most relevant water/clay ratio ('porosity') should be used in modelling and experimental systems based on thermodynamic calculations illustrating that the water/clay ratio affected the magnitude and nature of The flow experiments were also performed in this study for which the solid/solution ratio was not identified. 2 The value was calculated by dividing the initial amount of solid by the total volume of through-flowing fluid. 3 The volume of solution is that of the pores in compacted specimens.
alteration. The dissolution rates in compacted sandbentonite mixtures were determined experimentally and formulated under hydroxide ion activities, a OH -, of 0.04-0.57 mol dm -3 and temperatures of 50-170°C Yamaguchi et al., 2007). The dissolution rates in the mixtures, however, were smaller than that in pulverized montmorillonite and depended more heavily on a OH -. Yamaguchi et al. (2007) suggested that different a OHin pore water from that in the external solution might be responsible for this dependence, although this was not verified. Sawaguchi et al. (2013) studied the dissolution rates of compacted pure montmorillonite at 130°C and showed greater dissolution rates than those in the mixtures. The explanation given for the difference was that the dissolution of montmorillonite in the mixtures was inhibited by decreasing the a OHstemming from the dissolution of accessory minerals such as SiO 2 polymorphs (quartz, chalcedony, etc.) in the mixtures.
In the present study, the dependence of the dissolution rate of compacted montmorillonite (Kunipia-F) on OHactivity and temperature were investigated experimentally to clarify the effect of accessory minerals on the dissolution rate under compaction.
FIG. 1. Schematic drawing of the experimental configuration of the saturation process before the alteration of compacted montmorillonite.
the experimental conditions were the same as in a previous study by the present authors , except for the temperature conditions. The montmorillonite (Kunipia-F, a commercial product of Kunimine Industries Co. Ltd., Japan) was compacted in specimen holders to obtain a dry density of 1060 kg m -3 or porosity of 0.6. The porosity of 0.6 was selected by considering porosity variations of 0.4-0.8 through the experiment for the mixtures . The Si-, Al-and Ca-adjusted NaCl-NaOH mixed solutions used in the previous experiment  were also used in the present study ( Table 2). The alkaline solutions were prepared by adding NaCl, NaOH, sodium metasilicate  Table 2. Potassium, although usually present in cementitious water, was not added to the experimental solutions because the dissolution rate of montmorillonite in KOH-NaOH mixed solutions where Kfeldspar was formed was identical to that in an NaOH solution (Yamaguchi et al., 2007). The specimen holder was soaked in the alkaline solution in a Teflon ® container assembly at room temperature under vacuum for 1 week to saturate the specimen (Fig. 1). Then, the assembly was contacted with 0.065 dm 3 of the alkaline solution in a sealed stainless steel container and kept at 50, 70 and 90°C in an oven to observe montmorillonite dissolution over a reasonable time period. The alkaline solution was replaced by fresh solution when necessary to minimize changes in the concentration of OH -. The necessity was decided by monitoring the pH of mockup assemblies each week. When the measured value of pH decreased by >0.1 pH units from the initial value, the alkaline solution in all assemblies of the same condition was exchanged for the initial solution.
Therefore the variations of pH in the assemblies were kept within ±0.2 pH units, during the alteration period. The pH was measured at ∼25°C using a ROSS ® combination-glass pH electrode (Orion 8172BNWP, Thermo Scientific, Inc., USA) calibrated with 7.00 and 10.01 pH buffers and 1 mol dm -3 NaOH solutions. Each of the assemblies was taken from the oven at the fixed sampling time. The compacted montmorillonite specimen was pushed out from the holder by 2 mm at a time using a screw presser and was sliced with a steel blade. Each sliced piece was weighed, freeze-dried at -110°C, ground using a mortar and pestle, redried at 90°C and then analyzed for the amount of remaining montmorillonite by the conventional methylene-blue adsorption titration method (Hang & Brindley, 1970;Brindley & Thompson, 1970;Japan Bentonite Manufacturers' Association, 1991). In the present study, 2.5 × 10 −4 kg of crushed sample and of pure montmorillonite (Kunipia-F) for calibration were used. Furthermore, the remaining reacted solid was analyzed by powder X-ray diffraction (XRD, Geigerflex, Rigaku Corp., Japan). The concentrations of Al and Ca in the contacted aqueous solutions were determined by inductively-coupled plasma atomic emission spectroscopy (ICP-AES, SII SPS 7800, SII Nanotechnology Inc., Japan) after filtering the solutions through a 0.45 μm PTFE (polytetrafluoroethylene) filter and acidifying into 0.5 N HNO 3 . The concentration of Si was determined as SiO 2 by means of the molybdosilicate method (Rand et al., 1976; Japanese Standard  Association, 2004; The Japan Society for Analytical Chemistry Hokkaido, 1994) and by UV spectrometry (U-3210, HITACHI, Ltd., Japan) after the filtration.

R E S U LT S A N D D I S C U S S I O N
The weight of montmorillonite in each sliced piece (W mont , kg) was determined using equation 1: where W dry is the dry weight of each slice (kg), V titrb (alter) is the total volume of methylene-blue solution added to titrate the altered sample (dm 3 ), V titrb ( pure) is the total volume of methylene-blue solution added to titrate the pure montmorillonite (dm 3 ). It is known empirically that slight differences in experimental procedure have a large influence on a result in the methylene-blue adsorption titration method (Miyoshi et al., 2015). In the present study, efforts were made to minimize errors: (1) by having a single analyst perform a series of experiments; (2) by following a standard testing method; and (3) by using a micropipette to drop the fixed amount of suspension on the filter paper. The error in the results of the titration was estimated to be ±7.4 × 10 -6 kg at 1σ, which brings an error of ±9.9 × 10 -5 kg at 1σ into W mont . The montmorillonite content of each of the slices decreased with alteration time (Fig. 2). The XRD patterns of each slice also showed that the decrease in montmorillonite and the formation of secondary analcime (NaAlSi 2 O 6 ), proceeded over time (Fig. 3). The alteration of primary minerals proceeded both at the centre and the edges of the specimens facing the alkaline solution so that uniform dissolution of montmorillonite in the compacted specimens was considered hereafter. The diffusive mass transfer of OHwould be much greater than the rate of consumption in the compacted montmorillonite. The above-mentioned dissolution and formation of minerals is qualitatively consistent with the concentrations of Si and Al in the contacting solutions (Fig. 4). The concentrations of Si tend to be much greater than the initial values in all alkaline solutions, due to the dissolution of montmorillonite. The formation of analcime in the solutions rich in Si and Na concentrations probably consumed Al, which decreased the Al concentration. Although determination of the concentration of Ca was difficult due to interference with the high-concentration of Na, an increase in the concentrations of Ca was observed, probably due to the dissolution of montmorillonite, which contains ∼0.36 wt.% of exchangeable Ca (Ito et al., 1993).
The amount of montmorillonite in the compacted montmorillonite decreased linearly with time over several months under the conditions of OHconcentration and temperature used (Fig. 5). The greater concentration of OHin the external solution and/or temperature resulted in the faster, linear decrease in montmorillonite content in accordance with previous studies Yamaguchi et al., 2007;Sawaguchi et al., 2013). The montmorillonite density can be expressed as: where ρ(t) is the amount of montmorillonite in the altered compacted specimen (kg m -3 ), ρ(0) is the initial density (kg m -3 ), R A is the dissolution rate of compacted montmorillonite (kg m -3 s -1 ) and t is the time after the contact with the alkaline solution (s). The calculated dissolution rates by equation 2 are listed in Table 3   were determined by multiplying [OH -] by the activity coefficients evaluated using the SIT (specific ion interaction) model (Lemire et al., 2001). The dissolution rates of compacted montmorillonite, including previously obtained data at 130°C, were formulated as a function of a OH -(mol dm -3 ) and temperature, T (K), by multiple regression analysis: R A ¼ 10 (4:5+2:5) Á (a OH À ) 1:3+0:4 Á e (À55000+16000)=RT (3) where R is the gas constant (8.314 J mol -1 K -1 ). The errors indicate the confidence limit of 68% (1σ). Equation 3 was compared with the formula of the dissolution rate of montmorillonite in the mixtures (Yamaguchi et al., 2007): R A = 3500· (a OH -) 1.4 ·e -51000/RT (Fig. 6). The dissolution rate is greater in the compacted montmorillonite than in the mixtures. In a previous study performed at 130°C , the possibility that the dissolution of montmorillonite in the mixtures was inhibited by the increase of Si concentration in pore water and/or by the effect of the difference between bentonite specimens used in each experiment was ruled out. Therefore the difference in dissolution rates would be caused by the negative shift of, or decrease in, a OHin the mixtures accompanied by dissolution of accessory minerals such as SiO 2 polymorphs. Assuming that the shift would be linear by 0.05, 0.1 or 0.2 mol dm -3 or logarithmic by 0.3, 0.4 or 0.5 units, the dissolution rates of montmorillonite in the mixtures were estimated (Fig. 7). The curves given by shifting equation 3 by 0.05 mol dm -3 and by 0.4 logarithmic units reproduce the dissolution rate data in the mixtures well. Furthermore, trial calculations by considering the shift of a OHwere performed to reproduce the montmorillonite dissolution experiment of Yamaguchi et al. (2013), who compacted sandbentonite mixture specimens immersed in 1 mol dm -3 NaOH solution at 90°C for 14, 27, 51 and 67 days. Figure 8 shows the montmorillonite densities after each alteration period and the calculated values obtained using the dissolution rates from equation 3 and those after shifting. The calculated values obtained by shifting equation 3 by 0.3 logarithmic units reproduced the experimental data better than by 0.05 mol dm -3 . These results indicate that the dissolution rate in the mixtures can be explained by equation 3 considering a shift of the a OHin the mixtures, and that the shift may be logarithmic. The shift in logarithmic scale is more consistent than the shift in linear scale with the dependence on compacted montmorillonite, 1.3 ± 0.4, being closer to that for the mixtures, 1.4 ± 0.3. The shift in linear scale is not favoured because it gives an extremely low dissolution rate when applied to lower pH ranges.
The activation energy obtained (55 ± 16 kJ/mol) is consistent with that reported for the mixtures (51 kJ/mol, Yamaguchi et al., 2007) and with that determined for smectite dissolution in KOH solution (52 ± 4 kJ/mol, Bauer & Berger, 1998). Yamaguchi et al. (2007) compared the formula of dissolution rate of montmorillonite in the mixtures with that of pure pulverized montmorillonite determined by flow-through experiments  and pointed out that the difference in the dependence on a OHmade a large difference in the dissolution rate when the dependence was extrapolated to repository conditions. The formula obtained in the present study also gives a much lower dissolution rate than Sato et al. (2004) for a OH -, < 0.01 mol dm -3 . However it gives greater dissolution rate than Yamaguchi et al. (2007) by ∼0.3 orders of magnitude.
In the present study, the dissolution rate of montmorillonite free of accessory minerals, under compacted conditions was formulated. In order to FIG. 8. The results of dissolution experiments for the compacted sand-bentonite mixtures   provide reliable values for the long-term prediction of bentonite buffer performance under disposal conditions by using the dissolution rate equation, further study is necessary to quantify the decrease in a OHaccompanied by dissolution of accessory minerals such as SiO 2 polymorphs.
C O N C L U S I O N S The dissolution rate of the compacted montmorillonite (Kunipia-F) was formulated as a function of a OHand temperature. The existence of accessory minerals decreased the dissolution rate of montmorillonite. The dissolution rate model developed can be used not only for pure montmorillonite but also for bentonite mixtures by taking into account the negative shift of a OHin the mixtures accompanied by dissolution of accessory SiO 2 polymorphs.