The combined effect of ph and temperature on smectite dissolution rate under acidic conditions

Transcription

1 The effect of ph and temperature on smectite dissolution -1-11/10/04 The combined effect of ph and temperature on smectite dissolution rate under acidic conditions Keren Amram and Jiwchar Ganor* Department of Geological and Environmental Sciences, Ben-Gurion University of the Negev, P. O. Box 653, Beer-Sheva 84105, Israel. Phone Fax WEB Published in Geochimica et Cosmochimica Acta (2005) 69, Submitted to: Geochimica et Cosmochimica Acta, April, 2004 Revised and resubmitted, October, 2004 * Corresponding author.

2 The effect of ph and temperature on smectite dissolution -2-11/10/04 ABSTRACT The main goal of this paper is to propose a new rate law describing the combined effect of ph (1 to 4.5) and temperature (25 C to 70 C) on smectite dissolution rate, under far from equilibrium conditions, as a step towards establishing the full rate law of smectite dissolution under acidic conditions. Dissolution experiments were carried out using non-stirred flowthrough reactors fully immersed in a thermostatic water bath held at a constant temperature of 25.0 C, 50.0 C or 70.0 C ± 0.1 C. Smectite dissolution rates were obtained based on the release of silicon and aluminum at steady state. The results show good agreement between these two estimates of smectite dissolution rate. Low Al/Si ratios were obtained in experiments that were conducted at ph 4. These low Al/Si ratios are explained by precipitation of gibbsite and/or diaspore. Dissolution rate increases with temperature and decreases with increasing ph. Dissolution rates of experiments in which G r -21kcal mol 1, are not affected by deviation from equilibrium. Dissolution rates in most experiments are not affected by the addition of up to 0.3 M NaNO 3 to the input solution. A simple model is used to describe the combined effect of ph and temperature on smectite dissolution rate. According to this model, dissolution rate is linearly proportional to the concentration of adsorbed protons on the mineral surface, and proton adsorption is described using a Langmuir adsorption isotherm. All experimental results at ph<4 were fitted to the model using a multiple non-linear regression. The resulting rate law is: e10700/ RT a Rate = 220 e 17460/ RT e10700/ RT a where R is the gas constant, T is the temperature (K) and a H + is the activity of protons in solution. According to the model, the dependence of dissolution rate on temperature is affected by the activation energy and the adsorption enthalpy. The activation energy obtained from the fitting (17±2 kcal/mol) is within error equal to the average value of 15 kcal/mol of apparent activation energies for silicates dissolution rate (Lasaga et al., 1994). The obtained net H + H +

3 The effect of ph and temperature on smectite dissolution -3-11/10/04 enthalpy of adsorption (-11±2 kcal/mol) is within the range of 7.9 to 23.1 kcal/mol, experimentally obtained for oxides (Sverjensky and Sahai, 1998). 1 INTRODUCTION Smectites are phyllosilicates with two tetrahedrally coordinated layers and one octahedrally coordinated layer. Montmorillonites are the most abundant smectites in nature. The structural formula of an ideal montmorillonite is: M 0.66 [Mg 0.66 Al 3.34 ][Si 8 ]O 20 (OH) 4 (H 2 O) n. M 0.66 represents 0.66 monovalent cations or 0.33 divalent cations. In natural montmorillonites, the tetrahedrally coordinated Si is substituted by up to 0.4 Al atoms, and octahedrally coordinated Al is considerably substituted by Mg, Fe III or Fe II. The structural charge resulting from the tetrahedral and octahedral substitution is compensated by the presence of exchangeable interlayer cations, denoted by M in the structural formula. Smectites are generated in soils and sedimentary deposits by weathering, diagenesis or hydrothermal processes, which can either involve degradation and transformation of precursor phyllosilicates or precipitation from solution. Smectite is the most abundant mineral in soils of temperate climates that are derived from basic igneous rocks, and from the clay fraction of soils of arid climates derived from granitic pediment (Boettinger and Southard, 1995). Smectite, particularly montmorillonitic smectite, is the principal constituent of bentonite clay deposits. These have been formed by the alteration of eruptive igneous rocks of basic to intermediate composition. Smectite-rich argillites and bentonites have been recognized as suitable clays to be used as a sealant material in the multibarrier systems designed for storage of high level nuclear waste in burial repositories (Chapman and McKinley, 1989). Due to its osmotic swelling capacity (and consequently its plasticity and impermeability), smectite impedes groundwater interaction with the metal canisters. The cation exchange reactions immobilize undesirable cations from the radioactive waste, and retard its leakage from the canisters towards the local groundwater. However, the durability of the smectite itself under confinement conditions is a key parameter that must also be taken into consideration. For a repository in an argillaceous

4 The effect of ph and temperature on smectite dissolution -4-11/10/04 formation, acidification may result from oxidation of pyrite, a common mineral phase in claystone. 1.1 Previous studies on smectite dissolution rate and the motivation for the present study Only a few studies of smectite dissolution kinetics have been carried out in the last decade (Furrer et al., 1993; Zysset and Schindler, 1996; Bauer and Berger, 1998; Cama et al., 2000; Huertas et al., 2001; Metz, 2001; Metz et al., 2004a). All the experiment at 25 C and the experiments of Metz (2001) and Metz et al. (2004a) at 50 C were conducted under acidic conditions, whereas those at 20, 35, 40 and 60 to 80 C were conducted under basic (mostly extremely basic) conditions. Furrer et al. (1993) and Zysset and Schindler (1996) conducted both batch and flow-through dissolution experiments to study the proton-promoted dissolution kinetics of K-montmorillonite. The dissolution experiments were conducted at 25 C in HCl / KCl solutions (ph = 1 to 5 and [KCl] = 0.03 to 1.0 M). The dissolution rate increased with KCl concentrations and decreased with ph. At ph 3 the dissolution rate was inhibited by aluminum (Furrer et al., 1993; Zysset and Schindler, 1996). Their observed released rate ratio, R Si /R Al, depends on both the ph and KCl concentration. Bauer and Berger (1998) conducted batch experiments at 35 and 80ºC in concentrated KOH solutions (0.1 to 4 M) to study the dissolution kinetics of industrial (Ibeco and Ceca) montmorillonite. They found that under very basic conditions (11.5 ph 13.9) smectite dissolved independently of aqueous silica or aluminum concentrations. They proposed a non-linear dependency of smectite dissolution rate, Rate = k 0. 15± 0.06 a OH -. The apparent activation energy was found to be 13 ± 1 kcal/mole. Huertas et al. (2001) measured bentonite dissolution in granitic solutions (ph 7.6 to 8.5) in a semi-batch reactor at 20, 40 and 60 C. Sato et al. (2002; 2003) conducted flow-through experiments on Na-montmorillonite dissolution at 30, 50 and 70 C at ph 8.6 to 13.3 in NaOH-NaCl and KOH-KCl solutions. The effect of deviation from equilibrium on the dissolution rate of smectite was studied by Cama et al. (2000) at ph 8.8 and 80 C, and by Metz (2001) at ph 3 and 50 C. Both studies show that smectite dissolution rate is independent of deviation from equilibrium in the range of G r <-30 kcal mol 1. In contrast to the observations of Furrer et al. (1993) and Zysset and Schindler (1996), Metz

5 The effect of ph and temperature on smectite dissolution -5-11/10/04 (2001) found that far from equilibrium smectite dissolution rate is independent of both Al and Si concentrations. Besides dissolution kinetic studies on smectite powders employing batch and flowthrough reactors, a few papers about in-situ atomic force microscope (AFM) studies on dissolution rates of single-cell smectite crystals have been recently published (Bickmore et al., 1998; Bickmore et al., 1999; Bosbach et al., 2000; Bickmore et al., 2001). Smectite dissolution rates determined in AFM studies appear to be faster than the rates determined in flow-through or batch reactor studies at the same temperature and ph. As part of our attempt to establish the full rate law of smectite dissolution under acidic conditions, the present study introduces a new data set examining the effect of both temperature (25 C to 70 C) and ph (1 to 4.5) on smectite dissolution rate. Cama et al. (2002) demonstrated that in order to study the effect of temperature on kaolinite dissolution rate one should distinguish between two effects: 1) the effect of temperature on the rate coefficient, which may be modeled using the Arrhenius equation; and 2) the effect of temperature on the adsorption of protons on the surface. The last effect cannot be studied independently of the ph effect on the rate. Therefore, they presented a model describing the combined effect of ph and temperature on kaolinite dissolution rate. The main objective of the present paper is to model the experimental data using a rate law describing the combined effect of ph and temperature, following the approach of Cama et al. (2002). 2 MATERIALS AND METHODS 2.1 Characterization and pretreatment of smectite The smectite sample used in this study is the SAz-1, an international reference sample of the Clay Mineral Society Source Clay Repository. Details about the origin of these standard clay samples were published by Van Olphen and Fripiat (1979) and Moll (2001). The sample is not a pure smectite, containing considerable amounts of silica phases (quartz, cristobalite and/or amorphous silica) and minor quantities of alkali-feldspars, plagioclase and carbonates. The structural formula of pure SAz-1 was determined by Metz et al. (2001; 2004a) to be K 0.02 Na 0.05 Ca 0.41 Mg 0.18 [Mg 1.11 Fe 0.17 Al 2.77 ][Al 0.30 Si 7.70 ]O 20 (OH) 4. Using this structural

6 The effect of ph and temperature on smectite dissolution -6-11/10/04 formula and the whole rock analysis, Metz et al. (2004a) calculated that the raw smectite contains about 87% smectite, 8% excess SiO 2 (probably mostly in amorphous silica). The rest is mostly hydrated water. The smectite was pretreated in N HNO 3 at 70 C for a few months, using the procedure described in Ganor et al. (1995). The sample is composed of smectite aggregates ranging in size from less than a micron to more than 100 micron (Metz et al., 2004a). BET surface areas of raw SAz-1 sample determined in previous studies vary between 34 and 97 m 2 g -1 (Metz et al., 2004b). Following dissolution experiments the BET surface area of SAz-1 increased to 127±13 m 2 g -1. Using atomic force microscope, Metz et al. (2004b) determined the edge surface area of SAz-1 to be 4.9 ± 0.7 m 2 g Experimental setting Dissolution experiments were carried out using non-stirred flow-through reactors (ca. 35 ml in volume) fully immersed in a thermostatic water-bath held at a constant temperature of 25.0 C, 50.0 C or 70.0 C ± 0.1 C (Fig. 1). The reaction cells were composed of two chambers, a lower chamber of 33-mm inner diameter and an upper chamber of 26-mm inner diameter. The two chambers were separated by a fine (5 µm) nylon mesh, on which smectite powder was placed. Some more details of the experimental procedure can be found in Metz and Ganor (2001). 2.3 Solutions and Analyses Input solutions were prepared at specific ph by diluting 1M HNO 3 with double deionized water. In experiments designed to study the effect of ionic strength, different amounts of NaNO 3 have been added into the input solution. Input and output solutions were analyzed for Al, Si, and ph. Total Al and Si were analyzed colorimetrically with a UV-visible spectrophotometer, using the Catechol violet method (Dougan and Wilson, 1974) and Molybdate blue method (Koroleff, 1976), respectively. The uncertainty in measured Al and Si was better than ± 5% for concentrations above 4 µm. The precision dropped to ±15% and 33% for measurements at low concentrations of 2 and 0.5 µm, respectively. The ph was measured at experimental

7 The effect of ph and temperature on smectite dissolution -7-11/10/04 temperature on an unstirred aliquot of solution using a semi-micro Orion Ross combination electrode. The reported accuracy is ±0.02 ph units (±4.5% in H + activities). 3 CALCULATIONS The overall dissolution reaction of smectite sample SAz-1 under acidic conditions can be expressed as: (1) ( K K Na Ca Na Mg )( Mg Ca Fe 0.17 Al Mg )( Al Si Fe ) O ( OH ) Al H H SiO + 6.8H O The dissolution rate, Rate, (mol g -1 s -1 ) in steady state was based on the release of Al and Si according to the expression: (2) q ν j Rate = ( C j, out C j, inp ) m where C j,inp and C j,out are the concentrations of component j (Al or Si) in the input and the output solutions, respectively (mol m -3 ), ν j is the stoichiometry coefficient of j in the dissolution reaction, t is time (s), m is the sample mass (g) and q is the fluid volume flux through the system (m 3 s -1 ). Note that in our formalism, the rate is defined to be negative for dissolution and positive for precipitation. The common practice in experimental kinetics is to normalize the dissolution rate to the total surface area of the pure mineral, which is measured by the Brunauer-Emmett-Teller (BET) method, (Brunauer et al., 1938). In contrast to this common practice, Furrer et al. (1993), Schlabach et al. (1999) and Zysset and Schindler (1996) normalized their smectite dissolution rate data to the sample mass. They argued that normalization to surface area is not appropriate, as long as it is not possible to measure exactly the extent of the edge surface area of a smectite powder. Indeed, Metz et al. (2004b) showed that there is no correlation between the total and the edge surface area of smectite, and as a result the BET surface area cannot serve as a proxy for the reactive surface area of smectite. Therefore, we normalized the dissolution rates by the mass of the smectite. The initial mass of the smectite in each experiment was calculated from the product of the starting mass and the estimated percentage of the smectite in SAz-1 (87%, Metz et al., 2004a). Following each stage (i.e., the time

8 The effect of ph and temperature on smectite dissolution -8-11/10/04 between the replacements of two sequential output bottles), the remaining mass of each mineral was updated based on the release rate of Al and the duration of the stage. The release rate of Al and not of Si was used in the updating procedure following the conclusion of Metz et al. (2004a) that the initial fast release of Si, before steady state, mainly reflects the hydrolysis of a fast dissolving silica phase, while the initially slow release of Al reflects the dissolution of the smectite itself. The error in the calculated rate is estimated using the Gaussian error propagation method (Barrante, 1974) from the equation: (3) 2 P P = i xi 1/ 2 i ( x 2 ) where P is the uncertainty of the calculated parameter and x i is the estimated uncertainty of the measurements of the quantity x i. The degree of saturation of the solution with respect to smectite dissolution is calculated in terms of the Gibbs free energy of reaction G r (4) IAP G r = RT ln( ) K eq where R is the gas constant, T is the absolute temperature, IAP is the ion activity product of the solution, and K eq is the solubility constant. The activity coefficients and the activities of the different species in solution were calculated using the EQ3NR module of the EQ3/6 software package (Wolery, 1992). Errors ( P) in the above-calculated parameters (P), i.e., IAP and G r, were estimated according to the Gaussian error propagation equation (3). In the absence of a specific K eq value for smectite sample SAz-1, the solubility constant of the EQ3/6 thermodynamic data-set data0.com.r22a (Wolery, 1992) for Ca-endmember smectite (Ca 0.33 [Mg 0.66 Al 3.34 ][Si 8 ]O 20 (OH) 4 ) were used in the calculations. This proxy for the solubility constant was selected so the calculation would be consistent with those of Metz (2001). 4 RESULTS The variations of output Al and Si concentrations in three representative flow-through experiments as a function of time are shown in Fig. 2. Figures of all dissolution experiments

9 The effect of ph and temperature on smectite dissolution -9-11/10/04 are presented in Amram (2002). Each of the experiments was composed of 1 to 5 stages, where new stages were initiated by a change in the flow rate (e.g., , Fig. 2b) and/or in the composition of the input solution (e.g., , Fig. 2c). The vertical lines in Fig. 2 delineate the different stages. Much of the noise in the non-steady state data results from instabilities in flow rate. In several experiments, an established steady state was disturbed by a long period of instabilities in flow, and thereafter a new steady state was established. In such cases the new steady state was reported as a different stage (e.g., , Fig. 2b). The experimental conditions of all the stages are compiled in Table 1. The first two digits in the names of the experiments denote the experimental temperature. The last digit in the name (following a dot) marked the consecutive stage number of the experiment. Al and Si analyses used to calculate steady-state compositions are denoted by open symbols in Fig. 2. Si concentrations were usually higher at the onset of the experiments (Fig. 2), after which Si concentrations decreased until steady state was attained. A reversed trend was observed for the release of Al, which increased usually from very low starting concentrations to higher concentrations at steady state. The ph decreased continuously until steady state, where ph in output solutions was close to the ph in the input solution (generally up to 0.2 ph units higher). Duration of experiments varies but mostly surpasses 1500 hours, and some last for more than h. The time required to achieve the first steady state varied considerably. It was usually shorter under conditions in which the dissolution rate was fast (high temperature and low ph) and under higher ionic strength. In most of the experiments, steady state was easily maintained for several hundred hours (up to 2500 h), as long as the flow rate was stable (e.g., Figs. 2a and b). The exceptions are experiments conducted under conditions in which smectite dissolved very fast. In these experiments the mass of smectite decreased significantly with time, and as a result the output Al and Si concentrations decreased as well (e.g., , Fig. 2c). Several studies (e.g., Walther, 1996; Gautier et al., 2001) indicate that the amount of time prior to steady state may influence the resulting steady-state dissolution rate. Figure 2b shows the change in Al and Si concentrations in a multi-stage experiment (50-03) at 50 C in which the flow rate varied between and 0.04 ml/min. The experiment attained the first steady state after about 6500 h. The dissolution rate at

10 The effect of ph and temperature on smectite dissolution /10/04 steady state was 2.8±0.5x10-12 mol g -1 s -1. After about 2000 h at steady state the flow rate increased to 0.04 and a change in concentration was observed. The dissolution rate at this steady state was the same, 2.8±0.5x10-12 mol g -1 s -1. Following a period of instability in flow rate a third steady state was obtained in which the rate was 2.6±0.5x10-12 mol g -1 s -1. Regardless of the changes with time the same dissolution rate was observed in the three stages. Release of elements was highly incongruent during the first few hundreds to few thousands hours of the experiments. Metz et al. (2004a) showed that the enhanced release of Si reflects the hydrolysis of a silica phase which dissolves faster than smectite, while the initially slow release of Al reflects the dissolution of smectite itself. As we thoroughly discussed this initial non-congruent stage of smectite dissolution in Metz et al. (2004a), the present study is discussing only the steady state dissolution rate. Smectite dissolution rates at steady state (eq. (2)) were obtained based on the release of silicon (Rate Si ), and aluminum (Rate Al ) at steady state for each stage (Table 1). Figure 3 plots the dissolution rates evaluated based on the release of Si versus those obtained based on the release of Al. The solid lines in Fig. 3 are the 1/1 diagonals. Taking into account the appropriate errors, a good agreement between the two estimates of smectite dissolution rate is observed in most of the experiments. In these experiments, the steady-state dissolution rates are measured at conditions of under-saturation with respect to gibbsite, diaspore, kaolinite and boehmite. Some experiments conducted at ph 4 in which equilibrium with respect to these minerals is achieved and, consequently, incongruent dissolution occurs. In these experiments the dissolution rate is based only on Si release. In general, pretreated smectite samples were used in the flow-through experiments. Raw samples of smectite SAz-1 underwent nine flow-through experiments (Table 1). The same kinetic behavior was observed in experiments with pretreated and raw SAz-1.

11 The effect of ph and temperature on smectite dissolution /10/04 5 DISCUSSION 5.1 Are the calculated steady-state dissolution rates influenced by the presence of accessory phases in the smectite sample? As the dissolved sample contains about 8 wt.% excess SiO 2 it is important to verify that the dissolution of the Si-rich accessory phases does not affect the release of Si at steady-state. Metz et al. (2004a) discussed the general case in which a mixture of a major phase (smectite in the case of SAz-1) and a minor phase (Si-rich phase) is dissolved. Following their argumentations, there are three possible scenarios: 1) the half life of the smectite is significantly shorter than that of the Si-rich phase (i.e., the smectite dissolves faster); 2) the two phases have similar half lives; and 3) the half life of the smectite is significantly longer than that of the Si-rich phase. In the first case, the Si-rich phase is both less abundance and less reactive than the smectite and therefore the contribution of the Si-rich phase is expected to be negligible. In the last case, the Si-rich phase would be extinct before steady state, due to its shorter half life (e.g., Fig. 6 of Metz et al. (2004a)). Only in the second case, in which the Si-rich phase has a half life similar to that of the smectite, the steady state would reflect the dissolution rate of the two phases. The following observations indicate that this is not the case: 1) Metz et al. (2004a) showed that the initial fast release of Si during the dissolution of SAz-1 (Fig. 2) reflects the hydrolysis of a silica phase which dissolves faster than smectite. This fast dissolving Si-rich phase is accounted for more than 60% of the excess Si in SAz-1 (Metz et al., 2004a). 2) Metz et al. (2004a) measured the composition of raw SAz-1 using XPS. Since this measurement has an information depth of 6 nm, the measured composition is hardly obscured by contaminants, and is therefore a good approximation for the composition of pure SAz-1. In the present study, the average Al/Si ratio at steady state is within error equal to the Al/Si ratio of the XPS measurement of the raw sample, indicating that the steady-state ratio is not significantly influenced by the dissolution of phases with significantly different Al/Si ratio. 3) Metz et al. (2004a) recovered powder from their flow-through experiments with SAz-1 and found that the recovered powder contained significantly less excess Si. The average steady state Al/Si ratio in the present study is within error equal to the Al/Si ratio of in the recovered

12 The effect of ph and temperature on smectite dissolution /10/04 powder, which was analyzed by both SEM-EDS and XPS. Therefore, we conclude that the release of Al and Si at steady-state is due to the dissolution of the smectite itself, and can be used to calculate the dissolution rate of the smectite. 5.2 Separating the direct effects of ph and temperature from effects of other environmental variables In order to model the effects of ph and temperature on smectite dissolution rate it is important to separate between direct and indirect effects of the environmental variables involved. By direct effect we mean an effect related to surface processes and therefore, one that can be used to understand the reaction mechanism. In addition to temperature and ph, other environmental variables such as output Al and Si concentrations, ionic strength, and the degree of saturation vary between the experiments. Therefore, their possible effects on the smectite dissolution rate are examined below The effect of degree of saturation The Gibbs free energy of reaction ( G r ) of the smectite dissolution reaction is a strong function of ph and of Al and Si concentrations. If the dissolution rate varied due to changes in G r in the different experiments, as the ph and the output concentrations varied, then the calculated effect of ph would include a spurious contribution. This problem would be minimal only in the "far-from-equilibrium" dissolution plateau region, which is defined as the region in rate vs. G space where there is no direct effect of the degree of saturation on dissolution rate. Metz (2001) studied the effect of deviation from equilibrium on smectite dissolution rate under acidic conditions. According to his study, near equilibrium (0 G r -20 kcal/mole) the rates increase gradually with increasing undersaturation. Far from equilibrium, at G r -30 kcal/mole, the dissolution rate is much faster and is independent of the degree of saturation. The transition between these two regions occurs somewhere in the range of -20 G r -30 kcal/mole. The G r during the first steady state in experiment (Fig. 2b) was G r =-20.1 kcal/mol. In order to examine if this experiment is conducted under the conditions of the dissolution plateau, we increased the flow rate by a factor of 3.2. As a result the Al and Si concentration decreased and the average G r during the new steady

13 The effect of ph and temperature on smectite dissolution /10/04 state was kcal/mol. Figure 2b shows that the concentration of Si decreases by the same factor (3.2) as the increase in flow rate, and as a result the dissolution rates during the two steady states were the same. If the experiment was close to equilibrium during the first steady state, the consequential decrease in G r would bring to an increase in smectite dissolution rate. Therefore, we conclude that the dissolution rates in experiments with G r kcal/mol were independent of the deviation from equilibrium. According to the results of experiment the dissolution plateau region for smectite is in the range of G r kcal/mol, which was the G r range of most of the experiments (Table 1). The results of Mogollon et al. (1996) showed that the dissolution plateau for gibbsite at 25 C was in very good agreement with the results of Nagy and Lasaga (1992) at 80 C. Assuming that the dissolution plateau for smectite is similarly independent of temperature, the dissolution rates in most of the experiments were independent of the deviation from equilibrium. Some experiments, however, were conducted under close to equilibrium conditions (Table 1). All these experiments where conducted at ph>4. Therefore, these experiments are not used in the fitting of the proposed model The effects of silicon and aluminum on smectite dissolution rate Metz (2001) examined the effects of silicon and aluminum on smectite dissolution rate at 50 C. He found that under far from equilibrium conditions ( G r -30 kcal mol -1 ) smectite dissolution rate is independent both of Al concentration (ranging between 3 to 16 µm at ph 3 and between 21 and 139 µm at ph2) and of Si concentration (ranging between 9 and 41 µm at ph 3 and between 56 and 329 µm at ph2). This rate independency is also supported by an observation from the present study that few experiments that were conducted at the same ph and temperature and with different steady-state Al and Si concentrations showed similar dissolution rate. We suggest that the changes in far from equilibrium dissolution rates observed in the present study (Fig. 4) are not significantly influenced by the variability in Al and Si concentration. It is important to note that the present study ranges of Al ( µm) and Si ( µm) concentrations are significantly larger than the ranges which were thoroughly studied by Metz (2001). Therefore, we cannot rule out the possibility that some of the experimental noise may be related to effects of Al or Si on dissolution rate.

14 The effect of ph and temperature on smectite dissolution /10/ The effects of ionic strength, Na + - and NO 3 To establish the effect of ph on dissolution rate, one should conduct a series of experiments that are identical in all factors except ph. However, adding acid changes the ionic strength and the concentration of the balancing anion. Therefore, studying the effect of ph may be conducted using two possible experimental designs: 1) The input solution contains only an acid and therefore the concentration of the balancing anion (NO 3 in the present study) varies in the different experiments and is equal to the H + concentration; 2) The input solution is composed of a mixture of an acid and a salt (NaNO 3 in the present study), so the concentration of the balancing anion and the ionic strength are the same in all the experiments. In this design the concentration of the salt cation varies in the different experiments. In order to assess the effects of ionic strength and Na and NO 3 concentrations on the determination of the observed ph dependence of smectite dissolution rate, we used both experimental designs. About half of the experiments were conducted with input solution composed solely of HNO 3. The ionic strength in these experiments ranges from M - (ph 4.5) to 0.1 M (ph 1). In the rest of the experiments the ionic strength and NO 3 concentration of ~ 0.32 M were maintained by adding suitable amounts of NaNO 3 into the input solutions. By doing this, Na + - concentration increases as the ph increases whilst NO 3 concentration remains constant (~0.32 M). Figure 4 compares dissolution rates in experiments conducted under constant ionic strength to those obtained without adding NaNO 3, i.e., where H + =NO - 3. In most experiments at ph<4 the differences between the two sets are small. The exceptions are the 25 C experiments at ph < 1.8 and the 50ºC at ph of 3 and 2, where the dissolution rate under constant ionic strength is significantly slower than that in the experiments in which H + =NO - 3. Above ph of 4 the dissolution rates under constant ionic strength are significantly slower at 50 and 70 C and faster at 25 C than those in the experiments in which H + =NO - 3. Taking into account the experimental noise, it is hard either to prove that the observed differences between the experiments represent a real effect of the addition of salt, nor to rule out this possibility. For the purpose of modeling the effects of ph and temperature on smectite dissolution rate we decided to use the results of all the experiments conducted at ph<4, regardless of their ionic strength. As the effect on the rate is

15 The effect of ph and temperature on smectite dissolution /10/04 small to insignificant, the results of the modeling will be only slightly influenced by this decision. 5.3 Modeling the effect of ph and temperature on dissolution rate The model proposed below is a simple version of the model proposed by Cama et al. (2002). The model is based on two assumptions: 1) The proton promoted reaction mechanism consists of fast adsorption of a proton on a surface site followed by a slow hydrolysis step; and 2) The adsorption of the protons on the surface site may be described by a simple Langmuir adsorption isotherm: (5) X H, ads = F b a H + 1+ b a H + where F is the maximum surface coverage of protons on the reactive surface site, b is a constant related to the energy of adsorption and a H+ is the activity of protons in solution. Adsorption of a proton on a surface site close to the metal influences the bond strength and thus affects the dissolution rate. If steady-state conditions are maintained, the rate of this reaction path is (Lasaga, 1981): Rate s ρ (6) = k X H, ads where k (s -1 ) is the rate coefficient, Rate is the observed dissolution rate (mol g -1 s -1 ), s is the specific surface area (m 2 g -1 ), ρ (mol m -2 ) is the density of reactive surface sites on the mineral surface and X H,ads is the molar fraction of the surface site that is protonated. Substituting the Langmuir adsorption isotherm (equation (5)) into equation (6), gives: (7) Rate b a = k F s ρ 1+ b a H + H + Both the rate coefficient, k, and the adsorption coefficient, b, in equation (7) depend on temperature. The temperature dependence of the dissolution rate generally follows the Arrhenius law:

16 The effect of ph and temperature on smectite dissolution /10/04 (8) k = Ae E a / RT where A (s -1 ) is the pre-exponential factor, E a is the apparent activation energy, R is the gas constant and T is the temperature (K). The temperature dependence of the adsorption coefficient may be evaluated recalling that the b constant in the Langmuir adsorption isotherm is the equilibrium constant of the protonation reaction, and therefore its temperature dependence may be written as: (9) 0 0 S / R H / RT b = e e 0 H / RT = K0 e where S 0 (cal mol -1 K -1 ) is the entropy, and H 0 (kcal mol -1 ) is the net enthalpy of adsorption. Following Sverjensky and Sahai (1998), the standard states for both surface and aqueous species are assumed to reflect hypothetical 1 molal solutions referenced to infinite dilution and a surface potential of zero at 25 C. The temperature dependence of the b constant in the Langmuir adsorption isotherm may be described using equation (9) and assuming that the heat capacity, C p, is equal to zero, and therefore H 0 is temperature independent. This last assumption was examined and justified for kaolinite by Cama et al. (2002). The combined effect of ph and temperature on smectite dissolution rate may be described by substituting equations. (8) and (9) into equation (7): (10) Rate = A e s ρ Ea / RT K0 e F 1+ K e 0 H 0 / RT a H 0 / RT H + a H + The coefficients k'=a. F. s. ρ, E a, K 0 and H 0 were calculated from a multiple non-linear regression of equation (10) using least squares. For the fitting we used all the experimental results at 25, 50 and 70 C at ph<4. The resulting coefficients are k'=220±750 mol g -1 s -1, E a =17460±2000 cal mol -1, K 0 =3± and H 0 =-10700±2500 cal mol -1. The regression coefficient is R 2 =0.93. Substituting these values into equation (10) yields, (11) Rate = 220 e 17460/ RT e10700/ RT a e10700/ RT a H + H +

17 The effect of ph and temperature on smectite dissolution /10/04 A comparison between the prediction of equation (11) and the experimental data at 25 C, 50 C and 70 C is shown in Fig. 4. The obtained activation energy, 17±2 kcal/mol, is similar to the activation energy of the dissolution of the edge site of kaolinite (22 kcal/mol, Cama et al., 2002) and is within error equal to the average value of 15 kcal/mol of apparent activation energies for silicates dissolution rate (Lasaga et al., 1994). The obtained net enthalpy of adsorption of -11±2 kcal/mol is within the range of 7.9 to 23.1 kcal/mol, experimentally obtained for oxides (Sverjensky and Sahai, 1998). The reasonable activation energy and net enthalpy of adsorption, although not proving the proposed model, provide support for its validity. 5.4 Comparing the prediction of the model with measured adsorption isotherms The relative adsorption/desorption of protons on mineral surfaces is commonly measured using potentiometric surface titration. The term "surface titration" is somewhat misleading as the measurements of the adsorption of proton onto the mineral surface are based on changes in the ph of the solution. The relative surface concentration of protons is determined by mass balance between the proton (or hydroxide) added to solution and the measured proton concentration in solution after equilibration using the so-called proton consumption function + V Cs = ( CA CB [ H ] + [ OH ]) (12) A where C s is the change in surface concentration of protons (mol m -2 ), C A and C B are the concentrations of the acid and base added (mol l -1 ), respectively, [H + ] and [OH - ] are the solution concentrations of H + and OH - after equilibration (mol l -1 ), V is the fluid volume (l) and A is the total surface area (m 2 ). It is important to note that the value of the proton surface charge obtained by surface titration is arbitrary until a value of zero charge is established (Davis and Kent, 1990; Ganor et al., 2003), i.e., the titration measures a relative change in surface concentration (as defined by the proton consumption function, Eq. (12)) and not the absolute concentrations (Schroth and Sposito, 1997). Therefore, the calculation of the absolute proton adsorption density is based on an assumption regarding the ph PZNPC. Figure 5 shows the surface titration data obtained by Zysset and Schindler (1996) at 25 C between ph

18 The effect of ph and temperature on smectite dissolution /10/04 4 and 1. According to the interpretation of Zysset and Schindler (1996) the surface protonation at this ph range is controlled by both adsorption of H + on aluminol site and ion exchange reaction: at ph 2.5 protonation of the aluminol is dominant (black dots in Fig. 5), whereas at ph<2.5 ion exchange significantly contributes to the concentration of adsorbed H + (squares in Fig. 5). A byproduct of the fitting of the proposed model (eq. (10)) is that it predicts the molar fraction of surface protonation. This prediction of the proposed model may be compared to protonation data of Zysset and Schindler (1996). Such a comparison is not straightforward. Zysset and Schindler (1996) measured the concentration of adsorbed H + ions (mol g -1 ), which results from several reactions including adsorption and cation exchange (Stadler and Schindler, 1993; Zysset and Schindler, 1996). In contrast, the fitting of the proposed model predicts the molar fraction (and not the total concentration) of a protonated edge site that governs the dissolution rate under the examined ph range. It is important to note that the rate law proposed in the present study assumes that the dissolution rate is proportional to the concentration of protons that are adsorbed on the reactive edge surface site and is independent of protonation due to ion exchange reaction. This is the major difference between the present study rate law and that of Zysset and Schindler (1996), who proposed that the rate is linearly proportional to the sum of the proton concentrations on the aluminol and the cation exchange sites (see eq. 20 in Zysset and Schindler, 1996). Based on the present study prediction, the total concentration of adsorbed H + ions ({H + }, mol g -1 ) may be calculated by multiplying the predicted molar fraction at the reactive edge site by the total concentration of this site (θ, mol g -1 ). (13) H 0 / RT K e a + { H } = { H } + H = H + 0 θ { } H 0 0 / RT 1+ K0 e ah e10700/ RT a + θ e10700/ RT a H + H + In the calculations we assume that the concentration of adsorbed H + ions on sites that are not influencing the rate is constant in the examined ph range and their sum equals {H + } 0. The term {H + } 0 may include contribution from the permanent charge of the mineral surface as well as the real surface charge at the ph that was assumed to be the ph PZNPC. The coefficients

19 The effect of ph and temperature on smectite dissolution /10/04 {H + } 0 and θ may be obtained by fitting Eq. (13) to adsorption isotherm data using least squares. As the dissolution rate is assumed to be independent of the extent of ion exchange, and as according to Zysset and Schindler (1996) ion exchange significantly contributes to the observed {H + } at ph<2.5, only the seven data points at ph 2.5 (black dots in Fig. 5) were used for the fitting. The solid line in Fig. 5 is the best-fit curve obtained for the surface titration of Zysset and Schindler (1996) at 25 C in the ph range of 2.5 to 4. The obtained coefficients are {H + } 0 =8.3± mol g -1 and θ=2.42± mol g -1, and the regression coefficient R 2 = Figure 5 shows that the model proposed in the present study predicts the concentration of adsorbed H +, between ph 1.5 and 4, although only the data at ph 2.5 were used for the fitting. Following Zysset and Schindler (1996), we suggest that the excess concentration of adsorbed H + below ph 1.5 is due to ion exchange. According to Zysset and Schindler interpretations below ph 2.5 ion exchange significantly contributes to the concentration of adsorbed H +. In contrast, the prediction of the present study indicates that the contribution of ion exchange is significant only below ph 1.5. Surface titration curves are commonly interpreted in terms of protonation and deprotonation of surface sites (see for example Davis and Kent, 1990; Parks, 1990; Lutzenkirchen and Kienzler, 2002, and references therein). This interpretation involves a critical stage in which the amount of protonation and deprotonation reactions is assumed. In most cases, this assumption is based on the minimum amount of sites that are required to obtain an adequate agreement between the surface speciation model and the observations. Mathematically, it is always possible to add more reactions while keeping (or improving) the quality of the fitting. The comparison between the predictions of the present study and the observations of Zysset and Schindler (1996) is based on the assumption that between ph 2.5 and 4 the protons consumption is controlled by the protonation of a single surface site. Neither the present study, nor the study of Zysset and Schindler (1996) provided any evidence supporting this assumption. Taking into account that smectite contained many different surface and interlayer sites that may adsorb and exchange protons, this assumption may be questioned. Moreover, Stadler and Schindler (1993) argued that smectite protonation below ph 4 is possibly dominated by ion exchange. Therefore, although the predictions of the proposed model show a very good agreement with the total concentration of adsorbed H +

20 The effect of ph and temperature on smectite dissolution /10/04 between ph 1.5 and 4, this agreement cannot serve as a proof for the proposed model, and the obtained value of the total concentration of the reactive edge site (θ) should be regarded with caution. 5.5 Comparison of the results of the present study to those of previous studies Figure 6 compares smectite dissolution rates that were obtained in the present study to results of Furrer et al. (1993), Zysset and Schindler (1996) and Metz et al. (2004a). Furrer et al. (1993) conducted both flow-through and batch dissolution experiments with K- montmorillonite sample SWy-1. All the experiments were conducted at 25 C, using HCl/KCl solutions. In most of their experiments the ionic strength was adjusted to 0.1M. The rates of the flow-through dissolution experiments of Furrer et al. (1993) are generally slower than the rates of their batch experiments (Fig. 6a). A comparison between the results of the present study and those of Furrer et al. (1993) shows a similarity between the dissolution rates in the flow-through experiments of the present study and those in the batch experiments of Furrer et al. (1993). Zysset and Schindler (1996) conducted three sets of batch dissolution experiments in which the solutions contained 0.03, 0.1 and 1M KCl, using the same sample as Furrer et al. (1993). The experiments conducted at ph<2 and 0.03M KCl display stoichiometric dissolution. The ionic strength in these experiments of Zysset and Schindler (1996) were above the ionic strength of the present study experiments, which were conducted at the same ph without salt and below those with salt. Their dissolution rates (Fig. 6a) were slightly slower than those of both data sets of the present study. The experiments conducted by Zysset and Schindler (1996) at ph>2 and 0.03M KCl display non stoichiometric dissolution. The present study (stoichiometric) rates are located between the rates that were calculated by Zysset and Schindler (1996) from the release of Si and those from the release of Al (Fig. 6b). The rates obtained by Zysset and Schindler (1996) in experiments conducted with 0.1 and 1 M KCl show a significant enhancement of rate in comparison to the experiments conducted under relatively low ionic strength. Adding NaNO 3 in the present study hardly had an effect on the dissolution rate, and in some experiments, adding NaNO 3 even inhibited the dissolution. Figure 6b also shows that the effect of ph on dissolution rate is smaller in the experiments conducted by Zysset and Schindler (1996) with 0.1 and 1 M KCl than the effect

21 The effect of ph and temperature on smectite dissolution /10/04 observed in the present study. We do not have a good explanation for the differences between the observation of the present study and those of Zysset and Schindler (1996). It is interesting to note that the dissolution rate observed by Furrer et al. (1993) at ionic strength of 0.1M resembled the rate that was calculated based on Si by Zysset and Schindler (1996) at 0.03M. Metz et al. (2004a) measured the dissolution rate of smectite sample SAz-1 at 50 C and ph 2 and 3. The input solution in the far-from equilibrium experiments of Metz et al. (2004a) was composed of pure HClO 4 (no salt was added). Figure 6c shows that the dissolution rate of smectite is not significantly influenced by the type of balancing anion (ClO - 4 vs. NO - 3 ). 6 SUMMARY AND CONCLUSIONS Steady-state smectite dissolution rates were examined using non-mixed flow-through reactors. The experiments were conducted at 25 C, 50 C and 70 C in a ph range of 1 to 4.5. Most of the experiments were conducted under far-from equilibrium conditions ( G r < -21 kcal mol -1 ). Some experiments at ph>4, were conducted under conditions closer to equilibrium. These experiments were not used in the fitting of the proposed model. In general, smectite dissolution rate increases with temperature and decreases with ph. The experimental results were fitted to a model, which is a simple variation of the model proposed by Cama et al. (2002) for the proton-promoted reaction of kaolinite. According to our model, the dissolution rate is proportional to the concentration of protons that are adsorbed on the reactive edge surface site, which may be expressed using the Langmuir adsorption isotherm. We do not have any information regarding the identity of this active site. Zysset and Schindler (1996) postulated that the active site that controls the rate under acidic conditions is either Al-O-Si or Al-OH-Al group. The results of the present study neither support nor contradict this suggestion. The dependence of the dissolution rate on temperature is affected by the activation energy and the adsorption enthalpy. From fitting our results to the proposed model we found activation energy of 17±2 kcal/mol and enthalpy of -11±2 kcal/mol.

22 The effect of ph and temperature on smectite dissolution /10/04 Acknowledgments. This research was supported by THE ISRAEL SCIENCE FOUNDATION (grant No. 174/01). We gratefully acknowledge thorough reviews by the associate editor, Jacques Schott, and by Stephan.J. Köhler and Andreas Bauer. We thank Volker Metz for both fruitful and knowledgeable discussions and for conducting the thermodynamic calculations. The technical assistance of Ester Shani, Nivi Kesler, Ruth Talby and Gony Yagoda is greatly acknowledged.

23 The effect of ph and temperature on smectite dissolution /10/04 1 Table 1: Experimental conditions and results Experiment sample NaNO 3 mass Flow rate SS time p H [Al] [Si] Rate Al Rate Si G r (M) (g) (ml min -1 ) (days) input output (µm) (µm) (mol g -1 s -1 ) (mol g -1 s -1 ) (kcal mol -1 ) 25 C * treated E-11 ±5E E-11 ±5E * treated E-11 ±1E E-11 ±1E * treated E-11 ±1E E-11 ±1E raw E-11 ±3E E-11 ±3E treated E-11 ±3E E-11 ±3E * treated E-11 ±6E E-11 ±6E * treated E-11 ±6E E-11 ±6E treated E-11 ±2E E-11 ±2E raw E-11 ±4E E-11 ±3E raw E-11 ±5E E-11 ±5E * treated E-11 ±3E E-11 ±3E * treated E-11 ±3E E-11 ±2E * treated E-11 ±3E E-12 ±2E * treated E-11 ±2E E-12 ±2E treated E-12 ±1E E-12 ±1E treated E-11 ±4E E-11 ±3E treated E-12 ±2E E-12 ±2E treated E-12 ±8E E-12 ±7E * treated E-12 ±1E E-12 ±1E raw E-12 ±2E E-12 ±1E raw E-12 ±3E E-12 ±1E treated E-12 ±1E E-12 ±4E * treated E-12 ±9E E-12 ±3E * treated E-12 ±9E E-12 ±3E * treated E-12 ±2E E-12 ±5E * treated E-12 ±1E E-12 ±7E raw E-13 ±3E E-12 ±2E raw E-13 ±3E E-13 ±1E * treated E-12 ±3E E-12 ±3E treated E-13 ±4E E-13 ±5E * treated E-13 ±3E E-13 ±2E * Experiment that were conducted with NaNO 3 ; SS time = time from the beginning of the experiment to steady state.