An Experimental Study on the Ion-Exchange Behavior of the Smectite of Cabo de Gata (Almería, Spain): FEBEX Bentonite

Journal of Colloid and Interface Science 239, 409–416 (2001) doi:10.1006/jcis.2001.7605, available online at http://www.idealibrary.com on

An Experimental Study on the Ion-Exchange Behavior of the Smectite of Cabo de Gata (Almer´ıa, Spain): FEBEX Bentonite F. Javier Huertas,∗, 1 Pablo Carretero,† Jordi Delgado,† Jos´e Linares,∗ and Javier Samper† ∗ Department of Earth Sciences and Environmental Chemistry, Estaci´on Experimental del Zaid´ın, CSIC, Profesor Albareda 1, 18008 Granada, Spain; ˜ Campus de Elvina ˜ s/n, 15192 A Coruna, ˜ Spain and †ETS de Ingenieros de Caminos, Canales y Puertos, Universidade da Coruna, Received May 24, 2000; accepted April 2, 2001

Binary cation-exchange processes (Na–K, Na–Mg, and Na–Ca) have been investigated at 25◦ C in samples of bentonite from Cabo de Gata (Almer´ıa, SE Spain). Exchange isotherms and Vanselow selectivity coefficients indicate that the preference of this bentonite for K+ , Ca2+ , and Mg2+ ions is greater than that for the Na+ ion. The change in ij KV as a function of the equivalent fraction of cation M, EM , is interpreted on the basis of the presence of different types of adsorption/exchange sites (tetrahedral, octahedral, and edge sites). For the Na/K exchange, K Na KV exhibits a steep decrease for EK < 0.25–0.3. This is interpreted as due to preferential exchange at the tetrahedral sites. However, the opposite behavior is observed in the case of Ca and Mg, with a noticeable increase of iNa KV for E Ca (E Mg ) > 0.6–0.8. The compared selectivities of Ca and Mg for this smectite are nearly equal. As far as the exchanger is concerned, the Na–Ca and Na–Mg exchange is nearly ideal, in the raoultian sense. Thermodynamic exchange constants and the corresponding Gibbs free energies of reaction have been computed from ij KV coefficients. This allow us to quantify the exchanger’s affinity sequence, which appears to be Na+ < K+ < Mg2+ < Ca2+ . ° C 2001

Academic Press

Key Words: smectite; ion exchange; selectivity coefficients; nuclear waste repository.

INTRODUCTION

Isomorphic substitutions in the structure of clay minerals produce structural charge imbalances that are usually equilibrated via ion adsorption. For the smectite group minerals, the charge excess is permanent and negative in sign, and it becomes apparent in the interlayer space where it is counterbalanced by the specific adsorption of cations. Due to this behavior and the loose bonding of the interlayer cations with the clay structure, smectites are excellent cation exchangers. In soils and sediments Na+ , K+ , Mg2+ , and Ca2+ are the most common interlayer cations but, depending on particularities of the chemical system, heavy metals or radionuclides can be also exchanged (1–5). The exchange reactions involving alkaline and alkaline earth cations in smectites have been extensively studied because of 1 To whom correspondence should be addressed. Fax: +34 958 129 600. E-mail: [email protected].

their obvious interest for soil science. The affinity of smectites for alkaline metals increases with the ionic radius of cations (Cs > Rb > K > Na > Li) (6–7), which appears to be related to the decrease in the effective ionic radius of the hydrated ions and to the reduction of their corresponding hydration energy (8). The overall applicability of the affinity series has been reported in numerous studies (see the extensive review by Bruggenwert and Kamphorst (9)). Based on the same theoretical approach, a similar sequence can be postulated for alkaline earth cations (Ba > Sr > Ca > Mg) (10). However, available literature data are controversial, to some extent, in this respect. Sposito et al. (11–13) have observed nearly identical affinities for Ca and Mg in Wyoming bentonite, while Suarez and Zahow (10) pointed out an affinity slightly higher for Ca than for Mg on the same material. A higher selectivity for Ca than for Mg has also been reported in experiments performed on montmorillonitic soils by Fletcher et al. (14). Such a difference in clay behavior has been attributed to (a) effects induced during sample preparation (10, 15), (b) the presence of organic matter (10, 14), or (c) intrinsic chemical interactions between the exchanger and the exchange complex. With regard to the latter, Maes and Cremer (16) have shown that the charge density of a clay plays a significant role in its ionic affinity. For example, Cs selectivity increases at the expense of that of Na when octahedral substitutions increase. Isomorphic substitutions at the tetrahedral sheet of smectites enhance the development of strongly bonded inner-sphere complexes (17), which are also favored by the increase in their ionic radii. Therefore, Ca should be preferred over Mg based on its hydration energy but the selectivity should decrease whenever surface charge density decreases and tetrahedral replacements take over (10). The high exchange capacity of smectite, as well as its thermal, rheological, and hydraulic properties, makes it an extremely suitable material for many practical applications. Reference concepts behind the engineering design of waste facilities take advantage of these properties. In the case of a high-level nuclear waste (HLW) repository, smectite clay group minerals comprise an engineered bentonite barrier and, sometimes, a significant component of a clayey host rock (18). In crystalline rocks, HLW is conceptualized to be stored inside bentonite-lined metal canisters distributed along horizontal drifts or vertical shafts (18–19).

409

0021-9797/01 $35.00

C 2001 by Academic Press Copyright ° All rights of reproduction in any form reserved.

410

HUERTAS ET AL.

In this context and within the framework of the FEBEX Project (20–21) several experiments scoping various time and spatial scales are being performed. Among them, the most significant is the so-called in situ test. This experiment consists of a drift of 17-m length and 2.3-m diameter excavated in a granitic formation at Grimsel, Switzerland, filled with two electric heaters and a bentonite barrier of 0.45-m thickness (20). The final purpose of the experiment is twofold: first, a hands-on technological demonstration on the constructability of HLW repositories; and second, modeling with computer codes the thermal, hydrodynamic, geochemical, and mechanical processes taking place on the bentonite barrier. Reactive transport modeling of the experiments is being pursued (21) using real time monitored data as well as information gathered from shorter term laboratory experiments. By the summer of year 2001, the in situ experiment will be decommissioned and samples will be taken and examined for geochemical changes induced by hydration and heating. At this point, the experimental observation will be compared to model predictions to check the predictive capabilities of current thermohydrogeochemical models (21). In this paper we report experimental results on the binary exchange reactions between the FEBEX bentonite and Na, K, Mg, and Ca. These results are used to calculate their selectivity coefficients and derive thermodynamic data needed to model cation-exchange reactions.

and dried to 60◦ C in a conventional oven. Before the experiment the clay was kept stored in polyethylene bottles at room temperature and humidity. Mineralogical analyses of the samples were carried out by X-ray diffraction (XRD). Several XRD patterns were recorded on powder specimens and on oriented and glycolated (<2-µmsize fraction) mounts using a Philips PW1730 diffractometer with CuK α radiation, Ni filter, and a graphite monochromator. Semiquantitative analysis based on the relative corrected areas of selected peaks from the X-ray diffractograms yielded an average mineralogical distribution of 96% dioctahedral smectite, 1.5% cristobalite, 1% quartz, 1% calcite, and trace amounts of K-feldspar. XRD of the oriented and glycolated mounts showed that it was actually an R = 0 mixed-layer illite/smectite (I/S), with 15% illite layers (27). Wet chemical analyses of the bulk bentonite (Table 1) were performed following the method described by Shapiro (28). The cation-exchange capacity (CEC) and exchangeable cations were determined separately according to the procedures of the Soil Conservation Service (29), consisting of saturation with NH+ 4 at with a NaCl solution. pH 7 and subsequent displacement of NH+ 4 The resulting CEC of the FEBEX sample is 102 ± 3 meq/100g of dry clay (Table 1). After correction for the presence of nonsmectitic mineral phases identified by XRD, the structural formula of the mixed-layer I/S is calculated as Na0.179 K0.201 Mg0.124 Ca0.157 [Al2.854 Fe0.356 Mg0.885 ]

MATERIALS AND METHODS

[Si7.656 Al0.344 ]O20 (OH)4 .

Sample Pretreatment and Preparation of the Starting Material Several tons of bentonite were sampled from a quarry located in Cortijo de Archidona in Serrata de N´ıjar (Cabo de Gata, SE Spain) to perform the different FEBEX experiments. Hereafter, this material will be called the FEBEX bentonite (20). The origin of the bentonite rock is associated with hydrothermal alteration processes taking place over tuffaceous volcanic rocks (for a detailed description, see Refs. 22–26). After being sampled from the quarry, the bentonite was subjected to a homogenization process according to the protocol described in ENRESA (20) and kept stored in big bags under atmospheric moisture conditions. Subset samples taken from the big bags were distributed to laboratories participating in the FEBEX project. One of these subsets was used in the present study. The bulk bentonite sample was dry ground in an agate mortar, mechanically rehomogenized,

Cation-exchange data (Table 1) indicate that only 0.016 K+ ion per unit cell is amenable for ion exchange. The remaining 0.185 ion corresponds to the illite layers that do not participate in exchange reactions. To perform the exchange experiments, the FEBEX bentonite was homoionized to a Na-saturated form. A bentonite suspension was prepared by adding 50 g of ground bentonite to 500 mL of a 0.5 M NaCl solution. The suspension was periodically handshaken. After 24 h contact, the supernatant was decanted and separated from the clay slurry. A new batch of 500 mL of the 0.5 M NaCl solution was added to the slurry, repeating the cycle four times. The final suspension was centrifuged at 8000 rpm for 10 min, the liquid phase discarded and the solid rinsed repeatedly with a water/acetone 1 : 1 mixture up to a negative result of

TABLE 1 Chemical Analyses of a Sample of the FEBEX Smectite

Total % Exchangeable cations (meq/100 g) CEC (meq/100 g)

SiO2

Al2 O3

Fe2 O3

CaO

MgO

Na2 O

K2 O

TiO2

P2 O5

MnO

H2 O+

Total

58.92

19.48

3.48

2.51 42

4.83 32

2.28 25

1.21 2.5

0.27

0.06

0.06

7.09

100.19 101.5

Note. Oxides are given as weight percentages while exchangeable cations are expressed as meq/100 g dry sample.

102

411

ION-EXCHANGE BEHAVIOR OF FEBEX BENTONITE

the AgNO3 /AgCl test. Once the treatment was completed, the solid was oven-dried at 60◦ C, reground in an agate mortar, and kept stored until needed. To check that Na+ ion adsorption was fully achieved, one of the samples was reacted with a pH 7, 1 M NH4 Ac solution to test the degree of success in the homoionization process. Flame emission spectrometry analyses (Scharlau Science) indicated that the Na content of this sample was 103 ± 4 meq/100 g dry clay, consistent with the measured CEC value. Hereafter, the homoionized Na-bentonite will be referred as the starting material in the exchange experiments.

is that of Vanselow (31), according to which a generic binary ion-exchange reaction can be represented by £ ¤ £ ¤ +z j i z i M j X Z j ex + z j Mi+z [1] aq ⇔ z i M j aq + z j Mi X z i ex , where Mi and M j are the exchangeable cations, X represents a functional group of the exchanger, z i and z j are the electrical charges of the exchangeable cations, and aq and ex refer to the aqueous solution and exchanger phases, respectively. The conditional Vanselow constant, K V , is defined as z

i j KV

Experimental Cation-Exchange Isotherms The starting Na-smectite was reacted at room temperature (25◦ C) with mixed-salt solutions of NaCl/KCl, NaCl/MgCl2 , or NaCl/CaCl2 at the equilibrium pH of the salt solution (close to neutrality; no buffers were used to avoid interferences with the adsorption/exchange reaction), keeping a total cation normality of 0.5 eq L−1 . Aqueous speciation of the bathing solutions performed with EQ3NR (30) indicate that the most abundant Ca, Mg, and K aqueous species are the free ions. Only 10% of total magnesium is present as MgCl2 , 3% of total calcium is present as CaCl2 , and KCl is nearly absent in the solution. Isotherm exchange experiments were performed using 10 experimental points, which cover the complete range of corresponding binary equivalent fractions. Each data point was duplicated, which means that, after the end of the experiment, no less than 20 points per isotherm were available. Two grams of the starting material was placed inside a polyethylene centrifuge tube and left to react for 72 h with 10 mL of one of the mixed-salt initial solutions. The clay/water suspension was periodically hand-shaken to prevent lump formation. Once the reaction time was completed, the water/clay mixture was centrifuged (5000 rpm) and the supernatant solution was analyzed for the cation concentrations (Na and, depending on the starting solution, K, Ca, or Mg), whereas the solid was rinsed with a 1 : 1 water/acetone solution until the AgNO3 /AgCl test was considered successful. Exchangeable cations were replaced in the solid with the aid of a 1 M, pH 7 ammonium acetate solution. Cations were analyzed by atomic flame emission (Na, K) and absorption (Ca, Mg) spectroscopy (Perkin Elmer 1100B, running an air–acetylene flame). The analytical errors in Na and K determinations were lower than 2% and, in the case of Ca and Mg, lower than 3%. The amount of adsorbed cations was estimated from the analysis of the smectite exchange complex, after exchange with ammonium acetate. Data Analysis The mathematical expressions accounting for ion-exchange processes are stoichiometric mass-action equations in which conditional equilibrium constants (also known as selectivity coefficients) are used because they cannot be considered genuine thermodynamic equilibrium constants. Several conventions are employed to express the values of selectivity coefficients in the literature. Their formalism depends on the definition of the exchanger itself (31–34). One of the most popular formulations

=

zi aM NMji j z

zi aMj i NM j

,

[2]

where a stands for the aqueous activities of cations i and j, and N is the mole fraction of the cations i and j in the exchanger. The driving force for the exchange reaction is the gradient in chemical potential existing between the exchanger and the liquid phase (35). From a thermodynamic point of view, the equilibrium constant of reaction [1] relates aqueous and solid activities through the mass-action activity ratio: z

i j K ex

=

j i a zj,aq ai,ex

z

j i ai,aq a zj,ex

.

[3]

The activity of the species in solution is given by their mole concentration, m i,aq , times the activity coefficient, γi . A model accounting for potential deviations from ideality in the exchanger is needed for adsorbed ions, ai,ex = f i Ni,ex ,

[4]

where f i is the rational activity coefficient of ion i in the exchanger and Ni,ex is the mole fraction of ion i in the solid exchanger. Under standard state (pure homoionic form in equilibrium with an infinitely dilute solution of the exchanger ion at any temperature and pressure (35)) and reference conditions (Raoult’s law), the activity coefficient of the exchanger is equal to one and Eq. [3] becomes i j K ex

=

(γ j m j )zi ( f i Ni )z j . (γi m i )z j ( f j N j )zi

[5]

From Eqs. [2] and [5] one can relate the thermodynamic equilibrium constant to the conditional Vanselow constant: z

i j K ex

= ij K V

fi j . f jzi

[6]

Sposito (36) presents a detailed review of the thermodynamic relationships of ion exchange processes and shows that, by performing adequate transformations of mole fractions (N ) into equivalent fractions (E), it is possible to compute the solid activity coefficients, f j and f i , from Vanselow constants by means of ZEi z i ln f j =

E i ln ij K V

−

ln ij K V d E i 0

[7]

412

HUERTAS ET AL.

Z1 z j ln f i = −(1 −

E i ) ln ij K V

+

ln ij K V d E i ,

[8]

Ei

where the equivalent fractions of cations i and j in the solid, E i and E j , are defined by the following expressions: Ei =

|z i |Ni |z i |Ni + |z j |N j

[9]

Ej =

|z j |N j . |z i |Ni + |z j |N j

[10]

It is possible then to compute the thermodynamic exchange constant, ij K ex , by taking the logarithm of Eq. [6] and substituting into [7] and [8]: Z1 ln ij K ex

=

ln ij K V d E i .

[11]

0

In fact, to evaluate accurately the thermodynamic exchange constant and the activity coefficients of the exchanger it is necessary to add some extra terms to Eqs. [7], [8], and [11] to take into account the variable amount of water adsorbed and its corresponding activity in the exchanger (36). Due to the difficulties embedded with the experimental determination of water activity in solids, these terms are usually ignored, although it is acknowledged that their contribution to activity coefficients is small but positive. On the other hand, Fletcher (35) suggests that ignoring water activity terms can lead to underestimates of activity coefficients by as much as 20% for monovalent and 30% for divalent ions in solutions with ionic strengths around 2 mol kg−1 . This effect is less pronounced in dilute solutions, such as those considered in our study. Therefore, under our experimental conditions the underestimation of activity coefficients is expected to be smaller than what Fletcher (35) proposes. A mathematical expression relating the change in ijK V as a function of the exchanger composition, within the investigated range, can be obtained by least-squares fitting of the experimental observations to a third-order polynomial expansion (Margules type) of the corresponding equivalent fractions, E i (36), ln ij K V = a0 + a1 E i + a2 E i2 + a3 E i3 + · · · + an E in , [12]

RESULTS AND DISCUSSION

FIG. 1. Experimental isotherms corresponding to (a) Na/K, (b) Na/Ca, and (c) Na/Mg binary exchange reactions. Dashed lines represent nonpreferential exchange isotherms and solid lines are computed isotherms by using Margules expansion for ln iNaK V (Eq. [12]).

Experimental isotherms relating the equivalent fraction of cations measured in the solid (E i ) to the corresponding ones in equilibrium with the liquid phase ( E˜ i ) are plotted in Fig. 1. For each binary Na/cation system, data points delineate different L-shaped exchange isotherms. Sposito (36) makes clear that before concluding cation preferences, it is mandatory to compare experimental results with nonpreferential homovalent and

heterovalent isotherms of the corresponding cation (dashed lines in Fig. 1). In fact, the shape of the isotherm depends not only on the solution’s concentration but also on the charge of the exchanging ions. It is clear from Fig. 1 that the preference of the FEBEX bentonite for K, Mg, and Ca is higher than that for Na. Vanselow selectivity coefficients for each binary Na/cation exchange reaction have been computed according to Eq. [2].

where a0 , a1 , a2 , . . . , an are fitting constants.

ION-EXCHANGE BEHAVIOR OF FEBEX BENTONITE

413

of the smectite to have higher affinity for K than for Na. The exchange constant decreases as K progressively replaces Na in the smectite. This tendency changes when the potassium equivalent fraction, E K , reaches a value around 0.25–0.3. From this point to E K ≈ 0.7, the Vanselow coefficient remains nearly constant, irrespective of E K . Finally, it decreases again for E K > 0.8. Gast (6–7) has reported the same behavior in ij K V for the Na/K exchange as well as for Na/Rb and Na/Cs when the smectite becomes saturated in these cations. The crystal chemical nature of the different exchange positions in the smectite structure (tetrahedral, octahedral, and edge sites) accounts for the behavior observed in the Na/K exchange isotherm. According to Sposito (17), tetrahedral sites complex cations strongly, especially those able to form innersphere complexes (e.g., K+ ), whereas octahedral positions enhance the Lewis character of ditrigonal cavities and thus favor complexing with outer-sphere pairs (e.g., Na+ , Mg2+ , or Ca2+ ). Based on this, one should expect K Na K V to reach a higher value when exchange occurs at the tetrahedral sites than when it occurs at the octahedral sites. Data in Fig. 2a are consistent with this observation because K Na K V decreases for E K values smaller than 0.25–0.3. This range of values is very close to the relative amount of tetrahedral charge (36%) in the FEBEX smectite, and the observed difference may be due to the presence of nonexchangeable K+ ions fixed at tetrahedral sites in illite layers. Besides, it is possible to describe the Na/K isotherm data using two segments corresponding to Langmuir-type (linearized) sorption equations (Fig. 3). This observation points to the possibility of at least two types of surface exchange sites, each involving a different amount of sorption energy. The crossover of both lines occurs at E K ≈ 0.23, indicating the transition between the relative abundance between the different types of sites. The final smooth decrease in K Na K V observed in Fig. 2a (E K > 0.8) can be associated with ion exchange taking place mainly on the external surface of the smectite.

FIG. 2. Measured (symbols) and computed (lines) Vanselow selectivity coefficients for (a) Na/K, (b) Na/Ca, and (c) Na/Mg binary exchange reactions as a function of the equivalent fraction in the solid phase. Lines represent the polynomial fit obtained with Eq. [12] (Table 2).

They are represented in Fig. 2. Activity coefficients of aqueous species were calculated with the EQ3NR software (30), which uses an extended version of the Debye–H¨uckel equation (B-dot). For the Na/K exchange reaction, Vanselow coefficients are greater than one. This is in agreement with the tendency

FIG. 3. Plot of the average values of the Na/K exchange isotherm in a double inverse graph. The experimental points fit to straight lines, which correspond to different types of exchange sites. The crossing point occurs at E K(solid) = 0.23.

414

HUERTAS ET AL.

Vanselow selectivity coefficients for the Na/Mg isotherm indicate the preference of smectite for the divalent cation. They are Mg nearly constant (Na K V ≈ 5.6) for E Mg < 0.6, but they increase up to a value of 10.2 for the nearly Mg-saturated smectite. The interpretation of the Na/Ca exchange resembles that of Na/Mg although the selectivity coefficients are greater (Ca Na K V ≈ 7.0 for E Ca < 0.6) for the latter. The preference of the smectite for Mg or Ca instead of Na has been already reported by several authors (e.g., 9, 14, 37). One of the factors controlling this behavior appears to be the higher hydration energy of both cations with respect to that of Na+ . In contrast with the Na/K exchange, no special preference is observed in the Na/Ca or Na/Mg exchange over the tetrahedral sites (Figs. 2b and 2c). This should be the case for cations that form outer-sphere complexes but iNa K V remains roughly constant for E < 0.6–0.8. However, when the equivalent fraction of alkaline earth cations becomes greater than ∼0.6–0.8, the smectite exhibits stronger selectivity for divalent ions than for Na. When the Na/Ca ratio is low, Ca2+ adsorption on the smectite platelets allows their stacking to form quasicrystals (38–39). Once Mg or Ca concentrations are large enough, the interlayer space of the quasicrystals is saturated with respect to the divalent cation (E M2+ > 0.7) (11) and the external surface becomes the most significant location for the Na+ /M2+ exchange. Thus, the smectite shows higher Na+ /M2+ selectivity when the interlayer space of the quasicrystal is nearly saturated in Mg or Ca and the external surface becomes saturated in divalent cations (Figs. 2b and 2c). A comparison of the selectivity coefficients for the Na/Mg and Na/Ca exchange reactions indicates that the smectite has a slightly higher affinity for Ca than for Mg, which agrees with the findings of Suarez and Zahow (10) for Wyoming bentonite. The ij K V values for the three experimental isotherms have been fitted to Margules-type, third-order polynomial functions (Eq. [12]). Fitting coefficients are given in Table 2. With the obtained coefficients it is possible to compute also binary Na/cation isotherms, which serves as a consistency check (Figs. 1a–1c). This can be easily achieved reordering properly Eq. [2] and replacing mole by equivalent fractions (36). With the aid of Eqs. [7] and [8] and the ij K V ’s calculated from Eq. [12], it is possible to compute rational activity coefficients for the solid phase (Fig. 4). The shape of the curve describing the activity coefficients as a function of exchanger composition depends on the ability of the fitting ij K V function to reproduce the experimental data. Moreover, for equivalent fractions of Ca and Mg less than 0.12 and 0.09, respectively, and those of K larger TABLE 2 Least-Squares Fitting Constants for the ln ij K V Function (Eq. [12]) and the Mean Error (σ) of the Estimate of ln ij K V Reaction

a0

a1

a2

a3

σ

Na/K Na/Mg Na/Ca

2.455 1.728 0.887

−6.176 0.187 9.149

12.558 −1.721 −21.412

−8.332 2.039 14.523

±0.202 ±0.262 ±0.190

FIG. 4. Solid-phase activity coefficients as a function of equivalent fractions for each binary exchange. (a) Na and K; (b) Na and Ca; (c) Na and Mg.

than 0.92, the curves are extrapolations and should be taken with caution. Computed data indicate that for the Na/Ca and Na/Mg exchange reactions the FEBEX bentonite behaves like an ideal exchanger, in the raoultian sense. Only at high Mg and Ca contents do the activity coefficients deviate significantly from unity. The similarity in the activity coefficients of Ca and Mg can be expected from the similar selectivity of the FEBEX bentonite for Ca and Mg. The steep rise in the activity coefficient of Ca at low equivalent fractions could be an artifact introduced by the extrapolation of the Ca Na K V function outside the experimental range, from which no clear conclusion can be drawn. However,

415

ION-EXCHANGE BEHAVIOR OF FEBEX BENTONITE

at high Ca content, there is a significant increase in the activity coefficient of Na in the exchanger. The case of potassium is somewhat different. At low K content (E K < 0.2) the activity coefficient is less than 1. However, it becomes greater than unity for equivalent fractions greater than 0.2. This behavior suggests that the Na/K exchange in the FEBEX bentonite is slightly nonideal. The solid-phase activities (i.e., the rational activity coefficient of ion i times its corresponding mole fraction, f i Ni ) are

TABLE 3 Thermodynamic Equilibrium Constants, ij K ex , and Gibbs Free Energies for the Na/K, Na/Mg, and Na/Ca Exchange Reactions Reaction

M K Na ex

NaX + K+ ⇔ KX + Na+ 2NaX + Mg2+ ⇔ MgX2 + 2Na+ 2NaX + Ca2+ ⇔ CaX2 + 2Na+

1.47 1.78 1.95

M 1G ◦ ex Na

(kJ mol−1 )

−0.954 −1.428 −1.654

plotted in Fig. 5. Experimentally derived data points and activity coefficients fall close to the ideal behavior, where activity equals mole fraction. This is especially true in the case of Na/Ca binary where the points stand above the 1 : 1 line. Small positive and negative deviations occur for Na/Mg and Na/K, although they are close to the experimental uncertainty. In practice, this means that activity corrections for the FEBEX exchanger will be of minor importance in computing its ion-exchange behavior. The thermodynamic equilibrium constant, ij K ex , for the three exchange reactions can be evaluated through integration of ij K V , in logarithmic form, over the complete range of equivalent fractions (Eq. [11]). From them, the Gibbs free energy of the exchange reaction, 1G ex , can be computed (Table 3). As expected, 1G ex clearly stresses the smaller affinity for Na by the smectite compared to the rest of cations. In fact, the affinity sequence appears to be Na+ < K+ < Mg2+ < Ca2+ . The alkalineearth cations are preferentially adsorbed over the ditrigonal cavities of the smectite compared to the alkaline cations (40), which form outer-sphere complexes. On the other hand, the affinity of smectite for cations with similar charge increases with decreasing ionic radius. The reason for this behavior lies in the hydration energy, 1HHyd , which is a major contributor to the energetic balance of the exchange reaction. As the ionic radius increases, 1HHyd decreases (41), which favors the adsorption of larger cations, as K+ with respect to Na+ (36). Combination of ij K ex data from Table 3 makes possible the calculation of additional exchange reactions, for instance, that of Ca and Mg exchange in the FEBEX bentonite, which is given Mg Ca by Ca Mg K ex = Na K ex /Na K ex = 1.09. This figure suggests that the higher selectivity for Ca than for Mg is rather small, as previously reported by Sposito et al. (11), who suggests a nonpreferential behavior with a Ca Mg K ex of 1.02. ACKNOWLEDGMENTS This research was supported with funds provided by the FEBEX Project (European Comission FI4W-CT95-0006), ENRESA (703227 and 703231), and Xunta de Galicia (XUGA 11802B98). The authors thank M. D. Mingorance for the analyses of Ca and Mg.

REFERENCES FIG. 5. Variation of the cation activity in the smectite as a function of the composition of the solid phase (molar fraction) for (a) Na/K exchange, (b) Na/Ca exchange, and (c) Na/Mg exchange. Bisector lines correspond to ideal behavior.

1. Inskeep, W. P., and Baham, J., Soil Sci. Soc. Am. J. 47, 660 (1983). 2. Fergusson, J. E., “The Heavy Elements: Chemistry, Environmental Impact and Health Effects.” Pergamon Press, Oxford, 1990. 3. Zachara, J. M., and McKinley, J. P., Aquat. Sci. 55, 250 (1993).

416

HUERTAS ET AL.

4. McKinley, J. P., Zachara, J. M., Smith, S. C., and Turner, G. D., Clays Clay Miner. 43, 586 (1996). 5. Turner, G. D., Zachara, J. M., McKinley, J. P., and Smith, S. C., Geochim. Cosmochim. Acta 60, 3399 (1996). 6. Gast, R. G., Soil Sci. Soc. Am. J. 33, 37 (1969). 7. Gast, R. G., Soil Sci. Soc. Am. J. 36, 14 (1972). 8. Bolt, G. H., Soils Sci. 79, 267 (1955). 9. Bruggenwert, M. G. M., and Kamphorst, A., in “Soil Chemistry. B. Physicochemical Models” (G. H. Bolt, Ed.), Vol. 5B, p. 141. Elsevier, Amsterdam, 1983. 10. Suarez, D. L., and Zahow, M. F., Soil Sci. Soc. Am. J. 53, 52 (1989). 11. Sposito, G., Holtzclaw, K. M., Charlet, L., Jouany, C., and Page, A. L., Soil Sci. Soc. Am. J. 47, 51 (1983). 12. Sposito, G., Holtzclaw, K. M., Jouany, C., and Charlet, L., Soil Sci. Soc. Am. J. 47, 917 (1983). 13. Sposito, G., Jouany, C., Holtzclaw, K. M., and LeVesque, C. S., Soil Sci. Soc. Am. J. 47, 1081 (1983). 14. Fletcher, P., Sposito, G., and LeVesque, C. S., Soil Sci. Soc. Am. J. 48, 1016 (1984). 15. Sposito, G., Soil Sci. Soc. Am. J. 55, 965 (1991). 16. Maes, A., and Cremer, A., in “Soil Chemistry. B. Physico-chemical Models” (G. H. Bolt, Ed.), Vol. 5B, p. 205. Elsevier, Amsterdam, 1983. 17. Sposito, G., “The Surface Chemistry of Soils.” Oxford Univ. Press, New York, (1984). 18. Krauskopf, K. B., “Radioactive Waste Disposal and Geology.” Chapman & Hall, London, 1991. 19. Chapman, N. A., and McKinley, I. G., “The Geological Disposal of Nuclear Waste.” Wiley, Chichester, 1989. 20. ENRESA, Eds., “FEBEX: Full-Scale Engineered Barriers Experiments in Crystalline Host Rock. Etapa Preoperacional. Informe de S´ıntesis,” Technical Report 09/97. ENRESA, Madrid, 1997. 21. ENRESA, Eds., “FEBEX Project: Full-Scale Engineered Barriers Experiment for a Deep Geological Repository for High Level Radioactive Waste in Crystalline Host Rock. Final report,” Technical Report 1/2000. ENRESA, Madrid, 2000. 22. Linares, J., Huertas, F., Lachica, M., and Reyes, E., Proc. Intl. Clay Conf. 1, 351 (1972).

23. Caballero, E., Fernandes Porto, M., Linares, J., Huertas, F., and Reyes, E., Estudios Geol. 39, 121 (1983). 24. Leone, G., Reyes, E., Cortecci, G., Pochini, A., and Linares, J., Clay Miner. 18, 227 (1983). 25. Fern´andez Soler, J. M., “El Volcanismo Calco-alcalino en el Parque Natural de Cabo de Gata-N´ıjar (Almer´ıa): Estudio Volcanol´ogico y Petrol´ogico.” Ph.D. dissertation, University of Granada, 1992. 26. Linares, J., Huertas, F., Reyes, E., Caballero, E., Barahona, E., Guardiola, J. L., Y´an˜ ez, J., Romero, E., and Delgado, A., “Investigaci´on de Bentonitas como Material de Sellado para Almacenamientos de Residuos Radiactivos de Alta Actividad.” Technical Report 01/93. ENRESA, Madrid, 1993. 27. Moore, D. M., and Reynolds, Jr., R. C., “X-Ray Diffraction and the Identification and Analysis of Clay Minerals.” Oxford Univ. Press, New York, 1997. 28. Shapiro, L., US Geol. Survey Bull. 1401, 76 (1975). 29. Soil Conservation Service, “Soil Survey Laboratory Methods and Procedure for Collecting Soil Samples.” USDA, Washington, DC, 1972. 30. Wolery, T. J., “EQ3NR, a Computer Program for Geochemical Aqueous Speciation–Solubility Calculations: Theoretical Manual, User’s Guide, and Related Documentation (Version 7.0).” URCL-MA-110662 PT III. Lawrence Livermore National Laboratory, 1992. 31. Vanselow, A. P., Soil Sci. 33, 95 (1932). 32. Gapon, E. N., J. Gen. Chem. 3, 144 (1933). 33. Gaines, G. L., and Thomas, H. C., J. Chem. Phys. 21, 714 (1953). 34. Appelo, C. A. J., and Postma, D., “Geochemistry, Groundwater and Pollution.” Balkema, Rotterdam, 1993. 35. Fletcher, P., “Chemical Thermodynamics for Earth Scientists.” Longman, Burt Mill, 1993. 36. Sposito, G., “The Thermodynamics of Soil Solutions.” Oxford Univ. Press, New York, 1981. 37. Huertas, F. J., Cuadros, J., and Linares, J., Appl. Geochem. 10, 347 (1995). 38. Shainberg, I., and Kaiserman, A., Soil Sci. Soc. Am. J. 33, 547 (1969). 39. Verburg K., and Baveye P., Clays Clays Miner. 42, 207 (1994). 40. McBridge, M. B., “Environmental Chemistry of Soils.” Oxford Univ. Press, New York, 1994. 41. Marcus, Y., “Ion Properties.” Decker, New York, 1997.