Journal of Colloid and Interface Science 230, 150156 (2000) doi:10.1006/jcis.2000.7003, available online at http://www.idealibrary.

com on

On the Applicability of DLVO Theory to the Prediction of Clay Colloids Stability

Tiziana Missana1 and Andr es Adell
Departamento de Impacto Ambiental de la Energ a, CIEMAT, Avenida Complutense, 22, Madrid, Spain E-mail: tiziana.missana@ciemat.es Received March 6, 2000; accepted May 30, 2000

The stability behavior of Namontmorillonite colloids has been studied by combining the analysis of their surface charge properties and time-resolved dynamic light scattering experiments. The chemical surface model for several types of clays, including montmorillonite, has to take into account the double surface charge contribution due to their permanent structural charge and to their pH-dependent charge, which is developed at the edge sites, therefore, these stability studies were carried out as a function of both ionic strength and pH. DLVO theory is largely applied for the prediction of the stability of many colloidal systems, including the natural ones. This work shows that the stability behavior of Namontmorillonite colloids cannot be satisfactorily reproduced by DLVO theory, using the surface parameters experimentally obtained. Particularly, this theory is unable to explain their pHdependent stability behavior caused by the small charge at the edge sites. Based on these results, a literature review of DLVO stability prediction of clay colloids was performed. It conrmed that this theory is not capable of taking into account the double contribution to the total surface charge and, at the same time, pointed out the main uncertainties related to the appropriate use of the input parameters for the calculation as, for example, the Hamaker constant C 2000 Academic Press or the surface potential. Key Words: montmorillonite; DLVO theory; surface charge; dynamic light scattering.

INTRODUCTION

Stability behavior of clays is of great importance in several industrial processes as well as in soil chemistry and environmental science. Bentonite clay is a relevant material in the waste management programs of many countries. Due to its high sorption capability, it is able to retain relevant contaminants (1) and it is, thus, considered as a potential backll material in high-level radioactive waste deep geological repositories. It has recently been shown that bentonite colloids can be generated at the host rock/clay interface of such repositories (2). Colloids with high sorption capability may constitute an additional mechanism for radionuclide migration toward the biosphere,
1

To whom correspondence should be addressed. 150

provided they are stable. Consequently, the stability behavior of these colloids has a key role in the repository performance assessment. Montmorillonite, which is the main component of bentonite, is a 2 : 1 layer (2 tetrahedral and 1 octahedral) clay. The main contribution to the surface charge in these clay minerals is permanent and it is due to isomorphic substitutions within their layer structure. In these layers, atoms of lower valence usually replace atoms with higher valence, leading to an excess of negative charge, which is compensated by the presence of cations near the surface. At the edge of the clay particles, where the primary bonds of the tetrahedral and octahedral layer are broken, and surface functional groups (SOH) similar to those present on the surface of an alumina/silica particle exist (3), an electric double layer is formed by the adsorption/desorption of hydroxyl ions (4, 5). This produces an additional pH-dependent contribution to the surface charge. Since the stability behavior of a colloidal suspension is closely related to the surface charge properties of the suspended solid, the contribution of both the permanent and pH-dependent charge should be considered when stability predictions for clay materials are carried out. Not too much work has been published on the prediction of the stability of bentonite/montmorillonite colloids. In addition, even if the experimental data presented in the published papers were quite similar, the predictions (using DLVO theory) often show contradictory aspects. The aim of this work is to relate the stability behavior of Namontmorillonite colloids to the surface properties of the clay and to discuss to what extent the classical DLVO theory can be applied to the prediction of the stability of clay colloids. Prior to the stability experiments, the surface charge properties were experimentally determined by means of acidbase potentiometric titrations and electrokinetic measurements. Surface potentials, which are needed for the application of DLVO theory, were calculated using the experimental data and the GouyChapman equation. The coagulation kinetics was studied by means of the timeresolved dynamic light scattering technique (6), which was recently demonstrated as being a valuable tool for the study of the coagulation kinetics of clays (7).

0021-9797/00 $35.00
C 2000 by Academic Press Copyright All rights of reproduction in any form reserved.

DLVO THEORY AND CLAY COLLOIDS STABILITY

151

The results showed that the small pH-dependent charge, developed at the edge of clay colloids, greatly inuence their stability behavior and that the DLVO theory is not particularly suitable to predict clay colloid stability, mainly because it is not able to take into account this charge contribution. A discussion on the appropriate use of the input parameters needed for DLVO calculation was also carried out.
EXPERIMENTAL AND METHODS

1 10 3 M). The suspensions were stirred and purged with N2 prior to titrations, which were also carried out under a N2 atmosphere to minimize the effects of CO2 . A reference solution was obtained by ltering the initial suspension through a 0.01 m lter membrane and it was titrated by the same procedure used in the colloidal suspension. The charge Q H was estimated by subtracting the titration curve of the reference solution from the titration curve of the clay suspension at the corresponding ionic strength (11, 15). In fact, it can be easily demonstrated that QH = suspension suspension 1 reference reference Ca Ca Cb Cb S + Y suspension Y reference ,

Suspension Preparation The material used in this work is a Spanish clay, commonly called FEBEX (8) bentonite. FEBEX bentonite is a clay with a smectite content greater than 90% (93 2%), with quartz (2 1%), plagioclase (3 1%), cristobalite (2 1%), potassic feldspar, calcite, and trydimite as accessory minerals. The main characteristics of this clay are summarized in Table 1 and further details on its physico-chemical characteristics can be found elsewhere (9). The clay was puried and converted to its Na form following the method proposed by Bayens and Bradbury (10) and the colloidal fraction was obtained by centrifuging. Dialysis bags were used for conditioning the suspension at different ionic strengths, with NaClO4 as the electrolyte. The initial pH of the suspension was 7.0 0.5. The concentration of the colloidal material was determined by weighing 100 ml of the suspension, after drying by oven (at 105110C for 48 h). The correction needed due to the presence of the electrolyte salts in the suspension was taken into account. AcidBase Potentiometric Titrations The pH-dependent surface charge, due to the adsorption/desorption of hydroxyl ions ( Q H ), is usually determined, both in oxy-hydroxides (11) and in clays (10, 12), by acidbase potentiometric titrations. Different methods for determining the surface charge from titration experiments have been proposed (13), some of them being more reliable if the material is appreciably soluble. In a separate work, titration experiments onto FEBEX bentonite and the comparison of the results obtained with different methods are discussed (14). In this work, the results of a fast titration method were used. Fast titrations were carried out by performing continuous additions of NaOH or HCl (5 l) every 3 min to Na montmorillonite suspensions (1.92 0.2 g/L) conditioned with NaClO4 at three different ionic strengths (1 10 1 , 1 10 2 ,
TABLE 1 Summary of the Main Physico-Chemical Characteristics of FEBEX Bentonite
Total surface area External surface area (BET) Specic weight CEC Radius of bentonite colloids 725 m2 /g 33 m2 /g 2.7 g/cm3 103 mequiv/100 g 250 50 nm

where Ca and Cb are the quantity of acid and base (in mol L 1 ) added to the suspension or to the reference solution and S the solid concentration (in kg L 1 ). The term Y accounts for constituents that can consume H+ or OH , other than the surface functional groups SOH, including impurities or species from clay dissolution. The difference (Y suspension Y reference ) is assumed to be negligible because the reference solution is equilibrated and titrated the same way as the solid suspension (16), and the same amount of dissolved ions may be assumed to exist in both systems. Furthermore, the amount of solid dissolved is minimized if fast titrations are carried out. -Potential Measurements The same suspensions used for titration experiments were used for -potential measurements as a function of pH and ionic strength. These measurements were carried out by means of the laser Doppler electrophoresis technique using a Malvern Zetamaster apparatus equipped with a 5-mW HeNe laser ( = 633 nm). Particle Size Measurements and Time-Resolved Dynamic Light Scattering Experiments A batch of clay suspension (1.4 0.2 g/L) conditioned with NaClO4 at the ionic strength of 1 10 3 M was put into an ultrasonic bath 1 h before the experiment to obtain a well-dispersed system and was used as the starting sample. The coagulation kinetics was studied as a function of ionic strength and pH. The ionic strength was increased by adding aliquots of 1 M NaClO4 to the initial suspension. The changes in pH were obtained by adding NaOH and HCl (0.1 M). The particle size distribution was determined by photon correlation spectroscopy (PCS) by means of a 4700 Malvern apparatus equipped with a Spectra Physics 4-W argon laser ( = 514 nm). The photomultiplier was set to a scattering angle of 90 and the signal was transmitted to a Malvern 7032 digital autocorrelator. The average diffusion coefcient ( D ) was determined using the cumulant method (17) and the mean hydrodynamic radius of the particles (r ) was calculated using the

152 StokesEinstein equation, D= kT , 6 r

MISSANA AND ADELL

[1]

(k is the Boltzmann constant, T the absolute temperature, and the viscosity of the uid). The stability ratio, W , which represents the ratio between the total number of collisions and the number of collision that results in an aggregation process, can be expressed as W = kf , ks [2]

where kf is the rate of fast coagulation and ks is the rate of slow coagulation. The rate of fast coagulation refers to the coagulation that is limited only by the rate of diffusion, i.e., when the potential barrier does not exist. The stability ratio represents the effectiveness of the potential barrier in preventing particles from coagulation. The variation of the mean radius in a colloidal suspension depends on the concentration of the suspension, C , and its rate of coagulation, as follows: dr = ks FC . dt t !0 [3]

FIG. 1. Surface charge developed by the ionization of the SOH groups at the edges of the clay colloids in suspensions conditioned at (j) 1 10 1 , (d) 1 10 2 , and (m) 1 10 3 M of NaClO4 .

F is an optical factor depending on the scattering angle, the radius of the particle (r ), and the wavelength of the light (18). The stability ratio can be, therefore, experimentally determined by the expression dr f C t !0 W = dt dr C
dt t !0

[4]

and evaluation of (drh /dt )|t !0 from the initial slope of the curves that represent the evolution of the mean hydrodynamic diameter with time ( f refers to the fast coagulation regime). The mean hydrodynamic diameter of the particles in the initial suspension was 200250 nm. The evolution of this diameter was observed for approximately 1 h, by carrying out the single measurements every minute and using an experimental time of 60 s. The fast coagulation regime was clearly observed at a ionic strength of 0.1 M; however, rapid coagulation takes place in the original suspension (pH 7) at an ionic strength of approximately 1 10 2 M.
RESULTS

dependence on the ionic strength and a clear intersection point between the curves is not observed. Q H is positive when pH <6.57 and negative when pH >6.57. In the alkaline range, the degree of deprotonation increases with the ionic strength, which is in agreement with the behavior of oxy-hydroxides. At pH <7, the dependence on the ionic strength is anomalous, and this clearly indicates that another charge-determining mechanism has to be considered (H+ /Na+ exchange) (12). These results agree quite well, at least qualitatively, with the results of titration of different Namontmorillonites obtained by other authors (10, 12, 19). The concentration of the surface functional groups, SOH, was estimated with the FITEQL code (20), being approximately 5 meq/100 g. Considering that the total CEC, which approximately represents the magnitude of the permanent charge ( Q PER ), is 103 meq/100 g, the charge contribution of the functional groups SOH ( Q H ) is only 5% of the total charge. The total surface charge density of the clay particle, 0, is the sum of the structural charge density in the layers, PER , , that can be calculated from and the variable charge density, H Q H ( H = Q H F / A, where F is the Faraday constant and A the specic surface area). Therefore, an estimation of the total charge at the surface of the clay, as a function of pH, can be obtained by adding these two contributions. In the GouyChapman model, the charge in the diffuse layer is p ze90 [5] d = 0.1174 I sinh 2kT and 0+ d = 0,
2 where d is in C/m and 90 in V and where k is the Boltzmann constant, T the absolute temperature, e the electron charge, z the electrolyte valence, and I the solution ionic strength in mol/L.

Surface Charge and Potential Figure 1 shows the results obtained from acidbase titration experiments for the suspension at the three ionic strengths studied. The curves of Q H as a function of pH only show a small

DLVO THEORY AND CLAY COLLOIDS STABILITY

153

FIG. 2. Surface potentials calculated by means of the GouyChapman equation as a function of the pH and at different ionic strengths: (j) 1 10 1 (d) 1 10 2 , and (m) 1 10 3 M.

FIG. 4. strength.

-Potential curves for clay colloids as a function of the ionic

The GouyChapman equation was used to calculate the surface potential, 90 , from the experimentally determined surface charge. Figure 2 shows the surface potentials for bentonite colloids, calculated by Eq. [5] as a function of the pH and at different ionic strengths. The value of these potentials is quite high: approximately 105, 160, and 220 mV at the ionic strengths of 1 10 1 , 1 10 2 , and 1 10 3 M, respectively. Calculations with a more complex model than the GouyChapman theory, for a different Namontmorillonite but with similar CEC (96 meq/100 g), lead to surface potential values of the same order of magnitude and even higher (21). A feature that has to be remarked on is that the variation of the surface potentials with the pH is very small (only a few mV when pH varies from 4 to 10) at all the ionic strengths considered. Figure 3 shows the -potential as a function of pH of the suspension conditioned at the lower ionic strength (1 10 3 M).

As can be seen in the gure, Namontmorillonite colloids show fairly low negative potential (approximately 30 mV), almost constant in all the pH ranges considered. -Potential measurements account for both clay charge contributions; however, since the magnitude of the pH-dependent charge is very small compared to that of the permanent charge, its effects can only be slightly observed. Figure 4 shows the behavior of the -potential of the suspension when the ionic strength is changed. The experimental -potential is always much lower than the surface potential calculated with the GouyChapman theory and a clear dependence on the potential from the ionic strength cannot be seen. This behavior is quite uncommon in colloidal systems; therefore, reasonable doubt arises as to whether the experimental -potential may be used in any theoretical calculation on clay colloids instead of the calculated surface potential. Time-Resolved Coagulation Experiments Figures 5 and 6 show the results of the kinetic experiments carried out, as a function of the pH, with the suspension conditioned at 1 10 3 M and the stability ratio calculated with Eq. [4] from these experiments, respectively. As can be clearly observed, at a xed ionic strength, the stability behavior of the clay colloids strongly depends on the pH. The curves obtained for pH >6.5 (pH 9.5 and 10.5) are substantially at and no change in the mean size was observed during the experiment. The curve at pH 6.5 shows a small increase in the size during the experimental time and at pH 5 the destabilization of colloids is practically immediate. The curve at pH 3.5 is very similar to the curve obtained in the fast coagulation regime ( I = 0.1 M and pH 7), which has been included in the gure for comparison. A difference in 5 orders of magnitude exists between the stability ratio of the suspension at pH 4 and 10. As mentioned before, at ionic strengths higher than 10 2 M, even at basic pH, the suspension was unstable.

FIG. 3.

-Potential curves for clay colloids as a function of pH.

154

MISSANA AND ADELL

FIG. 5. Mean hydrodynamic diameter as a function of time for clay colloids at different pH in 1 10 3 M NaClO4 . The dashed curve corresponds to the curve obtained in the fast coagulation regime ( I = 0.1 M).

FIG. 7. DLVO total interaction energy curves obtained using the Gouy Chapman surface potential 9 at the ionic strength (r) 1 10 3 , (d) 1 10 2 , and (j) 1 10 1 M. The Hamaker constant is 6 10 20 J.

DLVO Stability Calculations The DLVO theory is widely used to calculate the balance between the repulsive and attractive forces that act when two particles approach each other. The main contributions to the interaction energy are the double-layer electrostatic energy (Vedl ) and the van der Waals forces (Vvdw ). The rst term is repulsive, as expected when two similar potential distributions interfere, while the second one is an attractive term. Another repulsive term, often less important than the previous two, is the shortrange Born energy (Vb ), which can be considered in the calculations of the total interaction energy. The total interaction energy of two identical spherical particles of radius (r ) (Vedl ), as a function of the distance of closest approach, is (22)
2 ln[1 + exp( h )], Vedl = 2r ""0 90

where 90 is the surface potential and the product " "0 the relative permittivity of water. The van der Waals energy (Vvdw ) between two identical spherical particles can be written as (23) h (4r + h ) AH 2r 2 ln + Vvdw = 6 (2r + h )2 h (4r + h ) 2 2r , [7] + (2r + h )2 where AH is the effective Hamaker constant. The total interaction energy of two bentonite colloids, calculated from Eqs. [6] and [7] and using the surface potentials of Fig. 3 (considered at pH 7), is shown in Fig. 7. The Hamaker constant considered for these calculations is 6 10 20 J. Using these potentials, a potential barrier high enough to prevent coagulation would exist at all the ionic strengths considered. This prediction clearly contradicts the previous experimental results obtained: in fact, at pH 7 rapid coagulation was observed, at least at the two higher ionic strengths. At the ionic strength 1 10 3 M, colloid stability was observed, but only at pH >7. The use of the DLVO theory for stability calculations at lower pH implies the use of practically the same input potential (approximately 220 mV, Fig. 2), and the predictions would be again in contradiction with the experimental results.
DISCUSSION

[6]

FIG. 6. Stability ratio as a function of the pH calculated from the curves plotted in Fig. 5.

The prediction of the stability of Namontmorillonite colloids with DLVO theory has been observed to be far from their actual behavior. For the discussion of the above results, the input parameters needed for DLVO theory calculations will be analyzed and, at this scope, different works published by other authors on

DLVO THEORY AND CLAY COLLOIDS STABILITY

155

clay colloids stability will also be considered. A very important parameter used in DLVO calculation is the surface potential. This parameter can be estimated using different theoretical models, but in many cases, the experimental -potential is assimilated to the surface potential. The increase of the ionic strength in a suspension causes the contraction of the double layer and a decrease of the surface potential. Indeed, in other colloidal systems, the -potential is observed to decrease when increasing the ionic strength, which is in agreement with the double-layer theory. In this case, the use of the -potential as the surface potential might be considered an acceptable approach. In the case of these clay colloids, there is not any evidence of a decrease in potential when the ionic strength is increased, as can be seen in our experimental results (Fig. 4) and as that already reported in the literature (24, 25). In addition, the differences between calculated potentials and experimental -potentials are, in general, quite large. In spite of the fact that, in this case, the use of the -potential as the surface potential seems to be a very rough approximation, it has been used for clay colloids stability predictions. For example, recent works on bentonite colloids, based on DLVO theory, predicted their occulation in a 1 : 1 electrolyte concentration of 1 10 3 M (26, 27). In these works, the -potential (between 10 and 30 mV) was used in the calculations with a Hamaker constant of 1 10 19 J. Even if their experimental values are not so far from ours, their predictions are totally different from those we presented in Fig. 7. The fact that they use the experimental -potential as the surface potential instead of calculated surface potential values (which differ by about 1 order of magnitude) could obviously be responsible for the totally different stability predictions (23). The other input parameter in DLVO formulation is the Hamaker constant. The van der Waals attractive term in the expression of the total interaction energy depends on the Hamaker constant (Eq. [7]); the higher this constant, the higher the attractive term. The direct measurement of this constant is extremely difcult and its determination has been the object of considerable research efforts. A difference in an order of magnitude in this constant could lead to completely different results in the stability predictions. In spite of this fact, different authors use different values for the Hamaker constant, without providing explanation for their specic choice. Generally, the values used for natural colloids are those estimated for silica colloids. Other authors observed that coagulation experiments with clay colloids could be satisfactorily reproduced by means of the classical DLVO theory using rather high values of the Hamaker constant and reasonable potential values (without specifying the values) (28). The reason why rather high values of the Hamaker constant should be assumed can be explained, for example, by considering our previous experimental results. Due to the pH-dependent stability behavior, we cannot actually establish an electrolyte critical coagulation concentration (CCC) unless a pH value is xed. Anyway, we can assume a range in which coagula-

tion is observed to start in our experiments, i.e., 1 10 3 to 1 10 2 M. The expression given for the CCC is (4) 90 4 ! tanh ze 4kT 36 CCC = 8 10 . [8] H 2 z6 If all the known parameters in Eq. [8] are substituted, a value for the Hamaker constant between 1 10 17 and 1 10 18 J is obtained, which is essentially higher than the values normally considered for colloidal suspensions (1 10 19 to 1 10 20 J). On the other hand, published values for the Hamaker constant of montmorillonite in aqueous solution based on coagulation experiments (29) or theoretical calculations (30) are 2.72 10 20 and 2.25 10 20 J, respectively, which are even smaller than the values reported for a silica colloid (47 10 20 J) (23) and the value that has been used in this work (6 10 20 J). The fact that a very high value of the Hamaker constant would be necessary to t the experimental results is probably due to the need of accounting for the existence of attractive forces different from the van der Waals forces, which are not considered in DLVO theory. For example, DLVO theory does not take into account short-range interactions such as hydration energy and swelling energy of this kind of clays, which might have an attractive component (31, 32). Apart from the uncertainties related to these input parameters, the main aspect in which DLVO theory fails in predicting clay colloid stability is related to the pH-dependent effects. The pH dependence of the clay stability is an expected phenomenon, which is due to the small charge developed at the edge of the clay particles. This charge is positive at pH <6.5, as shown by means of potentiometric titrations (Fig. 1). This means that in these clay colloids two different double layers exist that, in certain conditions, are of opposite signs. Negatively charged faces and positively charged edges can produce edgeto-Face (EF) aggregation effects that are responsible for lower stability at pH <6.5 and enhanced stability at higher pH. This well-known effect has been reported, for example, for kaolinite (7) and Hmontmorillonite (33) colloidal suspensions. In the study of Hmontmorillonite, the authors experimentally nd a signicant difference in the stability by varying the pH. They performed DLVO calculations trying to theoretically justify this behavior. The experimental coagulation values that they found at pH 2, 4, and 8 were approximately 100, 200250, and 350400 mmol/dm3 of a monovalent electrolyte. At the ionic strength of 100 mmol/dm3 , the heights of the peaks of the total interaction energy that they calculated with the DLVO theory were 0.14 mJ/m2 at pH 2, 0.8 mJ/m2 at pH 4, and 1.5 mJ/m2 at pH 8. The authors considered that these differences in the potential barrier were enough to justify their experimental behavior. Besides the fact that all these values would be 15 kT (considering a particles size of 100 nm approximately), even at pH 2, indicating stability in all the cases, the potentials that they calculated, at this ionic strength at pH 4

156

MISSANA AND ADELL

and 2, only differed by about 15 mV. It seems quite unlikely that with such a small difference the theoretical predictions would be signicantly different. This fact has already been mentioned in relation to our experimental results. Considering the results presented in Figs. 2 and 3 (9 and , respectively), the difference in the potential values observed between pH 4 and pH 10 at a ionic strength of 10 3 M is not higher than 10 mV. As mentioned before, very different behaviors in the DLVO curves are not observed if two potential values that differ by only 10 mV are used in DLVO equations. However, completely different behaviors are observed in the stability of the clay suspension at pH 4 and 10. Surface potentials of montmorillonite colloids as well as of other clay materials that mainly carry structural negative charge always show a very small dependence on the pH. This is logical, considering that the edge charge represents only a small fraction of the whole charge. The same behavior is also observed for -potentials. The existence of this edge charge distribution, which is very small in absolute values with respect to the total charge, cannot be taken into account by the DLVO theory.
CONCLUSIONS

REFERENCES
1. Horseman, S. T., Higgo, J. J. W., Alexander, J., and Harrington, J. F., Water, gas and solute movement through argillaceous media. NEA Report CC96/1. OECD, Paris, 1996. 2. Missana, T., Turrero, M. J., and Melon, A., Mater. Res. Soc. Symp. Proc. 556, 647 (1999). 3. van Holphen, H., in An introduction to clay colloids chemistry. Wiley Interscience, New York, 1977. 4. Stumm, W., and Morgan, J. J., in Aquatic Chemistry. Wiley, New York, 1981. 5. Sposito, G., in The surface chemistry of soils. Oxford Univ. Press, Oxford, U.K., 1984. 6. Novich, B. E., and Ring, T. A., Clays Clay Miner. 32, 400 (1984). 7. Kretzschmar, R., Holthoff, H., and Sticker, H., J. Colloid Interface Sci. 202, 95 (1998). 8. The FEBEX (Full scale Enginereed Barrier.EXperiment) project consists of an in situ test, in natural conditions and at full scale, and a mock-up test at almost full scale and a series of a laboratory tests to complement the information from the two large scale tests. The aim of this project was to study the behavior of components in the near eld of a repository of high-level wastes in crystalline rock. 9. FEBEX bentonite: Origin Properties and fabrications of blocks. Enresa Technical Publication 05/98, Madrid, 1998. 10. Bayens, B., and Bradbury, M. H., PSI, Technical Report, Bericht n 95/10 Willingen, 1995. 11. Dzombak, D. A., and Morel, F. M. M., in Surface Complexation Modelling. Chap. 3, Wiley, New York, 1990. 12. Wanner, H., Albinsson, Y., Karland, O., Wieland, E., Wersin, P., and Charlet, L., Radiochim. Acta 66 / 67, 157 (1994). 13. Shultess, C. P., and Sparks, D. L., Soil Sci. Soc. Am. 50, 14061411 (1986). 14. Missana, T., Turrero, M. J., and Adell, A., Mater. Res. Soc. Symp. Proc. 605 (2000), in press. 15. Sposito, G., in Environmental Particles (J. Bufe and H. P. Van Leeuewen, Eds.), pp. 291314, Lewis, Boca Raton, FL, 1992. 16. Lu, W., and Smith, E. H., Geochim. Cosmochim. Acta 60(18), 3363 (1996). 17. High performance system for particle characterisation: PCS 4700 Malvern, User Manual. Worchester, (1990). 18. Einarson, M. B., and Berg, J. C., J. Colloid Interface Sci. 155, 165 (1993). 19. Avena, M. J., Cabrol, R., and de Pauli, C. P., Clays Clay Miner. 38(4), 356 (1990). 20. Herbelin, A. L., and Westall, J. C., FITEQLA computer program for determination of chemical equilibrium constants from experimental data. 1996. 21. Avena, M. J., and de Pauli, C. P., J. Colloid Interface Sci. 202, 195204 (1998). 22. Hogg, R., Healy, T., and Fuersternau, D., Trans. Faraday Soc. 69, 1638 (1966). 23. Hiemenz, P. C., and Rajagoplan, R., in Principles of colloids and surface chemistry, 3rd ed, Chap. 10, Dekker, New York, 1997. 24. Sondi, I., Biscan, J., and Pravdic, V., J. Colloid Interface Sci. 178, 514 (1996). 25. Hunter, R. J., in Zeta Potential in Colloid Science. Academic Press, London, 1981. 26. Kurosawa, S., Kato, H., Veta, S., Yokoyama, K., and Fujihara, H., Mater. Res. Soc. Symp. Proc. 556 (1998). 27. Kurosawa, S., Yui, M., and Yoshikawa, H., Mater. Res. Soc. Symp. Proc. 465, 963 (1997). 28. Vold, R. D., and Vold, M. J., in Colloid and Interface Chemistry. Chap. 9, AddisonWesley, Reading, MA, 1983. 29. Helmy, A. K., and Ferreiro, E. A., J. Electroanal. Chem. 57, 103 (1974). 30. Helmy, A. K., J. Colloid Interface Sci. 207, 128 (1998). 31. Norrish, K., Discuss Faraday Soc. 18, 120 (1954). 32. McBride, M. B., Clays Clay Miner. 45(4), 598 (1997). 33. Tombacz, E., Abraham, I., Gilde, M., and Szanto, F., Colloids Surf. 49, 75 (1990).

This work showed that the application of DLVO theory to the prediction of clay colloid stability implies a high degree of uncertainty. The experimental data presented in different papers on the stability or the surface charge properties of clay colloids are often quite similar, but the stability predictions (using DLVO theory) might show contradictory aspects, depending on the choice of the input parameters, particularly the Hamaker constant or the surface potential. The different uses of these parameters can lead to completely different predictions. Furthermore, this theory is not able to take into account the contribution of the pH-dependent charge to the stability behavior of the clay. At pH <6.5, the layers and the edges are of opposite signs and aggregation effects between edges and faces are responsible for lower stability. Since the absolute value of the charge developed at particle edges is much smaller than the permanent negative value, it can only induce a very small variation in the magnitude of the surface potential. In fact, at pH 4 and 10 surface potentials only differ by about 10 mV and this difference does not produce signicant differences in the DLVO total energy interaction curves, in disagreement with the experimental results. All the above-mentioned points indicate that this theory is not particularly suitable to predicting the stability of clay colloids since it can lead to erroneous prediction, which would bias the performance assessment study of a waste repository.
ACKNOWLEDGMENTS
This work has been partially funded by the CIEMAT-ENRESA association. The authors are very grateful to Berta de la Cruz for her assistance in the linguistic revision of the manuscript.