Surface protonation data of kaolinite reevaluation based on dissolution experiments

 

Journal of Colloid and Interface Science, 264, 67-75.

Ganor, Jiwchar *, Jordi Cama1, and Volker Metz2

 

Department of Geological and Environmental Sciences, Ben-Gurion University of the Negev,

P. O. Box 653, Beer-Sheva 84105, Israel.

 

E-mail ganor@bgumail.bgu.ac.il

WEB http://www.bgu.ac.il/geol/ganor/

 

1 Present address: Department of Environmental Geology

Institute of Earth Sciences "Jaume Almera", CSIC

Lluís Solé i Sabarís s/n, Barcelona 08028, Catalonia (Spain)

E-mail jcama@ija.csic.es

WEB http://www.ija.csic.es/cat/

 

2 Present address: Institut für Nukleare Entsorgung,

Forschungszentrum Karlsruhe (FZK-INE),

P. O. Box 3640, Karlsruhe 76021, Germany

E-mail volker@ine.fzk.de

 

 

 

Submitted to: Journal of Colloid and Interface Science, December, 2002.

Revised and re-submitted, February, 2002

 

 


ABSTRACT

The aim of the present study is to compare available surface titration curves of kaolinite, to explain the differences between them, and to constrain their interpretation based on predictions of surface protonation that emerged from dissolution experiments. Comparison of six surface titration curves obtained at 25C reveals significant discrepancies, both in the shape of the curves and in the pH of the point of zero net proton charge (pHPZNPC). Based on an analysis of the different sites available for adsorption on kaolinite surfaces we conclude that different kaolinite samples are expected to have similar pHPZNPC. Therefore, the major reason for the differences in the observed surface protonation is related to the different way in which the pHPZNPC was determined. To compare the titration curves, some of the curves were recalculated so that the proton surface concentrations of all the titration curves would be zero around pH=5. As a result, we obtained a good agreement between the titration curves. A prediction of the molar fraction of protonated sites was retrieved from modeling of kaolinite dissolution reaction and was compared to the protonation data obtained from surface titration. The model successfully predicts the surface protonation data of most of the surface titrations.

Key words: kaolinite, clay, surface-protonation, adsorption-isotherms, dissolution.

 


introduction

Knowledge of surface protonation is essential for any attempt to understand kinetics of mineral dissolution and for interpreting adsorption of other ions on mineral surfaces. As a result, many studies of surface titration curves of kaolinite have been published in the last decade in the geochemical literature (1-12). Although these titration curves are different from each other, no attempt was made to compare the curves and to explain the discrepancies between them. This lack of agreement between the isotherms makes questionable the usage of any of them for predicting surface protonation. Some of these titration curves were used to model the pH dependence of kaolinite dissolution rate (2, 13). In each of these studies, a surface titration was measured and fitted to a surface speciation model. Thereafter, the resulting surface speciation was used in order to interpret results of dissolution experiments.

Cama et al. (14) examined the effect of pH on kaolinite dissolution rate under far-from equilibrium conditions. They proposed a kinetic model that includes surface protonation terms. The aim of the present study is to compare available surface titration curves of kaolinite, to explain the differences between them, and to use the rate law proposed by Cama et al. (14) in order to constrain the interpretation of the titration curves.

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

     [1]       

where DCs is the change in surface concentration of protons (mol m-2), CA and CB 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 (m2). Part of the change in solution pH may be a result of proton consumption and release by other reactions such as mineral dissolution, ion exchange and speciation in solution. Therefore the mass balance calculation of the surface protonation must be corrected for these reactions, and as Huertas et al. (11) show, this correction is not negligible under low and high pH conditions.

Surface titration curves are commonly interpreted in terms of protonation and deprotonation of amphoteric surface sites (see for example 15, 16, 17, and references therein). The interpretation of surface titration curves usually involves several stages. 1) Choice of a surface complexation model comprising the selection of a protonation mechanism (1pK or 2pK approach), deciding how to deal with surface heterogeneity (single- or multi-site model), and choosing among the available electrical double layer models (constant capacitance model, diffuse layer model, basic Stern model, triple layer model etc.); 2) fitting of the model parameters arising from the first stage to the experimental data; 3) calculating surface speciation and surface charge based on 1) and 2). The possible adjustable parameters are the equilibrium constants, the total surface density of each of the sites and parameters related to the electrostatic corrections (e.g., the capacitance in the constant capacitance model). One of the major problems with many of the surface complexation models is that they contain many adjustable parameters and as a result different models may fit equilibrium data equally well, and different combinations of adjustable parameters yield curves that adequately describe the experimental data (18). As a result, in many cases one cannot use the modeling for estimating parameters with true physical meaning (19). In the present paper we use independent information on surface speciation that emerges from modeling of dissolution rate data to constrain the interpretation of the adsorption data.

DISCUSSION

Comparing surface titration curves of kaolinite

Comparison of literature surface titration curves

Figure 1a compares surface titration curves obtained at 25C by Wieland and Stumm (2), Brady et al. (4), Schroth and Sposito (5), Ward and Brady (8) and Huertas et al. (11). These studies were chosen for comparison for the following three reasons: 1) They present titration curves that were extended in the acidic region to pH lower than 3.5; 2) The data are presented in tables or in figures that are large enough so the data may be retrieved accurately; and 3) They took into account possible proton consumption and release by other reactions such as mineral dissolution. The experimental conditions of each titration are summarized in Table 1. The curve obtained by Huertas et al. (11) is similar to those obtained by Schroth and Sposito (5) but significantly different from the curves obtained by Brady et al. (4) and Ward and Brady (8) (Fig. 1a). The curve obtained by Wieland and Stumm (2) is different from all the other curves. The pH of the point of zero net proton charge (pHPZNPC) ranges in these titrations from 3.9 to 7.5. In order to examine the significance of the differences between the curves, one should know the uncertainties associated with the measurements of the titration curves. Huertas et al. (11) repeated some of the titration points and estimated that the error in the measurements is less than 10%. If this error reflects one standard deviation, then within 95% confidence limit each data point of Huertas et al. (11) is precise within 20% (error bars in Fig. 1a). Assuming that the precision associated with the data of Brady et al. (4) is similar, error bars of 20% were added to the data points of Brady et al., as well. According to the error bars in Fig. 1a, the differences between the titration curve of Brady et al. (4) and that of Huertas et al. (11) are significant. Part of these discrepancies might be explained by different assumptions regarding the correction procedure for protons released and consumed by other reactions. However, the existence of strong discrepancies under neutral pH conditions, in which kaolinite dissolution rate is very slow, indicates that the correction procedure is not the major reason for the differences in the titration curves.

The differences in the shape of the isotherms may be explained by differences in kaolinite samples, differences in sample impurities (as was suggested for oxide by 20), and/or differences in the composition and concentrations of the background electrolytes (Table 1). However, the similarity between the isotherm of sample KGa-1 using a background electrolyte of LiCl determined by Schroth and Sposito (5) and the isotherms of Twiggs County kaolinite using a background electrolyte of KClO4 determined by Huertas et al. (11) seems to indicate that the observed differences between the isotherms of sample KGa-1 using a background electrolyte of NaCl determined by Brady et al. (4) and the isotherms of the same sample determined by Huertas et al. (11) are not related to sample identity and purity nor to the difference in composition of the background electrolytes. Huertas et al. (11) showed that titration curves obtained using ionic strength of 0.001M, 0.01M and 0.1M are similar and therefore, the differences in the titration curves obtained by Huertas et al. (11) and Brady et al. (4) cannot be attributed to differences in concentrations of the background electrolyte either.

Explanation for the discrepancies between the adsorption isotherms

We suggest that the major reason for the differences in the observed surface protonation is due to the different ways in which the pHPZNPC was determined. 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 (15), i.e., the titration measures a relative change in surface concentration (as defined by the proton consumption function, Eq. [1]) and not the absolute concentrations (5). Therefore, the calculation of the absolute proton adsorption density is based on an assumption regarding the pHPZNPC. Reported values of pHPZNPC of kaolinite vary significantly and range from 3 to 7.5 (5). Based on several observations of previous studies, Wieland and Stumm (2) assume that the pHPZNPC of the edge surfaces of kaolinite is 7.5 and therefore they recalculate proton surface concentration to be equal to zero at pH 7.5. Schroth and Sposito (5) measured surface concentrations of both protons and ions of the background electrolyte solution (Li+ and Cl-) and the permanent surface charge density using the method of Chorover and Sposito (21). Using these measurements they evaluated the absolute proton surface charge and the pHPZNPC based on surface charge balance. They obtained pHPZNPC values of 5 and 5.4 for kaolinite reference samples KGa-1 and KGa-2, respectively. In contrast, Brady et al. (4), Ward and Brady (8) and Huertas et al. (11) did not reset their measured surface concentration. Therefore, their reference point is the initial surface concentration before adding the acid or the base. Brady et al. (4) used a starting suspension of 4 g kaolinite (KGa-1 from Washington County, Georgia) having a surface area of approximately 10 m2 g-1 in 50 ml background electrolyte solution of 0.1M NaCl, whereas Huertas et al. (11) used a suspension with a solid/solution ratio of 25 g l-1 composed of kaolinite (from Twiggs County, Georgia) having a surface area of about 8 m2 g-1 with background electrolyte solution of 0.001M to 0.1M KClO4. The surface area to volume ratio of the starting suspensions were 800 m2 l-1 and 200 m2 l-1 in the experiments of Brady et al. (4) and Huertas et al. (11), respectively (Table 1). As the mineral was introduced into the electrolyte solution, its surfaces exchange protons with the solution until they approach equilibrium. Even if we assume that the surface charge of the two kaolinites were the same before they were introduced into the solution, that the initial pHs of the two solutions were originally identical and that the equilibration was not influenced by the different composition of the electrolytes, the pH of the two solutions in equilibrium with the mineral surface should be different due to the differences in surface area to volume ratio. The change in solution pH would be higher in the suspension with the higher surface area to volume ratio. The measured surface protonation in each of these studies represents the change in surface protonation with respect to the pH of the solution after its initial equilibration with the kaolinite, i.e., this pH was de facto assumed to be the pHPZNPC. Moreover, the titration curve of Brady et al. (4) is composed of two independent titration curves, one below pH 3.9 and the other above pH 4.3. Each of these two curves started with an initial stage in which kaolinite suspension was equilibrated in a background electrolyte solution of 0.1M NaCl. After equilibration, the pH of one suspension was 3.9 and of the other 4.3. The first suspension was titrated with HCl and the other with NaOH. The measured surface protonation in each of these two curves represents the change in surface protonation with respect to the pH of the solution after its initial equilibration with the kaolinite (P. Brady, private communication). Accordingly, two different values of pH of point of zero charge (3.9 and 4.3) were reported by Brady et al. (4).

As each of the above isotherms was retrieved by measuring changes in surface protonation with respect to a different (assumed) point of reference, it is impossible to compare them without defining a common point of reference. In the present paper we compare the different titration curves, assuming that the observed differences in the pHPZNPC are mostly artifacts of the differences in points of reference, and the titration curves of all the kaolinite samples may be shifted so they would have similar pHPZNPC. The validity of this assumption is examined below.

The factors that control the pHPZNPC of kaolinite

Numerous authors studied the nature of the surface charge of kaolinite (2, 4, 5, 8, 11, 22-25). According to these studies the surface charge of kaolinite is the sum of a permanent (pH-independent) and a non-permanent (pH-dependent) surface charges. The kaolinite basal surfaces carry a negative permanent surface charge, which is attributed to substitution of Si4+ by Al3+ in the tetrahedral layer of the kaolinite and to isomorphic substitution in other clays that are present as contamination in the sample. Due to the small cation-exchange capacity of kaolinite, several models neglect the permanent negative charge on the tetrahedral basal plane (e.g., 4). The non-permanent (pH-dependent) surface charge of kaolinite is a results of some or all of the following processes: 1) protonation and deprotonation of aluminol (>AlOH) on the edge; 2) deprotonation of silanol (>SiOH) on the edge; and 3) protonation and deprotonation of basal plane hydroxyl groups that are coordinated to two underlying aluminum atoms (>Al2OH). The latter site is less reactive than the aluminol on the edge and it is protonated at lower pH. As a result, several studies presented models of kaolinite surface charge that did not include contributions from basal planes (e.g., 4, 23). Huertas et al. (11) suggested that most of the basal plane aluminum surface sites (>Al2OH) are not protonated nor deprotonated at the pH range of 3 to 9. At this pH range, the non-permanent surface charge of kaolinite is dominated by the surface charge of the edges, i.e., by the difference between the amount of protonated Al edges (>AlOH2+) and the amount of deprotonated Si edges (>SiO-). As all the reported values of pHPZNPC of kaolinite ranged from 3 to 7.5 (5), in which the basal aluminol site is not ionized, the absolute pHPZNPC should be the pH in which the concentration of >AlOH2+ at equilibrium with the solution is equal to that of >SiO-. As only the edge contributes to the charge at the pHPZNPC of kaolinite, changes in the surface area ratio of the edge to that of the basal plane would not affect the pHPZNPC. The ratio of total aluminol to total silanol edge sites is about constant in different kaolinite samples, and is mostly determined by the kaolinite structure and the electrostatic valence principle of Pauling. Therefore, changing the total edge surface area would likewise not affect the pHPZNPC. Hence, at equilibrium, different kaolinite samples are expected to have similar pHPZNPC, although small differences in pHPZNPC may result from differences in the density of ion substitution and defects on the edge surfaces. It is important to note that differences in the permanent charge of the different samples would affect the pH of the point of zero net charge (pHPZNC), but not the pH of the point of zero net proton charge (pHPZNPC). Our conclusion that different kaolinite samples have similar pHPZPNC is supported by the observations of Schroth and Sposito (5), who obtained similar pHPZNPC values (5 and 5.4) for samples KGa-1 and KGa-2. Difference in values of pHPZNPC of the same kaolinite sample (KGa-1) as obtained by Schroth and Sposito (5) (pHPZNPC =5) and Brady et al. (4) (pHPZNPC =4) is yet another indication that the wide range of literature pHPZNPC does not reflect differences between kaolinite samples.

Recalculating surface titration curves

The above discussion indicates that the observed differences in the pHPZNPC are mostly artifacts of the differences in points of reference. Therefore, in order to compare all the titration curves they should be shifted so they would have similar pHPZNPC. We recalculate the titrations of Wieland and Stumm (2), Brady et al. (4) and Ward and Brady (8), so the proton surface concentration equaled zero at pH=5, which is the pHPZNPC of the titration of sample KGa-1 that was measured by Schroth and Sposito (5) (Fig. 1b). This value of pHPZNPC is also similar to value (5.1) estimated from theoretical considerations by Sverjensky and Sahai (26) for both the constant capacitance model and the diffuse double layer model. The isotherms of Huertas et al. (11) and Schroth and Sposito (5) were not recalculated because their pHPZNPC is around 5. The recalculations of the curves were done by adding constant values of -9.10-7, 4.4.10-7, 7.10-7 and 5.2.10-7 mol m-2 to each data point of the titration curves of Wieland and Stumm (2), Brady et al. (4) at pH4.3, Brady et al. (4) at pH3.9, and Ward and Brady (8), respectively. As the adsorption isotherm of Brady et al. (4) is composed of two independent titration curves, a different value was added to the data points of each of the curves. The titration curve below pH 3.9 does not have data points around pH 5, and therefore the value added to each data points was selected so the first two data points would match the equivalent data points in the titration curve of the same kaolinite sample (KGa-1) that was measured by Schroth and Sposito (5).

Figure 1b shows a good agreement between most of the recalculated curves. The recalculated curve of Wieland and Stumm (2) is similar to the other curves above pH 4.5 but shows a strong deviation below it. We do not have any explanation why the curve of Wieland and Stumm (2) is different from all the other curves, and hence it will not be further discussed in the present paper. The other curves show similar trends as can also be seen in Fig. 1c, which is an enlargement of the pH range of 3 to 5. At this pH range, we have an independent prediction of the molar fraction of surface protonation that will be used below to constrain the interpretation of the adsorption isotherms.

Comparing predictions of surface protonation that emerged from dissolution experiments with measured adsorption isotherms

As the original titration curve of Ward and Brady (8) is similar to that of Brady et al. (4) and the curves of Schroth and Sposito (5) are not significantly different from that of Huertas et al. (11), the following discussion is mainly concentrating on the differences between the curve of Brady et al. (4) and that of Huertas et al. (11). These two curves were chosen to represent the other curves as they include more data points in the examined pH range. As we showed above, there are significant differences between the shape of the original titration curve of Brady et al. (4) and that of Huertas et al. (11). These differences disappear after correcting for the different way in which the pHPZNPC was determined. Despite these differences, both titration curves were successfully interpreted by fitting the experimental data to surface complexation models. Brady et al. (4) used a two sites (aluminol and silanol) and three reactions (a protonation reaction of the aluminol and deprotonation reactions of both sites) model. Huertas et al. (11), who extended the titration to a more acidic range, used a third site in their model (edge aluminol, basal aluminol and edge silanol) and 4 reactions (protonation reactions of the two aluminol sites and deprotonation reactions of an aluminol and a silanol sites). In both studies, the equilibrium constants and the total surface density of each of the sites were adjusted in order to fit the data to the models. Independently of these studies, Cama et al. (14) predicted the equilibrium constants of the protonated reactions based on results of kaolinite dissolution reaction at pH 0.5 to 4.5. These predictions are used below to constrain the interpretation of the adsorption isotherms, in the relevant pH range.

Cama et al. (14) examined the effect of pH on kaolinite dissolution rate. They proposed that two independent parallel reaction paths control the rate of kaolinite dissolution at pH range of 0.5 to 4.5. Each reaction path consists of fast adsorption of a proton on a different surface site followed by a slow hydrolysis step. Cama et al. (14) described the effect of temperature and pH on kaolinite dissolution under acidic conditions by:

     [2]       

where rr (mol m‑2) is the density of reactive surface sites on the mineral surface, A (s-1) is the pre-exponential factor, Ea (kcal mol-1) is the apparent activation energy, K0 (M-1) is a constant related to the entropy of the proton adsorption, DH0 (kcal mol-1) is the net enthalpy of adsorption, aH+ is the activity of protons in the solution, R (kcal K-1 mol-1) is the gas constant, T is the temperature (K) and the subscripts 1 and 2 refer to the first and second reaction paths, respectively. The parameters in the rate law were determined by fitting their entire data set of kaolinite dissolution rates, obtained at 25, 50 and 70C and in the pH range from 0.5 to 4.5, to Eq. [2] using a multiple non-linear regression. The resulting rate law is:

     [3]       

The first reaction path solely dominates the overall rate above pH 2.5, whereas the second mechanism controls it below pH 0.5. Between pH 0.5 and 2.5 the two reaction paths significantly influence the rate. As a result, the fitting to the second reaction path is based on very few data points, and therefore, the results of the fitting to the first reaction path is much more trustworthy than the fitting to the second reaction path . Due to the poor constraints of the fitting of the coefficients of the second reaction path, they are not usable for calculation of the activation energy and the adsorption enthalpy, and different combinations of the coefficients in the second term in Eq. [3] yield similar curves that suitably describe the experimental data (14).

A by-product of the fitting of the proposed model (eq. [3]) is that it predicts the molar fraction of protonated sites on each of the two surface sites. These predictions of the proposed model may be compared to protonation data obtained from surface titration. Such a comparison is not straightforward. Surface titration of kaolinite measures the net surface charge density (mol m-2), which results from a permanent (pH-independent) charge and from several reactions, occurring on a variety of crystallographically distinct sites (4, 11). The protonation of some of these sites may affect the kaolinite dissolution rate, whereas the contribution of other protonated sites to the overall dissolution rate may be insignificant. In contrast, the fitting of the proposed model predicts the molar fraction (and not the total concentration) of protonated sites that govern the dissolution rate under the examined pH range.

The proton surface density (Gi) on the two surface sites (S1 and S2) may be calculated by multiplying the predicted molar fraction of the site by its total surface density (qi). In the calculations we assume that the proton surface density of sites that are not influencing the rate is constant in the examined pH range and their sum equals G0. The term G0 may include contribution from the permanent charge of the mineral surface as well as the real surface charge at the pH that was arbitrarily assumed to be the pHPZNPC. The predicted total proton surface density is:

     [4]       

The coefficients G0, q1 and q2 may be obtained by fitting Eq. [4] to adsorption isotherm data using least squares. The thick solid line in Fig. 2a is the best-fit curve obtained for the surface titration of Huertas et al. (11) at 25C in the pH range of 2.2 to 5. The obtained coefficients are G0=‑4.10-8 mol m-2, q1=8.10-7 mol m-2 (0.5 sites nm-2), q2=1.10‑4 mol m-2 (60 sites nm-2) and the regression coefficient R2=0.985. The dotted and dashed lines in Fig. 2a are plots of the second and the third terms in Eq. [4], respectively, i.e., of the pH dependencies of G1 and G2. As there are few points in the surface protonation data of kaolinite samples KGa-1 and KGa-2 determined by Schroth and Sposito (5) we did not fit the model to these adsorption isotherms. However, Fig. 2a shows that the surface protonation data of kaolinite samples KGa-1 and KGa-2 determined by Schroth and Sposito (5) exhibit an excellent agreement with the best fit curve of the data of Huertas et al. (11), although they were not used for the curve fitting.

Based on crystallographic considerations, Sposito (27) estimated the density of reactive sites on phyllosilicate edges to be approximately 8 sites nm-2 of edge surface area. In order to determine the surface density of the reactive sites on the edge per total surface area, this figure should be multiplied by the percentage of the edge surface area. Sposito (27) estimated that the edge surface area of kaolinite is 8% of the total surface area, resulting in a total edge surface density of 0.6 sites nm-2 (28), which is in good agreement with the total surface density obtained for the first site (0.5 sites nm-2). Based on Scanning Force Microscopy measurements, Brady et al. (4) show that the edge surface area may be as high as 47% resulting in a total edge surface density of up to 3.8 sites nm-2. The total surface density obtained for the second site (60 sites nm-2) is therefore unrealistic, even if one considers adsorption on basal surface planes ( 4 sites nm2). As was discussed above, the parameters describing the second reaction path are not well constrained and therefore one cannot use them to predict the proton surface coverage. For example, it is possible to fit the model of Cama et al. (14) (Eq. [4]) to the dissolution rate data forcing the value of K02 to be equal to a wide range of values without significantly changing the quality of the fitting nor the parameters related to the fitting of the first reaction path. Thus it is possible to adjust the parameters of the second reaction path so one would gain a realistic surface density of less than 4 sites nm2 for the second site. The excellent fitting of the surface charge prediction of our proposed model to the independent surface charge measurements of Huertas et al. (11) and the reasonable (and well constrained) value of the total surface density of the first site strengthen the model proposed by Cama et al. (14), and indicates that the adsorption isotherm of Huertas et al. (11) may be used for interpretation of dissolution experiments. The predictions of the proposed model regarding the equilibrium constant of the deprotonation of the first site at 25C () and the total surface density (q1=8.10-7 mol m-2) of this site are similar to the equilibrium constant (4.8.10-5) and total surface density (7.5.10-7 mol m-2) of the aluminol edge site obtained by Huertas et al. (11) using a non electrostatic model.

As the isotherms of Brady et al. (4) and Ward and Brady (8) were conducted at pH > 3, only the constraints imposed by the first surface site may be examined, although close to pH 3 the isotherms may be somewhat affected by the second surface site. The solid and the doted lines in Fig. 2b are the best-fit curves of the first two terms (G0+G1) in Eq. [4] to the original surface titrations of Brady et al. (4) and Ward and Brady (8), respectively, in the pH range of 3.1 to 5. The obtained coefficients and the regression coefficient are summarized in Table 2. The model of Cama et al. (14) may be successfully fitted to the surface protonation data of Huertas et al. (11), Schroth and Sposito (5) and Ward and Brady (8) (Fig. 2a and b), with regression coefficients better than 0.97 (Table 2). In contrast, the proposed model does not predict the isotherm of Brady et al. (4) (Fig. 2b).

Fig. 2c shows fitting of the proposed model to the recalculated surface titrations of Brady et al. (4) and Ward and Brady (8) in the pH range of 3.1 to 5. The obtained coefficients and the regression coefficient are summarized in Table 2. The model of Cama et al. (14) may be successfully fitted to the recalculated surface protonation data of both Brady et al. (4) and Ward and Brady (8), with regression coefficients of 0.984 and 0.965, respectively (Table 2). As the original and the recalculated isotherms of Ward and Brady (8) differ only by a constant value of 5.2.10-7 mol m-2, they have the same shape, and therefore the fitting of the model to the original data (Fig. 2b) is essentially identical to the fitting to the recalculated data (Fig. 2c). As was discussed above, the original isotherm of Brady et al. (4) is composed of two independent titration curves. Therefore the recalculated isotherm, which was calculated by adding different values to the data points of each of these curves, is significantly different in shape from the original isotherm of Brady et al. (4). As a result the model of Cama et al. (14) may be successfully fitted to the recalculated surface protonation data of Brady et al. (4) (R2= 0.984, Fig. 2c), although the proposed model does not predict the original isotherm of Brady et al. (4) (R2= 0.77, Fig. 2b).

Summary and conclusions

Strong discrepancies were observed between six surface titration curves of kaolinite that were reexamined in the present study. These discrepancies are explained by the different way in which the pHPZNPC was determined. After correcting for the differences in the pHPZNPC a good agreement between the general shapes of most of these titration curves was obtained. It is important to note that calculation of surface potential is based on the absolute surface charge. Therefore, when the pHPZNPC is incorrectly assumed and the proton surface concentration is not recalculated according to the real pHPZNPC, an error would be introduced to the calculations of the surface potential and of the electrostatic correction factor.

Cama et al. (14) examined the effect of pH on kaolinite dissolution rate. They proposed that two independent parallel reaction paths control the rate of kaolinite dissolution at pH range of 0.5 to 4.5. A by-product of the fitting of the proposed model to dissolution rate data is that it predicts the molar fraction of protonated sites on each of the two surface sites. These predictions of the proposed model are compared to protonation data obtained from surface titration. The model of Cama et al. (14) successfully predicts the surface protonation data of Huertas et al. (11), Ward and Brady (8) and of Schroth and Sposito (5). As the shapes of the original and the recalculated isotherms of Ward and Brady (8) are identical, the first term of the model of Cama et al. (14) successfully predicts both the original and the recalculated surface protonation data. In contrast, the original isotherm of Brady et al. (4) is composed of two independent titration curves, and therefore the shape of the combined curve includes a false plateau between the pHPZNPC of the first curve (3.9) and that of the second curve (4.3) (Fig. 1a). This plateau disappears as the curve is recalculated using a sole pHPZNPC, and therefore the shape of the recalculated isotherm is significantly different from that of the original isotherm. The disagreement between the prediction of the model of Cama et al. (14) and the original isotherm of Brady et al. (4) (Fig. 2b) is due to the existence of this plateau. Therefore, the model of Cama et al. (14) predicts the recalculated surface protonation data of Brady et al. (4) but not its original isotherm.

The excellent fitting of the surface charge prediction of the model of Cama et al. (14) to the independent surface charge measurements of Huertas et al. (11) and the reasonable value of the total surface density of the first site strengthen the model, and indicate that the adsorption isotherm of Huertas et al. (11) may be used for interpretation of dissolution experiments. The parameters of the model may be adjusted in order to gain a reasonable surface density for the second site too. However, dissolution experiments under very acidic conditions (pH<0.5) are required in order to constrain this value.

The results of the present study support the conclusions of Huertas et al. (13) and Cama et al. (14) that kaolinite dissolution under acidic conditions is controlled by proton adsorption on two different surface sites. The first reaction site controls the overall dissolution rate at pH2.5, whereas the second site controls it below pH 0.5. Between pH 0.5 and 2.5 both reaction sites influence the rate. The total surface density obtained for the first site (0.5 sites nm-2) is similar to the density of reactive sites on kaolinite edges (0.6 sites nm-2). This observation supports the suggestion of Huertas et al. (13) that proton adsorption on the edges controls kaolinite dissolution at mildly acidic conditions (up to pH 3). Their suggestion that adsorption on the less reactive basal planes controls the rate under more acidic conditions is neither supported nor contradicted by the results of the present study, as the total surface density of the second site is not well constrained.


Acknowledgments. This research was supported by grant # ES-66-96 from the Israel Ministry of Energy and Infrastructure and by the Belfer Foundation for Energy and Environmental Research. We wish to express our gratitude to J. Huertas, P. Brady and J. Lützenkirchen for fruitful discussions and to E. Shani and E. Hammer for their technical assistance.

 


                                                               1                    Table 1: Experimental conditions used in measuring surface titration curves

 

NR = not reported, SA = total surface area, V = volume, I = ionic strength.

Values shown in italics are of titration curves that are not shown in Fig. 1.

                        2          Table 2: parameters obtained by fitting the predictions of the model of Cama et al. (14) to literature surface protonation data


FIGURE CAPTIONS

1                     Fig. 1 Surface proton density of kaolinite obtained by surface titration. (a) Comparison of titration curves at 25C. (b) Recalculation of the titration curves so the proton surface concentrations would be zero at pH=5 (see text). (c) An enlargement of the pH range of 3 to 5 of Fig. 1b. Source data: W&S 1992 - Wieland and Stumm (2); BCN 1996 - Brady et al. (4); HCW 1998 - Huertas et al. (11); W&B 1998 - Ward and Brady (8); S&S 1997 - Schroth and Sposito (5).

2                     Fig. 2 Comparing the prediction of the model proposed by Cama et al. (14) to surface protonation data. (a) Best-fit curve of Eq. [4] to surface titration curve. The thick solid line is a curve fit to the data of Huertas et al. (11). The dotted and dashed lines are plots of the second and the third terms of the fitting to the data of Huertas et al. (11); (b) best-fit curves of the first two terms (G0+G1) in Eq. [4] to the original surface titrations of Brady et al. (4) and Ward and Brady (8); (c) best-fit curves of the first two terms (G0+G1) in Eq. [4] to the recalculated surface titrations of Brady et al. (4) and Ward and Brady (8). Source data is as in Fig. 1.


 

references

1. Motta, M. M. and Miranda, C. F., Soil. Sci. Soc. Am. J. 53, 380-385, (1989).

2. Wieland, E. and Stumm, W., Geochim. Cosmochim. Acta 56, 3339-3355, (1992).

3. Fox, I. and Malati, M. A., Journal of chemical technology and biotechnology 57, 97-107, (1993).

4. Brady, P. V., Cygan, R. T. and Nagy, K. L., J. Colloid Interface Sci. 183, 356-364, (1996).

5. Schroth, B. L. and Sposito, G., Clays and Clay Min. 45, 85-91, (1997).

6. Blockhaus, F., Sequaris, J.-M., Narres, H. D. and Schwuger, M. J., J. Colloid Interface Sci. 186, 234-247, (1997).

7. Angove, M. J., Johnson, B. B. and Wells, J. D., Colloids Surfaces A 126, 137-147, (1997).

8. Ward, D. B. and Brady, P. V., Clays and Clay Min. 46, 453-465, (1998).

9. Weerasooriya, R., Wickramarathne, H. U. S. and Dharmagunawardhane, H. A., Colloids Surfaces A 144, 267-273, (1998).

10. Angove, M. J., Johnson, B. B. and Wells, J. D., J. Colloid Interface Sci. 204, 93-103, (1998).

11. Huertas, J. F., Chou, L. and Wollast, R., Geochim. Cosmochim. Acta 62, 417-431, (1998).

12. Ikhsan, J., Johnson, B. B. and Wells, J. D., J. Colloid Interface Sci. 217, 403-410, (1999).

13. Huertas, J. F., Chou, L. and Wollast, R., Geochim. Cosmochim. Acta 63, 3261-3275, (1999).

14. Cama, J., Metz, V. and Ganor, J., Geochim. Cosmochim. Acta 66, 3913-3926, (2002).

15. Davis, J. A. and Kent, D. B., in Reviews in Mineralogy, Mineralogical Society of America, Washington, (1990).

16. J. Lützenkirchen, B. Kienzler, in: Modelling in Natural Sciences, Springer-Verlag, New York, 2003

17. Parks, G. A., in Reviews in Mineralogy, Mineralogical Society of America, Washington, (1990).

18. Westall, J. C. and Hohl, H., Adv. Colloid Interface Sci. 12, 265-294, (1980).

19. Majone, M., Papini, M. P. and Rolle, E., J. Colloid Interface Sci. 179, 412-425, (1996).

20. Kosmulski, M., J. Colloid Interface Sci. 253, 77-87, (2002).

21. Chorover, J. and Sposito, G., in University of California at Berkeley, Berkeley, (1993).

22. Bolland, M. D. A., Posner, A. M. and Quirk, J. P., Clays and Clay Min. 28, 412-418, (1980).

23. Carroll-Webb, S. A. and Walther, J. V., Geochim. Cosmochim. Acta 52, 2609-2623, (1988).

24. Zhou, Z. and Gunter, W. D., Clays and Clay Min. 40, 365-368, (1992).

25. Xie, Z. and Walther, J. V., Geochim. Cosmochim. Acta 56, 3357-3363, (1992).

26. Sverjensky, D. A. and Sahai, N., Geochim. Cosmochim. Acta 60, 3773-3797, (1996).

27. Sposito, G., The surface chemistry of soils, Oxford University Press, New York, (1984).

28. Nagy, K. L., in Chemical Weathering Rates of Silicate Minerals, Mineralogical Society of America, Washington, (1995).