Geochimica et Cosmochimica Acta, Vol. 65, No. 21, pp. 3643–3655, 2001 Copyright © 2001 Elsevier Science Ltd Printed in the USA. All rights reserved 0016-7037/01 $20.00 ! .00 Pergamon PII S0016-7037(00)00709-8 Interpretation and prediction of triple-layer model capacitances and the structure of the oxide– electrolyte–water interface DIMITRI A. SVERJENSKY Morton K. Blaustein Department of Earth and Planetary Sciences, Johns Hopkins University, Baltimore, MD 21218, USA (Received November 9, 2000; accepted in revised form June 25, 2001) Abstract—The interpretation of mineral–water interactions in processes such as surface-charge development, adsorption of aqueous ions onto surfaces, and the kinetics of dissolution at the scale of the oxide– electrolyte– water interface has been greatly facilitated by models of the electric double layer. In models that explicitly account for adsorption of the electrolyte ions, a critical parameter is the integral electric capacitance that expresses the charge at the surface relative to the drop in electric potential at some distance away from the surface where the electrolyte ions adsorb. Despite the widespread application of such surface complexation models, much uncertainty surrounds the choice of values of the integral capacitance because it appears to depend on the specific oxide and type of electrolyte, yet it cannot be directly measured. In the present study, it is shown that triple-layer model capacitances (C1), obtained in a consistent manner from regression of surface-charge data referring to a wide range of ionic strengths, electrolyte types, and mineral surfaces, fall into two groups: on rutile, anatase, and magnetite, values of C1 increase with decreasing crystallographic radius of the electrolyte cation from Cs! to Li!; on quartz, amorphous silica, goethite, hematite, and alumina, values of C1 increase with decreasing hydrated electrolyte cation radius from Li! to Cs!. The triple-layer model capacitances on both groups of solids can be described by a model of the mineral–water interface with physically reasonable parameters consistent with X-ray standing-wave studies of the rutile–water interface (Fenter et al., 2000). The model specifies a layer of chemisorbed water molecules at the surface and a layer of adsorbed electrolyte cations. On rutile, anatase, and magnetite, the layer of chemisorbed water molecules is interpenetrated by the layer of electrolyte cations that adsorb close to the surface as dehydrated, inner-sphere complexes. On quartz, amorphous silica, goethite, hematite, and alumina, the layer of chemisorbed water molecules varies from 0.0 to as much as 6.0 Å and the electrolyte cations form hydrated, outer-sphere complexes. The model capacitances are consistent with interfacial dielectric constants ranging from 20 to 62. Triple-layer model capacitances can now be predicted for oxides in either alkali or alkaline earth electrolyte solutions that have not been studied experimentally. In addition, predictions can be made of the structure of the oxide– electrolyte–water interface for many oxides and electrolytes. Copyright © 2001 Elsevier Science Ltd plane of adsorption (!0) to the drop in potential at a distance " according to 1. INTRODUCTION Recent interpretations of mineral–water interactions at the scale of the oxide– electrolyte–water interface have employed models of the electric double layer to provide a quantitative understanding of processes such as surface-charge development, adsorption of aqueous ions onto mineral surfaces, and the kinetics of dissolution of oxides and silicates (Hiemstra et al., 1989a,b; Davis and Kent, 1990; Dzombak and Morel, 1990; Stumm and Wieland, 1990; Hayes et al., 1991; Goldberg, 1992; Dove, 1994; Katz and Hayes, 1995; Lützenkirchen et al., 1995; Crawford et al., 1996a,b,c; Hayes and Katz, 1996; Hiemstra and van Riemsdijk, 1996; Kosmulski, 1996, 1999; Ludwig et al., 1996; Sverjensky and Sahai, 1996; Venema et al., 1996a,b; Dove and Nix, 1997; Lützenkirchen, 1997, 1998; Robertson and Leckie, 1997; Sahai and Sverjensky, 1997a,b; Stumm, 1997; Felmy and Rustad, 1998; Criscenti and Sverjensky, 1999). Most such models contain an integral electrical capacitance parameter that expresses the charge at the surface relative to the drop in electrical potential at some distance away from the surface. For example, in the triple-layer model, the inner-layer capacitance (C1) relates the charge at the innermost C1 " !0/(#0 # #") (1) In Eqn. 1, #0 and #" refer to the potentials at the 0-plane (the innermost plane, where protons and hydroxyls adsorb) and the "-plane (where the electrolyte ions adsorb), respectively. Because the surface potentials in Eqn. 1 cannot be directly measured, the value of the capacitance C1 is critical to relate the potentials to the proton surface charge (!0). Despite the widespread application of surface complexation models of the oxide– electrolyte–water interface, much uncertainty surrounds the choice of values of the integral capacitance (Westall and Hohl, 1980; Hayes et al., 1991; Lützenkirchen, 1998, 1999). There are two main reasons for this. First, in contrast to the classical studies of the Hg-water interface (Bockris and Reddy, 1970), capacitances at the oxide– electrolyte–water interface cannot be directly measured (capacitances have been inferred from the combination of multiple kinds of experimental data: Sprycha, 1984; however, from surfacecharge data alone, capacitances are generally obtained as described below). Second, even though the capacitance can be interpreted by analogy with a parallel-plate capacitor (Davis et al., 1978), the capacitor parameters involved are also poorly *Author to whom correspondence should be addressed ([email protected]). 3643 3644 D. A. Sverjensky known. For example, when the 0-plane and the "-plane of the triple-layer model are treated as the plates of a capacitor, C1 " 8.854$int/" (2) where C1 has units of %F ! cm#2, &int refers to the interfacial dielectric constant of the water between the 0-plane and the "-plane, and " (Å) refers to the distance separating the two planes. Although it is likely that 6 $ &int $ 78 and that 2 $ " $ 4 Å, respectively (Hayes et al., 1991), the actual magnitudes of &int and " are not known, which has prevented the use of Eqn. 2 as a means of predicting capacitances. Recent progress in the application of X-ray standing waves to the mineral–water interface is quantifying interfacial distances relevant to electric double-layer models for the first time (e.g., Fenter et al., 2000). Capacitances have generally been obtained along with other surface complexation parameters by fitting experimentally derived surface-charge data. Where sufficient data are not available, it has been suggested that the capacitances be set to a value of 80 %F ! cm#2 (e.g., Hayes et al., 1991). However, it has also been demonstrated that capacitances at the oxide– electrolyte–water interface are a systematic function of the size of the electrolyte ions (Davis et al., 1978; Jang and Fuerstenau, 1986; Colic et al., 1991; Kallay et al., 1994; Sahai and Sverjensky, 1997b; Kitamura et al., 1999), and even that systematic trends with electrolyte ion size can be different on different oxide surfaces (Kallay et al., 1994). For example, on hematite (Colic et al., 1991), model capacitances increase with decreasing size of the hydrated ions. This trend is the same (although the changes are much larger) as the trends for electrolyte ion adsorption on Hg (Bockris and Reddy, 1970) and AgI (Lyklema and Overbeek, 1961), trends that have been termed lyotropic. However, on rutile, a systematic trend opposite to the lyotropic series has been inferred (Jang and Fuerstenau, 1986; Kallay et al., 1994). On rutile, model capacitances increase with increasing size of the hydrated ions. Such behavior is inconsistent with Eqn. 2. This inconsistency can be resolved by suggesting that the adsorbed electrolyte ions on rutile have crystallographic ionic radii. The latter would vary inversely with capacitance, as required by Eqn. 2. In turn, this suggests that the ions adsorbed on rutile behave as though they are dehydrated. Taken together, all of the above systematics suggest that model capacitances depend on the specific oxide and the type of electrolyte (Dumont et al., 1990; Kallay et al., 1994; Johnson et al., 1999a,b) and on the state of hydration of the electrolyte ions. In the present study, triple-layer model capacitances are interpreted taking into account dependencies on the type of oxide, the type of electrolyte, and the state of hydration of the electrolyte ions. The interpretation of model capacitances requires a set of capacitance values and associated double-layer parameters obtained with a single model in a consistent manner. Numerous capacitances for the oxide– electrolyte–water interface have been reported by fitting surface charge and/or electrokinetic data with a variety of electric double-layer models and a great variety of assumptions (e.g., Hayes et al., 1991). However, in most studies, only a small number of solids and electrolyte types have been investigated. As a consequence, the capacitance values from different studies are generally not comparable. In the investigation reported here, capacitances for Fig. 1. A diagrammatic representation of the proposed structure of the oxide– electrolyte–water interface. The distance of separation of the 0- and the "-planes is assumed to be made up of a layer of water molecules and a layer of electrolyte cations. Both layers contribute to the overall integral capacitance (C1) between the 0- and the "-planes. As a first approximation, it is assumed that the interfacial dielectric constant of water (&int) is the same for the two layers. a wide variety of oxides and electrolyte types are analyzed by building on and adding to results previously obtained by use of a consistent formulation of the triple-layer model (Sahai and Sverjensky, 1997a). A preliminary interpretation of triple-layer model capacitances resulted in a very approximate correlation for estimating values of C1 based on the properties of the electrolyte alone (Sahai and Sverjensky, 1997b) but did not specifically account for the type of oxide or the state of hydration of the adsorbed ions. The new interpretation of model capacitances presented below is facilitated by recent X-ray reflectivity studies of the rutile– electrolyte–water interface (Fenter et al., 2000). The latter provide fundamental constraints on distances of adsorbed species from mineral surfaces, which helps to ensure that the interpretation of model capacitances is consistent with physically reasonable values of the interfacial dielectric constant and the distance " (Eqn. 2). The interpretation of model capacitances presented below enables prediction of capacitances for systems that have not been studied experimentally. It also enables prediction of the distance " for the triple-layer model, which quantifies an important model parameter for the oxide– electrolyte–water interface— one that can be tested with further experimental X-ray studies. 2. MODEL FOR INNER-LAYER CAPACITANCES AT THE OXIDE–ELECTROLYTE–WATER INTERFACE According to the triple-layer model, ions at the oxide– electrolyte–water interface are specifically bound at two planes: the 0-plane and the "-plane (Fig. 1). These two planes are planes of uniform electric potential (#0 and #", respectively). The locations of these two planes relative to the surface of the oxide are not required for application of the model. However, in the Prediction of electric double-layer capacitances present study, the separation of the planes is of importance to interpret the capacitance. It should be noted that the surfaces of oxides are not expected to be planar. Instead, they will contain oxygens at different distances from a given reference plane (Koretsky et al., 1998). Because surface charge may arise from and be associated with these different types of oxygens, the 0-plane represents a planar approximation to the average contributions from the different surface species. The "-plane represents the locations of adsorbed electrolyte ions complexed to surface sites at the 0-plane (Davis et al., 1978). It is assumed here that the distance separating the 0- and "-planes and the overall capacitance (C1) is influenced by two factors: the size and state of hydration of the adsorbing electrolyte ions at the "-plane, and the presence of water molecules between the 0- and "-planes (Fig. 1). First, the size and the state of hydration of the adsorbing electrolyte ions can be expected to vary more strongly for the electrolyte cations than for the anions (Helgeson and Kirkham, 1976). This expectation is supported by recent systematic electrokinetic studies of a series of 1:1 electrolytes with a common cation or a common anion (Johnson et al., 1999a,b). These studies show that changes in the zeta potential are primarily associated with cation properties rather than anion properties. Consequently, as a first approximation, it is assumed that variation of C1 with electrolyte type is attributed to variations in the radius and hydration state of the electrolyte cation. The radius rM of the electrolyte cation in Figure 1 will represent either a hydrated cation radius or a crystallographic radius (i.e., a dehydrated cation), depending on the mineral. In the calculations described below, hydrated cation radii will be used for cations adsorbing onto those minerals behaving according to the lyotropic series (e.g., hematite). Crystallographic radii will be used for cations adsorbing onto those minerals behaving opposite to the lyotropic series (e.g., rutile). The choice of radii is discussed below. Second, the presence of water molecules between the 0- and "-planes and an explicit role for adsorbed water molecules is included in the present model. The presence of water molecules adsorbed to the surface is to be expected from considerations of the termination of the bulk crystal structure (Koretsky et al., 1998). Abundant infrared evidence also suggests that water molecules chemisorb to the surface of oxides and metals (Little, 1966; Koretsky et al., 1997). At the oxide– electrolyte–water interface, the details of the configuration of the water molecules and the cations are not known. However, on metal surfaces in water, the nature of the adsorbed water is established from experiments (Toney et al., 1994) and interpretations of measured capacitances (Bockris and Reddy, 1970). For example, in interpretations of the capacitances at the Hg-water interface (Bockris and Reddy, 1970), it is assumed that the distance " is made up of two layers: a layer of water molecules chemisorbed to the surface and a layer of electrolyte cations and anions. One possible configuration of the two layers is represented in Figure 1. Here, only one species in each layer has been depicted, an electrolyte cation with radius rM and a single water molecule with radius rH2O. It can be seen in Figure 1 that the cation is quite close to the surface of the solid—that is, that " $ 2rH2O ! rM. Other possible configurations might correspond to a much thicker layer of water and the situation where " % 2rH2O ! rM. Given the uncertainties surrounding the coordination state of the cations and the amount of water at the interface on oxides, 3645 it is convenient to define the distance parameter r1 (Fig. 1) where " " r1 ! rM (3) The parameter r1 expresses the distance of closest approach of the cation to the 0-plane, which is presumably related to the thickness of the layer of water, the orientations of the water molecules in the layer, and the coordination of the cation relative to the water molecules with which it is in contact. It is next assumed that the two layers depicted in Figure 1 can be treated as capacitors in a series. As a consequence, the capacitance C1 can be expressed (Bockris and Reddy, 1970) by ! " ! 2rH20 r1 ! rM # 2rH20 1 " ! C1 (8.854)$int,1 (8.854)$int,2 " (4) In Eqn. 4, the numerator of each term on the right-hand side of the equation represents the thickness of each layer. The denominator of each term contains the effective dielectric constant of water in each layer, &int,1 and &int,2. Although it might be expected that &int,1 and &int,2 would be different (e.g., Bockris and Reddy, 1970), independent values of such dielectric constants are not available. Consequently, as a first approximation, it is assumed in the present study that &int,1 " &int,2 " &int. This assumption eliminates the dependence in Eqn. 4 on the radius of the water molecules, which reduces the number of unknown parameters resulting in 1 rM r1 " ! C1 (8.854)$int (8.854)$int (5) Eqn. 5 provides a basis for regression of model capacitances (1/C1) in terms of the electrolyte cation size (rM). In the present study, the parameters &int and r1 are parameters to be determined by regression of the capacitances for a single mineral (or a small group of related minerals) and a wide range of electrolyte types. 3. APPLICATION TO TRIPLE-LAYER MODEL CAPACITANCES 3.1. Selection of Crystallographic and Hydrated Radii The application of Eqn. 5 requires radii for the adsorbing electrolyte cations. Crystallographic radii (rM,cr) for monovalent and divalent electrolyte cations (Shannon and Prewitt, 1969; Shock and Helgeson, 1988) are listed in Table 1. It should be noted that these radii refer specifically to sixfold coordination. The effects of coordination state on the radius of adsorbed cations at the oxide–water interface are as yet unknown. Hydrated radii (rM,hyd) listed in Table 1 for Li!, Na!, Be!!, Mg!!, Ca!!, Sr!!, and Ba!! were assumed equal to the Stokes radii computed from mobilities (Robinson and Stokes, 1959). For the cations K!, Rb!, Cs!, and NH! 4 , the hydrated radii were assumed to be equal to the crystallographic radii, reflecting the lack of a permanent hydration shell (Robinson and Stokes, 1959; Kitamura et al., 1999). Although other choices of radii could be made, the values in Table 1 permit a simple interpretation of triple-layer model capacitances, resulting in the integration of a wide variety of experimentally derived data. Refinement of the radii chosen here will no doubt 3646 D. A. Sverjensky Table 1. Crystallographic and hydrated radii (Å) for electrolyte cations. Ion rM.cra rM.hydb Li! Na! K! Rb! Cs! Tl! NH4! N(CH4)3! Be2! Mg2! Ca2! Sr2! Ba2! Ra2! 0.74 1.02 1.38 1.49 1.70 1.50 1.47c 3.47d 0.27e 0.72 1.00 1.16 1.36 1.43c 2.37 1.83 1.38 1.49 1.70 1.50 1.47 3.47 4.08 3.46 3.09 3.09 2.88 1.43 a Crystallographic radii for VI-fold coordination from Shannon and Prewitt (1969) unless otherwise stated. b Hydrated radii calculated from limiting equivalent conductivities (Robinson and Stokes, 1959, p. 126) or set equal to the crystallographic radius. c Shock and Helgeson (1988). d Estimated by Robinson and Stokes (1959, p. 122). e Radius for IV-fold coordination from Shannon and Prewitt (1969). result from further experiments involving X-ray reflectivity measurements. 3.2. Regression of Capacitances Model capacitances derived from regression of surface protonation data as a function of pH and ionic strength are given in Table 2. It should be emphasized that the capacitances refer to experimental studies of surface protonation from many investigators over a wide range of ionic strengths (Bolt, 1957; Bérubé and Bruyn, 1968a,b; Abendroth, 1970; Breeuwsma and Lyklema, 1973; Huang and Stumm, 1973; Yates, 1975; Riese, 1982; Blesa et al., 1984; Sprycha, 1984, 1989; Fokkink et al., 1987; Hayes, 1987; Liang, 1988; Hayes et al., 1991; Casey, 1994; Kallay et al., 1994; Lumsden and Evans, 1995; Huang, 1996; Venema et al., 1996a; Machesky et al., 1998; Criscenti and Sverjensky, 1999; Criscenti, 2000). The derivation of many of the capacitances given in Table 2 by an internally consistent method of regression of surface-charge data has already been described (Sahai and Sverjensky, 1997a,b; Criscenti and Sverjensky, 1999; Criscenti, 2000). Additional capacitances referring to rutile and silica in N(CH3)4Cl, rutile in LiCl, NaCl, and CsCl, and quartz in KNO3 and CaCl2 were included in the present study (Appendix) to extend the ranges of minerals and electrolyte types reported previously. It should be noted that the Table 2. Inner-layer capacitances (C1, %F.cm#2; taken from Sahai and Sverjensky, 1997a, unless otherwise stated) consistent with the triple-layer model derived from regression of surface-charge experiments referring to wide ranges of ionic strength. Electrolyte Solid C1 Source of surface charge data N(CH3)4Cl CsCl KNO3 KNO3 KNO3 NaCl NaClO4 NaNO3 NaCl LiCl LiNO3 LiCl LiCl NaCl CsCl KNO3 KNO3 CaCl2 NaCl KNO3 N(CH3)4Cl NaCl NaCl KCl NaCl NaNO3 NaNO3 NaCl NaNO3 NaCl '-TiO2 '-TiO2 '-TiO2 Fe3O4 '-TiO2 '-TiO2 '-TiO2 '-TiO2 "-TiO2 "-TiO2 '-TiO2 '-TiO2 Fe2O3 Fe2O3 Fe2O3 Fe2O3 Fe2O3 quartz quartz quartz am. SiO2 am. SiO2 am. SiO2 am. SiO2 FeOOHd FeOOHd FeOOHd (-Al2O3 '-Al2O3 (-Al2O3 55 95 110 120 125 120 125 130 130 140 145 155 80 90 90 95 95 81 100 105 50 95 100 120 60 60 60 90 100 110 Yates (1975)a Kallay et al. (1994)a Yates (1975)a Blesa et al. (1984) Fokkink et al. (1987) Machesky et al. (1998)a Bérubé and de Bruyn (1968b)a Bérubé and de Bruyn (1968a) Sprycha (1984) Sprycha (1984) Yates (1975) Kallay et al. (1994)a Breeuwsma and Lyklerna (1973) Liang (1988) Breeuwsma and Lyklema (1973) Yates (1975) Fokkink et al. (1987) Riese (1982)a Riese (1982) Huang (1996)a Casey (1994)a Casey (1994) Bolt (1957) Abendroth (1970) Lumsden and Evans (1995) Hayes (1987)b Venema et al. (1996a)c Huang and Stumm (1973) Hayes et al. (1991)b Sprycha (1989) a Fitted in the present study (Appendix). Criscenti and Sverjensky (1999). c Criscenti (2000). d All goethites included here have BET surface areas greater than 50 m2.g#1. b Prediction of electric double-layer capacitances 3647 1 rM,cr 0.75 " ! C1 (8.854)26.1 (8.854)26.1 Fig. 2. The inverse of the triple-layer model capacitances (1/C1) of rutile, anatase, and magnetite as a function of the crystallographic radius of the electrolyte cation (rM). capacitances obtained in this manner are linked to the equilibrium constants for cation and anion adsorption that were obtained during the regression analyses. Only surface-charge data referring to a wide range of ionic strengths were used. Such data enable separate evaluation of the capacitance and the electrolyte ion equilibrium constants. Consequently, when the site density for an oxide is known from tritium exchange experiments, and when the values of the surface protonation constants are obtained independently (e.g., Sverjensky and Sahai, 1996), a unique capacitance and pair of electrolyte equilibrium constants can be obtained (see also Hayes et al., 1991). 3.2.1. Rutile, Anatase, and Magnetite It can be seen in Table 2 that the first group of solids—rutile, anatase, and magnetite— have model capacitances that increase ! from 55 to 155 %F ! cm#2 in the series N(CH3)! $ 4 $ Cs ! ! ! K $ Na $ Li , which is in order of decreasing crystallographic radius of the electrolyte cation. As already noted above, this order is in agreement with the results of Kallay et al. (1994). It is the opposite of the lyotropic series. Because the variation in model capacitances reflects crystallographic radii, it seems reasonable to postulate as a first approximation that the monovalent cations are dehydrated when adsorbed to rutile, anatase and magnetite. In fact, dehydration of cations adsorbed on these minerals might be expected to be favored. The work required to remove waters of solvation from cations near mineral surfaces is smaller when the minerals have higher dielectric constants (James and Healy, 1972). Although the dielectric constant of anatase is not known, rutile and magnetite have such high dielectric constants that the calculated work of dehydration for cations near these minerals is negligible (James and Healy, 1972). Taking account of the trend with crystallographic radius in Table 2, the model capacitances for rutile, anatase, and magnetite have been regressed with Eqn. 5 in terms of the crystallographic radius (rM,cr) from Table 1 yielding the solid line in Figure 2 and the equation (6) It can be seen in Figure 2 that the model capacitances are clearly consistent with Eqn. 6 for rutile, anatase, and magnetite in a range of 1:1 electrolyte types. The discrepancies between calculated and experimentally derived model capacitances are mostly less than &10 %F ! cm#2 (Fig. 4). Furthermore, the model capacitances are consistent with physically reasonable parameters (&int and r1) for the oxide– electrolyte–water interface. The effective interfacial dielectric constant near the "-plane (&int " 26.1) is physically reasonable because it lies between the value of the dielectric constant at the interface (6) and the bulk water value of 78 (Bockris and Khan, 1993; Boily, 1999). Comparisons of the distances derived above from analyzing model capacitances can be made with the results of X-ray scattering studies for Rb and Sr on rutile (Fenter et al., 2000). Although triple-layer model capacitances for the adsorption of Rb! and Sr!! on rutile are not available experimentally, the results of the regression described above permit prediction of these capacitances and of the distance ". For example, from Eqns. 3 and 6 and the crystallographic radii in Table 1, it follows that values of " for Rb! on rutile can be calculated with " (Rb! on TiO2) " 0.75 ! rRb!,cr " 2.24 Å (7) This value of " " 2.24 Å is a predicted separation of the 0- and the "-planes in Figure 1. It also represents a minimum distance of the Rb! from the rutile surface. The total distance can be estimated by adding on the distance of the 0-plane from a surface reference plane. Following Fenter et al. (2000), the reference plane is selected to be the last Ti-O lattice plane parallel to (110). Although the precise location of the 0-plane relative to the reference plane is subject to uncertainty, crystal chemical considerations give rise to two different oxygen distances relative to the oxygens in the last Ti-O lattice plane (Koretsky et al., 1998): bridging oxygens (%Ti2O, at '1.3 Å), and oxygens on chemisorbed water attached to %O5Ti groups (i.e., %O5Ti(H2O) with H2O at '1.9 Å). Assuming that all of these oxygens could contribute to surface charge on the 0-plane (i.e., the oxygens at 0, 1.3, and 1.9 Å), the average distance of the 0-plane from the surface is '1.1 Å. Combining this estimate with the prediction of " given by Eqn. 7 results in an estimate for the distance of Rb! from the surface of rutile in a Rb-electrolyte solution: Rb! distance " 2.24 ! 1.1 " 3.3 Å. (8) ! X-ray studies have established that the Rb ion is 3.4 Å from the (110) surface of rutile in a Rb-electrolyte solution (Fenter et al., 2000), which agrees well with the estimate in Eqn. 8. Another check on the use of Eqns. 3 and 6 for prediction of values of " is provided by X-ray standing wave results for Sr!! adsorption on rutile in Na-salt solutions (Fenter et al., 2000). Assuming that Sr!! adsorbs on the "-plane and that the distance " is determined primarily by the dominant electrolyte cation in solution (i.e., Na!), it can be predicted with Eqn. 6 and the radius of the Na! ion (Table 1) that " (Na! on TiO2) " 0.75 ! 1.02 " 1.77 Å (9) for rutile in solutions of Na-salts. Combining this estimate with the estimate of the distance of the 0-plane used above (1.1 Å), results in an estimate for the distance of Na! or Sr!! from the surface of rutile in a Na-electrolyte solution given by Na! or Sr!! distance " 1.77 ! 1.1 " 2.9 Å (10) 3648 D. A. Sverjensky Fig. 3. The inverse of the triple-layer model capacitances (1/C1) of hematite, quartz, silica, goethite, and alumina as a function of the hydrated radius of the electrolyte cation (rM). The result in Eqn. 10 compares very favorably with the experimentally derived distance of 2.8 Å for the Sr!! ion adsorbed to the (110) surface of rutile in a Na!-electrolyte solution (Fenter et al., 2000). The above results strongly support the use of crystallographic radii (Table 1) as a first approximation to the sizes of monovalent and divalent cations adsorbed on rutile. The results suggest that the structures of the interfaces of rutile, anatase and Prediction of electric double-layer capacitances 3649 rutile (e.g., for hematite, quartz, and rutile, values of the dielectric constant are 25, 4.6, and 121, respectively; Sverjensky and Sahai, 1996). Consequently, hydration of cations adsorbed on these minerals might be favored because the work required to remove the waters of solvation near low dielectric constant mineral surfaces is substantial (James and Healy, 1972). Taking account of the hydrated radii (rM,hyd) in Table 1, the model capacitances for hematite, quartz, and amorphous silica have been regressed with Eqn. 5, yielding the solid lines in Figure 3 and the equations Hematite: 1 rM,hyd 3.5 " ! C1 (8.854)53 (8.854)53 (11) Quartz: rM,hyd 1 3.7 " ! C1 (8.854)62 (8.854)62 (12) rM,hyd 0.0 1 " ! C1 (8.854)19.7 (8.854)19.7 (13) Amorphous silica: Fig. 4. Comparison of triple-layer model capacitances derived by fitting experimental surface-charge data with theoretically calculated capacitances. It can be seen in Figure 3 that the model capacitances are clearly consistent with Eqns. 11 through 13 for hematite, quartz, and amorphous silica in a range of 1:1 and 2:1 electrolytes. The discrepancies between calculated and experimentally derived model capacitances are again mostly less than &10 %F ! cm#2 (Fig. 4), and the model capacitances are again consistent with physically reasonable parameters (&int and r1) for the oxide– electrolyte–water interface. However, no independent X-ray results are available yet to test these results. It can be noted from Eqns. 11 through 13 that the regression results for r1 for hematite and quartz (3.5 and 3.7 Å, respectively) are much larger than obtained above for rutile (0.75 Å). This result suggests that the hydrated cations are much further from the surfaces of hematite and quartz than the cations adsorbed to rutile. However, it should also be noted that in the case of quartz, this result is heavily dependent on the data point for CaCl2, which largely determines the slope of the calculated line in Figure 3. The much larger interfacial dielectric constants for hematite and quartz compared with rutile suggest that the water near these interfaces, on average, is less oriented and has a greater similarity to bulk water than the water bound to rutile. Capacitances for goethites and aluminas are also plotted in Figure 3. It can be seen in Figure 3 that the data for goethite and alumina are restricted to the Na-salts. Also, in the case of the goethites, only those samples that had reported BET surface areas greater than 50 m2 ! g#1 were used. Goethites analyzed in the present study with BET surface areas less than 50 m2 ! g#1 were not used seemed to have anomalously high surface charge and capacitance characteristics, a feature also noted by Hiemstra and van Riemsdijk (1991). The data in Figure 3 for goethite and aluminas serve only to define very preliminary values of the distance (r1). Assuming that the interfacial dielectric constant for these solids is the same as that for hematite, as a first approximation, results in the following equations: magnetite in electrolyte solutions are very similar. Because the distance r1 " 0.75 Å on all these surfaces is much less than the radius of a water molecule (1.4Å), the alkali cations are adsorbing very close to the 0-plane, and therefore very close to the surface. Even though they may be coordinated to one or more of the chemisorbed water molecules at the surface, the alkali ions are probably dehydrated compared with their state in aqueous solution. 3.2.2. Hematite, Quartz, Amorphous Silica, Goethite, and Alumina It can be seen in Table 2 that hematite, quartz, and amorphous silica have model capacitances that tend to increase for a series of cations such as Li $ Na $ K and N(CH3)4 $ Na $ K, which are in order of decreasing hydrated radius (Table 1). As already noted above, this order is in agreement with the lyotropic series for the Hg-water and AgI-water interfaces. Because the variation in model capacitances reflects hydrated cation radii, it seems reasonable to postulate as a first approximation that the monovalent cations are hydrated when adsorbed to hematite, quartz, and amorphous silica. All these minerals have dielectric constants much lower than those of Table 3. Predicted inner-layer capacitances (C1, %F.cm#2) consistent with the triple-layer model.a Solid '-TiO2 "-TiO2 Fe3O4 '-MnO2 Fe2O3 FeOOH '-Al2O3 (-Al2O3 Quartz am. SiO2 a Li! Na! K! Rb! Cs! NH4! N(CH4)3! Mg!! Ca!! Sr!! Ba!! 155 155 155 155 80 56 89 89 90 73 131 131 131 131 88 60 99 99 99 95 108 108 108 108 96 63 107 107 108 126 103 103 103 103 94 63 107 107 106 119 94 94 94 94 90 61 103 103 102 103 104 104 104 104 94 63 107 107 113 119 55 55 55 55 67 50 74 74 77 50 157 157 157 157 67 50 74 74 77 50 132 132 132 132 71 52 78 78 81 56 121 121 121 121 71 52 78 78 81 56 110 110 110 110 74 53 81 81 83 61 Calculated with Eqns. 6 and 11 through 15 and radii from Table 1. 3650 D. A. Sverjensky Table 4. Predicted distances " (Å) consistent with the triple-layer model.a Solid Li! Na! K! Rb! Cs! NH4!! N(CH4)3! Mg!! Ca!! Sr!! Ba!! '-TiO2 "-TiO2 Fe3O4 '-MnO2 Fe2O3 FeOOH '-Al2O3 (-Al2O3 '-SiO2 am. SiO2 1.5 1.5 1.5 1.5 5.9 8.4 5.3 5.3 6.1 2.4 1.8 1.8 1.8 1.8 5.3 7.8 4.7 4.7 5.5 1.8 2.1 2.1 2.1 2.1 4.9 7.4 4.3 4.3 5.1 1.4 2.2 2.2 2.2 2.2 5.0 7.5 4.4 4.4 5.2 1.5 2.5 2.5 2.5 2.5 5.2 7.7 4.6 4.6 5.4 1.7 2.2 2.2 2.2 2.2 5.0 7.5 4.4 4.4 5.2 1.5 4.2 4.2 4.2 4.2 7.0 9.5 6.4 6.4 7.2 3.5 1.5 1.5 1.5 1.5 7.0 9.5 6.4 6.4 7.2 3.5 1.8 1.8 1.8 1.8 6.6 9.1 6.0 6.0 6.8 3.1 1.9 1.9 1.9 1.9 6.6 9.1 6.0 6.0 6.8 3.1 2.1 2.1 2.1 2.1 6.4 8.9 5.8 5.8 6.6 2.9 a Calculated with Eqns. 16 through 21 and radii from Table 1. 1 rM,hyd 6.0 " ! C1 (8.854)53 (8.854)53 (14) rM,hyd 1 2.9 Aluminas: " ! C1 (8.854)53 (8.854)53 (15) Goethite: In Eqn. 15, it can be seen that the apparent value of r1 for goethite is larger than for any of the other minerals depicted in Figures 2 and 3, in part because the actual capacitance value for these goethites is so small (only 60 %F ! cm#2; Table 2). Additional experimental data for surfacecharge development on goethite, aluminas, and quartz in a range of different electrolytes are needed. " " 3.5 ! rM,hyd. (17) " " 3.7 ! rM,hyd. (18) " " 0.0 ! rM,hyd. (19) For quartz, For amorphous silica, For corundum and (-alumina, " " 2.9 ! rM,hyd. 4. PREDICTION OF TRIPLE-LAYER MODEL CAPACITANCES AND MODEL STRUCTURES OF OXIDE–ELECTROLYTE–WATER INTERFACES (20) For goethite (with BET surface areas greater than 50 m2!g#1), " " 6.0 ! rM,hyd. 4.1. Prediction of Capacitances The close agreement between the capacitances calculated with Eqn. 6 and Eqns. 11 through 15 and the experimentally derived capacitances can be seen more clearly in Figure 4. It should be emphasized that the latter are a set of model capacitances that were obtained by fitting surface-charge data over a wide range of ionic strengths. They refer to a specific set of assumptions detailed previously (Sahai and Sverjensky, 1997a,b) and are subject to uncertainties of at least &10 %F ! cm#2, which is of the same order as most of the discrepancies in Figure 4. Uncertainties of this order contribute uncertainties in calculations of surface charge of less than &2 %C ! cm#2 at an ionic strength of 0.1 mol/L and a pH '4 units from the zero point of charge. Overall, the close agreement between the calculated and experimentally derived capacitances suggests that Eqn. 6 and Eqns. 11 through 15 and the distances (r1) given in Figures 2 and 3 can be used to make predictions of triple-layer model capacitances for a wide range of electrolytes and solids. The results of such predictions are given in Table 3 for a selection of alkalis and other monovalent and divalent cations. Because most of the data used to calibrate Eqn. 6 and Eqns. 11 through 15 refer to monovalent cations, the predictions in Table 3 for these ions should be the most reliable. It must be emphasized that the predictions for divalent cations are preliminary values only. 4.2. Prediction of Model Structure of the Oxide–Water Interface The model developed in the present study for calculation of capacitances also has implications for the structure of the oxide– electrolyte– water interface in the context of the triple-layer model. In particular, calculation of the distance " for the triple-layer model can be made with Eqn. 3 because the distance r1 is known for a number of minerals of interest. The regression calculations depicted in Figures 2 and 3 suggest that the following relations for calculating " can be used: For rutile, anatase, and magnetite, " ) 0.75 * r M,cr. For hematite, (16) (21) Predicted values of ", established by use of Eqns. 16 through 21 are given in Table 4 for a wide variety of electrolytes. To illustrate the predicted differences between adsorption of dehydrated and hydrated ions, values of the alkali ion distances from the surfaces of rutile and quartz are depicted in Figure 5. It should be emphasized that the capacitance model presented above predicts only the distance " in the triple-layer model (i.e., r1 ! rM in Fig. 5). The alkali ions on both rutile and quartz in Figure 5 are at a total distance from the surface equal to the predicted values of " (" r1 ! rM, from Table 4) plus the distance of the 0-plane from a reference plane. In Figure 5, the surface reference plane is at 0 Å. For rutile, the 0-plane may be '1.1 Å from the reference plane (see above). For quartz, following the same crystal– chemical reasoning previously used for rutile, the 0-plane is depicted '0.4 Å from the reference plane (Fig. 5). The 0-plane location for rutile depicted in Figure 5 results in a calculated Rb! ion distance from the rutile surface very close to the experimental value (3.4 Å; Fenter et al., 2000). The rest of the results in Table 4 and Figure 5 represent model estimates that offer a more detailed picture of the oxide– electrolyte–water interface than has previously been offered within the context of the triple-layer model. For example, from the distances in Figure 5, it can be inferred that the Li! ion should be closer to the surface of rutile than the Cs! ion. However, for quartz, it is predicted that the Cs! ion will be much closer to the surface than the Li! ion. Predictions such as these can be tested with further X-ray standing-wave studies. 4.3. Implications for Surface Complexation of Metals at the Oxide–Electrolyte–Water Interface Taken together, the triple-layer capacitance model developed here and the results of X-ray standing-wave experiments imply that the alkali cations are bound very close to the surfaces of rutile, anatase, and magnetite. The capacitance model specifies a layer of chemisorbed water molecules at the surface and a layer of adsorbed alkali cations. Because the distance r1 " 0.75 Å (Fig. 1) on these surfaces is much less Prediction of electric double-layer capacitances 3651 adsorbed on the "-plane of these minerals will form inner-sphere complexes. The results of X-ray standing-wave and EXAFS studies of Sr2! adsorption on rutile (Fenter et al., 2000) are consistent with this suggestion. However, on minerals such as quartz, amorphous silica, and similar solids, trace amounts of Ca2! or Sr2! adsorbed on the "-plane should form outer-sphere complexes. This expectation is consistent with the results of X-ray standing-wave and EXAFS studies of Sr2! adsorption on silica and goethite (O’Day et al., 2000; Sahai et al., 2000) and hydrous ferric oxide (Axe et al., 1998). 5. CONCLUSIONS Fig. 5. Predicted distances of adsorption of alkali cations from the surfaces of rutile and quartz. Note all the distances are given relative to a mineral-surface reference plane, which is the last metal– oxygen plane of the solid. The distance of the 0-plane was estimated from crystalchemical considerations (see text). The position of the alkalis relative to the 0-plane (i.e., " " r1 ! rM) was calculated with the capacitance model summarized in the text. On rutile, the electrolyte cations are dehydrated, whereas on quartz they are hydrated. than the radius of a water molecule (1.4 Å), it appears that the layer of chemisorbed water molecules is interpenetrated by the layer of alkali cations. Even though the cations may be coordinated to one or more of the chemisorbed water molecules at the surface, they are dehydrated compared with their state in aqueous solution. This conclusion strongly suggests that the alkali cations form inner-sphere complexes with the surface functional groups and/or the chemisorbed water molecules on rutile, anatase and magnetite. In contrast, on minerals such as quartz, amorphous silica, goethite, hematite, and alumina, the alkali cations are probably hydrated, forming outer-sphere complexes. If the alkali cations of a background electrolyte determine the structure of the oxide–water interface by forming inner-sphere complexes on the "-plane of the triple-layer model for rutile and similar solids, it follows that trace amounts of other metals (e.g., Ca2!, Sr2!), if also (1) The triple-layer model capacitances (C1) for different mineral surfaces fall into two groups: on rutile, anatase, and magnetite, values of C1 increase with decreasing crystallographic radius from Cs! to Li!; on quartz, amorphous silica, goethite, hematite, and alumina, values of C1 increase with decreasing hydrated cation radius from Li! to Cs!. (2) The triple-layer model capacitances on both groups of solids can be described by a model of the interface with physically reasonable parameters. The model specifies a layer of chemisorbed water molecules at the surface and a layer of adsorbed electrolyte cations. On rutile, anatase, and magnetite, the layer of chemisorbed water molecules is interpenetrated by the layer of electrolyte cations. On quartz, amorphous silica, goethite, hematite, and alumina, the layer of chemisorbed water molecules varies in thickness. On both groups of minerals, the model capacitances are consistent with interfacial dielectric constants ranging from 20 to 62. (3) It is inferred from the present study, and from the X-ray standing wave results of Fenter et al. (2000), that the alkali and alkaline earth cations adsorb close to the surfaces of rutile, anatase, and magnetite as dehydrated inner-sphere complexes. In contrast, these cations adsorb further away from the surfaces of quartz, amorphous silica, goethite, hematite, and alumina as hydrated, outer-sphere complexes. (4) Triple-layer model capacitances can be predicted for systems that have not been studied experimentally—for example, for oxides immersed in all alkali and alkaline earth electrolyte solutions. Similarly, the distance " for the triple-layer model and estimates of the total distance from a surface reference plane can be predicted for many systems that have not been studied experimentally. It should be emphasized that the capacitances predicted in this article refer specifically to the triple-layer model of the oxide– electrolyte– water interface. It seems likely that a similar approach could be taken with other surface complexation models that explicitly treat electrolyte ion adsorption provided that an extensive reanalysis of all the surfacecharge data is undertaken. In addition, the predicted capacitances summarized above refer only to single oxide and electrolyte systems. When transition and heavy metals are also present in the system, it can be assumed, as a first approximation, that the capacitance of the oxide– electrolyte–water interface is unaffected (e.g., Criscenti and Sverjensky, 1999). Acknowledgments—Discussions with W. Casey, L. Criscenti, J. A. Davis, P. Fenter, M. Machesky, C. Koretsky, J. Kubicki, and N. Sahai helped the development of this work significantly, as did reviews by J. Dyer and D. Sparks. Financial support was provided by DuPont (through Noel Scrivner), NSF grant EAR 9526623, and DOE grant DE-FG02-96ER-14616. I also thank Harold C. Helgeson for his everinspiring example of how to do theoretical geochemistry, as well as his insistence on the importance of having lunch. Associate editor: D. L. Sparks REFERENCES Abendroth R. P. (1970) Behavior of a pyrogenic silica in simple electrolytes. J. Colloid Interface Sci. 34, 591–596. Axe L., Bunker G., Anderson P. R., and Tyson T. A. (1998) An EXAFS analysis of strontium at the hydrous ferric oxide surface. J. Colloid Interface Sci. 199, 44 –52. Bérubé Y. G. and Bruyn P. L. D. (1968a) Adsorption at the rutile– solution interface I. Thermodynamic and experimental study. J. Colloid Interface Sci. 28, 305–323. Bérubé Y. G. and Bruyn P. L. D. (1968b) Adsorption at the rutile– 3652 D. A. Sverjensky solution interface II. Model of the electrochemical double layer. J. Colloid Interface Sci. 28, 92–105. Blesa M. A., Figliolia N. M., Maroto A. J. G., and Regazzoni A. E. (1984) The influence of temperature on the interface magnetiteaqueous electrolyte solution. J. Colloid Interface Sci. 101, 410 – 418. Bockris J. O. and Reddy A. K. N. (1970) Modern Electrochemistry. Plenum Press. Bockris J. O. and Khan S. U. M. (1993) Surface Electrochemistry: A Molecular Level Approach. Plenum Press. Boily J.-F. (1999) The surface complexation of ions at the goethite('FeOOH)–water interface: A multisite complexation approach. Ph.D. thesis. Umeå University. Bolt G. H. (1957) Determination of the charge density of silica sols. J. Phys. Chem. 61, 1166 –1169. Breeuwsma A. and Lyklema J. (1973) Physical and chemical adsorption of ions in the electrical double layer on hematite ('-Fe2O3). J. Colloid Interface Sci. 43, 437– 448. Casey W. H. (1994) Enthalpy changes for Bronsted acid-base reactions on silica. J. Colloid Interface Sci. 163, 407– 419. Colic M., Fuerstenau D. W., Kallay N., and Matijevic E. (1991) Lyotropic effect in surface charge, electrokinetics, and coagulation of a hematite dispersion. Colloid Surfaces 59, 169 –185. Crawford R. J., Harding I. H., and Mainwaring D. E. (1996a) Adsorption and coprecipitation of multiple heavy metal ions onto the hydrated oxides of iron and chromium. Langmuir 9, 3057–3062. Crawford R. J., Harding I. H., and Mainwaring D. E. (1996b) Adsorption and coprecipitation of single heavy metal ions onto the hydrated oxides of iron and chromium. Langmuir 9, 3050 –3056. Crawford R. J., Harding I. H., and Mainwaring D. E. (1996c) The zeta potential of iron and chromium hydrous oxides during adsorption and coprecipitation of aqueous heavy metals. J. Colloid Interface Sci. 181, 561–570. Criscenti L. J. (2000) Metal adsorption onto oxides from aqueous solutions: The influence of the electrolyte, ionic strngth and surface coverage on surface complexation. Ph.D. thesis. Johns Hopkins University. Criscenti L. J. and Sverjensky D. A. (1999) The role of electrolyte # # 2! anions ClO# ) adsorption on 4 , NO3 , and Cl in divalent metal (M oxide and hydroxide surfaces in salt solutions. Am. J. Sci. 299, 828 – 899. Davis J. A., James R. O., and Leckie J. O. (1978) Surface ionization and complexation at the oxide/water interface I. Computation of electrical double layer properties in simple electrolytes. J. Colloid Interface Sci. 63, 480 – 499. Davis J. A. and Kent D. B. (1990) Surface complexation modeling in aqueous geochemistry. In Mineral–Water Interface Geochemistry (eds. M. F. Hochella Jr. and A. F. White), Vol. 23, pp. 177–259. Mineralogical Society of America. Dove P. M. (1994) The dissolution kinetics of quartz in sodium chloride solutions at 25°C to 300°C. Am. J. Sci. 294, 665–712. Dove P. M. and Nix C. J. (1997) The influence of the alkaline earth cations, magnesium, calcium, and barium on the dissolution kinetics of quartz. Geochim. Cosmochim. Acta 61, 3329 –3340. Dumont F., Warlus J., and Watillon A. (1990) Influence of the point of zero charge of titanium dioxides hydrosols on the ionic adsorption sequence. J. Colloid Interface Sci. 138, 543–554. Dzombak D. A. and Morel F. M. M. (1990) Surface Complexation Modeling.J. Wiley. Felmy A. R. and Rustad J. R. (1998) Molecular statics calculations of proton binding to goethite surfaces: Thermodynamic modeling of the surface charging and protonation of goethite in aqueous solution. Geochim. Cosmochim. Acta 62, 25–31. Fenter P., et al. (2000) Electrical double-layer structure at the rutile– water interface as observed in situ with small-period x-ray standing waves. J. Colloid Interface Sci. 225, 154 –165. Fokkink L. G. J., Keizer A. D., and Lyklema J. (1987) Specific ion adsorption on oxides: Surface charge adjustment and proton stoichiometry. J. Colloid Interface Sci. 118, 454 – 462. Goldberg S. (1992) Use of surface complexation models in soil chemical systems. Adv. Agron. 47, 233–329. Hayes K. F. (1987) Equilibrium, spectroscopic, and kinetic studies of ion adsorption at the oxide–aqueous interface. Ph.D. thesis. Stanford University. Hayes K. F., Redden G., Ela W., and Leckie J. O. (1991) Surface complexation models: An evaluation of model parameter estimation using FITEQL and oxide mineral titration data. J. Colloid Interface Sci. 142, 448 – 469. Hayes K. F. and Katz L. E. (1996) Application of x-ray adsorption spectroscopy for surface complexation modeling of metal ion sorption. In Physics and Chemistry of Mineral Surfaces (ed. P. V. Brady), pp. 147–223. CRC Press. Helgeson H. C. and Kirkham D. H. (1976) Theoretical prediction of the thermodynamic properties of aqueous electrolytes at high pressures and temperatures. III. Equation of state for aqueous species at infinite dilution. Am. J. Sci. 276, 97–240. Hiemstra T., van Riemsdijk W. H., and Bolt G. H. (1989a) Multisite proton adsorption modeling at the solid/solution interface of (hydr)oxides: A new approach I. Model description and evaluation reaction constants. J. Colloid Interface Sci. 133, 91–104. Hiemstra T., Wit J. C. M. D., and van Riemsdijk W. H. (1989b) Multisite proton adsorption modeling at the solid/solution interface of (hydr)oxides: A new approach II. Applications to various important (hydr)oxides. J. Colloid Interface Sci. 133, 105–117. Hiemstra T. and van Riemsdijk W. H. (1991) Physical chemical interpretation of primary charging behavior of metal (hydr) oxides. Colloids Surfaces 59, 7–25. Hiemstra T. and van Riemsdijk W. H. (1996) A surface structural approach to ion adsorption: The charge distribution (CD) model. J. Colloid Interface Sci. 179, 488 –508. Huang P. (1996) The effects of the adsorption of metal ions and surfactant behavior on the interfacial behavior of silicate minerals. Ph.D. thesis. University of California, Berkeley. Huang C. and Stumm W. (1973) Specific Adsorption of Cations on Hydrous (-Al2O3. J. Colloid Interface Sci. 43, 409 – 420. James R. O. and Healy T. W. (1972) Adsorption of hydrolyzable metal ions at the oxide–water interface III. A thermodynamic model of adsorption. J. Colloid Interface Sci. 40, 65– 81. Jang H. M. and Fuerstenau D. W. (1986) The specific adsorption of alkaline-earth cations at the rutile/water interface. Colloids Surfaces 21, 235–257. Johnson S. B., Franks G. V., Scales P. J., and Healy T. W. (1999a) The binding of monovalent electrolyte ions on '-alumina. II. The shear yield stress of concentrated suspensions. Langmuir 15, 2844 –2853. Johnson S. B., Scales P. J., and Healy T. W. (1999b) The binding of monovalent electrolyte ions on '-alumina. I. Electroacoustic studies at high electrolyte concentrations. Langmuir 15, 2836 –2843. Kallay N., Colic M., Fuerstenau D. W., Jang H. M., and Matijevic E. (1994) Lyotropic effect in surface charge, electrokinetics, and coagulation of a rutile dispersion. Colloid Polymer Sci. 272, 554 –561. Katz L. E. and Hayes K. F. (1995) Surface complexation modeling I. Strategy for modeling monomer complex formation at moderate surface coverage. J. Colloid Interface Sci. 170, 477– 490. Kitamura A., Fujiwara K., Yamamoto T., Nishikawa S., and Moriyama H. (1999) Analysis of adsorption behavior of cations onto quartz surface by electrical double-layer method. J. Nucl. Sci. Technol. 36, 1167–1175. Koretsky C. M., Sverjensky D. A., Salisbury J. W., and D’Aria D. M. (1997) Detection of surface hydroxyl species on quartz, (-alumina and feldspars using diffuse reflectance infrared spectroscopy. Geochim. Cosmochim. Acta 61, 2193–2210. Koretsky C. M., Sverjensky D. A., and Sahai N. (1998) A model of surface site types on oxide and silicate minerals based on crystal chemistry: Implications for site types and densities, multi-site adsorption, surface infrared spectroscopy, and dissolution kinetics. Am. J. Sci. 298, 349 – 438. Kosmulski M. (1996) Adsorption of cadmium on alumina and silica: Analysis of the values of stability constants of surface complexes calculated for different parameters of triple layer model. Colloids Surfaces 117, 201–214. Kosmulski M. (1999) How to handle the ion adsorption data with variable solid-to-liquid ratios by means of FITEQL. Colloids Surfaces 397, 408 – 407. Liang L. (1988) Effects of surface chemistry on kinetics of coagulation of submicron iron oxide particles ('-Fe2O3) in water. Ph.D. thesis. California Institute of Technology. Prediction of electric double-layer capacitances Little L. H. (1966) Infrared Spectra of Adsorbed Species.Academic Press. Ludwig C., Devidal J.-L., and Casey W. H. (1996) The effect of different functional groups on the ligand-promoted dissolution of NiO and other oxide minerals. Geochim. Cosmochim. Acta 60, 213– 224. Lumsden D. G. and Evans L. J. (1995) Surface complexation model parameters for goethite ('-FeOOH). J. Colloid Interface Sci. 164, 119 –125. Lützenkirchen J. (1997) Ionic strength effects on cation sorption to oxides: Macroscopic observations and their significance in microscopic interpretation. J. Colloid Interface Sci. 195, 149 –155. Lützenkirchen J. (1998) Parameter estimation for the triple layer model. Analysis of conventional methods and suggestion of alternative possibilities. J. Colloid Interface Sci. 204, 119 –127. Lützenkirchen J. (1999) Parameter estimation for the constant capacitance surface complexation model: Analysis of parameter interdependencies. J. Colloid Interface Sci. 210, 384 –390. Lützenkirchen J., Magnico P., and Behra P. (1995) Constraints upon electrolyte binding constants in triple-layer model calculations and consequences of the choice of the thermodynamic framework. J. Colloid Interface Sci. 170, 326 –334. Lyklema J. and Overbeek J. T. G. (1961) Electrochemistry of silver iodide: The capacity of the double layer at the silver iodide–water interface. J. Colloid Sci. 16, 595– 608. Machesky M. L., Wesolowski D. J., Palmer D. A., and Ichiro-Hayashi K. (1998) Potentiometric titrations of rutile suspensions to 250°C. J. Colloid Interface Sci. 200, 298 –309. O’Day P. A., Newville M., Neuhoff P. S., Sahai N., and Carroll S. A. (2000) X-ray adsorption spectroscopy of strontium(II) coordination I. Static and thermal disorder in crystalline, hydrated, and precipitated solids and in aqueous solution. J. Colloid Interface Sci. 222, 184 –197. Riese A. C. (1982) Adsorption of radium and thorium onto quartz and kaolinite: A comparison of solution/surface equilibrium models. Ph.D. thesis. Colorado School of Mines. Robertson A. P. and Leckie J. O. (1997) Cation binding predictions of surface complexation models: Effects of pH, ionic strength, cation loading, surface complex, and model fit. J. Colloid Interface Sci. 188, 444 – 472. Robinson R. A. and Stokes R. H. (1959) Electrolyte Solutions.Butterworths. Sahai N. and Sverjensky D. A. (1997a) Evaluation of internallyconsistent parameters for the triple-layer model by the systematic analysis of oxide surface titration data. Geochim. Cosmochim. Acta 61, 2801–2826. Sahai N. and Sverjensky D. A. (1997b) Solvation and electrostatic model for specific electrolyte adsorption. Geochim. Cosmochim. Acta 61, 2827–2848. Sahai N., Carroll S. A., Roberts S., and O’Day P. A. (2000) X-ray adsorption spectroscopy of strontium(II) coordination II. Sorption and precipitation at kaolinite, amorphous silica and goethite surfaces. J. Colloid Interface Sci. 222, 198 –212. Shannon R. D. and Prewitt C. T. (1969) Effective ionic radii in oxides and fluorides. Acta Cryst. B25, 915–946. Shock E. L. and Helgeson H. C. (1988) Calculation of the thermody- 3653 namic and transport properties of aqueous species at high pressures and temperatures: Correlation algorithms for ionic aqueous species and equation of state predictions to 5 kb and 1000 C. Geochim. Cosmochim. Acta 52, 2009 –2036. Sprycha R. (1984) Surface charge and adsorption of background electrolyte ions at anatase/electrolyte interface. J. Colloid Interface Sci. 102, 173–185. Sprycha R. (1989) Electrical double layer at alumina/electrolyte interface: I. Surface charge and zeta potential. J. Colloid Interface Sci. 127, 1–11. Stumm W. (1997) Reactivity at the mineral–water interface: Dissolution and inhibition. Colloids Surfaces 120, 143–166. Stumm W. and Wieland E. (1990) Dissolution of oxide and silicate minerals: Rates depend on surface speciation. In Aquatic Chemical Kinetics (ed. W. Stumm), pp. 367– 400. J. Wiley. Sverjensky D. A. and Sahai N. (1996) Theoretical prediction of singlesite surface protonation equilibrium constants for oxides and silicates in water. Geochim. Cosmochim. Acta 60, 3773–3798. Toney MF, et al. (1994) Voltage-dependent ordering of water molecules at an electrode– electrolyte interface. Nature 368, 444 – 446. Venema P., Hiemstra T., and van Riemsdijk W. H. (1996a) Comparison of different site-binding models for cation sorption: Description of pH dependency, salt dependency, and cation-proton exchange. J. Colloid Interface Sci. 181, 45–59. Venema P., Hiemstra T., and van Riemsdijk W. H. (1996b) Multisite adsorption of cadmium on goethite. J. Colloid Interface Sci. 183, 515–527. Westall J. C. and Hohl H. (1980) A comparison of electrostatic models for the oxide/solution interface. Adv. Colloid Interface Sci. 12, 265– 294. Yates D. E. (1975) The structure of the oxide/aqueous electrolyte interface. Ph.D. thesis. University of Melbourne. APPENDIX Figures A1 and A2 contain plots of surface-charge data as functions of pH and ionic strength. The solid curves in the figures represent calculations carried out in the present study by use of the parameters in Table A1 and the extended triple-layer approach described previously (Sahai and Sverjensky, 1997a,b; Criscenti and Sverjensky, 1999). In this approach, the fitted parameters are the capacitance (C1) and the equilibrium constants for the binding of the electrolyte cation (log KM) and anion (log KL) in the case of 1:1 electrolytes. In the case of quartz in CaCl2, the only fit parameter was the capacitance (see Table A1, footnote g). All other parameters in the Table A1 were estimated in a manner consistent with the extended triple-layer approach. The site densities (Ns) were taken from Sahai and Sverjensky (1997a,b) and Criscenti and Sverjensky (1999) and are consistent with the results of tritium exchange experiments. In the case of the surface-charge data for quartz in CaCl2 (Riese, 1982), the site density in Table 1 refers to tritium exchange experiments carried out by Riese (1982). The values of the surface protonation equilibrium constants in the Table A1 were calculated from the ZPC (" 0.5[log K1 ! log K2]) reported by the authors who had carried out the surface-charge experiments, and from the theoretically estimated values of (pK (" log K2 # log K1) published previously (Sverjensky and Sahai, 1996). 3654 D. A. Sverjensky Fig. A1. Plot of surface-charge data as functions of pH and ionic strength for rutile in different electrolyte solutions. The symbols represent experimental data. The curves were generated in the present study (see text). Prediction of electric double-layer capacitances 3655 Fig. A2. Plot of surface-charge data as functions of pH and ionic strength for quartz and amorphous silica in different electrolyte solutions. The symbols represent experimental data. The curves were generated in the present study (see text). Table A1. Triple-layer model parameters derived from regression of surface-charge experiments referring to wide ranges of ionic strength using the methods described by Sahai and Sverjensky (1997a). Salt (mL) Solid Nsa log K1b log K2c log KMd log KLe C1f Source of surface charge data N(CH3)4Cl CsCl KNO3 NaCl NaClO4 LiCl CaCl2 KNO3 N(CH3)4Cl '-TiO2 '-TiO2 '-TiO2 '-TiO2 '-TiO2 '-TiO2 quartz quartz am. SiO2 12.5 12.5 12.5 12.5 12.5 12.5 4.5 11.4 4.6 2.6 2.8 2.6 2.2 2.6 2.8 #2.2 #2.2 #1.2 9.0 9.2 9.0 8.6 9.0 9.2 6.2 6.2 7.0 3.4 2.5 2.1 2.5 2.0 2.8 —g 0.2 2.0 3.4 2.4 2.2 2.3 1.9 2.2 0.2 2.0 0.7 55 95 110 120 125 155 81 105 50 Yates (1975) Kallay et al. (1994) Yates (1975) Machesky et al. (1998) Bérubé and de Bruyn (1968b) Kallay et al. (1994) Riese (1982) Huang (1996) Casey (1994) Site density (nm#2). Refers to the equilibrium %SOH ! H! " %SOH! 2. c Refers to the equilibrium %SO# ! H! " %SOH. d # ! # Refers to the equilibrium %SO ! M " %SO _M!. e # # " %SOH! Refers to the equilibrium %SOH! 2 ! L 2 _L . f Capacitance (C1) has units of %F. cm#2. g The Ca equilibria used to fit the data for quartz in CaCl2 are based on independent fits to data for the adsorption of trace amounts of calcium in KNO3 and NaCl solutions (D. A. Sverjensky, unpublished data) which resulted in the following equilibria and equilibrium constants (adjusted for differences in site densities and molarity/mole fraction conversions): %SOH ! Ca!! ! H2O " %SO#_Ca(OH)! ! 2H! (log K " #14.0); and 2%SOH ! Ca!! " (%SO#)2_Ca!! ! 2H! (log K " #7.1); the value of C1 given for quartz in CaCl2 solutions was the only adjustable parameter in the regression of the data depicted in Figure A2. a b
© Copyright 2026 Paperzz