Journal of Colloid and Interface Science 293 (2006) 312321 www.elsevier.

com/locate/jcis

A new surface structural approach to ion adsorption: Tracing the location of electrolyte ions
Rasoul Rahnemaie 1 , Tjisse Hiemstra , Willem H. van Riemsdijk
Department of Soil Quality, Wageningen University, P.O. Box 8005, 6700 EC Wageningen, The Netherlands Received 27 April 2005; accepted 28 June 2005 Available online 8 August 2005

Abstract Electrolyte ions differ in size leading to the possibility that the distance of closest approach to a charged surface differs for different ions. So far, ions bound as outersphere complexes have been treated as point charges present at one or two electrostatic plane(s). However, in a multicomponent system, each electrolyte ion may have its own distance of approach and corresponding electrostatic plane with an ion-specic capacitance. It is preferable to make the capacitance of the compact part of the double layer a general characteristic of the solidsolution interface. A new surface structural approach is presented that may account for variation in size of electrolyte ions. In this approach, the location of the charge of the outersphere surface complexes is described using the concept of charge distribution in which the ion charge is allowed to be distributed over two electrostatic planes. It was shown that the concept can successfully describe the pH dependent proton binding and the shift in the isoelectric point (IEP) in the presence of variety of monovalent electrolyte ions, including Li+ , Na+ , K+ , Cs+ , Cl , NO , and ClO with a common set of parameters. The new concept also sheds more light on the degree of hydration of the ions when 3 4 present as outersphere complexes. Interpretation of the charge distribution values obtained shows that Cl ions are located relatively close to the surface. The large alkali ions K+ , Cs+ , and Rb+ are at the largest distance. Li+ , Na+ , NO , and ClO are present at intermediate 3 4 positions. 2005 Elsevier Inc. All rights reserved.
Keywords: Electrolyte ion; Diffuse double layer; Basic Stern; Three plane model; Adsorption; Iron oxide; Goethite; CD model; MUSIC model

1. Introduction Usually, mineral surfaces are charged. Surface charge is compensated by counterions of opposite charge as rst suggested by Helmholtz in 1835 [1]. The combination of surface charge and counter charge is called the double layer. Without chemical interaction of ions with the mineral surface, electrolyte ions have a distribution pattern as derived almost a century ago by Gouy [2] and Chapman [3]. This structure is known as the diffuse double layer (DDL) and it can be experimentally tested with surface force apparatus (SFA) [4].
* Corresponding author. Fax: + 31 317 48 37 66.

E-mail address: tjisse.hiemstra@wur.nl (T. Hiemstra).

1 Present address: Department of Soil Science, Tarbiat Modares Univer-

sity, P.O. Box 14115-336, Tehran, Iran. 0021-9797/$ see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2005.06.089

Near the surface, electrolyte ions cannot be treated as point charges. Counterions have a nite size, which implies that the ions have a minimum distance of approach to the surface. Stern [5] has described the concept of minimum charge separation. The compact part of DDL, between surface- and countercharge, is called the Stern layer. In terms of electrostatics, the compact part of the double layer can be considered as a plate condenser with a certain capacitance. An important parameter, relating layer thickness and capacitance, is the dielectric constant of the medium in the condenser. The precise value of this macroscopic quantity when applied to interfaces is unknown. Electrolyte ions may, in addition to electrostatic forces, also be bound due to specic weak interactions with the surface [69]. This phenomenon is called ion pair formation and the complexes are named outersphere complexes. In some cases, a simple electrolyte ion like Rb+ may even

R. Rahnemaie et al. / Journal of Colloid and Interface Science 293 (2006) 312321

313

form an innersphere complex as shown recently by Fenter et al. [10]. In such surface complexes, one or more ligands of the adsorbed ion are common with the metal ions of the solid. The charge of metal oxide surfaces is related to the relative adsorption of protons by surface oxygens. The proton binding is pH dependent. An important factor in the H binding is the electrostatic potential, which is created by the accumulation of charge at the surface. In principle, charge and electrostatic eld are localized. However, for surfaces the local interaction can be described very well with a smeared out approach as was shown by Borkovec [11]. This mean eld approach is generally applied in surface complexation models. Many models have been proposed to describe the relation between the solution composition and the charge properties of colloids [79,1124]. These surface complexation models are a combination of chemical equilibria and a chosen electrostatic model. From the perspective of physical reality, a minimum prerequisite to account for the variable inuence of the electrolyte concentration is the use of a diffuse double layer in the model. According to Stern, the ions should have a minimum distance of approach, which has to be described by the model. The combination is called the basic Stern (BS) model. It can be considered as the simplest model for the description of the variable proton charge of metal oxide surfaces [25]. Moreover, the BS model accounts for physically realistic features. If necessary, ion pair formation can be implemented within the BS concept, locating these outersphere complexes at the minimum distance of approach. The BS approach can be extended with an additional layer to change the double layer prole. A known representative of this class of models is the triple layer (TL) model [9]. In the classical TL model, the distance of minimum approach of the electrolyte ions in the diffuse double layer is greatly enlarged by the choice of a very low capacitance for the second layer. This results in a much smaller interaction between ions in the DDL and proton charge at the surface. Due to this constraint, ion pair formation is a priori needed in order to get sufcient surface charge in this model approach. Increase of the electrolyte ion afnity for ion pair formation at the metal oxide surface will lead to more surface charge due to an improved screening of the electrostatic eld. Generally, electrolyte ion afnities have been derived from the analysis of the charging behavior of oxides with a surface complexation model (e.g., [17]). A systematic analysis for the charging behavior of Ti-oxides with the BS model [26], has shown that the afnity constants for electrolyte cations are generally larger than those for anions. A higher afnity of cations over anions will lead to an upward shift of the isoelectric point, in particular at high electrolyte concentrations [27]. This phenomenon can be observed with electro-acoustic-phoresis, as has been shown for gibbsite [28], alumina [29], zirconia and rutile [30], anatase [31], and silica [32]. It indicates that the larger afnity of electrolyte cations versus anions is a general phenomenon. For

goethite, the variation in afnity for various electrolyte anions has been derived from the interpretation of a consistent set of titration data in NaClO4 , NaNO3, and NaCl media with the BS model [33]. In that study, arbitrarily the afnity of the surface for Na+ and NO was assumed to be equal and no 3 attempt was made to determine the binding constant of Na+ in detail. In the present study, we will derive the afnity of electrolyte cations and anions using new surface charge data for goethite, measured in a series of monovalent electrolyte media (LiNO3 , LiCl, NaNO3 , NaCl, KNO3 , and CsNO3 ). Preliminary modeling indicated that any difference in electrolyte afnity is most markedly observed at high pH and relatively high electrolyte concentrations. Such conditions were applied in our experiments. In a companion paper [34], we will describe the interaction of divalent electrolyte ions (Ca2+ , Mg2+ , and SO2 ). The binding of these ions may be 4 a combination of innersphere and outersphere complexation. The compact part of the double layer is crucial in interface chemistry since changes in potential are most profound there. From this perspective, the location of the electrolyte ions is an important issue. It determines the electrostatic contribution ( Gelec ) to the overall afnity ( Goverall ) of the electrolyte ions, i.e., Goverall = Gintr + Gelec . In most model approaches, the electrolyte cations and anions are located at the same position, i.e., they experience the same electrostatic eld strength. Sverjensky [35] has applied this approach for electrolyte ions and has shown that the capacitance varied for various salts. This was interpreted as a variation in the mean distance of approach of the salts. In principle, each electrolyte ion will have its own size and location in the double layer prole. In a model approach, this would lead to an ion specic electrostatic plane with corresponding capacitance. The question arises as to what extent macroscopic data can reveal information on the location of the electrolyte ions in the double layer prole. In the present study, we will analyze the primary charging behavior of a metal oxide in detail. We will use a charge distribution (CD) approach for outersphere complexation. In the data analysis, the charge of the electrolyte ions is free to be distributed between two electrostatic planes. This approach may reveal the location of the charge in the double layer prole. In order to nd a common set of parameters, all experimental data will be analyzed together and treated as one data set. The results will be interpreted in terms of the double layer structure. 1.1. MUSIC model 1.1.1. Reactive groups In the lattice of goethite (-FeOOH), two different types of triply coordinated oxygens exist, one nonprotonated (Fe3 O) and one protonated (Fe3 OH) oxygen. The difference in proton afnity of both triply coordinated oxygens is linked to a difference in the distances between the oxygen and the coordinating Fe-ions. Both types of oxygens are also

314

R. Rahnemaie et al. / Journal of Colloid and Interface Science 293 (2006) 312321

found at the surface of the dominant 110 face and the 100 face [3638]. The triply coordinated surface groups of type I are non-protonated over a wide pH range, i.e., present as Fe3 OI . The other type of triply coordinated surface group (type II) has a higher proton afnity and can be present as Fe3 OII H. Charge can be attributed to these surface groups by applying the Pauling bond valence concept [39,40]. Based on the concept of equal charge distribution [41], the 1/2 triply coordinated oxygen can be represented as Fe3 OI +1/2 . and the triply coordinated hydroxyl as Fe3 OII H The triply coordinated oxygens and hydroxyls of the bulk mineral can also be found at the surface. In addition, the interface of goethite has surface groups with a lower Fe coordination. The doubly coordinated oxygens will have a high afnity for protons, because the Fe2 O1 is highly under-saturated in terms of charge [40]. It leads to the formation of a stable and uncharged surface species Fe2 OH0 , which does not easily accept a second proton. Singly coordinated surface oxygens (FeO3/2 ) are highly unstable and are always transformed into FeOH1/2 in aqueous systems [39,42]. Depending on the pH, the hydroxyl (FeOH1/2 ) may accept a second proton forming a surface water group +1/2 (FeOH2 ). 1.1.2. Surface structure and site density At the 110 and 100 faces, one row of singly coordinated surface groups (FeOH(H)) is found on a unit cell basis, which is equivalent to a site density Ns of 3 nm2 at the 110 face [43]. In addition, one row of non-H-reactive doubly coordinated groups (Fe2 OH) is present. At the 110 and 100 faces, one also has three rows of triply coordinated groups (Fe3 OH), one row with the low proton afnity (type I) and two rows with the high proton afnity (type II). The large difference in proton afnity of both types of triply coordinated surface groups leads to a lower effective site density for the proton reactive one [40], since the charge of one row of Fe3 O1/2 (type I) cancels against one row of Fe3 OH+1/2 (type II). The apparent site density of the proton reactive triply coordinated surface group (type II) therefore is 3 nm2 . The afnity constant of the type II group is estimated to be relatively close to that of the singly coordinated group, i.e., log K 4 [40]. To simplify, the afnity constants of the proton reactive groups (FeOH1/2 and 1/2 Fe3 OII ) are set equal in the modeling [40,44]. It can be shown that with this simplication the logarithm of the afnity constant is equal to the value of the PZC, in the absence of ion pair formation or in case of symmetrical ion pair formation. Faces like the 021 and 001 faces terminate the goethite needles at the top ends of the crystals. These crystal faces have equal numbers of singly and doubly coordinated surface groups (Ns = 78 nm2 ). In the modeling, the protonation constant of the singly coordinated surface group at these faces is taken equal to that of the 110 face. The overall effective site density of goethite is calculated assuming 90% 110/100 face and 10% 021/001 face leading

to 2.7 + 0.75 = 3.45 nm2 for the singly coordinated and 2.7 nm2 for the triply coordinated surface group. 1.2. Primary protonation reactions Based on the above analysis, the protonation of the crystal face can be represented by two protonation reactions: FeOH1/2 + H+ (aq) Fe3 O1/2 + H+ (aq) FeOH2
+1/2

(1) (2)

Fe3 OH+1/2 .

The proton reactive surface groups may interact with electrolyte ions, forming outersphere surface complexes, resulting in the formation of the ion pairs FeOH1/2 C+z , +1/2 FeOH2 Az and Fe3 O1/2 C+z , Fe3 OH+1/2 Az , in which C+z represents the electrolyte cation and Az the anion. To simplify the modeling, the proton afnity constants for singly and triply coordinated surface groups are set equal, log KH,FeOH = log KH,Fe3 O . The same is done for the ion pair formation constants for singly and triply coordinated surface groups [44]. 1.3. Locating the ion charge We discussed above that in the simplest approach the outersphere complexes of electrolyte ions are located at a position determined by the minimum distance of approach of ions to the surface. These complexes do not have common ligands with surface groups. Many ions may form innersphere complexes with surface groups. In these surface complexes, one or more ligands of the adsorbed ion are common with the surface. The other ligands are orientated towards the solution. In a model approach, they are at a distinct distance from the surface in the compact part of the double layer. According to the bond valence concept, this leads to a distribution of charge in the compact part of the double layer [44]. In the charge distribution (CD) model, part of the ion charge is attributed to a surface plane and the other part is attributed to a second electrostatic plane, at some distance from the surface. In the simplest picture, the second electrostatic plane may coincide with the plane of minimum distance of approach of the electrolyte ions. It implies that the electrolyte ions may have a rather large interaction with innersphere complexes. The interaction can be reduced by locating the outer ligands of the innersphere complexes at a smaller distance from the surface in a separate electrostatic plane. This model has been proposed by Hiemstra and van Riemsdijk [44] and is named the three plane (TP) model. Innersphere complexation of PO3 has been best described with the TP model [44]. How4 ever, SO2 adsorption has been best described with the BS 4 approach [45]. To unify both descriptions in one concept, Rietra et al. [45] suggested to locate the anion at the 1-plane and the cation at the 2-plane. In Fig. 1, the corresponding TP model is schematically represented.

R. Rahnemaie et al. / Journal of Colloid and Interface Science 293 (2006) 312321

315

bottles. The acid solutions were kept in glass bottles to avoid contamination by organic materials. Ultra pure water (0.018 dS/m) was used throughout the experiments. The ultra pure water was pre-boiled to remove dissolved CO2 before use in the experiments. The preparation of the goethite suspension was based on the method of Atkinson et al. [12], which has been described in more detail by Hiemstra et al. [47]. A freshly prepared 0.5 M Fe(NO3 )3 solution was slowly titrated with 2.5 M NaOH to pH 12.0. The prepared suspension was aged for 4 days at 60 C and subsequently dialyzed for at least three weeks in ultra pure water, which was refreshed twice a day. The BET(N2 ) specic surface area of our goethite equaled 98.6 m2 g1 .
Fig. 1. A schematic representation of the TP model at the goethitewater interface. In the TP model, there are three electrostatic planes at the surface. The common ligands of innersphere surface complexes with the surface groups are located at the 0-plane. The solution-oriented ligands of the innersphere complexes are located at the 1-plane. Electrolyte ions may be located at the 1- or 2-plane. The 2-plane coincides with the head end of the DDL.

2.2. Charging behavior of goethite A salt-free suspension of goethite was acidied to pH 3.55 with 1 M HNO3 and continuously purged with puried, moist N2 for 24 h to remove (bi)carbonate. Subsamples of this stock suspension were used throughout the titration experiments. The charge of the goethite in the stock suspension served as the reference point to calculate the initial surface charge present in the different systems at the start of the various titrations. Certain volumes of a specied salt solution were added to the series of individual batches of goethite suspension. This leads to a desired ionic strength and a xed initial goethite concentration for the series of titrations in different background electrolyte ions. A certain starting pH of the suspension is then obtained and a relative change in the charge of the suspension can be calculated. This relative charge with respect to the stock suspension is different for the different electrolytes and different salt levels. Each batch of goethite suspension, which contained a known amount of a specied salt, was titrated from its initial pH to a pH of about 10.5. Internal calibration of the various titration curves relatively to the stock suspension makes our approach distinctive from a set of classical titrations without internal calibration/consistency. To get reliable results at high pH, it is necessary to use a high solid-solution ratio to minimize the inuence of corrections for the blank. In our experiments, the initial goethite concentration was 16.52 g L1 that is equivalent with 1628.4 m2 L1 . Individual titrations were performed at different concentrations of various electrolyte ions, i.e., 0.04, 0.10, and 0.20 M solutions of LiCl, LiNO3 , and NaCl, and 0.20 M solutions of NaNO3 , KNO3 , and CsNO3 . Each titration was started after equilibration for half an hour. After each addition of base, the suspension was rst equilibrated for 15 min. Then, monitoring of the drift in pH was started. The nal data point was collected, when the drift became lower than 0.02 mV min1 for 3 successive readings (each 1 min). The titrations were done with 0.100 M NaOH. This procedure will introduce a small amount of Na+ ions into the background solution. Their concentration is actually neg-

1.4. A new approach Electrolyte ions differ in size. It implies that the double layer will accommodate these different ions at different distances from the surface. As a consequence, one has to dene, in principle, for each ion an individual electrostatic plane. This idea is not very attractive, in particular in multicomponent systems. Therefore, electrolyte ions have often been located at only one electrostatic location. In case of the use of the above given three plane model (Fig. 1), it is possible to accommodate the electrolyte ions within the second Stern layer, the layer between the 1- and 2-plane. Depending on the relative position of the electrolyte ions in the double layer prole, the charge of the electrolyte ion is divided over both electrostatic planes at either side of the second Stern layer. Small ions will attribute more of their charge to the 1-plane. Large ions do the opposite. The relative locations of the electrolyte ions in the double layer can be found by determining the charge distribution coefcients from an accurate and internally calibrated data set. The above-suggested model for outersphere complexes can be considered as an extension of the CD model. Actually, the charge distribution of the outersphere ions is related to the size of the ion, while the charge distribution of innersphere complexes can be related to the number of common ligands and the bond strength [44,46].

2. Material and methods 2.1. Synthesis and characterization of goethite To avoid (bi)carbonate contamination, all chemicals (Merck p.a.) were made under a puried N2 atmosphere. To be free of silica, these solutions were kept in polyethylene

316

R. Rahnemaie et al. / Journal of Colloid and Interface Science 293 (2006) 312321

Fig. 2. The relative proton adsorption on goethite as affected by the type of cation and anion of the background electrolytes. The concentration of added salts is 0.2 molar. The different symbols on the gure show the exact measured values and the lines are interpolations between the experimental data.

ligible compared with the studied background electrolytes. Nevertheless, for each data point, the amounts of all added ions were taken into account during modeling. In addition, we accounted for the change of the suspension concentration during the titration. In order to compare the one-way titration data with the more common forwardbackward titrations, a sample of goethite suspension was titrated with 0.015, 0.05, and 0.2 M NaNO3 . The various electrolyte additions were done at low pH, followed by a forward and backward titration of the suspension. A double-junction AgAgCl reference electrode and a glass electrode were used for the titration experiments. The outer junction of the reference electrode was lled with a solution of 0.125 M NaNO3 and 0.875 M KNO3 . The mobility of the positive and the negative ions in this solution is about the same, i.e., the diffusion potential over the outer junction is independent of the salt concentration [48]. The titration was done using the Wallingford automatic computerized titrator [49]. The electrodes were calibrated with 35 commercial (Merck) buffer solutions, ranging from pH 2 to 12. The temperature was xed at 20 0.1 C using a thermostated reaction cell, while the entire apparatus was present in a constant temperature room (22 1 C). In the calculation, the water dissociation and dielectric constants were corrected for 20 C and the activity coefcients were calculated for each data point individually.

Fig. 3. The effect of type of cation or anion on the proton adsorption on goethite. (a) illustrates the effect of a change of Li+ for Na+ in a chloride solution and (b) a change of Cl for NO in a lithium solution. 3

3. Results and discussion 3.1. Treatment of goethite titration data For each addition of base, a change in surface charge can be calculated relative to the starting point of the titration. Fig. 2 illustrates an example of the relative surface charge of goethite in 0.2 molar solutions of a series of electrolytes. The data in Fig. 2 show that for a given concentration of electrolyte ions, the proton desorption is affected by the type

of cation as well as anion present. A close look at the data in Fig. 2 reveals that the major difference in the starting point is due to the specic type of anion, i.e., nitrate or chloride. The gure also shows that for a particular anion, the charging curve is strongly different for different cations, which points to different interactions with the charged surface. The slope of titration curves reect the strength of the interaction between the surface groups and the adsorbing ions, which for outersphere surface complexes may depend on the distance to the surface (electrostatic contribution) and/or the intrinsic afnity. In Fig. 3, the particular effect of a cation or an anion is illustrated more clearly, showing the charging curves at a change of Li+ for Na+ in a chloride solution (Fig. 3a) and at a change of Cl for NO in a lithium solu3 tion (Fig. 3b). Symmetrical ion pair formation on goethite has often been assumed [33,44,50]. This leads to the possibility to use the common intersection point (CIP) of the titration data as the pristine point of zero charge (PPZC), which is used to transform the relative surface charge of goethite to an absolute scale. As a starting point, this idea was applied to the titration data of sodium nitrate and the afnity constants of sodium and nitrate were set equal. However, further modeling of the data revealed that actually the afnity of none of the electrolyte ions was equal. This means that the PPZC cannot be obtained unequivocally with these ions as background electrolytes. The uncertainty in the de-

R. Rahnemaie et al. / Journal of Colloid and Interface Science 293 (2006) 312321

317

termination of the zero charge is related to the difference in the afnity between the cation and the anion of the background electrolyte solution. Model calculations show that in case of the use of NaNO3 or LiCl, the surface charge present in the CIP is almost zero with a deviation smaller than 0.05 mol charge m2 , in contrast with the charge in the CIP in case of LiNO3 or NaCl. Therefore, we used the CIP in the presence of NaNO3 , as the zero reference point. Based on this perception, the goethite surface charge data were scaled from the relative to the absolute charge by identifying the CIP in NaNO3 (at pH 9.0) as zero charge. 3.2. Modeling of electrolyte ion interaction Monovalent electrolyte ions are mainly adsorbed as outersphere surface complexes. The outersphere surface complexes do not share ligands with the metal ion of the reactive surface groups. These ions are mainly adsorbed via electrostatic interactions with the surface. The following equations show the way that the cation and anion binding reactions were formulated for data modeling: FeOH1/2 + C+1 (aq) FeOH+1/2 + A1 (aq) FeOH1/2 C+1 ,
+1/2 FeOH2 A1 ,

Table 1 The charge allocation (z) and the afnity constants (log K) of interaction of protons and monovalent cations and anions with goethite as derived from modeling of the goethite titration data using the charge distribution model of the outersphere complexes. The charge of electrolyte ions is distributed over the 1- and 2-planes. Sum of z1 and z2 is equal to the valence of ions involved. The tted capacitance for the rst and second Stern layer are respectively C1 = 0.98 0.01 and C2 = 0.73 0.05 F m2 . The standard deviation is given for the tted parameters Ions H+ Li+ Na+ K+ Cs+ Cl NO 3 ClO 4 1 0 0 0 0 0 0 0 z0 z1 0 0.460.03 0.590.07 0 0 1a 0.700.02 0.570.03 z2 0 0.540.03 0.410.07 1a 1a 0 0.300.02 0.430.03 log K 9.0b 0.46 0.04 0.38 0.08 0.38 0.14 1.5c 0.53 0.02 0.43 0.02 0.520.06d

(3) (4)

a The charge was placed by denition since the model showed that almost the total charge remains at this plane. b The proton afnity was derived from the titration data in the presence of NaNO3 (see text). c The log K value was not tted but chosen, due to the low sensitivity of the data. d The afnity and CD values for ClO were optimized on the goethite 4 titration data of Rietra et al. [33].

where the symbol emphasizes the weak chemical interaction between surface groups and adsorbed ions, and C+1 and A1 represent the monovalent cations and anions, respectively. The equations are written only using the singly coordinated surface groups. However, the ion pair formation with triply coordinated surface groups was formulated similarly. As mentioned above, the data of Fig. 2 show that electrolyte cations and anions may have a different electrostatic interaction with the surface. This can be due to differences in the intrinsic afnity constant of Eqs. (3) and (4) but differences in electrostatic interaction may contribute too. The latter can be interpreted in terms of difference in distance to the surface. 3.3. Model results As mentioned in the introduction, the titration data will be used to reveal the position of the charge of the electrolyte ions in the double layer prole. In a model, a variation in distance can be implemented by distributing the charge between two electrostatic positions. In our approach, the charge distribution and the intrinsic electrolyte afnity constants are both derived from the titration data. The tted value of charge distribution of the various outersphere complexes reects the relative position of the ions. The tted capacitance of the inner and outer Stern layer is a measure of the absolute distance. The charge distribution of electrolyte ions was optimized on the titration data. The CD values found for Cl , K+ , and Cs+ ions in a fully free optimization, were slightly larger

than the formal valence of these ions. Therefore, we restricted the maximum charged allowed to be present on the electrostatic planes to the valence of the ions involved (1 or +1). This restriction was necessary for Cl on the 1-plane and for K+ and Cs+ on the 2-plane. The tted parameter values are given in Table 1 for monovalent electrolyte ions. The parameter values in Table 1 show that charge distribution of the Li+ and Na+ ions is different from that of K+ and Cs+ . It implies that both sets of ions have a different electrostatic position. The charge distribution of Li+ and Na+ is rather similar, i.e., these ions have a quite similar location in the electrostatic eld. However, Li+ and Na+ ions have a quite different inuence on the charging behavior, as shown in Fig. 2. In LiCl and LiNO3 , the value of the PZC is about 8.5 while it is about 9 in NaCl and NaNO3 . This is also found by modeling (Figs. 4a4d). Our analysis shows that the major difference in behavior of Li+ and Na+ ions is due to different intrinsic afnities and not to a different electrostatic position. Parameters in a t can be correlated. This may complicate unequivocal determination of parameters. In our approach, an internal correlation between log K and CD was observed. The correlation coefcient was found to be r = 0.4, 0.7, and 0.9 for respectively NO , Li+ , and Na+ . 3 From a model point of view, K+ and Cs+ are important ions, most dominantly determining the capacitance of the second Stern layer (C2 value). It is important to note that there are only data available at one concentration level for K+ and for Cs+ in contrast to three levels for other cations. From a statistical point of view, we cannot exclude that the C2 value might be slightly higher than the suggested value. Model calculations show that the C2 value might fall in the

318

R. Rahnemaie et al. / Journal of Colloid and Interface Science 293 (2006) 312321

Fig. 4. The experimental data and model description of the proton adsorption on goethite in different monovalent electrolyte solutions. The CD model for outersphere complexation was used to describe the titration data.

range of about 0.71.5, since exclusions of the titration data in the presence of CsNO3 and/or KNO3 from the modeling leads to a higher C2 value. There are several goethite titration data sets in Refs. [33, 5056]. These titrations have been done in the presence of NaNO3 , NaCl, or NaClO4 . The reported PZC values range from 8.9 to 9.4. Although there are differences in PZC values, the model parameters found in this study still predict most of these titration data reasonably well as long as the data refer to goethites with a rather similar specic surface area (>80 m2 g1 ). Goethites with a lower specic surface area show a higher slope that implies a larger capacitance is needed to describe the data. This was found for the data of Vangeen et al. [54], Robertson and Leckie [55], and Villalo-

bos et al. [53]. For goethite, it has been shown that the slope of the charging curves depend on the surface area [47]. Recently, Sverjensky [57] has interpreted the titration data for a series of goethites using a revised TP model and showed that the differences in charging can be expressed as a variation in the capacitance value of the inner layer. The difference might be due to surface roughness [57]. Also in our approach, surface roughness would in principle also affect the capacitance of the inner Stern layer. The capacitance of the inner Stern layer can be made adjustable in order to t the data without changing the capacitance of the outer layer in which charge distribution for the outer sphere complexes takes place. Rietra et al. [33] have studied the proton adsorption in the presence of chloride, nitrate, and perchlorate on goethite.

R. Rahnemaie et al. / Journal of Colloid and Interface Science 293 (2006) 312321

319

The above model-derived parameters (capacitances, and afnity of H+ , Na+ , Cl , and NO ) were applied to this 3 data set, yielding a good description of the data (R 2 = 0.99). From the titration data in the Na perchlorate solution, the CD and adsorption afnity of ClO were derived (Table 1). 4 3.4. The capacitance of the Stern layer The capacitance of the Stern layer can be interpreted in terms of a distance, d, between the two electrostatic planes of the double layer [58], according to: r 0 C= (5) , d in which 0 and r are respectively the absolute (8.85 1012 C V1 m1 ) and relative dielectric constant. Note that in case of two Stern layers, we may write for the total distance d = d1 + d2 , i.e., 1/C = 1/C1 + 1/C2 . In principle, the distance of approach of electrolyte ions may be different, leading to an ion dependent capacitance. It is preferable to make the capacitance a general characteristic of the solid/solution interface in case of application to multi-component systems. In the present approach, the Stern layer capacitance is taken as a constant and the effect of the position of the ion charge in the compact part of the double layer prole is taken into account via the charge distribution. In this approach, the capacitance of the inner Stern layer is determined by the minimum distance of approach of the ion closest to the surface (i.e., Cl ), while the capacitance of the second Stern layer is determined by the minimum distance of approach of the electrolyte ions that are the furthest away from the surface, i.e., in this study K+ and Cs+ . Table 1 shows that the tted values of the capacitances are very similar (10.7 F m2 ). 3.5. Distances For interpretation of the capacitances, a value for the dielectric constant is required. This property may change in the double layer prole. A point of concern is the use of the macroscopic quantity in Eq. (5) that is applied to a microscale. Nevertheless, we will try to relate the capacitance to distances to sharpen our picture of the double layer. If water in the outer Stern layer has properties similar to that of pure water (r = 80 at 20 C), a capacitance C2 of about 0.7 F m2 leads to d2 10 (1 109 m). This distance should be considered as a maximum. Due to high concentration of ions, one may consider a lower dielectric constant of about 60 [59] for the outer Stern layer. This leads to a distance of about 8 . It should be noticed that if the dielectric properties (r ) depend on the local electrolyte concentration the capacitance value might not be truly constant, but would depend on the ion loading. If the dielectric properties of the inner Stern layer capacitance are similar to that of water, a capacitance of about 1 F m2 is equivalent to a distance of about 7 , being equivalent with about 2.5 water molecules. In the modeling, the

Fig. 5. The double layer structure based on the interpretation of the CD of outersphere complexes (Cl , K+ , Na+ , Li+ ) in the compact part of the double layer of metal (hydr)oxides in case the inner and outer Stern layer have a different dielectric constant. The location of the Li+ ions is rather comparable with Na+ ions.

Cl ion was found to determine the capacitance of the inner layer. The distance d1 7 would lead to the picture in which Cl is bound primarily as a hydrated ion at the interface. However, if the value of the relative dielectric constant is lower, for instance half way between that of goethite (r = 11) and water or a concentrated electrolyte solution (r = 8060), a capacitance value C1 of about 1 F/m2 is equivalent to d1 43 . In that case, the Cl ion is directly coordinating to the surface groups. A distance of about 34 corresponds with the radius of Cl (1.8 ) plus the distance of half a water molecule (1.4 ). This representation is given in Fig. 5. A lower value of the capacitance is in line with the analysis of the surface charge of metal oxides given by Sverjensky [57]. The given representation gives credit to the idea that ion pair formation is stimulated by local electrostatic forces, which are larger when the opposite charges are as close as possible together. Recently, Predota et al. [60] have simulated with molecular dynamics the binding of Cl at the TiO2 solidsolution interface. They found that the Cl ions were directly coordinating with the surface groups. The K+ ion is found to determine the largest minimum distance of approach (d = d1 + d2 3 + 8 = 11 . The high value for the distance of the second Stern layer (d2 = 8 ) might imply that the large electrolyte ions (such as K+ , Rb+ , Cs+ ) remain hydrated with a sheet of primary and secondary water molecules. The given distance d2 8 can be explained as the sum of the ionic radius of the electrolyte ion (r = 1.33, 1.47, 1.67 ) and the size of two water molecules (5.6 ). The position of K+ is given in Fig. 5. For rutile, it has been suggested, based on the Standing Wave spectroscopy, that Rb+ may form innersphere complex on the 110 face [10]. In our analysis, we found no indications for innersphere complexation of alkali ions such as

320

R. Rahnemaie et al. / Journal of Colloid and Interface Science 293 (2006) 312321

4. Conclusions In the solidsolution interface of minerals, the CD approach is an effective way to account for variation in size of electrolyte ions using only two electrostatic planes. The TP model can be used as a framework to accommodate simultaneously electrolyte ions with different size if the charge of the ions is distributed. The minimum distance of approach of adsorbed electrolyte ions depends on the nite size of ions and their degree of hydration, which determine their relative distances to the surface of minerals. Cl ions are relatively close to the surface. The large alkali ions K+ , Cs+ , and Rb+ are at the largest distance. Li+ , Na+ , NO , and ClO are present at intermediate 3 4 positions. The capacitance of inner Stern layer is determined by the minimum distance of approach of the ion closest to the surface, while the capacitance of the outer layer is determined by the minimum distance of approach of the ion furthest away from the surface.

Fig. 6. The double layer structure based on the interpretation of the CD of outersphere complexes of anions (Cl , NO , and ClO ) in the compact 3 4 part of the double layer of metal (hydr) oxides.

Cs+ and K+ . This may suggest that the binding mechanism for Rb+ might be different on goethite and rutile. This suggestion is supported by the difference in adsorption mechanism found for Sr2+ on goethite (outersphere complex) and rutile (innersphere complex) by spectroscopy [10,61]. The Na+ and Li+ ions were found to distribute their charge about equally over the 1- and 2-plane (Table 1). It indicates that both ions have a position that is intermediate between the Cl ion and the K+ ion. Interpretation of this conclusion points to the presence of Na+ and Li+ in the double layer that are hydrated with primary water molecules, but the ions stay at a larger distance of the surface than Cl , due to the maintenance of a hydration shell of ions (Fig. 5). The Cl ion coordinates primarily with surface groups in this picture in contrast to Na+ . In this double layer picture, the location of the anions can also be indicated (Fig. 6). For NO , the charge distribution 3 factor indicates that a considerable part of the charge (2/3) is in the 1-plane. This might imply that NO partly pene3 trates into the layer where Cl is also bound. This might be with one or two ligands. A precise structure cannot be given. We have also modeled the charge distribution of ClO . Like 4 Na+ , this ion has a charge distribution in which the charge is about equally distributed over the 1- and 2-plane. This might be interpreted as a structure in which the bound ClO keeps 4 a layer of primary water molecules (Fig. 6). Finally, it is noted that very recently Sverjensky [57] has proposed to increase the capacitance of the outer layer in the TP model from 0.2 F m2 to a value equal to the inner capacitance, i.e., C2 1 F m2 . As argued by Hiemstra and van Riemsdijk [58] such a capacitance value is more in line with a realistic physical chemical picture of the compact part of the double layer. A value of C2 1 F m2 is roughly equivalent to the diameter of a hydrated ion if the dielectric constant at the second layer is equal to that of an electrolyte solution.

Acknowledgments The authors thank the Ministry of Science, Research, and Technology of Iran (MSRT) for nancial support. We also thank Mr. A. Korteweg (Laboratory of Physical and Colloid Chemistry) for the BET analysis.

References
[1] D.L. Sparks, Soil Physical Chemistry, second ed., CRC Press, Boca Raton, FL, 1999, p. 409. [2] G. Gouy, Ann. Phys. (Paris) Ser. 4 (9) (1910) 457. [3] D.L. Chapman, Philos. Mag. 6 (25) (1913) 475. [4] J.N. Israelachvili, Intermolecular and Surface Forces, second ed., Academic Press, London, 1991, p. 450. [5] O. Stern, Z. Electrochem. 30 (1924) 508. [6] D.C. Grahame, Chem. Rev. 41 (3) (1947) 441. [7] D.E. Yates, S. Levine, T.W. Healy, J. Chem. Soc. Faraday Trans. 70 (1974) 1807. [8] D.E. Yates, The Structure of the Oxide/Aqueous Electrolyte Interface, Ph.D. thesis, University of Melbourne, 1975. [9] J.A. Davis, R.O. James, J.O. Leckie, J. Colloid Interface Sci. 63 (1978) 480. [10] P. Fenter, L. Cheng, S. Rihs, M. Machesky, M.J. Bedzyk, N.C. Sturchio, J. Colloid Interface Sci. 225 (1) (2000) 154. [11] M. Borkovec, Langmuir 13 (10) (1997) 2608. [12] R.J. Atkinson, A.M. Posner, J.P. Quirk, J. Phys. Chem. 71 (1967) 550. [13] J.W. Bowden, S. Nagarajah, N.J. Barrow, A.M. Posner, J.P. Quirk, Austral. J. Soil Res. 18 (1) (1980) 49. [14] J. Westall, H. Hohl, Adv. Colloid Interface Sci. 12 (4) (1980) 265. [15] D.A. Dzombak, F.M.M. Morel, Surface Complexation Modeling: Hydrous Ferric Oxide, Wiley, New York, 1990. [16] W. Piasecki, W. Rudzinski, R. Charmas, J. Phys. Chem. B 105 (40) (2001) 9755. [17] N. Sahai, D.A. Sverjensky, Geochim. Cosmochim. Acta 61 (14) (1997) 2801.

R. Rahnemaie et al. / Journal of Colloid and Interface Science 293 (2006) 312321

321

[18] S. Goldberg, G. Sposito, Soil Sci. Soc. Am. J. 48 (1984) 772. [19] A.R. Felmy, J.R. Rustad, Geochim. Cosmochim. Acta 62 (1) (1998) 25. [20] T. Hiemstra, W.H. Vanriemsdijk, M.G.M. Bruggenwert, Neth. J. Agric. Sci. 35 (3) (1987) 281. [21] M. Schudel, S.H. Behrens, H. Holthoff, R. Kretzschmar, M. Borkovec, J. Colloid Interface Sci. 196 (2) (1997) 241. [22] M.L. Machesky, D.J. Wesolowski, D.A. Palmer, K. Ichiro-Hayashi, J. Colloid Interface Sci. 200 (2) (1998) 298. [23] I. Christl, R. Kretzschmar, Geochim. Cosmochim. Acta 63 (1920) (1999) 2929. [24] L. Sigg, W. Stumm, Colloids Surf. 2 (2) (1981) 101. [25] J. Lutzenkirchen, Environ. Sci. Technol. 32 (20) (1998) 3149. [26] K. Bourikas, T. Hiemstra, W.H. Van Riemsdijk, Langmuir 17 (3) (2001) 749. [27] A. Breeuwsma, J. Lyklema, J. Colloid Interface Sci. 43 (2) (1973) 437. [28] W.N. Rowlands, R.W. Obrien, R.J. Hunter, V. Patrick, J. Colloid Interface Sci. 188 (2) (1997) 325. [29] S.B. Johnson, P.J. Scales, T.W. Healy, Langmuir 15 (8) (1999) 2836. [30] M. Kosmulski, J.B. Rosenholm, J. Phys. Chem. 100 (28) (1996) 11,681. [31] J. Gustafsson, P. Mikkola, M. Jokinen, J.B. Rosenholm, Colloids Surf. A Physicochem. Eng. Aspects 175 (3) (2000) 349. [32] M. Kosmulski, E. Matijevic, Colloid Polym. Sci. 270 (10) (1992) 1046. [33] R. Rietra, T. Hiemstra, W.H. Van Riemsdijk, J. Colloid Interface Sci. 229 (1) (2000) 199. [34] R. Rahnemaie, T. Hiemstra, W.H. Van Riemsdijk, J. Colloid Interface Sci. (2005), submitted for publication. [35] D.A. Sverjensky, Geochim. Cosmochim. Acta 65 (21) (2001) 3643. [36] T. Hiemstra, W.H. Van Riemsdijk, J. Colloid Interface Sci. 225 (1) (2000) 94. [37] P.G. Weidler, T. Schwinn, H.E. Gaub, Clays Clay Miner. 44 (4) (1996) 437. [38] P.G. Weidler, S.J. Hug, T.P. Wetche, T. Hiemstra, Geochim. Cosmochim. Acta 62 (2122) (1998) 3407. [39] T. Hiemstra, W.H. Vanriemsdijk, G.H. Bolt, J. Colloid Interface Sci. 133 (1) (1989) 91. [40] T. Hiemstra, P. Venema, W.H. Vanriemsdijk, J. Colloid Interface Sci. 184 (2) (1996) 680.

[41] L. Pauling, J. Am. Chem. Soc. 51 (1929) 1010. [42] J.R. Rustad, A.R. Felmy, B.P. Hay, Geochim. Cosmochim. Acta 60 (9) (1996) 1563. [43] V. Barron, J. Torrent, J. Colloid Interface Sci. 177 (2) (1996) 407. [44] T. Hiemstra, W.H. Vanriemsdijk, J. Colloid Interface Sci. 179 (2) (1996) 488. [45] R. Rietra, T. Hiemstra, W.H. Van Riemsdijk, J. Colloid Interface Sci. 218 (2) (1999) 511. [46] T. Hiemstra, W.H. Van Riemsdijk, On the relationship between surface structure and ion complexation of oxidesolution interfaces, in: A.T. Hubbard (Ed.), Encyclopedia of Surface and Colloid Science, Dekker, New York, 2002. [47] T. Hiemstra, J.C.M. Dewit, W.H. Vanriemsdijk, J. Colloid Interface Sci. 133 (1) (1989) 105. [48] D.A. McInnes, The Principles of Electrochemistry, Dover, New York, 1961. [49] D.G. Kinniburgh, C.J. Milne, P. Venema, Soil Sci. Soc. Am. J. 59 (2) (1995) 417. [50] P. Venema, T. Hiemstra, W.H. Vanriemsdijk, J. Colloid Interface Sci. 183 (2) (1996) 515. [51] P.W. Li, L.K. Koopal, T. Hiemstra, J.C.L. Meeussen, W.H. Van Riemsdijk, Geochim. Cosmochim. Acta 69 (2) (2005) 325. [52] J.F. Boily, J. Lutzenkirchen, O. Balmes, J. Beattie, S. Sjoberg, Colloids Surf. A Physicochem. Eng. Aspects 179 (1) (2001) 11. [53] M. Villalobos, J.O. Leckie, J. Colloid Interface Sci. 235 (1) (2001) 15. [54] A. Vangeen, A.P. Robertson, J.O. Leckie, Geochim. Cosmochim. Acta 58 (9) (1994) 2073. [55] A.P. Robertson, J.O. Leckie, J. Colloid Interface Sci. 188 (2) (1997) 444. [56] D.G. Lumsdon, L.J. Evans, J. Colloid Interface Sci. 164 (1) (1994) 119. [57] D.A. Sverjensky, Geochim. Cosmochim. Acta 69 (2) (2005) 225. [58] T. Hiemstra, W.H. Vanriemsdijk, Colloids Surf. 59 (1991) 7. [59] J.O.M. Bockris, A.K.N. Reddy, Modern Electrochemistry, vol. 1, Plenum, New York, 1970. [60] M. Predota, A.V. Bandura, P.T. Cummings, J.D. Kubicki, D.J. Wesolowski, A.A. Chialvo, M.L. Machesky, J. Phys. Chem. B 108 (32) (2004) 12,049. [61] N. Sahai, S.A. Carroll, S. Roberts, P.A. ODay, J. Colloid Interface Sci. 222 (2) (2000) 198.