Surface complexation model for multisite adsorption of copper(II) onto kaolinite

Geochimica et Cosmochimica Acta, Vol. 69, No. 15, pp. 3733–3745, 2005 Copyright © 2005 Elsevier Ltd Printed in the USA. All rights reserved 0016-7037/05 $30.00 ⫹ .00

doi:10.1016/j.gca.2004.12.029

Surface complexation model for multisite adsorption of copper(II) onto kaolinite CAROLINE L. PEACOCK and DAVID M. SHERMAN* Department of Earth Sciences, University of Bristol, Bristol, BS8 1RJ, United Kingdom (Received July 12, 2004; accepted in revised form December 15, 2004)

Abstract—We measured the adsorption of Cu(II) onto kaolinite from pH 3–7 at constant ionic strength. EXAFS spectra show that Cu(II) adsorbs as (CuO4Hn)n⫺6 and binuclear (Cu2O6Hn)n⫺8 inner-sphere complexes on variable-charge ⬅AlOH sites and as Cu2⫹ on ion exchangeable ⬅X--H⫹ sites. Sorption isotherms and EXAFS spectra show that surface precipitates have not formed at least up to pH 6.5. Inner-sphere complexes are bound to the kaolinite surface by corner-sharing with two or three edge-sharing Al(O,OH)6 polyhedra. Our interpretation of the EXAFS data are supported by ab initio (density functional theory) geometries of analog clusters simulating Cu complexes on the {110} and {010} crystal edges and at the ditrigonal cavity sites on the {001}. Having identified the bidentate (⬅AlOH)2Cu(OH)20, tridentate (⬅Al3O(OH)2)Cu2(OH)30 and ⬅X--Cu2⫹ surface complexes, the experimental copper(II) adsorption data can be fit to the reactions 2 ⬅ AlOH ⫹ Cu2⫹ ⫹ 2H2O ⫽ 共⬅AlOH兲2Cu共OH兲2 ⫹ 2H⫹ 0

3共⬅AlOH兲 ⫹ 2Cu2⫹ ⫹ 3H2O ⫽ 共⬅Al3O(OH)2兲Cu2共OH兲3 ⫹ 4H⫹ 0

and ⬅X⫺--H⫹ ⫹ Cu2⫹ ⫽ ⬅ X⫺--Cu2⫹ ⫹ X⫹.

(A1)

Copyright © 2005 Elsevier Ltd bonding and thus there are no interlayer spaces. Isomorphic substitution of Al3⫹ for Si4⫹ in the tetrahedral sheet (e.g., van Olphen, 1977) results in a small fixed negative charge within the siloxane layer, which accounts for the small cation exchange capacity (CEC) of kaolinite (generally ⬍0.02 mol charge/kg in chemically pure kaolinites; Talibudeen, 1981). On the face of the tetrahedral silica sheet, permanent negative charge is located at the ditrigonal siloxane cavities (e.g., Davis and Kent, 1990) and is commonly represented as ⬅X⫺ surface sites. Kaolinite CEC has also been suggested to occur due to the presence of a small amount of aluminosilicate gel coating (Ferris and Jepson, 1975) and/or contamination by small amounts of 2:1 phyllosilicate minerals (e.g., Bolland et al., 1976, 1980; Lim et al., 1980). In addition to permanent, negatively charged ⬅X⫺ sites, kaolinite has amphoteric, variablecharge ⬅SOH sites at the crystal edges and on the octahedral alumina sheet. ⬅SOH sites represent silanol ⬅SiO(H) sites at the crystal edges (contributing only to the negative charge through the formation of ⬅SiO⫺) and aluminol ⬅AlOH(H) sites at the crystal edges and on the octahedral sheet (both protonation and deprotonation can occur to form ⬅AlOH2⫹, ⬅AlOH, and ⬅AlO⫺). ⬅Al2OH(H) sites also exist at the gibbsite basal plane and contribute to the positive surface charge in the lowest pH regime through the formation of ⬅Al2OH2⫹ (Huertas et al., 1998). Previous studies examining the interaction between Cu(II) and kaolinite have tended to focus on either the modeling of adsorption behavior displayed in experimental adsorption edges and isotherms, or direct spectroscopic investigation of the metal-mineral association. Two types of modeling approach have been followed: surface complexation modeling (SCM) and a more empiric consideration involving the use of Langmuir or similar equations to

1. INTRODUCTION

The aqueous geochemistry of copper can be strongly controlled by sorption onto iron and manganese (hydr)oxides and clay minerals. In soils, copper is concentrated into the clay fraction (Le Riche and Weir, 1963) presumably by sorption onto clay and colloidal FeOOH phases. Kaolinite, montmorillonite and illite are the most common clay minerals in soil systems (Du et al., 1997). Whether copper is associated with the phyllosilicate minerals vs. the iron hydroxides is unclear. The copper-kaolinite association and mechanism of retention in altered copper-bearing rocks is also undetermined. In the Rakha-Chapri mining block (Singhbhum copper belt, India) supergene clay alteration products extend down to ⬃60 m and are comprised predominantly of kaolinite. Tenginkai et al. (1991) reported copper as a component of the kaolinitic clay minerals and suggested association via surface adsorption and/or lattice binding. Mookherjee and Tenginkai (1987) also reported fixation of substantial Cu by supergene clay minerals. The exact mechanism of retention is ambiguous. This study presents a surface complexation model for the adsorption of copper on kaolinite that is consistent with spectroscopic data and ab initio calculations. A surface complexation model based on actual surface species will be more reliable when applied to modeling reactive transport of Cu in complex natural systems. Kaolinite is a 1:1 aluminosilicate consisting of a tetrahedral silica sheet bonded to an octahedral alumina sheet via the sharing of oxygen atoms between silicon and aluminium atoms in neighboring sheets. Consecutive layers stack by hydrogen

* Author to whom correspondence ([email protected]).

should

be

addressed 3733

3734

C. L. Peacock and D. M. Sherman

describe adsorption data (e.g., Spark et al., 1995; Majone et al., 1996). Using a SCM framework, the adsorption of Cu(II) to kaolinite has been successfully described by adsorption onto the variable-charge ⬅SOH sites (e.g., Jung et al., 1998). Several authors, however, have noted that adsorption of heavy metals onto various clay minerals occurs in two or more stages: a first sorption process at low pH due to ion exchange with the permanent, negatively charged ⬅X⫺ sites, and a second step at higher pH due to inner-sphere complexation on the variablecharge ⬅SOH sites. Schindler et al. (1987), Angove et al. (1998) and Ikhsan et al. (1999) have successfully modeled the adsorption of Cu(II) to kaolinite by invoking (⬅X2)2⫺--Cu2⫹ ion exchange complexes at low pH and inner-sphere (⬅SO)2Cu0 and/or ⬅SOCu⫹ surface complexes on the variable-charge ⬅SOH sites at higher pH. These previous modeling studies, however, invoke outer-sphere ion exchange complexes within the constant capacitance surface complexation model (CCM). The CCM has only one electrostatic plane for adsorption and thus does not technically allow for outer-sphere complexation. Using a SCM approach, Majone et al. (1996) modeled ⬅X⫺--Cu2⫹ ion exchange complexes. There have been few direct spectroscopic investigations of Cu(II)-kaolinite adsorption. Using electron spin resonance (ESR), McBride (1976, 1978) reported an even distribution of sorption sites over basal planes and crystal edges and planar Cu(H2O)42⫹ complexes aligned parallel to planar kaolinite surfaces. Using EXAFS spectroscopy, the presence of small multinuclear clusters bound by inner-sphere complexation has been reported for the adsorption of Co(II) on kaolinite and SiO2 (e.g., O’Day et al., 1991, 1994) and for the adsorption of Pb(II) on ␥-Al2O3 (Chisholm-Brause et al., 1990). Following our development of a SCM constrained by results from spectroscopy for Cu(II) adsorption to goethite, hematite and lepidocrocite (Peacock and Sherman, 2004), we fit our kaolinite sorption edge and isotherms based on surface species determined from EXAFS. The interpretation of EXAFS spectra is aided using first-principles (density functional theory) calculations of surface complex geometries. We report the mechanism by which copper(II) sorbs to kaolinite and determine realistic surface complex stability constants that can be used for geochemical modeling. 2. EXPERIMENTAL METHODS 2.1. Mineral Preparation and Characterization Kaolinite from the ECC International Lee Moor China Clay Pit, Cornwall, UK, was used without further treatment. Mineral identity and purity was confirmed by X-ray powder diffraction (XRD) analysis of randomly orientated powder samples. Additional characterization is given by Graveling et al. (1997). Surface area was measured by BET to be 12.2 ⫾ 3 m2/g. SEM revealed a typical pseudohexagonal-plate crystal morphology, with crystallites typically 1–2 ␮m diameter. In some cases, the crystal edges appeared rough and chipped. 2.2. Potentiometric Titration Potentiometric titrations of a kaolinite suspension were carried out at three salt concentrations (0.01 M, 0.03 M and 0.1 M NaNO3) following the method of Hayes et al. (1991). Dried solid kaolinite (0.25 g) was suspended in 50 mL of preboiled, nitrogen-purged (⬍ 1 ppm CO2 (g)) 18.2 m⍀ Milli-Q water and further purged with nitrogen (⬍ 1 ppm CO2 (g)) overnight before titrations. Initial pH after overnight purging was approximately pH 5. Titrations were performed at 25°C in an air-tight reactor with constant stirring to prevent settling. Base (0.1 M NaOH,

free from carbonate), acid (0.1 M HNO3) and salt solutions (3 M NaNO3) were prepared from stock solutions and added via an automated titrator. A nitrogen atmosphere (⬍ 1 ppm CO2 (g)) was maintained throughout the experiment. Electrolyte was added to adjust the ionic strength to 0.01 M and acid then added to gradually lower the pH to approximately pH 3. Incremental addition of base then produced a titration from approximately pH 3–10. After each incremental addition of base, up to 10 min were allowed for pH equilibration (or drift below 0.5 mV/min). The suspension was returned to pH ⬃ 3 by reverse acid titration, electrolyte added to adjust the ionic strength to the next level and the titration repeated following the same method. The titration was completed within 24 h and we observed no significant hysteresis between the acid and base titration legs. We used a pin-tip double junction glass combination electrode (Sentek) with a salt bridge of 3 M NaNO3. The electrode was calibrated potentiometrically following the method of Gans and O’Sullivan (2000). Dissolved Al and Si were measured after titration by inductivelycoupled plasma atomic emission spectrometry (ICP-AES) to assess the contribution of dissolution to the proton balance. In agreement with Brady et al. (1996), dissolved levels of Al and Si were too low (⬍1 ppm; ⬍4 ⫻ 10⫺5 mol/L) to significantly effect the charge balance equations. The base leg of the titrations are reported here and used to optimize acid-base parameters for use in mineral-copper surface complexation modeling. 2.3. Sample Synthesis Kaolinite batch experiments were prepared with copper (II) aqueous solution using AR grade reagents and 18.2 m⍀ Milli-Q water. All solutions and resulting experimental suspensions were purged with Ar (g) or N2 (g) (⬍ 1 ppm CO2 (g)) and all adsorption experiments were conducted at 25°C. pH measurements were calibrated to ⫾ 0.05 pH units using Whatman NBS grade buffers. 2.3.1. pH adsorption edge experiments Copper (II) stock solution was prepared at 100 ppm from Cu(NO3)2 · 3H2O. Adsorption pH experiments at 25 ppm [Cu]total were prepared by adding 7.5 mL of 100 ppm Cu stock solution to 0.1 g kaolinite in 22.5 mL of 0.1 M NaNO3. Sorbent concentration in solution was therefore 3.33 g/L. The resulting suspensions were constantly stirred and initial pH was recorded after stabilization to two decimal places. Suspension pH was then varied from pH 3–7 by the dropwise addition (⬍ 1 mL) of HNO3/NaOH and left to equilibrate for 10 min. Suspension pH was then recorded after stabilization to two decimal places. Adsorption pH experiments were then sealed under N2 (g) (⬍ 1 ppm CO2 (g)) and shaken continuously for 1 week. The pH was measured again at the end of the week; values changed by ⬍0.3 pH units. Adsorption of Cu to kaolinite at 25 ppm [Cu]total was investigated with EXAFS spectroscopy of specific samples from the adsorption edge at pH ⬃ 4.4, 5.8 and 6.5. Kaolinite samples at pH ⬃ 4.4, 5.8 and 6.5 contained 0.12, 0.42 and 0.66 wt% copper with estimated surface coverage (calculated from the optimized total active surface site density, Table 2) at 7.6, 27.1 and 42.6% respectively. 2.3.2. Constant-pH isotherm experiments Constant pH isotherms were measured at pH 5.8, 6.5 and 7.0 and prepared under N2 (g) (⬍ 1 ppm CO2 (g)). Copper II stock solution was prepared at 100 ppm from Cu(NO3)2 · 3H2O. Kaolinite constant pH experiments were prepared by adding 3–15 mL of 100 ppm Cu stock solution to 0.1 g kaolinite in 27–15 mL of 0.1 M NaNO3. Sorbent concentration in solution was therefore 3.33 g/L, and [Cu]total ranged from 10 –50 ppm. The resulting suspensions were constantly stirred and initial pH was recorded after stabilization to two decimal places. Suspension pH was then adjusted to that of the experiment by the dropwise addition (⬍ 1 mL) of NaOH and left to equilibrate for 10 min. Suspension pH was then recorded after stabilization to two decimal places. The resulting 30 mL batch suspensions were sealed under N2 (g) (⬍ 1 ppm CO2 (g)) and shaken

Multisite sorption of Cu on kaolinite

3735

continuously for 1 week. The pH was measured again at the end of the week; values changed by ⬍0.3 pH units. Batch adsorption samples were separated by centrifugation (5000 rpm for 10 –15 min) into an adsorption sample (thick paste) for spectroscopic analysis and a clear supernate for determination of total copper concentration. Supernates were filtered using 0.2 ␮m cellulose nitrate membrane filters, acidified with 1% HNO3 and analyzed for copper by ICP-AES. All adsorption samples were spectroscopically analyzed either immediately after centrifugation or after storage at 1 to 4°C for a maximum of 48 h. 2.4. EXAFS Data Collection and Analysis 2.4.1. Data collection EXAFS fluorescence spectra of the copper K edge (8.979 keV) were collected on station 16.5 at the CLRC Synchrotron Radiation Source, Daresbury Laboratory, UK. Adsorption samples were presented to the X-ray beam as a wet paste held by Sellotape in a 2 mm-thick Teflon slide with a 4 ⫻ 15 mm sample slot. During data collection, storage ring energy was 2.0 GeV and the beam current varied between 130 and 240 mA. The monochromator was set to reject 50% of the incoming beam to minimize higher harmonics in the EXAFS spectrum. EXAFS data were collated from up to 10 fluorescence mode scans using an Ortec 18-element solid state detector.

Fig. 1. Speciation of copper(II) as a function of pH. [Cu]total ⫽ 3.94 ⫻ 10⫺4 molal (⬃25 ppm) in 0.1 M NaNO3.

optimized using a Newton-Raphson method and Broydon-Fletcher update of the Hessian matrix as coded in ADF 2.0. During the geometry optimizations the total energies were converged to ⫾5 kJ/mol. 2.6. Surface Complexation Modeling

2.4.2. Data analysis EXAFS data reduction was performed using Daresbury Laboratory software (EXCALIB, and EXBACK, Dent and Mosselmans, 1992). EXCALIB was used to calibrate from monochromator position (millidegrees) to energy (eV) and to average multiple spectra from individual samples. EXBACK was used to define the start of the EXAFS oscillations (determined from the inflection point on the K edge) and perform background subtraction. The preedge was fit to a linear function and the postedge background to two second-order polynomial segments. EXAFS were fit in the small atom approximation and we allowed for multiple scattering as coded in EXCURV98 (Binsted, 1998). The phase-shift functions used in the curve fitting were derived by ab initio methods in EXCURV98 using Hedin-Lundqvist potentials (Hedin and Lundqvist, 1969) and von Barth ground states. No Fourier filtering was performed during the data analysis. The inclusion of multiple scattering improved the fit in the 3.3– 4.5 Å region where some of the features result from O-O scattering within the square planer CuO46⫺ clusters and scattering involving the next nearest neighbor atoms (Al and Si). Multiple scattering calculations require specification of the full three dimensional structure of the Cu coordination environment (i.e., bond angles in addition to bond lengths). This was done using a hypothetical model cluster with C1 symmetry. Note that the multiple-scattering contributions were calculated self-consistently during the EXAFS fits. Multiple scattering path lengths were limited to 10 Å.

The program FITEQL v3.2 (Herbelin and Westall, 1996) was used to fit the acid-base behavior of the kaolinite surface and subsequently the adsorption behavior of copper on kaolinite to a surface complexation model. The extended constant capacitance model (ECCM (Nilsson et al. (1996)) was used to account for surface electrostatics. FITEQL is used extensively for the calculation of chemical equilibrium constants in metal adsorption studies (e.g., Robertson and Leckie, 1998; Tadanier and Eick, 2002; Peacock and Sherman, 2004) including the fitting of potentiometric titration data and adsorption data involving minerals with permanent negative charge such as kaolinite (e.g., Schindler et al., 1987; Angove et al., 1998; Ikhsan et al., 1999). The quality of the fits produced is given by: V(Y) ⫽ (Y ⁄ SY)2 ⁄ (np ∗ nII ⫺ nu)

(1)

where Y is the actual error in the mass balance equation, SY is the estimated experimental error given by FITEQL and the reciprocal of the variance SY is the weighting factor. np is the number of data points, nII is the number of chemical components with known total and free concentrations, and nu is the number of adjustable parameters (Lumsdon and Evans, 1994; Gao and Mucci, 2001). A good fit to experimental metal binding data are indicated by a value of V(Y) between 0.1 and 20 (Herbelin and Westall, 1996). 3. RESULTS AND DISCUSSION

3.1. Sorption of Cu2ⴙ on Kaolinite 2.5. Density Functional Calculations Quantum mechanical calculation of cluster geometries and energies were performed using the ADF 2.0 code (Velde et al., 2001) which implements density functional theory for finite clusters and molecules using the linear combination of atomic orbital formalism. Molecular orbitals in the ADF code are constructed from Slater-type atomic orbitals, consisting of a Cartesian part rkrxkxykyzkz with kx ⫹ ky ⫹ kz ⫽ l (l ⫽ angular momentum quantum number) and an exponential part e⫺␣r. Density functional theory allows a very large basis set to be used: for all atoms an uncontracted, triple-zeta basis set with polarization functions (i.e., 1s2s2p3s3s=3s⬙3p3p=3p⬙ ⫹ 3d for aluminium and silicon, 1s2s2s=2s⬙2p2p=2p⬙ ⫹ 3d for oxygen, 1s2s2p3s3p3d3d=3d⬙4s4s=4s⬙ ⫹ 4p for copper and 1s1s=1s⬙ ⫹ 2p for hydrogen) was used. The charge density was also fit to a Slater-type orbital basis set. For all atoms except hydrogen, frozen core orbitals (i.e., 1s, 2s, 2p for Al and Si; 1s for O and 1s, 2s, 2p, 3s and 3p for Cu) were used. The Vosko et al. (1980) parameterization for the local exchangecorrelation functionals together with generalized gradient corrections of Perdew et al. (1992) was used. The geometries of the clusters were

3.1.1. Aqueous speciation of Cu2⫹ The aqueous speciation of Cu2⫹ at 25 ppm [Cu]total (calculated by suppressing the formation of CuO (s)) is shown in Figure 1 as a function of pH. In 0.1 M NaNO3 between pH ⬃ 2– 6.6, Cu(II) occurs predominantly as the Cu2⫹ aqueous cation (in agreement with Baes and Mesmer, 1976). Above pH ⬃ 6.6, Cu(OH)2 (s) limits the concentration of aqueous Cu2⫹ and is the major form of Cu(II) present in the system. 3.1.2. Adsorption pH edge data At 25 ppm [Cu]total and between pH ⬃ 2– 6.6, Cu(II) likely sorbs as Cu2⫹ (aq) and we find a sigmoid adsorption edge for kaolinite (Fig. 2). The shape of our adsorption edge is in good agreement with previous studies of Cu2⫹ adsorption onto kaolinite (e.g., Ikhsan et al., 1999) where significant sorption

3736

C. L. Peacock and D. M. Sherman

Fig. 2. Adsorption of copper(II) ions to kaolinite (3.33 g/L) as a function of pH at I ⫽ 0.1 M NaNO3 and 25°C, after 1 week equilibration time with 25 ppm [Cu]total. Symbols are data points, lines are ECCM fits showing total and individual surface species (large dashed line ⫽ mononuclear complex; small dashed line ⫽ binuclear complex and dotted line ⫽ ⬅X⫺--Cu2⫹ complex).

Following Karthikeyan and Elliott (1999), we plot data points (square) on the kaolinite constant pH isotherms (Fig. 3) corresponding to the adsorption pH edge condition at pH ⬃ 5.8, 6.5 and 7.0 represented in Figure 2. At pH 5.8, the experimental constant pH isotherm data and adsorption pH edge condition are consistent with the thermodynamic data presented in Figure 1 showing no Cu(OH)2 (s) precipitation at pH 5.8. At pH 6.5, the adsorption pH edge condition is outside the region of Cu(OH)2 (s) precipitation and is thus consistent with the precipitation of Cu(OH)2 (s) after ⬃30 –35 ppm [Cu]total. At pH 7.0, the adsorption pH edge condition is within the region of Cu(OH)2 (s) precipitation and is thus consistent with the precipitation of Cu(OH)2 (s) after ⬃ 20 ppm [Cu]total. However, at pH 7.0, precipitation is at a higher pH than our last EXAFS spectra of adsorption pH edge data and is present only at the very final adsorption pH edge regime. In 0.1 M NaNO3 background electrolyte and similar [Cu]total (35 ppm; 5.44 ⫻ 10⫺4 M), Du et al. (1997) have included the

(⬃15%) was observed in the low pH regime (pH ⬃ 3– 4) with 0.005 M KNO3 background electrolyte. Spark et al. (1995) reported an apparent dependence of adsorption in the low pH regime on KNO3 background electrolyte concentration, with the first adsorption process almost completely absent at higher ionic strengths. Adsorption in the low pH regime of our edge data (Fig. 2), measured in 0.1 M NaNO3 background electrolyte, is presumably a reflection of i) using a NaNO3 rather than KNO3 background electrolyte and ii) the ratio of background to adsorbing cation used (since the first adsorption process is likely to be due to ion exchange). Spark et al. (1995) reported that, at intermediate background electrolyte concentrations, electrolytes containing K⫹ suppressed the first adsorption process more than those containing Na⫹ (i.e., K⫹ ions compete more effectively with the adsorbing metal ions than Na⫹ ions). Ikhsan et al. (1999) used K⫹: Cu2⫹ ⫽ 50:1 and reported significant adsorption in the low pH regime whilst Spark et al. (1995) at 0.1 M background electrolyte used K⫹:Cu2⫹ ⫽ 1000:1 and reported an absence of adsorption in the low pH regime. Our experimental adsorption pH edge conditions (using NaNO3 at Na⫹:Cu2⫹ ⫽ 250:1) are therefore more permitting of the first adsorption process. 3.1.3. Constant pH isotherm data Constant pH isotherms (Fig. 3) were measured at pH 5.8, 6.5 and 7.0 and thus correspond to adsorption at the mid, upper and final points of the pH edge experiment respectively. Data are plotted as final aqueous [Cu] (log mol/L) against the surface density of adsorbed ions, ⌫ (log mol/m2). Saturation of CuO (s) and Cu(OH)2 (s) is predicted to occur when log [Cu2⫹] mol/L is ⫺5.8 (⬃0.1 ppm) and ⫺4.8 to ⫺4.5 (⬃1–2 ppm), respectively. At pH 5.8 we find no Cu(OH)2 (s) precipitation and this is consistent with the thermodynamic data presented in Figure 1. At pH 6.5 we find precipitation of Cu(OH)2 (s) after ⬃30 –35 ppm [Cu]total. The predominance of Cu(OH)2 (s) precipitate ⬎25 ppm [Cu]total is consistent with the thermodynamic data presented in Figure 1. We find no Cu(OH)2 (s) precipitation at our adsorption pH edge [Cu]total. At pH 7.0 we find precipitation of Cu(OH)2 (s) after ⬃20 ppm [Cu]total.

Fig. 3. Adsorption of copper(II) ions to kaolinite (3.33 g/L) at pH 5.8, 6.5 and 7.0 (constant), I ⫽ 0.1 M NaNO3, and 25°C, after 1 week equilibration time with 10 –50 ppm [Cu]total. Plotted as final aqueous [Cu] (log mol/L) against the surface density of adsorbed ions, ⌫ (log mol/m2). Symbols are data points, lines are ECCM fits showing total surface species.

Multisite sorption of Cu on kaolinite

3737

Fig. 4. (a) EXAFS and (b) Fourier transform of EXAFS for Cu(II) on kaolinite adsorption samples equilibrated with 25 ppm [Cu]total.

formation of bulk and/or surface precipitation of Cu(OH)2 (s) in their surface complexation model to best describe the observed Cu2⫹ adsorption after pH ⬃ 6.6. Their modeling results indicate the onset of bulk and surface Cu(OH)2 (s) precipitation to be at the very end of our adsorption edge (at pH ⬃ 6.8 and ⬃ 7.1 respectively) with predominance of these products occurring at a higher pH regime than investigated in this study (at pH ⬃ 7.2 and 8.1 respectively). In the upper pH range of our adsorption edge, Du et al. (1997) report the predominance of multinuclear surface complexes and this is consistent with our EXAFS results. Presumably, the occurrence of Cu(OH)2 (s) after pH 6.6 is a result of the somewhat higher [Cu]total used in the adsorption pH edge experiments of Du et al. (1997) (this is consistent with our isotherm data at pH 6.5 where the precipitation of Cu(OH)2 (s) appears to occur at [Cu]total ⬃ 35 ppm). As discussed below, EXAFS spectra are consistent with the absence of precipitation on the mineral surface, at least up to pH 6.5, in the pH edge experiments. Our pH edge data can, therefore, be used to develop a surface complexation model rather than a model including bulk/surface precipitation. 3.1.4. Cu K-Edge EXAFS spectroscopy and Ab initio molecular geometries Cu K-edge EXAFS (and Fourier transforms of the EXAFS) for wet-paste kaolinite adsorption samples are shown in Figure 4 and summarized in Table 1. Note, again, that we are fitting the spectra in terms of single-atom shells in a cluster with C1 symmetry to allow for self-consistent inclusion of multiple scattering. At pH ⬃ 5.8 and 6.5 we find the copper first-shell coordi-

nation environment to have 4.0 O at 1.85–2.05 Å consistent with the protonated square-planar (CuO4Hn)n⫺6 ion. The fourfold (vs. six-fold) coordination is expected given the JahnTeller distortion of the d9 Cu2⫹ ion. Inclusion of an axial oxygen with a larger distance (2.29 Å) and Debye-Waller factor than the equatorial oxygens in the Cu coordination shell gives a slight improvement to the fits. The range of Cu-O distances and small Debye-Waller factors of the first four O neighbors may be an artifact of fitting the four oxygens to four distinct shells. Attempts to constrain the Cu-O distances to be equal (but with larger Debye-Waller factors) gave less satisfactory fits. The ab initio geometries (discussed below) predict that the four shortest Cu-O bond lengths in the surface complexes at pH ⬃ 5.8 and 6.5 range from 1.99 to 2.08 Å. Beyond the oxygen shells, we find 0.3 (at pH 5.8) and 0.5 (at pH 6.5) next-nearest-neighbor atoms (Cu) at 2.94 and 2.96 Å respectively. We interpret the ⬃ 3.0 Å distance to result from polymerization of (CuO4Hn)n⫺6 complexes to give (Cu2O6Hn)n⫺8 dimers. We find additional next-nearest-neighbor shells corresponding to 2 Al atoms at 3.69 and 3.72 Å (at pH 5.8) and 3.69 and 3.74 Å (at pH 6.5). We interpret these distances as resulting from bidentate corner-sharing between (CuO4Hn)n⫺6 complexes and edge-sharing Al(O,OH)6 polyhedra (Fig. 5a). The features in the Fourier transform of the EXAFS at distances greater than 3.3 Å appear to result from multiple scattering. To help verify the structural model for the (CuO4Hn)n⫺6 surface complex, we calculated the optimized geometries for clusters analogous to bidentate mononuclear corner-sharing surface complexes using density functional theory. The domi-

3738

C. L. Peacock and D. M. Sherman Table 1. EXAFS fits for Cu(II) sorbed to kaolinite.a

(a) Mononuclear and binuclear complex pH (wt %Cu) 5.8 (0.42) 6.5 (0.66)

NO R(Cu-O1) (2␴2)

NO R(Cu-O2) (2␴2)

NO R(Cu-O3) (2␴2)

NO R(Cu-O4) (2␴2)

NO R(Cu-O5) (2␴2)

NCu R(Cu-Cu) (2␴2)

NAl R(Cu-Al1) (2␴2)

NAl R(Cu-Al2) (2␴2)

1.0 1.85 (0.002) 1.0 1.89 (0.002)

1.0 1.95 (0.002) 1.0 1.93 (0.002)

1.0 1.97 (0.002) 1.0 1.99 (0.002)

1.0 2.05 (0.002) 1.0 2.00 (0.002)

1.0 2.29 (0.015) 1.0 2.29 (0.015)

0.3 2.94 (0.008) 0.5 2.96 (0.008)

1.0 3.69 (0.014) 1.0 3.69 (0.007)

1.0 3.72 (0.014) 1.0 3.74 (0.015)

X2 (R %) 1.3 (16.6) 3.62 (32.5)

(b) Mononuclear ditrigonal cavity complex

pH (wt % Cu)

NO R(Cu-O1) (2␴2)

NO R(Cu-O2) (2␴2)

NO R(Cu-O3) (2␴2)

X2 (R %)

4.4 (0.12)

1.0 1.92 (0.015)

1.0 2.04 (0.015)

0.50 2.28 (0.015)

1.77 (20.8)

a

NA is number of atoms of type A; R is distance (Å); 2␴2 is Debye-Waller factor (Å2).

nant aqueous complex below pH ⬃ 7 is Cu(H2O)42⫹; hydroxyl complexes are unimportant (Fig. 1). However, the ab initio geometries predict that, if the coordination number of Cu is 4 (as indicated by the EXAFS), the surface complexes must be (⬅AlOH)2Cu(OH)2 (Fig. 5a) and not (⬅AlOH)2Cu(OH2)2 or (⬅AlO)2Cu(OH2)2. Protonation of the (⬅AlOH)2Cu(OH)2 complex leads to the release of two water molecules to give two-fold coordinated Cu (i.e., (⬅AlOH)2Cu ⫹ 2H2O). The (⬅AlO)2Cu(OH2)2 tautomer of (⬅AlOH)2Cu(OH)2 is higher in energy by 0.19 eV (18 kJ/mol). Analogous surface complexes (i.e., (⬅FeOH)2Cu(OH)2) were also found for Cu2⫹ sorbed on iron(III) (hydr)oxides (Peacock and Sherman, 2004). We hypothesize that the Lewis acidity of Cu2⫹ is enhanced by surface complexation. The ab initio predicted bond lengths (Fig. 5a) are in reasonable agreement with those observed via EXAFS for Cu adsorption on kaolinite (Table 1a). Bidentate mononuclear (and tridentate binuclear) cornersharing surface complexes analogous to those predicted with ab initio calculations are able to occur on the {110} and {010} faces of kaolinite (setting C1). The EXAFS data show no evidence for Cu(OH)2 precipitation at pH 6.5. The absence of Cu(OH)2 precipitation at the [Cu]total used in our edge experiment is also indicated by the sorption isotherm at pH 6.5 discussed above. As will be shown below, the surface complexation model obtained by fitting the sorption edges is consistent with the complexes shown in Figure 5. At pH ⬃ 4.4, we find the copper first-shell coordination environment to have 2.0 O at 1.92 and 2.04 Å. We find 0.5 O at 2.28 Å. The low coordination number under acidic conditions is in agreement with predictions from ab initio cluster calculations: when the two OH ligands on a (⬅Si2O)Cu(OH)2 surface complex are protonated, one of the resulting H2O molecules dissociates from the sorbed Cu to leave a (⬅Si2OCu(H2O) surface complex (Fig. 5b). The best fit to the EXAFS data were achieved with 3 (O) shells only; including an additional 3 next-nearest-neighbor Si atoms near the predicted distances gave no statistical improvement to the fit. The ab initio predicted bond lengths (Fig. 7b) are in reasonable agreement with those observed via EXAFS for Cu adsorption on kaolinite (Table 4b 1b).

3.2. Surface Complexation Modeling 3.2.1. Equilibria at the mineral surface The kaolinite mineral surface is modeled using a multisite model, similar to that used in the previous studies of Schindler et al. (1987), Angove et al. (1998) and Ikhsan et al. (1999). Two types of active surface functional group are assumed: permanent, negatively charged ⬅X⫺ sites on the tetrahedral silica sheet and amphoteric, variable-charge ⬅SOH sites at the crystal edges and on the octahedral alumina sheet. The ⬅X⫺ sites are assumed to undergo an exchange reaction with ions of the background electrolyte at higher pH: ⬅X⫺ -- H⫹ ⫹ Na⫹ ⫽ ⬅ X⫺ -- Na⫹ ⫹ H⫹ ⫺

log Kie

(2)

⫹

where ⬅X --Na represents a sodium ion from the background electrolyte electrostatically bound to a permanent negatively charged ⬅X⫺ site via outer-sphere complexation. The amphoteric adsorption site may exist in one of three protonation states; ⬅SOH2⫹, ⬅SOH and ⬅SO⫺. Surface acidity constants are assigned to the reactions: ⬅SOH ⫹ H⫹ ⫽ ⬅ SOH2⫹ ⬅SOH ⫽ ⬅ SO⫺ ⫹ H⫹

log Ka1 log Ka2

(3) (4)

where S is a nonspecific surface metal ion and ⬅SOH2⫹, ⬅SOH and ⬅SO⫺ are representative surface species. No interaction between the species adsorbed on the two different adsorption sites is considered. The amphoteric treatment of a single surface site is generally recognized as a convenient modeling framework rather than a precise representation of actual functional groups existing at the mineral surface (Rustad et al., 1996). For kaolinite, a crystallographic consideration of the mineral surface shows both silanol (⬅SiOH) and aluminol (⬅AlOH) amphoteric sites, represented in the modeling framework as simply ⬅SOH. As such, reactions (3) and (4) underestimate somewhat the complexity of a mineral surface. However, in agreement with previous studies of adsorption on kaolinite (e.g., Schindler et al., 1987; Angove et al., 1998; Huertas et al., 1998; Ikhsan et

Multisite sorption of Cu on kaolinite

Fig. 5. Cu(II) ab initio molecular geometry clusters. (a) Ditrigonal cavity complex. (b) Bidentate corner-sharing mononuclear cluster. Bond lengths in Å. Both clusters give Cu-Al and Cu-O bond lengths in reasonable agreement with those observed in the EXAFS.

3739

3740

C. L. Peacock and D. M. Sherman

al., 1999) it is likely that ⬅SOH groups involved in adsorption are predominantly ⬅AlOH groups, able to both protonate and deprotonate as represented by reactions (3) and (4). Huertas et al. (1998), considering only amphoteric, variable-charge surface sites, have attributed the charging behavior of kaolinite to acidic Al2OH2⫹ sites in the lowest pH regime, weak acidic AlOH2⫹ sites between pH ⬃ 3– 6, weak basic SiO⫺ sites at pH ⱖ6 and finally basic AlO⫺ sites at pH ⱖ9. Thus in the pH range of our potentiometric titrations (pH ⬃ 3–10) and our adsorption pH edge (pH ⬃ 3–7) and constant pH isotherm (pH 5.8, 6.5 and 7.0) experiments, surface charge due to amphoteric surface groups is likely to be determined predominantly by AlOH sites. An extended version of the CCM (Schindler and Gamsjager, 1972; Stumm et al., 1976; Hohl and Stumm, 1976) was used to describe the electric double layer properties of the mineral surface. The extended constant capacitance model (ECCM, Nilsson et al., 1996) allows both inner- and outer-sphere complexation at the mineral surface via two planes of adsorption: a surface plane for strongly bound ions, including H⫹ (innersphere complexation), and a ␤-plane for weakly bound ions (outer-sphere complexation). Constant capacitance values C1 and C2 are assigned to the inner layer (between the surface plane and the ␤-plane) and outer layer (between the ␤-plane and the bulk solution) respectively and are related to the total capacitance Ctot by the equation for two capacitances in series (Westall, 1986): 1 ⁄ Ctot ⫽ 1 ⁄ C1 ⫹ 1 ⁄ C2

(5)

When only inner-sphere complexation is considered, the ECCM becomes the traditional CCM (with the specific capacitance equal to Ctot) and equilibrium constants determined in the CCM can therefore be included in the new model providing the specific capacitance is equal to Ctot (Nilsson et al., 1996). The ECCM was recently used by Lackovic et al. (2003) to successfully describe the adsorption of Cd(II) onto kaolinite with outer-sphere adsorption onto ⬅X⫺ sites [2(⬅X⫺--K⫹) ⫹ Cd2⫹ ⫽ (⬅X2)2⫺--Cd2⫹ ⫹ 2K⫹] and inner-sphere adsorption onto ⬅SOH sites [2⬅SOH ⫹ Cd2⫹ ⫽ (SO)2Cd ⫹ 2H⫹]. Mineral surface area was determined by BET analysis. Active total surface site density (SSD) was determined by fitting potentiometric titration data rather than fixed at a crystallographically determined value. This was done based on our SEM images, where crystallites appear smaller (1–2 ␮m vs. ⬃10 ␮m diameter) and with rougher edges than an ideal kaolinite that would be used to determine a crystallographic SSD. Upon determining a total active SSD (from fitting potentiometric titration data), a value greater than the crystallographically determined total number of surface sites can be rationalized by increasing the BET surface area measurement (as FITEQL works with the concentration of active surface sites in moles/ L). Total active SSD therefore need not be unreasonably high if the BET surface area estimate is adjusted accordingly. An underestimation of the surface area of kaolinite by the BET method is argued by Xie and Walther (1992) who state that the gas molecules are too large to access small channels along the crystal edges where acid-base reactions may take place. (These authors predict the “real” kaolinite surface area to be 10 –20 times greater than the BET surface.) Surface complexation involving ions of the background electrolyte was not considered.

3.2.2. Modeling potentiometric titration data Proton adsorption on kaolinite was successfully described by the ECCM considering two types of adsorption sites (with fixed and variable charge, respectively). The multisite ECCM has six adjustable model parameters: a total SSD, three acidity constants (two surface acidity constants for SOH sites and one equilibrium constant for the ion exchange X⫺ site; reactions (3), (4) and (2) respectively) and two capacitances (C1 and C2, related via 1/Ctot ⫽ 1/C1 ⫹ 1/C2). Attempts to simultaneously fit all parameters did not converge; thus we adopted the modeling approach of Hayes et al. (1991). Briefly, an initial estimate for total SSD was obtained by considering the minimum number of total sites required to account for the maximum adsorption displayed by our adsorption pH edge data. With total SSD fixed, values of C1 and C2 were systematically varied according to Eqn. 5 at different fixed values of ⬅X⫺--H⫹:⬅SOH site density ratio. The sensitivity analysis was then repeated at varying values of total SSD to yield the value of total SSD (with ⬅X⫺--H⫹:⬅SOH site density ratio) and C1 and C2 to produce the lowest goodness of fit parameter (V(Y)). With total SSD (including ⬅X⫺--H⫹: ⬅SOH site density ratio) now fit, C1 and C2 were then varied systematically (according to Eqn. 5) over a wider range of values as a check of the C1 and C2 values already obtained. It has been brought to our attention that FITEQL v. 3.2 does not account for the dilution of the solid sorbent concentration as the potentiometric titration progresses. To correct for this, we fit our potentiometric titration data at 0.01, 0.03 and 0.1 M at the solid solution ratio calculated from the midpoint of the relevant base leg raw titration data. We find that log Ka1, log Ka2 and log Kie changed by ⬍0.2 log units overall, when compared to potentiometric titration fits generated using the initial solid solution ratio throughout. Subsequently, log K’s for the surface sorption complexes (generated below) changed by ⬍0.03 log units. Log Ka1, log Ka2 and log Kie reported here (Table 2) are those generated with the adjusted solid solution ratio; log K’s for surface sorption complexes (Table 4) are those generated using log Ka1, log Ka2 and log Kie reported in Table 2. Optimized acid-base parameter combinations are listed in Table 2 and the potentiometric titration data with model fits shown on Figure 6. In agreement with previous kaolinite potentiometric titrations (e.g., Schindler et al., 1987; Huertas et al., 1998), Figure 6 shows an increasing degree of protonation with decreasing ionic strength at low pH. This pattern is typical of a mineral with permanent negative charge (Kraepiel et al., 1998). Titration curves for ionic strengths 0.01 M, 0.03 M and 0.1 M are similar and show ionic strength to only slightly influence the proton adsorption and desorption reactions (in agreement with Huertas et al., 1998). We report the experimental pHPZC at pH ⬃ 4.7 (within ⫾ 0.1 pH units) for the three ionic strengths measured. This value lies within the range of reported experimental values (e.g., ⬃4, Carroll-Webb and Walther, 1988; 4.68, Ikhsan et al., 1999; 5.28, Brady et al., 1996; ⬃5.5, Huertas et al., 1998). Model fits of the acid-base data (Fig. 5) show the ECCM produces a very good replication of the data. Optimized total SSD (12.0 sites/nm2, Table 2) does indeed appear to be greater than a crystallographically determined total number of surface sites. Determined crystallographically, total

Multisite sorption of Cu on kaolinite

3741

Table 2. Acid-base fits used in mineral-Cu surface complexation modeling. ECCM pH PZCa surface area (m2/g)b total surface site density (sites/nm2)c (mol/g)c [⬅X⫺--H⫹] (sites/nm2)c (mol/g)c [⬅SOH] (sites/nm2)c (mol/g)c log Kalc log Ka2c log Kiec C1 (F/m2)c C2 (F/m2)c Ctot (F/m2)c V(Y)c

4.7 12.2 12.0 2.43 ⫻ 10⫺4 0.96 1.95 ⫻ 10⫺5 11.0 2.23 ⫻ 10⫺4 2.52 ⫺7.51 ⫺3.06 3.0 7.0 2.1 3.21

a

Determined from potentiometric titration data (this study). Determined from BET analysis (this study). Determined from FITEQL simulation of potentiometric titration data (this study). b c

⬅AlOH surface sites is ⬃5.3 sites/nm2 on the {110} and {010} and total ditrigonal cavity sites (represented here as ⬅X⫺--H⫹ and present on the {001}) is ⬃4.4 sites/nm2. Irrespective of the aspect ratio, we can therefore approximate a lower total SSD limit of ⬃4.4 sites/nm2 and an upper total SSD limit of 5.3 sites/nm2. This translates to an upper surface area measurement of ⬃33.4 m2/g and a lower surface area measurement of ⬃27.7 m2/g respectively. These are less than 3 times greater than the BET measured surface area (12.23 m2/g) and well within the estimated possible underestimation of the surface area of kaolinite by BET (Xie and Walther, 1992 predict the “real” kaolinite surface area to be 10 –20 times greater than the BET surface.) The pH distribution of the ⬅X⫺--H⫹/⬅X⫺--Na⫹ and ⬅SOH surface sites in 0.1 M NaNO3 is shown on Figure 7 (calculated from the optimized acid-base parameter combinations listed in Table 2). In the pH range of our adsorption pH edge (pH 3–7) and constant pH isotherm (pH 5.8, 6.5 and 7.0) experiments ⬅SOH0 (and thus AlOH0) is the predominant variable-charge site. Below pH ⬃ 5 (in the region of sorption onto the permanent, negatively charged sites) ⬅X⫺ are present primarily as ⬅X⫺--H⫹. Our surface complexation modeling therefore models Cu2⫹ adsorption onto the ⬅SOH0 and ⬅X⫺--H⫹ sites (rather than onto the ⬅SO⫺ and ⬅X⫺--Na⫹ sites).

3.2.3. Modeling Cu adsorption data The observed copper adsorption data were replicated in the ECCM using the optimized acid-base parameter combinations (Table 2). Equilibria for reactions occurring in solution (Eqns. 5– 8, Table 3) were taken from Baes and Mesmer (1976) which provides an internally consistent set of stability constants that includes the dimer complex that forms in solution.

Fig. 6. Kaolinite potentiometric titration data at I ⫽ 0.01, 0.03 and 0.1 NaNO3 and 25°C, shown as total [H⫹] in mol/L. 5 g/L kaolinite. Symbols are data points, lines are ECCM model fits.

3.2.4. Cu2⫹ complexation at the surface of kaolinite A number of possible surface complexes (Eqns. 10 –12, Table 3) were used in the attempt to model the observed copper adsorption. We include multinuclear surface complexation (Eqn. 12, Table 3) in our modeling based on our direct spectroscopic evidence. This inclusion is in keeping with that of Katz and Hayes (1995a,b) who noted the need for multinuclear complexes to explain adsorption at moderate to high surface coverages. The ECCM fit to the copper(II) adsorption data are shown on Figure 2 and summarized in Table 4. Inner-sphere surface species (11) to (12), involving only ⬅SOH surface sites, were initially considered in a single-site, singlespecies framework for adsorption between pH 3–7. ⬅SOH surface site density was fixed at the value optimized in the fitting of our potentiometric titration data (Table 2). Bidentate mononuclear (⬅AlOH)2Cu(OH)20 surface complexes (Eqn. 11, Table 3; V(Y) ⫽ 449) and tridentate binuclear

3742

C. L. Peacock and D. M. Sherman

Fig. 7. Species distribution for a) ⬅SOH sites b) ⬅X⫺--H⫹/⬅X⫺--Na⫹ sites as a function of pH at 0.1 M NaNO3 ionic strength. Calculated using the optimized acid-base parameters at I ⫽ 0.1 M (Table 2).

(⬅Al3O(OH)2)Cu2(OH)30 surface complexes (Eqn. 12 Table 3; V(Y) ⫽ 575) provide the two best single-site, singlespecies fits to the observed copper adsorption data. In light of (and in agreement with) our EXAFS data, the multispecies adsorption of both mononuclear (⬅AlOH)2Cu(OH)20 and binuclear (⬅Al3O(OH)2)Cu2(OH)30 surface complexes provides the best single-site fit to the observed copper adsorption data (V(Y) ⫽ 413). We find the adsorption of both mono- and binuclear surface complexes accounts for adsorption at moderate to high surface coverages in the higher pH range of the adsorption edge. To improve the fit to the

observed copper adsorption at lower surface coverage (below pH ⬃ 5) and thus improve the fit overall, outer-sphere association with the ⬅X⫺--H⫹ sites (Eqn. 10, Table 3) was considered in conjunction with mono- and binuclear surface complexation on the ⬅SOH sites. ⬅X⫺--H⫹ and ⬅SOH surface site densities were fixed at the values optimized in the fitting of our potentiometric titration data (Table 2) and a multisite (both ⬅SOH and ⬅X⫺--H⫹) modeling approach was thus adopted. We find this multisite modeling approach provides the best overall fit to the adsorption data (V(Y) ⫽ 2.96, see Table 4) by accounting for adsorption at higher surface coverage in the upper pH range and sorption at low surface coverage in the low pH (pH ⬃ 3–5) range. Monodentate inner-sphere surface complexes (e.g., ⬅AlOHCuOH⫹) were also initially considered in a single-site, single-species framework for adsorption between pH 3–7 (to ensure our modeling approach was not biased towards surface complexes identified with EXAFS). However, in agreement with our EXAFS results, we cannot fit our observed copper adsorption data with monodentate inner-sphere complexes (either in isolation or in combination with bidentate inner-sphere complexes) in a single-site, single-species; single site, multispecies or multisite, multispecies modeling framework. In light of our EXAFS measurements, we therefore fit the sorption edges using the bidentate mononuclear (⬅AlOH)2Cu(OH)20 complex, 2 ⬅ AlOH ⫹ Cu2⫹ ⫹ 2H2O ⫽ 共⬅AlOH兲2Cu共OH兲2 ⫹ 2H⫹ 0

(6) the tridentate binuclear (⬅Al3O(OH)2)Cu2(OH)30 complex, 3 ⬅ AlOH ⫹ 2Cu2⫹ ⫹ 3H2O ⫽ 共⬅Al3O共OH兲2兲Cu2共OH兲0 ⫹ 4H⫹ 3

(7)

and the monodentate mononuclear ⬅X⫺--Cu2⫹ complex ⬅X⫺ -- H⫹ ⫹ Cu2⫹ ⫽ ⬅ X⫺ -- Cu2⫹ ⫹ H⫹

(8)

Table 3. Mineral-Cu surface complexation model reactions. Species Kaolinite surface (1) SOH (2) SOH2⫹ (3) SO⫺ (4) X⫺--H⫹ (5) X⫺--Na⫹ Cu(II) Solution speciation (6) CuOH⫹ (7) Cu2(OH)22⫹ (8) Cu(OH)2 (9) H2O Surface complexes (10) X⫺--Cu2⫹ (11) (SOH)2Cu(OH)20 (12) (S3O(OH)2)Cu2(OH)30 a b

From Baes and Mesmer (1976). From Gunnarsson et al. (2000).

Mass action relation SOH SOH ⫹ H⫹ ⫽ SOH2⫹ SOH ⫽ SO⫺ ⫹ H⫹ X⫺--H⫹ X⫺--H⫹ ⫹ Na⫹ ⫽ X⫺--Na⫹ Cu2⫹ ⫹ H2O ⫽ CuOH⫹ ⫹ H⫹ 2Cu2⫹ ⫹ 2H2O ⫽ Cu2(OH)22⫹ ⫹ 2H⫹ Cu2⫹ ⫹ 2H2O ⫽ Cu(OH)2 ⫹ 2H⫹ H2O ⫽ 2OH⫺ ⫹ H⫹ X⫺--H⫹ ⫹ Cu2⫹ ⫽ X⫺--Cu2⫹ ⫹ H⫹ 2SOH ⫹ Cu2⫹ ⫹ 2H2O ⫽ (SOH)2Cu(OH)20 ⫹ 2H⫹ 3SOH ⫹ 2Cu2⫹ ⫹ 3H2O ⫽ (S3O(OH)2)Cu2(OH)30 ⫹ 4H⫹

Equilibrium constant — Kal Ka2 — Kie KHyd.1 (10⫺8.2)a KHyd.2(10⫺10.59)a KHyd.3(10⫺17.5)a KW (10⫺13.79)b K10 K11 K12

Multisite sorption of Cu on kaolinite

Table 5. Predicted complexation of Cu2⫹ to kaolinite (experimental data from Ikhsan et al., 1999).

Table 4. Predicted complexation of Cu(II) to kaolinite. Predicted metal complexes

ECCM Surface complexation model

log K10a 1.55 log K11a ⫺7.88 a log K12 ⫺16.63 V(Y) 2.96 log K10: X⫺--H⫹ ⫹ Cu2⫹ ⫽ X⫺--Cu2⫹ ⫹ H⫹ log K11: 2SOH ⫹ Cu2⫹ ⫹ 2H2O ⫽ (SOH)2Cu(OH)20 ⫹ 2H⫹ log K12: 3SOH ⫹ 2Cu2⫹ ⫹ 3H2O ⫽ (S3O(OH)2)Cu2(OH)30 ⫹ 4H⫹ a

From simulation of Cu sorption data (this study).

with stability constants K共共⬅AlOH兲2Cu共OH兲02兲 ⫽ K共共⬅Al3O共OH兲2兲Cu2共OH兲03兲 ⫽

3743

兵(⬅AlOH)2Cu(OH)20其关H⫹兴2 兵⬅AlOH其2关Cu2⫹兴

(9)

兵(⬅Al3O共OH)2兲Cu2(OH)30其关H⫹兴4 兵⬅AlOH其3ⱍCu2⫹ⱍ2

(10)

and K共⬅X⫺--Cu2⫹兲 ⫽

兵⬅X⫺ -- Cu2⫹其关H⫹兴 兵⬅X⫺ -- H⫹其关Cu2⫹兴

(11)

respectively, where the surface species concentrations are given as mole fractions of surface sites. We are making the approximation that the fraction of available sites is independent of the configuration of surface sites involved in sorption. This will breakdown as we approach full coverage; however, our maximum surface coverage is estimated to be only 42.6%. At increased surface coverage, we might expect tri- and polynuclear complexes to form; these would be difficult to resolve from the EXAFS spectra as there is an uncertainty of ⫾ 0.5 in the number of copper neighbors at ⬃2.95 Å. Indeed, we cannot accurately resolve the relative fractions of mononuclear vs. dimer copper surface complexes on the ⬅AlOH-sites using EXAFS spectroscopy. EXAFS shows that the dimer is present at pH 5.8 and this is consistent with our surface complexation modeling of the sorption edges (Fig. 3). 3.2.5. Test of our surface complexation model We are able to fit our constant pH isotherm data (Fig. 3) to the surface complexation model proposed for the (pH edge) adsorption of copper to kaolinite. Using the log K constants derived in the pH edge data surface complexation modeling (Table 4) we fit our isotherm data in the ECCM. The fits are shown on Figure 3. Following Tamura and Furuichi (1997), theoretical surface density of adsorbed ions is the sum of the densities of the three types of surface complexes. Fits for constant pH isotherm data at pH 7.0 and 6.5 reflect the formation of Cu(OH)2 (s) precipitate after ⬃20 ppm [Cu]total and ⬃30 ppm [Cu]total respectively. To test our surface complexation model, we have fit our proposed surface complexes to previously published copper adsorption data on kaolinite (Ikhsan et al., 1999) at lower [Cu]total and background electrolyte concentration ([BE]). We have modeled Cu2⫹ adsorption to kaolinite in the ECCM at [Cu]total ⫽ 1.0 ⫻ 10⫺4 M, [BE] ⫽ 0.005 M KNO3 (K1).

pHPZC surface area (m2/g) total surface site density (sites/nm2) (mol/g) [⬅X⫺--H⫹] (sites/nm2) (mol/g) [⬅SOH] (sites/nm2) (mol/g) log Ka1 log Ka2 log Kie Ctot (F/m2) C1 (F/m2) C2 (F/m2) Metal complex

Parametera 4.68 14.73 3.3 8.1 ⫻ 10⫺5 1.36 3.33 ⫻ 10⫺5 1.96 4.79 ⫻ 10⫺5 3.96 ⫺7.24 ⫺2.85 2.1 3.0 7.0 Experimental condition K1

log K10b ⫺1.06 log K11b ⫺7.62 log K12b ⫺16.67 V(Y) 5.92 log K10: X⫺--H⫹ ⫹ Cu2⫹ ⫽ X⫺--Cu2⫹ ⫹ H⫹ 2⫹ 0 log K11: 2SOH ⫹ Cu ⫹ 2H2O ⫽ (SOH)2Cu(OH)2 ⫹ 2H⫹ log K12: 3SOH ⫹ 2Cu2⫹ ⫹ 3H2O ⫽ (S3O(OH)2)Cu2(OH)30 ⫹ 4H⫹ K1: [Cu]total ⫽ 1.0 ⫻ 10⫺4 M, [BE] ⫽ 0.005 M KNO3 a

As reported in Ikhsan et al. (1999). From simulation of Cu sorption data (data from Ikhsan et al., 1999, simulation this study). b

We used our values for C1 and C2 which equal the Ctot value of 2.1 F/m2 of Ikhsan et al. (1999) (1/C1 ⫹ 1/C2 ⫽ 1/Ctot so 1/3 ⫹ 1/7 ⫽ 1/2.1) and subsequently all additional surface complexation model parameters were as reported by Ikhsan et al. (1999), listed here in Table 5. We are able to successfully fit the data of Ikhsan et al. (1999) to the formation of bidentate mononuclear (⬅AlOH)2Cu(OH)20 and tridentate binuclear (⬅Al3O(OH)2)Cu2(OH)30 complexes, and monodentate mononuclear ⬅X⫺--Cu2⫹ complexes. The results of our modeling are shown in Figure 8 and summarized in Table 5. We produce a fit to the experimental data comparable to that produced by Ikhsan et al. (1999; although it is noted that the fit to the Cu sorption data presented by Ikhsan et al. was in fact generated by a model that represented the best compromise to fit titration, isotherm and edge data). We find that the stability constants for the bidentate mononuclear and tridentate binuclear surface complexes are similar (within 0.26 and 0.04 respectively) to that predicted for our own Cu2⫹ kaolinite adsorption data (at [Cu]total ⫽ 3.94 ⫻ 10⫺4 M; [BE] ⫽ 0.1 M NaNO3). In passing, we find the maximum amount of Cu2⫹ sorption on the ⬅X⫺ sites is approximately equal for both data sets (at ⬃4 ⫻ 10⫺5 to 6 ⫻ 10⫺5 mol/L Cu sorbed) despite ⬅X⫺ sorption on the kaolinite used by Ikhsan et al. (1999) being able to occur at a greater number of ⬅X⫺ sites (Ikhsan et al. (1999) ⬅X⫺ sites ⫽ 3.33 ⫻ 10⫺5 moles/g compared to our ⬅X⫺ site total at 1.95 ⫻ 10⫺5 moles/g). The amount of sorption associated with ⬅X⫺ sites therefore appears to be controlled more by

3744

C. L. Peacock and D. M. Sherman

Having identified the bidentate (⬅AlOH)2Cu(OH)20, tridentate (⬅Al3O(OH)2)Cu2(OH)30 and ⬅X⫺--Cu2⫹ surface complexes, the experimental copper(II) adsorption data can be fit to the reactions 2 ⬅ AlOH ⫹ Cu2⫹ ⫹ 2H2O ⫽ 共⬅AlOH兲2Cu共OH兲2 ⫹ 2H⫹ 0

3共⬅AlOH兲 ⫹ 2Cu2⫹ ⫹ 3H2O ⫽ 共⬅Al3O(OH)2兲Cu2共OH兲3 ⫹ 4H⫹ 0

(12)

and ⬅X⫺ -- H⫹ ⫹ Cu2⫹ ⫽ ⬅ X⫺ -- Cu2⫹ ⫹ H⫹ . Fig. 8. Adsorption of copper(II) ions to kaolinite as a function of pH. Symbols are data points (from Ikhsan et al. (1999)), lines are ECCM fits (this study) showing total and individual surface species (large dashed line ⫽ mononuclear complex; small dashed line ⫽ binuclear complex and dotted line ⫽ ⬅X⫺--Cu2⫹ complex).

the capacity of the kaolinite siloxane surface to sorb Cu2⫹ cations than the availability of suitable sites. The low pH adsorption data (below pH 4.0) of Ikhsan et al. (1999) appears to be fit more successfully by our surface complexation model than the model proposed by Ikhsan et al. (1999; as noted above, however, the model of Ikhsan et al. represents a best compromise). Furthermore, in the pH region shown to have bidentate (⬅AlOH)2Cu(OH)20 and tridentate binuclear (⬅Al3O(OH)2)Cu2(OH)30 complexes in this study (pH ⬎ ⬃ 5), we model the formation of both complexes (at lower [Cu]total and background electrolyte concentration) and this is consistent with our EXAFS results. Our surface complexation model disagrees with previous studies which invoked bidentate (⬅X2)2⫺--Cu2⫹ ion exchange complexes at low pH (e.g., Angove et al., 1998; Ikhsan et al., 1999) and monodentate surface species in addition to bidentate complexes at higher pH (e.g., Schindler et al., 1987). We find that Cu2⫹ adsorbs as a monodentate outer-sphere complex at the ion exchange sites and we see no evidence for monodentate surface complexation on the ⬅AlOH sites. However, this study does find the formation of bidentate inner-sphere complexes in the higher pH regime (where (⬅AlOH)2Cu(OH)20 are the same in terms of FITEQL modeling as (⬅SO)2M0 proposed by Angove et al., 1998, and Ikhsan et al., 1999). This study also invokes binuclear surface complexes in the higher pH regime on the ⬅AlOH sites based on direct spectroscopic evidence. 4. CONCLUSIONS

We measured the adsorption of Cu(II) onto kaolinite from pH 3–7 at constant ionic strength. EXAFS spectra show that Cu(II) adsorbs as (CuO4Hn)n⫺6 and binuclear (Cu2O6Hn)n⫺8 inner-sphere complexes on variable-charge ⬅AlOH sites and as Cu2⫹ on ion exchangeable ⬅X⫺--H⫹ sites. Sorption isotherms and EXAFS spectra show that surface precipitates have not formed at least up to pH 6.5. Inner-sphere complexes are bound to the kaolinite surface by corner-sharing with two or three edge-sharing Al(O,OH)6 polyhedra. Our interpretation of the EXAFS data are supported by ab initio (density functional theory) geometries of analog clusters simulating Cu complexes on the {110} and {010} crystal edges and at the ditrigonal cavity sites on the {001}.

In an encouraging test of our model we are also able to fit copper adsorption data at lower [Cu]total and background electrolyte concentration. Our surface complexation model disagrees with previous studies which invoked monodentate innersphere surface species in addition to bidentate complexes (e.g., Schindler et al., 1987). This shows that it is difficult to unambiguously fit sorption edges to a surface complexation model without spectroscopic data. Acknowledgments—Thanks are due to P. Chung Choi for assistance with ICP-AES analysis, Paul Moir-riche and Chris Corrigan at Daresbury Materials Support Laboratory for XRD, and Bob Billsborrow at Daresbury Laboratory for support at Station 16.5. CLP was supported by a NERC studentship. Associate editor: G. R. Helz REFERENCES Angove M. J., Johnson B. B., and Wells J. D. (1998) The influence of temperature on the adsorption of cadmium(II) and cobalt(II) on kaolinite.J. Colloid Interface Sci. 204, 93–103. Baes C. F. and Mesmer R. E. (1976) The Hydrolysis of Cations. Wiley. Binsted N. (1998) EXCURV98: The Manual. CLRC Daresbury Laboratory, Warrington, UK. Bolland M. D. A., Posner A. M., and Quirk J. P. (1976) Surface charge in kaolinites in aqueous suspension. Aust. J. Soil Res. 14, 197–216. Bolland M. D. A., Posner A. M., and Quirk J. P. (1980) pH-independent and pH-dependent surface charges on kaolinite. Clays Clay Min. 28, 412– 418. Brady P. V., Cygan R. T., and Nagy K. L. (1996) Molecular controls on kaolinite surface charge. J. Colloid Interface Sci. 183, 356 –364. Carroll-Webb S. A. and Walther J. V. (1988) A surface complex reaction model for the pH-dependence of corundum and kaolinite dissolution rates. Geochim. Cosmochim. Acta 52, 2609 –2623. Chisholm-Brause C. J., Hayes K. F., Roe A. L., Brown G. E. Jr., Parks G. A., and Leckie J. O. (1990) Spectroscopic investigation of Pb(II) complexes at the ␥-Al2O3/water interface. Geochim. Cosmochim. Acta 54, 1897–1909. Davis J. A. and Kent D. B. (1990) Surface complexation modeling in aqueous geochemistry. In Mineral-Water Interface Geochemistry,Vol. 23 (eds. M. F. HochellaJr. and A. F. White). pp. 177–260. Mineralogical Society of America. Dent A. J. and Mosselmans J. F. W. (1992) A Guide to EXBACK, EXCALIB and EXCURV92. CLRC Daresbury Laboratory, Warrington, UK. Du Q., Sun Z., Forsling W., and Tang H. (1997) Adsorption of copper at aqueous illite surfaces. J. Colloid Interface Sci. 187, 232–242. Ferris A. P. and Jepson W. B. (1975) The exchange capacities of kaolinite and the preparation of homoionic clays. J. Colloid Interface Sci. 51, 245–259. Gans P. and O’Sullivan B. (2000) GLEE: A new computer program for glass electrode evaluation. Talanta 51, 33–37. Gao Y. and Mucci A. (2001) Acid base reactions, phosphate and arsenate complexation and their competitive adsorption at the sur-

Multisite sorption of Cu on kaolinite face of goethite in 0.7 M NaCl solution. Geochim. Cosmochim. Acta 65, 2361–2378. Graveling G. J., Ragnarsdottir K. V., Allen G. C., Eastman J., Brady P. V., Balsley S. D., and Skuse D. R. (1997) Controls on polyacrylamide adsorption to quartz, kaolinite and feldspar. Geochim. Cosmochim. Acta 61, 3515–3523. Gunnarsson M., Jakobsson A., Ekberg S., Albinsson Y., and Ahlberg E. (2000) Sorption studies of cobalt(II) on colloidal hematite using potentiometry and radioactive tracer technique. J. Colloid Interface Sci. 231, 326 –336. 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. Hedin L. and Lundqvist S. (1969) Effects of electron-electron and electron-photon interactions on the one-electron states of solids. Solid State Physics 23, 1–181. Herbelin A. and Westall J. (1996) A Computer Program for Determination of Chem. Equilibrium Constants from Experimental Data. Version 3.2. Department of Chemistry, Oregon State University, Corvallis. Hohl H. and Stumm W. (1976) Interaction of Pb2⫹ with hydrous gamma-Al2O3. J. Colloid Interface Sci. 55, 281–288. Huertas F. J., Chou L., and Wollast R. (1998) Mechanism of kaolinite dissolution at room temperature and pressure: Part 1. Surface speciation. Geochim. Cosmochim. Acta 62, 417– 431. Ikhsan J., Johnson B., and Wells J. D. (1999) A comparative study of the adsorption of transition metals on kaolinite. J. Colloid Interface Sci. 217, 403– 410. Jung J., Cho Y., and Hahn P. (1998) Comparative study of Cu2⫹ adsorption on goethite, hematite and kaolinite: Mechanistic modelling approach. Bull. Korean Chem. Soc. 19, 324 –327. Karthikeyan K. G. and Elliott H. A. (1999) Surface complexation modeling of copper sorption by hydrous oxides of iron and aluminium. J. Colloid Interface Sci. 220, 88 –95. Katz L. E. and Hayes K. F. (1995a) Surface complexation modeling. I. Strategy for modeling monomer complex formation at moderate surface coverage. J. Colloid Interface Sci. 170, 477– 490. Katz L. E. and Hayes K. F. (1995b) Surface complexation modeling. II. Strategy for modeling polymer and precipitation reactions at high surface coverage. J. Colloid Interface Sci. 170, 491–501. Kraepiel A. M. L., Keller K., and Morel F. M. M. (1998) On the acid-base chemistry of permanently charged minerals. Environ. Sci. Technol. 32, 2829 –2838. Lackovic K., Angove M. J., Wells J. D., and Johnson B. B. (2003) Modeling the adsorption of Cd(II) onto Muloorina illite and related clay minerals. J. Colloid Interface Sci. 257, 31– 40. Le Riche H. H. and Weir A. H. (1963) A method of studying trace elements in soil fractions. J. Soil Sci. 14, 225–235. Lim C. H., Jackson M. L., Koons R. D., and Helmke P. A. (1980) Kaolins: Sources of differences in cation-exchange capacities and cesium retention. Clays Clay Min. 28, 223–229. Lumsdon D. G. and Evans L. J. (1994) Surface complexation model parameters for goethite (␣-FeOOH). J. Colloid Interface Sci. 164, 119 –125. Majone M., Papini M. P., and Rolle E. (1996) Modeling lead adsorption on clays by models with and without electrostatic terms. J. Colloid Interface Sci. 179, 412– 425. McBride M. B. (1976) Origin and position of exchange sites in kaolinite: An ESR study. Clays Clay Min. 24, 88 –92. McBride M. B. (1978) Copper(II) interactions with kaolinite: Factors controlling adsorption. Clays Clay Min. 26, 101–106. Mookherjee A. and Tenginkai S. G. (1987) Some unusual geochemical features of the oxidized zone at the central sector of the Singhbhum copper belt, India. Chem. Geol. 60 (1– 4), 51– 62.

3745

Nilsson N., Persson P., Lovgren L., and Sjoberg S. (1996) Competitive surface complexation of o-phthalate and phosphate on goethite (␣-FeOOH) particles. Geochim. Cosmochim. Acta 60, 4385– 4395. O’Day P. A., Parks G. A. and Brown G. E. (1991) EXAFS study of aqueous Co(II) sorption complexes on kaolinite and quartz surfaces. In X-ray Absorption Fine Structure (ed. S. S. Hasnain), pp. 260 –262. Ellis Horwood, London. O’Day P. A., Parks G. A., and Brown G. E. (1994) Molecular structure and binding sites of cobalt(II) surface complexes on kaolinite from x-ray absorption spectroscopy. Clays Clay Min. 42 (3), 337–355. Peacock C. L. and Sherman D. M. (2004) Copper(II) sorption onto goethite, hematite and lepidocrocite: A surface complexation model based on ab initio molecular geometries and EXAFS spectroscopy. Geochim. Cosmochim. Acta 68, 1723–1733. Perdew J. P., Chevary J. A., Vosko S. H., Jackson K. A., Pederson M. R., Singh D. J., and Fiolhais C. (1992) Atoms, molecules, solids and surfaces—Applications of the generalised gradient approximation for exchange and correlation. Phy. Rev. B 46, 6671– 6687. Robertson A. P. and Leckie J. O. (1998) Acid/base, copper binding and Cu2⫹/H⫹ exchange properties of goethite, an experimental and modelling study. Environ. Sci. Technol. 32, 2519 –2530. Rustad J. R., Felmy A. R., and Hay B. P. (1996) Molecular statics calculations of proton binding to goethite surfaces: A new approach to estimation of stability constants for multisite surface complexation models. Geochim. Cosmochim. Acta 60, 1563–1576. Schindler P. W. and Gamsjager H. (1972) Acid-base reactions of the TiO2 (Anatase)-water interface and the point of zero charge of TiO2 suspensions. Kolloid-Z. U. Z. Polymere 250, 759 –765. Schindler P. W., Liechti P., and Westall J. C. (1987) Adsorption of copper, cadmium and lead from aqueous solution to the kaolinite/ water interface. Netherlands J. Agric. Sci. 35, 219 –230. Spark K. M., Wells J. D., and Johnson B. B. (1995) Characterizing trace metal adsorption on kaolinite. Eur. J. Soil Sci. 46, 633– 640. Stumm W., Hohl H., and Dalang F. (1976) Interaction of metal ions with hydrous oxides surfaces. Croatica Chim. Acta 48, 491–504. Tadanier C. J. and Eick M. J. (2002) Formulating the charge-distribution multisite surface complexation model using FITEQL. Soil. Sci. Soc. Am. J. 66, 1505–1517. Talibudeen O. (1981) Cation exchange in soils. In The Chemistry of Soil Processes (eds. D. J. Greenland and M. H. B. Hayes). pp. 115–177. Wiley. Tamura H. and Furuichi R. (1997) Adsorption affinity of divalent heavy metal ions for metal oxides evaluated by modeling with the Frumkin isotherm. J. Colloid Interface Sci. 195, 241–249. te Velde G., Bickelhaupt F. M., Baerends E. J., Fonseca Guerra C., van Gisbergen S. J. A., Snijders J. G., and Ziegler T. (2001) Chemistry with ADF. J. Comp. Chem. 22, 931–967. Tenginkai S. G., Ugarkar A. G., Koti M. V., and Mookherjee A. (1991) Copper-bearing clay-minerals of the oxidized zone of the RakhaChapri block, Singhbhum copper belt, India. Proc. Indian Acad. Sci. Earth Planet. Sci. 100 (1), 13–29. van Olphen H. (1977) Introduction to Clay Colloid Chemistry. 2nd ed Wiley-Interscience. Vosko S. H., Wilk K. and Nusair M. (1980) Accurate spin-dependent electron liquid correlation energy for local spin density calculations: A critical analysis. Can. J. Phys. 58, 1200 –1205. Westall J. C. (1986) Reactions at the oxide-solution interface: Chemical and electrostatic models. In Geochemical Processes at Mineral Surfaces (eds. J. A. Davis and K. F. Hayes), pp. 54 –78. Symposium Series 323. American Chemical Society. Xie Z. and Walther J. V. (1992) Incongruent dissolution and surface area of kaolinite. Geochim. Cosmochim. Acta 56, 3357–3363.