Sorption of Nickel on Carbonate Fluoroapatites

Journal of Colloid and Interface Science 239, 303–313 (2001) doi:10.1006/jcis.2001.7583, available online at http://www.idealibrary.com on Sorption of Nickel on Carbonate Fluoroapatites Jane Perrone,∗, † Blandine Fourest,†,1 and Eric Giffaut∗ ∗ ANDRA DS/MA, 92298 Châtenay Malabry Cedex, France; and †IPN, Groupe de Radiochimie, Université Paris Sud, 91406 Orsay Cedex, France E-mail: [email protected] Received November 14, 2000; accepted March 22, 2001 The retention properties of a synthetic carbonate fluoroapatite and a natural francolite are compared in the present work from an investigation of the sorption of 63 Ni at tracer scale amounts onto these solids. Two different surface complexation models were successively used to fit the experimental adsorption isotherms obtained under various experimental conditions: the nonelectrostatic model and the constant capacitance model. The results are essentially described by two main equilibria involving one proton in acidic media and three protons in basic media. The corresponding thermodynamic constants are in agreement for both models. Modeling gives also close values for both solids, despite their distinct solubility and surface acidity. °C 2001 Academic Press Key Words: apatite; sorption; surface complexation; nickel; modeling. INTRODUCTION Geochemical studies have demonstrated the great stability of apatitic minerals as well as their capacity to retain durably a large variety of trace elements, particularly actinides, rareearth elements, and other heavy metals. These high retention capacities are due to: — their particular crystalline structure that allows isomorphous substitutions and diffusion phenomena, — complexation reactions with the functional groups of the surface, and — the formation of insoluble compounds via dissolution– precipitation processes. Apatites could therefore be used for the remediation of metalcontaminated soils and waters (1–4) or the confinement of industrial or nuclear wastes (5, 6). In particular, they are regarded as possible additives to the engineered barriers of a deep geological nuclear waste repository. As a matter of fact, the optimization of the chemical properties of the engineered barriers necessitates the use of materials with either specific or high-sorbing capacities. For such use, it is not only important to evaluate the uptake or release of radionuclides, but also to identify the sorption mechanisms, which directly determine the durability of the im1 To whom correspondence should be addressed. mobilization. In fact, the migration behavior of the radionuclides will essentially depend on the extent of sorption reversibility. Carbonate fluoroapatite (francolite) is the main phosphate mineral present in sedimentary phosphorites. It is microcrystalline and differs from synthetic pure apatites because of extensive and complex substitutions in apatite structure. These substitutions result in tremendous variations in chemical reactivity and stability of carbonate fluoroapatites. It has been shown that their solubility and reactivity increase with their carbonate content (7–10). Their sorption capacities and the mechanisms involved may therefore differ from those of pure synthetic hydroxyapatites. Although the immobilization of divalent cations has been extensively examined (mainly on synthetic pure hydroxyapatites), the sorption mechanisms involved are still not well understood, particularly in the case of nickel, whose long-lived isotopes are of great importance for the safety of a deep geological nuclear waste repository. In fact, it has been shown that isomorphous substitutions contribute to a large extent to the retention of heavy divalent cations like cadmium (11, 18–22), lead (12, 20, 23–24), and strontium (25–29). Other cations like Zn2+ , Mg2+ , Ba2+ , Cu2+ , Co2+ , and Ni2+ can also be exchanged, but to a much lesser extent. Suzuki et al. (11, 12) established a correlation between the radii and electronegativity values of the divalent cations and their ability to substitute for calcium into hydroxyapatite. The radii of easily removed cations like Pb2+ or Cd2+ are very close to that of Ca2+ and these ions are highly electronegative, whereas Mg2+ and Ba2+ , which are hardly removed, have radii larger than that of Ca2+ and low electronegativity values. Ions with high electronegativity values but small radii like Cu2+ , Co2+ , Mn2+ , and Ni2+ show intermediate behavior. But in many cases, cation exchange is not the predominant mechanism, and sorption results from an adsorption process (4, 13). Reichert and Binner (13) performed sorption experiments of binary or ternary mixtures of cobalt, chromium, iron (III), nickel, aluminum, copper, and lead on hydroxyapatites. They observed that cations with higher electronegativity values are preferentially adsorbed, and that the sorption of the divalent cation is favored for cations with different valences but similar electronegativity values. 303 0021-9797/01 $35.00 C 2001 by Academic Press Copyright ° All rights of reproduction in any form reserved. 304 PERRONE, FOUREST, AND GIFFAUT Moreover, the reaction of heavy metals with mineral francolite has rarely been studied (1) and may be complicated because of the coexistence of phosphate, carbonate, and fluoride ions and other cations in the structure of mineral apatite and their possible presence in solution. The aim of this study was therefore to identify the mechanism of nickel sorption onto carbonate fluoroapatites, and to quantify this sorption. Because of the large variety of mineral francolites, two materials were selected: a well-characterized synthetic carbonate fluoroapatite and a natural francolite from Morocco. The effect of various parameters (pH, ionic strength, solid/solution ratio, . . .) on sorption was carefully examined. In order to enable further predictive calculations, the experimental data were subsequently modeled with the nonelectrostatic and the constant capacitance surface complexation models. EXPERIMENTAL Solid Preparation A. Synthetic carbonate fluoroapatite. The synthetic carbonate fluoroapatite has been prepared according to the method described by Tomson and Nancollas (14) and revised by Jahnke (8) and, more recently, Régnier et al. (15). The apparatus used for the synthesis is a reaction vessel initially containing 2 liters of 1 M KNO3 solution thermostated at 70◦ C. Reagents, in the form of 1 M Ca(NO3 )2 , 4H2 O, 0.5 M K2 HPO4 , 0.3 M KF, and 0.1 M KHCO3 solutions, were added to the vessel content via a peristaltic pump at a rate of 2 ml/h. Important variations of the ionic strength were avoided using a 1 M KNO3 solution as the reaction medium. This electrolyte was chosen because K+ and + NO− 3 substitute into apatite to a much lesser extent than Na and − Cl . At the end of the synthesis, the solid was collected, washed with deionized water, and dried in an oven at 70◦ C. The precise stoichiometry of the synthetic carbonate fluoroapatite was determined using both the PIXE (proton–induced X-ray emission) and the PIGE (proton-induced gamma-ray emission) techniques. The carbonate content of the solid was confirmed by measuring the angular difference between the 004 and the 410 peaks on its X-ray diffraction pattern (17, 18). Its chemical composition is Ca10 (PO4 )5 (CO3 )F2.72 (OH)0.28 . B. Mineral francolite. The mineral apatite used in this study comes from the sedimentary phosphate rock deposit of Oulad Abdoun in Morocco, and was supplied by the BRGM (Bureau de Recherche Géologique et Minière). The sample is essentially made of carbonate fluoroapatite having a composition of Ca10 (PO4 )4.68 (CO3 )1.32 F1.87 OH1.45 , but also contains small amounts of calcite and quartz. Prior to its use, the solid was crushed and sieved. The fraction of particle size lower than 50 µm was collected, washed with deionized water, and dried at 70◦ C. Solutions The ionic strength of the solutions was fixed at 0.5, 0.1, or 0.05 M using potassium nitrate as an indifferent electrolyte, and their pH was made to vary between 2 to 13 by adding small amounts of concentrated nitric acid or potassium hydroxide solutions. Potassium nitrate and nitric acid solutions were prepared respectively by dissolving weighed amounts of potassium nitrate salt and by diluting concentrated nitric acid into fresh deionized water. The diluted HNO3 solutions were then calibrated with standard potassium hydroxide solutions. Sample Characterization A. Physical characterization of the powders. Particle size, shape, and crystallinity have been studied by laser granulometry, scanning electron microscopy, and X-ray diffraction methods. Specific area of the powders was obtained by the BET nitrogen adsorption method using a Coulter SA 3100 apparatus. B. Solubility. The solubility of our solids has been carefully examined. The experiments consisted of equilibrating weighed amounts of solid with 10 ml of solution in 15-ml high density polyethylene Nalgene vials. After contact times ranging from 1 to 120 days, the samples were centrifuged at 3500 rpm for 30 min. to achieve a complete separation of the solid and liquid phases. The final pH was measured and the supernatants were analyzed by capillary electrophoresis to determine the total concentrations of calcium, phosphate, and fluoride ions. The capillary electrophoresis apparatus used is a modular system consisting of a Spectraphoresis 100 injector coupled with a highvoltage power supply Prime Vision VIII from Europhor, and a scanning UV detector Prime Vision IV from Europhor. The solubilities are relatively high, especially in acidic media, and for both solids, no significant evolution of the measured solubility was observed beyond 24 h of contact time, indicating that equilibrium is rapidly attained. Concerning the synthetic carbonate fluoroapatite, the measured concentrations of phosphate and fluoride anions vary respectively from 5 × 10−2 to 1 × 10−5 M, and from 2.7 × 10−2 to 6 × 10−5 M in the pH range 2–12. In addition, results show that the solid dissolves congruently according to the reaction Ca10 (PO4 )5 (CO3 )(F, OH)3 2− − ↔ 10Ca2+ + 5PO3− 4 + CO3 + 3(F, OH) . K S,0 Thus, considering all the possible complexation and acid–base reactions involving the dissolved species, the solubility product of the solid could be calculated (10). Its value at zero ionic strength is K S,0 = 10−103±2 . In contrast, the mineral francolite dissolves incongruently, agreeing with the results obtained by Jahnke (8) for synthetic carbonate fluoroapatites of similar carbonate content. Calcium concentrations vary from 0.2 to 4 × 10−4 M in the pH range 2–12, whereas beyond pH 6 phosphate concentrations are lower than the detection limit of capillary electrophoresis (∼10−5 M). Thus, no solubility product could be calculated for this solid. 305 SORPTION OF NICKEL ON CARBONATE FLUOROAPATITES TABLE 1 Physical and Surface Properties of Starting Materials Material Synthetic apatite Mineral francolite Particle A Ns size (µm) (m2 /g) (sites · nm−2 ) 30 34 8.8 13.9 3.1 3.3 IEP PZC 6.3 ± 0.2 6.4 ± 0.2 4.8 ± 0.2 8.6 ± 0.2 C. Surface characterization. The isoelectric point has been determined from electrophoretic mobility measurements using a Coulter DELSA 440 device. Potentiometric titrations have also been performed (10) at different ionic strength values in order to determine the point of zero charge (PZC) and the surface site densities of our solids. Experiments were carried out at 25◦ C under an argon gas atmosphere, with a Radiometer Tacussel S.A. automatic system for precise titration. Prior to titration, the solids were equilibrated for 24 h with 150 ml of 0.5 or 0.1 M KNO3 solutions to which 3 ml of 0.5 or 0.1 M KOH solutions were added. Nitric acid was then used as a titrant. To limit the uptake of H+ or OH− by the acids and bases originating from the dissolution of the solid during the titration, the delay between two consecutive additions of acid did not exceed 2 min. The bulk physical and surface characteristics of starting materials are summarized in Table 1. Sorption Measurements Static sorption experiments were carried out in batches as follows: a given amount of apatite was equilibrated with 10 ml of solution for 24 h in 15-ml high-density polyethylene Nalgene vials; after this hydration step, nickel was added to the suspension from a 10−6 M 63-nickel nitrate stock solution in order to reach initial concentrations ranging from 5 × 10−10 to 1 × 10−8 M in the samples. After a fixed contact time, the samples were centrifuged at 3500 rpm for 30 min. to separate the solid and liquid phases. An aliquot of the supernatant was then analyzed for its nickel content using a TRICARB 2700 TR alpha–beta liquid scintillator from Packard Instruments. A final pH measurement was made on the remaining supernatant and taken as the representative pH of the experiment. In order to determine the time required for completion of the sorption reactions, a series of kinetic studies was performed on both synthetic and mineral apatite. The sorption time was made to vary from 2 h to 30 days. For both solids, an equilibrium was reached after 8 days. Since no evolution of the sorption was observed beyond this time, a contact time of 8 days was chosen for all further experiments. RESULTS Nickel/Solution Interactions Direct solubility measurements and predictive calculations were performed to examine the possible influence of phosphate, fluoride, or carbonate ions on the solubility and speciation of nickel. The experimental solubility study consisted of adding an aliquot of a concentrated nickel nitrate solution to solutions equilibrated with the synthetic or the mineral apatite in order to have an initial concentration of 10−3 M. After 8 days, the samples were ultracentrifuged for 1 h at 50,000 rpm to eliminate possible precipitates. The supernatants were analyzed for their nickel content and the pH was measured to check that it did not vary. The hydrolysis and complexation reactions of nickel in solution, as well as the dissolution/precipitation reactions considered for the calculations, and the corresponding equilibrium constants at zero ionic strength are listed in Table 2. Concentrations of the anionic species were derived from the solubility study of the solids. Predictive calculations indicate that in both cases no precipitation of nickel is expected, which is in good agreement with the results obtained for the mineral francolite, but for the synthetic compound a slight decrease of nickel concentration is observed between pH 8 and pH 10 (see Fig. 1), which could be due to the precipitation of a fluoride or phosphate compound (in this pH range, the solutions equilibrated with the synthetic apatite contain higher concentrations of phosphate and fluoride ions than those equilibrated with the mineral one). The corresponding speciation diagrams are given in Figs. 2a and 2b. In the case of the synthetic apatite, Ni2+ , NiF+ and NiH2 PO+ 4 are the main species in acidic and neutral medium, TABLE 2 Thermodynamical Data Used for Nickel Speciation Calculations Hydrolysis and complexation reactions Log K 0 Ni2+ + H2 O ↔ Ni(OH)+ + H+ −9.6 (44) Ni2+ + 2H2 O ↔ Ni(OH)0aq + 2H+ −20.2 (44) + Ni2+ + 3H2 O ↔ Ni(OH)− 3 + 3H −30 (44) + Ni2+ + 4H2 O ↔ Ni(OH)2− 4 + 4H −44 (44) Ni2+ + F− ↔ NiF+ 1.10 (45) Ni 2+ Ni2+ Ni2+ Ni2+ Ni2+ Ni2+ Ni2+ Ni2+ 0 + CO2− 3 ↔ Ni(CO3 )aq 2− + 2CO3 ↔ Ni(CO3 )2− 2 + + HCO− 3 ↔ NiHCO3 − + PO3− 4 ↔ NiPO4 3− + + H + PO4 ↔ Ni(HPO4 )0aq + + 2H+ + PO3− 4 ↔ NiH2 PO4 + − + NO3 ↔ NiNO3 0 + 2NO− 3 ↔ Ni(NO3 )aq Dissolution reactions 4.3 (46) 10.10 (46) 13.40 (46) 8.37 (47) 15.28 (47) 21.10 (47) 0.4 0.6 Log K s0 Ni(CO3 )s ↔ Ni2+ + CO2− 3 −6.84 NiF2 s ↔ Ni2+ + 2F− −0.61 Ni(OH)2 s ↔ Ni2+ + 2OH− −10.8 NiHPO4 s ↔ Ni2+ + HPO3− 4 −33.48 Ni3 (PO4 )2 s ↔ 3Ni2+ + 2 PO3− 4 −31.2 306 PERRONE, FOUREST, AND GIFFAUT Characterization of the Adsorption Mechanism FIG. 1. Theoretical and experimental solubility of nickel in the solutions equilibrated with the synthetic and the mineral carbonate fluoroapatite; [Ni] = 10−3 M in 0.1 M KNO3 at 25◦ C and t = 8 days. The dashed line represents calculated solubility, while synthetic carbonate fluoroapatite and mineral francolite are represented by shaded diamonds and open circles, respectively. At tracer scale, precipitation of nickel can be excluded, but it is still not clear whether cation exchange plays an important role in the overall process of nickel immobilization by carbonate fluoroapatites. Cation exchange can easily be characterized by comparing the amount of adsorbed nickel and the amount of calcium present in solution at the end of the experiment to the amount of calcium released by the dissolution of the solid. Such experiments could not be performed at tracer scale because the initial concentration of the sorbate would be too low in comparison to the solubility of the solids and to the detection limit of capillary electrophoresis (∼10−6 M for Ca2+ and Ni2+ ); no variation of calcium concentration would be observed in this case. Thus, a series of sorption experiments was conducted with initial nickel concentrations of 2 and 5 × 10−4 M, respectively, for the synthetic and the mineral apatite. We can see from Fig. 3 that the removal of nickel by carbonate fluoroapatites is not followed by a significant increase of then the NiPO− 4 complex prevails between pH 9 and 11. In the case of the francolite, Ni2+ and the hydrolyzed forms NiOH+ and NiOH2 aq are the dominant species. Above pH 11, the negatively charged complex Ni(OH)− 3 prevails in both cases. FIG. 2. Speciation diagrams of nickel versus pH; [Ni]tot = 10−3 M, KNO3 0.1 M. Solutions equilibrated with (a) the synthetic carbonate fluoroapatite and (b) the mineral francolite. FIG. 3. Variation of calcium concentration in solution before and after nickel adsorption and compared with the concentration of adsorbed nickel for (a) the synthetic carbonate fluoroapatite: [Ni]tot = 2 × 10−4 M, and (b) the mineral francolite: [Ni]tot = 5 × 10−4 M. SORPTION OF NICKEL ON CARBONATE FLUOROAPATITES 307 FIG. 4. Adsorption of nickel versus pH onto 10 g · l−1 suspensions of apatite, with [Ni]tot = 10−8 M in 0.1 M KNO3 at 25◦ C. d, synthetic apatite; s, francolite. FIG. 5. Effect of the solid/solution ratio (m/V) on the adsorption of nickel on the synthetic carbonate fluoroapatite, [Ni]tot = 10−8 M in = 0.1 M KNO3 at 25◦ C. d, m/V = 10 g/l; h, m/V = 5 g/l; m, m/V = 1 g/l. the calcium concentration in the solution. Thus, even if the possibility of cation exchange cannot be categorically excluded, it seems that the removal of nickel from the solution mainly results from adsorption reactions, in agreement with the conclusions of Reichert and Binner (13). KNO3 , suggesting that adsorption takes place at the surface of the solid via the formation of inner sphere complexes with the functional groups of the surface. Effect of pH Figure 5 shows for the synthetic carbonate fluoroapatite the percentage of adsorbed nickel versus pH at different solid/solution ratios. The three curves have the same shape but, for a solid/solution ratio of 1 g · l−1 , the adsorption edge is shifted toward higher pH values, in good agreement with the decreased number of adsorption sites. Comparable results were obtained for the mineral compound. Figure 4 shows the percentage of adsorbed nickel versus pH for (a) the synthetic apatite and (b) the mineral francolite. In both cases the solid/solution ratio was 10 g · l−1 and nickel concentration 10−8 M. Nickel adsorption on the synthetic carbonate fluoroapatite is negligible for the lower pH values and increases progressively from pH 4. A maximum is reached at pH 8; then above pH 11 the amount of adsorbed nickel decreases as the pH increases. Potentiometric titrations and zeta-potential measurements show that the point of zero charge of this solid is at pH 6.4. This means that nickel adsorption increases as the surface charge is less positive and becomes negative. As for the synthetic compound, nickel adsorption on the mineral francolite becomes significant above pH 4, but in this case, it increases more rapidly and the adsorption edge is reached at pH 7, i.e., below the point of zero charge of the solid (pH 8.6). This means that nickel is strongly adsorbed despite the fact that either nickel aqueous species or the sorbing surface has net positive charges. Moreover, in both cases, the increase of the adsorption percentage is spread over at least 3 pH units, which indicates that several surface complexes are formed. Effect of the Apatite/Solution Ratio Effect of Ionic Strength No significant influence of the electrolyte concentration is observed on the adsorption curves obtained at 0.5, 0.1, and 0.01 M FIG. 6. Effect of nickel concentration on its adsorption onto 10 g · l−1 suspensions of the synthetic carbonate fluoroapatite in KNO3 0.1 M at 25◦ C. , 1 × 10−8 M; s, 5 × 10−10 M. 308 PERRONE, FOUREST, AND GIFFAUT ria can be expressed as SOH + H+ ↔ SOH+ 2 SOH ↔ SO− + H+ with intrinsic stability constants defined as {SOH+ 2 }s + = K int [1] {SOH}s {H+ }s and − = K int FIG. 7. Reversibility of nickel adsorption onto a 10 g · l−1 synthetic apatite suspension; [Ni]tot = 10−8 M in 0.1 M KNO3 at 25◦ C. h, adsorption; d, desorption. {SO− }s {H+ }s {SOH}s [2] , where { }s denote the activity of species at the solid–solution interface. Identically, the adsorption of metal cations can be written as SOH + M z+ ↔ SOM(z−1)+ + H+ Effect of Nickel Concentration We have reported in Fig. 6 the adsorption curves obtained for initial nickel concentrations equal to 10−8 and 5 × 10−10 M. The results clearly show that, at tracer scale, there is no influence of the concentration of the sorbate on the measured adsorption percentages. Reversibility At the end of the sorption experiment, the supernatant was eliminated and the amount of liquid that still remained in the vial was determined by weighing. The supernatant was then replaced by an equivalent volume of fresh solution at the same pH value. After a desorption time equivalent to the sorption time, the sample was centrifuged, and the concentration of nickel was measured, in order to allow the calculation of the desorbed amount of nickel. Adsorption and desorption measurements for the synthetic apatite are compared in Fig. 7. Similar results were obtained for the mineral francolite. It can be deduced that nickel adsorption is reversible, which means that it is governed by equilibrated reactions that can be described using thermodynamical surface complexation models. MODELING THE ADSORPTION CURVES Model Background The studies of the surface charge characteristics of apatites indicate that H+ and OH− are potential determining ions, i.e., that the development of surface charge at the apatite–water interface is due to amphoteric dissociation reactions of surface functional groups through the uptake or release of H+ or OH− ions (10, 30–34). These protonation and deprotonation equilib- [3] with K int = © ª SOM(z−1)+ s {H+ }s {SOH}s {M z+ }s . These intrinsic constants are independent of surface charge and coverage, but since interfacial activities of surface complexes are not directly accessible by experiment, surface equilibria are usually described by apparent or conditional stability constants. The apparent stability constant of a surface complex is then defined as (35) K = ¤ £ SOM(z−1)+ s {H+ } [SOH]{M z+ } , [4] where [ ] are surface concentrations and { } activities in solution. Apparent and intrinsic stability constants can be related by considering that the nonideal behavior of surface species derives from coulombic interactions with a mean surface potential. Typically, an equation of the following form is assumed, K = K int exp µ ¶ −z F9 , RT [5] where z is the charge of the sorbed cation, 9 is the surface potential (in volts), and F is Faraday’s constant (C · mol−1 ). Several theoretical descriptions of the electrical double layer (EDL) formed between the solid surface and the aqueous solution exist in the literature (35–38). In the nonelectrostatic model (NEM), the electrostatic interactions are assumed to be negligible, with 9 equal to 0 (38). In this study, both the nonelectrostatic model and the constant capacitance model were used to fit our experimental data. The latter was chosen because most of the experiments were carried 309 SORPTION OF NICKEL ON CARBONATE FLUOROAPATITES out at 0.1 M high ionic strength. According to the constant capacitance model (CCM), the surface charge (σ ) and potential (9) are linked via the relation 9 = σ/C, these solids as the mean behavior of all the surface sites (noted SOH ) involved in the protonation and deprotonation equilibria SOH + H+ ↔ SOH+ 2 [6] where C is the capacitance of the electrical double layer (in F · m−2 ), which depends on the ionic strength of the solution (36). Acid–Base Properties of the Solids In the case of nonoxide salt minerals, two different hydration sites are usually considered: hydroxylated surface cations, and also protonated surface anions. Spectroscopic techniques like XPS (X-ray photoelectron spectroscopy) or LEED (low-energy electron diffraction) showed the existence of these two kinds of sites in the case of calcium carbonate (49) and recently surface complexation models that incorporate both cation and anion surface sites have been used to interpret the acid–base and surface complexation properties of carbonates and sulfides (40, 41). So far, the only thermodynamic constant values of proton reactions at the apatite–water interface reported in the literature are those estimated by Cases et al. (33) and Wu et al. (34). It is important to note that in both cases no accurate physicochemical characterization of the apatite/water interface has been performed to determine the nature of the reacting species. Cases et al. modeled the surface charge of sedimentary apatites by assuming that the surface of sedimentary apatites does not differ from that of hydroxyapatite, i.e., it is mainly constituted of calcium atoms and phosphate groups. Hence, the surface charge depends on the reactions PO4 H2 ↔ PO4 H− + H+ + PO4 H− ↔ PO2− 4 +H Ca(H2 O)+ ↔ Ca(OH) + H+ Ca(OH) ↔ CaO− + H+ . More recently, Wu et al. showed that the acid–base properties of fluoroapatite could be successfully described by considering these two reactions: SOH ↔ SO− + H+ Moreover, in a recent investigation of the fixation of U(VI) and Eu(III) on various zirconium and thorium phosphate compounds using spectroscopic techniques (XPS, EXAFS), Drot and Simoni (42) showed that the adsorbed species are only bound to the oxygens of the phosphate groups. Thus, due to the lack of precise characterization of the active sites of our solids, we considered the acid–base properties of K −. To estimate the values of the corresponding equilibrium constants, the NEM and the CCM were successively applied to the potentiometric titration data using the computer program FITEQL, version 3.2 (43), which is an iterative nonlinear leastsquares optimization program based on the Gauss method. The goodness of fit is given by the factor WSOS/DF (WSOS is the weighed sum of squares and DF the total degrees of freedom of the system). A good agreement between the experimental and calculated values is obtained for WSOS/DF values comprised between 0.1 and 20. One critical aspect in modeling the potentiometric titration data is the relatively high solubility of apatites. In fact, since the suspensions are always saturated with respect to the solids, the concentrations of acids and bases originating from the dissolution cannot be neglected, even for the pH values corresponding to the lowest values of the solubility. However, we showed that an apatite/solution equilibrium is reached after 24 h. Moreover, during the potentiometric titrations, two consecutive additions of nitric acid were separated at the most by a few minutes. Hence, the saturation of the suspensions is unlikely, and the uptake of H+ or OH− ions by the dissolved species during the experiments is considered to be negligible. In addition, Wu et al. (34) obtained similar values of the acidity constants of a fluoroapatite (the solubility of which is comparable to that of our synthetic compound) when assuming saturation or when neglecting the dissolution. Values of the surface parameters for the synthetic and the mineral apatite are listed in Table 3. The surface site densities of the solids have been approximately determined from the potentiometric titration curves and were used as initial guess values during modeling. We can note that, for the synthetic carbonate fluoroapatite, the obtained value TABLE 3 Surface Parameters for Synthetic and Mineral Apatite Assumed in This Study + PO4 H ↔ PO− 4 +H Ca(H2 O)+ ↔ Ca(OH) + H+ . K+ Model NEM CCM C = 18 F · m−2 NEM CCM C = 70 F · m−2 A Ns (m2 · g−1 ) (sites · nm−2 ) log K + log K − WSOS/ DF a. Synthetic carbonate fluoroapatite 8.8 3.7 5.72 ± 0.1 −7.5 ± 0.1 8.8 3.8 5.98 ± 0.1 −7.22 ± 0.1 4.5 2.9 b. Mineral francolite 13.3 7.15 ± 0.1 −10.2 ± 0.1 14 7.18 ± 0.1 −10.2 ± 0.1 16.2 8.7 13.9 13.9 310 PERRONE, FOUREST, AND GIFFAUT is very close to the experimental one, while it is much higher for the mineral francolite (14 instead of 3.3 sites · nm−2 ). For both solids, the values of log K + and log K − determined with the two models are very close. This is due to the high values of the electrical double-layer capacitance used for fitting the experimental data with the CCM: respectively 18 F · m−2 for the synthetic apatite and 70 F · m−2 for the mineral one. The surface capacitance values usually reported for oxides and hydroxides are only on the order of 1 F · m−2 in solutions of comparable ionic strength, but high values have already been reported for sulfides (40), carbonates (41), and fluoroapatite (34). Physically, they correspond to a thin, highly structured (nondiffuse) double layer, capable of accommodating high charge densities; they are also in good agreement with the electrophoretic mobility measurements that show a weak dependence of the surface charge on the ionic strength and high surface charge values at pH values far from the PZC. Moreover, the fact that the values of the surface acidity constants are so close indicates a weak contribution of the electrostatic interactions. We can also note that the surface of the synthetic apatite is more acidic than the surface of the mineral apatite. Such difference might be explained by the difference in crystallinity and chemical composition between the two solids. In this study, the adsorption experiments have been performed at tracer scale, which means that [Ni]tot ≪ Cs . Then Cs = [SOH] · αs [10] αs = 1 + K + [H+ ] + K − [H+ ]−1 . [11] with Identically for nickel, we can write [Ni]solution = [Ni2+ ] · αNi with αNi = 1 + X X j K j [NO− 3] + j X X ¤m X £ 3− ¤n £ + K m CO2− K n PO4 3 + X ¤o X £ p + K o HPO2− K p [H2 PO− 4 4] . o K l [F− ]l l + m (2−n)+ SOH + Ni2+ + (n − 1)H2 O ↔ SONi(OH)(n−1) + nH βi [H+ ]−i + i Modeling the Adsorption Curves Using the NEM By assuming that only the Ni2+ ion and its hydrolyzed forms can adsorb on carbonate fluoroapatites, surface complexation reactions can be written as [12] n [13] p The formula of nickel partition coefficient becomes Kd = Kn K n · CS αNi · αS · [H+ ]n [14] or with Kn = £ (2−n)+ ¤ SONi(OH)(n−1) [H+ ]n [SOH][Ni2+ ] log . [7] The nickel partition coefficient between solid and liquid phase is then given by £ (2−n)+ ¤ SONi(OH)(n−1) [Ni]apatite K n · [SOH] · [Ni2+ ] Kd = = = . [Ni]solution [Ni]solution [Ni]solution · [H+ ]n [8] If Cs represents the total surface site concentration, the surface site conservation equation is − Cs = [SOH] + [SOH+ 2 ] + [SO ] + n X £ i=1 (2−n)+ ¤ SONi(OH)(n−1) . [9] µ K d · αNi · αS Cs ¶ = n · pH + log K n = log B. [15] K d is an experimental data, αNi and αS are easily calculated using the speciation constants of nickel and surface sites, and Cs is readjusted for the lowest pH values (pH < 5) by calculating the amount of solid that dissolved during the experiment. The stoichiometry of the surface complexes is then graphically extracted by plotting log B versus pH (48). The values of the corresponding slope, n, and of the intercept with the y axis, log K n , values are adjusted in order to obtain the best agreement between the experimental and calculated log B values. This procedure was applied to the different experimental data sets, and coherent results that are averaged in Table 4 were obtained. Neither the apatite/solution ratio nor the nickel concentration seem to have an influence on the nature of the surface complexes and on the corresponding stability constants. A. Case of the synthetic carbonate fluoroapatite. The log B versus pH curves can respectively be decomposed into three linear segments (see pH ranges and corresponding slopes in 311 SORPTION OF NICKEL ON CARBONATE FLUOROAPATITES TABLE 4 Application of Kurbatov’s Model to the Adsorption of Nickel on Carbonate Fluoroapatites Synthetic apatite Mineral apatite pH range Slope value Theoretical slope value Correlation coefficient <6.5 6.5–9.5 >9.5 <7 >7 −0.02 1.17 2.27 −0.09 −2.85 0 1 2 0 2 0.96 0.95 0.99 0.96 0.99 Note. [Ni]tot = 10−8 M, [apatite] = 10 g · l−1 , [KNO3 ] = 0.1 M. Table 4); they correspond to the surface complexation equilibria (Fig. 8) (1) SOH + Ni2+ ↔ SOHNi2+ (pH < 6.5; n = 0) K1 (2) SOH + Ni2+ ↔ SONi+ + H+ (6.5 < pH < 9.5; n = 1.39) K2 (3) SOH + Ni2+ + H2 O ↔ SONi(OH) + 2H+ (pH > 9.5; n = 2.27). K3 The mean values of the corresponding equilibrium constants are given in Table 5. B. Case of the mineral francolite. The log B versus pH curve presents only two linear parts, which imply only two surface complexation reactions (see Fig. 9): (1) SOH + Ni2+ ↔ SOHNi2+ (pH < 7; n = 0) K1 (3) SOH + Ni2+ + H2 O ↔ SONi(OH) + 2H+ (pH > 7; n = 2.85) K3. FIG. 9. Modeling nickel adsorption on the mineral francolite using the NEM; [Ni]tot = 10−8 M, m/V = 10 g · l−1 , [KNO3 ] = 0.1 M. The solid line denotes the modeling, and the solid circles denote experimental values. Equilibrium (2) involving one proton is no more involved, probably because the francolite surface sites are completely deprotonated at relatively high pH values (log K − = −10.2), when nickel is already partially hydrolyzed. It is also interesting to note that the values of K 1 and K 3 are similar for the synthetic and natural apatite (see Table 5). Modeling the Adsorption Curves Using the CCM In this modeling step, stoichiometries and formation constants of the surface complexes are extracted from experimental data using the FITEQL program. Neither the surface capacitance (equal to 18 or 70 F · m−2 ) nor the surface acidity constants or surface site concentrations (values previously determined and used for NEM) have been adjusted. A. Synthetic carbonate fluoroapatite. Two different sets of equilibria can describe the adsorption curves. The first one corresponds to the set previously determined using the NEM, and the second one consists of the reactions (1), (2), and (4) such as − (4) SOH + Ni2+ + PO3− 4 ↔ SOHNiPO4 . K4 The use of surface speciation codes often leads to several mathematically acceptable sets of reactions. In most cases, some of TABLE 5 Mean Values of the Equilibrium Constants of Nickel Complexation at the synthetic and Mineral Apatite Surface Determined Using the NEM Synthetic apatite FIG. 8. Modeling nickel adsorption on the synthetic carbonate fluoroapatite using the NEM; [Ni]tot = 10−8 M, m/V = 10 g · l−1 , [KNO3 ] = 0.1 M. The solid line denotes the modeling and the solid circles denote experimental values. Log K 1 Log K 2 Log K 3 3.5 ± 0.1 −3.2 ± 0.2 −11.5 ± 0.2 Mineral apatite 4.1 ± 0.1 −13.9 ± 0.2 312 PERRONE, FOUREST, AND GIFFAUT K 3 are not much changed by the suppression of Reactions (2) and (4). It is also interesting to note that the values of K 1 and K 3 (reported in Table 6) are very similar to those estimated with the NEM for both solids, which is in good agreement with the modeling of the acid–base properties of the solids and corroborates the hypothesis that chemical interactions prevail in the adsorption process. SUMMARY FIG. 10. Modeling nickel adsorption on the synthetic carbonate fluoroapatite using the CCM; [Ni]tot = 10−8 M, m/V = 10 g · l−1 , [KNO3 ] = 0.1 M. The solid line denotes the modeling, and the solid circles denote experimental values. them can easily be excluded on the basis of chemical or stereochemical considerations, but in the present case, both solutions are chemically acceptable. However, although it leads to the most satisfying adjustment of the experimental curves, the solution involving reaction (4) appears as the most questionable one. As a matter of fact, the surface complex formed by reaction (4) is much larger than the surface complex formed by reaction (2) and a competition between the phosphate groups present in the solution and those situated on the solid surface is expected to occur. Hence, the most acceptable solution is the set of reactions (1), (2), and (3) (see Fig. 10). B. Mineral francolite. In the case of the mineral francolite, a satisfying description of the experimental data was achieved using the reactions (1) and (3). This result is in complete agreement with that obtained by using the simpler NEM. It tends also to show that equilibria (1) and (3) prevail for the two apatites under comparison, at least in acidic (Reaction (1)) or basic (Reaction (3)) media. Reactions (1) and (3) are needed to obtain the best fit of the sorption curves, in the case of the synthetic apatite, but not really to explain the sorption results. As a matter of fact, K 1 and TABLE 6 Mean Values of the Equilibrium Constants of Nickel Complexation at the synthetic and Mineral Apatite Surface Determined Using the Constant Capacitance Model Synthetic apatite Log K 1 Log K 2 Log K 3 WSOS/DF 3.71 ± 0.1 −3.84 ± 0.2 −12.16 ± 0.2 1.9 Mineral apatite 4.15 ± 0.1 −14.01 ± 0.2 1.8 — As expected, Ni is highly sorbed onto carbonate fluoroapatites, with high sorption percentages and distribution coefficients reaching values on the order of 10 m3 kg−1 in neutral and basic media. — The sorption process seems to result from the formation of inner sphere complexes with the functional groups situated at the surface of the solids. — Modeling sorption isotherms leads to the determination of two thermodynamic equilibria. The values of the corresponding constants are similar for the two solids under consideration, the synthetic carbonate fluoroapatite, and the natural francolite. They do not also depend on the model chosen for the fitting. The chemical interactions seem to prevail over the electrostatic interactions. — The high content of impurities and carbonate of francolite do not modify the retention properties of this natural solid. This interesting behavior needs to be confirmed by investigating the sorption of other radionuclides, specially heavy elements for which a specific solution should be found concerning the elaboration of the engineered barriers of a nuclear waste repository. REFERENCES 1. Chen, X., Wright, J. V., Conca, J. L., and Peurrung, L. M., Environ. Sci. Technol. 31, 624 (1997). 2. Arey, J. S., Seaman, J. C., and Bertsch, P. M., Environ. Sci. Technol. 33, 337 (1999). 3. Murray, F. H., Brown, J. R., Fyfe, W. S., and Kronberg, B. I., Can. Mineral. 21, (1983). 4. Xu, Y., Schwartz, F. W., and Traina, S. J., Environ. Sci. Technol. 28, 1472 (1994). 5. Gauglitz, R., Holterdorf, M., Franke, W., and Marx, G., Radiochim. Acta 58/59, 253 (1992). 6. Ohnuki, T., Kozai, N., and Isobe, H., J. Nucl. Sci. Technol. 34, 58 (1997). 7. Chien, S. H., and Black, C. A., Soil Sci. Soc. Am. J. 40, 234 (1976). 8. Jahnke, R. A., Am. J. Sci. 284, 58 (1984). 9. LeGeros, R. Z., and LeGeros, J. P., in “Phosphate Minerals” (J. O. Nriagu and P. B. Moore, Eds.), p. 351. Springer-Verlag, Berlin, 1984. 10. Perrone, J., “Etude des propriétés de rétention des carbonate fluoroapatites vis-à-vis de Ni(II), Am(III) et Th(IV),” Thesis n◦ 5817. Université Paris XI, 1999. 11. Suzuki, T., Hatsushika, T., and Miyake, M., J. Chem. Soc. Faraday Trans. 1. 77, 1059 (1981). 12. Suzuki, T., Hatsushika, T., and Hayakawa, Y., J. Chem. Soc. Faraday Trans. 1. 78, 3605 (1982). 13. Reichert, J., and Binner, J. G. P., J. Mater. Sci. 31, 1231 (1996). 14. Tomson, M. B., and Nancollas, G. H., Science 200, 1059 (1969). SORPTION OF NICKEL ON CARBONATE FLUOROAPATITES 15. Régnier, P., Lasaga, A. C., and Berner, R. A., Am. Mineral. 17, 809 (1995). 16. Gulbrandsen, R. A., U.S. Geol. Surv. Prof. Paper. 700B, B9–B13 (1970). 17. Schuffert, J. D., Kastner, M., Emanuele, G., and Jahnke R. A., Geochim. Cosmochim. Acta 55, 2323 (1990). 18. Mandjiny, S., Zouboulis, A. I., and Matis, K. A., Sep. Sci.Technol. 30, 2963 (1995). 19. Sery, A., Manceau, A., and Neville-Greaves, G., Am. Mineral. 81, 864 (1996). 20. Jeanjean, J., Vincent, U., and Fedoroff, M., J. Solid State Chem. 108, 68 (1994). 21. Jeanjean, J., Fedoroff, M., Faverjon, F., and Vincent, U., J. Mater. Sci. 31 (1996). 22. Toulhoat, N., Potocek, V., Neskovic, C., Fedoroff, M., Jeanjean, J., and Vincent, U., Radiochim. Acta 74, 257 (1996). 23. Takeuchi, Y., and Arai H., J. Chem. Eng. Jpn. 23, 75 (1996). 24. Suzuki, T., Ishigaki, K., and Miyake, M., J. Chem. Soc. Faraday Trans. 1. 80, 3157 (1984). 25. Torstenfelt, B., Andersson, K., and Allard B., Chem. Geol. 36, 123 (1982). 26. Bacic, S., Komarov, V. F., and Vukovic, Z., in “Radiation Protection. Selected Topics, International Symposium, Dubrovnik, 2–6 October, 1989.” 27. Lazic, S., and Vukovic, Z., J. Radioanal. Nucl. Chem.: Art. 149, 161 (1991). 28. Lazic, S., Radiochim. Acta 62, 95 (1993). 29. Pandey, S., Patel, P. N., and Pujari, M., J. Surface. Sci. Technol. 6, 265 (1990). 30. Somasundaran, P., J. Colloid Interface Sci. 27, 659 (1968). 31. Bell, L. C., Posner, A. M., and Quirk, J. P., J. Colloid Interface Sci. 42, 250 (1973). 32. Chander, S., and Fuerstenau, D. W., J. Colloid Interface Sci. 70, 506 (1979). 33. Cases, J. M., Jacquier, P., Smani, S. M., Poirier, J. E., and Bottero J. Y., Ind. Miner. Mines Carrières: Tech. Janvier–Février, 1–12 (1989). 313 34. Wu, L., Forsling, W., and Schindler, P. W., J. Colloid Interface Sci. 147, 178 (1991). 35. Dzombak, D. A., and Morel, F. M. M., “Hydrous Ferric Oxide: Surface Complexation Modeling.” Wiley, New York, 1990. 36. Stumm, W., and Morgan, J. J., “Aquatic Chemistry,” second ed. Wiley, New York, 1981. 37. Sposito, G., J. Colloid Interface Sci. 91, 329 1983. 38. Davis, J. A., and Kent, D. B., in “Mineral-Water Interface Geochemistry” (M. F. Hochella and F. White, Eds.), Rev. Mineral., No. 23, p. 177, 1990. 39. Stipp, S. L., and Hochella, M. F., Jr., Geochim. Cosmochim. Acta 55, 1723 (1991). 40. Rönngren, L., Sjöberg, S., Sun, Z., Forsling, W., and Schindler, P. W., J. Colloid Interface Sci. 145, 396 (1991). 41. Van Cappellen, Ph., Charlet, L., Stumm, W., and Wersin, P., Geochim. Cosmichim. Acta 57, 3505 (1993). 42. Drot, R., and Simoni, E., Langmuir 15, 4820–4827 (1999). 43. Herbelin, A., and Westall, J., FITEQL, a computer program for determination of chemical equilibrium constants from experimental data. Version 3.2. Department of Chemistry, Oregon State University, Corvallis, OR (1996). 44. Baes, C. F., and Mesmer, R. E., Am. J. Sci. 281, 935 (1981). 45. Chandratillake, M. R., and Newton, G. W. A., Chemval project. Comparison of thermodynamical databases used in geochemical modeling. EUR 11891. EN, 1988. 46. Bradbury, M. H., and Baeyens, B., J. Contam. Hydrol. 27, 223 (1997). 47. Mattigod, S. V., and Sposito, G., in “Chemical Modeling in Aqueous Systems, A. C. S. Symposium” (E. A. Jenne, Eds.), Washington, DC, 1979. 48. Kurbatov, M. H., and Wood, G. B., J. Phys. Chem. 56, 698 (1952).