Langmuir 2009, 25, 2355-2362
2355
Phase Transformations, Ion-Exchange, Adsorption, and Dissolution Processes in Aquatic Fluorapatite Systems Åsa Bengtsson,* Andrei Shchukarev, Per Persson, and Staffan Sjo¨berg Department of Chemistry, Umeå UniVersity, SE-901 87, Sweden ReceiVed September 25, 2008. ReVised Manuscript ReceiVed NoVember 18, 2008 A synthetic fluorapatite was prepared that undergoes a phase transformation generated during a dialysis step. A surface layer with the composition Ca9(HPO4)2(PO4)4F2 is formed, which is suggested to form as one calcium atom is replaced by two protons. A surface complexation model, based upon XPS measurements, potentiometric titration data, batch experiments, and zeta-potential measurements was presented. The CaOH and OPO3H2 sites were assumed to have similar protolytic properties as in a corresponding nonstoichiometric HAP (Ca8.4(HPO4)1.6(PO4)4.4(OH)0.4) system. Besides a determination of the solubility product of Ca9(HPO4)2(PO4)4F2, two additional surface complexation reactions were introduced; one that accounts for a F/OH ion exchange reaction, resulting in the release of quite high fluoride concentrations (∼1 mM) that turned out to be dependent on the surface area of the particles. Furthermore, to explain the lowering of pHiep from around 8 in nonstoichiometric HAP suspensions to about 5.7 in FAP suspensions, a reaction that lowers the surface charge due to the readsorption of fluoride ions to the positively charged Ca sites was introduced: ≡CaOH2+ + F- H ≡CaF + H2O. The resulting model also agrees with predictions based upon XPS and ATR-FTIR observations claiming the formation of CaF2(s) in the most acidic pH range.
1. Introduction Apatites are a diverse class of phosphate minerals that are important in a great variety of natural and industrial processes. Hydroxyapatite (Ca5(PO4)3OH), HAP, is very similar to the biological calcium-deficient carbonate HAP that is the main inorganic constituent in mammalian bone and teeth.1 It has therefore been suggested as a good material for artificial bone and tooth implants. It is believed that fluoride ions present in saliva strengthen the enamel in teeth by completely or partially converting the biological calcium-deficient carbonate HAP in the enamel to a less-soluble fluorapatite (Ca5(PO4)3F), FAP.2,3 The lower solubility of FAP can be understood from both bulk structural and surface considerations. The lattice energy of FAP is larger than that of HAP, thus FAP is thermodynamically more stable, that is less soluble.4 The fluoride ion sits in the plane of three adjacent calcium atoms of the bulk and is thereby providing more stability to the molecular units. The hydroxyl ion of the HAP bulk sits, on the other hand, out of this plane, and consequently provides less stability to this unit. Also, because it is larger than the fluoride ion, vacancies are common occurrences in naturally as well as synthetic HAP particles. Surface stabilization by fluoride ions is an important contributor to the lower solubility of FAP, but it has been suggested that fluorhydroxy-apatites with variable F/OH ratios exhibit even lower reactivity compared to the pure fluor- and hydroxyapatites due to strong OH-F hydrogen bonding in the apatite structure.5 Fluoride ions can also (re)adsorb to calcium surface groups and promote a network of surface hydrogen bonding with vicinal surface OH groups (e.g., protonated phosphate and calcium hydroxyl surface groups) and consequently enhance the stability of surface atoms. Some studies * To whom correspondence should be addressed. E-mail: asa.bengtsson@ chem.umu.se. Fax: + 46 90 786 91 95. (1) Driessens, F. C. M.; Verbeeck, R. M. H. J. Cryst. Growth. 1981, 53(1), 55–62. (2) Knappwost, A. Angew. Chem. 1956, 68, 371. (3) Misra, D. N. J. Colloid Interface Sci. 1999, 220, 387–391. (4) Flora, N. J.; C, H. Y.; Jenkins, H. D. B. Inorg. Chem. 2004, 43, 2340–2345. (5) Veiderma, M.; To˜nsuaadu, K.; Knubovets, R.; Peld, M. J. Org. Chem. 2005, 690, 2638–2643.
have in fact shown that fluoride adsorbed to the HAP surface is more effective than fluoride incorporated in the structure of HAP at inhibiting mineral dissolution.6 As the solubility of a mineral more reflects the composition of surface layer(s) and not the composition of the bulk phase, a consequence of this variability is the formation of compounds with different solubility products. This is the third article in a series dealing with surface coordination chemistry and surface complexation modeling of different apatite systems. The first article7 describes goethite(R-FeOOH) promoted dissolution of FAP. The results clearly demonstrate goethite being a very strong sink for phosphate ions, which drives the dissolution for FAP and thereby alters the phosphate speciation in this two-mineral system. The second article8 deals with solubility characteristics and surfacecomplexation properties of a nonstoichiometric HAP (Ca8.4(HPO4)1.6(PO4)4.4(OH)0.4). The variation of the Ca/P ratio with pH was interpreted as a readsorption of phosphate ions to positively charged Ca sites at the surface. In the present study, we investigate the dissolution of a wellcharacterized synthetic FAP (Ca9(HPO4)2(PO4)4F2) over a large pH interval. The multitechnique approach employed in this study (batch experiments, potentiometric titrations, zeta-potential measurements, XPS and ATR-FTIR spectroscopy) provides calibration data for a surface complexation model coupled to a FAP dissolution model. The proposed model provides a relationship between the surface speciation and mineral dissolution in the 2-11 pH range.
2. Experimental Section 2.1. Synthesis and Characterization of Fluorapatite. FAP was synthesized in the laboratory according to the method by Penel et al.9 A solution of 1000 mL of 0.4 M Ca(NO3)2 · 4H2O (Scharlau) was added dropwise to a boiling solution of 1000 mL 0.24 M (6) Wong, L.; Cutress, T. W.; Duncan, J. F. J. Dent. Res. 1987, 66, 1735–1741. (7) Bengtsson, Å.; Lindegren, M.; Sjo¨berg, S.; Persson, P. Appl. Geochem. 2007, 22(9), 2016–2028. (8) Bengtsson, Å.; Shchukarev, A.; Persson, P.; Sjo¨berg, S. Geochim. Cosmochim. Acta. 2009, 73(2), 257–267. (9) Penel, G.; Leroy, G.; Rey, C.; Sombret, B.; Huvenne, J. P.; Bres, E. J. Mater. Sci.: Mater. Med. 1997, 8(5), 271–276.
10.1021/la803137u CCC: $40.75 2009 American Chemical Society Published on Web 01/13/2009
2356 Langmuir, Vol. 25, No. 4, 2009
Bengtsson et al.
Table 1. Thermodynamic Data (I ) 0.1 M, 25 °C) Used in the Model Calculationsa species
log β
H+
Ca2+
HPO42-
Na+
F-
≡CaOH
≡ OPO3H2
≡F
H Ca2+ HPO42Na+ F≡CaOH ≡OPO3H2 ≡F OHPO43H2PO4H3PO4 Ca(OH)+ Ca(PO4)Ca(HPO4) Ca(H2PO4)+ HF HF2CaF+ ≡OPO3H≡OPO3Na≡CaOH2+ ≡CaOPO3H≡CaF ≡OH Ca8.4(HPO4)1.6(PO4)4.4(OH)0.4b Ca9(HPO4)2(PO4)4(F)2b CaF2
0 0 0 0 0 0 0 0 -13.78 -11.65 6.74 8.65 -12.9 -6.45 1.73 7.32 2.92 3.51 0.82 -1.11 -11.08 8.41 11.63 11.40 -8.78 23.27 48.08 9.73
1 0 0 0 0 0 0 0 -1 -1 1 2 -1 -1 0 1 1 1 0 -1 -2 1 1 1 -1 -4.8 -4 0
0 1 0 0 0 0 0 0 0 0 0 0 1 1 1 1 0 0 1 0 0 0 0 0 0 8.4 9 1
0 0 1 0 0 0 0 0 0 1 1 1 0 1 1 1 0 0 0 0 0 0 1 0 0 6 6 0
0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0
0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 2 1 0 0 0 0 1 -1 0 2 2
0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 0
0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0
+
phase sln sln sln sln sln srfc srfc srfc sln sln sln sln sln sln sln sln sln sln sln srfc srfc srfc srfc srfc srfc sld sld sld
a sln, srfc, and sld stands for solution, surface, and solid respectively. b The solid phase included is the composition of the surface layers of HAP (Bengtsson et al.8 and FAP (this study)). All constants, except those for the surface complexation model, were taken from Smith and Martell.14 The surface complexes were taken from Bengtsson et al.8 and this study.
Figure 1. Atomic ratios at the surface of the FAP pastes equilibrated for 3.5 months, in the pH range 2-11 determined by XPS. The analyses were performed both in liquid nitrogen (LN) and in room temperature (RT).
(NH4)2HPO4 (Merck) and 0.18 M NH4F (Merck) over a 1 h period. Solution pH was kept to a value of about 9 by addition of small aliquots of 25% NH3 (Scharlau). The product was matured for an additional hour at 80 °C and then cooled to room temperature. It was then washed with deionized and boiled water and dialysed in Millipore 12-14000 D tubes for about three months. The product was dried at room temperature, gently mortared, and then stored in polyethene bottles. The product was characterized by X-ray diffraction and was found to produce patterns characteristic for FAP. The specific surface area, measured by the single point BET (N2) method (Micrometrics Flowsorb II 2300), was 10.7 m2/g. The elemental composition of the bulk was analyzed by Analytica AB (Luleå, Sweden) using ion selective electrode (ISE), inductively coupled plasma atomic emission spectrometry (ICP-AES), or sector field mass spectrometry (ICP-
SFMS) and the ratios between Ca, P, and F were found to be Ca/P 1.7, Ca/F 5.5, and P/F 3.2. 2.2. Titrations. Potentiometric titrations were performed to study the dissolution of FAP and possible surface complexation reactions at high pH. The automatic system used was designed and built at the Department of Chemistry, Umeå University.10 The cell was immersed in an oil thermostat at a constant temperature of 25 ( 0.05 °C. The free H+ concentration was determined by measuring the emf (E) of the cell: - Ag, AgCl(s) | ionic medium (0.1 M NaCl) | equilibrium solution | glass electrode + (10) Ginstrup, O. Chem. Instr. 1973, 4, 141–155.
Aquatic Fluorapatite Systems
Langmuir, Vol. 25, No. 4, 2009 2357
E0 (in mV) is an apparatus constant for the cell. Ej (in mV) is the liquid junction potential between the solution and the salt bridge and was calculated according to Sjo¨berg et al.11 The titration vessel contained suspensions of FAP in 0.1 M Na(Cl), and the initial suspension density of fluorapatite was 10 g/L. The suspensions had been equilibrated in ionic medium for at least 1 month before starting the titrations. The titrations were started as coulometric titrations (to generate hydroxide ions at a constant volume) from the pH of suspension (pH 5.3 for a 10 g/L suspension) to pH 10, thereby preserving the initial suspension density, and then as acidimetric titrations by the addition of standardized HCl to pH ≈ 5.3. Titrant addition was automatically carried out when the electrodes reached a drift of 0.2 mV/h. This criterion was monitored by emf measurements every 30 min with a minimum reading period of 4 h prior titrant addition and was always met within 10 h. The titration vessel was continuously purged with a low flow of N2(g) to avoid any interference from atmospheric O2(g) and CO2(g). The suspension was also continuously stirred with a polyethylene propeller, thereby eliminating the risk of grinding the particles and changing their surface area, as would otherwise occur with a magnetic stirring bar. Aliquots of the suspension were regularly collected during the titrations, and these aliquots were used for the zeta-potential measurements. 2.3. Batch Experiments. 2.3.1. Dissolution. FAP dissolution was studied in a series of batches. A suspension of FAP and appropriate amounts of acid or base to cover a large pH interval were mixed in polyethene centrifuge tubes. Two of the batches were
prepared as three identical replicates. The total suspension density of FAP in each batch was 7.7 g/L, representing a total surface area of 82 m2/L. Atmospheric CO2(g) was eliminated from the solutions with N2(g), throughout the manipulations. The batches were equilibrated on an end-overend rotator for 3.5 months, and after measuring the final pH values, all samples were centrifuged at 4000 rpm for 20 min. The resulting wet pastes and a small volume of the corresponding supernatants were immediately analyzed with ATR-FTIR spectroscopy. The procedure for this analysis is described in section 2.5. The remaining supernatant was filtered thorough a 0.22 µm filter, acidified to pH < 2, and then stored at 4 °C until it was analyzed for dissolved calcium, phosphate, and fluoride in solution. Calcium was analyzed using atomic absorption spectrometry (PerkinElmer 3110) and for phosphate and fluoride using liquid ion chromatography (Metrohm) with conductivity detection. 2.3.2. Ion Exchange. A series of batches was prepared to investigate the effect on pH and pF in solution due to the ion exchange between water molecules in solution and the fluoride on the surface of FAP. Batches with suspension densities of 0.5, 1.0, 2.0, 4.0, 5.0, 7.7, 8.0, and 10.0 g/L were equilibrated on an end-overend rotator for 2 months. Each batch was prepared as three individual replicates. Values of pH were measured in the suspensions before they were centrifuged at 4000 rpm for 20 min. The supernatants were filtered through a 0.22 µm Millipor filter, acidified, and analyzed for fluoride using IC. 2.3.3. Kinetics. Studies were performed to investigate the kinetics of the ion exchange between water molecules in solution and the fluoride on the surface of the FAP. A series of batches containing 0.050 ( 0.0003 g FAP and 5 mL 0.1 M NaCl (to achieve a suspension density of 10 g/L) were equilibrated in test tubes on an end-overend rotator and sequentially withdrawn after 5 min, 30 min, 40 min, 1 h, 3 h, 6 h, 1 day, 2 days, 3 days, 6 days, 17 days, and 27 days. Each batch was prepared as three identical replicates. Values of pH were measured in the suspensions before they were centrifuged at 4000 rpm for 20 min. The supernatants were filtered through a 0.22 µm Millipor filter, acidified, and analyzed for phosphate and fluoride using IC, and for calcium using AAS in the same aforementioned methods. 2.4. X-ray Photoelectron Spectroscopy (XPS). The centrifuged pastes from the experimental series were analyzed with a Kratos Axis Ultra electron spectrometer using a monochromatic Al KR source operated at 150 W, a hybrid lens system with magnetic lens, a and charge neutralizer. The fast-freezing technique was used to preserve the water content in the vacuum. This procedure12 involves precooling in the end of the sample transfer rod (10 min at -170 °C) and then waiting 45 s after loading the wet paste before pumping the introducing chamber. After pumping to 10-5 Pa, the frozen paste was transferred to the precooled (-160 °C) manipulator where it was kept until a base vacuum of 2 - 4 (10-7) Pa in the analysis chamber was reached. A wide spectrum (pass energy 160 eV) and narrow scans of all detected elements (pass energy 20 eV) were acquired. The temperature of the manipulator did not exceed -150 °C, nor were there detectable pressure changes during the whole experiment. The wide spectrum taken at the end of experiment was identical to the one taken at the beginning. After the measurements at liquid nitrogen temperature, the sample was kept in the analysis chamber overnight to warm to room temperature to remove water. The measurements were repeated the next day to follow the changes at the FAP surface caused by water loss. The binding energy (BE) scale was referenced to the C1s line of aliphatic carbon, set at 285.0 eV. Processing of the spectra was accomplished with Kratos software and the CasaXPS program package. 2.5. Attenuated Total Reflectance Fourier Transform (ATRFTIR) Spectroscopy. A Bruker IFS 66 v/s spectrometer fitted with a deuterated triglycine sulfate (DTGS) detector was used to collect ATR-FTIR spectra of the empty cell, the supernatant, and the paste for each sample. The spectra were recorded in vacuum (5 mbar)
(11) Sjo¨berg, S.; Ha¨gglund, Y.; Nordin, A.; Ingri, N. Marine Chem. 1983, 13, 35–44.
(12) Shchukarev, A.; Rosenqvist, J.; Sjoberg, S. J. Electron Spectrosc. Relat. Phenom. 2004, 137-40, 171–176.
Figure 2. ATR-FTIR spectra of the wet FAP pastes at pH 3-11.
The E of the cell was measured and the free H+ concentration was calculated with:
E ) E0 + 59.157 log[H+] + Ej
(1)
2358 Langmuir, Vol. 25, No. 4, 2009
Bengtsson et al.
Figure 3. (a) pH, (b) calcium, (c) phosphate, and (d) fluoride dissolved from FAP as a function of time. [FAP] ) 10 g/L, I ) 0.1 M NaCl. The results are presented as the mean of three individual experiments and error bars represent the standard deviations.
with a horizontal ATR accessory and a diamond crystal as the reflection element (SensIR Technologies). The FAP pastes and supernatants were applied to the diamond crystal surface directly, and then a vacuum proof lid was placed over the samples to protect them from drying. Five hundred scans were collected for each sample over the range 370-7500 cm-1, and the angle of incidence for the setup was approximately 45°, which is far from the critical angle. Sample spectra were interpreted after subtracting spectra of the empty cell and the corresponding supernatant using the OPUS 4.0 software. 2.6. Zeta-Potential Measurements. Zeta-potential measurements of FAP were performed in the pH 4-11 range using a Malvern Instruments Zetasizer 4 instrument. The suspension density of FAP was 10 g/L and the ionic strength 0.1 M Na(Cl). The resulting isoelectric point was found to be at pHiep ) 5.7, which is the same value obtained by Jarlbring et al.13 for a synthetic FAP with a specific surface area of 17.7 m2/g and suspension density of 25 g/L.
3. Surface Complexation Modeling The dissolution of FAP is strongly pH dependent and involves the release of calcium, phosphate, and fluoride ions. Furthermore, these ions will take part in protonation, complexation, and precipitation reactions. -(3-n) + PO3(n ) 1,2,3) 4 + nH S HnPO4 (n-1)+ + (n ) 0,1,2) Ca2++PO34 +nH S CaHnPO4 -
+
nF +H
S HF(1-n)+ (n ) 1,2) n -
Ca +2F S CaF2(s) 2+
(1a) (2) (3) (4)
Surface reactions such as adsorption and desorption of H+ as well as the readsorption of phosphate species according to reactions eqs 5-8 also have to be taken into account.8
≡CaOH + H+ S ≡CaOH+ 2
(5)
≡OPO3H2 S ≡OPO3H-+ H+
(6)
2≡CaOH+ 2 +HPO4 S ≡CaOPO3H + H2O +
-
+
≡OPO3H2 + Na S ≡OPO3Na + 2H
(7) (8)
The formation constants for these reactions were included in the calculations to quantify the dissolution of FAP and the coupled sorption reactions. The constants for eqs 1-4 were taken from Smith and Martell14 and eqs 5-8 are according to Bengtsson et al.8 Furthermore, the possibility that fluoride ions will take part in ion exchange as well as readsorption processes will be considered:
≡F + H2O S ≡OH + H+ + F≡CaOH2+
-
+ F S ≡CaF + H2O
(9) (10)
The formation of a number of charged surface species according to eqs 5-8 will result in a charge accumulation at the FAP surface. Therefore, the different surface equilibrium constants must be corrected for the Coulombic energy of the charged surface. The conditional constant, βs11(cond.) defining a protonation reaction of a generic surface site ≡SOH:
≡SOH + H+ S ≡SOH+ 2 is related to the intrinsic constant equation:
βs11(intr.)
βs11(intr.) ) βs11(cond.)eFψ⁄RT
(11) according to the
(12)
where Ψ is the acting surface potential calculated according to eq 13.
Aquatic Fluorapatite Systems
Langmuir, Vol. 25, No. 4, 2009 2359
Figure 6. Results of titrations of FAP in 0.1 M Na(Cl) and 25 °C at an initial suspension density of 10 g/L. The method of titrant addition is specified in the legend. The model line was derived from the constants in Table 1.
Figure 4. Dissolution of FAP illustrated by the concentrations of calcium, phosphate, and fluoride in solution in batch experiments equilibrated for 3.5 months. Suspension density ) 7.7 g/L; I ) 0.1 M Na(Cl); T ) 25 °C. (a) experimental solution composition and model predictions (lines) and (b) element ratios (no model shown, the lines were derived as a visual aid.).
Figure 7. Surface speciation of FAP (7.7 g/L) equilibrated in 0.1 M NaCl at 25 °C from the model of Table 1.
charge distributions of the solid water interface. The strategy in the present modeling approach is to try to explain experimental data with as few parameters as possible so the models used in the present work are developed assuming a constant capacitance for the electrical double layer at the charge surface.15 The activity coefficients of the different species in solution are controlled by an ionic medium of constant ionic strength (I ) 0.1 M). In addition the activity coefficients of the surface species were assumed to be constant. When necessary, the individual activity coefficients γi of an ionic species were recalculated to an ionic strength (I) of 0.1 utilizing Davis equation,16 where zi denotes the ionic charge number: Figure 5. Simulated and measured pH and pF as a function of the mass of FAP equilibrated in 0.1 M NaCl. (O) and (b) denote the experimental values of pH and pF, respectively. The lines represent the calculated pH and pF values as a function of suspension density of FAP and are derived using the constants in Table 1. The results are presented as a mean of three individual experiments, and the deviation between the replicates were found to be less than 2%.
Ψ ) TσF ⁄ (sAC)
(13)
where Ψ is the surface potential (V), C is a constant with the dimensions of specific capacitance (CV-1/m2), Tσ is the molarity of total surface charge (mol/L), s is the specific surface area (m2/g), and A is the suspension density (g/L). Different more or less sophisticated surface complexation models are in use including electrical double-layer theories that allow for mechanistic multilayer, multisite assumptions with
log γi ) -0.509zi2(I1⁄2 ⁄ (1 + I1⁄2) - 0.3I)
(14)
In the different calculations the computer code WinSGW,17 which is based on the SOLGASWATER algorithm,18 was used. The thermodynamic data used for the calculations are listed in Table 1. The equilibrium model shown in Table 1 is defined by the eight components: H+, Ca2+, Na+, HPO42-, F-, ≡CaOH, (13) Jarlbring, M.; Gunneriusson, L.; Forsling, W. J. Colloid Interface Sci. 2005, 285(1), 206–211. (14) Smith, R. M.; Martell A. E. Inorganic Complexes; Plenum Press: New York, 1976. (15) Schindler, P. W.; Gamsja¨ger, H. Kolloid-Z. Z. Polym. 1972, 250, 759– 765. (16) Davis, C. W. Ion Association;Butterworth-Heinemann: Woburn, MA, 1962. (17) Karlsson, M; Lindgren, J. 2006.http://www.dagger.mine.nu/MAJO/ winsgw.htmTH. (18) Eriksson, G. Anal. Chim. Acta 1979, 112(4), 375–383.
2360 Langmuir, Vol. 25, No. 4, 2009
Bengtsson et al.
≡OPO3H2, and ≡F. A general equilibrium reaction can be written as given by:
pH++qCa2++rNa++sHPO24+tF- + u ≡CaOH + V ≡OPO3H2 + x ≡ F S (H+)p(Ca2+)q(Na+)r(HPO24 )s × (F-)t(≡CaOH)u(≡OPO3H2)V(≡F)x (15) This equation defines the formation constant βpqrstuVx. The total concentration of the different components is obtained by summarizing over the different columns, viz.
[H+]tot ) [H+]tot(sln) + [H+]tot(srfc) + [H+]tot(sld) ) [H+] + [OH-] - [PO34 ] + [H2PO4 ] + 2[H3PO4] - [Ca(OH) ] -
+
[Ca(PO4) ] + [Ca(H2PO4) ] + [HF] + [HF2]+ -
-
[≡CaOH+ 2 ] - [≡PO4H ] - 2[≡OPO3Na ] + [≡CaOPO3H-] + [≡CaF]-[≡OH]4nCa9(HPO4)2(PO4)4(F)2)4.8n(Ca8.4(HPO4)1.6(PO4)4.4(OH)0.4) (16) Here, n denotes the number of moles of the solid phases per liter of solution. In the experiments, acid (H+) or base (OH-) have been added. This equation shows that these additions will be consumed due to i) dissolution of HAP, ii) surface complexation reactions, and iii) complexation in solution. In addition to these data, the total concentration of calcium and phosphate in solution is known from chemical analysis.
[Ca2+]tot(sln) ) [Ca2+] + [Ca(OH)+] + [Ca(PO4)-] + [Ca(HPO4)] + [Ca(H2PO4)+] + [CaF+] (17) [HPO42-]tot(sln) ) [PO43-] + [HPO42-] + [H2PO4-] + [H3PO4] + [Ca(PO4)-] + [Ca(HPO4)] + [Ca(H2PO4)+] (18) [F-]tot(sln) ) [F-] + [HF] + 2[HF2-] + [CaF+] (19) These experimental values can also be compared with model dependent calculated values, which can be derived from Table 1. In the different model calculations error squares of sums (U):
U(H) ) Σ([H+]totcalcd - [H+]totexptl)2
(20)
U(Ca) ) Σ([Ca2+]tot(sln)calcd - [Ca2+]tot(sln)exptl)2 (21) U(P) ) Σ([HPO42-]tot(sln)calcd - [HPO42-]tot(sln)exptl)2 (22) U(F) ) Σ([F-]tot(sln)calcd - [F-]tot(sln)exptl)2
(23)
where used in fitting a model to experimental proton, [Ca2+]tot(sln) and [HPO42-]tot(sln) and [F-]tot(sln) data. The ultimate goal of the equilibrium analysis is to design a model that gives a good fit to minimize over all error squares U(H) + U(Ca) + U(P) + U(F). This analysis involves a determination of composition and stability of the FAP phase as well as the prevailing surface complexes. It is assumed that literature data describing complexation in solution is known, and no attempts will be made to refine any of the solution complexation constants. It is also assumed that the surface complexation constants eqs 5-8 are the same as those in the HAP system.8 This means that the number of unknown stability constants will be limited to the solubility product for FAP, ion exchange, and readsorption equilibria involving fluoride ions, as seen in eqs 9 and 10.
The strategy of the equilibrium analysis is the determination of: i) An approximate value of the solubility product of FAP minimizing U(Ca) and U(P) using data where the dissolution reaction is predominating (pH e 5). In these calculations the high solubility will cause significant changes of the ionic strengths, which are corrected for in the calculations using eq 14. Composition of the solid phase is obtained from XPS data ii) A determination of composition and stability of ion exchanged F- and the readsorption reaction involving F- from the potentiometric titration data. In these calculations the solubility product for FAP will be included. These calculations also involved a determination of the capacitance value C (eq 13). iii) By using an iterative procedure, constants obtained under i) and ii) will be refined, minimizing U(H) + U(Ca) + U(P) + U(F). A prerequisite for surface complexation modeling is a surface characterization in terms of surface site compositions and surface site densities.
4. Surface Site Characterization The protolytic surface properties of FAP particles are assumed to be controlled by different phosphate, fluoride, and calcium hydroxyl sites. Recent 1H, 31P, and 31P MAS NMR studies19 have shown the presence of active phosphorus surface sites as well as bulk phosphate of fluorapatite. The 31P MAS NMR spectrum of FAP at pH 5.9 (close to pHiep) revealed one dominant resonance line at 2.9 ppm, which was ascribed to bulk phosphate groups. Two weaker shoulders at 5.4 and 0.8 ppm were assigned to the unprotonated and protonated phosphate sites, respectively. A resonance peak at -4.5 ppm also appeared at pH of 3.5 from a ≡POxH2 surface species (most likely ≡PO3H2 at this low pH value). The protolytic properties of calcium sites at the surface of synthetic FAP particles were studied by Sandstro¨m et al.20 also using 1H and 31P MAS NMR methods. Three possible forms of calcium hydroxyl sites were suggested, viz. ≡CaOH2+, ≡CaOH, and ≡Ca(OH)2-. Furthermore, their mutual ratios were found to vary with pH. Alkalimetric titrations of synthetic carbonate-free fluorapatites13 were used to suggest site densities in the order of 2.34-2.95 sites/nm2 in ionic strengths of 0.1 and 0.5 mol/L NaNO3. A site density of 3.1 sites/nm2 was also proposed by titrimetry for synthetic carbonate fluorapatites.21 Furthermore, it was assumed that these sites have amphoteric properties. Surface site densities for natural fluorapatite were also calculated by Wu et al.,22 revealing a total crystallographic value of 12.3 sites/nm2. Only a fraction of these sites is however expected to be proton active. Kukura et al.23 performed radioisotope measurements to determine the site density of apatite and concluded that the calcium sites amount to 4.3 sites/nm2 and the phosphate sites to 3.0 sites/nm2. These values are in close agreement with the results from Jarlbring et al.13 and Perrone et al.21 and were chosen for the model calculation in our study.
5. Results and Discussion 5.1. Surface Analysis of FAP. The results of the XPS measurements of the pastes from the batch experiments are shown (19) Jarlbring, M.; Sandstrom, D. E.; Antzutkin, O. N.; Forsling, W. Langmuir 2006, 22(10), 4787–4792. (20) Sandstro¨m, D. E.; Jarlbring, M.; Antzutkin, O. N.; Forsling, W. Langmuir 2006, 22(26), 11060–11064. (21) Perrone, J.; Fourest, B.; Giffaut, E. J. Colloid Interface Sci. 2002, 249(2), 441–452. (22) Wu, L.; Forsling, W.; Schindler, P. W. J. Colloid Interface Sci. 1991, 147, 178–185.
Aquatic Fluorapatite Systems
in Figure 1 and provide information on the composition of the top atomic layers of the particles. The samples equilibrated in the pH 4-11 range reveal ratios of Ca/P ) 1.5, Ca/F ) 3.9-4.5, and P/F ) 2.5-2.9. The Ca/P ratio remains constant at 1.5 throughout the whole pH range and indicates that a Ca-depleted surface is formed during the dialysis step. Furthermore the constant value within this pH range indicates a predominating congruent dissolution process to take place. A small pH dependence on the surface composition of fluoride, due to F-/ OH- exchange reactions as it will be shown in the later sections, also causes the Ca/F and P/F ratios to increase slightly with increasing pH values. The samples equilibrated at pH < 4 reveal considerably lower Ca/F and P/F ratios and larger Ca/P ratios than those of the bulk material (Table 1). These results reveal a preferential accumulation of both Ca and F at the surface of the paste. Our thermodynamic calculations (Table 1) provide evidence that this enrichment is caused by the precipitation of CaF2(s). These XPS-results indicate that the composition of the Ca-depleted FAP surface has the composition Ca9(HPO4)2(PO4)4F2(s). This depletion is accomplished by the replacement of one Ca-atom by two protons, an exchange reaction also suggested by Chaı¨rat et al.24 Similar exchange reactions have also been proposed in the HAP system,8,25 in which one Ca and one hydroxyl group were replaced by one proton. As a whole, the XPS data provided strong evidence for a depletion of calcium from the FAP surface and an effect of pH on the Ca/F and P/F ratios due to fluoride adsorption, ion exchange, and CaF2(s) precipitation. The dissolution data and the models of the following section will also provide further evidence for these processes. The FAP pastes obtained from the batch experiments were also analyzed by means of ATR-FTIR (Figure 2). The spectra exhibited little changes over the range of pH values, in contrast to those of HAP,12 which revealed the presence of a band at 1130 cm-1, indicative of the activity of a protonated phosphate surface site. The FAP particles of this study (10.7 m2/g), in contrast to those of HAP (80.5 m2/g), are of considerably lower specific surface area and consequently have a lower surface throughput in the spectrometer. A 2D IR correlation spectroscopic analysis26 of these spectra, which typically retrieves more subtle band components in highly overlapped spectra, could not resolve any other additional features. There has also been studies showing bands in the 3500-3600 cm-1 region assigned to OH stretching modes in hydrogen bonds (OH-F)5,27 but no such bands could be detected in these spectra. The spectra of the FAP pastes therefore only provide information on the vibrations of the bulk components. As such, the bands of the 800-1200 cm-1 region arise from bulk deprotonated phosphate units.28 It is noteworthy to mention that the HAP particles exhibited similar spectra but that they also revealed the presence of pH-dependent features arising from the protonation of surface phosphate groups. It should thereby be expected that such groups also exist on the FAP surface, although they cannot be detected by vibration spectroscopy. This assumption is also corroborated by the results from the NMR studies of FAP suspensions19 as discussed in Section 4. 5.2. Kinetics. Experimental investigations of dissolution processes of minerals as complex as apatites require that the (23) Kukura, M.; Bell, L. C.; Posner, A. M.; Quirk, J. P. J. Phys. Chem. 1972, 76(6), 900–904. (24) Chaı¨rat, C.; Oelkers, E. H.; Schott, J.; Lartigue, J. E. Geochim. Cosmochim. Acta 2007, 71, 5888–5900. (25) Brown, P. W.; Martin, R. I. J. Phys. Chem. 1999, 103, 1671–1675. (26) Noda, I. Appl. Spectrosc. 1989, 44(4), 550–561.
Langmuir, Vol. 25, No. 4, 2009 2361
minerals be first conditioned before starting the experiments. Literature values for the solubility product log Ks0 of FAP range from -58.5 to -70.29,30 These large discrepancies could partly be due to kinetic factors, which could be more important than they traditionally have been accounted for. For example, Somasudaran31 showed the necessity of equilibrating a natural fluorapatite in aqueous solutions for at least 2 weeks until reaching constant values of zeta potential and pH. The results from batch experiments of FAP reacted in 0.1 M NaCl as a function of time are shown in Figure 3. The concentrations of calcium, phosphate, and fluoride in solution rapidly increased and pH rapidly decreased in the first 5 days. The suspensions were, however, still not fully equilibrated after 27 days of reaction time. Long-term reactions contributing to the sluggish dissolution kinetics of FAP, in addition to the ratedetermining surface bond-breaking mechanisms,32 also include the fluoride/water exchange reactions:33
≡F + H2O S ≡ OH + H+ + F-
(24)
5.3. Dissolution and Surface Complexation of FAP. Equilibrium values of the calcium, phosphate, and fluoride concentrations released from FAP during batch experiments as a function of pH are shown in Figure 4. The constant Ca/P ratio and the systematic trend of soluble concentrations of phosphate and calcium (part a of Figure 4) indicate apparent congruent dissolution behavior of the surface layer of FAP. Fluoride is however preferentially leached from FAP above pH 5.3 but is less soluble than phosphate and calcium below this value. The Ca/F and P/F ratios (part b of Figure 4) are in fact almost 100 in acidic conditions, which is consistent with the XPS data of Figure 1, and are the result from the precipitation of CaF2(s), as it will be further discussed in the following section. 5.4. pH0 as a Function of Suspension Density. An important observation is the fact that the initial pH of the suspensions without any addition of acid or base (pH0) was found to decrease with increasing suspension density, whereas pH0 of a HAP suspension is more or less independent of the suspension density and pH0 ≈ 8.1 (10 g/L solid),8 the value in a corresponding FAP suspension is 5.35. Furthermore, [F-]tot(sln) increases with increasing suspension density (c.f. Figure 5). This pH/pF dependency clearly shows upon a surface area related reaction and is ascribed to the ion exchange reaction between OH and F (eq 24). This ion exchange reaction might explain the lowering in pH0 with increasing solid concentration but cannot explain the drop in pHiep (FAP) as we are dealing with a reaction (eq 24) involving uncharged species. The challenge in the modeling work is to design a model that will explain these observations besides finding a model that will fit experimental batch and potentiometric titration data. 5.5. Potentiometric Titration Data. Potentiometric acid/base titrations of the FAP surface were performed in the pH range 5.3 e pH e 10.1, where there is very little dissolution taking place (Figure 4). Alkalimetric titrations were performed starting at the pH of suspension of 5.3 ( 0.05 with a suspension density of 10 g/L. When reaching pH 10.1, acid was added to titrate back to pH 5.3. Both the reproducibility and reversibility of the titrations (27) Ishikawa, T.; Teramachi, A.; Tanaka, H.; Yasukawa, A.; Kandori, K. Langmuir 2000, 16, 10221–10226. (28) Tejedor-Tejedor, I. M.; Anderson, M. A. Langmuir 1990, 6, 602–611. (29) Jaynes, W. F.; Moore, P. A.; Miller, D. M. J. EnViron. Qual. 1999, 28, 530–536. (30) Valsami-Jones, E.; Ragnarsdottir, K. V.; Putnis, A.; Bosbach, D.; Kemp, A. J.; Cressey, G. Chem. Geol. 1998, 151(1-4), 215–233. (31) Somasudaran, P. J. Colloid Interface Sci. 1968, 27, 659–666. (32) Lasaga, A. C.; Lu¨ttge, A. Am. Mineral. 2004, 89(4), 527–540. (33) Dorozhkin, S. V. J. Cryst. Growth. 1997, 182, 133–140.
2362 Langmuir, Vol. 25, No. 4, 2009
were confirmed by repeated sets of these titrations. The potentiometric titration data are visualized in Figure 6. Reproducibility and reversibility was found to be good, and the maximum number of hydroxide ions reacted per gram FAP was found to be around 0.2 mM. 5.6. Final Model. The equilibrium analysis of the present eight-component system assumes that the intrinsic constants for the surface reactions related to the surface components tCaOH and tOPO3H2 are the same for the Ca-depleted FAP surface as for the nonstoichiometric HAP (Table 1). The calculations were initiated by a determination of the solubility product for the Ca-depleted FAP surface layer. By minimizing U(Ca) + U(P), an initial guess value was obtained that could be used to account for the dissolution of FAP during the potentiometric titrations. By minimizing U(H) for the potentiometric titration data,a good fit was obtain by introducing the ion exchange reaction that is a pH0(calcd) that agrees with pH0(exptl). However, it was found that the calculated pHiep with Ψ ) 0 attained a value typical for the HAP system (∼8). By assuming a readsorption of fluoride ions to the positively charged calcium sites ≡CaOH2+ surface charge was lowered. In these calculations, a total concentration of ≡F sites of 0.8 mM for a suspension density of 7.7 g/L was assumed based upon a corresponding experimental limiting value at high pH (part a of Figure 4). The refinement of the different constants gave the following results: log β(≡F + H2O S ≡ OH + H++F-)) -8.78 ( 0.13 log β(≡CaOH + F-+H+ S ≡CaF + H2O) ) 11.40 ( 0.33
2log β(9 Ca2++6 HPO4+2 F- S
Ca9(HPO4)2(PO4)4F2(s)+4 H+)) 48.08 ( 0.45
And the capacitance: C ) 3.4 ( 0.3 F/m2 The resulting fit to experimental data is given in part a of Figure 4, and in Figures 5 and 6. As can be seen the fit is good with respect to calculated Ca, P, H with some deviation for F in the most acidic range where the formation of CaF2(s) takes place. This fit could be improved by slightly increasing its solubility. Furthermore, Ψ ) 0 was obtained with pH ) 5.3, which is in acceptable agreement with pHiep ) 5.7 considering the low buffer capacity with respect to surface charge within the interval 4 e pH e 9 (cf. Figure 7). It is also obvious from Figure 7 that the surface shows zwitterionic properties, that is the coexistence of negatively charged phosphate sites and positively
Bengtsson et al.
charged Ca sites. This diagram also shows upon an extensive ion exchange process related to eq 9, with a mixture of ≡F and ≡OH sites in the pH interval 4-7. This means that this process is the main pH buffer around pH ≈ 5, and not Ca and/or P sites as suggested in earlier studies13,21,22,24 This process is also the explanation to why the dissolution of a basic mineral like FAP is generating slightly acidic suspensions.
Summary A synthetic fluorapatite was prepared that undergoes a phase transformation generated during a dialysis step. A surface layer with the composition Ca9(HPO4)2(PO4)4F2 is formed, which is suggested to form as one calcium atom is replaced by two protons. Because of the low surface area, the presence of protonated phosphate groups could not be verified from ATR-FTIR measurements. However, recent NMR studies by Sandstro¨m et al.20 using a FAP synthesized in a similar manner provided evidence for the presence of surface protonated phosphate sites. A surface complexation model, based upon XPS measurements, potentiometric titration data, batch experiments, and zeta-potential measurements was presented. To keep the number of adjustable parameters at a minimum, the acid/base properties of the ≡CaOH and ≡OPO3H2 sites were assumed to be the same as in a corresponding nonstoichiometric HAP (Ca8.4(HPO4)1.6(PO4)4.4(OH)0.4) system. Besides a determination of the solubility product of Ca9(HPO4)2(PO4)4F2, two additional surface complexation reactions were introduced. One that accounts for a F/OH ion exchange reaction resulting in the release of quite high fluoride concentrations (∼1 mM) that turned out to be dependent on the surface area of the particles. Furthermore, to explain the lowering of pHiep from being around 8 in HAP suspensions to about 5.3 in FAP suspensions a reaction that lowers the surface charge due to the readsorption of fluoride ions to the positively charged Ca-sites reaction was introduced: ≡CaOH2+ + F- H ≡CaF + H2O. The resulting model also agrees with predictions based upon XPS and ATR-FTIR observations claiming the formation of CaF2(s) in the most acidic pH range. Acknowledgment. This work was supported by the Swedish Research Council and Georange. The Kempe Foundation, Sweden, is acknowledged for funding of the FTIR spectrometer, and the Wallenberg Foundation, Sweden, for funding the XPS. Jean-Franc¸ois Boily is gratefully acknowledged for linguistic corrections and valuable comments on the manuscript. LA803137U
