Environ. Sci. Technol. 2005, 39, 4042-4048
Sorption Mechanisms of Zinc on Hydroxyapatite: Systematic Uptake Studies and EXAFS Spectroscopy Analysis YOUNG J. LEE,* EVERT J. ELZINGA, AND RICHARD J. REEDER Department of Geosciences and Center for Environmental Molecular Science, State University of New York at Stony Brook, Stony Brook, New York 11794-2100
The systematics and mechanisms of Zn uptake by hydroxyapatite (HAP) in preequilibrated suspensions open to PCO2 were characterized using a combination of batch sorption experiments, X-ray diffraction (XRD), and extended X-ray absorption fine structure spectroscopy (EXAFS) over a wide range of pH and Zn concentrations. Sorption isotherms of Zn(II) on HAP at pH 5.0 and 7.3 show an initial steep slope at low Zn(II) concentrations, followed by a plateau up to [Zn] < ∼750 µM, suggesting Langmuirtype behavior. At [Zn] > 750 µM, a sharp rise in the pH 5.0 isotherm suggests precipitation, whereas slight continued uptake in the pH 7.3 isotherm is suggestive of an additional uptake mechanism. The sorption isotherm at pH 9.0 shows a steep uptake step at [Zn] e 0.8 µM, followed by an increasing linear trend up to [Zn] ) 5 µM, without any indication of a maximum, suggesting that precipitation is an important uptake process at this pH. Zn K edge EXAFS results show a first oxygen shell at 1.96-1.98 ( 0.02 Å in sorption samples with [Zn]tot e 250 µM at pH 5.0, 7.3, and 9.0, consistent with tetrahedral coordination. EXAFS results reveal additional P and Ca neighbors that support formation of an inner-sphere Zn surface complex where the Zn is coordinated to surface PO4 tetrahedra in a cornersharing bidentate fashion, bridging a Ca atom. In contrast, EXAFS and XRD data indicate that precipitation of Zn3(PO4)24H2O (hopeite) dominates the mode of Zn uptake at [Zn]tot g 3 mM at pH 5.0. Principal component analysis and linear combination fits of EXAFS data reveal a mixture of innersphere Zn surface complexation and precipitation of Zn5(OH)6(CO3)2 (hydrozincite) in sorption samples for [Zn]tot ) 5 mM at pH 7.3 and for [Zn]tot ) 1 mM at pH 9.0.
Introduction Hydroxyapatite (HAP) has been recognized as an effective sorbent for a variety of metals and radionuclides (1-7). HAP and materials containing HAP as a principal component (e.g., bone meal, bone char) have been widely used as soil amendments and as direct additives to water to induce immobilization of contaminants (2, 4, 5, 6). More recently, HAP has been tested as fill material in permeable reactive barriers (PRBs) for selective removal of metals and radio* Corresponding author current address: Department of Geology & Geophysics, University of Wisconsin-Madison, 1215 W. Dayton St., Madison, WI 53706; phone: (608)265-5796; fax:(608)262-0693; e-mail:
[email protected]. 4042
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 11, 2005
nuclides (5-7). Previous studies have shown that the mechanism by which HAP reduces dissolved metal concentrations depends on the metal and the experimental conditions employed. For example, several studies have demonstrated that Pb(II) is effectively immobilized by reaction with HAP via formation of an insoluble lead phosphate, where the required phosphate is derived from dissolution of the HAP (2-4). For reaction of Cd with HAP, however, several studies have suggested that adsorption, ion exchange at surface sites, and coprecipitation may be the dominant mechanisms of immobilization (2, 3). The stability of the final state of the bound metal will determine its potential for remobilization, and hence knowledge of the mechanism of immobilization is essential for predicting the long-term fate of contaminant species. Zinc is ubiquitous in most near-surface environments. Zinc plays an important role in essential biological functions but can be toxic at high concentrations (8, 9). Partitioning of aqueous zinc to solids such as metal-(hydr)oxide, carbonate, and phosphate mineral phases has been shown to influence the fate of zinc in natural systems (10-14). Studies addressing Zn(II) interaction with HAP have indicated different mechanisms of immobilization, depending on the experimental conditions. Xu et al. (3) examined the uptake of Zn(II) by synthetic HAP in batch experiments for a range of conditions. When HAP suspensions were allowed to preequilibrate (i.e., achieve saturation) prior to addition of Zn, the mechanism of uptake was interpreted to be via adsorption at surface sites over the pH range 5.6-7.2. In batch experiments without prior equilibration of suspensions with HAP, solution pH and [Ca] were observed to evolve during reaction with Zn, suggesting ion exchange or coprecipitation are possible mechanisms of uptake in addition to adsorption. Chen et al. (4) studied Zn(II) interaction with a natural carbonate fluorapatite in batch uptake experiments over a wide range of initial pH conditions, noting that the removal of Zn(II) was strongly dependent on initial and final pH values. Suspensions were not preequilibrated before reaction with Zn, and significant changes in solution pH occurred during reaction. At low pH, hopeite [Zn3(PO4)24H2O] was observed to precipitate, whereas at high pH zincite (ZnO) formed. Formation of these zinc solids in combination with observed changes in pH is consistent with partial dissolution of the apatite phase during reaction with Zn. Lusvardi et al. (15) also examined Zn(II) reacted HAP systems without preequilibration, and observed formation of hopeite initially followed by formation of scholzite [CaZn2(PO4)22H2O]. Precipitation of hopeite at low pH was also noted by Cheung et al. (16) in their study of Zn(II) interaction with bone char, which consists largely of poorly crystalline HAP. However, at pH ∼5, these authors attributed Zn(II) uptake to adsorption or ion exchange. It is evident that the mechanism of Zn interaction with HAP varies depending on the pH and conditions of reaction, especially whether systems are at bulk saturation with respect to HAP. Whereas the formation of precipitates can sometimes be demonstrated by XRD, no direct confirmation of other modes of Zn(II) interaction with HAP has been provided. Moreover, there is no systematic understanding of the combined influences of Zn concentration and pH on the mechanisms of Zn immobilization in the presence of HAP, or the role of dissolved CO2, an important component of near-surface solutions. The solution conditions are also not known for which Zn uptake by HAP is most favorable. In the present study, we combine direct spectroscopic evidence with systematic batch sorption experiments over a range of 10.1021/es048593r CCC: $30.25
2005 American Chemical Society Published on Web 04/29/2005
pH and [Zn] conditions. Our findings reveal changes in uptake mechanisms that we can relate to solution composition and changes in saturation with respect to zinc solids. Our results also provide a structural model for Zn adsorption complexes at the HAP surface.
Materials and Methods Samples and Preparation. The HAP used in this study is a precipitated reagent-grade tribasic calcium phosphate (from J. T. Baker) with an average particle size of 100 nm on the basis of TEM measurements. X-ray diffraction data revealed that all diffraction peaks were consistent with the HAP structure. A surface area of 56 m2 g-1 for the sample was determined with a multipoint BET isotherm using N2(g) as the adsorbate. Suspensions of 1 g L-1 HAP were equilibrated at atmosphere CO2 pressure (10-3.5 bar) for 3 weeks, with the final pH stabilizing at 7.3. Suspensions were also prepared at pH 3-9 by addition of HCl or NaOH while maintaining atmosphere CO2 pressure (10-3.5 bar) for 3 weeks. Final ionic strengths were in the range 0.006-0.05 M. After aging, dried powders were characterized by X-ray diffraction, with all observed peaks corresponding to those of HAP. FTIR spectra were also obtained before and after aging (Figure SI-1), with no differences observed. Weak bands at ∼1485, 1455, and 1417 cm-1 indicate the presence of minor carbonate in the HAP. Quantification using a calcium carbonate standard indicates ∼0.3 wt % CO3 in the sample; this amount is unaffected by aging. In all sorption experiments, Zn (from a ZnCl2 stock solution) was added to HAP suspensions preequilibrated at the desired pH. If necessary, the suspension pH was readjusted following addition of Zn by using HCl or NaOH solutions. Sorption experiments were conducted as a function of pH (over the range 3-9) for 250 µM total Zn. Zn-reacted suspensions were equilibrated on a reciprocal shaker table at room temperature for 24 h. Adsorption isotherms were also obtained as described above at pH 5.0, 7.3, and 9.0 after equilibration for 24 h. The suspensions were centrifuged for 15 min at 13000 rpm and filtered through 0.22-µm polypropylene membrane filters. The filtered supernatants were analyzed for aqueous Zn using direct-coupled plasma (DCP) spectrophotometry. A zinc phosphate reference sample, Zn3(PO4)24H2O, was synthesized at room temperature and at 50 °C using the method described by Pawlig and Trettin (17). XRD patterns for both these samples were consistent with that reported by Pawlig and Trettin for R-hopeite, and we assume that our material is also the R-hopeite form. Solution Speciation. Aqueous Zn speciation and saturation indexes for various Zn solids were calculated using the program PHREEQC and the LLNL database distributed with this program (18). Data for aqueous Zn species were replaced with those from the National Institute of Standards and Technology thermochemical database (19) and are listed in Table SI-1. A model speciation calculation is shown in Figure SI-2A for a total Zn(II) concentration of 100 µM, with the solution constrained to maintain equilibrium with respect to HAP and a PCO2 ) 10-3.5 bar. In addition to variation in the dominant Zn species with pH, total Ca and P concentrations vary over the conditions of the uptake experiments (Figure SI-2B). EXAFS Spectroscopy. Zn(II)-equilibrated solids were recovered from the suspensions by vacuum filtration and were retained as wet pastes. The pastes were loaded into Lucite sample holders and sealed with Kapton tape. EXAFS spectra were collected at beamlines X11A, X18B, X23A2, and X23B at the National Synchrotron Light Source, Brookhaven National Laboratory. Spectra were collected at the Zn K edge using a pair of Si(111) monochromator crystals at X11A and X23B, a channel-cut Si(111) crystal at X18B, or a Si(311) crystal
FIGURE 1. Zinc pH sorption edge on hydroxyapatite (HAP) with 250 µM total Zn(II) and PCO2 ) 10-3.5 bar. pair at X23A2, with detuning of 35-40% for harmonic rejection. An energy value of 9659 eV was assigned to the first inflection point in the absorption edge of a Zn metal reference foil. Multiple EXAFS spectra were collected in fluorescence mode using a Lytle detector at room temperature or in transmission mode depending on Zn(II) concentration of the samples; individual spectra were calibrated and averaged for analysis. Spectra for Zn model compounds were collected in transmission mode. EXAFS fitting was performed using the program WinXAS (20). A linear function for the preedge region and a secondorder polynomial in the postedge region were used for normalization. The χ(k) functions were extracted using a cubic spline and were Fourier transformed with k3 weighting over the approximate k range 2.1-12.4 Å-1. Theoretical phase shifts and amplitudes were calculated using the ab initio program FEFF7 (21) with model structures in which Zn substitutes for either the Ca1 or the Ca2 site in HAP. An amplitude reduction factor S02 ) 1.0 was determined from fits of the reference compounds ZnO and ZnCO3, both of which have uniform and distinct shells. No constraints were placed on the fitting parameters of the first-neighbor O shells. Initial fits of higher shells, however, revealed strong correlations between coordination number (CN) and the DebyeWaller factors (σ2). The fitted σ2 values were ∼0.009-0.012 Å2, which is typical of the range for the second-neighbor backscatterers in adsorption samples (14). Consequently, in final fits the σ2 values for higher shells were fixed at 0.010 Å2 for low [Zn] adsorption samples; for samples interpreted to contain solid precipitates, the σ2 values for higher shells were fixed at values determined from fits of respective reference solid. This facilitated comparison of results and reduced the number of independent variables in the fitting procedure. Estimated errors are (0.02 Å for the first-shell distances and (0.03-0.05 Å for more distant shells. Errors in the DebyeWaller parameters are estimated to be (0.002 Å2. For coordination number (CN), which is correlated with the Debye-Waller factor and the S02 factor, the error is roughly estimated at (15% for the first O shell and >25% for the more distance shells.
Results and Discussion Uptake of Zinc by Hydroxyapatite. The pH dependence of Zn sorption by HAP is shown in Figure 1. Zn sorption exhibited typical adsorption edge behavior, that is, a steep increase in the sorption of Zn with increasing pH until nearly all of the Zn was removed by HAP. The sorption of Zn increased from 31% to 85% of the total from pH 4.0 to 6.0. The sorption edge, 50% Zn(II) uptake, for these experimental conditions is approximately pH 4.5 (Figure 1). This dependence of sorption on pH may reflect both (i) a corresponding variation in the surface charge of HAP and (ii) variation in VOL. 39, NO. 11, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4043
FIGURE 2. (a) Sorption isotherms of Zn(II) on hydroxyapatite at pH 5.0 and 7.3 and (b) at pH 9.0 after equilibration for 24 h. Arrows in the inset in 2A indicate the Zn concentration above which the suspensions are supersaturated with respect to hopeite at pH 5.0 and hydrozincite at pH 7.3. Saturation with respect to hopeite in the pH 7.3 solution (at 4 µM Zn) is not shown. Samples indicated by solid symbols were investigated by EXAFS. See text for further explanation. the Lewis acid behavior of the surface functional groups (3, 22). It may alternatively be expressed as decreased competition between protons and Zn(II) for functional groups at the HAP surface. Significant Zn(II) sorption occurs at pH 5, which is lower than the point of zero charge (pHpzc) for HAP, which has been reported in the range 6.9-8.5 (3, 23-25), suggesting that inner-sphere Zn complexes form at the HAP surface. The sorption isotherms of Zn(II) on HAP at pH 5.0 and 7.3 show steep slopes at low equilibrium Zn(II) concentrations, followed by an approach to a plateau for [Zn] < ∼750 µM (see Figure 2a and inset), suggesting Langmuir-type behavior. At [Zn] > 750 µM, the pH 7.3 isotherm shows slight continued uptake, which could reflect the existence of a second uptake mechanism, such as precipitation. In the pH 5 isotherm, a marked increase in Zn uptake between 750 and 1000 µM strongly suggests precipitation. Aqueous speciation calculations indicate that pH 5.0 suspensions become saturated with respect to hopeite at ∼215 µM Zn and remain undersaturated with respect to hydrozincite over the Zn concentration studied here. In contrast, pH 7.3 suspensions in equilibrium with HAP and PCO2 ) 10-3.5 bar are saturated with respect to hopeite at ∼4 µM Zn and with respect to hydrozincite at 125 µM Zn. However, unambiguous evidence for precipitation is not observed in the pH 5.0 isotherm until much higher concentrations are reached. In the pH 7.3 isotherm, no clear indication of precipitation is evident, although the slight, continued uptake at higher [Zn] could reflect the appearance of a second uptake process, possibly precipitation. The sorption isotherm at pH 9.0 (Figure 2B) shows strong uptake up to equilibrium [Zn] ≈ 0.8 µM and increases in an 4044
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 11, 2005
approximately linear trend above 0.8 µM without any indication of reaching a plateau and thus displays nonLangmuir behavior. The depletion of dissolved Zn in the suspensions to concentrations less than 5-6 µM even for addition of 1 mM Zn from the stock solution suggests precipitation (or surface precipitation) as an important process at pH 9.0. These observations are consistent with speciation calculations, showing that HAP-saturated solutions at pH 9 become saturated with respect to hydrozincite [Zn5(OH)6(CO3)2] at [Zn] ) 0.4 µM and with respect to hopeite at [Zn] ) 6 µM. Although the number of available binding sites does not necessarily correspond to monolayer coverage, the Langmuir model has been applied to estimate maximum sorption densities (26). On the basis of the pH 5.0 and 7.3 isotherm regions below sorbed [Zn]surf ≈ 220 and 450 µmol/g, respectively, the calculated maximum Zn sorption densities range from 3.9 µmol/m2 at pH 5.0 to 7.9 µmol/m2 at pH 7.3 for the HAP surface. The calculated sorption density at pH 7.3 is similar to a theoretically calculated Ca site density of 7.2 µmol/m2 for the HAP surface (5, 27). XRD. The presence of precipitates in Zn(II)-reacted HAP suspensions was investigated by XRD (Figure SI-3). No precipitates were detected in the HAP-reacted samples at low and intermediate initial Zn concentrations ([Zn]tot < 250 µM) at pH 5.0, 7.3, and 9.0. XRD detected the presence of hopeite at [Zn]tot g 3 mM at pH 5.0, which is consistent with previous work showing that formation of hopeite is favored at high Zn concentrations in HAP systems at pH ≈ 5.5 but becomes less important as pH increases to 7.1 (4). For samples with [Zn]tot ) 5 mM at pH 7.3 and with 1 mM total Zn at pH 9.0, no Zn precipitation was detected, even though speciation calculations indicated that solutions were supersaturated with respect to hydrozincite, hopeite, zinc hydroxide, and amorphous Zn(OH)2(s). No evidence of Zn precipitation was found using scanning electron microscopy or energy-dispersive spectroscopy in a study examining adsorption of Zn on calcium carbonate (28) despite oversaturation with respect to hydrozincite. Hence, if precipitates did form in our system at pH 7.3 and 9.0 under the conditions of this study, they escaped detection by XRD. These macroscopic observations, however, cannot provide information on other modes of zinc uptake by HAP. Therefore, we use EXAFS to examine the local structural environments and coordination of zinc at the HAP surface. Zn K Edge X-ray Absorption Spectroscopy. The k3weighted raw and fitted χ (chi) functions and the corresponding radial structure functions (RSF, not corrected for phase shifts) for the various Zn sorption samples and reference compounds (including aqueous Zn2+) are shown in Figure 3. The sorption samples with [Zn]tot e 250 µM have similar k3-weighted χ (chi) functions (Figure 3A). The spectra are all in phase and do not show significant differences with Zn concentration or pH, suggesting that the local coordination environment for Zn reacted with HAP at low initial [Zn] and its sorption mechanism are similar over the pH range 5-9. The RSFs for these samples are also similar, showing a distinct peak in the low R region corresponding to the nearest oxygen shell, along with peaks indicative of secondneighbor scattering at higher R (Figure 3B). Moreover, the χ functions and RSFs of these low [Zn] sorption samples are dissimilar to those for the Zn solids hopeite, zincite, and hydrozincite. This suggests that no significant formation of these solid phases occurred at lower [Zn], although we cannot rule out the presence of a minor Zn precipitate component occurring as a separate phase. At [Zn]tot ) 5 mM at pH 7.3 and 1 mM total Zn at pH 9.0, however, the χ spectra exhibit subtle but important differences, showing peaks similar to those of hydrozincite, which is the stable Zn phase in HAP preequilibrated suspensions
FIGURE 3. (a) The Zn K edge k3-weighted χ(k) curves of sorption samples and reference compounds and (b) corresponding radial structure functions (not corrected for phase shifts). The dotted lines are best fits. open to air (14, 28). In contrast, the χ spectra (especially between 4.90 and 9.10 Å-1) and the RSFs for the sorption samples with [Zn]tot g 3 mM at pH 5.0 have a close similarity to those of hopeite. These observations are consistent with our XRD data indicating the presence of hopeite and are further supported by the speciation calculations showing supersaturation with respect to hopeite. Whereas we did not observe evidence for the presence of Zn solids in sorption samples with [Zn]tot e 250 µM, their presence is clearly evident in sorption samples at the higher Zn concentrations examined, suggesting changes in the dominant uptake mechanism that correlate with Zn concentration and pH. As a test of our interpretations of the occurrence of different Zn coordination environments in the sorption samples, we performed a principal component analysis (PCA) of the k3-weighted χ functions of the sorption samples described above and shown in Figure 3. The analysis revealed three significant components, consistent with the presence of three different Zn species in this data set. We then performed target transformations to identify the different Zn species. As with the hopeite and hydrozincite reference compounds, the pH 7.3 [Zn]tot ) 250 µM sorption sample was excluded from the PCA, for use as the endmember for adsorbed Zn in the target transformation. Hopeite and adsorbed Zn were positively identified as components, and hydrozincite was also found to be important, although the target transformation for this component produced a lower fit quality than for the other Zn components. Using these endmembers for Zn speciation in the samples indicated that the low [Zn] adsorption samples (at all pH values studied) are dominated by single Zn species: adsorbed Zn. This is consistent with the strong similarities found between the low [Zn] adsorption samples at all pH values. The higher [Zn] samples at pH 5 are also dominated by single Zn species: hopeite. This is consistent with the nearly identical spectra obtained for hopeite and the high [Zn] samples reacted at pH 5.0. In no case was a mixture of components observed for any of these samples.
For the two high [Zn] samples reacted at pH 7.3 and 9.0, however, PCA indicates the importance of both adsorbed Zn and hydrozincite, with approximately equal contributions. We therefore performed linear-combination (LC) fits of the k3-weighted χ data for these samples, using the χ functions of adsorbed Zn and hydrozincite as endmembers. The LC analysis yielded good fits of the experimental data (Figure SI-4), with estimated contributions of 60% and 40% (pH 9.0) and 51% and 49% (pH 7.3) for adsorbed Zn and hydrozincite, respectively. This finding indicates that Zn in these higher concentration samples is speciated as a mixture of adsorbed Zn and a hydrozincite precipitate. The onset of precipitation of hydrozincite with increasing Zn concentration may explain the continued uptake trends observed in the pH 7.3 and 9.0 isotherms. Previous workers (10, 28) studying Zn interactions with metal oxides and calcium carbonate concluded that a change in primary sorption mechanism from surface complexation to precipitation occurs as its concentration increases. Our best-fit EXAFS results for the sorption samples and reference compounds are shown in Tables 1 and 2. The most significant difference between the sorption and the aqueous Zn2+ samples is in the first oxygen shell. The bestfit distances for the oxygen shell in the sorption samples with [Zn]tot e 250 µM over the range of pH studied are the same at 1.96-1.98 ( 0.02 Å, with σ2 ) ∼0.008 Å2. These distances agree with the Zn-O distance of 1.95-1.98 Å determined by EXAFS spectroscopy for tetrahedral coordination of Zn (10-14). In contrast, the fitted Zn-O distance for the aqueous Zn2+ sample is 2.07 Å ( 0.02 Å, which is consistent with an octahedral coordination environment (13). This suggests that the coordination of Zn changes from octahedral to tetrahedral as it adsorbs at the HAP surface. Also, the normalized intensity of the Zn K edge white line for the low [Zn] adsorption samples is suppressed relative to the aqueous Zn2+ spectrum and the spectra for compounds with Zn in octahedral coordination, which supports the tetrahedral coordination of the adsorbed Zn (29). Also significant is that VOL. 39, NO. 11, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4045
TABLE 1. Best-Fit Results from Zn K Edge EXAFS Data for Zn(II) Sorption on HAP
a
sample
[Zn]tot µM
uptake (%)
pH
shell
CN
R (Å)
σ2 (Å2)
250ZnHAP5
250
58.9
5.0
Zn-O Zn-P Zn-Ca
4.4 1.3 2.1
1.96 3.22 3.33
0.008 0.010a 0.010a
3000ZnHAP5
3000
68.5
5.0
Zn-O Zn-P Zn-Zn
5.1 1.2 1.4
1.98 3.12 3.36
0.011 0.012 0.006
5000ZnHAP5
5000
69.6
5.0
Zn-O Zn-P Zn-Zn
5.3 1.6 1.5
1.98 3.15 3.36
0.012 0.015 0.007
50ZnHAP7
50
99.6
7.3
Zn-O Zn-P Zn-Ca
4.4 1.3 1.4
1.98 3.27 3.36
0.008 0.010a 0.010a
250ZnHAP7
250
97.6
7.3
Zn-O Zn-P Zn-Ca
4.3 0.7 1.1
1.97 3.21 3.32
0.008 0.010a 0.010a
5000ZnHAP7
5000
31.4
7.3
Zn-O Zn-Zn Zn-Zn
5.1 1.3 1.1
2.00 3.12 3.58
0.012 0.012a 0.009a
100ZnHAP9
100
99.3
9.0
Zn-O Zn-P Zn-Ca
4.4 0.7 0.9
1.97 3.18 3.30
0.008 0.010a 0.010a
1000ZnHAP9
1000
99.4
9.0
Zn-O Zn-Zn Zn-Zn
5.2 1.5 0.9
1.99 3.11 3.60
0.011 0.012a 0.009a
Fixed parameter. See text for estimated errors.
TABLE 2. Zn K Edge EXAFS Fit Results and XRD Data for Model Compounds shell
CN
R (Å)
σ2 (Å2)
Zn2+
Zn-O
6.6
2.07
0.009
hydrozincitea
Zn-Oe
5.3 5.2 2.1 2.3 3.4 5.2
2.03 2.044 3.15 3.172 3.54 3.576
0.011
6.0f 6.0 6.0f 6.0 6.0f 6.0 6.0f 6.0
2.10 2.111 2.99 2.964 3.19 3.231 3.66 3.673
0.006
4.0f 4.0 12.0f 12.0
1.96 1.979 3.22 3.229
0.006
4.9 4.8 1.7 3.2 1.0 2.0
1.98 2.034 3.13 3.169 3.37 3.399
0.012
compound (aq)
Zn-Zn Zn-Zn ZnCO3b
Zn-O Zn-C Zn-O Zn-Zn
zincitec
Zn-O Zn-Zn
hopeited
Zn-Oe Zn-Pe Zn-Zn
0.012 0.009
0.004 0.004 0.006
0.010
0.013 0.005
a XRD data from Ghose (30). b XRD data from Effenberger et al. (31). XRD data from Abrahams and Bernstein (32). d XRD data from Whitaker (33). e One shell was used to fit two or more Zn-X distances. f Fixed parameter. Italics indicate radial distance or average radial distance derived from XRD crystallographic data. c
the 1.96-1.98 Å Zn-O distance is shorter than the 2.05-2.11 Å distance reported by Barrea et al. (8) for Zn coprecipitated with biological carbonated HAP and interpreted to substitute in a Ca site. 4046
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 11, 2005
Inasmuch as the EXAFS data from the low [Zn] sorption samples show no indication of a contribution from a Zn solid or from a Zn coprecipitate, our strategy for fitting the peaks in the RSF in the range ∼3.2-3.3 Å (Figure 3B) was based on evaluating different combinations of P and Ca backscatterers, as components of the HAP surface, and Zn, as a second neighbor of a potential multinuclear surface complex (or a Zn precipitate). Fits using Zn or Zn and P shells were unsatisfactory, giving physically unrealistic fit parameters, thereby ruling out multinuclear surface complexes and Zn precipitates. Best fits were clearly obtained using P and Ca shells, with Zn-P and Zn-Ca distances of 3.18-3.27 Å and 3.30-3.36 Å, respectively. These distances, combined with the Zn-O distance 1.96-1.98 ( 0.02 Å and CN value 4.34.4, indicate that, over the pH range studied and for [Zn]tot e 250 µM, zinc dominantly forms inner-sphere tetrahedral adsorption complexes at the HAP surface. Similar fitting results found for the low [Zn] samples reacted at pH 5.0, 7.3, and 9.0 indicate that the difference in aqueous Zn speciation at pH 9.0 versus pH 5.0 and pH 7.3 (Figure SI-2A) appears to have no effect on the mechanism of Zn coordination to the HAP surface. For the Zn sorption samples with [Zn]tot g 3 mM at pH 5.0, however, our best-fit distance for the first oxygen shell is 1.98 ( 0.02 Å with a CN value ∼5 ( 1. Zn-P and Zn-Zn backscattering paths are observed at 3.12-3.15 Å and 3.36 Å, respectively (Table 1). These distances are consistent with our fit results for hopeite (Table 2), suggesting formation of this phase at the HAP surface or in solution for [Zn]tot g 3 mM at pH 5.0. For the sorption samples at pH 9.0 with 1 mM total Zn and at pH 7.3 with 5 mM total Zn, the best-fit distances for the first oxygen shell are 1.99-2.00 ( 0.02 Å with a D-W factor, σ2 ) 0.011-0.012 Å2. More significant are fits for the higher shells, which revealed Zn-Zn backscattering paths at 3.11-3.12 Å and 3.58-3.60 Å (Table 1). These distances are in reasonably good agreement with our best-fit distances of Zn-Zn backscattering in hydrozincite at 3.15 and 3.54 Å
FIGURE 4. Schematic model illustrating possible coordination of a Zn adsorption complex at the HAP surface. See text for further explanation. (Table 2), suggesting formation of this phase at pH 9.0 with 1 mM total Zn and at pH 7.3 with 5 mM total Zn concentration. With a Zn-O distance of 2.11 Å and a shortest Zn-Zn distance of 3.67 Å, ZnCO3 is an unlikely precipitate in either of these samples, even though initially supersaturated. For the sorption sample from the pH 7.3 and 5 mM total Zn experiment, for which speciation calculations indicated both hopeite and hydrozincite were supersaturated, fits using P and Zn shells consistent with hopeite yielded poorer fit results. Consequently, the EXAFS results support the formation of only hydrozincite and not hopeite or ZnCO3. For final fits given in Table 1, we fixed the σ2 values for the Zn-Zn shells in these sorption samples at the values determined for the hydrozincite reference. Although the Zn-Zn backscattering suggests hydrozincite precipitation in the pH 9.0 1 mM Zn and pH 7.3 5 mM Zn sorption samples, our best-fit result for the Zn-O distance (1.99 ( 0.02 Å) is slightly shorter than the mean Zn-O distance in hydrozincite (∼2.04 Å), whereas it is slightly longer than the Zn-O distance of adsorbed tetrahedral Zn (∼1.97 Å). Hydrozincite contains two similar octahedral Zn sites and one tetrahedral Zn site, with a tetrahedral:octahedral ratio of 2:3 (30), resulting in a weighted average Zn-O distance of ∼2.04 Å (14). The intermediate Zn-O distances found for these high concentration Zn samples are consistent with the presence of a mixture of hydrozincite and adsorbed Zn surface complexes, as established in the PCA and LC analyses described earlier. Precipitation of hydrozincite was favored over hopeite in the pH 7.3 5 mM Zn sorption sample, despite greater oversaturation with respect to hopeite than hydrozincite. The reasons for this are unclear. However, Chen et al. (4) observed formation of hopeite in Zn-reacted HAP systems preferentially at low pH (
