ARTICLE pubs.acs.org/est
Phosphate Bonding on Noncrystalline Al/Fe-Hydroxide Coprecipitates Yu-Ting Liu and Dean Hesterberg* Department of Soil Science, North Carolina State University, Raleigh, North Carolina, United States, 27695
bS Supporting Information ABSTRACT: Poorly crystalline minerals have high sorption capacities for environmentally important chemical species, but molecular-level mechanisms of sorption on complex mineral assemblages remain largely unknown. We determined the distribution of orthophosphate (PO4) bonding between Al and Fe in relation to structural properties of Al/Fe-hydroxide coprecipitates. Phosphate was sorbed at concentrations between 0.042 and 0.162 mol P mol1 AlþFe on coprecipitates containing 0, 20, 50, 75, or 100 mol % of metal as Al. Phosphorus XANES analyses showed preferential bonding of PO4 for Al on coprecipitates with 20 and 50 mol % Al, and no preference for either metal at 75 mol % Al, consistent with X-ray photoelectron spectroscopy (XPS) analyses of near-surface metal distributions. Structural ordering and the Fe-hydroxide domain size in coprecipitates decreased with increasing Al proportion, as shown by X-ray diffraction (XRD) and Fe EXAFS analyses. Structural interactions in coprecipitates imparted unique PO4 sorption properties compared with isolated Al- or Fe-hydroxide.
’ INTRODUCTION Phosphorus is essential for life, but its management in terrestrial environments requires balancing bioavailability and mobility to avoid contamination of water bodies.1,2 Sorption capacities of soils for PO4 are frequently gauged by acid-oxalateextractable Al and Fe, indicating that structurally disordered Aland Fe-(hydr)oxides are dominant sorbents.3 Such solids are common in soils and waterways,4 and their PO4 sorption capacities are generally greater than those of phyllosilicates and crystalline oxide minerals.5,6 In soils and other environmental systems, poorly ordered Al- and Fe-(hydr)oxides are likely associated into Al-substituted Fe-(hydr)oxides 7,8 or complex assemblages of Al- and Fe-hydroxides containing organic matter.9 Interactions between dissimilar solid phases and the distribution of sorbed PO4 between Al- and Fe-(hydr)oxides in complex particles 10,11 could be particularly important for PO4 mobilization under reducing conditions, where Fe(III)-(hydr)oxides undergo reductive dissolution.12 Whereas the distribution of PO4 between Al and Fe in physical mixtures depended on the crystallinities of the (hydr)oxide minerals involved,10,13 knowledge about molecular-scale sorption properties of Al/Fe-hydroxide coprecipitates is limited. Harvey and Rhue11 reported that PO4 sorption on poorly crystalline Al/Fe-hydroxide coprecipitates generally increased with increasing Al/Fe ratio as a result of increasing anion exchange capacities caused by changes in morphology and structure. In contrast, Masue et al.14 found that sorption of arsenate decreased with increasing Al/Fe molar ratio because of noncrystalline Al-hydroxide transformation to bayerite and gibbsite. Thus, oxyanion sorption on Al/Fe-hydroxide coprecipitates is r 2011 American Chemical Society
affected by solid-phase interactions and structural changes that affect the density and accessibility of binding sites. The objective of this research was to determine the relative distribution of sorbed PO4 bonding between Al and Fe in relation to surface and structural characteristics of Al/Fe-hydroxide coprecipitates. We combined sorption isotherm experiments with spectroscopic analyses that are sensitive to PO4 bonding with Fe vs Al (P K-edge XANES) 10,13,15 and local coordination of Fe-hydroxides (Fe K-edge EXAFS). Bulk and surface properties of the coprecipitates were also characterized using XRD, XPS, and other techniques.
’ MATERIALS AND METHODS Al/Fe-Hydroxide Coprecipitates (AFH). Samples of AFH were synthesized based on a ferrihydrite synthesis procedure.16 Analytical grade Fe(NO3)3 3 9H2O and Al(NO3)3 3 9H2O were dissolved into 0.5 L of CO2-free water13 to achieve a total metal (AlþFe) concentration of 0.2 mol L1, with Al/(AlþFe) molar ratios of 0, 0.1, 0.2, 0.5, 0.75, or 1.0 (0-, 10-, 20-, 50-, 75-, and 100AFH). Coprecipitation was induced by adjusting the solutions to pH 7.5 with 1.0 M KOH and stirring for 0.5 h.16 Precipitates were washed three times each with 1 M KCl and 0.01 M KCl solution by shaking for 30 min and centrifuging.13 Final stock suspensions of 0.2 L in 0.01 M KCl solution were adjusted to pH 6.0, characterized for solids concentration,13 and stored at 4 °C for a Received: May 10, 2011 Accepted: May 26, 2011 Published: June 30, 2011 6283
dx.doi.org/10.1021/es201597j | Environ. Sci. Technol. 2011, 45, 6283–6289
Environmental Science & Technology maximum of 2 weeks. A subsample of each suspension was washed with deionized water to remove excess salts, and freezedried prior to analyses of total Al and Fe concentration, XRD, XPS, transmission electron microscopy (TEM), Fourier-transform infrared (FTIR) spectroscopy, and specific surface [see “Chemical and Mineralogical Analyses of Al/Fe-Hydroxide Coprecipitates” in the Supporting Information (SI) for details]. Iron K-Edge X-ray Absorption Spectroscopy (XAS). The local coordination of Fe in AFH samples was characterized using Fe K-edge extended X-ray absorption fine structure (EXAFS) spectroscopy. A moist paste of each AFH sample was collected on a 0.2 μm Millipore polycarbonate filter membrane at a thickness calculated to yield unit edge step across the Fe K-edge near 7112 eV.17 The samples were sealed with 8 μm Kapton film and Kapton tape to avoid desiccation, and stored at 4 °C. Spectra were collected at Beamline X11B at the National Synchrotron Light Source, Brookhaven National Laboratory in Upton, NY. Details of XAS data collection and analyses are given in the SI under “XAS Data Collection and Analysis”. Phosphate Sorption Isotherms. Phosphate sorption isotherms for AFH samples were determined at pH 6.0 and 25 °C in a 0.01 M KCl background. Each sample contained 30.00 ( 0.05 g of suspension with a solids concentration of 1.50 g kg1. Various amounts of 0.01 M KH2PO4 solution were added to pH6 suspensions while vigorously stirring. See ref 13 for experimental details. Phosphorus K-Edge XANES Data Collection and Analysis. Samples for P XANES analysis were prepared at concentrations between 0.042 and 0.162 mol P mol1 AlþFe using the identical procedure of isotherms, but with a sample mass of 200.00 ( 0.05 g. A portion (∼0.2 g) of the moist solids was collected after centrifugation, mounted in acrylic sample holders, covered with 5 μm polypropylene X-ray film (Spex Industries, Edison, NJ) to inhibit desiccation, and stored at 4 °C or on ice until analyzed.13 Spectra were collected at Beamline BL16A1 at the National Synchrotron Radiation Research Center in Hsinchu, Taiwan. Details for data collection and processing are given in the SI under “XAS Data Collection and Analysis”.
ARTICLE
Figure 1. X-ray diffraction patterns for Al/Fe-hydroxide coprecipitates with 0, 10, 20, 50, 75, and 100 mol % of total metal as Al (0-, 10-, 20-, 50-, 75-, 100-AFH). Data were collected using unresolved Cu KR radiation.
’ RESULTS Characteristics of Al/Fe-Hydroxide Coprecipitates. Chemical analyses showed that g95% of Al and Fe were incorporated into the solids during hydrolysis. X-ray diffraction results for the 0and 10-AFH samples showed two broad peaks centered at 2.6 and 1.5 Å (Figure 1), consistent with two-line ferrihydrite.16 These peaks are also discernible for the 20-AFH sample, but samples with Al g 50 mol % show no evidence for ferrihydrite or crystalline solids (Figure 1). Moreover, no evidence for crystallization of Al-hydroxides was shown by FTIR spectra in any samples (Figure S1 in SI).18 A disproportionate loss of diagnostic ferrihydrite peaks between 10 and 20 mol % Al indicates that Al coprecipitation at g20 mol % disrupted structural ordering of ferrihydrite. However, sample dilution with noncrystalline Al-hydroxide at Al proportions of g50% could mask diffraction peaks from ferrihydrite, if present. Transmission electron microscope images (Figure 2) show changes in aggregate structure of AFH samples with increasing proportion of Al, particularly between the 50- and 75-AFH samples. Morphologies of the 20- and 50-AFH samples were similar to ferrihydrite (0-AFH), with no distinct microaggregates visible within the larger, more compact aggregates. The 75-AFH
Figure 2. TEM images of Al/Fe-hydroxide coprecipitates with 0, 20, 50, 75, and 100 mol % of total metal as Al (0-, 20-, 50-, 75-, 100-AFH).
sample was similar to that of noncrystalline Al-hydroxide (100-AFH), with loosely associated, pseudohexagonal microaggregates visible throughout aggregates. However, we found no clear trend in BET surface areas of the same freeze-dried samples: 187 ( 11, 181 ( 8, 227 ( 5, 184 ( 8, and 146 ( 8 m2 g1 for the 0-, 20, 50-, 75-, and 100-AFH samples, respectively. Anderson and Benjamin19 measured BET surface areas of ∼200, 297, and 41 m2 g1 for Al/Fe-hydroxide coprecipitates containing 0, 50, and 100 mol % Al. However, Wang et al.20 found that the BET surface area of dehydrated, poorly crystalline 6284
dx.doi.org/10.1021/es201597j |Environ. Sci. Technol. 2011, 45, 6283–6289
Environmental Science & Technology
ARTICLE
Table 1. Structural Parameters Obtained from Fe K-Edge EXAFS Fitting Analysis for Al/Fe-Hydroxide Coprecipitates with 0, 20, 50, and 75 mol % of Total Metal as Al (0-, 20-, 50-, 75-AFH)a,b path
R (Å)
CN
0
FeFe1
3.05 (0.006)
2.8 (0.4)
(0-AFH)
FeFe2
3.43 (0.006)
1.4 (0.6)
20
FeFe1
3.05 (0.006)
2.4 (0.4)
(20-AFH)
FeFe2
3.43 (0.006)
1.7 (0.5)
50
FeFe1
3.05 (0.006)
2.0 (0.3)
(50-AFH)
FeFe2
3.43 (0.006)
1.4 (0.5)
75 (75-AFH)
FeFe1 FeFe2
3.05 (0.006) 3.43 (0.006)
1.9 (0.4) 1.2 (0.6)
Al mol %
Figure 3. Iron K-edge EXAFS spectra for Al/Fe-hydroxide coprecipitates with 0, 20, 50, and 75 mol % of total metal as Al (0-, 20-, 50-, 75-AFH). (a) EXAFS k3-weighted χ(k) data; (b) magnitude of Fourier transformed k3 χ(k) data.
Al-oxyhydroxide underestimated the surface area accessible to water in hydrated samples. The proportions of Al relative to Fe [Al/(AlþFe)] detected by XPS analysis were 0, 45, 72, 83, and 100 mol % for the 0-, 20-, 50-, 75-, and 100-AFH samples. These results indicate a 2.3-, 1.4-, and 1.1-fold enrichment of near-surface Al relative to total Al in the 20-, 50-, and 75-AFH coprecipitates, respectively. Enrichment of Al on particle surfaces was previously reported for coprecipitates having Al/Fe = 1.19,21 The increasing consistency between nearsurface and total Al and Fe with increasing Al proportion could be a result of increasing particle disaggregation (Figure 2), increasing Al-hydroxide domain size, or decreasing Fe-hydroxide domain size. Iron K-Edge EXAFS. The Fe EXAFS spectra for samples with Al coprecipitates are similar to that of two-line ferrihydrite (0-AFH) [Figure 3a, and refs 22,23]. However, the amplitudes of oscillations at k ≈ 5.0, 7.5, and 8.5 Å1 decreased with increasing proportion of Al (Figure 3a arrows), corresponding to a systematic decrease in high-shell backscattering signals between 2.2 and 3.5 Å in the Fourier transformed (FT) spectra (Figure 3b, and ref 22). The data were fit (R-factor = 0.018) with an EXAFS model derived from a ferrihydrite structural model24 that accounted for the highest-amplitude, single-scattering FeO and FeFe paths out to 3.43 Å (Table 1; Figure S2 in SI). The EXAFS signals of the first-shell FeO path and two higher-shell FeFe paths in the model were stronger than those of the three higher-shell FeO paths (Figure S3 in SI). Modeling of the first-shell FeO coordination with parameters fit to a common value across all AFH samples showed an average of 5.7 ((0.3) oxygen atoms at 1.98 ((0.004) Å, with σ2 = 0.011 ((0.001) Å2 (Table 1). Average FeFe distances of 3.05 and 3.43 Å corresponded to edge- and corner-shared FeO6 octahedra, respectively.23,25 With increasing coprecipitated Al, the modeled coordination number (CN) for Fe at 3.05 Å showed a decreasing trend from 2.8 ((0.4) for 0-AFH to 1.9 ((0.4) for 75 AFH, whereas the CN for Fe at 3.43 Å varied irregularly between 1.2 ((0.6) and 1.7 ((0.6) (Table 1). The decreasing trend in CN for Fe at 3.05 Å accounts for the systematic decrease in the FT magnitude between 2.2 and 3.5 Å (Figure 3b). Although increasing structural disorder and isomorphic substitution of Al for Fe would also diminish the high-shell backscattering
a Fitting was done across the k range of 2.5 to 11.5 Å1 and an R range of 1.0 to 3.5 Å. Numbers in parentheses are uncertainties calculated for the EXAFS model. b All samples were fit simultaneously, yielding a normalized sum of squared residuals [R-factor = ∑(data-fit)2/∑data2) of 0.018 (1.8%)].Values of other EXAFS model parameters17 not shown above were either fixed or fit to a common value across all samples as follows:
• S02 = 0.83(0.18) (fixed amplitude reduction factor based on first-shell fitting of hematite). • ΔE = 2.31 (0.38) eV (fitted energy shift parameter). • R (fitted interatomic distances) were 1.98 (0.004) Å for FeO1, 3.41 (0.006) Å for FeO2, 3.62 (0.006) Å for FeO3, and 3.80 (0.006) Å for FeO4. • CN (fitted coordination number) was 5.7 (0.3) for FeO1; CN values for FeO2, FeO3, and FeO4 were constrained to those of edge-sharing (FeFe1) and cornersharing (FeFe2) octahedral: CNO2 = 1/4 CNFe2; CNO3 = 1/4 CNFe2 þ 1/2 CNFe1; CNO4 = 1/2 CNFe2 þ CNFe1 (see “XAS Data Collection and Analysis” in the SI for details). • σ2 (fitted mean-square displacements of interatomic distances) were 0.011 (0.001) Å2 for FeO1, 0.011 (0.002) Å2 for FeFe1 and FeFe2 paths, and 0.017 (0.003) Å2 for FeO2, FeO3, and FeO4 paths.
signal, preliminary fitting analyses indicated that σ2 values for the FeFe paths were the same within uncertainties across all samples (data not shown), and substituting Al into the model for either or both FeFe paths resulted in a poorer EXAFS fits (see SI, “EXAFS Model Evaluation of Al Substitution for Fe in Hydroxides”). Presumably, rapid synthesis of AFH at ambient temperature did not favor isomorphic substitution of Al for Fe, as has been shown for crystalline Fe-hydroxides synthesized under hydrothermal conditions.16 Although CNs modeled for higher-shell FeFe paths overlapped within uncertainties between individual AFH samples (Table 1), the systematic decrease in FT magnitude and modeled CN for the FeFe1 path at 3.05 Å with increasing Al suggested that Al disrupted edge-shared octahedral linkages.22 Based on a structural model proposed elsewhere,26,27 a decrease in octahedral linkages translates to a decrease in number of linked FeO6 polyhedra, that is, domain size, as illustrated in the SI (Figure S6 under “Conceptualized Fe Domain Size for Al/Fe-Hydroxide Coprecipitates”). Because two-line ferrihydrite consists of domains in the nanometer-size range (∼2 nm),24 a significant proportion of Fe occurs in terminal FeO6 octahedra at domain surfaces. With increasing Al coprecipitation, a decrease in highershell Fe coordination (Table 1) can be interpreted as an increase 6285
dx.doi.org/10.1021/es201597j |Environ. Sci. Technol. 2011, 45, 6283–6289
Environmental Science & Technology
ARTICLE
Figure 4. Phosphate sorption isotherms at pH 6.0 ( 0.05 and corresponding Freundlich models (solid lines) for Al/Fe-hydroxide coprecipitates with 0, 20, 50, 75, and 100 mol % of total metal as Al (0-, 20-, 50-, 75-, 100-AFH). The Freundlich models (qi = Aciβ, where qi is the fitted sorbed PO4 concentration for a given Al proportion (i) and dissolved concentration ci) had A and β parameters of 0.197 and 0.123 for 0-AFH; 0.203 and 0.117 for 20-AFH; 0.196 and 0.118 for 50-AFH; 0.255 and 0.096 for 75-AFH; 0.328 and 0.096 for 100-AFH. Dashed lines represent predicted PO4 sorption isotherms derived from weighted combinations of Freundlich models for end members of ferrihydrite (0-AFH, q0) and Al-hydroxide (100-AFH, q100) based on the relative proportions of Fe (1x) and Al (x) and in a given system, that is, qx,predicted = (1x) q0 þ x q100.13.
in the proportion of terminal octahedra.28,29 This result is consistent with a loss of short-range structural order shown by our XRD data, specifically for the 20-AFH sample (Figure 1). Based on these and XPS results showing an enrichment of nearsurface Al relative to Fe, we propose that our 20- and 50-AFH samples were largely composite domains consisting of core FeO6 polyhedra with AlO6 polyhedra integrated at their surfaces. Such a coreshell structure is plausible if Fe-hydroxide domains form first during hydrolysis and serve as a template for heterogeneous nucleation of Al-hydroxide domains.30 Saturation indices determined from aqueous speciation modeling of our mixed Al/Fe solutions between pH 4 and 8 using Visual Minteq 31,32 were five orders-of-magnitude greater for ferrihydrite than for noncrystalline Al-hydroxide (data no shown), consistent with solubility product constants [Fe(OH)3 = 1039; Al(OH)3 = 1034, ref 33]. If kinetics are not a limitation, ferrihydrite should precipitate first.34 However, in the 75-AFH sample, the micrometer-scale morphology of the coprecipitate is similar to that of noncrystalline Al-hydroxide (Figure 2), and XPS showed
