Formation of Crystalline Zn−Al Layered Double HydroxidePrecipitates on γ‑Alumina: The Role of Mineral DissolutionWei Li,*,† Kenneth J. T. Livi,‡ Wenqian Xu,§ Matthew G. Siebecker,† Yujun Wang,†,∥ Brian L. Phillips,⊥

and Donald L. Sparks†

†Environmental Soil Chemistry Group, Delaware Environmental Institute and Department of Plant and Soil Sciences, University ofDelaware, Newark, Delaware 19717, United States‡The High-Resolution Analytical Electron Microbeam Facility, Department of Earth and Planetary Sciences, The Johns HopkinsUniversity, Baltimore, Maryland 21218, United States§Department of Chemistry, Brookhaven National Lab, Upton, New York 11973, United States∥Key Laboratory of Soil Environment and Pollution Remediation, Institute of Soil Science, Chinese Academy of Sciences, Nanjing210008, People's Republic of China⊥Department of Geosciences and Center for Environmental Molecular Science, Stony Brook University, Stony Brook, New York,11794, United States

*S Supporting Information

ABSTRACT: To better understand the sequestration of toxic metals such as nickel(Ni), zinc (Zn), and cobalt (Co) as layered double hydroxide (LDH) phases in soils,we systematically examined the presence of Al and the role of mineral dissolutionduring Zn sorption/precipitation on γ-Al2O3 (γ-alumina) at pH 7.5 using extendedX-ray absorption fine structure spectroscopy (EXAFS), high-resolution transmissionelectron microscopy (HR-TEM), synchrotron-radiation powder X-ray diffraction(SR-XRD), and 27Al solid-state NMR. The EXAFS analysis indicates the formationof Zn−Al LDH precipitates at Zn concentration ≥0.4 mM, and both HR-TEM andSR-XRD reveal that these precipitates are crystalline. These precipitates yield a smallshoulder at δAl‑27 = +12.5 ppm in the 27Al solid-state NMR spectra, consistent withthe mixed octahedral Al/Zn chemical environment in typical Zn−Al LDHs. TheNMR analysis provides direct evidence for the existence of Al in the precipitates andthe migration from the dissolution of γ-alumina substrate. To further address thisissue, we compared the Zn sorption mechanism on a series of Al (hydr)oxides with similar chemical composition but differingdissolubility using EXAFS and TEM. These results suggest that, under the same experimental conditions, Zn−Al LDHprecipitates formed on γ-alumina and corundum but not on less soluble minerals such as bayerite, boehmite, and gibbsite, whichpoint outs that substrate mineral surface dissolution plays an important role in the formation of Zn−Al LDH precipitates.

■ INTRODUCTION

Mobility and bioavailability of heavy metals in soil and aquaticenvironments are heavily influenced by metal−mineralinteractions. It is well documented that transition metals suchas nickel (Ni), zinc (Zn), and cobalt (Co) can form mixedmetal−Al hydroxide surface precipitates on aluminum oxidesand Al-rich soil clays.1−7 Upon aging, these precipitates becomeless soluble, which is an important mechanism for toxic metalsequestration in natural environments. These precipitates areidentified as hydrotalcite-like surface phases, because the short-range structural order of these precipitates is similar tohydrotalcite minerals (e.g., takovite). The hydrotalcite mineralsare also named layered double hydroxides (LDHs), which havea chemical formula commonly represented as [Mez+1−xAl

3+x

(OH)2]q+(anionn−)q/n·yH2O, where Me2+ may be Co2+, Ni2+, or

Zn2+ and the anion can be nitrate, carbonate, or silicate.8−10

Even though it is clear that Me−Al LDHs can form in certainenvironments, there is a missing link between the mineralogy of

LDHs and the formation mechanism of LDH surfaceprecipitates. Such knowledge is important to improve surfacesorption/precipitation models and to better predict theenvironmental fate of toxic transition metals.To understand the LDH formation mechanism, a crucial step

is the characterization of the surface precipitates. Extended X-ray absorption fine structure spectroscopy (EXAFS) has beenextensively used to identify LDH phase in either labor-tory1−7,11−16 or natural environments.17−23 However, theinformation obtained from EXAFS alone is not adequate tofully characterize the LDH surface precipitates. Technically,EXAFS only provides average structural information over ashort-range order (usually less than 5 Å); therefore, EXAFS

Received: May 6, 2012Revised: September 24, 2012Accepted: October 8, 2012

Article

pubs.acs.org/est

© XXXX American Chemical Society A dx.doi.org/10.1021/es3018094 | Environ. Sci. Technol. XXXX, XXX, XXX−XXX

fails to determine if the LDH-like precipitates are crystalline oramorphous, which is important in understanding thestabilization of LDH precipitates. Crystalline LDHs haveperiodic interlayers between the mixed metal−Al hydroxidelayers, with interlayer distances being ∼7.5−8 Å depending onthe interlayer anions. The layered structure allows LDHprecipitates to be further reactive toward anions. Anotherlimitation of EXAFS is the challenge of analyzing light elements(e.g., Al), because light elements usually have weak back-scattering properties and thus do not contribute significantly tothe amplitudes in the background subtracted chi(k) EXAFSdata. This weakness makes it difficult to distinguish betweenLDH species and metal hydroxides with distorted structureswhile using EXAFS. This issue has been addressed byScheidegger and co-workers,4,5 who showed that adding anadditional Ni−Al scattering path (besides the Ni−Ni path) inthe second atomic shell would lead to better fitting. However,this only provided indirect evidence for the presence of Al; untilrecently, direct measurement of Al in the precipitates is stilllacking. Additionally, EXAFS is mainly used for structuralanalysis and can hardly provide information about the chemicalcomposition, crystallinity, and morphology, all of which is ofequal importance in understanding the LDH formation.For these reasons, we employed several state-of-the-art

techniques including synchrotron-radiation X-ray diffraction(SR-XRD), high-resolution transmission electron microscopy(HRTEM), and 27Al solid-state nuclear magnetic resonance(NMR) and traditional EXAFS to investigate zinc sorption/precipitation mechanisms on γ-alumina (γ-Al2O3). Theobjective of this study is to thoroughly characterize the Zn−Al LDH surface precipitates and elucidate the role of Al in itsformation. One important finding is the identification ofcrystalline LDH precipitates on γ-alumina at a low Znconcentration (∼26 ppm) using SR-XRD and HRTEM. Inaddition, the role of surface Al dissolution from the substrate asan important mechanism in the formation of LDH precipitatesis discussed.

■ EXPERIMENTAL SECTIONSorption Experiments. Aluminum oxide (aluminum oxide

C, Degussa), identified by powder X-ray diffraction as the γ-phase Al2O3 (γ-alumina), was used as the adsorbent. The γ-alumina powder exhibited a Brunauer−Emmett−Teller (BET)specific surface area of 136 m2 g−1 and a pHPZC of 9.1.24

Sorption of Zn on γ-alumina was conducted at ambientconditions using a batch technique. A 0.10 g aliquot of dry γ-alumina powder was added to 40 mL of 0.01 M NaNO3background electrolyte at pH 7.5. The pH was maintained byaddition of a HEPES buffer (pKa = 7.5 at 298 K) at a finalconcentration of 10 mM. Previous studies showed that HEPESdoes not interfere with transition metal sorption to mineralsurfaces.6,25 Small amounts of 50 mM Zn(NO3)2 solution wereadded into each tube to reach the desired initial Znconcentration. A speciation diagram of aqueous Zn is providedin the Supporting Information (Figure S1), suggesting at theexperimental pH that the dominant dissolved Zn species is Zn2+

(aq). A reaction time of 24 h was chosen based on the sorptionkinetics (Supporting Information, Figure S2). After thereaction, the samples were centrifuged to separate the solidand solution. The supernatant was filtered with a 0.2 μm filterand then analyzed for Zn by inductively coupled plasma−atomic emission spectroscopy (ICP-AES), while the solidsamples were then freeze-dried for further HRTEM, SR-XRD,

and solid-state NMR characterization. Replicate samples of thewet paste at initial Zn concentrations of 0.2, 0.4, 0.8, and 3 mMwere prepared for EXAFS analysis.

Synchrotron Radiation Powder X-ray Diffraction. SR-XRD patterns were recorded in transmission geometry with aPerkin−Elmer amorphous silicon detector at an incident X-rayenergy of 38794 eV (λ = 0.3196 Å) at beamline X7B at theNational Synchrotron Light Source (NSLS), BrookhavenNational Lab (Upton, NY). Dry powders were placed inkapton capillary tubes. Two-dimensional XRD patterns werecalibrated with lanthanum hexaboride (LaB6, NIST 660a) andintegrated to XRD profiles of intensity versus 2θ with Fit2d.26

X-ray Absorption Fine Structure Spectroscopy. EXAFSdata were collected on the sorption samples at beamline X11Aat the NSLS. All samples were mounted in a thin plastic sampleholder covered with Kapton tape and placed at 45° to theincident beam. Data were collected in both fluorescence andtransmission modes using a Lytle detector positioned 90° tothe beam. Zn foil was used for energy calibration. A pair ofSi(111) crystals for the monochromator were employed, whichwere detuned by 30% to suppress high order harmonic X-rays.Data processing was performed with the EXAFS data analysisprogram SIXPACK.27 Spectra were averaged after carefulenergy calibration using Zn foil (E0 = 9659 eV). The χ(k)function was Fourier transformed using k3 weighting, and allshell-by-shell fitting was done in R-space. A single thresholdenergy value (ΔE0) was allowed to vary during fitting. Theamplitude reduction factor, S0

2, was estimated based on thefitting of the spectra of a Zn solution and the Zn−Al LDHmodel compound and then applied to sorption samples.

High-Resolution Transmission Electron Microscopy.Small quantities of powdered reactants were dispersed indeionized water and ultrasonicated for three minutes. A 200mesh Cu grid with a lacey-carbon support film was dipped intothe suspension and dried. TEM analyses were made using aPhilips CM300 FEG microscope equipped with an Oxford lightelement energy dispersive X-ray spectroscopy (EDS) detectorand a Gatan GIF 200 CCD imaging system. The point-to-pointresolution of the TEM is better than 0.2 nm, and the lineresolution is 0.09 nm. Images were analyzed and processedusing the Gatan Digital Micrograph 1.8 software. The softwarepackage ES Vision4 was used to acquire and process the EDSspectra.

Solid-State NMR. Solid-state 27Al single-pulse MAS (SP/MAS) NMR spectra of Zn sorption samples were collected on a400 MHz Varian Inova spectrometer (9.4 T), with Larmorfrequencies of 130.2 MHz. Spectra were collected using aVarian/Chemagnetics T3-type probe with samples contained in3.2 mm ZrO2 rotors and a spinning rate of 20 kHz. A 6 μs 90°pulse was calibrated using the 1 M Al(NO3)3 solution standard,but only 1 μs rf (i.e., radio frequency) pulse length was chosenfor measuring the solid samples. The pulse delay was optimizedas 5 s, and approximately 400 scans were collected for eachspectrum to obtain an acceptable signal-to-noise ratio. The 27Alchemical shifts (δAl) are reported relative to an external 1 MAl(NO3)3 solution set to δAl = 0 ppm.

■ RESULTSSorption Isotherm. A sorption isotherm was carried out

over a Zn concentration range of 0.1−3 mM (Figure 1). Overthis range, the surface coverage increased from 0.44 to 11.3μmol m−2. The isotherm was fitted well using a Langmuirmodel (R2 = 0.97) and yielded a maximum sorption capacity of

Environmental Science & Technology Article

dx.doi.org/10.1021/es3018094 | Environ. Sci. Technol. XXXX, XXX, XXX−XXXB

12.6 μmol m−2. Several samples at different initial Znconcentrations were chosen for further analyses (Figure 1).X-ray Absorption Fine Structure Spectroscopy. Figure

2A shows that the background subtracted k3-weighted χ

functions of the adsorption samples with a Zn concentration≥0.4 mM are nearly identical to that of the Zn−Al LDH modelcompound. These results suggested Zn−Al LDH precipitatesformed in these sorption samples. This assignment was furthersupported by the shell-by-shell fitting (Figure 2B) of EXAFSdata. The first shell appeared at ∼1.3 Å (not corrected for phaseshift) due to a first oxygen shell contribution. The first shell canbe best fit with six O atoms at 2.06 ± 0.01 Å (SupportingInformation, Table S1), corresponding to a typical octahedralZn coordination. The Debye−Waller factor is about 0.01, largerthan a typical value (0.005) for first shell fitting, which suggestsa slight distortion of the Zn octahedron. The second shellappearing at ∼2.8 Å (not corrected for phase shift) cannot befit with either only a single Zn−Zn scattering path or a Zn−Alscattering. A better fit was achieved by using both a Zn−Al and

a Zn−Zn scattering path. The fitting results suggest the Zn−Zndistance is similar to the Zn−Al distance at ∼3.10 Å, which is ingood agreement with the fitting results of the Zn−Al LDHmodel compound.The XANES (Supporting Information, Figure S3) and

EXAFS (Figure 2) spectra of the 0.2 mM Zn sorption sampleis different from those of the Zn−Al LDH model compoundand aqueous Zn solution. Thus, formation of either surfaceprecipitates or outer-sphere surface complexes was excluded.The 0.2 mM Zn EXAFS spectrum (Figure 2) can best be fittedwith a first Zn−O shell with coordination number (CN) of 4 ata distance of 1.97 Å and a second Zn−Al shell (CN = 1) at2.99bÅ, which corresponds to an inner-sphere bidentatemononuclear complex.7 Our EXAFS results are in goodagreement with the previous research of Trainor et al.7 thatZn−Al LDH precipitates formed at high-sorption-densitysamples (≥1.5 μmol m−2) and inner-sphere mononuclearsorption complexes formed at low-sorption-density samples(<1.5 μmol m−2).

Synchrotron Radiation Powder X-ray Diffraction. SR-XRD was used to further analyze the long-range structuralorder of LDH precipitates. The XRD patterns of the 0.2 mMZn reacted sample are almost identical to bulk γ-alumina(Figure 3), except for very small differences at a range of 0−2°,

which is probably because the sorption sample has a differenthydration state. This is consistent with the prior EXAFSanalysis that no precipitates formed on this sample. In contrast,two additional diffraction peaks at about 2.4° and 4.8° wereobserved for the samples reacted with 0.4 mM Zn and higherconcentration (Figure 3). This translates to d-spacings of 0.76and 0.38 nm, which can be indexed to the (006) and (003)reflections of typical Zn−Al LDH materials.28 For the 3 mM Znreacted γ-alumina, more additional peaks were identified at2.4°, 4.8°, 7.1°, and 8.0°, all of which can be indexed to Zn−AlLDH.The novel SR-XRD data adds much complementary

structural information to the above EXAFS analysis. Forexample, the peak at 2.4° can be indexed to the (006) reflectionof Zn−Al LDHs, which corresponds to the interlayer distancebetween the Zn−Al hydroxide layers that EXAFS cannotdetect. The peaks at ∼4.8° could be indexed to the (003)

Figure 1. Sorption isotherm of Zn on γ-alumina, fitted with a one-siteLangmuir equation (dashed line). Sorption points (open circle) werelabeled with initial Zn concentration. Errors are within the size of eachsorption point. According the EXAFS analysis, Zn−Al LDH surfaceprecipitates will form only when sorption density is larger than 1.5μmol m−2.

Figure 2. EXAFS analysis of Zn adsorption/precipitation on γ-alumina. (A) k3 weighted χ functions; (B) the corresponding Fouriertransform (phase shift not corrected). Experimental and fitted data arepresented as black solid lines and red dashed lines, respectively.

Figure 3. SR-XRD analysis of nonreacted γ-alumina and Zn reacted γ-alumina at indicated concentrations. The wavelength of the X-raybeams is 0.03196 nm.

Environmental Science & Technology Article

dx.doi.org/10.1021/es3018094 | Environ. Sci. Technol. XXXX, XXX, XXX−XXXC

Figure 4. TEM images of 0.8 mM reacted γ-alumina (A, B), EDS analysis of the selected regions (C), and comparison of SAED patterns with SR-XRD (D).

Figure 5. TEM images of 3 mM reacted γ-alumina (A, B), EDS analysis of the selected regions (C), and comparison of SAED patterns with SR-XRD(D).

Environmental Science & Technology Article

dx.doi.org/10.1021/es3018094 | Environ. Sci. Technol. XXXX, XXX, XXX−XXXD

reflection and slightly shift to a lower diffraction angle as the Znconcentration increased from 0.4 to 3 mM. This change isconsistent with the increasing coordination number of the Zn−Zn/Al shell in the EXAFS spectra (Supporting Information,Table S1). For the 3 mM Zn reacted sample, we found that thepeaks indicative of LDH (Supporting Information, Figure S4)are as broad as those for bulk γ-alumina, which suggests theprecipitates are crystalline and of nanometer size.High-Resolution Transmission Electron Microscopy.

Since SR-XRD suggests the presence of some type of crystallineprecipitate, TEM analysis was undertaken to examine themorphology, elemental distribution, and crystallinity of selectedsamples. Figure S5A, B of the Supporting Information shows alow magnification and an HRTEM image of pristine γ-alumina,which demonstrates that the γ-alumina is crystalline with aparticle size around 10−20 nm. Figure S6 of the SupportingInformation contains a low magnification TEM image of the 0.2mM Zn sample and a selected area electron diffraction (SAED)pattern with a rotationally averaged intensity profile. The SAEDprofile compares well with the SR-XRD for pristine γ-alumina.SAED, EDS, and TEM data all indicate that very little Zn ispresent and that no separate precipitates have formed.After reaction with 0.8 mM Zn (Figure 4A), a separate

apparently needle-shaped precipitate (150 nm × 25 nm) wasobserved with a particle size larger than pristine γ-alumina.Upon tilting the sample by 40°, the precipitate was identified tohave a flake-shaped morphology. The flake morphology isconsistent with the layered structure of LDH materials. EDSanalysis of I (Figure 4B, C) showed the precipitate to be rich inZn. The presence of Al cannot be determined due to theoverlap of γ-alumina. Neighboring γ-alumina did not containsignificant Zn (analysis II). In addition to the dark needle/flakeprecipitate, there were regions of Zn-rich poorly crystallinematerial (region and analysis III in Figure 4A, C). The poorly

crystalline nature of region III was determined by the lack ofchanging diffraction contrast during sample tilting experimentswhile using a small objective aperture. In addition, low-doseHRTEM imaging was performed and indicated that the poorlycrystalline phase was amorphous. (Supporting Information,Figure S7B). Upon non-low-dose imaging conditions, theamorphous material would alter to a spinel ZnAl2O4 phase(Supporting Information, Figure S7E, F) No extra peaks otherthan those for the alumina were found in the rotationallyaveraged SAED profiles (Figure 4D) obtained from areas thatinclude the high contrast precipitate and the poorly crystallinematerial (see also the Supporting Information, Figure S7C, D).The lack of 00l reflections for LDH structures is conspicuous inlight of their presence in the SR-XRD profile. This may be dueto three factors: (1) the LDH structures may collapse in thevacuum of the electron microscope in a way that disrupts thesharpness of the 00l reflections; (2) the thickness of the flakesalong the hk0 direction may be too thick for proper diffraction;and (3) the >0.7 nm basal reflection is very close to the centralbeam and can easily overwhelm faint reflections.In the 3 mM Zn/γ-alumina reacted sample (Figure 5), a

higher concentration of the dark and poorly crystallineprecipitates were observed (Figure 5B). Again, the SAEDprofiles contained no extra peaks other than those that could beidentified for γ-alumina. However, the presence of an LDHphase is clearly indicated by the SR-XRD data, and the EDSdata show the LDH phase to be Zn rich. Analysis II indicatesthat the poorly crystalline material unambiguously contains Al.The presence of both crystalline and poorly crystalline Zn-

rich phases is in contrast with the observation that onlyamorphous Ni−Al LDH phases were found on pyrophyl-lite.31,32 The low concentrations of Zn on γ-alumina crystalsneither preclude sorption of Zn onto their surfaces nor does iteliminate the possibility that Zn surface precipitates have

Figure 6. (A) 27Al SP/MAS NMR analysis of 3 mM Zn reacted γ-alumina (green solid line), nonreacted γ-alumina (red dashed line); (B) a zoom inof (a) in the chemical shift range of +40 to +80 ppm that represents the tetrahedral Al; (C) 27Al SP/MAS NMR spectra of the two samplespresented in (A) and two additional model compounds in the chemical shift range of −25 to +25 ppm that represents the octahedral Al. Thesespectra were collected with single-pulse at a spinning rate of 20 kHz, a pulse delay of 5 s, and approximately 400 scans for each sample.

Environmental Science & Technology Article

dx.doi.org/10.1021/es3018094 | Environ. Sci. Technol. XXXX, XXX, XXX−XXXE

formed. However, if they have formed, they must be in lowconcentrations relative to the separate precipitates. What isknown is that separate crystals of Zn-rich LDH along with amore poorly crystallized or amorphous, and possible precursor,phase have formed.Solid-State NMR. The observation of separate crystalline

LDH precipitates with large particle size implies a large amountof Al mass transfer from γ-alumina to the LDH precipitates,which could be an important mechanism for LDH formation.Because NMR spectroscopy is quantitatively sensitive to thechanges of Al coordination environment (i.e., 4-, 5- and 6-coordinated Al), 27Al solid-state NMR analysis was carried outon several samples to examine the changes in the Alenvironment. The crystal structure of γ-alumina contains∼30% tetrahedral and 70% octahedral Al,33,34 which isconsistent with the intensities of NMR peaks occurring at δAl= +8 ppm and δAl = +68 ppm, respectively (Figure 6A). Ifdissolution occurs, substantial tetrahedral Al in the bulkstructure will dissolve in the solution and convert to octahedralAl in either the solution (e.g., aqueous Al(H2O)6

3+) or solidstate (e.g., Al(OH)3). Also, formation of large amounts of Zn−Al LDH would be reflected in the NMR spectrum, as thechemical shifts for the Al in Zn−Al LDH and the octahedral Alin γ-alumina differ by about 4−5 ppm. As compared to the bulkγ-alumina, the 27Al solid-state NMR spectrum (Figure 6C) forthe 3 mM Zn reacted sample showed a small shoulder at δAl =+12.5 ppm. The NMR signal at +12.5 ppm indicates a mixedZn and Al octahedral structure that is consistent with a Zn−AlLDH model compound (Figure 6C), whereas the signal at δAl‑27= +8 ppm is attributed to a spinel-like octahedral Alenvironment of γ-alumina. To further confirm the assignmentof the small shoulder to Zn−Al LDH precipitates, a sorptionsample was prepared at a higher Zn concentration of 10 mMand a higher temperature of 333 K, enhancing precipitation ofLDH phases. As predicted, a much more significant NMR peakat ∼δAl = +12.5 ppm was observed.In addition, a signal intensity reduction of the peak at δAl =

+68 ppm was observed (Figure 6B), indicating the trans-formation of some tetrahedral Al from the bulk γ-alumina tooctahedral Al in the Zn−Al LDH precipitates. However, thedifferences are very small, and more information cannot bededuced because the precipitates only amount to a smallfraction of the whole sample with respect to the majoritysubstrate, γ-alumina. In considering the 3 mM Zn reactedsample (0.1 g), for example, roughly 125 μmol of Zn wassequestrated in this sample. By assuming that all of the Zn is inthe form of the Zn−Al LDH and a Zn/Al ratio of 2, theprecipitates contain 62.5 μmol of Al (1.688 mg), that is, 3.2% ofthe total Al (53 mg) of the whole sample. Since the fraction oftetrahedral Al in γ-alumina is ∼30%, only a 1% difference wouldbe predicted in the tetrahedral Al before and after reaction,which agrees with our 27Al NMR analysis.Zn Sorption Mechanism on Different Al (Hydr)oxides.

To examine the role of mineral surface dissolution on Znsorption/precipitation, a series of Al oxides and hydroxideswere selected as adsorbents. For comparison purposes, sorptionsamples were prepared under the same experimentalconditions, that is, a Zn concentration of 0.8 mM buffered atpH 7.5 using 10 mM HEPES and a reaction time of 15 min.This short reaction time was chosen because our preliminaryresults suggested that formation of Zn−Al LDH precipitates onγ-alumina was rapid, within 15 min (Supporting Information,Figure S8). Figure 7 shows the background subtracted k3-

weighted χ functions of the sorption samples and thecorresponding Fourier transforms. Precipitates on γ-aluminaand corundum (α-Al2O3) were identified, as suggested by thetwo features shown at about 7.2 and 9.5 Å−1 in the χ functionsand the appearance of a significant second shell in the radialdistribution function (Figure 7). Further shell-to-shell fittingconfirms that the precipitates were Zn−Al LDH phases(Supporting Information, Table S2). In contrast, the EXAFSspectra for sorption samples of boehmite (γ-AlOOH), bayerite(β-Al(OH)3), and gibbsite (α-Al(OH)3) are similar to that of0.2 mM Zn adsorbed γ-alumina, and the FT transforms do notshow a significant second shell. Generally, these EXAFS spectracan be best fit with a first Zn−O shell with a CN of ∼4 at adistance of around 1.97 Å and a second Zn−Al shell (CN = 1−1.5) at 3.01 Å, suggesting a predominant inner-sphere bidentatemononuclear surface complexes.7

■ DISCUSSIONFormation of Crystalline Precipitates. The combined

SR-XRD and HRTEM analyses have identified the formation ofcrystalline Zn−Al LDH precipitates as the dominant Znsorption mechanism on γ-alumina. This suggests a dissolution−reprecipitation process that happens at a high enough Znconcentration in the solution. The observation of crystallineLDH precipitates strongly supports all the previous EXAFSinvestigations and adds important complementary information.These observations are consistent with previous research byPaulhiac and Clause,1 who identified well-crystallized Zn−AlLDH precipitates by XRD when reacting γ-alumina with a highconcentration Zn solution (10 mM) at neutral pH. It is worthnoting that much lower Zn concentrations (0.4−3 mM) wereused in this study. Besides the crystallinity, the TEM analysisclearly revealed that the precipitates were separate from bulk γ-alumina. This supports an early study by d’Espinose de laCaillerie et al.,2 who used a dialysis bag to isolate γ-aluminaduring the Ni (10−100 mM) impregnation experiment (pH 9and 333 K) and identified crystallized Ni−Al LDH precipitatesoutside of the bag. In this study, our TEM images show largepieces of crystallites with particle size much larger than γ-alumina substrate. The large particle size of the LDHprecipitates in this study is comparable to bulk LDH materials

Figure 7. EXAFS analysis of Zn adsorption/precipitation on differentAl (hydr)oxides. (A) k3 weighted χ functions; (B) the correspondingFourier transform (phase shift not corrected). Experimental and fitteddata are presented as black solid lines and red dashed lines,respectively.

Environmental Science & Technology Article

dx.doi.org/10.1021/es3018094 | Environ. Sci. Technol. XXXX, XXX, XXX−XXXF

synthesized by coprecipitation method.28,29 Furthermore, weemployed FTIR to analyze the anions type in these samples(Supporting Information, Figure S9), where IR peaks at 1355cm−1 were observed that showed increasing intensities as theZn concentration increases. This IR band is assigned to thecarbonate group,2,35 which suggests the anions in the interlayersof Zn−Al LDH are CO3

2−. The carbonate is orginated from theatmospherical CO2 during the reaction, since carbonate cancompete nitrate to form more stable carbonate−LDH.The observation of crystalline Zn−Al LDH precipitates

differs from previous TEM studies that amorphous Ni-containing precipitates were observed and associated/depositedon the pyrophyllite surface.31,32 This is probably due to thedissolved Si from pyrophyllite that prevents the crystallizationof Ni−Al LDH. It has been reported that coprecipitation of Znand Al in the presence of silicate would lead to a very poorlycrystallized hydrotalcite-like phase.36 The present study wasconducted with Si-free adsorbent (i.e., γ-alumina); therefore,crystalline LDH precipitates can be identified.Another interesting observation is that the HRTEM images

also showed the presence of poorly crystalline precipitates thatcould not be detected by diffraction methods (Figure 5). Sincethe samples were properly washed and sonicated prior to TEManalysis, it is unlikely that the poorly crystalline Zn phaseprecipitated during drying of the TEM sample. It is unclear ifthis precipitate is an LDH-like (or a-LDH) phase and why itformed. During the synthesis of Mg−Al LDH, Yang et al.35

observed amorphous colloidal aluminum hydroxide andsuggested this phase as a precursor for the LDH formation.In our study, the poorly crystalline precipitates might be aprecursor of a crystalline phase as well. It is not possible withEXAFS to determine whether the amorphous phase is LDH-like, since it is in the presence of a crystalline Zn-LDH phaseidentified by SR-XRD. Further understanding of the role of thisphase was hampered by the current technical limitations.Role of Al in Precipitation. The role of Al was shown to

be important in the formation of secondary LDH precipitates.Scheinost et al.37 indicated Ni−Al LDH seems to bethermodynamically favored when Al is available. Using DRS,they observed the formation of α-Ni(OH)2 on Al-poor minerals(e.g., talc and amorphous silica) whereas Ni−Al LDHprecipitates formed on Al-rich adsorbents (e.g., pyrophylliteand gibbsite). Gao et al.30 observed single crystal Zn−Alhydrotalcite while reacting Zn solution with an Al-bearing glasssubstrate. They also found that the precipitates grew on the Alsurface but not on the glass surface. Both Scheidegger et al.3

and Towle et al.4 suggested that adding an additional Me−Alscattering path (Me could be Ni or Co) would lead to a betterfitting of EXAFS data, suggesting the presence of Al in theLDH phase. All of these studies suggested Al was an essentialcomponent in the precipitates. However, direct characterizationof the presence of Al in the previous studies was hampered dueto experimental limitations. In our study, SR-XRD and NMRanalyses provided more direct evidence. X-ray diffractionpatterns reflect the long-range structural order of the entirecrystal lattice whereas EXAFS only provides average localstructural information with a short-range order at most of 5 Å.The 27Al solid-state NMR analysis provided additional directevidence for the presence of octahedral Al in the LDHprecipitates and some evidence for the dissolution of bulk γ-alumina.Since γ-alumina is the only source of Al in our research

system, migration of Al from γ-alumina to the precipitates is

critical for the formation of Zn−Al LDH phase. Dissolution ofthe mineral surface has been hypothesized to be the drivingforce for metal−Al LDH formation, but experimental evidenceis sparse. In this research, we show a clear link between mineralsurface dissolution and surface precipitation. Thus, wecompared Zn sorption/precipitation on a series of aluminum(hydro)xides (Figure 7). These minerals have similar chemicalcomposition but have different surface dissolution constants(Supporting Information, Table S3). Their well-known surfacechemistry shows the following dissolution trend:7 γ-alumina (γ-Al2O3) > corundum (γ-Al2O3) > bayerite (β-Al(OH)3) >boehmite (γ-AlOOH) > gibbsite (α-Al(OH)3). Under the sameexperimental conditions (i.e., Zn concentration of 0.8 mM,solid/solution ratio of 2.5 g/L, and reaction time of 15 min),Zn−Al LDH surface precipitates were detected on γ-aluminaand corundum using EXAFS. In contrast, on bayerite,boehmite, and gibbsite, no precipitates were observed butmainly adsorption complexes were identified. Additional time-dependent EXAFS studies were performed on γ-alumina andboehmite. As shown in Figure S8 and Table S4 of theSupporting Information, the results demonstrate that Zn−AlLDH precipitates formed on γ-alumina after reaction of 15 minand remained up to 7 days. In contrast, for Zn sorption onboehmite, only Zn inner-sphere mononuclear sorptioncomplexes formed during the initial 15 min and later convertedto Zn−Al LDH precipitates after 24 h. These results clearlyindicate that formation of precipitates is highly related tomineral surface dissolution. A similar relationship has beensuggested for Ni precipitation on pyrophyllite, montmorilollite,and gibbsite.12 Ni−Al LDH precipitates formed only 15 minafter reaction with pyrophyllite but required 2 days onmontmorillonite. Given that corundum has the lowest specificsurface area and the lowest surface loading, we conclude that acrucial step for surface precipitation is mineral dissolution.However, more detailed investigations will be needed toestablish a quantitative relationship between the kinetics ofsurface precipitation and mineral surface reactivity.

■ ASSOCIATED CONTENT*S Supporting InformationZn speciation diagram, Zn sorption kinetics, XANES spectra ofZn reacted γ-alumina, SR-XRD patterns for precipitates,additional TEM images, EXAFS of Zn sorption samples,FTIR spectra, and EXAFS fitting parameters. This material isavailable free of charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*Phone: (302) 831-3219; fax: (302) 831-6505; e-mail: weili@udel.edu.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe sincerely appreciate the helpful comments from threeanonymous reviewers and from the editor, Dr. Scherer. Thisresearch was funded by the National Science Foundation(NSF) through the Delaware EPSCoR program (grant no.EPS0814251). We thank Cathy Olsen at the University ofDelaware for assistance with the ICP-OES analyses and Dr.Kaumudi Pandya for help with XAS data collection at beamlineX11A of the NSLS. Y.J.W. is grateful to the Shanghai

Environmental Science & Technology Article

dx.doi.org/10.1021/es3018094 | Environ. Sci. Technol. XXXX, XXX, XXX−XXXG

Synchrotron Radiation Facility (SSRF) for use of thesynchrotron radiation facilities for analyzing the modelcompounds. Drs. Mengqiang Zhu (LBNL) and Paul Northrup(BNL) are acknowledged for assistance with EXAFS dataanalysis. Use of the National Synchrotron Light Source,Brookhaven National Laboratory, was supported by the U.S.Department of Energy, Office of Science, Office of Basic EnergySciences, under contract no. DE-AC02-98CH10886.

■ REFERENCES(1) Paulhiac, J. L.; Clause, O. Surface coprecipitation of Co(II),Ni(II), or Zn(II) with Al(III) ions during impregnation of γ-alumina atneutral pH. J. Am. Chem. Soc. 1993, 115, 11602−11603.(2) d’Espinose de la Caillerie, J.-B.; Kermarec, M.; Clause, O.Impregnation of γ-alumina with Ni(II) and Co(II) at neutral pH:Hydrotalcite-type coprecipitate formation and characterization. J. Am.Chem. Soc. 1995, 117, 11471−11481.(3) Scheidegger, A. M.; Lamble, G. M.; Sparks, D. L. Investigation ofNi sorption on pyrophyllite: An XAFS study. Environ. Sci. Technol.1996, 30, 548−554.(4) Scheidegger, A. M.; Lamble, G. M.; Sparks, D. L. Spectroscopicevidence for the formation of mixed-cation, hydroxide phases uponmetal sorption on clays and aluminum oxides. J. Colloid Interface Sci.1997, 186, 118−128.(5) Towle, S. N.; Bargar, J.; Brown, G. E.; Parks, G. A. Surfaceprecipitation of Co(II) (aq) on Al2O3. J. Colloid Interface Sci. 1997,187, 62−82.(6) Ford, R. G.; Sparks, D. L. The nature of Zn precipitates formed inthe presence of pyrophyllite. Environ. Sci. Technol. 2000, 34, 2479−2483.(7) Trainor, T. P.; Brown, G. E.; Parks, G. A. Adsorption andprecipitation of aqueous Zn(II) on alumina powders. J. ColloidInterface Sci. 2000, 231, 359−372.(8) Sparks, D. L. Environmental Soil Chemistry, 2nd ed.; AcademicPress: Boston, 2002.(9) Johnson, C. A.; Glasser, F. P. Hydrotalcite-like minerals(M2Al(OH) 6(CO3)0.5·xH2O, where M = Mg, Zn, Co, Ni) in theenvironment: synthesis, characterization and thermodynamic stability.Clays Clay Miner. 2003, 51, 1−8.(10) Sideris, P. J.; Nielsen, U. G.; Gan, Z.; Grey, C. P. Mg/Alordering in layered double hydroxides revealed by multinuclear NMRspectroscopy. Science 2008, 321, 113−117.(11) Thompson, H. A.; Parks, G. A.; Brown, G. E. Dynamicinteractions of dissolution, surface adsorption, and precipitation in anaging cobalt(II)-clay-water system. Geochim. Cosmochim. Acta 1999,63, 1767−1779.(12) Thompson, H. A.; Parks, G. A.; Brown, G. E. Formation andrelease of cobalt(II) sorption and precipitation products in agingkaolinite-water slurries. J. Colloid Interface Sci. 2000, 222, 241−253.(13) Scheidegger, A. M.; Strawn, D. G.; Lamble, G. M.; Sparks, D. L.The kinetics of mixed Ni-Al hydroxide formation on clay andaluminum oxide minerals: A time-resolved XAFS study. Geochim.Cosmochim. Acta 1998, 62, 2233−2245.(14) Ford, R. G.; Scheinost, A. C.; Scheckel, K. G.; Sparks, D. L. Thelink between clay mineral weathering and the stabilization of Nisurface precipitates. Environ. Sci. Technol. 1999, 33, 3140−3144.(15) Catalano, J. G.; Warner, J. A.; Brown, G. E. Sorption andprecipitation of Co(II) in Hanford sediments and alkaline aluminatesolutions. Appl. Geochem. 2005, 20, 193−205.(16) Peltier, E.; van der Lelie, D.; Sparks, D. L. Formation andstability of Ni−Al hydroxide phases in soils. Environ. Sci. Technol. 2010,44, 302−308.(17) Roberts, D. R.; Sparks, D. L. Zinc speciation in a smelter-contaminated soil profile using bulk and microspectroscopictechniques. Environ. Sci. Technol. 2002, 36, 1742−1750.(18) Juillot, F.; Morin, G.; Ildefonse, P.; Trainor, T. P.; Benedetti, M.;Galoisy, L.; Calas, G.; Brown, G. E. Occurrence of Zn/Al hydrotalcitein smelter-impacted soils from northern France: Evidence from

EXAFS spectroscopy and chemical extractions. Am. Mineral. 2003, 88,509−526.(19) Nachtegaal, M.; Marcus, M. A.; Sonke, J. E.; Vangronsveld, J.;Livi, K. J. T.; van Der Lelie, D.; Sparks, D. L. Effects of in situremediation on the speciation and bioavailability of zinc in a smeltercontaminated soil. Geochim. Cosmochim. Acta 2005, 69, 4649−4664.(20) Voegelin, A.; Kretzschmar, R. Formation and dissolution ofsingle and mixed Zn and Ni precipitates in soil: Evidence from columnexperiments and extended X-ray absorption fine structure spectrosco-py. Environ. Sci. Technol. 2005, 39, 5311−5318.(21) McNear, D. H.; Chaney, R. L.; Sparks, D. L. The effects of soiltype and chemical treatment on nickel speciation in refinery enrichedsoils: A multi-technique investigation. Geochim. Cosmochim. Acta 2007,71, 2190−2208.(22) Roberts, D. R.; Ford, R. G.; Sparks, D. L. Kinetics andmechanisms of Zn complexation on metal oxides using EXAFSspectroscopy. J. Colloid Interface Sci. 2003, 263, 364−376.(23) Brown, G. E.; Catalano, J. G.; Templeton, A. S.; Trainor, T. P.;Farges, F.; Bostick, B. C.; Kendelewicz, T.; Doyle, C. S.; Spormann, A.M.; Revill, K; Morin, G.; Juillot, F.; Calas, G. Environmental interfaces,heavy metals, microbes, and plants: Applications of XAFS spectros-copy and related synchrotron radiation methods to environmentalscience. Phys. Scr. 2005, T115, 80−87.(24) Li, W.; Harrington, R.; Tang, Y.; Kubicki, J. D.; Aryanpour, M.;Reeder, R. J.; Parise, J. B.; Phillips, B. L. Differential pair distributionfunction investigation on structure of arsenate adsorbed on nano-crystalline γ-alumina. Environ. Sci. Technol. 2011, 45, 9687−9692.(25) Nowack, B.; Sigg, L. Adsorption of EDTA and metal-EDTAcomplexes onto goethite. J. Colloid Interface Sci. 1996, 177, 106−121.(26) Hammersley, A. P. ESRF Internal Report, ESRF98HA01T,FIT2D V9.129 Reference Manual V3.1, 1998.(27) Webb, S. SIXPack: a graphical user interface for XAS analysisusing IFEFFIT. Phys. Scr. 2005, T115, 1011−1014.(28) Liu, Z.; Ma, R.; Ebina, Y.; Iyi, N.; Takada, K.; Sasaki, T. Generalsynthesis and delamination of highly crystalline transition-metal-bearing layered double hydroxides. Langmuir 2007, 23, 861−867.(29) Xu, Z. P.; Stevenson, G.; Lu, C. Q.; Lu, G. Q. Dispersion andsize control of layered double hydroxide nanoparticles in aqueoussolutions. J. Phys. Chem. B 2006, 110, 16923−16929.(30) Gao, Y. F.; Nagai, M.; Masuda, Y.; Sato, F.; Seo, W. S.;Koumoto, K. Surface precipitation of highly porous hydrotalcite-likefilm on Al from a zinc aqueous solution. Langmuir 2006, 22, 3521−3527.(31) Scheidegger, A. M.; Fendorf, M.; Sparks, D. L. Mechanisms ofnickel sorption on pyrophyllite: Macroscopic and microscopicapproaches. Soil Sci. Soc. Am. J. 1996, 60, 1763−1772.(32) Livi, K. J. T.; Senesi, G. S.; Scheinost, A. C.; Sparks, D. L.Microscopic examination of nanosized mixed Ni−Al hydroxide surfaceprecipitates on pyrophyllite. Environ. Sci. Technol. 2009, 43, 1299−1304.(33) Lee, M. H.; Cheng, C. F.; Heine, V.; Klinowski, J. Distributionof tetrahedral and octahedral Al sites in gamma alumina. Chem. Phys.Lett. 1997, 265, 673−676.(34) Ravenelle, R. M.; Copeland, J. R.; Kim, W. G.; Crittenden, J. C.;Sievers, C. Structural changes of γ-Al2O3-supported catalysts in hotliquid water. ACS Catal. 2011, 1, 552−561.(35) Yang, Y.; Zhao, X.; Zhu, Y.; Zhang, F. Transformationmechanism of magnesium and aluminum precursor solution intocrystallites of layered double hydroxide. Chem. Mater. 2012, 24, 81−87.(36) Depeg̀e, C.; El Metoui, F.; Forano, C.; de Roy, A.; Dupuis, J.;Besse, J. Polymerization of silicates in layered double hydroxides.Chem. Mater. 1996, 8, 952−960.(37) Scheinost, A. C.; Ford, R. G.; Sparks, D. L. The role of Al in theformation of secondary Ni precipitates on pyrophyllite, gibbsite, talc,and amorphous silica: A DRS Study. Geochim. Cosmochim. Acta 1999,63, 3193−3203.

Environmental Science & Technology Article

dx.doi.org/10.1021/es3018094 | Environ. Sci. Technol. XXXX, XXX, XXX−XXXH