Environ. Sci. Technol. 2004, 38, 5426-5432
EXAFS Study of Zn Sorption Mechanisms on Montmorillonite S H I N W O O L E E , †,‡ P A U L R . A N D E R S O N , * ,† GRANT B. BUNKER,§ AND CAHIT KARANFIL§ Department of Chemical and Environmental Engineering and Department of Biological, Chemical, and Physical Sciences, Illinois Institute of Technology, Chicago, Illinois 60616
Extended X-ray absorption fine structure (EXAFS) analysis of zinc sorption on montmorillonite showed that different types of surface complexes or surface precipitates were formed depending on the reaction time. With an initial zinc concentration of 10-3 M at neutral pH, zinc remained octahedrally coordinated with about six oxygen atoms at Zn-O bond distances of 2.02-2.07 Å for up to six months. For samples aged up to 11 days, the Zn-Zn contribution in the second shell suggested formation of multinuclear surface complexes or surface precipitates. For samples aged 20 days and more, Zn-Zn and Zn-Si contributions in the second shell suggested formation of mixed metal coprecipitates such as a Zn phyllosilicate-like phase. Formation of these mixed metal solids probably accounts for the slow continuous sorption reaction at aging times exceeding 20 days. Sequestration of Zn in mixed metal precipitates and the stability of these phases can reduce the concentration, mobility, and toxicity of Zn in soils or sediments.
Introduction Sorption on the surfaces of clay minerals and oxides plays an important role in the fate of metal ions in soil, sediment, and aquatic environmental systems. The final form of metal sorption products influences the mobility, bioavailability, and ultimately toxicity in the environment. Understanding metal speciation and sorption complexes can help to select appropriate management options and remediation methods for contaminated environments. X-ray absorption spectroscopy (XAS) is a state-of-the-art technique that can be used to provide molecular-level information about the local structure of a metal contaminant. Recent studies using extended X-ray absorption fine structure (EXAFS) spectroscopy suggest formation of mixed metal hydroxides of Zn, Co, or Ni in the presence of clay and oxide minerals at neutral to slightly high pH. There is, however, still discrepancy about the form of these phases because it can be difficult to separate backscattering contributions from Al or Si in the second shell. Trainor et al. (1) found that hydrotalcite-like phases were produced during Zn sorption on alumina (R-Al2O3) powders. Ford and Sparks (2) studied Zn partitioning to the pyrophyllite surface and suggested that a Zn-Al layered double hydroxide (LDH) was * Corresponding author phone: (312)567-3531; fax: (312)567-8874; e-mail:
[email protected]. † Department of Chemical and Environmental Engineering. ‡ Present address: Department of Plant and Soil Sciences, University of Delaware, Newark, DE 19716. § Department of Biological, Chemical, and Physical Sciences. 5426
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 20, 2004
formed during sorption. Because Zn solubility decreased, they hypothesized that this phase may be a precursor for a Zn phyllosilicate-like phase, a possible end product. Schlegel et al. (3) reported nucleation and epitaxial growth of Zncontaining phyllosilicate [e.g., Zn3Si2O5(OH)4 or Zn3Si4O10(OH)2] on a hectorite layer using polarized EXAFS (P-EXAFS) spectroscopy. Other studies have also observed coprecipitation of mixed Ni-Al LDH in various Al-containing clay and minerals including pyrophyllite, illite, kaolinite, gibbsite, alumina, and montmorillonite (4-7) and formation of R-type metal hydroxides in Al-free minerals including talc, silica, and rutile (8, 9). Ford et al. (10) proposed that the increased stability of sorbed nickel on pyrophyllite occurred because the initial phase was a Ni-Al LDH that gradually changed into a precursor Ni-Al phyllosilicate by the substitution of silica for nitrate within the Ni-Al LDH interlayer. Da¨hn et al. (11) used polarized EXAFS spectroscopy, observed formation of a Ni phyllosilicate phase on montmorillonite, and suggested that Scheidegger et al. (4, 5) may not have detected the formation of Ni phyllosilicate phases in their study of Ni on montmorillonite. Charlet and Manceau (12) and Manceau et al. (13) suggested formation of clay-like phases from their studies of Co and Ni uptake on amorphous silica (SiO2) and quartz (R-SiO2), and Towle et al. (14) also discussed surface precipitation of Co at metal oxide surfaces. By examining the sorption/precipitation reactions in this study, we identified sorption species and the local coordination environment of Zn sorbed on montmorillonite as a function of aging time. XAS was used to explore the possibility that formation of a mixed metal precipitate such as a Zn-Al hydrotalcite-like phase or a Zn phyllosilicate-like phase could be a reason for the second slow sorption reaction. Formation of these mixed metal precipitates would improve sequestration of zinc and reduce migration and bioavailability of zinc in soils or sediments.
Materials and Methods Batch Sorption Experiment and Sample Preparation. Batch experiments of Zn sorption onto montmorillonite were conducted at room temperature (∼22-24 °C) with fixed pH of 7.0 in 0.1 M NaNO3. Na-montmorillonite (Swy-2, obtained from the Clay Minerals Society) was used as received without any pretreatment. The N2-BET surface area of Na-montmorillonite was 28.26 ( 1.36 m2/g, consistent with the reported value of 31.82 ( 0.22 m2/g (15). The reported pH at the point of zero charge of Na-montmorillonite was ∼7.58.5 (16, 17). The pH was maintained using 20 mM PIPES [piperazine-N,N′-bis(2-ethanesulfonic acid)] buffer. Preliminary tests confirmed that pH remained constant, and the effect on Zn sorption was negligible. The initial Zn concentration was 10-3 M (stable with respect to bulk precipitation at pH ) 7), and the solid/liquid ratio was 10 g/L. Montmorillonite was in contact with the background electrolyte and buffer for 4 h before adding Zn(NO3)2. A stock solution [0.1 M of Zn(NO3)2, pH < 2] was added drop-by-drop to achieve the intended Zn solution concentration. Zn sorption experiments were conducted in the absence of atmospheric carbon dioxide. High-purity N2 gas was bubbled through the suspension and mixed by a stirrer inside of a chamber filled with N2 during reaction times of 20 days or one month. These preparation steps were conducted in 1-L borosilicate glass beakers; after that, the suspensions were transferred to HDPE bottles, sealed, and placed on a tumbler for aging tests. Although we cannot rule out the possibility, it seems unlikely that these changes (glass to HDPE, N2 bubbles to tumbling) would promote a precipitation reaction. 10.1021/es0350076 CCC: $27.50
2004 American Chemical Society Published on Web 09/17/2004
For the XAFS sample preparation, 50 mL of suspensions equilibrated up to 6 months was collected and centrifuged at 20 000g for 30 min to separate the solid and liquid phases. Supernatants were filtered again using a 0.2 µm pore membrane filter and acidified. Dissolved metal concentrations were measured by flame atomic absorption spectroscopy. The solid was freeze-dried and stored in a desiccator prior to XAFS measurements. All solutions were made with double distilled water (DDW) and ACS reagent grade chemicals. A zinc-aluminum reference material, Zn-Al hydrotalcite-like compound (HTlc), was precipitated at room temperature following the method of Cavani et al. (18). On the basis of X-ray diffraction, Cavani et al. (18) confirmed that their precipitate had a formula of Zn6Al2(OH)16CO3‚4H2O. XAFS Measurements and Analysis. Zn K-edge (9659 eV) XAFS measurements were conducted in fluorescence mode using a Photomultiplier Tube (PMT) detector with a Bent Laue Analyzer (19) over the energy range of 9560-10760 eV. The beam energy was calibrated by assigning the first inflection point of the zincite standard compound (9659 eV). These measurements were made at room temperature at the Materials Research Collaborative Access Team (MR-CAT) sector 10 or the Biophysics Collaborative Access Team (BioCAT) sector 18 at the Advanced Photon Source (APS), Argonne National Laboratory. Electron beam energies were 7 GeV with a storage beam current of about 100 mA. Si(111) doublecrystal monochromators were used with X-ray mirrors to reduce the contributions of higher order harmonics. Samples were mounted in a 2 mm thick acrylic holder and positioned at a 45° angle to the relative incident beam; the Bent Laue Analyzer and PMT detector were placed normal to the beam and at a 45° angle to the sample. Helium gas was used as a filling gas for the incident beam (I0) detector chamber. Multiscanned spectra of each sample were collected (five to 10 scans) to improve the signal-to-noise ratio; all scans were merged together in the first step of analysis. EXAFS data analysis was performed using the MacXAFS 4.0 (20) program. Background absorbance of the spectra was removed, and EXAFS (chi) data were plotted as a function of electron wavenumber k (inverse Å). The extracted EXAFS function was weighted by k3 to enhance the high-k region. Fourier transforms were conducted over the k range of 2-11.01 Å-1 for the Zn samples to produce the radial structure function (RSF). Single and multishell in the RSF were isolated and back-transformed (the R-interval was 1.03-2.07 Å for the first shell and 2.37-3.37 Å for the second shell), and then the filtered data were converted into an amplitude (envelope) function and a phase function. Using the back transformed data, fits in k-space were assessed using a nonlinear leastsquares fit procedure with a theoretical reference model compound generated by FEFF8 (21) with WebAtoms 1.4 (22). EXAFS data were fit with phase and amplitude reference structures of ZnAl2O4 and hemimorphite [Zn4Si2O7(OH)2‚H2O] to get the Zn-O, Zn-Zn, Zn-Al, and Zn-Si relationships. The amplitude reduction factor (S02), which accounts for energy loss due to multiple excitations, was set to 0.825 based on fitting the experimental zincite (ZnO) EXAFS spectrum to get the coordination number of Zn-O to its known value of four. For the sample analyses, the coordination number N, the bond distance R, and the Debye-Waller factor σ2 were allowed to float as adjustable parameters. A single energy shift, the ∆Ε0 value obtained using MacXAFS during the fitting procedure, was used to maintain the same edge shift for all shells in a multishell data set. For the second shell, the value of σ2 was constrained to be 0.01Å2 because of the high noise level and the difficulty of a multiple-shell analysis (23) and to decrease the degrees of freedom during the fitting process. The number of floating parameters in the fitting procedures did not exceed the maximum number of independent parameters.
FIGURE 1. Zn sorption on montmorillonite as a function of reaction time. Initial Zn ) 10-3 M, montmorillonite ) 10.0 g/L, and pH ) 7.0 in 0.1 M NaNO3.
Results and Discussion Macroscopic Assessment of Zn Sorption on Montmorillonite. Sorption of Zn onto montmorillonite (Figure 1) was initially fast, with 40% uptake occurring in the first 20 min. Subsequent uptake continued slowly; about 80% was sorbed after 6 months reaction/exposure time. Changes in uptake were insignificant after 4 months. These bulk sorption results follow the general behavior of metal sorption onto minerals or oxide materials. Results from numerous studies suggest that adsorption to available surface sites take place during the initial rapid uptake, while the subsequent slower uptake could be due to sorption to less reactive sites, diffusion, or formation of surface precipitates (4, 24-30). For example, in their study of Zn sorption on pyrophyllite, Ford and Sparks (30) concluded that the formation of a Zn-Al layered double hydroxide (LDH), a mixed metal hydroxide precipitate, was the reason for the slow sorption reaction step. They based their conclusion on spectroscopic studies of the sorption product. Sorption Samples on Montmorillonite Reference Compounds. As reference materials, XAFS spectra of three Zn model compounds, 1 mM Zn(NO3)2(aq) solution (pH 1.97), zincite (ZnO), and Zn-Al hydrotalcite-like compound (HTlc), were also measured with sorption samples. The EXAFS spectra (χ-function) and the Fourier transformed radial structure functions (RSFs) corresponding to the χ-function EXAFS spectra of the reference compounds are shown in Figure 2. The χ-function EXAFS spectrum of the 1 mM Zn(NO3)2(aq) solution, which shows a single sinusoidal oscillation due to the backscattering of oxygen atoms in the first shell, provides a good example of an outer-sphere complex of Zn. A similar characteristic can be seen in the RSF of a zinc nitrate solution, which has only a single peak at about 1.7 Å (phase shift uncorrected) indicating a single shell of backscattering atoms neighboring the central atom. Zn was coordinated octahedrally with six oxygen atoms at interatomic bond distance of 2.07 Å in Zn(NO3)2(aq) solution (Table 1). This result is in good agreement with other results of recent EXAFS studies. For example, Trainor et al. (1) found that 10 mM Zn(NO3)2(aq) solution (at pH 3.6) had a 6-fold coordination of zinc with the Zn-O distance of 2.07 Å in the first shell. Bochatay and Persson (23) also reported the Zn-O distance of Zn2+(aq) in acidic solution was 2.07 Å when the coordination number of the oxygen shell was held constant at six. Because of the contribution of higher-shell atoms in the coordination environment of Zn, the χ-function EXAFS spectra of zincite and Zn-Al HTlc have more distinct structural features than appear for aqueous Zn2+. Fourier VOL. 38, NO. 20, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5427
TABLE 1. Structural Parameters for Zn Sorbed on Montmorillonite and Reference Materials from EXAFS Analysisa first shell (Zn-O)
second shell (Zn-Zn)
sample reaction time 1h 1 day 11 days 20 days 1 month 2 months 6 months ZnO Zn-Al HTlc Zn(NO3)2 solution, 1mM
(Zn-Al/Si)
Zn loading (mg/g)
N
R (Å)
σ2 (Å2)
N
R (Å)
σ2 (Å2)
2.77 3.05 3.57 4.39 4.78 4.81 4.98
5.4 5.4 5.6 6.2 6.0 6.1 6.1
2.02 2.02 2.02 2.05 2.05 2.05 2.07
0.013 0.013 0.012 0.012 0.011 0.011 0.007
1.5 2.0 2.7 4.5 5.1 4.7 6.1
3.09 3.10 3.09 3.11 3.10 3.10 3.10
0.01 0.01 0.01 0.01 0.01 0.01 0.01
4.0 6.1 6.8
1.97 2.06 2.07
0.003 0.011 0.009
12.0 4.3
3.23 3.11
0.01 0.01
N
3.9 3.9 4.5 4.1
R (Å)
3.28 3.26 3.28 3.27
σ2 (Å2)
∆E0 (eV)
reduced χ2
0.01 0.01 0.01 0.01
-5.49 -5.61 -5.39 -5.66 -5.77 -5.57 -6.82
0.064 0.098 0.117 0.112 0.265 0.071 0.605
-8.63 -7.08 -4.74
0.538 0.152 0.015
a The initial concentration of Zn was 10-3 M with background concentration of 0.1 M NaNO , pH ) 7, and solid/liquid ratio of 10 g/L. N is 3 coordination number, R is interatomic distance, and σ2 is Debye-Waller factor. Values in italics indicate parameters that were fixed during the data analysis. The accuracy for the NZn-O, NZn-Zn, and NZn-Al/Si was ( 120, ( 145, and ( 160%; RZn-O, RZn-Zn, and RZn-Al/Si was ( 10.02, ( 10.02, and ( 10.09 Å, respectively.
FIGURE 2. (a) Normalized, background-subtracted, and k3-weighted EXAFS spectra (χ-function) of the Zn reference compounds. (b) RSF of Zn reference compounds obtained by Fourier transformation of the EXAFS spectra. transformed RSFs of these spectra indicate that some of this backscatter was due to the existence of higher shells. For zincite, the first shell Zn-O distance was 1.97 Å with four oxygen atoms, and the second shell had Zn-Zn distance of 3.23 Å (the second shell coordination numbers were held constant at 12 Zn atoms). For the Zn-Al HTlc coordination environment, the first shell Zn-O distance was 2.06 Å with a coordination number showing 6.1 oxygen atoms. The second shell of Zn-Al HTlc was fitted only with Zn-Zn 5428
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 20, 2004
contribution with coordination number of 4.3 at a radial distance of 3.11 Å. EXAFS Analysis of Zn Sorption on Montmorillonite. EXAFS spectra (χ-function) of Zn samples sorbed on montmorillonite for different reaction times are shown in Figure 3a. More complex features are shown in the χ-function EXAFS spectra of all samples, especially at k ranges greater than ∼4.7 Å-1. These spectra indicate the presence of higher coordination shells of backscattering atoms surrounding the central Zn atom. As reaction times increased, more highly structured features appear at higher k ranges, and new features can be seen at k ) ∼5 and ∼7.2 Å-1. The samples can be divided into two distinct groups, those reacted up to 11 days and those reacted over 20 days. It is clear from the χ-function EXAFS spectra of these two groups that the Zn sorption environment changed to different coordination structures with increasing reaction times. Figure 3b represents the Fourier transformed RSFs of Zn/ montmorillonite samples corresponding to the EXAFS spectra of Figure 3a. There are two distinct peaks representing the first coordination shell of the zinc atom. We believe the peaklike feature around R ) 2-2.4 Å is an artifact caused by interference from the first and second shells. Oxygen atoms are at ∼1.7 Å (phase shift uncorrected), and the second peak is at ∼2.8 Å in the RSFs of all Zn sorption samples. The presence of the second shells is probably due to the contribution of the second neighbor atoms (Zn-Zn, Zn-Al, and/or Zn-Si) surrounding the central atom, which indicates the formation of innersphere complexes, multinuclear complexes, or precipitates. As expected from the χ-function EXAFS spectra, the intensity of the second peak in the RSFs increases with increasing reaction times. Especially for samples that reacted over 20 days, the intensity of the second peak is considerably larger than it is for the samples that reacted up to 11 days. There is a greater contribution of ZnZn, Zn-Al, and/or Zn-Si backscattering with increasing age. Structural parameters fitted for the samples of Zn on montmorillonite are shown in Table 1. Fitting results of the first shell (Zn-O) show that Zn was coordinated with about six oxygen atoms for all samples. The Zn-O bond distances were from 2.02 to 2.07 Å. Although there was a small increase in distance with increased reaction times, these bond distances suggest that there was no substantial change in the first shell coordination environment due to aging. The first shell coordination numbers were ∼5.4-6.2 oxygen atoms for all samples, which also reveals that there was no consistent change in first shell coordination environment with increasing reaction times. These results of bond distance and coordination number indicate that Zn was in an octahedral
FIGURE 3. (a) Normalized, background-subtracted, and k3-weighted EXAFS spectra (χ-function) of the Zn sorbed onto montmorillonite. (b) RSF obtained by Fourier transformation of the EXAFS spectra. The initial concentration of Zn was 10-3 M with background concentration of 0.1 M NaNO3, pH ) 7, and solid/liquid ratio of 10 g/L. environment with six oxygen atoms at the montmorillonite surface. These results are in good agreement with the observation that Zn generally exists in a 4- or 6-fold coordination environment with first neighbor oxygen atoms. The length of the Zn-O bond is typically 1.92-1.99 Å for 4-fold coordination and 2.02-2.12 Å for 6-fold coordination (31, 32). Figure 4a shows the Fourier-filtered first-shell contribution of sorption samples. For all samples, fits with about six oxygen atoms work well. Results for the second shell fit can be divided into two groups (Table 1); one group includes samples with only single Zn-Zn backscattering, and the other has samples with both Zn-Zn and Zn-Al/Si backscattering. Samples reacted up to 11 days showed only Zn-Zn backscattering with an average coordination number and distance for the Zn-Zn bond of 1.5-2.7 and 3.09-3.10 Å, respectively. The presence of Zn second neighbor atoms indicates the possible formation of Zn polynuclear surface complexes or surface precipitates. The short distance of the Zn-Zn bond in these phases suggests that the structure of these solids is not crystalline hydroxide or oxide but a modified form appropriate to the montmorillonite structure. Attempts to include Al or Si atoms with Zn atoms in the second shell fitting for these samples
were not successful; the fitting parameters had very low accuracies and improbable values (fit results not shown). The absence of Al or Si contributions in the second shell indicates that the formation of mixed metal surface precipitates did not occur for samples reacted up to 11 days. Alternatively, the lack of Al or Si second neighbor backscattering could be due to the amplitude cancellation effect of Al or Si and Zn, where the weak amplitude of a light element such as Al and Si could be canceled by the overlapping of heavier Zn-Zn backscattering (5, 33, 34). Furthermore, the sensitivity of the XAFS signal for weak second-shell backscattering atoms such as Al or Si is low, and the small coordination number (
