Environ. Sci. Technol. 2010, 44, 8491–8496
X-ray Absorption Spectroscopic Investigation of Molybdenum Multinuclear Sorption Mechanism at the Goethite-Water Interface YUJI ARAI* Department of Entomology, Soils, and Plant Sciences, 270 Poole Agricultural Center, Clemson University, Clemson, South Carolina 29634-0315, United States
Received April 20, 2010. Revised manuscript received September 18, 2010. Accepted October 1, 2010.
Understanding in situ metalloid surface speciation on mineral surfaces is critical to predicting the bioavailability in surface and subsurface environments. In this study, Mo K-edge X-ray absorption spectroscopy (XAS) was used to elucidate Mo(VI) surface speciation at the goethite-water interface. Effects of pH and loading levels were investigated. X-ray absorption near edge structure (XANES) analysis indicated that the Mo(VI) coordination environment changes from tetrahedral to octahedral with decreasing pH. At near neutral pH, Mo(VI) predominantly remains as tetrahedral molecules, forming innersphere surface species via corner- and edge-sharing attachment with iron octahedral structures (interatomic distance of Mo-Fe (RMo-Fe) at ∼2.8 and ∼3.4 Å, respectively). In contrast, a mixture of surface species comprising tetrahedrally and octahedrally coordinated Mo(VI) exists at pH ∼3-4. While the same Mo(VI) tetrahedral surface species are present at acidic pH, there was an additional MoO6 polymer attachment on iron octahedral structures, resulting in a RMo-Fe at 3.53 Å. The coordination number (CN) of a Mo-Mo backscatterer gradually increased with increasing loading level, suggesting the formation of surface polymerization. Overall, there seems to be a transition from Mo(VI) tetrahedral to octahedral coordination environment with decreasing pH. The XAS findings further support a Mo(VI) inner-sphere adsorption mechanism that was previously suggested in the pressure-jump relaxation study by Zhang and Sparks (Soil Sci. Soc. Am. J. 1989, 53 (4), 1028-1034). pH-Dependent multinuclear Mo(VI) surface speciation may be important in predicting Mo(VI) transport process in the soil-water environment.
Introduction Molybdenum (Mo) is an important trace element in agroecosystems. It can be either beneficial or toxic to crops and ruminants. Whereas the majority of crops are insensitive to Mo toxicity in soils (2), ruminants develop molybdenosis that induces copper deficiency if they consume forage containing >5 mg/kg of total Mo (3-5). This is due to the strong complexation of thiomolybdate, MoS42-, with copper, a metal that provides critical enzymatic function to connective tissues (6). Therefore the application of Mo in synthetic fertilizers and/or in biosolids (e.g., sludge) on agricultural lands must be carefully managed. Various abiotic * E-mail:
[email protected]; phone: 864-656-2607. 10.1021/es101270g
2010 American Chemical Society
Published on Web 10/21/2010
and biotic factors (pH, redox conditions, ionic strength, adsorbent type, organic matter content, temperature, concentration, competitive adsorbates, solubility product effects, and nonreductive/reductive dissolution of adsorbate) greatly affect the reactivity, speciation, mobility, and bioavailability of Mo. The Mo retention process in soils and sediments is one of the most important rate-limiting factors controlling the dissolved Mo concentration in soil solutions and pore waters. In an oxic environment, Mo is present as Mo(VI), which readily complexes with mineral surfaces. Many researchers have investigated Mo(VI) adsorption mechanisms onto major soil minerals (e.g., iron oxyhydroxides, goethite, and ferrihydrite, kaolinite, illite, aluminum oxide), e.g., refs 7-10. The iron oxide content in soil is presumably important for the Mo retention e.g., ref 8. Molybdenum is known to strongly adsorb in soil minerals at acidic pH, e.g., refs 7, 8. Tetrahedral oxyanion (e.g., MoO42-) adsorption could occur through the formation of outer-sphere and or innersphere surface species. Inner-sphere complexes form via a ligand exchange reaction with a surface functional group. Outer-sphere complexes form mainly by electrostatic interactions and contain more than one water molecule between the adsorbate and the adsorbent functional groups. Several researchers suggested that Mo(VI) forms inner-sphere complexes via ligand exchange mechanisms on soil mineral surfaces. Electrophoretic mobility measurements provide indirect evidence of specific Mo(VI) adsorption, indicating by a large downward shift in the point of zero charge (PZC) of amorphous iron oxide after Mo adsorption (8). Zhang and Sparks (1989) uncovered further evidence supporting ligand exchange Mo sorption mechanisms on goethite surfaces using pressure-jump relaxation measurements (1). Molybdenum adsorption on goethite was independent of ionic strength (0.01-0.1 M NaNO3), and the uptake data were successfully modeled using the modified triple layer model assuming inner-sphere adsorption (1). Similarly, Rietra and co-workers (2001) predicted the formation of bidentate innersphere surface complexes on goethite surfaces using a CDMUSIC model (11). More recently, Lang and co-workers have investigated the effects of aging time on Mo partitioning on goethite surfaces using the pressure-jump relaxation technique (12). They attributed fast relaxation to Mo chemisorption, and slower relaxation to transport within suspension (e.g., film diffusion). They detected no aging effect on Mo surface species. To support macroscopic evidence and the results of indirect approaches, several molecular-scale investigations were conducted. Diffuse reflectance infrared Fourier transform (DRIFT) and Raman spectroscopic investigations have measured broad spectra of the adsorbed Mo on the amorphous iron oxyhydroxide surface and γ-Al2O3 (13-15). However, the bonding environments were not clearly understood due to poorly resolved v3 band splitting. In this study, I employed XAS to elucidate Mo(VI) adsorption mechanisms at the goethite-water interface. Specifically, effects of loading level and pH were investigated. The same experimental conditions described in Zhang and Sparks’ work (1) were also used to reproduce two sorption samples so that the pressure-jump relaxation data could be re-evaluated via XAS analysis.
Materials and Methods Goethite was synthesized according to the method described by Atkinson et al. (16). The synthesis method was chosen to VOL. 44, NO. 22, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8491
reproduce sorption samples similar to those prepared in the conditions of the previous pressure-jump relaxation study by Zhang and Sparks (1). The precipitates were washed several times by centrifugation and decantation with deionized water to remove nitrate. The goethite paste was freeze-dried, and residual ferrihydrite was removed by 0.001 M HCl extraction for 20 min. The five-point Brunauer-Emmett-Teller (BET) surface area was 60 m2 g-1. The powder X-ray diffraction (XRD) analysis revealed the diagnostic d-spacing at 4.17 Å. X-ray Absorption Spectroscopic Analysis. Molybdenum adsorption on goethite was conducted as a function of pH (4 and 7) and Mo concentration (Motot of 100-800 µM) at I ) 0.01 M NaNO3. The reagents (i.e., 10 mM sodium molybdate and 0.01 M sodium nitrate solutions) were prepared in deionized water. Hydration of goethite adsorbent was effected in two steps. In the first step, 100 mL of goethite suspensions were hydrated in 0.01 M NaCl solutions for 2 days. Next, the pH of the suspensions was adjusted to values ranging between 4 and 7, using either 0.1 M HNO3 or 0.1 M NaOH, and equilibrated for an additional 24 h. An appropriate amount of 5 mM sodium molybdenum stock solution was then added to achieve an initial Mo concentration of 0.11, 0.33, and 0.88 mM, total volume of 30 mL, and final suspension density of 0.1 g/L. The samples were reacted for 24 h on an orbital shaker operating at 300 rpm. Two additional samples were reproduced in 1-L reaction vessels according to the method described by Zhang and Sparks (1). The final pH vlaues were 3.2 and 7.2. The total Mo concentration, ionic strength, and suspension density were 4.5 mM, 0.01 M NaNO3, and 15.8 g L-1, respectively (1). The final pH was measured in an aliquot, and 30 mL of the suspension was removed and then centrifuged at 11 950g for 10 min. The supernatant was filtered through 0.2-µm filter papers. The total Mo concentrations of the filtrates were measured using inductively coupled plasma (ICP)-mass spectrometry (MS). The goethite paste was recovered, and was loaded into polycarbonate sample holders, which were then sealed with Mylar/ polycarbonate windows. The samples were wrapped with moist tissues and kept at ∼5 °C prior to XAS data collection. Room-temperature Mo K-edge (20 keV) fluorescence spectra were collected within 40 h of sample preparation at the Stanford Synchrotron Radiation Laboratory (SSRL) beamline 2-3 (Si(220) double crystal monochromator) using a Canberra 13-element Ge detector array equipped with two Al foils. A Mo reference foil was used to calibrate at the Mo K-edge absorption edge energy positions (20 keV). The program FEFF 6 (17) and the structural refinement data of lidgrenite (Cu3(OH)(MoO4)2) and bamfordite (FeMo2O6(OH)3 · (H2O)) were used to estimate backscattering phases and amplitude functions of single scattering (SS) Mo-O, Mo-Mo, and Mo-Fe (18, 19). XAS data reduction and analyses were performed using the IFEFFIT engine based interface, SixPACK (20). k3 Weighted XAS spectra were Fourier transformed at a K range of 2-12 Å-1, and were fit in R-space over the range of 0-4.0 Å for Mo sorbed samples and 0-4.5 Å for lindgrenite and 10 mM Na2MoO4 solution. The solution samples with [Mo]total < 1 mM were not collected due to the detection limit of the beamline. A total of 2-8 XAS spectra were collected in fluorescence mode with a lytle detector or a 13 element Ge array detector at room temperature except for the reference mineral. Lidgrenite was diluted in boron nitrite to achieve ∼65% transmission of X-ray spectra collected in a transmission mode. The XAS data reduction and shell-byshell fit analysis were performed with the SiXPACK/IFEFFIT interface (20) using the method described previously (21). In this study, I did not conduct the linear combination (LC) fit of XANES spectra to determine the ratio of Mo(VI) tetrahedral to octahedral due to lack of Mo(VI) octahedral reference compounds (e.g., monomers to dimers). Monomeric Mo(VI) octahedral (aq) species tend to form at low 8492
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 22, 2010
[Mo(VI)]total < 0.5 mM that is below the detection limit of XAS at the particular beamline. Bamfordite that contains Mo(VI) octahedral dimers was not available due to its rare occurrence.
Results and Discussion Aqueous Speciation. Prior to XAS analysis, Mo(VI) aqueous speciation was assessed using CHEAQS PRO (22). The data are summarized in Supporting Information (SI)-1. The initial reaction condition was used to estimate the percentages of solution speciation. The following equilibrium constants were used in the aqueous speciation calculation (SI-1) (23, 24). The Davies equation was used for activity correction of aqueous species. + (log K ) 3.6) MoO24 + H T HMoO4
(24)
+ 0 MoO24 + 2H T H2MoO4 (log K ) 7.28) (24)
+ 67MoO24 + 8H T Mo7O24 + 4H2O
(log K ) 52.68)
(24)
+ 5+ 4H2O 7MoO24 + 9H T Mo7O23(OH) (log K ) 58.11)
(24)
+ 47MoO24 + 10H T Mo7O22(OH)2 + 4H2O (log K ) 62.44)
(24)
+ 37MoO24 + 11H T Mo7O21(OH)3 + 4H2O (log K ) 67.39)
(25)
In samples at near-neutral pH, the dominant species is the deprotonated molybdenum oxyanion, MoO42- (>99%). In contrast, there are more than one aqueous species present in samples at acidic pH values (3-4). In addition to MoO6(aq), several different polymer species are present: Mo7O246-, HMo7O245-, H2Mo7O244-, and H3Mo7O283-. With increasing the initial [Mo(VI)] concentration from 110 to 880 µM at pH ∼4, the percentage of polymer Mo(VI) aqueous species increases. While nonprotonated Mo7O246- species increased from 0.03 to 1.27%, H3Mo7O243- increased from 1.24 to 48.12%. More importantly, one sorption sample from the previous pressure-jump relaxation study at pH 3.19 shows that ∼90% of total Mo(VI) aqueous species are H3Mo7O243-. XANES Analysis. Figure 1 shows XANES spectra of Mo(VI) reacted goethite samples, a 10 mM sodium Mo(VI) solution sample, and lidgrenite (Cu3(OH)(MoO4)2). The bottom two spectra are samples that are reproduced from the pressurejump relaxation study by Zhang and Sparks (1). The preedge feature (indicated by a black arrow in Figure 1) is attributed to a bound state transition (1s f 4d). The enhanced pre-edge feature is indicative of ModO bonds in the Mo tetrahedral configuration, e.g., refs 26, 27. As the structure of lidgrenite indicates (18), Mo(VI) coordination environment in the mineral is tetrahedral (a spectra in red in Figure 1). A similar pre-edge feature can be seen in the sodium Mo(VI) solution sample, suggesting that the Mo(VI) coordination environment is predominantly tetrahedral under the reaction conditions. The intensity of pre-edge features that is observed in lidgrenite diminishes in the Mo(VI) reacted goethite samples
FIGURE 1. (a) Normalized Mo(VI) K-edge XANES spectra of Mo(VI)-reacted goethite samples and lidgrenite, and 10 mM sodium Mo(VI) solution at pH 6.5. An arrow indicates the pre-edge feature arising from a bound state transition (1s f 4d) of Mo(VI).
FIGURE 3. (a) Normalized, background-subtracted k3-weighted Mo K-edge EXAFS spectra of Mo(VI)-reacted goethite at pH ∼3-4, (b) Fourier-transformed k3-weighted Mo K-edge EXAFS spectra of Mo(VI)-reacted goethite (solid lines) and nonlinear least-squares fits (open circles).
FIGURE 2. (a) Normalized, background-subtracted k3-weighted Mo K-edge EXAFS spectra of Mo(VI)-reacted goethite at pH ∼7, (b) Fourier-transformed k3-weighted Mo K-edge EXAFS spectra of Mo(VI)-reacted goethite (solid lines) and nonlinear leastsquares fits (open circles). Two black arrows in 2b indicate the Mo-Fe shells. at pH ∼3-4. This is consistent regardless of the initial [Mo] and the suspension density at pH 3-4. However, the feature becomes more pronounced in samples at near-neutral pH. This suggests that (1) Mo(VI) might be predominantly present as tetrahedral molecules at near-neutral pH including the solution sample, and (2) Mo(VI) coordination environment is altered at pH ∼4. The pH-dependent transition of Mo(VI) coordination might be occurring at the mineral-water interface. This is not an unusual assumption. In the wellknown UV-vis Mo blue analysis for dissolved phosphate,
the acidic solution is known to facilitate the formation of heteropolymolybdate complexes, P[Mo12O40]3- (28). Additional evidence for the change in the coordination chemistry is the spectra distortion at the post edge regions. A reference line (a dashed line at ∼20 030 eV) was aligned at the peak of a Mo(VI) tetrahedral sample, lidgrenite. Although the peak seems to occur in all samples at pH ∼7, including the solution sample, there was a shift in peak at pH ∼3-4. The post edge feature in pH ∼7 samples seems broadened and the peak occurs at ∼20 025 eV. Accordingly, the interpretation of XANES data was incorporated in the following EXAFS analysis. EXAFS Analysis. The raw and fitted k3-weighted χ functions for the Mo(VI) reacted samples are shown in Figures 2a and 3a. The structural parameters obtained from the linear least-squares fits of the EXAFS data are summarized in Table 1. Using Visual MINTEQ version 2.61 (29), solubility products in these systems were evaluated. All systems were undersaturated with respect to Mo trioxide, MoO3(s). The corresponding radial structure functions (RSFs), obtained by Fourier-transformation of the χ functions, are shown in Figures 2b and 3b. The Mo(IV) sorption samples show two distinct EXAFS spectra in k3-weighted χ functions (Figures 2a and 3a) at respective pH values. In pH ∼7 samples, smooth oscillations occur in the range 2-12 Å-1 (Figure 2a). The features are similar to the EXAFS spectra features of As(V) reacted iron oxyhydroxides, e.g., ref 30. On the contrary, oscillations in pH ∼3-4 samples were distorted at 6-10 Å-1, suggesting changes in surface speciation (Figure 3a). In all samples (Figures 2 and 3), the EXAFS spectra are dominated by backscattering from an O first-shell at ∼1.3 Å in the RSF (uncorrected for phase shift), but there seem to be different contributions from two smaller shells at 2.2-3.75 Å (uncorrected for phase shift) at pH ∼3 and 7, suggesting the effect of pH on Mo(VI) surface speciation at the goethite-water interface. VOL. 44, NO. 22, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8493
TABLE 1. Least-Squares Analyses of Mo K-edge Bulk XAS Analysisa samples
O
O
Mo
Fe1
Fe2
∆Eo (eV)
R-factor
pH 6.56 Na-Mo solution 11 mM
CN R(Å) σ2(Å2)
4.2(2) 1.763(3) 0.002(3)
ss
ss
ss
ss
-3 ( 1
0.004
pH 3.97 110 µM Γ ) 4.4
CN R(Å) σ2(Å2)
3.3(3) 1.742(4) 0.0052(5)
1.0(3) 2.05(2) 0.01
0.4(1) 2.54(2) 0.01b
0.8(2) 2.84(2) 0.01b
2.5(2) 3.537(9) 0.01b
-3 ( 1
0.003
pH 3.92 330 µM Γ ) 7.23
CN R(Å) σ2(Å2)
3.7(5) 1.734(7) 0.0066(9)
1.3(5) 2.03(2) 0.01b
0.6(2) 2.52(2) 0.01b
0.9(2) 2.84(2) 0.01b
2.6(4) 3.53(1) 0.01b
-1 ( 1
0.008
pH 4.08 880 µM Γ ) 26.54
CN R(Å) σ2(Å2)
3.4(4) 1.734(7) 0.0061(9)
0.8(4) 2.02(2) 0.01b
0.8(5) 2.52(3) 0.01b
0.8(2) 2.82(2) 0.01b
2.4(3) 3.52(1) 0.01b
-4 ( 1
0.008
pH 7.31 110 µM Γ ) 3.71
CN R(Å) σ2(Å2)
3.9(2) 1.767(2) 0.0036(2)
ss
ss
0.8(1) 2.82(2) 0.01b
1.4(4) 3.48(2) 0.01b
-2 ( 1
0.003
pH 7.37 330 µM Γ ) 5.26
CN R(Å) σ2(Å2)
3.9(2) 1.771(3) 0.0035(4)
ss
ss
0.6(1) 2.81(2) 0.01b
1.4(4) 3.48(4) 0.01b
-4 ( 2
0.004
pH 7.39 880 µM Γ ) 19.16
CN R(Å) σ2(Å2)
3.9(2) 1.766(3) 0.0031(3)
ss
ss
0.7(2) 2.81(2) 0.01b
1.4(4) 3.47(2) 0.01b
-3 ( 2
0.007
pH 3.19 4.95 mM Γ ) 4.49
CN R(Å) σ2(Å2)
2.9(4) 1.733(7) 0.0052(9)
1.0(4) 1.99(3) 0.01b
0.9(4) 2.49(3) 0.01b
0.3(1) 2.81(2) 0.01b
2.5(4) 3.54(2) 0.01b
-0.3 ( 2
0.093
pH 7.02 4.95 mM Γ ) 1.32
CN R(Å) σ2(Å2)
3.9(2) 1.766(3) 0.0032(4)
ss
ss
0.7(2) 2.81(2) 0.01b
1.4(4) 3.47(2) 0.01
-3 ( 1
0.003
a Goethite suspension density 0.1g/L except for the last two samples which are 15.8 g/L. Initial Mo concentration 110-880 (µM). Γ: Surface coverage (µM m-2). CN: Coordination number. R: Interatomic distance (Å). σ2: Debye-Waller factor (Å2). Fit quality confidence limit for parameters: Mo-O/Fe shells, CN: (20%, R (0.02 Å. b Fixed parameter. Fe1 is an edge-sharing coordination between MoO4 and FeO6 (R: ∼2.8 Å), Fe2 is a corner-sharing bidentate binuclear coordination of MoO4/ MoO6 on FeO6 (R: ∼3.43 Å and 3.53 Å) (See Figure 4).
The raw and fitted k3-weighted χ functions for reference compounds are shown in SI-2. As the XANES data indicated, the Mo(VI) solution sample predominantly contains Mo(VI) tetrahedral molecules (Figure 4a) with Mo-O CN 4.2 ( 0.2 and interatomic distance (RMo-O) ∼1.76 Å. Lidgrenite (Cu3(OH)(MoO4)2) also shows the Mo(VI) tetrahedral arrangement with two Mo-Cu distances at 3.4 and 3.7 Å corresponding to corner-sharing monodentate mononuclear attachment of Mo(VI) tetrahedral on copper octahedral (SI-2). In samples at near-neutral pH (Figure 2), following interatomic distances were first considered based on the XANES data. I assumed that Mo(VI) coordination environment is predominantly tetrahedral. The first Mo-O shells were fit with CN ∼4 at 1.74 Å (Table 1 and Figure 4a). I attempted to fit with an additional Mo-O at 2.2 Å that corresponds to the axial Mo-O distance of Mo(VI) octahedral (Figure 4b). However, the longer Mo-O distance could not be incorporated into the fit, indicating the surface species are predominantly comprised of Mo(VI) tetrahedral molecules. To fit the second and third shells, the interatomic distances of similar tetrahedral molecules (i.e., As(V)) on iron oxyhydroxides were considered (30, 31). Edge-sharing bidentate mononuclear (Figure 4c) and corner-sharing bidentate binuclear species (Figure 4d) were successfully fit at ∼2.8 and ∼3.5 Å. These features in RSF are indicated by black arrows in Figure 2b. The presence of these Fe backscatters is shown in residual EXAFS spectra (i.e., subtracting the fit 8494
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 22, 2010
with Mo-O shells) (SI-3c and d). The molecular configuration of sorbed species is illustrated in Figure 4c and d. I also considered the triangular Mo-O-O-Mo 4 legged paths. When it was included in the fit, the fit quality decreased by ∼30% in reduced χ2 values, suggesting its insignificance. This is probably attributed to the shorter Fourier transform range k
