ATR–FTIR and Density Functional Theory Study of the Structures

Department of Geosciences, The Pennsylvania State University, University Park, Pennsylvania 16802, United ... Cite this:Langmuir 2012, 28, 41, 14573-1...
0 downloads 0 Views 973KB Size
Article pubs.acs.org/Langmuir

ATR−FTIR and Density Functional Theory Study of the Structures, Energetics, and Vibrational Spectra of Phosphate Adsorbed onto Goethite James D. Kubicki,*,† Kristian W. Paul,‡ Lara Kabalan,† Qing Zhu,† Michael K. Mrozik,† Masoud Aryanpour,§ Andro-Marc Pierre-Louis,∥ and Daniel R. Strongin∥ †

Department of Geosciences, The Pennsylvania State University, University Park, Pennsylvania 16802, United States DuPont Crop Protection Stine-Haskell Research Center, Newark, Delaware 19714, United States § Department of Mechanical Engineering, Stanford University, Stanford, California 94305, United States ∥ Department of Chemistry, Temple University, Philadelphia, Pennsylvania 19122, United States ‡

S Supporting Information *

ABSTRACT: Periodic plane-wave density functional theory (DFT) and molecular cluster hybrid molecular orbital−DFT (MO−DFT) calculations were performed on models of phosphate surface complexes on the (100), (010), (001), (101), and (210) surfaces of α-FeOOH (goethite). Binding energies of monodentate and bidentate HPO42− surface complexes were compared to H2PO4− outersphere complexes. Both the average potential energies from DFT molecular dynamics (DFT−MD) simulations and energy minimizations were used to estimate adsorption energies for each configuration. Molecular clusters were extracted from the energy-minimized structures of the periodic systems and subjected to energy reminimization and frequency analysis with MO−DFT. The modeled P−O and P---Fe distances were consistent with EXAFS data for the arsenate oxyanion that is an analog of phosphate, and the interatomic distances predicted by the clusters were similar to those of the periodic models. Calculated vibrational frequencies from these clusters were then correlated with observed infrared bands. Configurations that resulted in favorable adsorption energies were also found to produce theoretical vibrational frequencies that correlated well with experiment. The relative stability of monodentate versus bidentate configurations was a function of the goethite surface under consideration. Overall, our results show that phosphate adsorption onto goethite occurs as a variety of surface complexes depending on the habit of the mineral (i.e., surfaces present) and solution pH. Previous IR spectroscopic studies may have been difficult to interpret because the observed spectra averaged the structural properties of three or more configurations on any given sample with multiple surfaces.



desorption from goethite surfaces,11 thereby affecting phosphate and arsenate behavior in the environment. The adsorption of negatively charged oxyanions such as phosphate also affects the surface charge of environmentally important iron oxide substrates.12 By generating a more negatively charged surface, phosphate adsorption subsequently influences the adsorption of both metal cations and natural organic acids. The latter is particularly important because the adsorption of natural organic matter to iron oxides in soils plays a significant role in soil fertility13 and the global C cycle.14 One estimation by surface titration shows a decrease from 0.622 to 0.026 C m−2 in surface charge at pH 5.5.15 On the surfaces of goethite (α-FeOOH), Cd2+ adsorption is enhanced by the presence of phosphate. 16,17 The enhancement of Cd 2+ adsorption can be attributed to the change in surface charge brought about by phosphate adsorption. Modifying the surface charge by phosphate adsorption has also been proposed to

INTRODUCTION

Phosphate transport and bioavailability in soils are strongly affected by adsorption to aluminum oxides, iron oxides, and 1:1 silicate clays (e.g., kaolinite). Accordingly, a molecular-scale understanding of phosphate adsorption mechanisms has long been sought.1 The most extensively investigated soil mineral, with respect to phosphate adsorption, has been goethite (αFeOOH). Goethite is the most thermodynamically stable and abundant iron oxyhydroxide in soils. The absolute abundance of α-FeOOH in soils is 1 to 5%. However, because of its large specific surface area (approximately 50 to 200 m2 g−1), αFeOOH can represent 50 to 70% of the total surface area of a soil.2 The adsorption of phosphate by iron oxides is critical both in situations where phosphate may be a limiting nutrient for plant growth and where excess fertilization leads to eutrophication.3 Furthermore, in surface waters with suspended solids, soils, and groundwater aquifers, competitive adsorption occurs among the various oxyanion compounds (e.g., carbonate, arsenate, selenate, etc.) commonly present in natural waters.4−11 Because of this competition, arsenate can promote phosphate © 2012 American Chemical Society

Received: December 19, 2011 Revised: September 13, 2012 Published: September 17, 2012 14573

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

is not well understood. In this respect, molecular modeling can provide valuable information, but there is disagreement among published studies. For example, Paul et al.,33 Kwon and Kubicki,28 and Rahnemaie et al.29 performed DFT calculations for various sulfate and phosphate adsorption complexes, respectively. Sulfate was considered to adsorb as the deprotonated bidentate species33 whereas diprotonated bidentate, monoprotonated monodentate, and deprotonated monodentate phosphate were suggested28 as the best fit species to the IR spectra of Persson et al.23 and Arai and Sparks.24 However, Rahnemaie et al.29 concluded that a deprotonated bidentate species is dominant over a broad pH range and low surface loadings whereas a monoprotonated monodentate species is dominant for low pH and high surface loadings. Thus, the DFT results are at odds depending on whether one is mainly fitting the observed IR spectra28 or the isotherm data.29 The question of a bidentate versus monodentate surface complex for phosphate (and arsenate) remains a subject of continued debate. Recently, the traditional perception of a bidentate bridging complex has been challenged by the results of IR and EXAFS measurements.30 These authors reasoned that the Fe−As distance of 3.3 Å can correspond to a monodentate configuration, as the As−Co distance (3.25 Å) does in the structure of pentaaminecobalt(III) arsenate. On the basis of the similarities between Fourier transforms of As EXAFS spectra and that of pentaaminecobalt(III) arsenate, monodentate coordination for adsorbed arsenate on goethite was concluded. Moreover, with decreasing pD values a shift to higher frequencies was observed in the IR frequency of the As−O stretching mode, which does not support a deuterated bidentate species but rather a deuterated monodentate arsenate complex on goethite. These discrepancies may be resolved by employing periodic DFT and molecular cluster calculations. The molecular cluster calculations have proven to be effective for predicting vibrational spectra of many surface complexes, but this approximation allows full relaxation of all atoms such that the calculated energetics of the adsorption reaction will be systematically in error. Periodic DFT calculations include constraints on the surface complex imposed by the crystal structure, which will allow for a more realistic calculation of the energetics. This technique is computationally more expensive, so it has not been extensively tested for its ability to reproduce observed vibrational spectra of mineral surface complexes. A combination of the two approaches is used here to provide more accurate thermodynamics and structures (periodic models) and vibrational spectra (cluster models). To achieve a more comprehensive understanding of phosphate adsorption on goethite, it is vital to base interpretations on the actual surface properties of goethite particles. The macroscopic adsorption behavior of goethite particles is determined by the different reactivities of its crystalline surfaces. This difference in reactivity has been shown and emphasized recently by Villalobos et al.34,35 for the adsorption of protons, carbonate, chromate, and lead ions on three goethite nanoparticles having specific surface areas of 50, 70, and 94 m2 g−1. According to their model, a combination of (101) and (001) faces can describe the reactivity of goethite nanoparticles with high specific surface areas (SSAs). The (010) and (210) faces, in addition to the (101) and (001) faces, were required to describe the reactivity of goethite nanoparticles with low SSAs. Consequently, the differences in interpretation proposed in the papers cited above may be due

affect the adhesion of bacteria to Fe-coated surfaces, or the phosphoryl group in bacterial extracellular polymeric substances may bond directly with iron oxyhydroxides.18 To predict the adsorption behavior of compounds such as phosphate under experimental and field conditions, surface complexation modeling has been commonly employed. One such model is the multisite complexation (MUSIC) model of Hiemstra and co-workers.19 MUSIC utilizes surface-specific bond distances in order to predict adsorption edge and isotherm behavior; thus information on adsorption concentrations and structures is required in order to apply this modeling approach. Surface complex structural information is generally obtained via spectroscopic (e.g., attenuated total reflectance Fourier-transform infrared − ATR−FTIR) and Xray scattering (e.g., extended X-ray adsorption fine structure − EXAFS) methods often complemented by computational chemistry.20−22 Phosphate adsorption onto goethite has been discussed extensively because the vibrational spectra result in relatively broad bands that can be difficult to interpret. Further complicating the picture is the fact that both the number of bonds to the goethite surface and the protonation state of the adsorbed phosphate can be called into question.23,24 To the best of our knowledge, EXAFS results for phosphate adsorption to iron oxides have not been published. Weak X-ray scattering by P leads to EXAFS spectra with poor signal-to-noise ratios. For this reason, arsenate EXAFS spectra30 have been used as an analog for phosphate because these two oxyanions have similar chemical behavior. Infrared spectra of phosphate adsorbed onto goethite have been collected by a number of researchers.1,23−27 Studies of phosphate adsorption on α-FeOOH have resulted in numerous conflicting interpretations of binding mechanisms. Often a bidentate configuration has been assigned to the phosphate− goethite surface complex,25 but monodentate configurations have been proposed as well.23 For example, Tejedor−Tejedor and Anderson27 and Luengo et al.31 investigated the adsorption of phosphate at the α-FeOOH−water interface. Tejedor− Tejedor and Anderson27 proposed that phosphate formed protonated bidentate bridges, deprotonated bidentate bridges, and deprotonated monodentate complexes as a function of pH and phosphate concentration. Luengo et al.31 investigated phosphate adsorption and proposed that phosphate formed protonated and deprotonated bidentate bridging complexes as a function of pH. The specifics of monodentate versus bidentate also depend upon the pH under which the phosphate was adsorbed23,26 and the hydration conditions under which the IR spectra were collected.25 Spectroscopic studies such as those cited above have been used to parametrize surface complexation models. For example, Rahnemaie et al.29 used a charge distribution (CD) approach combined with density functional theory (DFT) calculations to predict that nonprotonated bidentate and singly protonated monodentate surface complexes form. Surface complexation models are fit to adsorption isotherm data, but the reactions used in the fitting can be constrained by spectroscopic information. The application of multisite models to adsorbed phosphate can be used to predict the macroscopic behavior of phosphate in the environment.32 Despite the advances achieved in our understanding of the surface complex structures of phosphate on α-FeOOH, uncertainty remains regarding the adsorption mechanisms of phosphate under more complex conditions. In particular, the competitive adsorption of phosphate, for example, with sulfate, 14574

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

Transmission Electron Microscopy (TEM). Transmission electron micrographs were collected using a JEOL JEM-1400 instrument equipped with a Gatan Dual Vision digital charge-coupled device (CCD) camera and high-brightness LaB6 electron source. Samples were prepared for TEM analysis by first sonicating a goethite (0.1 mg) suspension (for homogeneity) in DI water (1 mL) for approximately 10 min. A drop of the dispersed particle suspension was then placed on an amorphous holey-carbon film supported by a copper-mesh TEM grid. The goethite sample was dried on the grid support in air at room temperature (approximately 30 min). X-ray Powder Diffraction (XRD). Goethite samples were characterized by X-ray diffraction (XRD) using an Apex Duo Bruker Instrument Service (Mo Kα radiation) equipped with a CCD diffraction system. Samples were scanned at a wavelength of λ = 0.71073 Å (Mo Kα X-ray source) and at nearly constant irradiation volumes in the 2θ range of 5−50° using a step size of 0.02° (0.40 s/ step). Copper (Cu Kα) source radiation was not used in order to avoid interference from Fe fluorescence. Attenuated Total Reflectance−Fourier Transform Infrared (ATR− FTIR) Spectroscopy. A Nicolet 6700 spectrometer (Thermo Scientific) equipped with a smart Orbit attenuated total reflectance Fourier transform infrared (ATR−FTIR) accessory and a deuterated triglycine sulfate (DTGS) detector was used to collect vibrational spectra. Goethite samples to be analyzed were placed on a diamond ATR element in the form of a suspension and were dried with a stream of N2 to form a thin film. The film was rinsed with deionized water (DI, ∼10 mL) containing 10 mM NaCl to maintain the background electrolyte and to eliminate loosely bound goethite particles. Depending on the experiment, the reference water solution flowed across the goethite film at a specific pH value (4.22, 5.71, 7.51, or 7.96). Then, a 100 μM phosphate solution at the pH of interest (4.22, 5.71, 7.51, or 7.96) flowed (via syringe pump) through a reaction cell that enclosed the goethite film. The flow of solution was maintained at ∼0.4 mL/min. ATR−FTIR was carried out in situ, and spectral results for adsorbed phosphate on goethite nanoparticles were either reprocessed or subtracted from a reference spectrum of H2O that flowed over the same goethite thin film sample at the same pH. All reported spectra were manually baseline corrected in the 1200−800 or the 1200−900 cm−1 region by the magnitude of the strong goethite mode contributions near 889 and 796 cm−1. Nanogoethite and microgoethite (each sample having a mass of 0.216 g) were individually added to 30 mL of a 100 μM phosphate solution that also contained 10 mM NaCl. The phosphate/goethite suspensions were sonicated for 3−5 min to disperse the particles in solution. The desired pH condition for a particular experiment was adjusted with an appropriate volume of NaOH or HCl under vigorous stirring conditions. Phosphate/goethite suspensions were left to preequilibrate for 30 min at the specific pH of interest (i.e., 4.22, 5.71, 7.51, or 7.96) using a 718 STAT Titrino Metrohm (Brinkmann) pH meter. After equilibration, an aliquot of the phosphate/goethite suspension was centrifuged, and the supernatant was decanted, isolating the goethite particles. Before drying, all phosphate/goethite samples were rinsed three times with deionized H2O at the same pH. In these experiments, ATR−FTIR spectra were collected ex situ. Samples were individually placed on the diamond ATR element and dried with a stream of N2. In all experiments, ATR−FTIR data were collected at a resolution of 4 cm−1 for 200 coadded scans. All collected spectra were reprocessed against the clean ATR diamond element and manually baseline corrected in the phosphate region. The Omnic 7.3 software program was used for data acquisition, processing, and peak deconvolution procedures. ATR−FTIR spectra of aqueous phosphate were collected by using a Smart Orbit ATR (Thermo Scientific) diamond accessory and a Nicolet 6700 spectrometer (Thermo Scientific) equipped with a DTGS detector. Aqueous-phase IR spectra of sodium phosphate (Thermofisher analytical grade) at given pH values were collected by loading 2 drops of previously pH-adjusted 25 mM sodium phosphate solution on a diamond ATR element. In all of the FTIR experiments, data were collected as a single beam (resolution 4 cm−1) of 100

to changes in the goethite surfaces responsible for adsorbing phosphate in any given experiment (e.g., pH, hydration state, etc.). The goal of this study is to identify the most probable phosphate−goethite surface complex structures. To achieve this goal, ATR−FTIR spectra have been collected as a function of pH for two goethite samples of different sizes and habitats, and two computational approaches have been used: (1) periodic DFT molecular dynamics simulations and energy minimizations to estimate adsorption energies and (2) molecular cluster energy minimizations and frequency calculations to obtain IR-active vibrational frequencies and modes. The periodic DFT calculations are valuable for estimating surface complex structures and adsorption energies because they explicitly include details of the mineral surface and solvation effects by H2O molecules. Structures obtained by the periodic DFT calculations were used as a starting point for molecular cluster calculations to predict IR and Raman spectra for comparison with observed spectra. We argue that by modeling the structure, energy, and vibrational spectra of phosphate on five surfaces of goethite in a self-consistent manner, more definitive assignments to the vibrational spectra should be possible. Furthermore, once the correct phosphate surface complexes have been determined, surface complexation modeling can be applied in a more effective manner to predict macroscopic phosphate behavior in the environment. An understanding of how phosphate adsorbs to goethite can then be used to better understand how competitive adsorption, nucleation, growth, and phase transformation are affected by phosphate and similar oxyanions.



METHODS

Experimental Section. Synthesis and Solution Preparation. Both nanometer- and micrometer-sized goethite samples were synthesized by using a modified method adapted from Cornell and Schwertmann.36 To synthesize nanogoethite, a ferrihydrite precursor was first precipitated by adding 180 mL of 5 M NaOH to 100 mL of Fe(NO3)3·9H2O solution (0.1 M) at a pH of approximately 12.5. The resulting suspension was diluted to 2 L with deionized water and held in a closed Pyrex bottle in a 70 °C oven for 60 h. During the aging process, the red-brown ferrihydrite suspension transformed into a yellow, nanosized goethite precipitate. Micrometer-sized goethite was synthesized by first adding 50 g of Fe(NO3)3·9H2O to 830 mL of double-distilled water. Five molar (5 M) NaOH was added to the solution until a pH of ∼12 was achieved. The resulting suspension was placed in a sealed Pyrex bottle and aged in an oven at a temperature of 60 °C for a period of 24 h. Goethite from each synthesis protocol was dialyzed in deionized water (DI) for 5−10 days with frequent water changes to remove counterions (Na+ and NO3−) from the suspensions. Goethite sample suspensions were subsequently centrifuged, air dried, and individually stored in polyethylene bottles. The dried samples were crushed to produce finely divided goethite powders. All stock phosphate solutions were prepared from sodium dibasic anhydrous (NaH2PO4·H2O) purchased from Sigma-Aldrich. Characterization Techniques. Surface Area Measurement (BET). Specific surface areas of the samples were determined by the BET technique using N2 as the adsorption gas. Samples were degassed (i.e., desorbing residual superficial water) at 150 °C for 500 min prior to any BET measurements. The specific area was found to be 139.1 ± 0.8 m2 g−1 for nanogoethite and 67.6 ± 0.3 m2 g−1 for microgoethite using N2 BET surface area analysis. Villalobos et al.35 have reported similar BET values for synthesized goethite samples. 14575

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

Figure 1. Monodentate and bidentate HPO4− and outer-sphere H2PO4− near a (101) goethite surface.

Figure 2. Monodentate and bidentate HPO4− and outer-sphere H2PO4− near a (210) goethite surface. given protonation state (i.e., HPO42− for the inner-sphere and H2PO4− for the outer-sphere configurations). The five periodic systems studied consisted of Fe24O83H89P in (100) 9.03 × 9.24 × 22.04 Å3, (010) 9.24 × 9.95 × 20.00 Å3, (001) 9.95 × 9.03 × 20.46 Å3, (101) 10.97 × 9.03 × 18.56 Å3, and (210) 9.24 × 11.63 × 17.11 Å3. The magnetic ordering of the Fe atoms in the goethite slabs was set according to the experimentally observed antiferromagnetic pattern41 as in Kubicki et al.42 Projector-augmented plane-wave calculations43,44 were performed using VASP 5.2,45 the generalized gradient approximation (GGA) PBE pseudopotentials,46 and Vanderbilt-type ultrasoft pseudopotentials47 with an energy cutoff of 400 eV for the MD simulations and 500 eV for energy minimizations. The electron density cutoff was 1 × 10−4 eV. Partial occupancies were determined using the first-order scheme of the Methfessel−Paxton method with a 0.1 eV width. Initial magnetic moments of Fe atoms were assigned as alternate positive and negative values (±5) on the crystalline layers along the Miller direction [010] so as to obtain an antiferromagnetic goethite slab in each case. The onsite Coulomb interaction, the GGA+U method, was applied to the Fe atoms according to the approach of Dudarev et al.48 as implemented in VASP. The supercell lattice parameters were held fixed throughout the

coadded scans and the water spectrum was subtracted from each individual spectrum to get the final spectrum. Computations. Models of the H3PO4−goethite−water system were created by cleaving the respective surfaces(100), (010), (001), (101), and (210)34,35from the experimental crystal structure of bulk goethite (space group Pnma33) through the plane breaking the fewest number of Fe−O bonds. The resulting surfaces were consistent with previously studied surfaces when available.37,38 Periodic models of the surfaces were created with a vacuum space of approximately 10 Å between repeating goethite surfaces. HPO42− was added to the surface in monodentate and bidentate configurations as guided by previous results on molecular clusters.28 Outer-sphere configurations were created with H2PO4−, assuming the exchange of one H+ when forming (P)O−Fe bonds. Thirty-one H2O molecules were added to the system in order to mimic a density of approximately 1 g cm−3. Two H+ ions from H3PO4 were added to random H2O molecules in the model to form H3O+ ions that kept the simulation cells electrostatically neutral (Figures 1−5). Thus, the system is formally at a low pH (i.e., approximately 1.5), but all atoms are allowed to relax and H+ transfers are extremely rapid in this type of DFT−MD simulation.40 These model systems should therefore be representative of phosphate in its 14576

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

Figure 3. Monodentate and bidentate HPO4− and outer-sphere H2PO4− near a (010) goethite surface.

Figure 4. Monodentate and bidentate HPO4− and outer-sphere H2PO4− near a (001) goethite surface. simulations. A single k point was evaluated at the Γ point of the supercell, which provides acceptable energy results for such relatively large models. Partial energy minimizations were performed to lower the potential energy of the randomly added H2O molecules before the MD simulations. MD simulations were conducted for at least 6000 time steps (3 ps with a time step of 0.5 fs) and the Nosé49 thermostat (SMASS = 0.028) to control the temperature to 300 K. The average and standard deviations of the MD energies were extracted from the last 1000 steps (0.5 ps) of the simulations. The final configuration of the MD simulations was subjected to energy minimization. The ionic relaxations were considered to be converged when the change in the free energy of the system between two steps dropped below 2 × 10−2 eV Å−1. Upon completion of the DFT−MD simulations and energy minimizations, molecular clusters were extracted that represented the monodentate and bidentate configurations on each surface. Each cluster was created by selecting all atoms either covalently or H bonded to the central Fe−O−P surface complex. The clusters were terminated by OH− and H2O groups bonded to the Fe atoms in order to eliminate dangling bonds and maintain overall charge neutrality. Energy minimizations were performed using Gaussian 0350 with all Fe

atoms initially frozen and then allowed to relax. This stepwise energy minimization helps to ensure that the cluster model retains structural similarity to the original periodic surface configuration. The B3LYP exchange-correlation functional51,52 was used with the 6-31G(d,p) basis set.53 Upon completion of the energy minimizations, frequency calculations were conducted. The model harmonic frequencies were scaled by 0.96 before comparing with observation in order to account for anharmonicity, basis set effects, and approximations in electron correlation. (Note that this is an approximate scaling factor based on the Scott and Radom54 value for B3LYP/6-31G(d) of 0.9614.) In addition, because adsorption may occur at defect surface sites, clusters built on the model Fe2(OH)6(OH2)4 structure were also constructed and were subjected to energy minimization and frequency analyses as described above. Monodentate and bidentate phosphate with various protonation states were constructed with (eight H2O molecules) and without explicit H2O molecules of hydration. The calculated frequencies for these models were also compared to observed frequencies for both the wet and dry samples. 14577

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

Figure 5. Monodentate and bidentate HPO4− and outer-sphere H2PO4− near a (100) goethite surface.

Figure 6. XRD patterns for (a) microgoethite and (b) nanogoethite.



RESULTS AND DISCUSSION

To visualize the morphology of the two different synthesized samples, TEM was used to characterize the goethite of varying crystallinity that was observed from XRD analysis. Representative TEM images for the synthesized microgoethite and nanogoethite are shown in Figure 7. The TEM images show acicular nanorods with lengths measured to be 100−200 nm and widths of ca. 70 nm for nanogoethite (Figure 7a,b), whereas more polycrystalline microrods of goethite have lengths of between approximately 200 and 900 nm and ca. 50 nm width (Figure 7c,d). Nanogoethite exhibited a rough, irregular, starlike surface whereas microgoethite had narrow, rectangular needle crystalline rods with well-ordered smooth edges. In addition, the nanogoethite structure formed starlike (aggregates), poorly crystalline arrangements that may contain a small number of amorphous nanoparticles lying on top of the crystalline nanorod goethite phases contributing to a broader

Characterization of the Synthesized Goethite Samples. Figure 6 shows the XRD patterns for the microgoethite and nanogoethite synthesized materials. The microgoethite (Figure 6a) shows more intense Bragg diffraction lines with some slight differences in the sharpness of the d-spacing values as compared to the nanogoethite phase (Figure 6b). The XRD patterns are similar to those previously reported in the database of the International Center for Diffraction Data (ICDD). The most notable difference between the patterns is displayed between the 5 to 25 2θ° regions showing a slight shift in the diffraction lines from 15 to 13.5 2θ°. As a result of the synthesis methods, the degree of crystallinity is higher in microgoethite (67.6 ± 0.3 m2 g−1) than in nanogoethite (139.1 ± 0.8 m2 g−1) as deduced from the respective XRD patterns. 14578

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

Figure 7. TEM images of (a) acicular crystalline and (b) nanorods of goethite with measured lengths of approximately 237 nm. TEM images of (c) needles and (d) single microrods of goethite with measured lengths of approximately 900 nm. Note the more irregular habits of nanogoethite compared to microgoethite.

spectral shoulders at 966, 932, and 884 cm−1 (Figure 8c). At pH 7.96, the associated fitted spectrum shows modes at 1124, 1082, 1036, 1006, 966, 935, 887, and 865 cm−1 (Figure 8d). The IR bands experimentally observed in this study are similar to prior vibrational modes of adsorbed orthophosphate complexes on goethite (FeHPO4+) reported in a study by Tejedor−Tejedor and Anderson27 (Table 1). Tejedor−Tejedor and Anderson27 assigned modes at approximately 1120 and 1001−1088 cm −1 to the ν PO stretching modes of a monodentate protonated phosphate complex in the pH range of 3.6−6.0. Furthermore, the possible presence of a FeH2PO42+ complex at the lower end of the pH range had also been postulated in their study. In their study, the νPO stretching modes observed near 1089, 1044, and 945 cm−1 were assigned to either the nonprotonated bidentate (FeO)2 PO 2 or monoprotonated monodentate (FeO)2(OH)PO2 complexes at a higher pH of 9.2.31 These assignments will be discussed below (Extended Cluster Vibrational Frequencies Section). Microgoethite. Using a similar experimental procedure, we investigated the adsorption of phosphate at various pH values on microgoethite with ATR−FTIR. The resulting spectra are illustrated in Figure 9 in the range of 1200−900 cm−1. The fitted spectrum associated with adsorbed phosphate on microgoethite at the lower pH of 4.22 exhibits vibrational modes at 1157, 1122, 1091, 1044, 1009, and 982 cm−1 (Figure 9a). An increase in the pH to 5.71 resulted in the appearance of

and lower intensity in the Bragg diffraction reflection lines (Figure 6b). Attenuated Total Reflectance Fourier Transform Infrared (ATR−FTIR) Spectroscopy. Table 1 lists the phosphate-associated frequencies observed on goethite samples from Tejedor−Tejedor and Anderson,27 Persson et al.,23 Arai and Sparks,24 and this study. SI Table 1 lists IR and Raman frequencies for aqueous phosphate at various pH values. ATR− FTIR spectra of aqueous phosphate were collected at pH 4.22, 5.71, 7.51, and 7.96 and are illustrated in SI Figure 1. Nanogoethite. Figure 8 exhibits in situ ATR−FTIR results for the adsorbed phosphate on the nanogoethite phase for the 1200 to 800 cm−1 region. Note that surface loading decreases as the pH increases, necessitating greater amplification of the IR intensity in the spectra of the higher-pH samples. The ATR− FTIR of adsorbed phosphate vibrational modes on either goethite phase exhibit broad bands. The fitted pH 4.22 spectrum shows three main bands at 1084, 1044, and 1005 cm−1. The pH 4.22 IR spectrum also indicates the presence of significant peaks at 1176, 1117, 957, 876, and 857 cm−1 (Figure 8a). For the pH 5.71 spectrum (Figure 8b), the peak deconvoluted modes appeared mainly at 1122, 1089, 1039, and 1004 cm−1 with additional spectral intensity at 962, 932, and 876 cm−1 illustrated in Figure 8b. As the pH increased to 7.51, the associated fitted spectrum exhibited adsorbed phosphate modes at 1128, 1083, 1038, and 1009 cm−1 and 14579

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

Table 1. ATR−FTIR Frequencies of Adsorbed Phosphate at 100 μM on Different Nanogoethite and Microgoethite Phases at Various pH Values For Phosphate under Wet Sample Conditions 4.0 pH

Tejedor27

4.22 (micro)

4.2−5.7

(nano) 857 876

939

Persson23 876 939

957 1002

982 1009

1041

1044

1098 1120

1091 1122 1157 1195

1005

1001

5.71 (micro)

(nano)

941 954

876 932 962

1012

1004

1044 1084

1049

1043

1039

1117 1176

1122 1178

1090 1120 1164

1089 1122 1176

6.0

7.5

Tejedor27

Arai24

(micro)

(nano)

Persson23

(micro)

887

903 923 946

823 884 932 966

900 939

1009

1001

915 938 952 971 1005

952 996 1025 1045

1021

1097

1088

7.51

7.9

1022 1043 1084

1049

1124 1177

7.96

1039 1076 1095

1122

8

(nano)

Tejedor27

886 887 935 966 1006 1036 1082

1001 1023 1045 1088

1124 1129

For a 0.1 mM Dried Phosphate Sample after Adsorption 4.22 pH

(micro) 1003 1032 1059

5.71 (nano) 997 1017 1044

1091

1087

1107 1123 1138

1119

1155

1142 1156 1173

(micro) 963 1002 1033 1052 1068 1089 1109 1123 1137 1151 1172

1176

7.51 (nano) 1005 1030 1062 1083 1100 1113 1128 1143 1158 1182

1191 1198 1224

1202 1216 1249

1202

1235

(micro) 994 1017 1043 1058 1077 1087 1098 1110 1123 1135 1143 1155 1169 1179 1192 1204 1221 1254

7.96 (nano)

(micro)

962 1000 1022

965 998 1017

1068 1088 1099 1109 1122 1137

1050 1066 1087 1104 1121 1136

1156

1155

1175

1176

1223 1245

1200 1225 1244

(nano) 992 1017 1041 1057 1077 1087 1098 1112 1133 1144 1156 1170 1182 1192 1203 1223 1250

modes at 1120, 1090, 1043, and 1012 and two weak modes at 954 and 941 cm−1 (Figure 9b). The fitted IR spectrum of adsorbed phosphate at a nearly neutral pH of 7.51 exhibited relatively strong vibrational bands at 1084 and 1043 cm−1 with a mode fitted at 1022 cm−1 and weak bands at 1124, 946, 923, and 903 cm−1 (Figure 9c). As the pH increased to 7.96, the associated deconvoluted spectrum exhibited modes at 1095, 1076, 1039, 1005, 971, and 938 cm−1 and relatively weak peaks at 1160, 1129, and 915 cm−1 (Figure 9d). Although the phosphate modes associated with nanogoethite and microgoethite show similar trends with pH, a direct comparison shows that there are differences in the positions of modes for the two systems (Table 1). For example, at pH 4.22 phosphate exhibited modes at ∼1117, 1084, and 957 cm−1 when adsorbed on nanogoethite, whereas on microgoethite these modes are observed at ∼1157, 1091, and 982 cm−1 (Table 1). As the pH increased above 4.22 to 7.96, the differences in the location of the vibrational modes associated with phosphate adsorbed on nanogoethite and microgoethite persisted. These frequency variations between the two samples

Figure 8. Fitted ATR−FTIR data for adsorbed phosphate (100 μM) on nanogoethite at pH (a) 4.22, (b) 5.71, (c) 7.51, and (d) 7.96.

14580

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

modes on the microgoethite surface were not greatly affected as the pH was increased from 4.22 to 7.96, as illustrated in Figure 10a−d. However, the intensity of these phosphate modes decreased on the surface as the pH increased most likely because of less adsorption of phosphate at the higher pH. (Note the increasing scale factors in Figure 10.) To isolate the spectral features of adsorbed phosphate better, deconvolution of the spectra was carried out, and these analyses revealed modes near 1087, 1066, 1050, 1017, and 965 cm−1 as the pH was increased to 7.96. These changes with pH suggest different phosphate species coordinated as a function of pH. These other species also increase in concentration with increasing pH (Figure 10b−d). Tejedor−Tejedor and Anderson27 observed three set of bands under their experimental conditions. Their mode positions at 1123, 1006, and 982 cm−1 were assigned to either (FeO)2 (OH)PO monodentate protonated or to (FeO)(OH)2PO diprotonated bidentate complexes in the pH range of 3.6−6. The modes found at 1096 and 1044 cm−1 were assigned to the PO2 group and were attributed to either nonprotonated bidentate or the monoprotonated monodentate complexes in the pH range of 6 to 8.4. Their 1025 and 1001/998 cm−1 band assignments were correlated to the monodentate deprotonated complex at a pH above 9.6. Although these bands have been assigned to give information on the protonation/deprotonation of phosphate complexes on an identical crystalline goethite surface that could contain multiple surfaces, our ATR−FTIR results differ slightly and show that under a lower phosphate concentration the adsorption bands at any of the given experimental pH values are mostly indistinguishable from each other. Adsorbed phosphate species on the nanogoethite sample after drying are shown in Figure 11. The modes illustrated at

Figure 9. Fitted ATR−FTIR data for adsorbed phosphate (100 μM) on microgoethite at pH (a) 4.22, (b) 5.71, (c) 7.51, and (d) 7.96.

are probably due to differences in the crystalline goethite habits associated with nanogoethite and microgoethite. Prior studies, for example, have shown that the surfaces of high-specific-area (SA) goethite nanoparticles are mostly composed of (101) and (001) faces, whereas the surface of goethite with low SA contains an additional high degree of (010)/(210) faces (Pnma space group) with high surface density.34,35 Hence, although we were not able to characterize the surfaces present on the samples used in this study, the vibrational shifts between adsorbed phosphate modes on nanogoethite and microgoethite at a given pH are at least in part due to differences in the crystal habit between the two samples. Dried Samples. Figure 10 displays ATR−FTIR spectra of adsorbed phosphate on a microgoethite surface after being dried with a stream of N2. At pH 4.22, adsorbed phosphate exhibited strong modes at 1176, 1123, and 1003 cm−1 (Table 1 and Figure 10a). The positions of the adsorbed phosphate

Figure 11. Fitted ATR−FTIR data for adsorbed phosphate (100 μM) on nanogoethite at pH (a) 4.22, (b) 5.71, (c) 7.51, and (d) 7.96 after drying.

1119 and 1017 cm−1 appeared at pH 4.22 (Figure 11a). As the pH increased from 4.22 to 7.96, the position of the adsorbed phosphate modes changed slightly as shown by the peak fittings of additional small shoulder vibrational modes. Overall, the main positions of the 1123 and 1017 cm−1 modes were present in all of the spectra under all pH conditions. Furthermore, the spectral results showed the development of modes near 1087,

Figure 10. Fitted ATR−FTIR data for adsorbed phosphate (100 μM) on microgoethite at pH (a) 4.22, (b) 5.71, (c) 7.51, and (d) 7.96 after drying. 14581

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

Table 2. Comparison of Site Densitiesa for Goethiteb Faces Considered in This Study and in Villalobos et al.35 c surface site

(101)

(210)

(010)

(001)

(100)

Fe−OH or Fe−OH2 Fe2OH Fe3O or Fe3OH

3.03 (3.03) 3.03 (0.00d) 6.06 (3.03)e

7.44 (7.50) 7.44 (3.70)e 0.00 (0.00)

8.70 (9.10) 8.70 (4.55d) 0.00 (0.00)

3.34 (3.34) 3.34 (0.00d) 3.34 (3.34)

7.19 7.19 0.00

a Number of sites nm−2. bSpace group Pnma. cIn parentheses. dConsidered to be nonreactive in ref 35. eHalf of the surface is considered to be nonreactive in ref 35.

Table 3. P−O and P−Fe Distances (Å) in Goethite−Phosphate Surface Complexes on Various Surfacesa configuration P−O monodentate bidentate P---Fe monodentate bidentate

(101)

(210)

(010)

(001)

(100)

1.56 (1.56) 1.56 (1.58)

1.56 (1.57) 1.56 (1.55)

1.56 (1.57) 1.56 (1.57)

1.56 (1.56) 1.56 (1.59)

1.56 (1.56) 1.56 (1.57)

3.25 (3.20) 3.27 (3.21)

3.55 (3.29) 3.25 (3.19)

3.42 (3.38) 3.18 (3.13)

configuration

MDH0

P−O P---Fe

1.57 3.25

3.51 (3.32) 3.38 (3.33) 3.26 (3.22) 3.45 (3.37) Iron−Hydroxide Dimers with Eight H2O Molecules of Hydration MDH1

MDH2

BDH0

1.58 1.57 1.57 3.16 3.36 3.16 Iron−Hydroxide Dimers with No H2O Molecules of Hydration

BDH1

BDH2

1.57 3.22

1.57 3.19

configuration

MDH1

MDH2

BDH1

BDH2

P−O P---Fe

1.58 2.62

1.57 3.19

1.57 3.05

1.57 3.19

a

The numbers for the (101), (210), (010), (001), and (100) surfaces without parentheses are obtained from VASP energy minimization, and the numbers in parentheses are calculated from Gaussian structure optimization. All the values for the iron-hydroxide dimers were energy minimized using Gaussian 03.50

1057, 1041, and 992 cm−1 with additional modes around 1250 and 1182 cm−1 as the pH increased to 7.96 (Figure 11b−d). The mode at 1250 cm−1 has been assigned to the ν(C−O) stretching mode that is due to an increase in adsorbed carbonate species above pH 7 resulting from the reaction of the sample with atmospheric CO2,57 but this could also be due to a P−O−H bending mode. Similar to the microgoethite surface, the nanogoethite surface exhibits no significant changes in phosphate vibrational band positions as the pH increases. However, the nanogoethite surfaces did exhibit some differences from the microgoethite surfaces, notably the asymmetrical adsorbed phosphate mode on the microgoethite crystalline surface at 1176 cm−1 at high pH, compared to the analogous mode at 1123 cm−1 associated with the nanogoethite surface. The symmetrical adsorbed phosphate mode centered at 1000 cm−1 on the microgoethite surface also showed a shift to 1017 cm −1 , which is approximately 17 cm−1 higher than the analogous phosphate mode associated with nanogoethite. These results suggest that the structures of the phosphate complexes are sensitive to the goethite crystallographic habit on which they adsorb. Differences in the microscopic structure of the two different goethite surfaces play a crucial role in the binding of the adsorbed phosphate species. Persson et al.23 found similar results in a prior phosphate/goethite study (Table 1). At low pH (3 to 4), two main bands at 1178 and 1001 cm−1 dominated their phosphate/goethite vibrational data whereas at higher pH (10.8 to 12.8) the 1120 cm−1 band disappeared and two additional modes appeared at 1057 and 966 cm−1. Their findings agree well with those of Tejedor−Tejedor and Anderson27 as a function of pH. Although the cited prior works investigated phosphate adsorption complexes on goethite that

could contain multiple surfaces, our study shows that adsorbed phosphate on two different goethite samples leads to a variety of surface complexes depending on the crystal habits of the mineral. The changes in our IR spectra (Figures 10 and 11) with pH for the adsorbed phosphate complexes on the microgoethite and nanogoethite surfaces suggest that several phosphate species coexist. Again, it should be pointed out that the IR vibrational shifts for phosphate on the different samples are likely to be more strongly related to the presence of different adsorbed phosphate surface complexes due to structural variations between the goethite. The potential differences between surfaces present on nanogoethite and microgoethite will be examined in the Discussion Section below in light of the calculated frequencies for each surface model. Calculated Structures. The site density of terminal O atoms (i.e., O atoms bonded to only one Fe atom) is thought to control the adsorption of oxyanions such as chromate34,35 and, by inference, phosphate. In Table 2, the calculated site densities of FeO, Fe2O, and Fe3O atom types are compared to the values presented by Villalobos and co-workers. The values are in agreement with one another when one notes that Villalobos and co-workers assumed certain sites to be nonreactive and were not counted in their studies. A small discrepancy of 0.4 sites nm−2 exists for the Fe2OH sites on the (010) surface, which can be considered to be insignificant. The P−O and P---Fe distances for the periodic inner-sphere complexes on each goethite surface are listed in Table 3. The average P−O distance is 1.56 Å (1.57 to 1.58 Å for the iron hydroxide dimer models) regardless of which crystal face was modeled or whether phosphate was monodentate or bidentate. However, the range of P−O distance varies from 1.52 Å for PO bonds to 1.64 Å for P−OH bonds. Distinguishing among 14582

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

Table 4. Calculated Adsorption Energies (ΔEads) Relative to Outer-Sphere Configurations in kJ mol−1 with Standard Deviations from the MD Simulations at 300 K in Parenthesesa

a

configuration

(101)

(210)

(010)

(100)

(001)

monodentate (MD) monodentate (EM) bidentate (MD) bidentate (EM)

−46 (47) −44 −68 (52) −70

−73 (45) −90 −5 (50) −2

37 (45) 39 7 (46) −13

−14 (40) −44 −57 (42) −67

−54 (42) −50 51 (49) 51

ΔEads = E(outer sphere) − E(inner sphere).

disadvantages of this method are that fluctuations in the potential energy cause precision problems with the potential energy value and that smaller energy cutoffs (i.e., ENCUT = 400 eV) result in lower accuracy compared to energy minimizations. These problems are lessened by using the minimum potential energy obtained at 0 K with a larger energy cutoff (i.e., ENCUT = 500 eV), but these static structures may not represent reality as well as the configurations sampled via the DFT−MD simulations. (SI Figure 2 illustrates the convergence of the average energy from the MD simulations after approximately 4000 steps or 2 ps.) Although generally one cannot guarantee that the most stable energy is reached after such a short relaxation period, given the size of the model system and available computer resources these simulations represent reasonable attempts to relax the models away from the initial configurations. Consequently, we have used both approaches and compared the results for consistency. Table 4 lists the calculated adsorption energies of monodentate and bidentate surface complexes relative to the outer-sphere configuration for each crystal face. In general, the energy-minimized (EM) and DFT−MD adsorption energies are consistent in that the crystal faces with thermodynamically favorable adsorption energies as predicted with EM are also predicted to be favorable with DFT−MD. On the basis of these results, each surface is predicted to be capable of adsorbing phosphate as an inner-sphere complex, as either monodentate (i.e., (001) and (210)), bidentate (i.e., (010)), or both (i.e., (101) and (100)). (Note that the expected accuracy of the calculations is on the order of ±10 kJ mol−1, so we do not distinguish among ΔEads values in Table 4 that differ by less than this amount.) Overall, the most energetically favorable model configuration under these conditions is the monodentate (210) with ΔEads = −90 kJ mol−1 (Table 4). Other calculated ΔEads values fall in the range of approximately −10 to −70 kJ mol−1. Variations in the average DFT−MD values are even larger. Furthermore, the calculated adsorption energies in Table 4 do not consider entropy changes with adsorption. These values are in the range of enthalpies of adsorption of phosphate onto iron hydroxides55 and arsenate onto aluminum hydroxides,56 which should be reasonably analogous systems. There are some quantitative discrepancies among the calculated adsorption energies between the EM and DFT− MD methods. For example, in some cases the configurations determined with EM have ΔEads values that are significantly lower than the DFT−MD-predicted ΔEads values (Table 4 − monodentate (100) and (210)). Generally, however, the MDand EM-predicted adsorption energies are similar. We emphasize that both methods predict the same result with respect to whether the reactions are exothermic or endothermic except for the bidentate (010) complex, which has a small positive or negative value for the DFT−MD and EM methods (Table 4); these two values are not significantly different from

various P−O bond lengths may be possible using EXAFS, as was carried out previously for arsenate,20 but P is a relatively light element and difficult to quantify with EXAFS measurements. Consequently, we use the arsenate EXAFS data as an analog because phosphate and arsenate have similar pKa values and structures and compete for the same sites on goethite surfaces. As−O distances of 1.62 and 1.71 Å have been observed for adsorbed arsenate20 that suggest a short AsO bond and longer As−OH and/or As−O(−Fe) bonds. The analogous PO and P−OH bonds in aqueous [HPO4]2− are 1.55 and 1.66 Å, respectively, so the model surface PO and P−OH bond lengths are consistent with the observed values for adsorbed arsenate. The calculated periodic DFT P---Fe distances fall within the range of 3.05−3.45 Å for bidentate complexes and 2.62−3.55 Å for monodentate complexes (Table 3). Both configurations have values within the range observed for As---Fe distances when arsenate is adsorbed onto goethite (3.2−3.35 Å20). Loring et al.30 have examined crystalline pentaaminecobalt(III) arsenate where arsenate is monodentate coordinated with the Co(III) atom. Using EXAFS, they obtained an As---Co distance of 3.25 Å. This observation led them to the conclusion that the As---Fe distance of approximately 3.3 Å observed for arsenate adsorbed onto goethite need not necessarily exclude a monodentate configuration, as had been done by previous researchers.20,22 Our results are consistent with the possibility that a bent monodentate configuration of phosphate or arsenate could result in a P---Fe or As---Fe distance of approximately 3.3 Å on the (010) and (001) surfaces. The monodentate phosphate complexes on the (100), (101), and (210) surfaces are outside the range of expected secondnearest-neighbor distances, however. Table 3 also compares the interatomic distance from the periodic and cluster models. The largest difference in the average P−O distance between the periodic DFT and the extended molecular cluster results is 0.03 Å. Hence, the first coordination sphere for P is similar in both cases; the P---Fe distances are more variable, however. For example, a discrepancy of 0.26 Å exists for the P---Fe distance in the monodentate (210) and (100) complexes (Table 3); however, all the other P---Fe distances agree to within 0.1 Å and most of the values are significantly closer to each other. This is particularly true for the bidentate models because the phosphate group is more geometrically constrained (Table 3). These results suggest that the extended molecular clusters are not highly distorted relative to the periodic models with regard to the phosphate adsorption complex and should represent the short-ranged structure and P−O vibrational frequencies reasonably well. Adsorption Energies. Obtaining potential energies from averaging MD simulations has the advantage that it takes into account the thermal averaging of configurations that may more realistically represent the internal energy state of a system. The 14583

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

observed spectra under acidic and slightly basic pH conditions. These results are contrary to the expected trend with changing pH (i.e., the H2PO4− species present under acidic conditions and HPO42− present under neutral or basic conditions). Consequently, it is not logical to make simple one-to-one assignments of surface complexes for these spectra solely on the basis of these correlations. The (001) surface is predicted to have only monodentate surface species based on the calculated ΔEads, but the frequency correlations predict a match with experiment for the bidentate and monodentate HPO4− species for this surface (SI Table 2). The (101) surface could have both monodentate and bidentate HPO42− complexes on the basis of the calculated ΔEads values, and both configurations produced reasonable correlations with experimentally observed frequencies (SI Table 2). A periodic DFT energy minimization and molecular cluster frequency analysis was also run with the bidentate PO43− species. This model surface complex did correlate reasonably well with the pH 4.22−5.71 spectra of Persson et al.23 The calculated ΔEads values from the periodic DFT−MD simulation and energy minimization were −148 and −107 kJ mol−1, respectively. Thus, the deprotonated state should be favored. In this state, good correlations were found with observed frequencies (SI Table 2); hence, the calculated adsorption energies and frequency correlations are consistent for the (101) surface having a deprotonated, bidentate configuration. Once again, however, the deprotonated species matches spectra collected from the sample reacted at lower pH rather than higher pH. The monodentate HPO42− (210) surface complex was predicted to have the lowest energy-minimized ΔEads values (Table 4). This species produces reasonably good frequency correlations with experimental frequencies for the pH 7.96 microgoethite (SI Table 2). An additional monodentate H2PO4− (210) surface complex also correlated well with observed frequencies on microgoethite from pH 7.5 and 7.9 experiments (SI Table 2). The calculated ΔEads of this species from the VASP energy minimization calculations is −228 kJ mol−1. Thus, the relatively minor (210) surface may be highly reactive toward oxyanions and play a disproportionate role in their adsorption by goethite as suggested by Villalobos and coworkers for chromate.34,35 However, if the overall percentage of phosphate adsorbed onto the (210) surface is small, then this species will not dominate the observed IR spectrum. Outer Sphere versus Inner Sphere. For each face studied here, an inner-sphere complex could be found that had a calculated exothermic ΔEads. However, we did not calculate Gibbs free energies of reaction, so we cannot predict equilibrium constants between inner- and outer-sphere complexes. With some calculated ΔEads values that are within an uncertainty of 0 (e.g., the (010) bidentate complex), it is possible that outer-sphere complexes could be present in some proportion on different goethite surfaces. Because the vibrational frequencies of aqueous phosphate species overlap with the broad bands observed for phosphate adsorbed onto goethite (cf. Table 1 and SI Table 1), one cannot exclude outer-sphere complexes based on IR data. The rinsing of samples with phosphate-free water after goethite is exposed to aqueous phosphate is one way to attempt to remove outersphere phosphate, but even in this case, one may merely reequilibrate the proportions of inner- and outer-sphere species. Thus, a combination of spectroscopic methods such as IR/ Raman, EXAFS, and NMR would be ideal for determining phosphate speciation on any given goethite sample.

0, however. Hence, both methods would predict unfavorable or very weak adsorption. We consider the EM values to be more accurate and precise because the electron density is determined more stringently with this method (Methods Section). If we use the EM values only, the monodentate configuration would be preferred over the bidentate configuration on the (210) and (001) surfaces, whereas the bidentate configuration is favored on the (101) and (100) surfaces. Although the ΔE ads for the bidentate configuration on (010) is more favorable than for the monodentate complex, the values are not significantly different from 0 so no adsorption can be predicted on this crystal face. In the next section, values in Table 4 are used as evidence in combination with the vibrational frequency calculations to understand the surface complexation of phosphate species better on the different crystal faces of goethite. Vibrational Frequencies. A comparison of Raman frequencies for H3PO4 + H3O+·17(H2O), H2PO4−·18(H2O), and HPO42−·18(H2O) calculated with B3LYP/6-31G(d,p) (SI Table 1) and observed values for aqueous phosphate58,59 resulted in correlations with slopes = 0.99 to 1.10, intercepts = −62 to +9 cm−1, and R2 = 0.98 to 0.99 (SI Table 1). Thus, we expect the molecular cluster results to be able to reproduce observed spectra of phosphate on goethite to this level of accuracy if the configuration is to be considered possible. Note that the frequency scaling factors based on these correlations would be 0.91 to 1.01 so the average scaling factor of 0.96 is reasonable, but we expect a 5% error to be included in our analysis. A problem with attempting to correlate calculated vibrational frequencies with experiment is that the experimental vibrational bands vary between studies. Not only is pH a major determinant of the observed vibrational bands, but separate studies using similar pH conditions in adsorption studies report different values23,24,27,31 than those obtained in this work. Consequently, it is difficult to determine which model fits the observed spectra most accurately because the observations are not consistent with one another. Instead, in SI Tables 2−6, we correlate the calculated frequencies with a number of experimental spectra including our own on two types of goethite (i.e., nanogoethite and microgoethite both wet and dry). An examination of SI Tables 2−6 reveals that there are several surface complexes that meet our fitting criteria depending on the pH and type of goethite. In general, the molecular cluster frequency predictions of which configurations should be present on goethite surfaces are consistent with the calculated energetics discussed above. However, the correlation of model and experiment is complicated by other factors that are discussed below. Extended Cluster Vibrational Frequencies. Many of the configurations determined to be possible on the basis of calculated ΔEads values also result in good correlations with observed frequencies. Starting with the (100) surface, periodic DFT−EM predicts the stability of the monodentate and bidentate complexes, and the monoprotonated bidentate and all monodentate configurations correlate with at least one of the observed IR spectra (SI Table 2). For the (010) surface, DFT results favor a weakly bound bidentate complex (Table 4), but both monoprotonated bidentate and all monodentate configurations correlate with at least one spectrum across the pH range (SI Table 2). The H2PO4− monodentate configuration matches the pH 7.5 microgoethite spectrum, and the HPO42− monodentate calculated frequencies are consistent with 14584

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

Iron Hydroxide Dimer Vibrational Frequencies. The lack of simple assignments for the IR spectra based on the extended cluster calculations was addressed by calculating the frequencies of [H2PO4]−, [HPO4]2−, and [PO4]3− in monodentate and bidentate configurations on the iron hydroxide dimer as models for goethite surfaces.28 In addition, energy minimizations and frequency calculations were carried out to model both the wet and dry experiments discussed above. Performing periodic DFT−MD simulations, energy minimizations, and extended cluster optimization and frequency analysis on this extended matrix of model types would be impractical, so the iron hydroxide dimer approximation is a reasonable alternative approach. Linear correlation parameters for all such models against the wet nanogoethite and microgoethite samples examined in this study were determined (SI Tables 3 and 4). As found in a previous study,28 species such as the diprototonated bidentate (BDH2) complex can match experimental frequencies at pH 4.22 and deprotontated monodentate (MDH0) correlations at pH 7.51. However, because there are newly observed peaks in the current IR spectra and because the observed frequencies vary with goethite type, other models such as BDH0 and MDH2 also produce good correlations with at least one observed spectrum (SI Tables 3 and 4). In addition, the outersphere models can also result in good agreement with experimental spectra. However, none of the individual models reproduce all of the observed peaks in a given spectrum. We conclude that the spectra are the result of numerous species adsorbed onto each type of goethite at each pH. This inference prompted an effort to correlate observed frequencies with various combinations of calculated frequencies based on contributions that had reasonably accurate individual correlations. Statistical parameters for these composite correlations were tabulated (SI Tables 3−6). In most cases, the composite spectra produce improved correlations with observed frequencies and in all cases explain the existence of more of the observed peaks compared to the individual model complexes. Comparison of Nanogoethite and Microgoethite Surface Complexes. As discussed above, the nanogoethite and microgoethite spectra can be considerably different when phosphate is adsorbed under the same conditions. According to Villalobos and co-workers,34,35 this is probably due to the presence of different faces on each particle size. Here we discuss possible differences in the surfaces available for bonding phosphate on the basis of the correlations with our various surface models (SI Table 2). The limited ability of any individual surface model to describe all of the observed frequencies is discussed in the Composite Synthetic Spectra Section below, so we caution the reader about overinterpreting a single species correlating with observed spectra. Beginning with our pH 4.2 spectra, the nanogoethite spectrum results in reasonable correlations with (100), (001), and (101) surface models but not the (010) and (210) surfaces. In contrast, the microgoethite spectrum can be fit by only the (001) surface model. For pH 5.7, the surface models matching the nanogoethite spectrum are (100), (001), (010), and (101) and those matching the microgoethite spectrum are (100) and (010). At pH 7.9, the predicted surface adsorption sites are (100), (001), and (010) for nanogoethite and (100) and (210) for microgoethite. These results suggest that the nanogoethite surfaces are dominated by the (100), (001), and (010) sites with adsorption to the (101) surface only at low pH.

Microgoethite would also be dominated by the (100), (001), and (010) sites with adsorption to the (210) site only at higherpH conditions. Hence, we cannot infer significant differences in the surfaces present on nanogoethite and microgoethite from these calculated frequencies. Similar differences in the adsorption of oxalate onto nanorods and microrods of goethite have been observed previously.60 The frequency differences may instead be ascribed to different protonation states and bonding configurations between the two sizes of particles. For example, at pH 4.2, nanogoethite is correlated with the [HPO4]2− bidentate (100) whereas microgoethite fits the [H2PO4]− monodentate model frequencies better. Such differences can be interpreted as resulting from phosphate bonding to surface defects such as steps on goethite, with the nature of these defects changing with particle size. Because this study did not examine step or other defects on each surface, we cannot take this analysis further, but future calculations on such defect-adsorbed species would be useful. Composite Synthetic Spectra. Both the ΔEads values (Table 4) and frequency correlations (SI Tables 2−6) strongly suggest that phosphate adsorption on goethite at a given pH results in the presence of more than one structurally distinct phosphate surface complex. Although the −90 kJ mol−1 value for monodentate HPO42− on the (210) surface is the most favorable energetically of the HPO42− surface models, the computational error and neglect of entropic effects do not exclude other configurations. More importantly, the (210) surface comprises a small percentage of the overall surface area on goethite;35 hence, other sites are likely to be populated as the (210) surface is saturated. A significant number of surface complexes have also been shown to reproduce the observed IR frequencies (SI Tables 2−6), and this number does not include possible outer-sphere complexes that could also be present. To address the possibility of multiple phosphate binding complexes, synthetic model spectra have been created that can be compared to experimentally observed spectra that exhibit broad overlapping bands. Kwon and Kubicki28 used a similar technique to compare calculated frequencies with observed spectra. Even in surface complexes that are strongly correlated with experimental frequencies, single species do not closely match observed spectra because not all observed peaks are reproduced by a single species. The addition of a second or third set of calculated IR frequencies and intensities to the synthetic spectrum results in a significantly closer reproduction of the experimental data (SI Figures 3−5). Furthermore, a general trend toward more deprotonated species with increasing pH is evident (SI Tables 3 and 4). Although the composite spectra do not always match the observed spectra particularly well, there are two possible reasons for these discrepancies. First, the widths of all frequencies were assumed to be the same in the synthetic spectra, but this will not be the case in reality. Second, each complex was assumed to occur in equal proportions in the synthetic spectra because we have no separate knowledge of the individual surface concentrations for each species. Even with these differences, however, the comparisons of individual and composite synthetic spectra are consistent with the interpretation that more than one type of phosphate surface complex populates individual goethite crystal faces at a particular pH. 14585

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir





CONCLUSIONS Model results presented in this study contribute to our understanding of the characterization and adsorption of phosphate ions on varying goethite crystal faces. The model cannot definitively determine the phosphate surface complexes that occur on the goethite surface. This is due in part to the accuracy and precision limits of the calculated ΔEads and vibrational frequencies, but it is also due to the fact that powdered goethite samples used in various experiments can exhibit different faces.35 Without the characterization of the percent surface area represented by each set of faces in the goethite sample, it is not possible to deconvolute any signal averaging that occurs as a result of the presence of multiple species. Ideally, spectra would be collected using single crystals where each face could be studied separately. Where this is not possible, because growing large crystals of a given phase such as goethite is difficult, the next best approach is to follow the work of Villalobos and co-workers who have examined adsorption on the same phase in different crystal habits (i.e., varying the percentage of each face present). With this data in hand, deconvoluting the vibrational frequencies associated with the surface complexes on each face is theoretically possible. The main conclusion from this work is that phosphate adsorption onto goethite likely occurs via a number of mechanisms (monodentate, bidentate, and outer sphere) depending on the face where the adsorption is occurring. Such a varied adsorption process is likely a phenomenon that occurs for other oxyanions as well and perhaps for ionic adsorption in general. Past studies that investigated adsorption on bulk powders of a mineral probably also led to a mixture of multiple surface complexes, which explains the broad bands observed in the IR spectra and the common disagreements in the literature about assigning these spectra to specific surface complexes.



REFERENCES

(1) Parfitt, R. L.; Atkinson, R. J. Phosphate Adsorption on Goethite (α-FeOOOH). Nature 1976, 264, 740−742. (2) Schwertmann, U., Taylor, R. M., Dixon, J. B., Weed, S. B., Eds.; Minerals in Soil Environments, 2nd ed.; Book Series No. 1; Soil Science Society of America: Madison, WI, 1989; pp 379−438. (3) Arai, Y.; Sparks, D. L. Phosphate Reaction Dynamics in Soils and Soil Components: A Multiscale Approach. Adv. Agron. 2007, 94, 135− 179. (4) O’Reilly, S. E.; Strawn, D. G.; Sparks, D. L. Residence Time Effects on Arsenate Adsorption/Desorption Mechanisms on Goethite. Soil Sci. Soc. Am. J. 2001, 65, 67−77. (5) Vetterlein, D.; Szegedi, K.; Ackermann, J.; Mattusch, J.; Neue, H. U.; Tanneberg, H.; Jahn, R. Competitive Mobilization of Phosphate and Arsenate Associated with Goethite by Root Activity. J. Environ. Qual. 2007, 36, 1811−1820. (6) Zeng, H.; Fisher, B.; Giammar, D. E. Individual and Competitive Adsorption of Arsenate and Phosphate to a High-Surface-Area Iron Oxide-Based Sorbent. Environ. Sci. Technol. 2008, 42, 147−152. (7) Mihaljevic, M.; Ettler, V.; Sisr, L.; Sebek, O.; Strnad, L.; Vonaskova, V. Effect of Low Concentrations of Phosphate Ions on Extraction of Arsenic from Naturally Contaminated Soil. Bull. Environ. Contam. Toxicol. 2009, 83, 422−427. (8) Appelo, C. A. J.; Van der Weiden, M. J. J.; Tournassat, C.; Chaelet, L. Surface Complexation of Ferrous Iron and Carbonate on Ferrihydrite and the Mobilization of Arsenic. Environ. Sci. Technol. 2002, 36, 3096−3103. (9) Arai, Y.; Sparks, D. L.; Davis, J. A. Effects of Dissolved Carbonate on Arsenate Adsorption and Surface Speciation at the Hematite-Water Interface. Environ. Sci. Technol. 2004, 38, 817−824. (10) Wijnja, H.; Schulthess, C. P. Effect of Carbonate on the Adsorption of Selenate and Sulfate on Goethite. Soil Sci. Soc. Am. J. 2002, 66, 1190−1197. (11) Puccia, V.; Luengo, C; Avena, M. Phosphate Desorption Kinetics from Goethite as Induced by Arsenate. Colloids Surf., A 2009, 348, 221−227. (12) Hiemstra, T.; Antelo, J.; van Rotterdam, A. M. D.; van Riemsdijk, W. H. Nanoparticles in Natural Systems II: The Natural Oxide Fraction at Interaction with Natural Organic Matter and Phosphate. Geochim. Cosmochim. Acta 2010, 74, 59−69. (13) Franzluebbers, A. J. Achieving Soil Organic Carbon Sequestration with Conservation Agricultural Systems in the Southeastern United States. Soil Sci. Soc. Am. J. 2010, 74, 347−357. (14) Torn, M. S.; Trumbore, S. E.; Chadwick, O. A.; Vitousek, P. M.; Hendricks, D. M. Mineral Control of Soil Organic Carbon Storage and Turnover. Nature 1997, 389, 170−173. (15) Li, W.; Zhang, S.; Shana, X. Q. Surface Modification of Goethite by Phosphate for Enhancement of Cu and Cd Adsorption. Colloids Surf., A 2007, 293, 13−19. (16) Venema, P.; Hiemstra, T.; van Riemsdijk, W. H. Interaction of Cadmium with Phosphate on Goethite. J. Colloid Interface Sci. 1997, 192, 94−103. (17) Wang, K.; Xing, B. Mutual Effects of Cadmium and Phosphate on their Adsorption and Desorption by Goethite. Environ. Pollut. 2004, 127, 13−20. (18) Park, S. J.; Lee, C. G.; Kim, S. B. The Role of Phosphate in Bacterial Interaction with Iron-Coated Surfaces. Colloids Surf., B 2009, 68, 79−82. (19) Hiemstra, T.; Van Riemsdijk, W. H. A Surface Structural Approach to Ion Adsorption: The Charge Distribution (CD) Model. J. Colloid Interface Sci. 1996, 179, 488−508. (20) Sherman, D. M.; Randall, S. R. Surface Complexation of Arsenic(V) to Iron(III) (Hydr)oxides: Structural Mechanism from Ab Initio Molecular Geometries and EXAFS Spectroscopy. Geochim. Cosmochim. Acta 2003, 67, 4223−4230. (21) Stachowicz, M.; Hiemstra, T.; van Riemsdijk, W. H. Surface Speciation of As(III) and As(V) in Relation to Charge Distribution. J. Colloid Interface Sci. 2006, 302, 62−75.

ASSOCIATED CONTENT

S Supporting Information *

ATR−FTIR spectra of aqueous phosphate. Statistical information on frequency correlations. Energy versus time plots from molecular dynamics simulations. This material is available free of charge via the Internet at http://pubs.acs.org.



Article

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We greatly appreciate financial support from the National Science Foundation (NSF) Collaborative Research in Chemistry grant “Structure and Properties of Disordered IronOxyhydroxides” (CHE-0714121). The comments of Dane Morgan and Nathan Pinney (University of Wisconsin Madison), Clare Grey and Derek Middlemiss (University of Cambridge), as well as other members of the CRC grant are acknowledged. Computational support was provided by the Research Computing and Cyberinfrastructure group at The Pennsylvania State University. We thank Dr. Samantha Horvath for help in producing the synthetic IR spectra. Narayan Bhandari is acknowledged for acquiring infrared data for aqueous phosphate. 14586

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587

Langmuir

Article

(43) Blöch, P. E. Projector Augmented-Wave Method. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 17953−17979. (44) Kresse, G.; Joubert, D. From Ultrasoft Pseuodopotentials to the Projector Augmented-Wave Method. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 59, 1758−1775. (45) Kresse, G; Furthmüller, J. Efficient Iterative Schemes for Ab Initio Total-Energy Calculations Using a Plane-Wave Basis Set. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169−11186. (46) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868. (47) Laasonen, K.; Car, R.; Lee, C.; Vanderbilt, D. Implementation of Ultrasoft Pseudopotentials in Ab Initio Molecular Dynamics. Phys. Rev. B: Condens. Matter Mater. Phys. 1991, 43, 6796−6799. (48) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.; Sutton, A. P. Electron-Energy-Loss Spectra and the Structural Stability of Nickel Oxide: An LSDA+U Study. Phys. Rev. B: Condens. Matter Mater. Phys. 1998, 57, 1505−1509. (49) Nosé, S. Constant Temperature Molecular Dynamics Methods. Prog. Theor. Phys. Suppl. 1991, 103, 1−46. (50) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, revision C.01; Gaussian, Inc.: Wallingford, CT, 2004. (51) Becke, A. D. Density-Functional Thermochemistry. III. The Role of Exact Exchange. J. Chem. Phys. 1993, 98, 5648−5652. (52) Lee, C. T.; Yang, W. T.; Parr, R. G. Development of the ColleSalvetti Correlation-Energy Formula into a Functional of the Electron Density. Phys. Rev. B: Condens. Matter Mater. Phys. 1988, 37, 785−789. (53) Rassolov, V. A.; Ratner, M. A.; Pople, J. A.; Redfern, P. C.; Curtiss, L. A. 6-31G* Basis Set for Third-Row Atoms. J. Comput. Chem. 2001, 22, 976−984. (54) Scott, A. P.; Radom, L. Harmonic Vibrational Frequencies: An Evaluation of Hartree-Fock, Møller-Plesset Quadratic Configuration Interaction, Density Functional Theory, and Semi-Empirical Scale Factors. J. Phys. Chem. 1996, 100, 16502−16513. (55) Harvey, O. R.; Rhue, R. D. Kinetics and Energetic of Phosphate Sorption in a Multi-Component Al(III)-Fe(III) Hydr(oxide) Sorbent System. J. Colloid Interface Sci. 2008, 322, 384−393. (56) Kabengi, N. J.; Daroub, S. H.; Rhue, R. D. Energetics of Arsenate Sorption on Amorphous Aluminum Hydroxides Studied Using Flow Adsorption Calorimetry. J. Colloid Interface Sci. 2006, 297, 86−94. (57) Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coordination Compounds; Wiley: Hoboken, NJ, 2006. (58) Cherif, M.; Mgaidi, A.; Ammar, N.; Vallée, G; Fürst, W. A New Investigation of Aqueous Orthophosphoric Acid Speciation Using Raman Spectroscopy. J. Solution Chem. 2000, 29, 255−269. (59) Rudolph, W. W. Ph.D. Thesis. Technische Universität Dresden, 1986. (60) Cwiertny, D. M.; Hunter, G. J.; Pettibone, J. M.; Scherer, M. M.; Grassian, V. H. Surface Chemistry and Dissolution of α-FeOOH Nanorods and Microrods: Environmental Implications of SizeDependent Interactions with Oxalate. J. Phys. Chem. C 2009, 113, 2175−2186.

(22) Waychunas, G. A.; Rea, B. A.; Fuller, C. C.; Davis, J. A. Surface Chemistry of Ferrihydrite: Part 1. EXAFS Studies of the Geometry of Coprecipitated and Adsorbed Arsenate. Geochim. Cosmochim. Acta 1993, 57, 2251−2269. (23) Persson, P.; Nilsson, N.; Sjoberg, S. Structure and Bonding of Orthophosphate Ions at the Iron Oxide-Aqueous Interface. J. Colloid Interface Sci. 1996, 177, 263−275. (24) Arai, Y.; Sparks, D. L. ATR-FTIR Spectroscopic Investigation on Phosphate Adsorption Mechanisms at the Ferrihydrite-Water Interface. J. Colloid Interface Sci. 2001, 241, 317−326. (25) Parfitt, R. L.; Atkinson, R. J.; Smart, R. S. C. The Mechanism of Phosphate Fixation by Iron Oxides. Soil Sci. Soc. Am. J. 1975, 39, 837− 841. (26) Parfitt, R. L.; Russell, J. D.; Farmer, V. C. Confirmation of the Surface Structures of Goethite (α-FeOOH) and Phosphate Goethite by Infrared Spectroscopy. J. Chem. Soc., Faraday Trans. 1 1976, 72, 1082−1087. (27) Tejedor-Tejedor, M. I.; Anderson, M. A. Protonation of Phosphate on the Surface of Goethite as Studied by CIR-FTIR and Electrophoretic Mobility. Langmuir 1990, 6, 602−611. (28) Kwon, K. D.; Kubicki, J. D. Molecular Orbital Theory Study on Surface Complex Structures of Phosphates to Iron Hydroxides: Calculation of Vibrational Frequencies and Adsorption Energies. Langmuir 2004, 20, 9249−9254. (29) Rahnemaie, R.; Hiemstra, T.; van Riemsdijk, W. H. Geometry, Charge Distribution, and Surface Speciation of Phosphate on Goethite. Langmuir 2007, 23, 3680−3689. (30) Loring, J. S.; Sandström, M. H.; Norén, K.; Persson, P. Rethinking Arsenate Coordination at the Surface of Goethite. Chem. Eur. J. 2009, 15, 5063−5072. (31) Luengo, C.; Brigante, M.; Antelo, J.; Avena, M. Kinetics of Phosphate Adsorption of Goethite: Comparing Batch Adsorption and ATR-IR Measurements. J. Colloid Interface Sci. 2006, 300, 511−518. (32) Zhong, B.; Stanforth, R.; Wu, S. N.; Chen, J. P. Proton Interaction in Phosphate Adsorption onto Goethite. J. Colloid Interface Sci. 2007, 308, 40−48. (33) Paul, K.; Kubicki, J. D.; Sparks, D. L. Sulphate Adsorption at the Fe (Hydr)oxide-H2O Interface: Comparison of Cluster and Periodic Slab DFT Predictions. Eur. J. Soil Sci. 2007, 58, 978−988. (34) Villalobos, M.; Perez-Gallegos, A. Goethite Surface Reactivity: A Macroscopic Investigation Unifying Proton, Chromate, Carbonate, and Lead(II) Adsorption. J. Colloid Interface Sci. 2008, 326, 307−323. (35) Villalobos, M.; Cheney, M. A.; Alcaraz-Cienfuegos, J. Goethite Surface Reactivity: II. A Microscopic Site-Density Model That Describes Its Surface Area-Normalized Variability. J. Colloid Interface Sci. 2009, 336, 412−422. (36) Cornell, R. M.; Schwertmann, U. The Iron Oxides: Structure, Properties, Reactions, Occurrences and Uses; Wiley-VCH: Weinheim, Germany, 2003. (37) Gaboriaud, F.; Ehrhardt, J.-J. Effects of Different Crystal Faces on the Surface of Colloidal Goethite (α-FeOOH) Particles: An Experimental and Modeling Study. Geochim. Cosmochim. Acta 2003, 67, 967−983. (38) Rakovan, J; Becker, U; Hochella, M. F. Aspects of Goethite Surface Microtopography, Structure, Chemistry and Reactivity. Am. Mineral. 1999, 84, 884−894. (39) Kerisit, S.; Ilton, E. S.; Parker, S. C. Molecular Dynamics Simulations of Electrolyte Solutions at the (100) Goethite Surface. J. Phys. Chem. B 2006, 110, 20491−20501. (40) Kumar, A.; Park, M.; Huh, J. Y.; Lee, H. M.; Kim, K. S. Hydration Phenomena of Sodium and Potassium Hydroxides by Water Molecules. J. Phys. Chem. A 2006, 110, 12484−12493. (41) Coey, J. M. D.; Barry, A.; Brotto, J.-M.; Rakoto, H.; Brennan, S.; Mussel, W. N.; Collomb, A.; Fruchart, D. Spin Flop in Goethite. J. Phys.: Condens. Matter 1995, 7, 759−768. (42) Kubicki, J. D.; Paul, K. W.; Sparks, D. L. Periodic Density Functional Theory Calculations of Bulk and the (010) Surface of Goethite. Geochem. Trans. 2008, 9, 4. 14587

dx.doi.org/10.1021/la303111a | Langmuir 2012, 28, 14573−14587