Chemical Stabilization and Electrochemical Destabilization of the Iron

Oct 18, 2016 - Iron Keggin Ion in Water. Omid Sadeghi,. †. Clément Falaise,. †. Pedro I. Molina,. †. Ryan Hufschmid,. ‡. Charles F. Campana,...
0 downloads 0 Views 6MB Size
Download PDF
Article pubs.acs.org/IC

Chemical Stabilization and Electrochemical Destabilization of the Iron Keggin Ion in Water Omid Sadeghi,† Clément Falaise,† Pedro I. Molina,† Ryan Hufschmid,‡ Charles F. Campana,§ Bruce C. Noll,§ Nigel D. Browning,‡,∥ and May Nyman*,† †

Department of Chemistry, Oregon State University, Corvallis, Oregon 97331, United States Department of Materials Science and Engineering, University of Washington, Box 352120, Seattle, Washington 98195-2129, United States § Bruker AXS Inc, Madison, Wisconsin 53711, United States ∥ Fundamental and Computational Sciences Directorate, Pacific Northwest National Laboratory, Richland, Washington 99352, United States ‡

S Supporting Information *

ABSTRACT: The iron Keggin ion is identified as a structural building block in both magnetite and ferrihydrite, two important iron oxide phases in nature and in technology. Discrete molecular forms of the iron Keggin ion that can be both manipulated in water and chemically converted to the related metal oxides are important for understanding growth mechanisms, in particular, nonclassical nucleation in which cluster building units are preserved in the aggregation and condensation processes. Here we describe two iron Keggin ion structures, formulated as [Bi6FeO4Fe12O12(OH)12(CF 3 COO) 1 0 (H 2 O) 2 ] 3 + (Kegg-1) and [Bi 6 FeO 4 Fe 1 2 O 1 2 (OH)12(CF3COO)12]1+ (Kegg-2). Experimental and simulated Xray scattering studies show indefinite stability of these clusters in water from pH 1−3. The tridecameric iron Keggin-ion core is protected from hydrolysis by a synergistic effect of the capping Bi3+ cations and the trifluoroacetate ligands that, respectively, bond to the iron and bridge to the bismuth. By introducing electrons to the aqueous solution of clusters, we achieve complete separation of bismuth from the cluster, and the iron Keggin ion rapidly converts to magnetite and/or ferrihydrite, depending on the mechanism of reduction. In this strategy, we take advantage of the easily accessible reduction potential and crystallization energy of bismuth. Reduction was executed in bulk by chemical means, by voltammetry, and by secondary effects of transmission electron microscopy imaging of solutions. Prior, we showed a less stable analogue of the iron Keggin cluster converted to ferrihydrite simply upon dissolution. The prior and currently studied clusters with a range of reactivity provide a chemical system to study molecular cluster to metal oxide conversion processes in detail.

■

INTRODUCTION

including catalysis, ligand exchange, electrostatic association, or linking into extended networks.17−19 Clusters identified as molecular building blocks in geological materials have motivated chemists and geochemists to synthesize them in their discrete form. Among the most highly sought and elusive clusters is the iron Keggin ion, whose structure is evident in both ferrihydrite and magnetite.20 The iron Keggin structure in these materials consists of a central tetrahedral oxoanion FeO4, and each of the four oxos bridges to a trimer of edge-sharing FeO6 octahedra; a general formula of the Keggin cluster is FeO4Fe12(O,OH,H2O)36.21 In magnetite, the central tetrahedral iron is trivalent, and the four trimers of edge-sharing octahedra are mixed Fe2/3+O6 octahedra joined by edge-sharing of octahedra between the trimers. This structural

Molecular metal-oxo clusters resemble discrete units of bulk metal oxide materials and serve as preassembled building blocks in nature and in synthesis.1−7 Stable isolatable clusters that possess only ligands derived from water (H2O, OH−, and O2−) are generally limited to the groups 5 & 6 polyoxometalates (polyanions) and group 13 polyoxocations.8−10 A few additional exceptions include a polycationic Hf tetramer and anionic silicate and aluminosilicate octamers.11−13 By employing capping organic ligands such as carboxylates and amines, metal-oxo clusters of any metal cation can be isolated.14,15 These metals include low-valent transition metals with partially filled d orbital and lanthanides.16 However, we have limited ability to manipulate or exploit these ligated clusters in solution, because they generally have poor solubility, and the metal cations are not accessible. Accessibility yields functionality, © XXXX American Chemical Society

Received: July 18, 2016

A

DOI: 10.1021/acs.inorgchem.6b01694 Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

■

isomer is known as the ε-Keggin ion. In ferrihydrite, all iron is generally trivalent, and three of the four trimers are linked by edge sharing, while the forth trimer is joined to the other three by corner sharing. The first isolated iron Keggin ion cluster was synthesized and described in 2002 by Bino and co-workers,22 and it exhibits the α-Keggin topology, in which the iron trimers are all joined by corner sharing. With the formula [FeO4Fe12F24(OMe)12]5−, this cluster has no exterior aqua ligands nor known aqueous chemistry, because the methoxy ligands are unstable in water. More recently, we synthesized the α-Keggin cluster [Bi6FeO4Fe12O12(OH)12(CCl3COO)12]1+ (Kegg-3).21 Oxo and hydroxyl ligands bridge the iron centers, while trichloroacetate caps the irons in the terminal position. We show that bismuth is key to stabilizing this highly reactive Keggin ion in solution, while the trichloroacetic acid (TCA) ligands are useful for crystallizing it in the solid state. We also showed that removal of both the Bi and TCA ligand in aqueous solution by metathesis of BiOCl leads to rapid conversion to nanoferrihydrite, demonstrating that the iron Keggin ion could be an important prenucleation cluster for related iron oxide materials. Kegg-3, however, cannot be dissolved in water without adding electrolytes to displace the bismuth and TCA, which resulted in fast or slow conversion to iron oxide precipitates. These solubility and stability limitations challenge further fundamental studies or applications23 of the molecular iron Keggin ion, such as determining the stepwise mechanism of the transition from the iron Keggin ion to ferrihydrite or other related iron oxide materials. We found that the same synthetic procedure but employing trifluoroacetate (TFA) as a ligand instead of TCA yielded an iron Keggin ion that is highly soluble and stable in water. The cluster formulated as [Bi6FeO4Fe12O12(OH)12(CF3COO)10(H2O)2]3+ (Kegg-1) has structural differences from Kegg-3, most notably (1) two terminal acetate (TFA) ligands are replaced with water, and (2) disorder in oxo and hydroxyl ligands leads to different Fe−Fe distances within the cluster. The latter feature influences magnetic behavior and is the subject of a concurrent publication.24 In addition to structural characterization of Kegg-1, here we document the aging in solution by small-angle X-ray scattering (SAXS) in both water and acetone and compare this to Kegg-3. The scattering data in water shows an exemplary match to that which is simulated from the solid-state structure, in particular, features related to the electron density contrast between the Fe13 core, the Bi “shell”, and the TFA ligands. These data both highlight SAXS as a powerful tool for characterization of small molecular clusters in solution and reveal the importance of retaining the Bi cap for protection of the iron-oxo core against hydrolysis in water. However, introduction of electrons leads to rapid separation of metallic Bi from the Fe13 core and its subsequent conversion to a magnetic material that we identified as nanomagnetite. We introduced electrons to aqueous Kegg-1 by three different experimental means: (1) chemically by addition of hydrazine, (2) electrochemically in cyclic voltammetry experiments, (3) via secondary effects of electron beam irradiation in solution phase in situ transmission electron microscopy (TEM) experiments.25 In addition to the Fe13-to-ferrihydrite conversion by BiOCl metathesis, Fe13-to-magnetite conversion by reduction of bismuth reported here provides a second example of destabilization of the iron Keggin ion in water to form extended oxide lattices, while preserving the cluster building block.

Article

EXPERIMENTAL SECTION

Synthesis. Synthesis of Kegg-1: 6.64 g of Fe(NO3)3·9H2O (16 mmol) was dissolved in 15 mL of water, and the beaker was placed on a hot plate at 150 °C. To this solution, 2.12 g of Bi(NO3)3·5H2O (4 mmol) was added, and the mixture was reduced to 5 mL by boiling to obtain a clear red solution. This solution was cooled and then added to a 15 mL solution including 5.5 g of trifluoroacetic acid and 4.12 g of sodium bicarbonate in an ice bath. After it was stirred for 2 min, the orange-white solid was separated by filtration and dissolved in 20 mL of acetone. The red needle-shaped crystals (Figure S1) formed after 3 d with slow evaporation of the solvent. (Yield: ∼10% based on Bi). Compositional analysis was performed by Galbraith Laboratories, Knoxville TN: Results calcd (%): C 7.82%; H 1.07%; F 17.62%; Na 2.03%; Fe 16.03%; Bi 27.69%. Experimental (%): C 10.07%; H 1.12%; F 15.89%; Na 1.52%; Fe 14.4%; Bi 24.9%. Kegg-2 was formed by dissolution of Kegg-1 in water, and the solution was left to evaporate. After two weeks, yellow-orange plates formed (Figure S1). The remaining solution was colorless, indicating all of the iron Keggin clusters had recrystallized. Elemental analysis of the crystals: Results calcd (%): C 7.47%; H 1.01%; F 17.73%; Fe 17.37%; Bi 30%. Experimental (%): C 8.05%; H 1.32%; F 19.16%; Fe 14.3%; Bi 24.9%. Single-Crystal X-ray Diffraction. Kegg-1 (Bruker) The pale yellow needle specimen crystal was selected under a polarizing optical microscope and cut to ∼0.05 × 0.05 × 0.10 mm. The crystal was mounted with a small amount of oil-based cryoprotectant (Parabar oil) to the tip of a MiTeGen Micromount. This assembly was placed into the goniometer of a Bruker D8 VENTURE diffractometer equipped with an Oxford Cryosystems Cryostream 800 low-temperature device operating at 100 K. Data were collected using Mo Kα radiation (λ = 0.710 73 Å) generated by an Incoatec IμS HB microfocus tube equipped with a HELIOS multilayer mirror optic. Data were recorded with a Bruker PHOTON 100 CMOS detector. Data collection comprised several sets of 0.5° ω and φ scans. Data reduction was accomplished with SAINT.26 A multiscan absorption correction was applied using SADABS.27 Structure solution with XT and refinement with XL28 was performed in the APEX3 software suite.29 The structure was solved with JANA 2006 using the built-in chargeflipping algorithm, and the structure refinement was performed with the JANA2006 software by the full-matrix least-squares method.30 The structure was solved in the monoclinic space group C2/m (No. 12). The refinement satisfactorily converged (R1 > 7%) in this space group. The crystal structure is highly disordered. In fact, all the oxygen atoms (oxo and hydroxo) of the Fe13 core, as well as the trifluoroacetate ligands, are disordered (Figure S2). The calculations in lower symmetry (such as C2) did not improve the model. In an attempt to model the disorder, we tested for a modulated superstructure (substructure), but the careful analysis of the crystallographic data does not confirm this hypothesis. Many crystals from different batches were tested, but the disorder is still observed. The final refinements include anisotropic thermal parameters of all metallic atoms (Bi, Fe, Na). A rigid-body approach was used to model the carboxylate molecules surrounding the Fe13Bi6 unit. Kegg-2 (OSU) A crystal was selected under polarizing optical microscope and glued on to a glass fiber for a single-crystal X-ray diffraction experiment. X-ray intensity data were collected at low temperature (173 K, liquid N2 flow) on a Bruker X8-APEX2 CCD area-detector diffractometer using Mo Kα radiation (λ = 0.710 73 Å) with an optical fiber as collimator. Several sets of narrow data frames were collected with ω scans. Data reduction was accomplished using SAINT V7.53a.31 Absorption corrections were performed using SADABS.32 The structure was solved with JANA 2006 using the built-in charge-flipping algorithm, and the structure refinement was performed with the JANA2006. The final refinements include anisotropic thermal parameters of all atoms. The crystal structure contains voids that correspond to disordered TFA ligands (one per cluster) and water molecules. Electrochemistry. Electrochemical measurements were performed on an Epsilon Electrochemical Workstation (BASi). The working, B

DOI: 10.1021/acs.inorgchem.6b01694 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry Table 1. Crystallographic Information for Kegg-1 and Kegg-2 empirical formula formula weight temperature/K crystal system space group a/Å b/Å c/Å α/deg β/deg γ/deg volume/Å3 Z ρcalc g/cm3 μ/mm−1 F(000) crystal size/mm radiation 2Θ range for data collection/deg index ranges

reflections collected restraints/constrains/parameters goodness-of-fit on F2 final R indexes [I ≥ 2σ(I)] final R indexes [all data] largest diff. peak/hole/e·Å−3

Kegg-1

Kegg-2

Fe13Bi6C20F30Na4O58 3810 100(2) monoclinic C2/m 19.944(2) 19.251(2) 15.1010(16) 90 103.109(4) 90 5646.9(11) 2 3.468 11.061 3468 0.169 × 0.049 × 0.033 Mo Kα (λ = 0.710 73) 2.19 to 27.5 −25 ≤ h ≤ 25 0 ≤ k ≤ 24 0 ≤ l ≤ 19 13608 24/23/179 2.60 R1 = 0.0701 wR2 = 0.0818 R1 = 0.1253 wR2 = 0.0851 3.73/−3.88

Bi6Fe13C24F36O76 4168 173(2) cubic Pa3̅ 27.6602(5) 27.6602(5) 27.6602(5) 90 90 90 21162.45(115) Å3 8 2.6164 34.514 15296 0.12 × 0.08 × 0.03 Cu Kα (λ = 1.541 84) 2.77−67.7 −31 ≤ h ≤ 31 −29 ≤ k ≤ 32 −33 ≤ l ≤ 32 71074 0/0/475 1.16 0.0411 0.0463 0.0928 0.053 1.17/−1.32

reference, and auxiliary electrodes (BASi) were, respectively, glassy carbon (3 mm diameter) or a (1 × 3) cm2 indium tin oxide (ITO) slide, Ag/AgCl (Aq 3 M NaCl) and Pt wire. Before recording the voltammograms, the 0.4 mM solution of Kegg-1 in aqueous 0.2 M CF3COOH was degassed with argon for 15 min. The glassy carbon working electrode was polished with diamond suspensions of decreasing grain size (15, 6, and 1 μm), rinsed with water after each of the polishing routines, and finally rinsed with methanol. The ITO slide was washed with acetone and heated at 85 °C until dry. The voltammograms were recorded starting from the resting potential of the assembled electrochemical cell (open-circuit potential) and scanning first toward the anodic region of the potential range. In Situ Liquid Cell TEM Experiments. At PNNL, solutions for in situ experiments were prepared by mixing 2 mg of the iron Keggin cluster with 1 mL of Milli-Q water for Kegg-1 or acetone for Kegg-3 and stirred for 24 h. To prepare aqueous solutions of Kegg-3, first 0.05 g of cetyltrimethylammonium chloride (CTACl) was mixed with 1 mL of Milli-Q water. This aqueous surfactant solution was then mixed with 5 mg of TCA clusters and stirred for 24 h. All solutions were filtered with 0.45 μm Nylon syringe filters after mixing to remove any large aggregates. In Situ STEM. These experiments were performed on a probecorrected FEI Titan (PNNL) operated at 300 kV with a calibrated probe current at the sample of 8 pA for Kegg-3 and 19 pA for Kegg-1. Calibrated probe currents, along with magnification settings, were used to compute electron doses in STEM mode.33 Closed liquid cells were assembled with silicon chips with 50 nm thick silicon nitride windows in a Hummingbird Scientific holder. For each cell, 0.2 μL of solution was pipetted on the bottom window during assembly. Following in situ experiments, cells were opened to air to evaporate water for postmortem characterization. Posthoc STEM-electron energy loss spectroscopy (EELS; Gatan Quantum ERS) and EDS (Si[Li] EDAX) were

performed on the same FEI Titan, and a separate image-corrected 300 kV FEI Titan was used for selected area electron diffraction (SAED) and HRTEM. TEM of Chemically Reduced Kegg-1. The samples were prepared by dispensing 1 μL of diluted particles in acetone on Holey carbon grids and allowed to dry. The grids were then placed in a double tilt holder and inserted into the Titan G2 80−200 keV transmission electron microscope (TEM) operating at 200 keV. TEM diffraction experiments were conducted using SAD aperture 40, spot size 3, and condenser aperture 100. Measurements of the lattices spacings from these diffraction patterns were made in Digital Micrograph (DM) and also using Mitchell’s Diffraction Tools plugin to DM. Small-Angle X-ray Scattering. These data were collected on an Anton Paar SAXSess instrument utilizing Cu Kα radiation (1.54 Å) and line collimation. The instrument is equipped with a twodimensional image plate for data collection in the q = 0−2.5 Å−1 range. All solutions were contained in a sealed disposal 1.5 mm diameter capillary tube for SAXS measurements, and data collection time was 30 min. Likewise, background solutions were prepared and measured for 30 min. SAXSQUANT software was used for data collection and treatment (normalization, primary beam removal, background subtraction, desmearing, and smoothing to remove extra noise created by the desmearing routine). Differences in scattering intensity in the high q-range (∼q > 1 Å−1) from similar solutions is the result of imperfect background subtraction due to slight differences in capillary thicknesses and overlap between solvent scattering (increases at high q) and particle/cluster scattering. Radius of gyration Rg by the Guinier approximation and intensity I0 in the Guinier region of the scattering curve was also determined utilizing SAXSQUANT. All other analyses and fits to determine size, shape, size distribution, and pair distance distribution function (PDDF) were performed utilizing the C

DOI: 10.1021/acs.inorgchem.6b01694 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry IRENA macros within IgorPro 6.3 (Wavemetrics) software.34,35 Scattering curves were simulated using SolX software.36,37 Structural files containing the selected portion of the cluster were created as P1.xyz files, which were then converted to .pdh files for data import.38 Scattering curves from 0−2.5 Å−1 were simulated, which were then imported into Irena and treated in the same manner as experimental data. Other Characterization. Powder X-ray diffraction data was performed on Rigaku Ultima-IV with Cu Kα radiation. The diffraction patterns were obtained in the range from 10 to 90° (2θ) with a scan speed of 0.5°/min and step size of 0.05°. Infrared spectra were recorded in attenuated reflectance mode (ATR) using a Nicolet iS 10 spectrometer (Thermo Scientific). Thermogravimetric analysis-differential scanning calorimetry (TGA-DSC) were performed under air or nitrogen flow (100 mL min−1) with a heating rate of 10 °C/min on a SDT Q600 instrument (TA). Compositional analysis for Fe, Bi, Na, F, C, & H was performed by Galbraith Laboroatories (Knoxville, TN). The Fe, Bi, and Na ion content were determined by the GLI procedure ME-70 using a ICP-OES Optima 5300. C & H analyses were by the GLI Procedure ME-14 using the PerkinElmer 2400 Series II CHNS/O Analyzer. F content was determined by the GLI Procedure E9−3 using a using a Fisher Accumet AR25 Ion Meter.

acetate derivative. Terminal water ligands bound to Fe3+ clusters are not common, due to the high Lewis and Bronsted acidity of the Fe−OH2 complex. One other example is an aminocarboxylate-ligated tetramer with Fe−OH2 terminal ligands,39 and currently reported Kegg-1 provides a second rare example. The clusters are aligned into chains, bridged by four Na+ cations per cluster (Figure 1). The sodium is bound to

Figure 1. A view of the structure of Kegg-1, illustrating the linking of the Keggin ions by sodium cations. The terminal water ligands, Fe− OH2, point directly up and down in each of the three clusters (these are the polyhedra that do not have a ligand in the terminal position).

■

RESULTS AND DISCUSSION Solid-State Structures. Kegg-1 crystallizes in a monoclinic space group (C2/m). The crystal data collection and refinement parameters are summarized in Table 1. The basic cluster unit is similar to that of prior-reported Kegg-3 in that it is an αKeggin ion with Fe3+ in both the tetrahedral and octahedral sites. The six square “faces” of the Keggin cluster is capped by a Bi3+ cation, bonded to four oxo/hydroxyl ligands (μ2-O-Fe2) within the cluster. There is considerable disorder in the oxo/ hydroxyl ligands of the cluster. The central tetrahedral iron is surrounded by eight half-occupied oxo-ligands, typical of many α-Keggin ion structures with rotational disorder in the crystal lattice. This disorder was not observed prior in the structure of Fe13Bi6(TCA)12. Additionally, all 24 bridging oxo/hydroxyls are disordered (50% occupancy in each; O−O distance ∼0.7 Å) over two positions (μ2-O-Fe2 and μ2-OH-Fe2). The μ2-OH-Fe2 are “outer” with a smaller Fe−O−Fe bond angle ranging from 98 to 106° and a bond valence sum (BVS) ≈ 1.0−1.3 (Table S1). The μ2-O-Fe2 are “inner” with a bigger Fe−O−Fe bond angle ranging from 125 to 134° and a BVS ≈ 1.8−2.2. This differs from the structure of prior reported Kegg-3, which distinctly has the hydroxyl ligands within the iron triads and the oxo ligands bridging between the iron triads. While the oxo/ hydroxyls exhibit significant disorder in Kegg-1, the iron sites are fully occupied with no disorder. The Fe−Fe distances in this structure are all very similar ranging from 3.3 to 3.4 Å, including the distance from the tetrahedral Fe to any of the octahedral Fe, the Fe−Fe distances within triads, and the Fe− Fe distances between triads. In Kegg-3, there are three distinct Fe−Fe distances: 3.2 Å (within triads), 3.4 Å (tetrahedral Fe to octahedral Fe), and 3.6 Å (between triads). The effect of these differences on anti-ferromagnetic coupling and spin frustration are discussed elsewhere.24 The Bi−O bond distances that link Bi3+ to the cluster are very similar between the two analogues (∼2.2 Å). This is expected because in both ordered Kegg-3 and disordered Kegg-1, two OH− ligands and two O2− ligands link each Bi3+ to the Fe13 core. The TFA ligands are also disordered, as described above in the Experimental Section. Kegg-1 is a trivalent polycation (including the charge of the six bismuth), while the TCA-ligated cluster is monovalent. This is because in the TFA analogue, two of the terminal ligand positions are occupied by neutral water instead of the anionic

the TFA ligand oxygens. The Na−O bond distances to the TFA ligands and to lattice water all range from 2.4 to 2.6 Å. The sodium cations plus the cluster together provide a 7+ charge that must be neutralized. We determined the charge compensation is from four deprotonated ligands along with 1.5 disordered carbonate anions. Three lines of evidence are consistent with this hypothesis: free void space, infrared analysis, and compositional analysis. The free void space in the structure that we calculate with PLATON40 is 1742.4 Å3 per unit cell, meaning 871 Å3 per cluster. We estimate 578 Å3 is required for the four deprotonated TFA ligands (476 Å3) and 1.5 carbonate molecules (102 Å3; ∼17 Å3/per atom). The remaining free void space is filled by water molecules (∼16 water molecules). The presence of water and carbonate molecules are confirmed by IR spectroscopy (Figure S3); the carbonate is derived from the sodium carbonate used to deprotonate the ligand in the synthesis reaction. The number of water molecules was also confirmed by TGA (Figure S4). We observe 6.6% weight loss below 200 °C that we attribute to lattice water (theoretical weight loss for 16 water molecules is 6.4%). The formula based on these analyses is (Na4)[Bi6 FeO4Fe12O 12(OH) 12 (H2 O) 2(CF3 COO)10](CF3 COO)4· (CO3)1.5·16H2O. Kegg-2 was obtained simply by crystallization of Kegg-1 and is fully formulated as [Bi6FeO4Fe12(OH)12O12(CF3COO)12]· 15H2O[CF3COO]. It crystallizes in the cubic space group Pa3̅ with eight clusters per unit cell. The structure is different from both that of Kegg-1 and Kegg-3, in that there is no disorder whatsoever within the cluster, including the TFA ligands. The prior-reported Kegg-3 and the currently reported Kegg-1 both have two settings of each acetate ligand with 50% occupancy of each setting. The structure of Kegg-2 with complete order allows us to elucidate the interaction between the Bi and the ligands and develop a hypothesis regarding their synergistic role in stabilizing the Fe13 core. This is discussed later. The structure of the Fe13 core is much like that of Kegg-3 with the hydroxyl ligands between the edge-sharing octahedra of the four trimers, while the oxo ligands bridge the corner-sharing octahedra between trimers. The identity of the hydroxyls and oxos are D

DOI: 10.1021/acs.inorgchem.6b01694 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

than the TCA acid; and this is mostly likely due to the stronger H···F hydrogen bonds compared to H···Cl hydrogen bonds. To directly compare the solution stability of Kegg-1 and Kegg-3, we dissolved both these clusters in acetone and monitored the 50 mM solutions by SAXS with time (Figure 3).

confirmed by BVS calculations (Table S2). For all three of the structures discussed in this paper, the BVS for the hydroxide ligands are 1.2−1.3, and the BVS for the oxo ligands are 2.0− 2.1. The cluster is a monovalent cation, which requires charge balancing. In the case of Kegg-3, we located a single deprotonated ligand, and we had assumed this is also the case for Kegg-2, but it is presumably disordered, since we could not locate it. However, Kegg-2 has a free void space of 224 Å3, which could accommodate the disordered ligand. The ordered O2− and OH− ligands in Kegg-2 and Kegg-3 versus the disordered O2− and OH− ligands in Kegg-1 manifests slightly different overall shapes of the clusters. The disordered Kegg-1 is essentially spherical, and the nearestneighbor Fe−Fe distances are very similar. Kegg-2 and Kegg-3 are more tetrahedrally shaped with longer Fe−Fe distance between the corner-sharing octahedra and shorter Fe−Fe distances between the edge-sharing octahedra. All Fe−Fe distances of the three structures, including the central tetrahedral Fe to the 12 octahedral Fe, are summarized in Figure 2.

Figure 3. Scattering curves of 50 mM acetone solutions of Kegg-3 and Kegg-1 as a function of time.

Kegg-1 is quite stable, the scattering curve exhibits minimal change, even after 25 d, and no precipitate is observed. Most important, even the oscillatory region remains unchanged, suggesting that all the bismuth remains bound to the cluster (discussed in detail below). The Guinier region (below q = 0.2 Å) is flat, indicating spherical particles, and the radius of gyration (Rg) of 5.0 Å for the experimental data is in agreement with that of the simulated scattering data (Table 2). Rg is Table 2. Radius of Gyration (Rg) from Simulated and Experimental Scattering Data for Kegg-1 and Kegg-2 Figure 2. Fe−Fe distances (Å) in the three iron Keggin ion clusters illustrated the spherical shape of disordered Kegg-1 (all Fe−Fe distances from 3.3 to 3.4 Å, green), and the tetrahedral shape of ordered Kegg-2 (blue) and Kegg-3 (red) with edge-sharing Feoctahedral−Feoctahedral distance ≈ 3.2 Å, corner-sharing Feoctahedral− Feoctahedral distance ≈ 3.5 Å, and Fetetraheral−Feoctahedral ≈ 3.3−3.4 Å.

compound

solvent

Rg (Guinier) (Å)

Rg (PDDF) (Å)

Kegg-1 Kegg-1 Kegg-2 Kegg-1, simulated

water acetone water

5.0 5.2 5.0 4.9

5.1 5.3 5.1 4.7

defined as the root-mean-square measure of all mass-weighted vectors in the particle from the center of mass, which for

Solution Speciation and Stability of the TFA-Ligated Clusters. A big difference between Kegg-1 and Kegg-2 compared to Kegg-3 is the aqueous solubility. Kegg-1 and Kegg-2 are similar in their solution behavior, both in solubility and X-ray scattering in water (Figure S5), and here we focus on studies of Kegg-1 compared to Kegg-3. To dissolve Kegg-3 in water requires addition of an electrolyte, such as an alkali salt or a surfactant. Every instance of dissolution of Kegg-3 in water resulted in alteration of the cluster. With cetyltrimethylammonium chloride, for example, the bismuth rapidly precipitated as bismuth oxychloride, and the iron cluster converted to ferrihydrite nanoparticles.21 Kegg-1 dissolves readily without addition of electrolytes. Initially we thought the difference in charge (+3 for Kegg-1, +1 for Kegg-3) was the origin of the different aqueous dissolution behavior. Generally speaking, low charge-density ions exhibit poorer aqueous solubility. However, Kegg-2, also with a +1 charge, readily dissolves in water. Therefore, the higher solubility of the TFA-analogues is simply related to the solubility properties imparted by the ligands. We note also that the TFA acid exhibits higher aqueous solubility

spherical particles is Rg = R

( 35 ) (R = radius).

41

However, the

scattering data for Kegg-3 in acetone shows rapid aggregation (after 1 h) and eventual conversion to a gel. This is evident by the continued rise in the scattering intensity in the low-q region, monitored from 1 h of aging up to 40 h (Figure 3). Formation of polydisperse aggregates that include larger particles is responsible for this notable change in the scattering data. However, the data could not be fit to any reasonable model including cylindrical models or fractal aggregate models, and this suggests polydispersity and possible precipitation. Interestingly, in the region between q = 0.2 and 0.5 Å−1, the scattering remains constant, which suggests the primary particle size (the Fe13 core) remains constant, but these particles aggregate and associate as the Bi and ligands dissociate from the Fe13 core. In the high-q region of the scattering curve of Kegg3, instead of an oscillation, there is a relatively sharp feature (peak) around q = 1.2 Å−1. We attribute this to imperfect subtraction of the solvent peak; this peak is observable in the E

DOI: 10.1021/acs.inorgchem.6b01694 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 5 shows the SAXS curve of Kegg-1 in water (50 mM), along with the scattering curve simulated from the solid-state Xray structure. We determined a radius of gyration (Rg) of 5.0 Å for the experimental data and 4.9 Å for the simulated data (Table 2). The cluster shows an ideal monodisperse solution of discrete particles, indicated by 0 slope in the Guinier region (low-q region). This datum is particularly remarkable in that the experimental and simulated data match the first oscillation. In our experience with small molecular clusters like these iron Keggin ions, we generally only observe distinct oscillations if the cluster has a core−shell structure with bimodal electron density across its diameter (i.e., the uranyl peroxide capsules) and higher electron density in the shell.42 This is because the oscillatory region for small clusters (1−2 nm in radius) corresponds with background solvent scattering. The challenges associated with perfect background subtraction include structuring of both the solvent molecules and cluster counterions around the cluster. Kegg-1 does indeed have three different electron densities: (1) the Fe13 core, (2) the Bi caps as the inner shell, and (3) the fluorinated ligands as the outer shell, and these ligands should exhibit sufficient electron density contrast with the solvent. This motivated us to fit the scattering curve to the end of the first oscillation (q = 1.7 Å−1) with a core−shell−shell model in Irena, utilizing the general form factor for i concentric shells:43 1 F3(q) = [ρ V (R1)F1(q , R1) M3 1

scattering data of just acetone (Figure S6). However, the Kegg1 scattering curve (Figure 3) does not show this peak because the q-range is dominated by cluster scattering (the first oscillation). Similar to the prior electrolyte-aided aqueous dissolution of Kegg-3,21 the Bi ions and ligands detach from the Fe13 core as it starts to aggregate and form ferrihydrite. Immediate dissociation of Bi from Kegg-3 upon dissolution causes loss of the dense Bi shell, in addition to increasing polydispersity, thus diminishing the oscillation. While the difference in solution stability of the TFA-ligated clusters compared to the TCA-ligated cluster is very evident, there is no clear explanation. In fact most analyses led to the conclusion that the TCA analogue should be more stable. For example, because fluorine is considerably more electronwithdrawing, we would expect it to diminish the effective negative charge of the Fe13 core and therefore weaken the bond between Bi3+ and the cluster. However, the Bi−O bond distances and BVS are essentially identical between all three analogues. As mentioned above, the complete ordering of the ligands of Kegg-2 allows for unambiguous comprehension of the Bi coordination environment, Figure 4. Each Bi is bonded

N

+

∑ (ρi − ρi− 1)V (R i)F1(q , R i)] i=2

(1)

where M is the total particle volume, R1 is the core radius, Ri the radii of i-shells, V is volume of the concentric shells, and ρ is the electron scattering length densities of the core and different shells. Core−shell fits to the experimental and simulated scattering data are shown in Figure 5, and the fitting parameters are summarized in Table 3. These parameters total six and include the core radius, the thickness of the two shells, and the rho (Xray scattering length density; a measure of electron density normalized to the solvent) for the core and two shells. The numbers are not entirely realistic, especially the core rho values, so we report them with a cautionary note. However, they are consistent for both the experimental and simulated scattering data, and the trends are correct; that is, radius (Fe13) ≈ shell (TFA) > shell (Bi); and the rho(Bi) is substantially larger than both the Fe13 core and the TFA shell. For further conviction that the prominent oscillation in the scattering curve is owed to the trifunctional electron density across the radius of the cluster, we also simulated scattering data for the following hypothetical derivatives of Kegg-1 with differing electron density contrast: (1) Kegg-1 with 19 Fe atoms, converting the 6 Bi atoms to Fe in the simulation and (2) “Fe19” without the ligands. These simulated scattering curves are shown in Figure S7, and clearly there is a significant effect of the electron density profile of the cluster on the first oscillation. In both simulations, the first oscillation is diminished. The pair distance distribution function (PDDF; a probability distribution of scattering vectors through the particle) from the method of Moore for the experimental and simulated data are shown in Figure 6 (fits to data in Figures S8), and this too suggests a core−shell−shell structure of Kegg1.44 The large central peak with the shoulder to the left is

Figure 4. Bi coordination environment in Kegg-2. Bi (blue sphere) is bound to two O2− ligands of the corner-sharing iron octahedra (gray polyhedra, dark red sphere; Bi−O = 2.2 Å) and the OH− ligands of two edge-sharing iron octahedra (Bi−O = 2.4 Å), two Bi−O−C of the TFA ligands (Bi−O = 2.8 Å), and one water molecule (Bi−O = 2.6 Å).

to two acetate ligands (2.8 Å) and one water molecule (2.6 Å). The lone pair of Bi is apparently pointed toward the bottom of the page in Figure 4. Consider the O−C bond lengths of the ligands: Fe−O−C = 1.24−1.25 Å and the Bi−O−C = 1.23− 1.26 Å. The similarity of these suggests the C = O double bond is delocalized. Therefore, the C−O in proximity to the Bi has partial double-bond, or π-bond, character, which may create repulsion with the lone pair on bismuth. However, the electronwithdrawing character of the fluoride compared to chloride may diminish this effect and allow stable association with Bi, even upon dissolution in a ligating solvent such as water or acetone. Therefore, we can conclude that there is a synergistic effect between the ligand stabilizing the Bi and the Bi stabilizing the cluster. Below we discuss in detail the structure of dissolved Kegg-1 as determined by experimental and simulated SAXS curves. F

DOI: 10.1021/acs.inorgchem.6b01694 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 5. Experimental and simulated X-ray scattering data for Kegg-1 along with the core−shell−shell fits to these data with the refined parameters summarized in Table 3.

buffering pH (2.9) down to pH 0.8 (Figure 7). In comparison, the Al13 Keggin ion is stable between ∼pH 3 and 4.5.47 At pH

Table 3. Core−Shell−Shell Fitting Parameters for Experimental and Simulated Scattering Data for Kegg-1 parameter

experimental

simulated

core radius (Å) shell (1) thickness (Å) shell (2) thickness (Å) core rho shell (1) rho shell (2) rho total diameter (Å)

3.5 0.9 4.5 0.759 178 13.6 17.8

3.5 0.2 4.1 2.02 181 11.3 15.6

rho = X-ray scattering length density (1 × 1010 cm−1), normalized to solvent (water) scattering defined with a fixed value of 10.

a

Figure 7. Scattering curves of Kegg-1 as a function of pH indicates stability from pH = 0.8 to 2.9.

0.6 and 0.7, there is a decrease in the scattering intensity at low q, related to conversion of a fraction of the clusters to smaller species that do not scatter significantly (Rg values summarized in Table S3). The scattering intensity scales as r6 (r = radius of scattering species). The other evidence is in shallowing of the negative slope between q = 0.2 and 0.7 Å−1, which is an indication of increasing polydispersity toward smaller species. Aging of these acidified solutions for 1 d and 1 week (Figures S9 & S10) shows a trend toward more polydispersity and smaller species in all solutions except the three highest pH solutions. The trinuclear fragment of the Keggin ion is a likely decomposition product of acidified Kegg-1, as it is well-known, particularly ligated by carboxylates.48−50 We are uncertain of the fate of the Bi in the acidified solutions. However, no precipitate was ever observed, so it either remains bound to the iron species or is released into solution, likely in a mononuclear form, since the scattering intensity decreases rather than increases. With partial dissociation, while retaining solubility, it is possible that other Fe−Bi oxo clusters may be isolated. In prior study, addition of the Cl anion to a solution of Kegg3 yielded complete precipitation of bismuth and conversion of

Figure 6. PDDF of Kegg-1 in water (red) and simulated from the solid-state structure (blue).

characteristic of a core−shell structure with a more electrondense shell, while the smaller peak to the right is likewise characteristic of a less dense shell, such as an associated counterion or ligand shell.41,45 The aqueous solubility of Kegg-1 allows determination of its stability as a function of pH. This is important, because prior to our studies, a hypothetical Fe13 Keggin ion was predicted to be similar to the Al13 Keggin ion [AlO4Al12(OH)24(H2O)12]7+ in both structure and aqueous behavior.46 Addition of base to an aqueous solution of Kegg-1 simply resulted in formation of a red precipitate. However, Kegg-1 is considerably stable in acid; the solution scattering data exhibited no change from the selfG

DOI: 10.1021/acs.inorgchem.6b01694 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry the Fe13 core to colloidal ferrihydrite.21 Similar experiments with Kegg-1 again suggested more inert Bi−Ocluster bonds. These too formed precipitates, but these precipitates also contained iron, based on both energy-dispersive spectroscopy (EDS) and the observed orange color. The alkali cation for chloride affected the rate of precipitation, with LiCl inducing more rapid precipitation, while CsCl induced slow precipitation. This may be related to the better ability of Cs+ to stabilize the Fe13 core in solution, with partial removal of the bismuth. Nonetheless, the experimental evidence shows that BiOCl metathesis only partially separates the Bi from Fe13 for Kegg-1, while complete removal is accomplished for Kegg-3. In summary, all solution behaviors that we probed revealed much higher stability and solubility of Kegg-1 compared to Kegg-3, particularly with respect to Bi dissociation. We currently cannot rationalize this stark difference. All Fe−O and Bi−O bonds are similar between the two analogues, as are bond valence sums. The cluster distortion looks similar for the two, when considering different settings of the disorder for Kegg-1. The inductive effect (electron withdrawing) of the TFA ligand actually points toward a weaker Bi−O bond for this analogue. We are currently exploring this by computational studies, including generating electrostatic potential maps of the two clusters and also considering Fe−O and Bi−O binding energies of the two clusters. Introduction of Electrons to Kegg-1. In addition to identifying a controlled chemical pathway to convert Fe13 to ferrihydrite, we likewise endeavor to convert Fe13 to magnetite, since the Keggin geometry is recognized in both structures. This chemical process requires partial reduction of Fe3+ to Fe2+. Designing reproducible and well-controlled processes will then enable forthcoming mechanistic studies of Fe13 conversion to related iron oxides with partial or complete preservation of the cluster. Kegg-1 is an excellent candidate for controlled studies, because it does not destabilize upon dissolution via uncontrolled release of Bi (as demonstrated in the SAXS studies above). In this regard, hydrazine was added to boiling aqueous solution of Kegg-1 to yield a black magnetic precipitate and colorless solution (Figure S11). Powder X-ray diffraction (XRD) of the precipitate shows only metallic Bi (Figure S12), while the magnetic iron phase is X-ray amorphous. TEM imaging and elemental mapping reveals complete nanometerscale separation of bismuth from iron (Figures 8 & S13). Consistent with XRD of the material, the Bi metal is faceted

particles, while the iron-rich phases do not have any apparent crystallinity. However, SAED shows diffraction rings and spots that can be identified as magnetite and ferrihyrite. Electrochemical studies of aqueous Kegg-1 were also performed to further investigate the transformation of the cluster into ferrihydrite/magnetite. A cyclic voltammogram (Figure 9) was obtained from a solution of Kegg-1 in

Figure 9. Cyclic voltammogram of Kegg-1 (0.4 mM in 0.2 M CF3COOH) (blue dotted line, first segment: OCP to 1.2 V; blue solid line, full cycle, 1.2−1.0 V).

trifluoroacetic acid within the potential range of stability of the electrolyte solution. The starting point of the voltammogram was the open circuit potential (OCP, +490 mV vs Ag/ AgCl) of the electrochemical cell, and the scan progressed initially toward the anodic limit of the potential window. This voltammogram shows that the cluster can be electrochemically reduced in acidic aqueous solution, albeit this process is irreversible. No oxidation waves are observed in the first segment of the voltammogram (+490 to +1200 mV) as expected from the oxidation state of the metal centers in the cluster (Fe3+ and Bi3+). Upon scan reversal, a first composite reduction wave starting at +20 mV and peaking at −260 mV is assigned to the reduction of the cluster’s Fe3+ to Fe2+ and of Bi3+ to Bi0. This Bi0 deposits as an iron-free metallic film on the surface of the electrode as confirmed by EDS. Switching the direction of the potential scan at −1000 mV reveals two distinctive reoxidation waves. The first one, peaking at +110 mV, is a stripping wave caused by the reoxidation of Bi0 to Bi3+, which readily separates from the surface of the electrode and dissolves in the aqueous mixture of trifluoroacetic acid and Fe2+. The second wave (peaking at +960 mV) in the first full potential scan is considerably lower in intensity and can be ascribed to the reoxidation of the Fe2+ ions to Fe3+. These assignments are supported by the voltammograms of Fe(NO3)3 and Bi(NO3)3, obtained under the same experimental conditions as for Kegg-1 and shown in Figure S14. The addition of the Fe(NO3)3 and Bi(NO3)3 voltammograms closely resembles the voltammogram of Kegg-1 except for the position of the waves associated with the Bi3+/Bi0 redox process and the reduction of Fe3+ to Fe2+. The significant cathodic shift in the reduction of the Fe3+ centers in Kegg-1 with respect to free Fe3+ in Fe(NO3)3, and the similar but less pronounced shift with respect to free Bi3+ in Bi(NO3)3, reflects both the barrier

Figure 8. TEM image (left) and elemental mapping (middle) of hydrazine-reduced Kegg-1, showing complete phase separation of iron from bismuth. The separation is observed in both the morphology and elemental mapping. The Bi metal is faceted, while the iron oxide phase does not have a distinct shape. (right) SAED of a representative ironrich region. Labeled diffraction spots and rings (nm) are consistent with a mixture of ferrihydrite and magnetite. H

DOI: 10.1021/acs.inorgchem.6b01694 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 10. In situ STEM images of products of Fe13 in solution (A−C). Iron oxide/hydroxide particles self-assemble from Kegg-3 (A) after 24 h in acetone or in aqueous solution (B). Similar particles form in situ from Kegg-1, (C) which was otherwise stable in water. Posthoc analyses confirm structure and phase. Filtered HRTEM (D)51 shows the highly disordered polycrystalline structure. Lattice spacings in SAD pattern (E) match reasonably well with ferrihydrite, while iron white-line ratios from EELS indicate a mix of Fe2+ and Fe3+.

posed by the capping Bi3+ centers in the cluster to the reduction of the Fe3+ centers and the most favorable reduction of free Bi3+ compared to Bi3+ when forming part of the cluster. These observations suggest that the structure of the cluster is not maintained in solution upon reduction of its Bi3+ and Fe3+ centers even in an electrolyte solution containing an excess of the trifluoroacetate ligand, and hence they support the crucial role of Bi3+ to stabilize the Fe13 core. We used liquid cell scanning TEM (STEM) as a third mechanism to introduce electrons to Kegg-1 in water and Kegg-3 in acetone and water with in situ observation of resultant phase and morphology change. Both cluster solutions revealed formation of linear aggregates; examples are shown in Figure 10. The Kegg-3 self-assembled over the course of ∼24 h in both acetone and aqueous solutions. The Kegg-1 cluster upon dissolution in water is stable; however, similar aggregates formed under electron irradiation during STEM imaging. Posthoc inspection of the electron-beam induced colloids revealed a mixture of Fe2+ (EELS) and Fe3+, and a mixture of ferrihydrite and magnetite (electron diffraction), consistent

with the hydrazine-reduced Kegg-1. The reduction in the TEM imaging experiments is a secondary effect of irradiation.33 There was no evidence of the bismuth by EDS or EELS in the primary particles; however, there was residual bismuth in unreacted Kegg-1 elsewhere in the liquid cell. Again, complete separation of bismuth from iron was accomplished, preceding conversion of the iron Keggin ion to related simple oxide materials. Finally, we summarize the solution-phase conversion of the Fe13 clusters to (1) ferrihydrite (Kegg-3 upon dissolution, eq 2); (2) small clusters or monomers (Kegg-1 upon acidification, eq 3); and (3) magnetite upon addition of electrons (Kegg-1 & Kegg-3, eq 4). 4[Bi6FeO4 Fe12O12 (OH)12 (L)12 ]1 + → 26(Fe2O3·0.5H 2O) + 24Bi 3 + + 48L1 − + 20OH− + H 2O (2)

[Bi6FeO4 Fe12O12 (OH)12 (L)12 ]1 + + 44H+ ⇒13Fe3 + + 6Bi 3 + + 12L1 − + 28H 2O I

(3)

DOI: 10.1021/acs.inorgchem.6b01694 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

process is far more complex than simple aggregation, and certainly warrants investigation.

3[Bi6FeO4 Fe12O12 (OH)12 (L)12 ]1 + + 67e−1 → 13Fe3O4 + 18Bi + 36L1 − + 4H 2O + 28OH−

■

(4)

L = trifluoracetate or trichloroacetate ligand. Both eq 2 & eq 4 involve the release of hydroxide. The Fe13 Keggin ions, intermediate in dimension between a hydrated Fe3+ cation and an infinite-lattice iron oxyhydroxide, is formed in acid, below pH 2. Therefore, the release of hydroxide upon decomposition further destabilizes the cluster, which likely accelerates the conversion process, presumably initiated by OH− coordination to an Fe3+ center, or deprotonation of a bridging hydroxide. Upon acidification of Kegg-1 (see Figure 7), we observe conversion to smaller scattering species by SAXS; that is, destruction of the full cluster. This can be initiated by protonation of oxo and hydroxyl ligands, leading to solubilization of the clusters into monomers, as stated in eq 3. The hypothetical end-product and process is a reversion to Fe3+.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.6b01694. Microscope images of Kegg-1 and Kegg-2, view of disordered trifluoroacetate ligands, IR spectrum, TGADSC analysis, X-ray scattering curves, SAXS curve, simulated scattering data, photo of magnetic precipitate, XRD patterns, EDX map, cyclovoltammograms, tabulated BVS calculations (PDF) Crystallographic information files for Kegg-1 (CIF) Crystallographic information files for Kegg-2 (CIF)

■

■

AUTHOR INFORMATION

Corresponding Author

CONCLUSIONS Prior we had synthesized the first Fe13 Keggin ion from water that cannot be redissolved in water without simultaneous dissociation of the protecting shell of ligands and Bi3+ and, ultimately, conversion to ferrihydrite. Here we found that both aqueous solubility and stability, even down to pH ≈ 1, could be achieved by simply changing the ligand from a chlorinated acetate to a fluorinated acetate. The pH-dependent studies bracketed the stability range of this iron Keggin ion between 1 and 3. This is considerably more acidic than the aluminum Keggin ion to which a hypothetical iron Keggin was compared, prior to the current studies. Generally, Broensted acidities are higher for an iron oxyhydroxide than an aluminum oxyhydroxide. For example, the hydrolysis of Fe3+ to FeOH2+ takes place 2 pH units more acidic than Al3+, or the zero point of charge of Fe3+ colloids is 2−4 pH units more acidic than Al3+ colloids. This is why the aprotic mineral hematite (Fe2O3) forms easily in soils, whereas the corresponding aluminum mineral, corundum, forms only at high temperatures and pressures. Comparison of the experimental X-ray scattering datum to those simulated from X-ray structure provided an excellent match, and two independent paths of fitting both the experimental and simulated data revealed a core−shell−shell Fe13−Bi-ligand structure. The considerable stability of the trifluoroacetate-ligated cluster compared to the trichloroacetate-ligated cluster was not expected but could be explained by bridging of the ligands between the Fe13 core and the bismuth caps. Chemical reduction, voltammetry, and irradiation during in situ liquid-phase TEM all led to complete separation of the bismuth caps (plus ligands) from the Fe13 core, because Bi3+ both readily reduces and crystallizes. Fe3+ to Fe2+ also readily occurred, producing magnetic materials identified as magnetite or mixtures of ferrihydrite and magnetite. We have now identified multiple pathways, including bismuth metathesis and bismuth reduction, to convert the iron Keggin ion to related oxyhydroxide materials. We are continuing to study the stepwise processes by electrospray ionization mass spectroscopy, theory, calorimetry, total X-ray scattering, and more detailed in situ TEM investigations. This is important because the clusters contain a considerably higher Fe-octahedra/Fetetrahedra ratio (12:1) than ferrihydrite and magnetite (4:1 and 2:1, respectively) as well as different combinations of edgesharing and corner-sharing iron octahedra. Therefore, the

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The work led by and performed at Oregon State University was supported by the U.S. Department of Energy (DOE), Office of Basic Energy Sciences, Division of Materials Sciences and Engineering, under Award No. DE-SC0010802. A portion of this work was done as part of the Chemical Imaging Initiative conducted under the Laboratory Directed Research and Development program at Pacific Northwest National Laboratory (PNNL). PNNL is operated by Battelle for DOE under Contract No. DE-AC05-76RL01830. This work was performed in part using the William R. Wiley Environmental Molecular Sciences Laboratory, a U.S. DOE national scientific user facility sponsored by the DOE’s Office of Biological and Environmental Research and located at PNNL.

■

REFERENCES

(1) Nashner, M. S.; Frenkel, A. I.; Adler, D. L.; Shapley, J. R.; Nuzzo, R. G. Structural Characterization of Carbon-Supported Platinum− Ruthenium Nanoparticles from the Molecular Cluster Precursor PtRu5C(CO)16. J. Am. Chem. Soc. 1997, 119, 7760−7771. (2) Yu, F.; Turco, R. P. From molecular clusters to nanoparticles: Role of ambient ionization in tropospheric aerosol formation. J. Geophys. Res. Atmos. 2001, 106, 4797−4814. (3) Rozes, L.; Sanchez, C. Titanium oxo-clusters: precursors for a Lego-like construction of nanostructured hybrid materials. Chem. Soc. Rev. 2011, 40, 1006−30. (4) Chao, Y.; et al. In situ probing calcium carbonate formation by combining fast controlled precipitation method and small-angle X-ray scattering. Langmuir 2014, 30, 3303−9. (5) Gebauer, D.; Völkel, A.; Cölfen, H. Stable prenucleation calcium carbonate clusters. Science 2008, 322, 1819−22. (6) Habraken, W. J. E. M.; et al. Ion-association complexes unite classical and non-classical theories for the biomimetic nucleation of calcium phosphate. Nat. Commun. 2013, 4, 1507. (7) Jolivet, J.-P.; Chanéac, C.; Tronc, E. Iron oxide chemistry. From molecular clusters to extended solid networks. Chem. Commun. (Cambridge, U. K.) 2004, 477−483. (8) Nomiya, K.; Sakai, Y.; Matsunaga, S. Chemistry of Group IV Metal Ion-Containing Polyoxometalates. Eur. J. Inorg. Chem. 2011, 2011, 179−196. J

DOI: 10.1021/acs.inorgchem.6b01694 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry (9) Michot, L. J.; Montargès-Pelletier, E.; Lartiges, B. S.; d’Espinose de la Caillerie, J.-B.; Briois, V. Formation Mechanism of the Ga13 Keggin Ion: A Combined EXAFS and NMR Study. J. Am. Chem. Soc. 2000, 122, 6048−6056. (10) Burgert, R.; Stokes, S. T.; Bowen, K. H.; Schnöckel, H. Primary reaction steps of Al13- clusters in an HCl atmosphere: snapshots of the dissolution of a base metal. J. Am. Chem. Soc. 2006, 128, 7904−8. (11) Jackson, M. N.; et al. An overview of selected current approaches to the characterization of aqueous inorganic clusters. Dalton Trans. 2015, 44, 16982−7006. (12) Goudarzi, N. 27Al NMR study of the effect of aqueous and methanolic media on the distribution of tetraphenylammonium aluminosilicate species. J. Struct. Chem. 2015, 56, 250−258. (13) Ruther, R. E.; Baker, B. M.; Son, J.-H.; Casey, W. H.; Nyman, M. Hafnium sulfate prenucleation clusters and the Hf(18) polyoxometalate red herring. Inorg. Chem. 2014, 53, 4234−42. (14) D’Alessio, D.; et al. Lanthanoid ‘bottlebrush’ clusters: remarkably elongated metal-oxo core structures with controllable lengths. J. Am. Chem. Soc. 2014, 136, 15122−5. (15) Huang, L.; Zhang, J.; Cheng, L.; Yang, G.-Y. Poly(polyoxometalate)s assembled by {Ni6PW9} units: from ring-shaped Ni24-tetramers to rod-shaped Ni40-octamers. Chem. Commun. (Cambridge, U. K.) 2012, 48, 9658−60. (16) Miras, H. N.; Vilà-Nadal, L.; Cronin, L. Polyoxometalate based open-frameworks (POM-OFs). Chem. Soc. Rev. 2014, 43, 5679−99. (17) Rozes, L.; Steunou, N.; Fornasieri, G.; Sanchez, C. TitaniumOxo Clusters, Versatile Nanobuilding Blocks for the Design of Advanced Hybrid Materials. Monatsh. Chem. 2006, 137, 501−528. (18) Nyman, M.; Burns, P. C. A comprehensive comparison of transition-metal and actinyl polyoxometalates. Chem. Soc. Rev. 2012, 41, 7354−67. (19) Villa, E. M.; et al. Reaction dynamics of the decaniobate ion [H(x)Nb10O28](6‑x)‑ in water. Angew. Chem., Int. Ed. 2008, 47, 4844−6. (20) Michel, F. M.; et al. The structure of ferrihydrite, a nanocrystalline material. Science 2007, 316, 1726−9. (21) Sadeghi, O.; Zakharov, L. N.; Nyman, M. Crystal growth. Aqueous formation and manipulation of the iron-oxo Keggin ion. Science 2015, 347, 1359−62. (22) Bino, A.; Ardon, M.; Lee, D.; Spingler, B.; Lippard, S. J. Synthesis and structure of [Fe13O4F24(OMe)12]5‑: The first open-shell Keggin ion. J. Am. Chem. Soc. 2002, 124, 4578−4579. (23) Song, Y.-F.; Tsunashima, R. Recent advances on polyoxometalate-based molecular and composite materials. Chem. Soc. Rev. 2012, 41, 7384−402. (24) Bandeira, N. A. G.; Sadeghi, O.; Woods, T. J.; Zhang, Y.-Z.; Schnack, J.; Dunbar, K.; Nyman, M. An examination of the magnetic properties of homonuclear Fe(III) Keggin type polyoxometalates. Submitted to J. Phys. Chem. A 2016. (25) Woehl, T. J.; Evans, J. E.; Arslan, I.; Ristenpart, W. D.; Browning, N. D. Direct in situ determination of the mechanisms controlling nanoparticle nucleation and growth. ACS Nano 2012, 6, 8599−610. (26) SAINT V8.34; Bruker AXS Inc: Madison, WI, 2013. (27) Krause, L.; Herbst-Irmer, R.; Sheldrick, G. M.; Stalke, D. Comparison of silver and molybdenum microfocus X-ray sources for single-crystal structure determination. J. Appl. Crystallogr. 2015, 48, 3− 10. (28) Sheldrick, G. M. Crystal structure refinement with SHELXL. Acta Crystallogr., Sect. C: Struct. Chem. 2015, 71, 3−8. (29) APEX3 v2015.11; Bruker AXS Inc: Madison, WI, 2015. (30) Petricek, V.; Dusek, M.; Palatinus, L. Crystallographic computing system JANA2006: General features. Z. Kristallogr. Cryst. Mater. 2014, 229, 345−352. (31) SAINT Plus Version 7.53a; Bruker Anal. X-ray Syst.: Madison, WI, 2008. (32) Sheldrick, G. M. SADABS, Programs for Scaling and Absorption Correction of Area Detector Data; University of Göttingen: Germany, 1997.

(33) Abellan, P.; et al. Factors influencing quantitative liquid (scanning) transmission electron microscopy. Chem. Commun. 2014, 50, 4873. (34) Zhang, R.; Thiyagarajan, P.; Tiede, D. M. Probing protein fine structures by wide angle solution X-ray scattering. J. Appl. Crystallogr. 2000, 33, 565−568. (35) Ilavsky, J.; Jemian, P. R. Irena: tool suite for modeling and analysis of small-angle scattering. J. Appl. Crystallogr. 2009, 42, 347− 353. (36) Banfield, J. F.; Welch, S. a; Zhang, H.; Ebert, T. T.; Penn, R. L. Aggregation-based crystal growth and microstructure development in natural iron oxyhydroxide biomineralization products. Science 2000, 289, 751−754. (37) Tiede, D. M.; Zhang, R.; Chen, L. X.; Yu, L.; Lindsey, J. S. Structural characterization of modular supramolecular architectures in solution. J. Am. Chem. Soc. 2004, 126, 14054−14062. (38) O’Boyle, N. M.; et al. Open Babel: An open chemical toolbox. J. Cheminf. 2011, 3, 33. (39) Panasci, A. F.; Ohlin, C. A.; Harley, S. J.; Casey, W. H. Rates of water exchange on the [Fe4(OH)2(hpdta)2(H2O)4] molecule and its implications for geochemistry. Inorg. Chem. 2012, 51, 6731−8. (40) Spek, A. L.; et al. Structure validation in chemical crystallography. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2009, 65, 148−155. (41) Putnam, C. D.; Hammel, M.; Hura, G. L.; Tainer, J. A. X-ray solution scattering (SAXS) combined with crystallography and computation: defining accurate macromolecular structures, conformations and assemblies in solution. Q. Rev. Biophys. 2007, 40, 191−285. (42) Falaise, C.; Nyman, M. The key role of U28 in aqueous selfassembly of uranyl peroxide nanocages. Chem. - Eur. J. 2016, 22, 14678. (43) Pedersen, J. S. Modelling of Small-Angle Scattering Date from Colloids and Polymer Systems. Neutrons, X-rays Light Scatt. Methods Appl. to Soft Condens. Matter 2002, 391−420. (44) Moore, P. B. Small-angle scattering. Information content and error analysis. J. Appl. Crystallogr. 1980, 13, 168−175. (45) Nyman, May; Nyman, M.; Fullmer, L. B. In Trends in Polyoxometalates Research; Ruhlmann, L., Schaming, D., Eds.; Nova Science Publishers, Inc, 2015; pp 151−170. (46) Bradley, S. M.; Kydd, R. A. Comparison of the species formed upon base hydrolyses of gallium(III) and iron(III) aqueous solutions: the possibility of existence of an [FeO4Fe12(OH)24(H2O)12]7+ Polyoxocation. J. Chem. Soc., Dalton Trans. 1993, 15, 2407−2413. (47) Baes, C. F.; Mesmer, R. S. The Hydrolysis of Cations. John Wiley & Sons, New York, London, Sydney, Toronto 1976. 489 Seiten, Preis: £ 18.60. Berichte der Bunsengesellschaft für Phys. Chemie 1977, 81, 245−246. (48) Raptopoulou, C. P.; Sanakis, Y.; Boudalis, A. K.; Psycharis, V. Salicylaldoxime (H2salox) in iron(III) carboxylate chemistry: Synthesis, X-ray crystal structure, spectroscopic characterization and magnetic behavior of trinuclear oxo-centered complexes. Polyhedron 2005, 24, 711−721. (49) Amani, V.; Safari, N.; Khavasi, H. R. Solution and solid state characterization of oxo-centered trinuclear iron(III) acetate complexes [Fe3(μ3-O)(μ-OAc)6(L)3]+. Spectrochim. Acta, Part A 2012, 85, 17− 24. (50) Amini, M. M.; Yadavi, M. Influence of metal core of mixedmetal carboxylates in preparation of spinel: ZnFe2O(O2CCF3)6 as a single-source precursor for preparation of ZnFe2O4. Appl. Organomet. Chem. 2005, 19, 1164−1167. (51) Marks, L. D. Wiener-filter enhancement of noisy HREM images. Ultramicroscopy 1996, 62, 43−52.

K

DOI: 10.1021/acs.inorgchem.6b01694 Inorg. Chem. XXXX, XXX, XXX−XXX