Interactions Involving Lewis Acidic Aluminum Sites in Oxide

ARTICLE pubs.acs.org/JPCC

Interactions Involving Lewis Acidic Aluminum Sites in Oxide-Supported Perrhenate Catalysts Brian C. Vicente,† Ryan C. Nelson,† Anthony W. Moses,† Swarup Chattopadhyay,† and Susannah L. Scott*,†,‡ †

Department of Chemical Engineering and ‡Department of Chemistry and Biochemistry, University of California, Santa Barbara, California 93106, United States

bS Supporting Information ABSTRACT: In oxide-supported perrhenate catalysts, isolated Re(VII) becomes anchored to the support at surface hydroxyl sites. However, oxides that also possess Lewis acidity offer the possibility of additional bonding interactions with perrhenate, which may account for their enhanced activity, for example, in olefin metathesis. Evidence for such interactions was sought by X-ray absorption spectroscopy at the Re L1- and L3-edges, in combination with DFT modeling. Only perrhenate grafted onto silica using an anhydrous method shows a simple tSiOReO3 structure, with three equivalent RedO multiple bonds. Silica-supported perrhenate prepared by aqueous impregnation is not grafted, but interacts only weakly with the silica support as a tetrahedral ReO4
ion. On silica
alumina and γ-alumina, the anchored perrhenates also interact with adjacent Lewis acid sites. In particular, a terminal oxo ligand interacts with a coordinatively unsaturated Al atom, creating a well-defined Re
Al EXAFS path. In addition, an oxygen atom from the support is coordinated to Re, increasing the metal coordination number from four to five. Simulated vibrational spectra of DFT-optimized models indicate that these pentacoordinate Re sites are consistent with reported Raman spectra for perrhenate supported on γ-alumina.

’ INTRODUCTION Oxide-supported perrhenates are versatile heterogeneous catalysts used in the selective catalytic reduction of NOx by NH3,1 the epoxidation of olefins,2,3 and olefin metathesis.4,5 While their activation mechanisms are the subject of continuing investigation and debate,6
8 it is clear that the nature of the oxide support exerts a dramatic effect on catalyst performance. For example, perrhenate-modified silica
alumina is highly active in olefin metathesis near room temperature, while perrhenate on silica is inactive.9,10 Furthermore, the catalyst supported on silica
alumina is much more active at low Re loading than perrhenate-modified γ-alumina catalyst on a per Re site basis.4,6 Finally, γ-alumina-based catalysts are virtually inactive when their Re loading is below 3 wt %, but their activity increases sharply at higher loading,6 while for silica
alumina-based catalysts, activity increases linearly with Re loading up to ca. 3 wt %, above which a sublinear increase is observed.4 The origins of these support-dependent reactivities lie in structural and/or electronic differences between the supported perrhenate sites. Despite much investigation,8,11
13 little is known about the nature of these differences. On the basis of their vibrational spectra, a single structure with local C3v symmetry (S-OReO3, where S is a support cation) has been proposed to describe the perrhenate sites on γ-alumina, titania, and zirconia.14,15 They are believed to arise by anchoring perrhenate at surface hydroxyl sites during calcination,16 r 2011 American Chemical Society

as shown for silanol-containing supports in eq 1. tSiOH þ NH4 ReO4 f tSiOReO3 þ NH3 v þ H2 O v ð1Þ In support of monomer structures, no characteristic Re
O
Re vibrations12,14,17 or diffraction lines for extended rhenium oxide phases (e.g., Re2O7)18 have been detected at low Re loadings. While such signatures have been observed at higher loadings,6,19
21 they may be difficult to detect for materials with low metal content due to the low sensitivity of the measurements.22,23 Vibrational studies also indicate that the symmetry of the supported perrhenate sites in the dehydrated catalysts is lower than Td, since the υs(ReOx) mode is visible in both the IR and Raman spectra.14,15,17,24 The structures of these monomer “S
OReO3” sites (where S is a support cation) have been suggested to be analogous to those of molecular silanolates R3SiOReO3, and such compounds have been synthesized as models for the perrhenate sites in heterogeneous catalysts.25,26 However, the relevance of the molecular model compounds has yet to be established, since they do not show the same activity as the heterogeneous catalysts, for example, in olefin metathesis. The metathesis activity of supported perrhenates has long been associated with the presence of Received: October 15, 2010 Revised: March 20, 2011 Published: April 14, 2011 9012

dx.doi.org/10.1021/jp109929g | J. Phys. Chem. C 2011, 115, 9012–9024

The Journal of Physical Chemistry C

ARTICLE

Brønsted acidity in the support.27
30 Furthermore, the enhanced Brønsted acidity of certain hydroxyl groups on silica
alumina has been ascribed to the presence of neighboring Lewis acidic Al sites.20,31 This proximity of Brønsted and Lewis acid sites19,32 suggests that additional interactions with Lewis acid sites may be responsible for activating certain grafted perrhenates. Interestingly, the catalytic activity of the molecular complexes O3Re
[Al(OAr)
O]2
ReO3 for olefin metathesis was attributed to coordination of RedO to strongly Lewis acidic Al sites.33 We also note that the Lewis basicity of oxygen in the supporting oxides may play an important role in the activation of the heterogeneous catalysts. In this regard, it is noteworthy that the molecular complexes ROReO3 are themselves Lewis acidic, readily forming coordination complexes with bases such as alcohols and ethers,34
36 eq 2. R 0 R 00 O þ ReO3 ðORÞ f R 0 R 00 O
ReO3 ðORÞ

ð2Þ

Lewis acid
base interactions between grafted metal complexes and solid supports can be difficult to detect by vibrational spectroscopy but may be observed using X-ray absorption spectroscopy. Recently, we demonstrated that CH3ReO3 supported on amorphous silica
alumina interacts with both a Lewis acidic Al site and a neighboring Lewis basic O atom of the support.37,38 In several EXAFS studies of γ-alumina-supported perrhenate, no such interactions have been reported.11,39
42 However, in some cases the spectra were obtained for airexposed samples,11,39 and supported perrhenates are known to undergo rapid hydrolysis by atmospheric moisture.12 The air-free XANES of perrhenate on γ-alumina at the Re L1-edge was reported to be consistent with “AlOReO3” structure, although spectra of model compounds were not recorded for comparison.12 The EXAFS of ReOx/HZMS-5 was said to be consistent with tetrahedral ReO4
sites, even though the 27Al MAS NMR spectra suggested interactions between perrhenate and Lewis acidic Al sites in the zeolite framework.43 EXAFS curve-fitting of H3Re3(CO)12/γ-alumina after oxidation supported the existence of isolated, pentacoordinate ReO5 sites;44 however, no Re
Al paths were reported. Recently, we described a computational model for perrhenatemodified silica
alumina which displays additional interactions between the tSiOReO3 sites and neighboring support atoms.45 In this work, we provide XAS evidence for these interactions in perrhenates supported on γ-alumina and silica
alumina, and explore their consequences for the vibrational spectra of these materials.

’ EXPERIMENTAL AND COMPUTATIONAL METHODS Materials. Perrhenates supported on silica, γ-alumina, and silica
alumina were prepared by suspending silica (EvonikDegussa Aerosil 380, BET surface area 363 m2/g, nonporous), γ-alumina (Strem, 185 m2/g), or silica
alumina (GraceDavison Davicat 3113, 7.6 wt % Al, 573 m2/g) in 5 mL of an aqueous solution of NH4ReO4 (Aldrich, 99þ%). The NH4ReO4 concentration was chosen to achieve the desired Re loading. Each suspension was stirred for 2 h at room temperature, followed by heating at 80 °C overnight in air to remove the water. The materials were calcined in flowing O2 at 450 °C overnight and then evacuated at 450 °C and 10
4 Torr for 2 h. All solids were white after these treatments and were subsequently handled inside a glovebox (95%) was sublimed onto the treated silica at room temperature under vacuum, followed by overnight evacuation (10
4 Torr) to remove any physisorbed material. This perrhenate-modified silica contained 2.0 wt % Re. Propene Metathesis. A weighed amount (20 mg) of each catalyst was loaded into a glass batch reactor (volume ca. 120 mL) under Ar, after which the reactor was removed from the glovebox and evacuated. The section of the reactor containing the catalyst was immersed in a water bath at 25 °C to maintain isothermal reaction conditions. Propene (Praxair, 99.5%) was introduced at the desired pressure via a high vacuum manifold. Aliquots of 1.9 mL were expanded at timed intervals into an evacuated septum port separated from the reactor by a stopcock. One hundred microliter samples of the aliquot were removed with a gas-tight syringe via a septum. Gases were analyzed by flame ionization detection on a Shimadzu GC 2010 gas chromatograph equipped with a 30 m Supelco alumina sulfate PLOT capillary column (0.32 mm i.d.). The peak area of the small propane contaminant present in the propene was used as an internal standard. Crystal Structure Determination. (CH3)3SiOReO3 was synthesized by refluxing a solution of anhydrous Re2O7 (Strem, 99.99%) in neat (CH3)3SiOSi(CH3)3 (Aldrich, 99.5þ%), according to a literature procedure.46 Diffraction-quality single crystals were grown by allowing the solution to cool overnight, under N2. A colorless crystal of approximate dimensions 0.20  0.15  0.08 mm was mounted on a glass fiber and transferred to a Bruker CCD platform diffractometer. Data collection (20 s/ frame, 0.3°/frame for a sphere of diffraction data) and determination of unit cell parameters were accomplished using the program SMART.2 Raw frame data were processed using the program SAINT.3 Subsequent calculations were carried out using the program SHELXTL.4 The structure was solved by direct methods and refined on F2 by full-matrix least-squares techniques. All non-hydrogen atoms were refined anisotropically. X-ray Absorption Spectroscopy. Experiments at the Re L1and L3-edges were performed on beamlines 2-3 (bend) and 4-1 (wiggler), respectively, at the Stanford Synchrotron Radiation Lightsource (SSRL). It operates at 3.0 GeV with a current of 75
100 mA. In order to prevent adsorption of atmospheric moisture, the perrhenate-modified powders were packed into sample plates inside a glovebox and sealed under N2, as previously described.45,47 The EXAFS of the pure compounds NH4ReO4 and trimethylsilyl perrhenate were obtained by grinding each solid with BN (Sigma-Aldrich, 99%) under N2 to obtain 9013

dx.doi.org/10.1021/jp109929g |J. Phys. Chem. C 2011, 115, 9012–9024

The Journal of Physical Chemistry C

ARTICLE

a mixture containing ca. 2 wt % Re. For each sample, three data sweeps were averaged to improve the signal-to-noise ratio. The XANES regions of subsequent data sweeps were compared to verify that no sample decomposition had occurred. Upon completion of the data sweeps, air-sensitive samples were exposed to ambient atmosphere and an additional scan collected in order to verify that the samples were not inadvertently hydrolyzed prior to the initial data collection. EXAFS spectra were analyzed with Athena and Artemis48 (based on IFEFFIT, v. 1.2.11),49 using a previously described fitting procedure.38,47,50 To minimize the number of variable fit parameters, the value of S02 was fixed at the FEFF-calculated value for all of the Re-containing models considered in this work, 0.923. Initially, curvefits were attempted by fixing N at appropriate integer values. Once a reasonable fit was obtained, N was refined as a fit variable. This did not improve the fit quality significantly, and the original fixed values always lay within the error ranges of the fitted value. EXAFS fit parameters showing variable N results can be found in the Supporting Information. Gas Phase Analysis. The formation of hexamethyldisiloxane in the reaction of trimethylsilyl perrhenate with silica capped with hexamethyldisilazane was confirmed by sampling the reactor headspace. The analysis of volatile products was performed on a Shimadzu GCMS-QP2010 fitted with an Agilent DB-1 column (dimethylpolysiloxane, 30 m  0.25 mm i.d., 0.25 μm film thickness). A typical temperature program began with 5 min isothermal operation at 40 °C followed by a 25 °C min
1 ramp to 200 °C. The inlet and ion source temperatures were 250 and 260 °C, respectively. Calculation of Model Structures and Vibrational Spectra. Computations were performed using the DFT implementation in the Gaussian03 code, revision B.05.51 The B3PW91 density functional was used, with a mixed basis set for the orbitals. A fully uncontracted basis set, based on LANL2DZ, was used for the valence electrons of Re,52 augmented by two f functions (ζ = 1.14 and 0.40) in the full optimization. Re core electrons were treated by the Hay-Wadt relativistic effective core potential (ECP) given by the standard LANL2 parameter set (electron
electron and nucleus
electron). A 6-31G(d,p) basis set was used to describe the rest of the system. This approach was used successfully to describe the structure of CH3ReO3 attached to an Al-containing silsesquioxane cube, a model for the silica
alumina surface.37,38 Vibrational spectra were simulated using the vibrational frequencies and intensities produced by the DFT calculations. Spectra were generated by applying a Lorentzian function to each calculated frequency and summing over all vibrational frequencies after scaling each Lorentzian function by its corresponding DFT-calculated intensity, eq 3: Raman intensity ¼ ΣðIi ðΓ=2Þ=ððx
xi Þ2 þ ðΓ=2Þ2 ÞÞ

ð3Þ

where Ii is the DFT-calculated Raman activity of the vibration at wavenumber xi and Γ is the full width at half-maximum of the Lorentzian function, set arbitrarily to Γ = 12 cm
1. Vibrations arising primarily from the silsesquioxane framework are omitted from the simulated spectra.

’ RESULTS AND DISCUSSION Comparison of Propene Metathesis Activities. Perrhenates supported on silica, silica
alumina, and γ-alumina were tested for catalytic activity in the self-metathesis of propene, Figure 1. At 25 °C, the reaction is fastest over perrhenate-modified γ-alumina

Figure 1. Kinetics of propene metathesis at 25 °C in a constant-volume batch reactor containing 20.0 mg of perrhenate on γ-alumina (9.5 wt % Re, red solid triangle up), silica
alumina (0.9 wt % Re, purple solid circle), γ-alumina (2.5 wt % Re, black solid square), silica prepared via aqueous impregnation (1.1 wt % Re, blue solid diamond), and silica prepared via anhydrous grafting (2.0 wt % Re, green solid triangle down). Lines are drawn to guide the eye.

with high Re loading (9.5 wt % Re), followed by perrhenatemodified silica
alumina (0.9 wt % Re), then perrhenate-modified γ-alumina with low Re loading (2.5 wt % Re). No reaction was detected for the perrhenate-modified silicas, regardless of preparation method (aqueous impregnation or nonaqueous grafting). These results are consistent with previously reported kinetics of olefin metathesis over supported perrhenate catalysts4,6,9,10 and suggest that specific support interactions may be necessary for activating the catalyst. These dramatic support-dependent differences in activity led us to investigate possible differences in the structures of the supported perrhenates. EXAFS of Supported Perrhenates. Spectra at the Re L3-edge were recorded under air-free conditions for perrhenates that were supported on silica, γ-alumina, and silica
alumina via aqueous impregnation, followed by calcination. The results are compared in R space in Figure 2. Significant differences in the Re environments are apparent for the three supports. While all spectra contain a major peak at 1.4 Å in the R-space magnitude, the intensity of this feature differs for each sample, being largest for silica, followed by silica
alumina and then γ-alumina. In addition, weak features between 2.5 and 3.2 Å, caused by long-range single-scattering and/or multiple-scattering events, are different for each support, with perrhenate on γ-alumina showing the strongest feature in the FT magnitude, at 2.6 Å. These differences in the EXAFS are a starting point to explore possible variations in the perrhenate-support interactions. To create a FEFF model for curve-fitting and to assess the structural accuracy of our DFT calculations, we redetermined the structure of trimethylsilyl perrhenate, (CH3)3SiOReO3. Crystal Structure of Trimethylsilyl Perrhenate. Molecular metallasiloxanes can provide useful references for interpreting the structures and spectroscopic features of supported metal complexes.53,54 In particular, trimethylsilyl perrhenate was recently used to represent the geometry of oxide-supported perrhenates.3 Although its crystal structure was reported some time ago, the quality of that structure is marginal (final refinement factor 0.141).55 Only the Re atom was refined anisotropically. The lengths of the Si
C (2.11, 1.95, 1.69 Å) and RedO 9014

dx.doi.org/10.1021/jp109929g |J. Phys. Chem. C 2011, 115, 9012–9024

The Journal of Physical Chemistry C

ARTICLE

Figure 4. DFT-calculated structures for trimethylsilyl perrhenate, as well as a silsesquioxane monosilanol (model I), and perrhenate attached to the silsesquioxane silanolate (model IRe), representing the surface of silica and perrhenate-modified silica, respectively. Color scheme: Re (purple), O (red), Si (blue), C (black). Atoms of the silsesquioxane supporting framework are shown as Si (dark gray) and O (light gray). With the exception of the silanol proton (pink) in model I, structureterminating hydrogens have been omitted for clarity.

Table 1. Comparison of DFT-Calculated Distances (Å) for Trimethylsilyl Perrhenate, And Perrhenate Attached to One Corner of a Silsesquioxane Cube (Model IRe) (CH3)3SiOReO3

model IRea

RedO

1.692

1.689

Re
O(Si)

1.810

1.834

Re
Si

3.515

3.405

bond

Figure 2. The k3-weighted Re L3-edge EXAFS in R-space for perrhenates supported via aqueous impregnation of silica (1.1 wt % Re, black), γ-alumina (8.5 wt % Re, blue), and silica
alumina (1.0 wt %, red): top, FT magnitude; bottom, imaginary component of the FT.

Figure 3. Crystal structure of (CH3)3SiOReO3. Thermal ellipsoids are drawn at the 30% probability level. Hydrogen atoms are omitted for clarity. Selected bond lengths (Å): Re
O(2), 1.678(10); Re
O(3), 1.671(13); Re
O(4), 1.674(9); Re
O(1), 1.778(7); Si
O(1), 1.684(7); Si
C1, 1.851(9); Si
C2, 1.853(11); Si
C3, 1.819(11). Selected angles (deg): Si
O(1)
Re, 164.0(5).

bonds (1.55, 1.56, 1.71 Å) showed high variability, and the Re— OSi (1.67 Å) bond was reported to be unusually short. We redetermined the crystal structure of trimethylsilyl perrhenate, Figure 3. As expected, all three Si
C (1.851, 1.853, 1.819 Å) and terminal RedO (1.678, 1.671, 1.674 Å) bond

a

Perrhenate with local C3v symmetry, attached to a silsesquioxane cube (see Figure 4).

lengths are similar, while the Re
O1 bond distance (1.778 Å) is significantly longer than any of the three RedO multiple bonds. The Si
O1
Re angle is 164°; all other angles at Si and Re are close to their expected tetrahedral values. The final refinement factor for this structure is 0.054. Crystal data and details of the structure refinement are reported in Table S1 in the Supporting Information. Computational Modeling of Trimethylsilyl Perrhenate and Silica-Supported Perrhenate. To validate our computational approach for Re, we first calculated the structure of an isolated molecule of (CH3)3SiOReO3, Figure 4. The agreement between the calculated structure, Table 1, and the crystal structure is very good, with the exception of the Re
O
Si angle. The calculated value (174°) is significantly larger than the observed value (164°), resulting in a longer calculated Re
Si (nonbonded) distance. Packing effects apparent in the crystal structure, Figure S1 (Supporting Information), may cause the Re
O
Si unit to bend, resulting in a more acute angle in the crystal than for the isolated molecule. A silsesquioxane cube is a better electronic model for a silica surface than the trimethylsilyl fragment, due to its more oxygenrich environment.56 The cube also constrains the Si
O
Si angles and precludes rotation about the Si
O
Si bonds. To create a model for silica-supported perrhenate, the terminal hydroxyl group of a silsesquioxane monosilanol (model I) was replaced by perrhenate. The energy-minimized structure, also shown in Figure 4, is designated model IRe. Its RedO multiple bonds, 1.689 Å, are slightly shorter than those in (CH3)3SiOReO3, 9015

dx.doi.org/10.1021/jp109929g |J. Phys. Chem. C 2011, 115, 9012–9024

The Journal of Physical Chemistry C

ARTICLE

Table 2. Comparison of Bond Distances (d) Determined by Neutron and/or X-ray Diffraction with EXAFS Curvefit Distances (R) for Perrhenate Model Compounds compound NH4ReO4c (CH3)3SiOReO3

d

Na

RedO

4

1.73457
59 1.73 (0.01) 0.0014 (0.0002)

RedO

3

1.674e

1.71 (0.01) 0.0009 (0.0004)

Re
O

1

1.778 e

1.83 (0.02) 0.0037 (0.0027)

d (Å)

R (Å)b

σ2 (Å2)b

path

During the EXAFS fits, N was fixed at the integer values shown. S02 was fixed at 0.923. b Errors in the fit parameters are shown in parentheses. c For this fit, 3.5 e k e 14.0 Å
1; 0.9 e R e 2.2 Å; ΔE0 = (5.6 ( 2.5) eV. d For this fit, 3.5 e k e 13.0 Å
1, 0.9 e R e 2.0 Å; ΔE0 = (4.7 ( 1.3) eV. e This work. a

Figure 5. The k3-weighted Re L3-edge EXAFS in R-space (FT magnitude, red circles; imaginary component, black circles) for (a) NH4ReO4 and (b) trimethylsilyl perrhenate, showing curvefits (blue lines) generated using models based on their neutron diffraction and single-crystal X-ray diffraction structures, respectively.

while the Re
O(Si) bond length, 1.834 Å, is slightly longer, as expected due to greater electron-withdrawing character of the silsesquioxane ligand. The more acute Re
O
Si bond angle, 156°, results in a shorter Re
Si distance, 3.405 Å, compared to (CH3)3SiOReO3. This is also as expected because the electronpoor silsesquioxane silanolate provides less effective dπ
pπ overlap between orbitals on Re and the bridging oxygen. EXAFS of Model Compounds. First, we examined the EXAFS of polycrystalline NH4ReO4 and (CH3)3SiOReO3, in order to validate our curve-fitting procedure. For NH4ReO4, the curvefit to a model with Td symmetry and a single oxygen shell (model 0Re, N = 4) agrees well with the data, Figure 5a (R-space) and Figure S3 (Supporting Information) (k-space); all fit parameters are physically reasonable, Table 2. The fitted Re
O path length, 1.73 Å, agrees with the literature value of 1.734 Å,57
59 and the mean-square displacement, 0.0014 Å2, is as expected for a singlescattering path. The EXAFS of trimethylsilyl perrhenate is shown in Figure 5b. Despite having precisely the same coordination number (4) as NH4ReO4, (CH3)3SiOReO3 shows lower R-space intensity at 1.4 Å, as expected based on its lower symmetry (and the consequent interference between inequivalent Re
O paths). Analysis of this spectrum with the single-shell (0Re) model used to fit the NH4ReO4 spectrum resulted in a curvefit that does not reproduce the shape of the feature at 1.4 Å. A model with local C3v symmetry, involving three RedO multiple bonds and one

longer Re
O(Si) bond (model IRe), produced a much better fit, with mean-square displacements for both paths within the expected range for single-scattering paths, Table 2. The fitted bond distances are slightly longer than the RedO and Re— O(Si) distances determined by X-ray diffraction. Our attempts to fit the small feature observed at ca. 2.6 Å in R-space to a Re
Si path, an intermolecular Re
O path, or Re
O
O multiplescattering, were unsuccessful. It is likely that the feature arises from some combination of these paths. A complete list of the EXAFS curvefits attempted, showing the criteria for their acceptance or rejection, is presented in Table 3. EXAFS Curvefit Analysis of Silica-Supported Perrhenates. The EXAFS of perrhenate-modified silica prepared via aqueous impregnation is shown in Figure 6a. The curvefit to model 0Re matches the data well. While the curvefit generated using model IRe also matches the data, Figure S4 (Supporting Information), the mean-square displacement for the long Re
O path (N = 1, R = 1.77 Å) is negative, Table 4, indicating that this model does not accurately represent the structure. Consequently, we infer that perrhenate does not react with the hydroxyl groups of silica when dispersion is achieved from an aqueous solution, even after calcination to obtain an anhydrous material. A similar result was reported for the EXAFS of the Re sites in a calcined Pt
ReO x/SiO 2 catalyst prepared via aqueous impregnation.60 In order to covalently anchor perrhenate to silica, we developed an anhydrous alternative to the conventional aqueous impregnation method of preparation. First, dry silica was exposed to hexamethyldisilazane vapor, which caps terminal hydroxyls with trimethylsilyl groups, eq 4.61 Exposure of this material to volatile trimethylsilyl perrhenate vapor resulted in displacement of hexamethyldisiloxane, eq 5, yielding a perrhenate-modified silica (2.0 wt % Re). 2tSiOH þ ðCH3 Þ3 SiNðHÞSiðCH3 Þ3 f 2tSiOSiðCH3 Þ3 þ NH3 v

ð4Þ

tSiOSiðCH3 Þ3 þ ðCH3 Þ3 SiOReO3 f tSiOReO3 þ ðCH3 Þ3 SiOSiðCH3 Þ3 v

ð5Þ

The formation of hexamethyldisiloxane was confirmed by GC/ MS analysis of the reactor headspace. The EXAFS of the anhydrously modified silica is shown in Figure 6b. The maximum intensity of the feature at 1.4 Å in the FT magnitude spectrum closely resembles that of trimethylsilyl perrhenate itself, but it is lower than that of the perrhenatemodified silica prepared via aqueous impregnation, suggesting a lower symmetry for the anhydrously grafted sites. A curvefit 9016

dx.doi.org/10.1021/jp109929g |J. Phys. Chem. C 2011, 115, 9012–9024

The Journal of Physical Chemistry C

ARTICLE

Table 3. Comparison of EXAFS Curvefits for Perrhenate-Containing Materials to the Various Computational Models and Decisions on Their Acceptance/Rejection intensitya

model

NH4ReO4

15.6

0Re

fit accepted

(CH3)3OReO3

11.9

0Re IRe

failed to reproduce the shape of the feature at 1.4 Å fit accepted

Re/silica (aqueous)

14.5

0Re

fit accepted

IRe

negative σ2 for second Re
O shell

0Re

failed to reproduce the shape of the feature at 1.4 Å

IRe

fit accepted

0Re

model inconsistent with strong adsorption of perrhenate and its activity for olefin metathesis

IRe

unrealistically small σ2 for RedO path; negative σ2 for Re
O(Si) path; Re
Si distance unrealistically short

IIRe 0Re

fit accepted does not account for feature at 2.6 Å

IRe

large σ2 for Re
O(Al) path; Re
O(Al) distance unrealistically long; Re
Al distance unrealistically short

IIIRe

fit accepted

sample

Re/silica (anhydrous) Re/silica
alumina

Re/γ-alumina

a

13.3 13.7

11.8

curvefit evaluation

Intensity of the first R-space peak in the k3-weighted EXAFS spectrum.

Figure 6. The k3-weighted Re L3-edge EXAFS in R-space (FT magnitude, red circles; imaginary, black circles) for (a) perrhenate-modified silica made via aqueous impregnation (1.1 wt % Re), showing the curvefit (blue lines) to the Td-symmetric model 0Re and (b) perrhenatemodified silica made via anhydrous grafting of trimethylsilyl perrhenate (2.0 wt % Re), showing the curvefit (blue lines) generated using the locally C3v-symmetric model IRe.

using model 0Re did not reproduce the shape of the feature at 1.4 Å. However, the curvefit obtained using model IRe matches the data well, giving a short RedO path (N = 3) at 1.71 Å and a

longer Re
O path (N = 1) at 1.81 Å. There is a very weak feature in the EXAFS data at ca. 2.5 Å. We were unable to fit it using the Re
Si (nonbonded) path in model IRe; it likely arises from some combination of Re
Si and Re
O
O paths. This suggests that the second coordination sphere of Re is not well-ordered, presumably due to the lack of additional interactions between perrhenate and support atoms.37,38,47 EXAFS Curvefit Analysis of Perrhenates Supported on Silica
Alumina and γ-Alumina. The EXAFS of perrhenatemodified silica
alumina is shown in Figure 7a. The fit to model 0Re matches the data well between 0.9 and 1.8 Å in R-space. However, Td symmetry would imply that the perrhenate sites are largely unperturbed by the support. Since the tetrahedral perrhenate sites on silica (vide supra) show no activity for olefin metathesis (Figure 1), and since perrhenate supported on silica
alumina is active, we infer that the structure of the active Re sites on silica
alumina must differ from that of the inactive Re sites on silica. The curvefit generated using model IRe is not satisfactory; it yields either very small or negative mean-square displacements for both the RedO and Re—O(Si) paths, Table 5. This result suggests that one or more scattering paths are missing from the curvefit. Finally, the weak feature at ca. 2.6 Å in R-space can be reproduced using a Re
Si single-scattering path, suggesting that the second coordination sphere of this grafted perrhenate is more ordered than that of anhydrous perrhenate on silica, although the σ2 value for this path is large. We defer further analysis of the spectrum until a new model for the grafted site is developed (vide infra). The EXAFS of perrhenate-modified γ-alumina (8.5 wt % Re) is shown in Figure 7b. The EXAFS clearly differs from that of both perrhenate-modified silicas in the low intensity of the first peak in R-space. In addition, the feature at 2.6 Å in the Fourier transform magnitude is much more intense. However, the EXAFS curvefit using model IRe (modified appropriately with a Re
Al path instead of a Re
Si path) is unsatisfactory (Table 5). In particular, the fitted distance for the Re
O(Al) bond (2.10 Å) is unreasonably long compared to the Re
O(Si) distance for perrhenate-modified silica prepared via anhydrous grafting, 1.81 Å. In addition, the fitted Re
Al distance, 3.05 Å, is much shorter than either the Re
Si distance observed in trimethylsilyl perrhenate (3.515 Å) or the Re
Si distance calculated for model IRe (3.405 Å). The short Re
Al distance 9017

dx.doi.org/10.1021/jp109929g |J. Phys. Chem. C 2011, 115, 9012–9024

The Journal of Physical Chemistry C

ARTICLE

Table 4. Curvefit Parametersa for the EXAFS of Perrhenate-Modified Silicas deposition method

model

path

N

R (Å)

σ2 (Å2)

aqueous impregnation

0Reb

RedO

4

1.72 (0.01)

0.0017 (0.002)

IRe c

RedO

3

1.69 (0.01)

0.0007 (0.0012)

Re
O

1

1.77 (0.02)


0.0020 (0.0016)

anhydrous grafting

IRe d

RedO

3

1.71 (0.01)

0.0012 (0.0003)

Re
O

1

1.81 (0.02)

0.0041 (0.0015)

a N was fixed at the integer values shown, and S02 was fixed at 0.923. Standard errors in the fit parameters are reported in parentheses. b For this fit, 3.0 e k e 13.5 Å
1; 1.0 e R e 2.2 Å; ΔE0 = 2.7 ( 1.5 eV. c For this fit, 3.0 e k e 13.5 Å
1, 1.0 e R e 2.2 Å; ΔE0 = (0.8 ( 2.3) eV. d For this fit, 3.0 e k e 14.5 Å
1, 1.0 e R e 2.2 Å; ΔE0 = (2.2 ( 4.5) eV.

Table 5. Curvefit Parametersa Obtained Using Model IRe To Analyze the EXAFS of Perrhenate-Modified Silica-Alumina and γ-Alumina support

path

N

R (Å)

σ2 (Å2)

silica
alumina b

RedO Re
O(Si) Re
Si RedO Re
O(Al) Re
Al

3 1 1 3 1 1

1.70 (0.01) 1.80 (0.02) 3.06 (0.07) 1.73 (0.002) 2.10 (0.03) 3.05 (0.02)

0.0005 (0.0008)
0.0097 (0.0016) 0.0121 (0.0101) 0.0020 (0.0003) 0.0100 (0.0054) 0.0019 (0.0015)

γ-alumina c

a N was fixed at the integer values shown, and S02 was fixed at 0.923. Standard errors in the fit parameters are reported in parentheses. b 1.0 wt % Re. For this fit, 3.0 e k e 14.0 Å
1; 0.9 e R e 3.2 Å; ΔE0 = (2.6 ( 2.4) eV. c 8.5 wt % Re. The Re
Si path found in model IRe was redefined as a Re
Al path. For this fit, 3.0 e k e 14.5 Å
1; 0.9 e R e 3.3 Å; ΔE0 = (2.3 ( 1.6) eV.

Figure 7. The k3-weighted Re L3-edge EXAFS in R-space (FT magnitude, red circles; imaginary, black circles) showing curvefits (blue lines) generated using model IRe, for (a) perrhenate-modified silica
alumina (1.0 wt % Re) and (b) perrhenate-modified γ-alumina (8.5 wt % Re).

indicates a much more acute Re
O
Al angle than exists in model IRe for the analogous Re
O
Si angle. Very similar curvefits for the EXAFS of perrhenate on γ-alumina were recently reported.40,42 The geometric implications of the very long fitted Re
O(Al) bond length (2.08 Å) and the short fitted Re
Al distance (3.03 Å) were not discussed, although together they imply a rather acute ReOAl angle (ca. 100°, assuming an Al
O bond length of 1.86 Å).62 On the basis of these curvefit results, we conclude that the perrhenates supported on silica
alumina and γ-alumina are poorly described by structures closely related to model IRe, in which perrhenate interacts with the support via a single bridging oxygen. Unlike silica, both silica
alumina and γ-alumina possess Lewis acid sites. In order to generate plausible alternative

structures for these supported perrhenates, models were constructed that contain both Lewis and Brønsted acid sites. Computational Modeling of Silica
alumina, And Its Interactions with Perrhenate. When the Al content is low, silica
alumina is believed to contain isolated Lewis acid sites,63,64 at least some of which are located in close proximity to strong Brønsted acid sites.19,20 All of the hydroxyl groups are terminal, and are attached to Si.65 Therefore, a computational model for the silanol-terminated surface of dehydrated silica
alumina was constructed using a silsesquioxane cube with an Al-corner adjacent to the silanol (model II), Figure 8.38 To keep the cluster neutral while maintaining tetracoordination at Al (as demonstrated by Al K-edge XANES),64,66 the Al-substituted corner was capped with a neutral siloxane ligand, H3SiOSiH3. Omission of the capping siloxane did not substantially change the results, but is considered to be less physically meaningful. The terminal hydroxyl group was replaced by perrhenate, and the structure was reoptimized in order to generate a model for perrhenate-modified silica
alumina (Figure 8, model IIRe). In addition to the expected Si-ORe anchoring bond, two additional interactions between the perrhenate and the Al-substituted cube are evident. First, the Lewis acidic Al interacts with the perrhenate, resulting in significant elongation of one RedO bond (to 1.799 Å), compared to the two noninteracting RedO multiple bonds (1.679, 1.680 Å), Table 6. Second, a bridging oxygen (AlOSi) coordinates to the Lewis acidic Re center, at a Re
O distance of 2.078 Å. Consequently, the nonbonded Re
Si distance, 2.798 Å, is much shorter than that found in model IRe (3.405 Å). EXAFS Curvefit Analysis of Perrhenate-Modified Silica
Alumina, Using Model IIRe. The curvefit of model IIRe to the EXAFS of perrhenate-modified silica
alumina is shown in Figure 9. An attempt to include two unique Re
O shells (N = 2), 9018

dx.doi.org/10.1021/jp109929g |J. Phys. Chem. C 2011, 115, 9012–9024

The Journal of Physical Chemistry C

ARTICLE

Figure 8. DFT-calculated structures for a modified silsesquioxane cube as a model for the surface of silica
alumina (model II), as well as its perrhenate-modified analogue (model IIRe). Color scheme: Re (purple), O (red), Al (green), Si (blue), H (pink). Noninteracting atoms of the silsesquioxane framework are shown as Si (gray) and O (light gray). Capping hydrogen atoms have been omitted for clarity.

Table 6. Comparison of the Calculated Structure and Curvefit Parameters for the EXAFS of Perrhenate-Modified Silica
Alumina (1.0 wt % Re) EXAFS fit to model IIRea

model IIRe path

d (Å)

RedO

1.679

Re
O

1.799, 1.920

Nb

R (Å)

σ2 (Å2)

4

1.72 (0.01)

0.0024 (0.0002)

Re
O(support)

2.088

1

2.14 (0.02)

0.0054 (0.0020)

Re
Si

2.798

1

2.75 (0.05)

0.0154 (0.0078)

Re
Al

3.046

1

3.06 (0.04)

0.0087 (0.0041)

S0 was fixed at 0.923. Standard errors are shown in parentheses. For this fit, 3.0 e k e 14.0 Å
1; 0.9 e R e 3.2 Å; ΔE0 = (
0.3 ( 1.1) eV. b N was fixed at the integer values shown.

Figure 9. The k3-weighted Re L3-edge EXAFS in (a) R-space (FT magnitude, red circles; imaginary, black circles) and (b) k-space (black circles), for perrhenate-modified silica
alumina (1.0 wt % Re), showing the curvefit (blue lines) obtained using model IIRe.

one for the RedO ligands and one for the bridging oxygens (Re
O(Al/Si)), resulted in bond distances of 1.68 and 1.75 Å. These differ by less than the resolution of the EXAFS data set (ca. 0.1 Å). In addition, their difference in the DFT model may be exaggerated by the highly convex nature of the silsesquioxane cube model for the silica
alumina surface. For curvefitting, the two paths were therefore combined into a single scattering path (N = 4), resulting in an average distance of 1.72 Å. A long Re
O scattering path at 2.14 Å (N = 1), attributed to the interaction between Re and a support oxygen, was included in the EXAFS fit. The fit parameters are shown in Table 6. Minor differences between the Re
O and Re
O(support) distances for the curvefit and the model IIRe may be artifacts of the silsesquioxane model. The Re
O(support) interaction fixes the Re position relative to the silica
alumina surface. It follows that scattering paths involving the next-nearest-neighbors (i.e., Re
Si and Re
Al) should contribute significantly to the EXAFS. However, there is little R-space intensity in the region where these paths are expected (ca. 2.5 Å). We will return to a discussion of their contributions after further analysis of the EXAFS of perrhenate-modified γ-alumina (vide infra). Models for the γ-Alumina Surface and Its Interactions with Perrhenate. The dehydrated and partially dehydroxylated surface of γ-alumina is believed to contain two types of strong Lewis acid sites: four-coordinate, distorted tetrahedral Al sites;

five-coordinate, square pyramidal Al sites.67 In a DFT study of γalumina dehydration, tricoordinate Al sites appeared in appreciable amounts only after heating to 650 °C,68 and are therefore not considered further here. To construct a simple model for perrhenate-modified γ-alumina, model IIRe was used as a starting point. It was constructed by replacing the silanol (SiOH) corner of IIRe by an AlOH corner. Terminal hydroxyls are the least Brønsted acidic hydroxyl sites present on the γ-alumina surface69 but are reported to be the first to react with deposited perrhenate.16 To maintain overall charge neutrality, a [SiH3þ] fragment was added to a bridging oxygen of the cube framework. Although it does not attempt to model the γ-alumina subsurface, this silsesquioxane-based structure allows us to investigate potential interactions between the [Al
O
Al
OH] surface fragment and perrhenate, in order to extract bond distances and vibrational frequencies for comparison to experimental spectra. Figure 10 shows perrhenate attached to this structure, as model IIIRe. The interactions are similar to those predicted for perrhenate-modified silica
alumina in model IIRe (Figure 8). Re is bonded to both Al atoms via bridging oxygens, only one of which was derived from a terminal hydroxyl group. Coordination of the Al
O
Al bridging oxygen results in a pentacoordinate Re site with local symmetry close to C2v. Consequently, the local structure at Re is very similar to model IIRe, except that Re has two

a

2

9019

dx.doi.org/10.1021/jp109929g |J. Phys. Chem. C 2011, 115, 9012–9024

The Journal of Physical Chemistry C

ARTICLE

Figure 10. Calculated model structures for a terminal hydroxyl group of γ-alumina, as well as its perrhenate-modified analogue. Models III and IIIRe are DFT-optimized structures of a dialuminum-substituted silsesquioxane cube (framework not shown). Color scheme: Re (purple), O (red), Al (green), H (pink). Noninteracting O atoms (light gray) are also shown. See the Supporting Information for the full cube structure.

Al next-nearest neighbors instead of one Al and one Si next-nearest neighbor. A closely related structure was proposed as the precursor of the active site in another olefin metathesis catalyst, MoO3/ Al2O3.70 EXAFS Curvefit Analysis of Perrhenate-Modified γ-Alumina, Using Model IIIRe. The EXAFS of perrhenate-modified γ-alumina was analyzed using a FEFF model based on model IIIRe. The curvefit is shown in Figure 11. An attempt to fit two separate paths for the shorter RedO bonds (N = 2) and the longer bridging Re
O bonds (N = 2) resulted in distances of 1.70 and 1.75 Å, respectively. Since this difference is less than the resolution of the EXAFS data (ca. 0.1 Å), the two paths were combined and fit as a single RedO path (N = 4), resulting in an average distance of 1.73 Å, Table 7. A much longer Re
O bond (N = 1) at 2.05 Å is consistent with interaction of the Re center with an O in the γ-alumina framework. The Re
Al path (N = 2) reproduces the feature at 2.6 Å in the FT magnitude. Its short distance (3.04 Å) is expected due to the acute Re
O
Al angle in model IIIRe. When N for this path was refined simultaneously in all kn-weights (n = 1, 2, 3), the best value was found to be (2.0 ( 0.9). (The error is large because the variables N and σ2 are highly correlated.) Minor discrepancies between the fit parameters and the DFT-optimized model are likely due to the limited ability of the model to reflect the geometry of the γ-alumina surface, as well as heterogeneity in the Lewis acid sites. Effect of Re Loading on the Structures of Perrhenates Supported on γ-Alumina. Perrhenate-modified γ-alumina exhibits a highly nonlinear relationship between Re loading and catalytic activity in olefin metathesis.71 Some studies have suggested that an extended oxide overlayer forms at high Re loading (close to monolayer coverage)72 and that Re oxide oligomers were required to generate the active sites for olefin metathesis.73
75 If such structures were present, the EXAFS should contain evidence of Re
Re scattering paths. The material containing 8.5 wt % Re whose EXAFS is described above contains close to the maximum Re loading for this γ-alumina (9.5 wt %); however, attempts to fit a Re
Re path to the EXAFS feature at 2.6 Å were unsuccessful. Furthermore, the EXAFS of a perrhenate-modified γ-alumina containing only 2.5 wt % Re, far below monolayer coverage, showed identical intensity for the feature at 2.6 Å in the Fourier transform, Figure 12. These results are consistent with a Raman investigation that found no evidence for Re
O
Re linkages, even at high Re loading,12 although they raise an interesting question about the origin of the strong effect of loading on reactivity.

Figure 11. The k3-weighted Re L3-edge EXAFS for perrhenate-modified γ-alumina (8.5 wt % Re), in (a) R-space (FT magnitude, red circles; imaginary, black circles); and (b) k-space (black circles), showing the curvefit (blue lines) generated using model IIIRe.

Table 7. Comparison of Calculated and Curvefit Parameters for the EXAFS of Perrhenate-Modified γ-Alumina (8.5 wt % Re) EXAFS fita

model IIIRe path

d (Å)

Nb

R (Å)

σ2 (Å2)

4

1.73 (0.05)

0.0035 (0.0002)

RedO

1.685

Re
O

1.817, 1.885

Re
O(Al)

2.082

1

2.05 (0.02)

0.0065 (0.0027)

Re
Al

2.860, 3.012

2

3.05 (0.02)

0.0067 (0.0020)

S02 was fixed at 0.923. Standard errors are shown in parentheses. For this fit, 3.0 e k e 14.5 Å
1; 0.9 e R e 3.3 Å; ΔE0 = (1.9 ( 1.4) eV. b N was fixed at the integer values shown. a

The curvefit of model IIIRe to the EXAFS of the γ-alumina with low Re loading (2.5 wt % Re) resulted in similar fit parameters to those obtained for the EXAFS of γ-alumina containing 8.5 wt % Re (see Supporting Information). This suggests that Re is fully dispersed, not present as an extended oxide phase, in both materials. The origin of the nonlinear relationship between Re loading and catalytic activity is therefore unlikely to involve the greater reducibility of ReOx aggregates. An alternate explanation lies in the heterogeneity of the grafting sites present on γ-alumina.67 9020

dx.doi.org/10.1021/jp109929g |J. Phys. Chem. C 2011, 115, 9012–9024

The Journal of Physical Chemistry C

Figure 12. Comparison of the k3-weighted Re L3-edge EXAFS in Rspace (FT magnitude) for γ-alumina modified with perrhenate at loadings of 8.5 wt % Re (blue) and 2.5 wt % Re (red).

Figure 13. Possible structures for a bridging hydroxyl site on γ-alumina (model IV) and its perrhenate-modified analogue (model IVRe). Model IV was taken from an alumina slab calculation.68 Neither structure was reoptimized.

According to the IR spectra, the hydroxyl groups of γ-alumina that are most reactive toward perrhenate are the terminal (basic) hydroxyls.16 These sites are therefore populated first (i.e., occupied at low Re loading), but the resulting grafted perrhenates are inactive or only weakly active.16 Perrhenate reacts with bridging (acidic) hydroxyls to give highly active sites only after the inactive sites have formed.16 However, the strong similarity of the EXAFS for γ-alumina containing both 2.5 and 8.5 wt % Re suggests that the average local structures of the supported perrhenates are similar for both materials. In order to generate a plausible structure for perrhenate grafted via a bridging hydroxyl on γ-alumina, a cluster containing such a hydroxyl group was obtained from a published68 slab calculation for a γ-alumina surface (model IV, Figure 13). Replacing Hþ by ReO3þ, and allowing for Lewis acid
base interactions (Re
Osupport and Al
OdRe) similar to those present in models IIRe and IIIRe, the pentacoordinate Re site IVRe was constructed, with a geometry close to that of model IIIRe. The structural similarities between models IIIRe and IVRe, which differ only in their number of Al neighbors, would make it difficult to distinguish the two using EXAFS. Further analysis of high-R features in the EXAFS of perrhenate on silica
alumina. The EXAFS fit for perrhenate on silica
alumina shown in Figure 9 and Table 6 includes nextnearest-neighbor scattering paths which model a minor feature centered at 2.5 Å in R-space. A stronger scattering feature present at approximately the same position in the EXAFS of perrhenate

ARTICLE

Figure 14. Imaginary components of the FT-EXAFS spectrum for perrhenate on silica
alumina, the corresponding fit generated using model IIRe, and the individual contributions of three long-R paths to the fit.

supported on γ-alumina was shown to be a Re
Al singlescattering path (see above). Since both supports are capable of activating perrhenate for olefin metathesis, the active sites in both materials may have similar local structures, justifying the inclusion of such paths in the EXAFS fit for perrhenate on silica
alumina. Although it is not possible to distinguish between Re
Si and Re
Al paths (due to the very similar scattering properties of Si and Al), inspection of model IIRe suggests that the Re
Al path should be the longer of the two. The origin of the weak intensity in the high-R paths in the FTEXAFS of perrhenate on silica
alumina was investigated further. Figure 14 shows the individual Re
Osupport, Re
Si, and Re
Al curvefit paths. The Re
Si and Re
Al paths are clearly out of phase, leading to destructive interference of their EXAFS signals and a smaller feature in the FT magnitude than in the corresponding spectrum of perrhenate on γ-alumina. While inclusion of both Re
Si and Re
Al paths can be justified on chemical grounds, the mean-squared displacement for the Re
Si path, 0.0154 Å2, is large. Removing this path from the EXAFS fit did not significantly change the fit residual, although it did alter the fit parameters for the Re
Al path: R and σ2 increased to 3.10 ( 0.05 Å and 0.0117 ( 0.0066 Å2, respectively. High mean-squared displacements for paths involving nextnearest neighbors of Re suggest that the Re
Si and Re
Al paths are nonuniform, which is not unexpected given the amorphous nature of the silica
alumina support. XANES Comparison of Supported Perrhenates. Differences in the structures and local symmetries of perrhenates interacting with various supports should be evident in the XANES. This expectation is borne out in the XANES recorded at the Re L3edge (see Supporting Information). The spectra for all of the supported materials are dominated by intense white lines whose positions are consistent with a common oxidation state, Re(VII), as expected due to their preparation from perrhenate-based precursors and the avoidance of reducing conditions. However, there are significant differences in intensity, which may arise due to differences in the overall covalency of the Re
O bonds. Changes in the local coordination environment of Re are more readily interpreted by examining changes in the appearance of 9021

dx.doi.org/10.1021/jp109929g |J. Phys. Chem. C 2011, 115, 9012–9024

The Journal of Physical Chemistry C

Figure 15. Comparison of the Re L1-edge XANES for (a) two model compounds, NH4ReO4 (blue) and trimethylsilyl perrhenate (red), and (b) three supported perrhenates, perrhenate-modified silica prepared via aqueous impregnation (1.1 wt % Re, blue), perrhenate-modified silica prepared via anhydrous grafting (2.0 wt % Re, red), and perrhenatemodified γ-alumina (2.5 wt % Re, black).

the XANES at the L1-edge.76 The 2s f 6p excitation is preceded by a pre-edge feature due to a 2s f 5d transition, which is formally allowed only in noncentrosymmetric environments. An intense pre-edge peak is therefore expected for Re in a tetrahedral environment, with decreasing intensity as the symmetry at Re approaches octahedral. The L1 XANES for NH4ReO4, (CH3)3SiOReO3, perrhenatemodified silica (prepared via either aqueous impregnation of perrhenate or anhydrous grafting of trimethylsilyl perrhenate), and perrhenate-modified γ-alumina, are compared in Figure 15. All show a pre-edge peak between 12522 and 12524 eV. The intensity is only slightly higher for NH4ReO4, with Td symmetry, compared to trimethylsilyl perrhenate, with local C3v symmetry. Perrhenate-modified silica prepared via aqueous impregnation has a slightly more intense pre-edge peak than perrhenatemodified silica prepared via anhydrous grafting. This is consistent with our conclusions from EXAFS curvefitting that aqueous impregnation of silica results in a weakly interacting perrhenate that remains approximately tetrahedral, while anhydrous grafting onto silica leads to a supported perrhenate with C3v symmetry. Furthermore, features located at or just above the absorption edge (12535
12600 eV) in the XANES of trimethylsilyl perrhenate and perrhenate-modified silica (prepared by anhydrous grafting) show only minor differences, while the XANES of NH4ReO4 and perrhenate on silica prepared by aqueous impregnation in the same energy range are quite different.

ARTICLE

In the XANES of perrhenate-modified γ-alumina (2.5 wt % Re), the pre-edge peak is significantly less intense than for either of the model compounds or for either of the silica-supported perrhenates. This result is consistent with a previous investigation in which the intensity of the pre-edge feature for perrhenate-modified γ-alumina was found to be lower than that for ammonium perrhenate.12 Since the pre-edge intensity for the γ-alumina-supported perrhenate increased upon exposure to ambient moisture, the authors concluded that the local symmetry at Re increased from C3v to Td as the grafted perrhenate (assumed to be an “AlOReO3” site with no Lewis acid
base interactions) was hydrolyzed.12 However, this interpretation is not consistent with the high pre-edge intensity of the C3v-symmetric perrhenate in the model compound trimethylsilyl perrhenate. The weaker preedge intensity in the XANES of perrhenate-modified γ-alumina appears to be a consequence of local symmetry at Re that is substantially different from C3v. Vibrational Analysis of Models for Perrhenate-Modified Supports. Vibrational spectroscopy has been used extensively to characterize perrhenates dispersed on various oxides.14,15,17 In the IR, the metal
oxo stretching modes occur in a region of intensely absorbing lattice vibrations, thereby preventing direct observation of the vibrational modes of grafted perrhenates. Raman spectra have been reported for perrhenate supported on γ-alumina, although strong fluorescence precluded the observation of a Raman spectrum for perrhenate on silica
alumina.77 The structural differences found here for perrhenates supported on silica, silica
alumina and γ-alumina prompted us to seek differences in their calculated vibrational spectra, in order to assess whether reported spectra are consistent with the proposed structures. Table 8 shows the DFT-predicted vibrational frequencies associated with the Re
O bonds for models IRe, IIRe, and IIIRe. For energy calibration, the predicted and observed78 frequencies of (CH3)3SiOReO3 are also compared. As expected, DFT overestimates the measured vibrational frequencies by 60
80 cm
1. The simulated spectra of trimethylsilyl perrhenate and the three model structures are compared in Figure 16. The most intense bands are the υs(ReOx) modes, with less intense υa(ReOx) bands at lower energy. Replacing the [(CH3)3Siþ] fragment with the silsesquioxane cube (model IRe) results in an increase in the DFT-predicted frequencies of 7
18 cm
1, consistent with stronger RedO bonds resulting from the more electron-withdrawing supporting ligand; the υa(SiORe) mode is also significantly more intense. In contrast, the simulated spectra for models IIRe and IIIRe show shifts to lower energy relative to trimethylsilyl perrhenate. Although modes involving bridging oxygens (SiORe or AlORe) were not included in the simulated spectra, due to their strong coupling with vibrations of the silsesquioxane cubes, their calculated Raman intensities are low. In the experimental Raman spectrum of perrhenate-modified γ-alumina, υs(ReOx) appears at 1000 cm
1,79 which is slightly lower in energy than for trimethylsilyl perrhenate (1010 cm
1).72 The difference matches the predicted red shifts in the position of υs(ReOx) for models IIRe and IIIRe, in contrast to model IRe, for which υs(ReOx) is predicted to undergo a blue shift. The Raman band assigned to υa(ReOx) in the observed spectrum of perrhenatemodified γ-alumina is much lower in frequency (878 cm
1)79 compared to the same band for trimethylsilyl perrhenate (970 cm
1),72 although the simulated spectra for models IRe, IIRe, and IIIRe all show υa(ReOx) at the same or higher frequency as for trimethylsilyl perrhenate. This finding suggests 9022

dx.doi.org/10.1021/jp109929g |J. Phys. Chem. C 2011, 115, 9012–9024

The Journal of Physical Chemistry C

ARTICLE

Table 8. Comparison of Calculated Raman Frequencies (cm
1) for Perrhenate-Containing Models, With Observed Raman Shifts for Trimethylsilyl Perrhenate and Perrhenate on γ-Alumina mode

(CH3)3SiOReO3a

model IRe

model IIRe

model IIIRe

Re2O7/γ-aluminab

υs(RedOx)

1078 (1010)

1091

1071

1062c

1000

υa(RedOx) υa(SiORe)

1030 (970) 1003 (926)

1037 1021

1043 d

1029 d

878

Values in parentheses are reported Raman shifts for (CH3)3SiOReO3 in the gas phase.78 b Taken from ref 79: perrhenate-modified γ-alumina (1 wt % Re) was calcined at 500 °C and not subsequently exposed to atmospheric moisture.79 c The most intense of three vibrations between 1064 and 1055 cm
1, all of which are highly coupled with modes of the aluminosilsesquioxane cube. d Not reported, due to extensive coupling with modes of the aluminosilsesquioxane cube. a

’ ASSOCIATED CONTENT

bS

Supporting Information. Crystallographic packing diagram and refinement parameters for (CH3)3SiOReO3, additional EXAFS curve fits, Cartesian coordinates for DFT model structures, and CIF coordinates for Me3SiORe3. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Email: [email protected].

Figure 16. Simulated Raman spectra for (a) trimethylsilyl perrhenate (red), (b) model IRe (blue), (c) model IIRe (black), and (d) model IIIRe (green).

that the origin of the 878 cm
1 band may be more complex than previously assumed.

’ CONCLUSIONS The structures of perrhenates supported on silica, silica
alumina, and γ-alumina were probed by EXAFS and XANES, in combination with DFT modeling. While all of these supported perrhenates have been suggested to have local C3v symmetry, similar to trimethylsilyl perrhenate, the Re L3-edge EXAFS reveals subtle but important differences in their coordination environments. On silica, only nonaqueous grafting of perrhenate gives rise to a structure analogous to trimethylsilyl perrhenate, while aqueous impregnation followed by calcination does not appear to produce an anchored perrhenate. On γ-alumina and silica
alumina, at least some of the sites experience additional interactions between the grafted perrhenate and Lewis acid sites of the support, as predicted computationally, resulting in a local symmetry closer to C2v. Simulations of the Raman spectra for perrhenate-supported γ-alumina modeled by IIIRe are consistent with previously reported experimental results. Interactions between perrhenate and Lewis acid sites on γ-alumina and silica
alumina may play a key role in the olefininduced activation of these catalyst precursors for olefin metathesis. Silica contains no Lewis acid sites, and perrhenate-modified silicas, prepared by either aqueous impregnation or anhydrous grafting, are inactive for olefin metathesis. The requirement for a very specific interaction between perrhenate and an adjacent Lewis acid site may explain why so few sites are activated for olefin metathesis on a variety of different supports.

’ ACKNOWLEDGMENT This work was funded by the U.S. Department of Energy, Basic Energy Sciences, Catalysis Science Grant No. DE-FG0203ER15467, and the National Science Foundation under the auspices of the Center for New Technologies Through Catalysis (CENTC). Portions of this work were performed at the Stanford Synchrotron Radiation Lightsource, a national user facility operated by Stanford University on behalf of the U.S. Department of Energy, Office of Basic Energy Sciences. ’ REFERENCES (1) Wachs, I. E.; Deo, G.; Andreini, A.; Vuurman, M. A.; de Boer, M. J. Catal. 1996, 160, 322. (2) Mandelli, D.; van Vliet, M. C. A.; Arnold, U.; Sheldon, R. A.; Schuchardt, U. J. Mol. Catal. A: Chem. 2001, 168, 165. (3) Veljanovski, D.; Sakthivel, A.; Herrmann, W. A.; Kuhn, F. E. Adv. Synth. Catal. 2006, 348, 1752. (4) Andreini, A.; Xiaoding, X.; Mol, J. C. Appl. Catal. 1986, 27, 31. (5) Mol, J. C. J. Mol. Catal. 1994, 90, 185. (6) Mol, J. C. Catal. Today 1999, 51, 289. (7) Chen, X.; Zhang, X.; Chen, P. Angew. Chem., Int. Ed. 2003, 42, 3798. (8) Salameh, A.; Cop
eret, C.; Basset, J.-M.; Bohm, V. P. W.; Roper, M. Adv. Synth. Catal. 2007, 349, 238. (9) Lin, C. J.; Aldag, A. W.; Clark, A. J. Catal. 1976, 45, 287. (10) Andreini, A.; Mol, J. C. J. Mol. Catal. 1988, 46, 151. (11) Ellison, A.; Bickerstaffe, A.; Diakun, G.; Worthington, P. J. Mol. Catal. 1986, 36, 67. (12) Hardcastle, F. D.; Wachs, I. E.; Horsley, J. A.; Via, G. H. J. Mol. Catal. 1988, 46, 15. (13) Wachs, I. E. Catal. Today 1996, 27, 437. (14) Vuurman, M. A.; Stufkens, D. J.; Oskam, A.; Wachs, I. E. J. Mol. Catal. 1992, 76, 263. (15) Wang, L.; Hall, W. K. J. Catal. 1983, 82, 177. (16) Sibeijn, M.; Spronk, R.; van Veen, J. A. R.; Mol, J. C. Catal. Lett. 1991, 8, 201. (17) Vuurman, M. A.; Wachs, I. E. J. Phys. Chem. 1992, 96, 5008. 9023

dx.doi.org/10.1021/jp109929g |J. Phys. Chem. C 2011, 115, 9012–9024

The Journal of Physical Chemistry C (18) Ellison, A.; Coverdale, A. K.; Dearing, P. F. J. Mol. Catal. 1985, 28, 141. (19) Murrell, L. L.; Dispenziere, N. C. Catal. Lett. 1989, 2, 329. (20) Cr
epeau, G.; Montouillout, V.; Vimont, A.; Mariey, L.; Cseri, T.; Mauge, F. J. Phys. Chem. B 2006, 110, 15172. (21) Xu, B.; Sievers, C.; Lercher, J. A.; van Veen, J. A. R.; Giltay, P.; Prins, R.; van Bokhoven, J. A. J. Phys. Chem. C 2007, 111, 12075. (22) Guimond, S.; Abu Haija, M.; Kaya, S.; Lu, J.; Weissenrieder, J.; Shaikhutdinov, S.; Kuhlenbeck, H.; Freund, H. J.; Dobler, J.; Sauer, J. Top. Catal. 2006, 38, 117. (23) Moisii, C.; Curran, M. D.; van de Burgt, L. J.; Stiegman, A. E. J. Mater. Chem. 2005, 15, 3519. (24) Mitra, B.; Gao, X.; Wachs, I. E.; Hirt, A. M.; Deo, G. Phys. Chem. Chem. Phys. 2001, 3, 1144. (25) Schoop, T.; Roesky, H. W.; Noltemayer, M.; Schmidt, H.-G. Organometallics 1993, 12, 571. (26) Voigt, A.; Walawalkar, M. G.; Roesky, H. W. Chem. Rev. 1996, 96, 2205. (27) Ellison, A.; Coverdale, A. K.; Dearing, P. F. Appl. Catal. 1983, 8, 109. (28) Ellison, A.; Coverdale, A. K.; Dearing, P. F. J. Mol. Catal. 1985, 28, 141. (29) Xiaoding, X.; Mol, J. C. J. Chem. Soc., Chem. Commun. 1985, 631. (30) Xiaoding, X.; Mol, J. C.; Boelhouwer, C. J. Chem. Soc., Dalton Trans. 1 1986, 82, 2707. (31) Murrell, L. L.; Dispenziere, N. C. J. Catal. 1989, 117, 275. (32) Liu, X. S. J. Phys. Chem. C 2008, 112, 5066. (33) Doledec, G.; Commereuc, D. J. Mol. Catal. A: Chem. 2000, 161, 125. (34) Herrmann, W. A.; Correia, J. D. G.; Rauch, M. U.; Artus, G. R. J.; Kuhn, F. E. J. Mol. Catal. A: Chem. 1997, 118, 33. (35) Edwards, P. G.; Jokela, J.; Lehtonen, A.; Sillanpaa, R. J. Chem. Soc., Dalton Trans. 1998, 3287. (36) Herrmann, W. A.; Wojtczak, W. A.; Artus, G. R. J.; Kuhn, F. E.; Mattner, M. R. Inorg. Chem. 1997, 36, 465. (37) Moses, A. W.; Raab, C.; Ramsahye, N. A.; Chattopadhyay, S.; Eckert, J.; Chmelka, B. F.; Scott, S. L. J. Am. Chem. Soc. 2007, 129, 8912. (38) Moses, A. W.; Ramsahye, N. A.; Raab, C.; Leifeste, H. D.; Chattopadhyay, S. K.; Chmelka, B. F.; Eckert, J.; Scott, S. L. Organometallics 2006, 25, 2157. (39) Ellison, A.; Diakun, G.; Worthington, P. J. Mol. Catal. 1988, 46, 131. (40) Bare, S. R.; Ressler, T. Adv. Catal. 2009, 52, 339. (41) Rønning, M.; Nicholson, D.; Holmen, A. Catal. Lett. 2001, 72, 141. (42) Bare, S. R.; Kelly, S. D.; Vila, D; Boldingh, E.; Karapetrova, E.; Kas, J.; Mickelson, G. E.; Modica, F. S.; Yang, N.; Rehr, J. J. J. Phys. Chem. C 2011, 115, 5740. (43) Viswanadham, N.; Shido, T.; Sasaki, T.; Iwasawa, Y. J. Phys. Chem. B 2002, 106, 10955. (44) Lobo-Lapidus, R. J.; Gates, B. C. J. Catal. 2009, 268, 89. (45) Moses, A. W.; Leifeste, H. D.; Ramsahye, N. A.; Eckert, J.; Scott, S. L. Catal. Org. React. 2006, 115, 13. (46) Schmidt, M.; Schmidbaur, H. Inorg. Synth. 1967, 9, 149. (47) Deguns, E. W.; Taha, Z.; Meitzner, G. D.; Scott, S. L. J. Phys. Chem. B 2005, 109, 5005. (48) Ravel, B.; Newville, M. J. Synchrotron Radiat. 2005, 12, 537. (49) Newville, M. J. Synchrotron Radiat 2001, 8, 322. (50) Taha, Z. A.; Deguns, E. W.; Chattopadhyay, S.; Scott, S. L. Organometallics 2006, 25, 1891. (51) 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.;

ARTICLE

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.02; Gaussian, Inc.: Wallingford, CT, 2004. (52) Pietsch, M. A.; Russo, T. V.; Murphy, R. B.; Martin, R. L.; Rappe, A. K. Organometallics 1998, 17, 2716. (53) Duchateau, R. Chem. Rev. 2002, 102, 3525. (54) Chandrasekhar, V.; Boomishankar, R.; Nagendran, S. Chem. Rev. 2004, 104, 5847. (55) Sheldrick, G. M.; Sheldrick, W. S. J. Chem. Soc. A 1969, 2160. (56) Magg, N.; Immaraporn, B.; Giorgi, J. B.; Schroeder, T.; B€aumer, M.; D€obler, J.; Wu, Z.; Kondratenko, E.; Cherian, M.; Baerns, M.; Stair, P. C.; Sauer, J.; Freund, H.-J. J. Catal. 2004, 226, 88. (57) Kruger, G. J.; Reynhardt, E. C. Acta Crystallogr. 1978, B34, 259. (58) Powell, B. M.; Brown, R. J. C.; Harnden, A. M. C.; Reid, J. K. Acta Crystallogr., Sect. B: Struct. Sci. 1993, 49, 463. (59) Swainson, I. P.; Brown, R. J. C. Acta Crystallogr., Sect. B: Struct. Sci. 1997, 53, 76. (60) Ebashi, T.; Ishida, Y.; Nakagawa, Y.; Ito, S.; Kubota, T.; Tomishige, K. J. Phys. Chem. C 2010, 114, 6518. (61) Hair, M. L.; Hertl, W. J. Phys. Chem. 1971, 75, 2181. (62) Liz
arraga, R.; Holmstr€om, E.; Parker, S. C.; Arrouvel, C. Phys. Rev. B 2011, 83, 094201. (63) Omegna, A.; van Bokhoven, J. A.; Prins, R. J. Phys. Chem. B 2003, 107, 8854. (64) Omegna, A.; van Bokhoven, J. A.; Prins, R. J. Phys. Chem. B 2005, 109, 9280. (65) Hunger, M.; Freude, D.; Pfeifer, H.; Bremer, H.; Jank, M.; Wendlandt, K. P. Chem. Phys. Lett. 1983, 100, 29. (66) Kato, Y.; Shimizu, K.; Matsushita, N.; Yoshida, T.; Satsuma, A.; Hattori, T. Phys. Chem. Chem. Phys. 2001, 3, 1925. (67) Digne, M.; Sautet, P.; Raybaud, P.; Euzen, P.; Toulhoat, H. J. Catal. 2004, 226, 54. (68) Digne, M.; Sautet, P.; Raybaud, P.; Euzen, P.; Toulhoat, H. J. Catal. 2002, 211, 1. (69) Knozinger, H.; Ratnasamy, P. Catal. Rev.—Sci. Eng. 1978, 17, 31. (70) Handzlik, J.; Sautet, P. J. Catal. 2008, 256, 1. (71) Moulijn, J. A.; Mol, J. C. J. Mol. Catal. 1988, 46, 1. (72) Olsthoorn, A. A.; Boelhouwer, C. J. Catal. 1976, 44, 197. (73) Yide, X.; Jiasheng, H.; Zhiying, L.; Xiexian, G. J. Mol. Catal. 1991, 65, 275. (74) Yide, X.; Xinguang, W.; Yingzhen, S.; Yihua, Z.; Xiexian, G. J. Mol. Catal. 1986, 36, 79. (75) Nakamura, R.; Abe, F.; Echigoya, E. Chem. Lett. 1981, 51. (76) Bart, J. C. J. Adv. Catal. 1986, 34, 203. (77) Spronk, R.; Mol, J. C. New Frontiers in Catalysis; Guczi, L., Solymosi, F., T
et
enyi, P., Eds.; Elsevier: Budapest, Hungary, 1992. (78) Boal, D. H.; Ozin, G. A. Can. J. Chem. 1970, 48, 3026. (79) Kim, D. S.; Wachs, I. E. J. Catal. 1993, 141, 419.

9024

dx.doi.org/10.1021/jp109929g |J. Phys. Chem. C 2011, 115, 9012–9024