Molecular chemistry in a zeolite: Genesis of a zeolite y-supported ruthenium complex catalyst

Article abstract of DOI:10.1021/ja804265r

Dealuminated zeolite Y was used as a crystalline support for a mononuclear ruthenium complex synthesized from cis-Ru(acac)₂(C₂H ₄)₂. Infrared (IR) and extended X-ray absorption fine structure spectra indicated that the surface species were mononuclear ruthenium complexes, Ru(acac)(C₂H₄)₂²⁺, tightly bonded to the surface by two Ru-O bonds at Al³⁺ sites of the zeolite. The maximum loading of the anchored ruthenium complexes was one complex per two Al³⁺ sites; at higher loadings, some of the cis-Ru(acac)₂(C₂H₄)₂ was physisorbed. In the presence of ethylene and H₂, the surface-bound species entered into a catalytic cycle for ethylene dimerization and operated stably. IR data showed that at the start of the catalytic reaction, the acac ligand of the Ru(acac)(C₂H₄)₂²⁺ species was dissociated and captured by an Al³⁺ site. Ethylene dimerization proceeded ～600 times faster with a cofeed of ethylene and H₂ than without H₂. These results provide evidence of the importance of the cooperation of the Al³⁺ sites in the zeolite and the H₂ in the feed for the genesis of the catalytically active species. The results presented here demonstrate the usefulness of dealuminated zeolite Y as a nearly uniform support that allows precise synthesis of supported catalysts and detailed elucidation of their structures.

Full text of DOI:10.1021/ja804265r

Published on Web 09/12/2008
Molecular Chemistry in a Zeolite: Genesis of a Zeolite
Y-Supported Ruthenium Complex Catalyst
Isao Ogino and Bruce C. Gates*
Department of Chemical Engineering and Materials Science, UniVersity of California,
DaVis, California 95616
Received June 5, 2008; E-mail: bcgates@ucdavis.edu
Abstract: Dealuminated zeolite Y was used as a crystalline support for a mononuclear ruthenium complex
synthesized from cis-Ru(acac)2(C2H4)2. Infrared (IR) and extended X-ray absorption fine structure spectra
indicated that the surface species were mononuclear ruthenium complexes, Ru(acac)(C2H4)22+, tightly
bonded to the surface by two Ru-O bonds at Al3+ sites of the zeolite. The maximum loading of the anchored
ruthenium complexes was one complex per two Al3+ sites; at higher loadings, some of the cis-
Ru(acac)2(C2H4)2 was physisorbed. In the presence of ethylene and H2, the surface-bound species entered
into a catalytic cycle for ethylene dimerization and operated stably. IR data showed that at the start of the
2+
catalytic reaction, the acac ligand of the Ru(acac)(C2H4)2 species was dissociated and captured by an
Al3+ site. Ethylene dimerization proceeded ∼600 times faster with a cofeed of ethylene and H2 than without
H2. These results provide evidence of the importance of the cooperation of the Al3+ sites in the zeolite and
the H2 in the feed for the genesis of the catalytically active species. The results presented here demonstrate
the usefulness of dealuminated zeolite Y as a nearly uniform support that allows precise synthesis of
supported catalysts and detailed elucidation of their structures.
standard of uniformity10 and to catalyze ethylene hydrogena-
tion11 and acetylene cyclotrimerization.12
Introduction
Supported catalysts that are molecular analogues1-5 find
industrial application in processes such as alkene polymeriza-
tion6 and methanol carbonylation.7 They offer opportunities for
tailoring of properties by taking advantage of the structure of
the support to control access to catalytic sites.3,8,9 When such
catalysts are uniform in structure, they offer the prospective
advantages of molecular catalysts in solution, including high
selectivity, combined with those of solid catalysts generally,
such as ease of separation from products and lack of corrosion.
Like mononuclear rhodium complexes, mononuclear ruthe-
nium complexes are catalysts for numerous reactions in solu-
tion,13 including olefin metathesis,14-16 hydrogenation,17 asym-
metric hydrogenation,18,19 oxidation,20 C-C bond formation,21,22
23
and activation of CO2 and C-H bonds.24 The rich catalytic
chemistry of mononuclear ruthenium complexes has motivated
many researchers to anchor them to solid supports, typically
Zeolite supports, which are crystalline and offer well-defined
bonding sites for cationic metal complexes, are appealing
because of the prospects they offer for synthesis of unique
supported species, as illustrated in work with a rhodium complex
synthesized from Rh(acac)(C2H4)2 (acac ) C5H7O2-) on dealu-
minated zeolite Y. The supported complex was shown by
temperature-dependent 13C NMR spectroscopy to meet a high
(10) Ehresmann, J. O.; Kletnieks, P. W.; Liang, A.; Bhirud, V. A.;
Bagatchenko, O. P.; Lee, E. J.; Klaric, M.; Gates, B. C.; Haw, J. F.
Angew. Chem., Int. Ed. 2006, 45, 574.
(11) Liang, A. J.; Bhirud, V. A.; Ehresmann, J. O.; Kletnieks, P. W.; Haw,
J. F.; Gates, B. C. J. Phys. Chem. B 2005, 109, 24236.
(12) Kletnieks, P. W.; Liang, A. J.; Craciun, R.; Ehresmann, J. O.; Marcus,
D. M.; Bhirud, V. A.; Klaric, M. M.; Haymann, M. J.; Guenther, D. R.;
Bagatchenko, O. P.; Dixon, D. A.; Gates, B. C.; Haw, J. F.
Chem.sEur. J. 2007, 13, 7294.
(13) Murahashi, S. Ruthenium in Organic Synthesis; Wiley-VCH: Wein-
heim, Germany, 2004.
(14) Trnka, T. M.; Grubbs, R. H. Acc. Chem. Res. 2001, 34, 18.
(15) Grubbs, R. H. Angew. Chem., Int. Ed. 2006, 45, 3760.
(16) Nguyen, S. T.; Grubbs, R. H.; Ziller, J. W. J. Am. Chem. Soc. 1993,
115, 9858.
(1) Guzman, J.; Gates, B. C. Dalton Trans. 2003, 3303.
(2) Fierro-Gonzalez, J. C.; Kuba, S.; Hao, Y.; Gates, B. C. J. Phys. Chem.
B 2006, 110, 13326.
(3) Corma, A. Catal. ReV. Sci. Eng. 2004, 46, 369.
(4) Tada, M.; Iwasawa, Y. Chem. Commun. 2006, 2833.
(5) Surface and Interfacial Organometallic Chemistry and Catalysis;
Cope´ret, C., Chaudret, B., Eds.; Topics in Organometallic Chemistry,
Vol. 16; Springer: Berlin, 2005.
(17) Evans, D.; Osborn, J. A.; Jardine, F. H.; Wilkinson, G. Nature 1965,
208, 1203.
(18) Noyori, R.; Hashiguchi, S. Acc. Chem. Res. 1997, 30, 97.
(19) Noyori, R. Angew. Chem., Int. Ed. 2002, 41, 2008.
(20) Thompson, M. S.; Meyer, T. J. J. Am. Chem. Soc. 1982, 104, 4106.
(21) Trost, B. M.; Toste, F. D.; Pinkerton, A. B. Chem. ReV. 2001, 101,
2067.
(6) Hlatky, G. G. Chem. ReV. 2000, 100, 1347.
(7) Yoneda, N.; Kusano, S.; Yasui, M.; Pujado, P.; Wilcher, S. Appl.
Catal., A 2001, 221, 253.
(22) Kondo, T.; Takagi, D.; Tsujita, H.; Ura, Y.; Wada, K.; Mitsudo, T.
Angew. Chem., Int. Ed. 2007, 46, 5958.
(8) Notestein, J. M.; Solovyov, A.; Andrini, L. R.; Requejo, F. G.; Katz,
A.; Iglesia, E. J. Am. Chem. Soc. 2007, 129, 15585.
(9) Hicks, J. C.; Mullis, B. A.; Jones, C. W. J. Am. Chem. Soc. 2007,
129, 8426.
(23) Jessop, P. G.; Ikariya, T.; Noyori, R. Nature 1994, 368, 231.
(24) Murai, S.; Kakiuchi, F.; Sekine, S.; Tanaka, Y.; Kamatani, A.; Sonoda,
M.; Chatani, N. Nature 1993, 366, 529.
9
13338 J. AM. CHEM. SOC. 2008, 130, 13338–13346
10.1021/ja804265r CCC: $40.75
2008 American Chemical Society

Genesis of a Zeolite Y-Supported Ru Complex Catalyst
A R T I C L E S
metal oxides.25-27 However, there is still a lack of examples
providing key structural information about such catalysts,
including details concerning metal-support bonding. This
limitation is a reflection of the nonuniformity of the structures,
particularly that of the support surfaces.
In the present work, we have demonstrated that dealuminated
zeolite Y is an excellent support for ruthenium complex catalysts
and taken advantage of the uniformity of the zeolite to elucidate
the catalyst structure in detail. The catalyst was synthesized by
28
reaction of cis-Ru(acac)2(C2H4)2 (I) with the zeolite and its
structure characterized by infrared (IR) and extended X-ray
absorption fine structure (EXAFS) spectroscopies. The complex
enters into a catalytic cycle for ethylene dimerization, aided by
H2; the zeolite surface facilitates the genesis of the catalytically
active species.
Figure 1. IR spectra of calcined dealuminated zeolite HY (lower spectrum)
and the supported ruthenium complex, Sample 1 (upper spectrum).
Results
Synthesis of the Precursor I. The orange-colored crystals were
obtained in 43% yield by a reported method.28
Characterization of Surface Species Formed by Reaction
of Complex I with the Zeolite. Observations during the
Synthesis. The color of the synthesis mixture initially contain-
ing the zeolite and precursor I changed from yellow to dark-
orange within a few hours after the start of mixing. After
18 h, the solution had become colorless, consistent with the
complete uptake of I by the zeolite, corresponding to
approximately one Ru atom per unit cell of the zeolite
(approximately one Ru atom per eight zeolite supercages)
and a Ru/Al atomic ratio of ∼1/6. Removal of the solvent
by evacuation gave a solid with a light-pink color. Evidently,
the ruthenium was bonded to the zeolite. The as-prepared
sample was designated as Sample 1.
IR Evidence for Reaction of Zeolite Silanol Groups with I.
Figure 2. IR spectra of Sample 1 in 50% H2 in He flowing at room
temperature. The top spectrum corresponds to the sample before the
treatment in H2. The arrows indicate the directional changes of the
absorbance as a function of time.
The IR spectra indicated the presence of terminal (3743 cm-1
)
and acidic (3629 and 3565 cm-1) silanol groups in the
calcined zeolite.29 The bands characteristic of acidic silanol
groups decreased significantly in intensity relative to those
of the bare zeolite upon chemisorption of I, indicating
reaction of these silanol groups in the chemisorption process
(Figure 1).30 The band at 3629 cm-1 was assigned to OH
groups within the zeolite supercages.29,31 The critical diameter
of the precursor, ∼7 Å,28 was small enough to allow passage
of I through the zeolite apertures and into the supercages,
where it reacted with the OH groups. The other OH band, at
3565 cm-1, is representative of groups located within the
smaller zeolite ꢀ-cages, which were inaccessible to the
precursor.29,31 The decrease in the intensity of this band thus
suggests that ligands dissociated from ruthenium in I reacted
with these sites.
complicated by the possible overlap of bands due to ligands on
the complex with C-H vibrational bands characterizing the acac
ligands or species formed from them, which could either be
retained by ruthenium centers or dissociated and bonded at
aluminum sites in the zeolite.
Reactions of Sample 1 with H2 and D2. IR spectra were
recorded during the treatments to investigate the reactions of
Sample 1 (∼30 mg) with H2 (50% H2 in He flowing at a total
rate of 100 mL/min at room temperature and atmospheric
pressure) or a pulse of D2 (∼5 mL in He flowing at 100 mL/
min at room temperature and atmospheric pressure). The data
helped clarify the identifications of the bands in the C-H
stretching region.
The spectra of the H2-treated sample showed that C-H
stretching bands inferred to be evidence of ethylene ligands
(3010, 3030, and 3069 cm-1) decreased in intensity during the
treatment but did not disappear (Figure 2).32 In the experiments
using D2 instead of H2, the intensity of the deuterated ethane
(C2H4D2+) peak at m/z 32 increased after the beginning of the
flow of D2 (see Figure 3S in the Supporting Information; this
peak was undetectable in the mass spectra observed during H2
flow); simultaneously, the peaks at 3010, 3030, and 3069 cm-1
IR Evidence of Hydrocarbon Ligands on Sample 1. The IR
spectrum of Sample 1 included bands in the C-H stretching
region (Figure 2). The interpretation of the spectrum was
(25) Kaplan, A. W.; Bergman, R. G. Organometallics 1998, 17, 5072.
(26) Bru¨hwiler, D.; Frei, H. J. Phys. Chem. B 2003, 107, 8547.
(27) Yamaguchi, K.; Mori, K.; Mizugaki, T.; Ebitani, K.; Kaneda, K. J. Am.
Chem. Soc. 2000, 122, 7144.
(28) Bennett, M. A.; Byrnes, M. J.; Willis, A. C. Organometallics 2003,
22, 1018.
(29) Breck, D. W. Zeolite Molecular SieVes: Structure, Chemistry, and Use;
Wiley: New York, 1973.
(30) Miessner, H.; Burkhardt, I.; Gutschick, D.; Zecchina, A.; Morterra,
C.; Spoto, G. J. Chem. Soc., Faraday Trans. 1989, 85, 2113.
(31) Jacobs, P. A.; Uytterhoeven, J. B. J. Chem. Soc., Faraday Trans. 1973,
69, 359.
(32) Analysis of the gas-phase products by mass spectrometry was
inconclusive because the peaks in the mass spectra were broad and
small and the interpretation was complicated by traces of n-pentane
solvent from the synthesis that desorbed slowly.
9
J. AM. CHEM. SOC. VOL. 130, NO. 40, 2008 13339

A R T I C L E S
Ogino and Gates
acetylacetone on the zeolite,38,42 the new band at 1537 cm-1 in
the spectrum of Sample 1 was assigned to acac ligands
dissociated from ruthenium and bonded to zeolite Al3+ sites.
The bands at 1555 cm-1 and 1575 cm-1 in the spectrum of
Sample 1 were assigned to 2[γ(C-H)] and νs(CO)ring of acac
ligands bonded to Ru.38,43 By comparing the IR spectrum of
Sample 1 with literature data44 and the spectrum of the zeolite
with that of adsorbed acetylacetonate, we were able to assign
the shoulder at 1590 cm-1 to νs(CO)ring of acac ligands
coordinated to Al3+ sites. This assignment was supported by
the lack of this band in the spectrum of a silica-supported
ruthenium complex sample (Figure 5S in the Supporting
Information).
In summary, the IR data show that some of the acac ligands
in the precursor were dissociated from ruthenium upon adsorp-
tion of I and became bonded to Al3+ sites in the zeolite. To
estimate the number of acac ligands dissociated from the
ruthenium in Sample 1, peak deconvolution of the IR spectra
was performed, and the Ru-acac/Al-acac area ratio was
determined (Figure 10S in the Supporting Information). Samples
with higher ruthenium loadings (2 and 3 wt % Ru) were also
investigated and found to have smaller Al-acac/Ru-acac area
ratios than the sample containing only 1 wt % Ru (Figures
11S-13S in the Supporting Information); this provided a basis
for identification of the ruthenium complexes in the samples,
as described in the Discussion.
Further Evidence of Ligands on the Supported Mono-
nuclear Ruthenium Complex. Sample 1 was also characterized
by Ru K-edge EXAFS spectroscopy, as summarized in the
following section, which includes a summary of all of the structural
models considered in the EXAFS data fitting. The EXAFS results
were consistent with the IR spectra, showing that the supported
complex was mononuclear and bonded to the surface of the zeolite
with the ethylene ligands intact; structural details are given after
the following section on data analysis.
EXAFS Characterization of the Supported Mononuclear
Ruthenium Complex: Data Analysis. Figure 4A shows k2-
weighted EXAFS data characterizing Sample 1 and the best fit
described below. Figure 4B-D shows Fourier transforms (FTs)
of EXAFS data with various k-weightings together with the best
fits. Table 1 lists the six plausible structural models of the surface
species that were constructed on the basis of the data presented
above.
An assessment of the various candidate models is presented
below and summarized in Table 2; comments about each
model that was tested and an explanation of why all but one
of the models were rejected are provided. The model that
gave the best fit consisted of a mononuclear ruthenium
complex with two ethylene ligands and with one of the two
acac ligands in I replaced by zeolite oxygen atoms at the
acidic sites near Al3+, as shown below. An assessment of
Figure 3. IR spectra of (from bottom to top) calcined dealuminated zeolite
HY, acetylacetone adsorbed on the calcined zeolite, Ru(acac)2(C2H4)2, and
Sample 1.
corresponding to C-H vibrations declined in intensity and
disappeared (data not shown), confirming that they indicate
π-bonded ethylene ligands on the ruthenium.33 This result was
supported by the presence of a small band at 1278 cm-1 in the
spectrum of Sample 1 that could be assigned to the CdC
stretching mode coupled with the scissor mode of CH2 in
ethylene π-bonded to Ru (data not shown).34-37 The band
appeared to be small because intense bands characteristic of
the zeolite framework overlap it. The IR results were further
supported by the EXAFS results presented below, which
indicated a Ru-C coordination number of ∼4 (two ethylene
ligands per Ru atom) and an average Ru-C distance corre-
sponding to that of ethylene π-bonded to Ru.
Reaction of Acetylacetonate Ligands on Ruthenium. IR
spectra characterizing the sample formed by adsorption of
acetylacetone (C5H8O2) on the calcined zeolite (Figure 3)
included bands centered at 1537 and 1590 cm-1 [similar to those
characterizing Al(acac)3 (1534 cm-1)38] that were assigned to
acetylacetonate bonded to aluminum sites in the zeolite.
The IR spectrum characterizing Sample 1 included bands at
1521, 1537, 1555, and 1575 cm-1, with a small shoulder at
∼1590 cm-1. The peak at 1515 cm-1 in the spectrum of I,
characteristic of νas(CCC)ring in the acac ligands, shifted to 1521
cm-1 when I was adsorbed. A similar band shift was observed
when Ru(acac)3 was adsorbed on the zeolite (Figure 4S in the
Supporting Information). The 1521 cm-1 band characterizing
Sample 1 was thus assigned to νas(CCC)ring associated with the
acac ligands remaining on the ruthenium.39-41 On the basis of
IR spectra characterizing the sample prepared by adsorption of
(33) The spectrum reported for a supported sample prepared from
Rh(C2H4)2(acac) and zeolite Y similarly included a peak at 3069
cm-1 attributed to π-bonded ethylene ligands.11
(34) Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coor-
dination Compounds, 5th ed.; Wiley: New York, 1997; Part B, p 278.
(35) Mohsin, S. B.; Trenary, M.; Robota, H. J. J. Phys. Chem. 1988, 92,
5229.
(41) Van Der Voort, P.; van Welzenis, R.; de Ridder, M.; Brongersma,
H. H.; Baltes, M.; Mathieu, M.; van de Ven, P. C.; Vansant, E. F.
Langmuir 2002, 18, 4420.
(42) To check the assignment, control experiments were done with pure
silica as a support. The IR spectrum of the supported sample (Figure 5S
in the Supporting Information) included the band at 1521 cm-1
assigned to acac ligands on ruthenium but not that at 1537 cm-1
assigned to acac ligands bonded to Al3+, consistent with the lack of
Al3+ sites on the silica.
(36) Chesters, M. A.; De La Cruz, C.; Gardner, P.; McCash, E. M.; Pudney,
P.; Shahid, G.; Shepppard, M. J. Chem. Soc., Faraday Trans. 1990,
86, 2757.
(37) De La Cruz, C.; Sheppard, N. J. Chem. Soc., Faraday Trans. 1997,
93, 3569.
(38) Guzman, J.; Gates, B. C. Langmuir 2003, 19, 3897.
(39) Mikami, M.; Nakagawa, I.; Shimanouchi, T. Spectrochim. Acta, Part
A 1967, 23, 1037.
(43) Van der Voort, P.; Mitchell, M. B.; Vansant, E. F.; White, M. G.
Interface Sci. 1997, 5, 169.
(44) Baltes, M.; Van Der Voort, P.; Weckhuysen, B. M.; Rao, R. R.; Catana,
G.; Schoonheydt, R. A.; Vansant, E. F. Phys. Chem. Chem. Phys. 2000,
2, 2673.
(40) Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coor-
dination Compounds, 5th ed.; Wiley: New York, 1997; Part B, p 91.
9
13340 J. AM. CHEM. SOC. VOL. 130, NO. 40, 2008

Genesis of a Zeolite Y-Supported Ru Complex Catalyst
A R T I C L E S
Figure 4. EXAFS data and corresponding FT magnitudes and imaginary parts characterizing Sample 1 (∆k ) 3.67-15.63 Å-1, ∆R ) 0.8-3.0 Å, phase-
uncorrected): (A) k2-weighted EXAFS data, (B) k0-weighted FT, (C) k1-weighted FT, and (D) k2-weighted FT. The solid lines represent data and the dotted
lines represent fits. The arrows in (B-D) indicate a constructive/destructive feature attributed to the presence of the various shells.
Table 1. EXAFS Parameters Corresponding to the Six Candidate Models Characterizing Sample 1a
fit variances
model
backscatterer
N
103∆σ2 (Å2)
R (Å)
∆E0 (eV)
Abs
Im
εv2
A
O
C
Ru
4.0 ( 0.2
4.3 ( 0.5
3.0 ( 1.2
7.0 ( 0.3
7.4 ( 0.5
28 ( 4.2
2.08 ( 0.01
2.22 ( 0.01
3.04 ( 0.02
-3.0 ( 0.6
4.4 ( 0.7
6.5 ( 1.3
k0: 0.017
k2: 0.115
k0: 0.043
k2: 0.255
19.6
B
C
D
O
C
Al
4.0 ( 0.3
4.0 ( 0.5
1.4 ( 0.6
7.1 ( 0.4
6.0 ( 0.6
19 ( 5
2.06 ( 0.01
2.20 ( 0.01
3.07 ( 0.02
0.4 ( 0.6
7.9 ( 1.0
0.1 ( 2.0
k0: 0.020
k2: 0.133
k0: 0.041
k2: 0.331
22.9
17.6
25.5
O
C
Cl
4.0 ( 0.2
4.3 ( 0.5
1.8 ( 0.3
7.1 ( 0.3
7.3 ( 0.7
5.0 ( 0.5
2.07 ( 0.01
2.21 ( 0.01
3.31 ( 0.01
-1.7 ( 0.5
5.2 ( 1.0
-8.5 ( 2.2
k0: 0.009
k2: 0.059
k0: 0.020
k2: 0.129
O
C
Ru
Al
4.0 ( 0.3
4.4 ( 0.6
1.7 ( 0.7
0.6 ( 0.2
7.3 ( 0.4
7.7 ( 0.8
25 ( 4
2.07 ( 0.01
2.21 ( 0.01
2.89 ( 0.02
3.21 ( 0.01
-1.1 ( 0.6
5.5 ( 1.2
-14 ( 3
-19 ( 2
k0: 0.009
k2: 0.059
k0: 0.014
k2: 0.125
3.2 ( 1.0
E
F
O
C
Cl
Ru
4.0 ( 0.3
4.4 ( 0.6
2.1 ( 1.2
0.8 ( 1.1
7.2 ( 0.3
7.7 ( 0.8
10 ( 60
11 ( 13
2.07 ( 0.01
2.22 ( 0.01
3.05 ( 0.03
3.08 ( 0.01
-1.5 ( 0.6
4.9 ( 1.1
11 ( 4
k0: 0.010
k2: 0.053
k0: 0.019
k2: 0.119
18.0
20.6
4.0 ( 2.1
O
C
Cl
Al
4.0 ( 0.3
4.3 ( 0.6
1.9 ( 0.6
1.1 ( 0.2
7.1 ( 0.3
7.4 ( 0.7
3.9 ( 1.1
5.4 ( 1.9
2.07 ( 0.01
2.21 ( 0.01
3.03 ( 0.02
3.08 ( 0.01
-1.6 ( 0.6
5.1 ( 0.7
-1.2 ( 1.3
1.6 ( 1.0
k0: 0.013
k2: 0.053
k0: 0.015
k2: 0.135
a Fit details: R space, 3.67 < k < 15.63 Å-1, 0.8 < R < 3.0 Å. Notation: N, coordination number; R, interatomic distance; ∆σ2, Debye-Waller
parameter; ∆E0, inner potential correction; Abs, real part of the FT; Im, imaginary part of the FT; εv2, goodness-of-fit parameter; kn, weighting in the
Fourier transform.
the model is given in the Discussion. Details of the fitting
process and selection of the model are given in the following
paragraphs and in the Supporting Information.
The number of parameters used in fitting the data to each model
(12 or 16) was always less than the statistically justified number
(18) computed using the Nyquist theorem.45
Details of the EXAFS Data Fitting and Selection of the
Model. No fit determined with only two shells was adequate,
as shown by the poor agreement between the data and the best
fits. Consequently, two-shell models are not considered further
here. Each of the models considered in detail (three- and four-
shell models) included both Ru-O and Ru-C contributions.
Metal-light-atom backscatterer contributions such as Ru-C
and Ru-O cannot generally be distinguished with confidence
by EXAFS spectroscopy alone, but the IR and mass spectrom-
etry data provided the basis for the discrimination, demonstrating
(45) Lytle, F. W.; Sayers, D. E.; Stern, E. A. Phys. B 1989, 158, 701.
9
J. AM. CHEM. SOC. VOL. 130, NO. 40, 2008 13341

A R T I C L E S
Ogino and Gates
Table 2. Qualitative Summary of EXAFS Fitting Results for the Six Candidate Models Representing Sample 1
model
absorber/backscatterer contributions
comments regarding quality of fit of EXAFS data to the model
A
Ru-O, Ru-C, Ru-Ru
physically unrealistically large parameter value (∆σ2 ) 28 × 10-3 Å2)
for the Ru-Ru contribution; individual Ru-Ru shell not fit well on
application of phase and amplitude corrections
B
C
Ru-O, Ru-C, Ru-Al
Ru-O, Ru-C, Ru-Cl
worst values of the variances; physically unrealistically large parameter
value (∆σ2 ) 19 × 10-3 Å2) for the Ru-Al contribution
best goodness-of-fit value and physically realistic values for all
parameters; however, individual Ru-Cl shell not fit well on
application of phase and amplitude corrections
D
E
Ru-O, Ru-C, Ru-Ru, Ru-Al
Ru-O, Ru-C, Ru-Cl, Ru-Ru
worst goodness-of-fit value; physically unrealistically large parameter
values for the Ru-Ru (∆σ2 ) 25 × 10-3 Å2, ∆E0 ) -14 eV) and
Ru-Al (∆E0 ) -19 eV) contributions
physically unrealistically large parameter values for the Ru-Cl (∆σ2
10 × 10-3 Å2, ∆E0 ) 11 eV) and Ru-Ru (∆σ2 ) 11 × 10-3 Å2)
contributions; individual Ru-Cl and Ru-Ru shells not fit well on
application of phase and amplitude corrections
)
F
Ru-O, Ru-C, Ru-Cl, Ru-Al
model providing the best fit: goodness-of-fit value within the fluctuation
of that for C; physically realistic values for all parameters; good fit
with various k-weightings; individual contributions fit well
the presence of ethylene ligands in the initially prepared
supported ruthenium complex and their disappearance when the
sample was treated with H2 or D2. The formation of Ru-Osupport
bonds anchoring the ruthenium complex to the zeolite was
suggested simply by the observation that the precursor reacted
with the zeolite. Attempts to combine the Ru-O and Ru-C
contributions into a single contribution in the data fitting gave
an unrealistically large value of the Debye-Waller parameter
∆σ2 (0.015 Å2), and the fit was judged to lack physical
significance.
The six models differed from each other with regard to which
of the plausible contributions (Ru-Ru, Ru-Al, and/or a second
Ru-C) were included in addition to Ru-O and Ru-C. We
tested for (a) a Ru-Ru contribution, in order to check for cluster
formation resulting from reduction of the ruthenium (none was
found); (b) a Ru-Al contribution, because of the likely reaction
of the cationic ruthenium complex I with anionic sites in the
zeolite (this contribution was found to be significant); and (c)
a second Ru-C contribution (denoted Ru-Cl, where the “l”
subscript stands for “long”) arising from acac ligands on
ruthenium (this was also found to be significant).
Besides the Ru-O and Ru-C contributions, the three-shell
models A, B, and C included Ru-Ru, Ru-Al, and Ru-Cl
contributions, respectively. The four-shell models D, E, and F
included Ru-Ru and Ru-Al, Ru-Ru and Ru-Cl, and Ru-Cl
and Ru-Al contributions, respectively. Addition of the third
(and even the fourth) contribution led to essentially no changes
in the values of the parameters characterizing the first two shells
(Ru-O and Ru-C), leading us to conclude that these two
contributions were significant and necessary for satisfactory
modeling of the structure.
All of the models represented in Table 1 fit the overall data
well, with variances well below 1%.46 However, all of the
models except C and F gave at least one physically unrealistic
parameter value and thus were excluded, as summarized in Table
2 and the following paragraphs.
large inner potential correction (-14 eV),47 leading us to discard
it. Model E gave relatively large ∆σ2 values for the Ru-Cl
and Ru-Ru contributions (10 × 10-3 and 11 × 10-3 Å2,
respectively). Furthermore, the individual contributions of these
shells were not fitted well after the contributions were phase-
and amplitude-corrected (Figures 6S and 7S in the Supporting
Information) and the Ru-Cl contribution in E was characterized
by a large inner potential correction (>10 eV), leading to its
rejection. Thus, on the basis of the analyses for models A, D,
and E, we inferred that there was no Ru-Ru contribution (i.e.,
there were no clusters) that could be determined by EXAFS
spectroscopy.
Model B included a Ru-Al contribution, for which the fit
results indicated a large ∆σ2 value (19 × 10-3 Å2); thus, this
model lacked physical significance and was also rejected.
Inclusion of a Ru-Cl contribution instead of a Ru-Al
2
contribution (model C) gave the smallest value of the εv
(goodness-of-fit) parameter of all the models. However, the fit
of the third-shell (Ru-Cl) contribution was found to be
unsatisfactory after the phase and amplitude corrections were
applied (Figure 8S in the Supporting Information), and thus,
this model was also rejected.
Addition of a Ru-Al contribution to model C (to give the
four-shell model F) gave the best fit of any of the models, and
all of the parameter values were physically reasonable. This
model showed good fits of all of the individual shells, as shown
in Figure 9S in the Supporting Information. It also fit the data
well at all of the examined k-weightings (k0, k1, and k2), although
a constructive/destructive feature was evident in the Fourier
transforms (Figure 4B-D) at values of R near 2.8 Å; this
observation was expected because of the presence of the various
shells.46 The addition of the Ru-Al contribution increased the
goodness of fit by 17% over that of the 3-shell model C, but
the difference of the εv2 values was less than twice the value of
the fluctuations in this parameter (the value of the right-hand
side of eq 3 in the Supporting Information is 2).48,49 Model F
gave good fits of each of the individual contributions, and all
of the parameter values were physically realistic (Table 1). A
Each of the models A, D, and E included a Ru-Ru
contribution. Models A and D were excluded because they gave
unrealistically large ∆σ2 values for the Ru-Ru contribution (28
× 10-3 and 25 × 10-3 Å2, respectively). Furthermore, the
Ru-Ru contribution in D was characterized by an unrealistically
(47) Kelly, S. D.; Ravel, B. AIP Conf. Proc. 2007, 882, 132.
(48) Stern, E. A.; Newville, M.; Ravel, B.; Yacoby, Y.; Haskel, D. Phys.
B 1995, 209, 117.
(46) Koningsberger, D. C.; Mojet, B. L.; van Dorssen, G. E.; Ramaker,
(49) Kelly, S. D.; Kemmer, K. M.; Fryxell, G. E.; Liu, J.; Mattigod, S. V.;
Ferris, K. F. J. Phys. Chem. B 2001, 105, 6337.
D. E. Top. Catal. 2000, 10, 143.
9
13342 J. AM. CHEM. SOC. VOL. 130, NO. 40, 2008

Genesis of a Zeolite Y-Supported Ru Complex Catalyst
A R T I C L E S
clear result is that the EXAFS data demonstrated the existence
of a Ru-O and two Ru-C contributions.
Thus, Model F was judged to provide the best fit of the data
and is recommended. According to this model, the supported
mononuclear ruthenium complex is characterized by Ru-O and
Ru-C contributions at distances of 2.07 and 2.21 Å, respec-
tively. The former agrees with the range of values determined
crystallographically for the precursor complex I (2.055-2.080
Å).28 The Ru-O coordination number of ∼4 accounts for
bonding of the ruthenium to both the zeolite surface and the
acac ligands. The Ru-C contribution is consistent with the IR
and mass spectrometry data indicating ethylene ligands, and the
Ru-C coordination number of ∼4 indicates the presence of
two such ligands per Ru atom. The Ru-C distance essentially
matches the crystallographic results characterizing the precursor
complex I (2.183-2.212 Å). The Ru-Al contribution, at a
distance of 3.08 Å with a coordination number of ∼1, indicates,
as expected, that the cationic complex I reacted with the zeolite
to give a cationic complex bonded to sites where acidic silanol
groups had been present. Thus, the structure of the supported
complex represented below is in accord with all of the data.
Figure 5. Catalysis of ethylene dimerization in a flow reactor: (A) TOF
and (B) butene selectivity [(1) trans-2-butene, (•) 1-butene, (0) cis-2-butene]
as functions of time on stream. Feed composition: 4/4/93 C2H4/H2/He. Total
flow rate: 100 mL (NTP)/min.
Figure 6. Kinetics data: TOF as a function of the partial pressure of (A)
H2 (with PC H ) 4 kPa) and (B) C2H4 (with PH ) 4 kPa). Other conditions:
The Ru-C contribution with a coordination number of ∼2
at a distance of 3.03 Å (Table 1) was attributed to the two
carbonyl carbons in the acac ligand remaining on the ruthenium
(see the Discussion). The distance is longer than that observed
in the precursor I (∼2.93-2.98 Å, as estimated from crystal-
lographic data28), suggesting that the precursor and the supported
complex are not very close analogues.
Ptotal ) 101 kPa (balance He); room temperat2ure.
2
4
reactor experiment and reached a near-steady state (with a
1-butene/trans-2-butene/cis-2-butene molar ratio of 23/52/25)
after ∼11 h.
The TOF in the near-steady state was 0.06 mol of C4H8 (mol
of Ru)-1 s-1. The number of turnovers was 2800 when the
experiment was stopped after reaction for ∼14 h; clearly, the
reaction was catalytic.
Summary of EXAFS Fitting. In summary, the EXAFS results
indicate that the supported species was a cationic, mononuclear
diethylene ruthenium complex anchored tightly to the surface
at Al3+ sites by Ru-O bonds. The IR data reinforce this model,
as described further in the Discussion.
Increasing the C2H4 partial pressure led to increased selectiv-
ity for 1-butene formation. For example, an increase in the C2H4
partial pressure from 4 to 51 kPa at a H2 partial pressure of 4
kPa and a total pressure of 101 kPa led to an increase in the
selectivity for 1-butene formation from 23 to 92% as the
selectivities for trans-2-butene and cis-2-butene each decreased
to 4%.
Although ethylene dimerization might be expected to proceed
without H2,50-55 the reaction was much slower without H2 in
the feed stream (TOF ≈ 1 × 10-4 s-1), and then gave 1-butene
as the only detectable product. Thus, H2 played a role in forming
the catalytically active species.
Catalytic Dimerization of Ethylene. Blank Experiments.
Flow-reactor experiments were conducted with the bare zeolite
and with structures formed from adsorption of Ru(acac)3 on
the zeolite. No detectable conversion of C2H4 (or C2H4 + H2)
was observed at room temperature with either of these samples.
Reaction Products and Activity of the Catalyst. In contrast,
the supported ruthenium complex (Sample 1) was catalytically
active and gave mixtures of butenes with a trace amount of
ethane (<0.05% of the total product) when the feed to the flow
reactor was an equimolar mixture of C2H4 and H2 at room
temperature. Ethylene conversions were <5% at steady state,
mostly ranging from 1 to 3%. Therefore, the reactor was treated
as differential, and the reaction rate (turnover frequency, TOF)
was determined directly from the conversion data. Higher
conversions were also measured but are not reported here.
Figure 5A,B shows TOF and butene selectivity data as a
function of time on stream with a feed mixture containing C2H4
and H2 at room temperature, each at a partial pressure of 4 kPa
in helium, at a total pressure of 101 kPa. The TOF was nearly
independent of time on stream under these conditions. The
butene selectivity changed slightly at the beginning of the flow-
Figure 6A,B shows values of the TOF at near-steady state as
functions of the partial pressures of the reactants. The errors
shown on the plot were attributed primarily to uncertainties in
weighing of the catalyst in the glovebox. As described above,
(50) Laurenczy, G.; Merbach, A. E. J. Chem. Soc., Chem. Commun. 1993,
187.
(51) Yashima, T.; Ebisawa, M.; Hara, N. Chem. Lett. 1972, 473.
(52) Bass, J. S.; Kevan, L. J. Phys. Chem. 1990, 94, 1483.
(53) Hartmann, M.; Po¨ppl, A.; Kevan, L. J. Phys. Chem. 1996, 100, 9906.
(54) Takahashi, N.; Okura, I.; Keii, T. J. Am. Chem. Soc. 1975, 97, 7489.
(55) Baba, T.; Nakano, K.; Nishiyama, S.; Tsuruya, S.; Masai, M. Appl.
Catal. 1989, 52, 81.
9
J. AM. CHEM. SOC. VOL. 130, NO. 40, 2008 13343

A R T I C L E S
Ogino and Gates
Marquardt iterations to optimize the parameters. The peak
deconvolution (assuming the same molar extinction coefficient
for each species) gave an Al-acac/Ru-acac area ratio of 49/
51 for Sample 1, which contained 1 wt% Ru (see Figure 10S
in the Supporting Information), showing that essentially half
of the acac ligands were dissociated from the ruthenium and
bonded to Al3+ sites. This result was further supported by the
EXAFS data: the bonding Ru-O contribution with a coordina-
tion number of ∼4 corresponds to two O atoms of one acac
ligand and two O atoms of the zeolite, and the long Ru-C
contribution (Ru-Cl) with a coordination number of ∼2 arises
from two carbonyl carbons per acac ligand. The Ru-Cl
contribution was characterized by a longer distance (3.03 Å)
than the crystallographic value characterizing the precursor,28
indicating a distortion of the structure from that in the precursor
I.
Chemistry of Adsorption of I with Samples Containing 2
or 3 wt % Ru. On the basis of these results, we inferred that
one precursor molecule I reacted with two Al3+ sites (one for
anchoring the ruthenium complex and one for bonding to a
dissociated acac ligand) and that all of the ruthenium complexes
in this sample were dissociated and chemisorbed. By inference,
we expected a maximum loading of one chemisorbed ruthenium
complex per two Al atoms.
Figure 7. Time evolution of (A) IR spectra recorded at the start of the
catalytic reaction in a flow reactor and (B) the corresponding difference
spectra. The arrows indicate the directional changes of the absorbances of
individual peaks as a function of time.
H2 was required for the catalyst to initiate the reaction. As Figure
6A shows, the rate increased with increasing H2 partial pressure
up to ∼4 kPa and then decreased at higher H2 partial pressures.
This decrease suggests inhibition by H2. On the other hand, the
TOF increased monotonically with increasing C2H4 partial
pressure; the reaction order with respect to C2H4 was 0.8 (Figure
6B).
To approximately determine the value of this maximum by
experiment, we also investigated samples with higher ruthenium
loadings (2 and 3 wt%). The Al-acac/Ru-acac area ratios
determined from IR spectra characterizing these samples were
found to decrease with increasing ruthenium loading. Peak
deconvolutions of the Ru-acac and Al-acac bands, determined
as mentioned above, yielded Al-acac/Ru-acac ratios of 24/
76 and 18/82, respectively, for the samples containing 2 and 3
wt% Ru, respectively (Figures 12S and 13S in the Supporting
Information). Thus, the samples with Ru loadings of 2 and 3
wt % were characterized by lower Al-acac/Ru-acac ratios than
the sample containing 1 wt % Ru, demonstrating the presence
in the former samples of some ruthenium complexes that did
not exchange acac ligands with the zeolite. Thus, the experi-
ments confirmed that the maximum loading for dissociative
chemisorption of the ruthenium complex was in the range of
1-2 wt % Ru and that the available Al3+ sites were saturated
in the samples containing 2 and 3 wt % Ru. Some of the
ruthenium complexes in these samples were evidently phys-
isorbed, with the acac ligands retained by the ruthenium.
Because Sample 1 contained approximately one ruthenium
complex per six Al atoms and an amount of chemisorbed
ruthenium complex that did not exceed the maximum value,
we infer that not all of the Al3+ sites in the zeolite were
accessible to I or that some of the Al3+ sites were nonframework
sites59 and might not be suitable for anchoring the complexes;
the inaccessible sites were presumably in the ꢀ-cages, as
characterized by IR bands at 3565 cm-1 (Figure 1).
In contrast to the zeolite-supported Sample 1, the silica-
supported ruthenium sample was characterized by only a low
activity (TOF ) 7 × 10-3 s-1), even in the presence of H2
(C2H4 and H2, each at 4 kPa, in He). This result suggests that
catalytically active species barely formed on the amorphous
silica support, and a comparison of the two catalysts suggests
that Al3+ sites in the zeolite might be responsible for creating
catalytically active species.
IR Spectra of the Working Catalyst. To understand the roles
of H2 and zeolite Al3+ sites in the genesis of the catalytically
active species, experiments were conducted under the same flow
conditions stated above (C2H4 and H2, each at a partial pressure
of 4 kPa, in He; 100 mL/min total flow) with 30 mg of the
sample in an IR cell serving as a flow reactor.
Figure 7 shows the changes in the IR bands characteristic of
acac ligands bonded to aluminum sites (Al-acac) and to
ruthenium sites (Ru-acac) during the catalysis. The intensity
of the Ru-acac band decreased and that of the Al-acac band
began to increase when the flow of reactants started, suggesting
that dissociation of acac ligands from ruthenium and their
subsequent bonding to Al3+ sites in the zeolite yielded the
catalytically active species.
Discussion
Chemistry of Chemisorption of the Precursor Complex I
on the Zeolite. The IR data (Figure 3) showed that some of the
acac ligands were dissociated from the ruthenium after chemi-
sorption of I on the zeolite. The fraction of the acac ligands
dissociated from the ruthenium was estimated from the IR
spectra by deconvolution according to the Pearson VII model56-58
(as described in the Supporting Information). The deconvolution
was implemented by choosing five peaks (centered at 1521,
1537, 1555, 1575, and 1590 cm-1) and applying Levenberg-
To determine the value of maximum loading for chemisorp-
tion of the ruthenium more exactly, we performed the following
approximate analysis:
If we assume that the maximum loading of Ru to form
chemisorbed ruthenium complexes is 1 wt %, then x wt % Ru
(x > 1) would result in 1 wt % chemisorbed complexes and (x
- 1) wt % physisorbed species. This assumption would lead to
a value of 1/(2x) for the area ratio Al-acac/(Al-acac +
(56) Elderton, W. P; Johnson, N. L. Systems of Frequency CurVes;
Cambridge University Press: London, 1969.
(57) Chen, L.; Garland, M. Appl. Spectrosc. 2003, 57, 331.
(58) Bhan, A.; Allian, A. D.; Sunley, G. J.; Law, D. J.; Iglesia, E. J. Am.
Chem. Soc. 2007, 129, 4919.
(59) Shannon, R. D.; Gardner, K. H.; Staley, R. H.; Bergeret, G.; Gallezot,
P.; Auroux, A. J. Phys. Chem. 1985, 89, 4778.
9
13344 J. AM. CHEM. SOC. VOL. 130, NO. 40, 2008

Genesis of a Zeolite Y-Supported Ru Complex Catalyst
A R T I C L E S
coordination of a hydride, producing the catalytically active
Ru(CH3COO)(H)[BINAP].62-65
As these examples imply, the surface ruthenium species of
the as-prepared Sample 1 may enter into a catalytic cycle via
(1) heterolytic cleavage of H2 into a proton and a hydride on
the ruthenium complex, (2) protonation of the acac ligand, (3)
dissociation of the protonated acac ligand (the enol form of
C5H8O2), (4) coordination of the acetylacetone to an Al site of
the zeolite and deprotonation of the resultant species, and (5)
protonation of an oxygen atom of the zeolite bonded to the Al
site.
Conclusions
Zeolite-supported ruthenium complexes were prepared by the
reaction of I with dealuminated zeolite Y. The high degree of
structural uniformity of the supported species10 allowed a
detailed characterization of the chemistry of the supported
complex, including the bonding of the ruthenium to the support
surface. Precursor I reacted with acidic silanol groups on the
surface of the zeolite as one of two acac ligands in I was
dissociated from the ruthenium and bonded at an aluminum site
in the zeolite, and the ethylene ligands were retained by the
ruthenium. After this reaction, the ruthenium complexes re-
mained mononuclear, becoming bonded to the site where acidic
silanol groups had been present. The maximum loading of
ruthenium on the zeolite as chemisorbed species is ∼1 wt %.
The supported metal complex with 1 wt % loading of ruthenium
catalyzed ethylene dimerization at room temperature. H2
facilitated the formation of the catalytically active species by
inducing the dissociation of acac ligands from ruthenium; the
dissociated acac ligands were scavenged by aluminum sites in
the zeolite.
Figure 8. Values of the peak-area ratio Al-acac/(Al-acac + Ru-acac)
determined by deconvolution of IR spectra vs those calculated on the basis
of the assumption that the maximum loading of ruthenium is 1 wt%.
Ru-acac). The area ratios for the three samples with different
Ru loadings were calculated and compared with those deter-
mined from the IR data (Figure 8). The agreement is good,
pointing to the validity of the assumption. Thus, we conclude
that the maximum loading of Ru is ∼1 wt%.
Structural Model of the Supported Ruthenium Complex. On
the basis of the IR and EXAFS data representing Sample 1, we
infer that chemisorption of the precursor I led to the replacement
of one acac ligand per Ru atom with two oxygen atoms of the
support, producing the surface-bound species shown in the
structural model given in the Results.
The surface-bound species modeled in this structure is
formally a coordinatively saturated 18-electron species if one
assumes that the surface oxygen atoms are two-electron donors
(i.e., assuming a formal charge of -1 on an AlO2 unit): 6 e-
for Ru(II), 4 e- for acac, 4 e- for the zeolite oxygen ligand,
and 4 e- for the two ethylene ligands. The implications of the
coordinative saturation in regard to catalysis are presented
below.
Experimental Methods
Materials and Procedures. Sample handling was carried out
with standard Schlenk techniques, and samples were prepared and
stored in a glovebox under dry Ar. Glassware was dried at 120 °C
overnight prior to use in syntheses. Immediately before use in
syntheses, n-pentane solvent was distilled from Na/benzophenone
and then deoxygenated by sparging of N2. Ru(acac)3 (Strem, 99%),
was used to synthesize I. He, H2, and C2H4 (all Airgas, UHP grade)
were purified by passage through traps containing activated zeolite
4A and reduced Cu/Al2O3 to remove traces of moisture and O2,
respectively. D2 (Cambridge Isotopes, 99.6% D2 + 0.4% HD) was
used as received. Dealuminated zeolite HY [Zeolyst, CBV760, Si/
Al ratio ) 30 (atomic)] and silica (Degussa, Aerosil 200) were
used as supports.
Reaction of the Supported Ruthenium Complex to Enter
the Catalytic Cycle. If the proposed structural model of the
surface species in the initially prepared sample (Sample 1) is
valid, ligand dissociation must take place for this species to
enter the catalytic ethylene dimerization cycle. This reaction
proceeded much faster with a cofeed of C2H4 and H2 than
without H2. This result suggests that H2 played a role in the
creation of the catalytically active species, as indicated by the
IR data (Figure 7).
Several ruthenium(II) complexes in solution have been
reported to enter catalytic cycles through the dissociation of
ligands via heterolytic cleavage of H2.60 The ruthenium com-
plexes in these examples stay in the same formal oxidation state
(II, d6 metal). For example, RuCl2(PPh3)317 was reported to react
with H2 as H2 is activated heterolytically, forming RuHCl(PPh3)3
and HCl.61 A precatalyst for asymmetric hydrogenation,
Ru(CH3COO)2[BINAP], is believed to enter a catalytic cycle
via dissociation of an CH3COO- ligand as CH3COOH and
Synthesis of the Ruthenium Complex Precursor I. The
ruthenium complex I was synthesized by a literature method,28 and
1
the H and 13C NMR spectra and IR spectra essentially matched
the literature values, as follows: 1H NMR (500 MHz, C6D6): δ 5.26
(s, 2H, γ-CH), 3.72-3.87 (m, 8H, C2H4), 1.84, 1.88 (each s, 6H,
CH3). 13C NMR (500 MHz, C6D6): δ 187.2 (CdO), 185.7 (CdO),
98.7 (γ-C), 78.7 (C2H4), 28.1 (CH3), 27.5 (CH3). IR (KBr): 1576,
1515 cm-1 (acac).
Synthesis of Zeolite-Supported Ruthenium Complexes. The
zeolite was calcined by heating from room temperature to 773 K
in flowing dry O2 over a period of 3 h followed by a period of
constant temperature (773 K) for 4 h in flowing O2 and then 16 h
(60) Kubas, G. J. Chem. ReV. 2007, 107, 4152.
(63) Noyori, R.; Kitamura, M.; Ohkuma, T. Proc. Natl. Acad. Sci. U.S.A.
2004, 101, 5356.
(61) Crabtree, R. H. The Organometallic Chemistry of the Transition Metals,
4th ed.; Wiley-Interscience: Hoboken, NJ, 2005; p 251.
(64) Wiles, J. A.; Bergens, S. H.; Young, V. G. J. Am. Chem. Soc. 1997,
119, 2940.
(62) Kitamura, M.; Tsukamoto, M.; Bessho, Y.; Yoshimura, M.; Kobs, U.;
Widhalm, M.; Noyori, R. J. Am. Chem. Soc. 2002, 124, 6649.
(65) Ashby, M. T.; Halpern, J. J. Am. Chem. Soc. 1991, 113, 589.
9
J. AM. CHEM. SOC. VOL. 130, NO. 40, 2008 13345

A R T I C L E S
Ogino and Gates
under vacuum. The zeolite under vacuum was then cooled to room
temperature and used immediately for sample preparation, as
follows:
To 2.0 g of calcined zeolite in a 100 mL Schlenk flask with a
stir bar was added 0.072 g of complex I [Ru/Al ≈ 1/6 (atomic)].
To this mixture was added ∼30 mL of freshly distilled n-pentane.
The mixture was stirred at room temperature, and after 18 h, the
n-pentane was removed by evacuation.
Samples with higher ruthenium loadings were prepared similarly.
A zeolite-supported sample prepared from Ru(acac)3 was prepared
similarly, using Ru(acac)3 instead of I as the precursor. A silica-
supported sample was prepared similarly by using silica instead of
the zeolite. The ruthenium loading was 1 wt %.
Acetylacetone Adsorbed on Zeolite. To 1.2 g of calcined zeolite
in a 50 mL Schenk flask with a stir bar was added ∼20 mL of
freshly distilled n-pentane; 0.012 g of acetylacetone (Aldrich, 0.12
mmol) was added to this mixture by syringe. The resultant mixture
was stirred for 18 h at room temperature, and the solvent was then
removed by evacuation.
Ethylene Dimerization Catalysis. The zeolite-supported sample
was tested as a catalyst with a reactant mixture consisting of
ethylene, H2, and He. The reaction was performed in a once-through
flow reactor (i.d. ) 1.0 cm). The supported sample (10 mg) was
diluted with 6 g of inert, nonporous R-Al2O3 (Fisher) in a glovebox
and then loaded into the reactor and transferred to the flow system
without coming into contact with air. The partial pressures of
ethylene and H2 were varied individually, and the total pressure
was atmospheric. The flow rate of the feed mixture was maintained
at 100 mL (NTP)/min by adjusting the helium flow rate when the
composition was changed.
Products were analyzed by gas chromatography using an HP-
6890 gas chromatograph equipped with a 50 m × 0.53 mm DB-
624 capillary column (J & W Scientific) and an FID detector. In a
typical run, ethylene and H2, each at a partial pressure of 4.0 kPa,
flowed through the catalyst bed. The effluent gas was sampled every
30 min and analyzed. The ethylene conversions were <5%, and
the reactor was treated as differential.
IR spectra were baseline-corrected with OPUS software (Bruker).
Deconvolution of the IR spectra was implemented using the Pearson
VII model (the equation is shown in the Supporting Information).
The fit parameters were optimized by Levenberg-Marquardt
iterations.
The effluent gas was analyzed by mass spectrometry (Pfeiffer
Vacuum, Omni Star).
X-ray Absorption Spectroscopy of Supported Ruthenium
Complexes. Ru K-edge X-ray absorption spectra were measured
at the Stanford Synchrotron Radiation Laboratory (SSRL) on the
unfocused 30-pole 1.45 T wiggler beam line 10-2 under standard
ring operating conditions, with a storage-ring electron energy of 3
GeV and a ring current of 60-100 mA. A Si(220) double-crystal
monochromator was used for selection of the beam energy. The
monochromator was detuned to 75% of maximum intensity to
reduce the interference of higher harmonics present in the X-ray
beam.
The mass of each sample was chosen to give an absorbance
between 1.5 and 3.0 calculated at 50 eV above the absorption edge.
The sample was placed in a cell66 and maintained under vacuum
(<10-7 kPa) at liquid nitrogen temperature during the data
collection. X-ray intensity data were collected in transmission mode
by use of ion chambers mounted on each end of the sample cell.
The energy was calibrated by simultaneous measurement of the
absorption of a ruthenium metal foil. The beam exiting the ring
first passed through an ion chamber containing N2, and then through
the sample cell, an ion chamber containing argon, a cell containing
the ruthenium foil, and finally an ion chamber containing argon.
Data were measured at values of the wave vector k up to 16 Å-1
.
The first inflection point of the spectrum of the foil was assigned
to be 22117 eV. The EXAFS data analysis procedure is summarized
in detail in the Supporting Information.
Acknowledgment. This research was supported by the U.S.
Department of Energy, Office of Energy Research, Basic Energy
Sciences Grant FG02-04ER15513. We acknowledge beam time and
the support of the DOE Division of Materials Sciences for its role
in the operation and development of beam line 10-2 at the Stanford
Synchrotron Radiation Laboratory, and we thank the beam line staff.
We also thank Dr. Y. Ping of the University of California, Davis,
for collecting solution NMR data.
NMR Spectroscopy of the Ruthenium Complex Precursor.
NMR spectra were recorded at 293.6 K with a Bruker Avance 500
spectrometer (1H at 500 MHz, 13C at 125.8 MHz). The chemical
shifts (δ) for 1H and 13C are given in parts per million referenced
to the residual solvent peaks (7.16 ppm for 1H and 128.06 ppm for
13C, C6D6, Aldrich 99.5%).
Note Added after ASAP Publication. After this paper was
published ASAP September 12, 2008, the formula for the mono-
nuclear ruthenium complex was corrected in two places in the
Abstract. The corrected version was published October 1, 2008.
IR Spectroscopy of Zeolite-Supported Ruthenium Complexes
and Analysis of Effluent Gases. Spectra were collected in transmis-
sion mode with a Bruker IFS 66v Fourier transform spectrometer
with a spectral resolution of 2 cm-1. The solid sample, a pressed
wafer, was held at room temperature under dynamic vacuum
(<10-4 kPa). Each spectrum was the average of 64 scans. In some
experiments, the sample was in contact with flowing gas; ∼30 mg
of the solid sample in the glovebox was loaded into a cell that
served as a flow reactor (In-situ Research Institute, Inc., South Bend,
IN). The cell was sealed and connected into a flow system that
allowed recording of spectra while the reactant gases flowed through
the cell at the reaction temperature. In experiments with D2, ∼5
mL of D2 was pulsed into a stream of He flowing at 100 mL/min
through the cell at room temperature.
Supporting Information Available: EXAFS data analysis
procedure, 1H and 13C NMR spectra of I, IR spectra of zeolite-
supported ruthenium complexes, IR spectrum of a silica-
supported ruthenium complex, Fourier transforms of EXAFS
data, and IR spectra and their deconvolutions. This material is
available free of charge via the Internet at http://pubs.acs.org.
JA804265R
(66) Jentoft, R. E.; Deutsch, S. E.; Gates, B. C. ReV. Sci. Instrum. 1996,
67, 2111.
9
13346 J. AM. CHEM. SOC. VOL. 130, NO. 40, 2008