Grafted metallocalixarenes as single-site surface organometallic catalysts

Article abstract of DOI:10.1021/ja0470259

Metallocalixarenes were grafted onto silica using a surface organometallic approach and shown to be active and selective catalysts for epoxidation of alkenes using organic hydroperoxides. Calixarene-Ti^IV precursors were anchored at surface dens

Full text of DOI:10.1021/ja0470259

Published on Web 11/24/2004
Grafted Metallocalixarenes as Single-Site Surface
Organometallic Catalysts
Justin M. Notestein, Enrique Iglesia, and Alexander Katz*
Contribution from the Department of Chemical Engineering,
UniVersity of California at Berkeley, Berkeley, California 94720-1462
Received May 20, 2004; E-mail: katz@cchem.berkeley.edu
Abstract: Metallocalixarenes were grafted onto silica using a surface organometallic approach and shown
to be active and selective catalysts for epoxidation of alkenes using organic hydroperoxides. Calixarene-
TiIV precursors were anchored at surface densities from 0.1 to near-monolayer coverages (0.025-0.25
calixarene nm-2). Several spectroscopic methods independently detected calixarene-TiIV connectivity before
and after epoxidation catalysis. Kinetic analyses of cyclohexene epoxidation confirmed that the active sites
were anchored on the silica surface and were significantly more active than their homogeneous analogues.
The steric bulk and multidentate binding of the calixarenes led to structural stability and to single-site behavior
during epoxidation catalysis. Rate constants were independent of surface density for cyclohexene
epoxidation with tert-butyl hydroperoxide (11.1 ( 0.3 M-2 s-1) or cumene hydroperoxide (25 ( 2 M-2 s-1).
The materials and methods reported here allow the assembly of robust surface organometallic structures
in which the active sites behave as isolated species, even near saturation monolayer coverages. In turn,
this makes possible the rational design and synthesis of a class of heterogeneous oxide catalysts with
atomic-scale precision at the active site.
Introduction
environment, defined from the top by a calixarene molecule
and from the bottom by a silica surface. The rigidity and steric
Many heterogeneous catalysts currently in practice are
chemically complex and consist of oligomeric active site
structures that are difficult to optimize via atomic-scale control
at the active site. A promising new method for the design and
synthesis of heterogeneous catalysts with control of active site
structure involves the use of emerging concepts in surface
organometallic chemistry. Surface organometallic catalysts
consist of a transition-metal-containing active site that is grafted
onto a solid support, which serves as a rigid ligand for the metal
center.1-5 These catalysts can exhibit interesting support effects,
including large rate enhancements, when active sites are
anchored onto supports instead of acting as homogeneous
entities in solution.2,3 These rate enhancements resemble those
reported for systems consisting of hybrid organic-inorganic
materials,6 and reflect the synergy between the catalytically
active site and a support surface that acts as a ligand.
bulk of both ligandsscalixarene and silicasensure metal atom
site isolation, even during kinetic processes that require ligand
exchange; such stability is essential for durable and robust
heterogeneous catalysts.
Our synthesis of grafted metallocalixarene catalysts is based
on anchored calixarene structures 1 and 2. These have been
synthesized according to previously reported procedures by
treating a calixarenetriol or -tetrol compound with an activated
silica containing chloride ligands.7 We previously showed that
this process leads to the grafting of a calixarene monolayer and
to a 10 Å decrease in the average pore radius of the silica
support, consistent with the expected thickness of a single
calixarene layer. Each calixarene in 1 was shown to act as a
stoichiometric adsorbent for one toluene molecule, a result that
is consistent with an open cone conformer for each anchored
calixarene.7
Koiland molecules8 originally inspired the synthesis of grafted
calixarenes 1 and 2. Reports of isostructural Ti dimers of
calixarenes,9 as well as structures in which calixarenes serve
as a well-defined oxo surface for several other metals,10
suggested to us that the Si atom in 2 could be successfully
replaced by transition metals of catalytic interest. We report
We report here a new strategy to control the interplay between
metal centers and organic ligands or surfaces in grafted
organometallic catalysts. Our approach involves confining an
isolated metal atom within a pseudotetrahedral oxo coordination
(1) Cope´ret, C.; Chabanas, M.; Saint-Arroman, R. P.; Basset, J. M. Angew.
Chem., Int. Ed. 2003, 42, 156-181.
(2) Attfield, M. P.; Sankar, G.; Thomas, J. M. Catal. Lett. 2000, 70, 155-
158.
(3) Meunier, D.; Piechaczyk, A.; de Mallmann, A.; Basset, J. M. Angew. Chem.,
Int. Ed. 1999, 38, 3540-3542.
(4) Jarupatrakorn, J.; Tilley, T. D. J. Am. Chem. Soc. 2002, 124, 8380-8388.
(5) Bouh, A. O.; Rice, G. L.; Scott, S. L. J. Am. Chem. Soc. 1999, 121, 7201-
7210.
(7) Katz, A.; Da Costa, P.; Lam, A. C. P.; Notestein, J. M. Chem. Mater. 2002,
14, 3364-3368.
(8) Delaigue, X.; Hosseini, M. W.; de Cain, A.; Fischer, J. Tetrahedron Lett.
1993, 34, 3285-3288.
(9) Olmstead, M. M.; Sigel, G.; Hope, H.; Xu, X.; Power, P. P. J. Am. Chem.
Soc. 1985, 107, 8087-8091.
(6) Bass, J. D.; Anderson, S. L.; Katz, A. Angew. Chem., Int. Ed. 2003, 42,
5219-5222.
(10) Floriani, C.; Floriani-Moro, R. AdV. Organomet. Chem. 2001, 47, 167-
233.
9
16478
J. AM. CHEM. SOC. 2004, 126, 16478-16486
10.1021/ja0470259 CCC: $27.50 © 2004 American Chemical Society

Metallocalixarenes as Organometallic Catalysts
A R T I C L E S
here the Ti analogue 3 of material 2. Material 3 is based on
molecular precursor 4b, which has been synthesized and
characterized using single-crystal X-ray diffraction.11
ing existing heterogeneous catalysts and elucidating their
catalytic mechanisms.
Experimental Section
Precursor Synthesis. Calixarene-TiIVCl 4b was prepared by
refluxing 1.0 equiv of dimethoxycalixarene 5a (Acros, 99%) with TiCl4
in toluene (Aldrich, 1.0 M) for 48 h followed by recrystallization from
hexane according to previously published procedures.11 Elemental
analysis (Anal. Calcd for 4b‚(C7H8)0.5: C, 73.79; H, 7.55. Found: C,
1
73.97; H, 7.82) and H NMR (400 MHz, CDCl3, 298 K) (δ 1.21 (m,
36H, But), 3.34 (dd, 4H, J ) 12.4 Hz, exo-CH2), 4.23 (s, 3H, OCH3),
4.35 (d, 2H, J ) 12.4 Hz, endo-CH2), 4.78 (d, 2H, J ) 12.4 Hz, endo-
CH2), 7.05 (m, 8H, ArH)) agreed with published values. The four
doublets in the methylene region are characteristic of the lower
symmetry of the mono-demethylated product 4b. Calixarene 5a has
only two methylene doublets. Species 4b could be stored for several
weeks as a dark red toluene solution under Schlenk conditions without
detectable degradation.
Immobilization of 4b To Yield 3-75. In the naming scheme used
here, 3-75 indicates a material with 75 µmol of calixarene-TiIV per g
of catalyst. Chromatography silica gel (300 mg, 60 Å pores, 250-500
µm, Selecto) was partially dehydroxylated under dynamic vacuum at
500 °C for 24 h. The silica was transferred to a flask with 50 mL of
anhydrous toluene. A 25 µmol sample of 4b (as a toluene solution)
was added and the suspension refluxed under N2. The red color of 4b
was gradually transferred from the solution to the solid. After 0.5 h,
70 µL of 2,6-di-tert-butylpyridine (0.25 mmol) was added, and the
suspension was returned to reflux. After 24 h, the solid was filtered,
washed with ∼300 mL of hot anhydrous toluene, and dried under
dynamic vacuum at 25 °C for at least 4 h and at 250 °C for 1 h.
Materials with identical catalytic behavior were made with and without
recrystallization of 4b, thus enabling the entire catalyst synthesis to
occur in a single pot using commercially available reagents.
Our objective in substituting Ti within the grafted metallo-
calixarene architecture is to incorporate epoxidation catalytic
activity in the same manner as for titanium silicalite (TS-1), in
which isolated Ti atoms replace Si atoms within the zeolite
framework.12-14 We demonstrate the utility of 3 as a highly
active, stable, and selective catalyst for alkene epoxidation.
Grafted titanium complexes catalyze epoxidation reactions but
lose activity and selectivity during catalysis, especially at high
surface Ti loadings, which fail to maintain metal site isolation
and tend to form Ti-oxo oligomers.15-17 Some previous
attempts to combine organic and inorganic ligands around Ti
metal centers have led to more active catalysts, but also to
sterically hindered sites, leaching, or ligand exchange during
epoxidation catalysis.2,4,18-20 We show here that material 3 is
active and stable during epoxidation catalysis and shows single-
site catalytic behavior. These conclusions are supported by
spectroscopic characterization and measurements of intrinsic
epoxidation rate constants on 3; these rate constants are large
and do not change with reaction time or Ti surface density.
The type of grafted metallocalixarene material reported here
for Ti active species is directly extendable to other transition
metals and to the synthesis of isolated active sites for a wide
range of chemical transformations. Also, because of their single-
site nature, these materials provide robust strategies for optimiz-
ing catalyst activity and selectivity via calixarene ligand design
and synthesis, which exploit the existing diversity and flexibility
of metallocalixarene synthetic chemistry.10,21 We anticipate
broad utility for grafted metallocalixarenes as new surface
organometallic catalysts and as a powerful platform for improv-
Immobilization of 5b To Yield 2-116. Using our previously
published procedures,7 a suspension of 1.00 g of chlorinated silica gel
was refluxed in 50 mL of toluene with 160 mg of 5b22 (0.24 mmol)
and 1.0 mL of NEt3 (7.2 mmol, 2 equiv relative to surface silanols) to
yield the silicon-containing analogue of 3.
Catalysis. A 50 mg sample of catalyst (7 µmol of Ti for 3-138) and
∼300 mg of 4A or 3A molecular sieves were added to a 50 mL round-
bottom flask with magnetic stirring. The reactor was sealed and
degassed under dynamic vacuum at 25 °C for 1 h and at 70 °C for 1
h. The reactor was cooled under Ar backflow, charged with 40 mL of
octane (solvent) and 6.5 mmol of substrate, sealed with a PTFE-coated
rubber septum, and transferred to a 60 °C oil bath. After thermal
equilibration, 1.5 mmol of tert-butyl hydroperoxide (TBHP; Fluka, ∼5.5
M solution in nonane with 4A molecular sieves) or cumyl hydroper-
oxide (CHP; Fluka, ∼80% solution in cumene, 4A molecular sieves
added) was injected to start the reaction. Aliquots (100 µL) were
removed periodically using a syringe, filtered to remove the catalyst,
and analyzed using an Agilent 6890 GC system equipped with a flame
ionization detector using an HP-1 methylsilicone capillary column.
Cumene or nonane present in the hydroperoxide solution was used as
an internal standard. Details of reagent purification and of product
identification are given in the Supporting Information.
(11) Zanotti-Gerosa, A.; Solari, E.; Giannini, L.; Floriani, C.; Re, N.; Chiesi-
Villa, A.; Rizzoli, C. Inorg. Chim. Acta 1998, 270, 298-311.
(12) Saxton, R. J. Top. Catal. 1999, 9, 43-57.
Analytical Characterization. Thermogravimetric analysis (TGA)
was performed on a TA Instruments TGA 2950 system using a flow
of dry synthetic air at a ramp rate of 5 °C/min. Carbon content was
measured by the Berkeley Microanalytical Laboratory using a Perkin-
Elmer 2400 Series II combustion analyzer. Titanium content was
measured by Quantitative Technologies, Inc. using inductively coupled
plasma mass spectrometry (ICP-MS).
(13) Bellussi, G.; Rigutto, M. S. Stud. Surf. Sci. Catal. 1994, 85, 177-213.
(14) Davis, M. E.; Katz, A.; Ahmad, W. R. Chem. Mater. 1996, 8, 1820-1839.
(15) Davis, R. J.; Liu, Z. Chem. Mater. 1997, 9, 2311-2324.
(16) Thomas, J. M. Angew. Chem., Int. Ed. 1999, 38, 3588-3628.
(17) Gao, X.; Wachs, I. E. Catal. Today 1999, 51, 233-254.
(18) Crocker, M.; Herold, R. H. M.; Orpen, A. G.; Overgaag, M. T. A. J. Chem.
Soc., Dalton Trans. 1999, 3791-3804.
(19) Arends, I. W. C. E.; Sheldon, R. A. Appl. Catal., A 2001, 212, 175-187.
(20) Hutter, R.; Mallat, T.; Baiker, A. J. Catal. 1995, 153, 177-189.
(21) Wieser, C.; Dieleman, C. B.; Matt, D. Coord. Chem. ReV. 1997, 165, 93-
161.
(22) Groenen, L. C.; Rue¨l, B. H. M.; Casnati, A.; Verboom, W.; Pochini, A.;
Ungaro, R.; Reinhoudt, D. N. Tetrahedron 1991, 47, 8379-8384.
9
J. AM. CHEM. SOC. VOL. 126, NO. 50, 2004 16479

A R T I C L E S
Notestein et al.
Scheme 1. Precursor Synthesis and Grafting Proceduresa
a Conditions: (i) 5a with 1 equiv of TiCl4 in toluene, 60 °C, 24 h, as synthesized by Radius;27 (ii) reflux 5a with 1 equiv of TiCl4 in toluene, 48 h; (iii)
reflux 4b with partially dehydroxylated silica for 24 h in toluene with 10 equiv of 2,6 di-tert-butylpyridine; (iv) reflux 5b with chlorinated silica for 24 h
in toluene with 2 equiv of N(Et)3.
calixarene nm-2 (0.20 mmol g-1) when an excess of 4b was
refluxed in the presence of silica for 280 h. After this point,
additional 4b had no effect on the ultimate surface coverage of
3 determined by TGA and elemental analysis. We note that the
silanol density remains much higher than the calixarene active
site density for all materials synthesized and is not appreciably
changed upon calixarene grafting for silica pretreatment tem-
peratures below 1000 °C.26
A Ti atom:calixarene molecule ratio of 1.0 ( 0.1 was
independently measured in 3 at all surface coverages (0.03-
0.25 nm-2) by combustion analysis and ICP-MS. These
equimolar ratios strongly indicate that the molecular precursor
4b was grafted intact. Bulk calixarene contents were estimated
by total weight loss in TGA above 250 °C (Supporting
Information Figure S1). The onset of weight loss upon heating
3 in air is quite dramatic, with a sharp maximum in the
derivative plot centered between 360 and 380 °C. Calixarene
Spectroscopy. UV-vis spectroscopy was performed on a Varian
Cary 400 Bio UV-vis spectrophotometer equipped with a Harrick
Praying Mantis accessory for diffuse reflectance measurements on solids
at room temperature. Compressed PTFE powders were used as a
reference. Fluorescence spectra were measured using a Hitachi F-4500
fluorescence spectrophotometer with a solids reflectance accessory.
Solid-state 13C CP/MAS NMR was performed at the Caltech solid-
state NMR facility using a Bruker DSX500 spectrometer operating at
500 MHz.
Results and Discussion
Materials Synthesis and General Characterization. The
synthetic scheme for grafting metallocalixarenes is depicted in
Scheme 1 and described in the Experimental Section. The
catalyst synthesis can be generalized to the large library of
upper- and lower-rim substituted calixarenes,23 but care must
be exercised to prevent the formation of multinuclear spe-
cies.9,24,25 Precursor 4b was immobilized with quantitative yield
up to a surface density of 0.18 calixarene nm-2 (0.15 mmol
g-1). The surface density was increased to a maximum of 0.25
(24) Clegg, W.; Elsegood, M. R. J.; Teat, S. J.; Redshaw, C.; Gibson, V. C. J.
Chem. Soc., Dalton Trans. 1998, 18, 3037-3039.
(25) Cotton, F. A.; Dikarev, E. V.; Murillo, C. A.; Petrukhina, M. A. Inorg.
Chim. Acta 2002, 332, 41-46.
(23) Gutsche, C. D. Calixarenes; Royal Society of Chemistry: Cambridge, U.K.,
1992.
(26) Zhuravlev, L. T. Colloids Surf., A 2000, 173, 1-38.
9
16480 J. AM. CHEM. SOC. VOL. 126, NO. 50, 2004

Metallocalixarenes as Organometallic Catalysts
A R T I C L E S
Table 1. Summary of Materials Synthesized
concn,
pore vol,a
cm3/g
surface density,b
[Ti],c
wt %
1
2
ID
µ
mol g-
calixarene nm-
Ti:calixarene
3
3
3
3
3
3
2
203
146
138
75
51
25
0.42
0.48
N/D
N/D
0.58
N/D
0.54
0.25
0.18
0.17
0.09
0.06
0.03
0.14
0.97
0.69
0.65
0.35
0.23
0.11
1.00
0.97
0.99
0.98
d
d
116
a The silica support has a pore volume of 0.69 cm3/g. b Determined by
combustion analysis. c Determined by ICP-MS. d A very low carbon content
places large errors on the Ti:calix ratio. Nonetheless, values between 0.9
and 1.0 are always obtained.
contents were measured from combustion analysis and TGA,
and both methods agreed to within 10%. Nitrogen physisorption
data and BJH pore size distributions are similar to those reported
previously for 17 and show decreasing total pore volume and
average pore radius as calixarene loading is increased (Sup-
porting Information Figure S2). The decrease in pore volume
corresponds to an average calixarene-TiIV molecular volume
of 2.3 nm3, in reasonable agreement with the molecular
dimensions predicted from molecular models (1.4 nm × 1.4
nm × 1.0 nm ) 2.0 nm3). Additional materials characterization
data are provided in Table 1.
Figure 1. 13C CP/MAS NMR spectra of immobilized materials 2 (M )
Si) and 3 (M ) Ti). An asterisk indicates a spinning sideband or surface
methoxide. The appearance of a distinct resonance 7′ is indicative of
phenoxide coordination to Ti. The inset highlights the downfield shift of
resonance 4 in 3 relative to 2 due to methyl ether coordination to Ti.
Materials Spectroscopic Characterization. NMR spectros-
copy provides additional compelling evidence that the atomic
connectivity between Ti and calixarene in 3 is retained during
grafting. Close comparison of 13C NMR spectra for materials 2
and 3 in the solid state and for molecular precursors 5a28 and
4a27 in deuterated chloroform shows that all four spectra are
similar, with the exception of two resonances, 4 and 7′. In the
spectra of the precursors, resonances 4 and 7′ are shifted
downfield 8 and 13 ppm, respectively, when Ti-containing 4a
is compared to unmetalated 5a. Likewise, in the solid-state
spectrum of 3, resonances 4 and 7′ are shifted by 2 and 11 ppm,
respectively, relative to the resonances in 2 (Figure 1). These
shifts are consistent with expected changes in electron density
upon phenoxide (resonance 7′) and methyl ether (resonance 4,
inset) coordination to Ti. The latter coordination represents a
weak dative bond inferred from interatomic distances between
Ti and methyl ether oxygen in structures derived from single-
crystal X-ray diffraction of 4a,27 4b,11 and related titanocalix-
arenes.9 The resonance assignment for carbon 4 was verified
by isotopically enriching the methoxy position in grafted
calixarene materials.
calixarene ligands and also ruling out the presence of extended
Ti-O-Ti connectivity29 even after calcination. Fluorescence
emission spectra (λex ) 270 nm; Supporting Information Figure
S3) of 3 show strong quenching relative to those of both the
unmetalated calixarene 2 and the calcined 6. This quenching is
a common feature of d0 metal-dye complexes,30 and is
consistent with Ti attachment to the calixarene in 3.
We use the position of the calixarene-TiIV LMCT edge as a
probe of catalyst structure, in much the same way as edge
energies are used to characterize dispersed metal oxides.31,32
As seen in Figure 2B, extrapolation of the linear region to the
intercept gives the same edge energy of 2.18 ( 0.02 eV (or
585 nm) for all grafted materials, irrespective of titanium
loading. This strongly suggests that the structure of the
predominant grafted species is independent of calixarene-TiIV
surface density. In dispersed metal oxides, the edge energy is
unchanged at low surface density, because of the predominant
presence of isolated oxo species, and then decreases as metal
centers become linked by bridging oxygens with increasing
surface coverage.31,32 We attribute the constant edge energy in
3 up to the maximum surface loading possible to structural
isolation enforced by the steric bulk of the calixarene and silica
surface as ligands. The edge energy of the precursor molecule
4b lies at 1.90 eV, 0.28 eV lower than that of the grafted 3,
reflecting the change of ligand from R3SiO- in 3 to Cl- in 4b
and confirming that the Ti precursor has been covalently grafted,
instead of merely physisorbed, on the surface.
UV-vis and fluorescence spectroscopies were used to probe
the surface structures formed by grafting of 4b onto silica. The
absorption spectra of precursor 4b and of immobilized 3 and 2
are shown in Figure 2A. All calixarenes exhibit an absorption
maximum at 270-290 nm arising from π-π* transitions, but
only calixarene-TiIV complexes 3 and 4b show an additional
broad absorption feature at 290-500 nm, which we assign to
calixarene-TiIV charge transfer (LMCT) bands. Material 6 was
produced by treatment of 3 in dry air at 500 °C. This process
removed all absorption features in the 290-500 nm range,
confirming that these electronic transitions arise from the
(29) Guidotti, M.; Ravasio, N.; Psaro, R.; Ferraris, G.; Moretti, G. J. Catal.
2003, 214, 242-250.
(30) Rurack, K. Spectrochim. Acta, Part A 2001, 57, 2161-2195.
(31) Barton, D. G.; Shtein, M.; Wilson, R. D.; Soled, S. L.; Iglesia, E. J. Phys.
Chem. B 1999, 103, 630-640.
(27) Radius, U. Inorg. Chem. 2001, 40, 6637-6642.
(28) 13C NMR (400 MHz, C6D6, 298 K) δ 30.63, 31.84 (C(CH3)3), 32.25 (CH2),
33.56, 33.85 (C(CH3)3), 62.79 (OMe), 125.58, 125.93 (Arm), 127.99, 132.34
(Aro), 141.20, 146.46 (Arp), 151.03, 151.63 (Ari).
(32) Gao, X.; Bare, S. R.; Fierro, J. L. G.; Banares, M. A.; Wachs, I. E. J.
Phys. Chem. B 1998, 102, 5653-5666.
9
J. AM. CHEM. SOC. VOL. 126, NO. 50, 2004 16481

A R T I C L E S
Notestein et al.
Figure 2. (A) UV-vis absorption spectra of (a) 2, (b) 4b, (c) 3, and (d) 3 after 25 turnovers to cyclohexene epoxide at 60 °C, 10 min, CHP oxidant in
octane. (B) Absorption edge region of representative catalysts (e) 3-51, (f) 3-146, and (g) 3-203. Extending the linear region of each curve to the baseline
gives an absorption edge energy of 2.18 ( 0.02 eV.
Table 2. Representative Epoxidation Results at 60 °C in Octanea
conversionc
%
k,b
M-2 s-
1
1
no.
catalyst
oxidant
reactant
TOF0,b h-
TON
time, h
1
2
3
4
5
6
7
8
9
10
11
12
13
14
4bd
3-25
3-51
3-75
3-138
3-146
3-203
3-51
3-146
3-203
3-138e
3-146
3-146
3-146
TBHP
TBHP
TBHP
TBHP
TBHP
TBHP
TBHP
CHP
CHP
CHP
CHP
CHP
cyclohexene
cyclohexene
cyclohexene
cyclohexene
cyclohexene
cyclohexene
cyclohexene
cyclohexene
cyclohexene
cyclohexene
cyclohexene
cyclooctene
1-octene
0.1
11.0
11.1
10.8
11.5
11.0
11.3
24.5
27.7
8
181
188
195
180
181
183
381
416
308
1080
690
7
4
15
54
58
92
77
83
97
74
98
97
98
99
22
39
0.5
8
2
5
2
2
2
2
2
633
343
367
165
156
130
338
162
115
202
170
42
24.8
1
f
0.5
1.3
4
f
CHP
CHP
0.5
7.6
cis-stilbene
25
61
3
Literature Comparisons
40
15
16
17
18
19
20
TiCl4‚SiO2
TBHP
TBHP
TBHP
TBHP
CHP
cyclohexene
cyclooctene
cyclohexene
cyclohexene
cyclohexene
cyclohexene
N/A
100
202
172
460
540
90
1120
122
230
72
90
84
50
91
82
98
1.5
22
1
0.5
1.5
2
CpTiSi7O12‚MCM-4141
N/A
N/A
N/A
N/A
N/A
Cp2TiCl2‚MCM-4134
42
dry TiO2/SiO2
20
TiO2/SiO2
Ti(OSiR3)4‚SBA-154
CHP
1500
600
a General reaction conditions are given in the Experimental Section. b k determined from r ) k[cat][P][A] based on the observed epoxide production rate.
TOF0 ) (moles of epoxide/mole of Ti)/hour extrapolated to initial time. c Total conversion of peroxide at a given time. d A 5 µmol sample of 4b, 5 mL of
octane, and 0.65 mmol of cyclohexene. e Only 7 mL of octane used to increase all reactant concentrations. f Reaction was extremely rapid, and insufficient
data could be collected to calculate a meaningful rate constant.
Catalytic Behavior of 3. Material 3 can be expected to be
an active and selective epoxidation catalyst as a result of its
rigid immobilization within a pseudotetrahedral geometry, which
is often cited as a requirement for heterogeneous epoxidation
catalysis on Ti centers.17,18,33-35 Epoxidation rates and selectivi-
ties for several unfunctionalized alkenes (Table 2) were
measured at 60 °C using CHP or TBHP as oxidants and dry
octane as solvent. Material 3 catalyzes cyclohexene epoxidation
with both TBHP and CHP; rates were slightly higher with CHP.
Cyclohexene epoxidation on catalyst 3 typically proceeded to
>95% conversion and >95% selectivity to the epoxide (based
on hydroperoxide). Epoxidation of 1-octene on 3 and cyclo-
hexene on 4b occurs at a very low rate, and as such,
unproductive thermal and surface-catalyzed decomposition of
the hydroperoxide accounts for much of the total conversion,
and the selectivity decreases to 30% and 66%, respectively.
(33) Abbenhuis, H. C. L. Chem.sEur. J. 2000, 6, 25-32.
(34) Maschmeyer, T.; Rey, F.; Sankar, G.; Thomas, J. M. Nature 1995, 378,
159-162.
(35) Imamura, S.; Nakai, T.; Utani, K.; Kanai, H. J. Catal. 1996, 161, 495-
497.
9
16482 J. AM. CHEM. SOC. VOL. 126, NO. 50, 2004

Metallocalixarenes as Organometallic Catalysts
A R T I C L E S
Figure 3. Cyclohexene epoxidation in octane at 60 °C catalyzed by 3: (90) 3-203, (bO) 3-146, (>) 3-139, (1) 3-75, ([]) 3-51, (2) 3-25. Open symbols
are for CHP, and closed symbols are for TBHP. (A) TON to epoxide, CHP oxidant. Lines are fit using the first-order rate law proposed in the text. (B)
Conversion X to epoxide, CHP oxidant. Lines are obtained by linear regression of the data points shown. The slope is proportional to the pseudo-first-order
rate constant k1. (C) Second-order rate constants k2 for CHP (s) and TBHP (---) plotted against total Ti concentration in the reactor. Lines are obtained by
linear regression of the data points shown. The slope represents the intrinsic rate constant for epoxidation of cyclohexene with each oxidant.
Regardless, for all surface coverages and substrates, no side
products (i.e., other C6 oxides for cyclohexene) were detected
by gas chromatography, essentially ruling out contributions from
competing radical pathways.36 Also, epoxidation of cis-stilbene
proceeded with complete retention of configuration to only cis-
stilbene oxide, which confirms that a nonradical mechanism is
responsible for epoxidation in our catalysts.37
The epoxidation of cyclooctene and 1-octene proceeded at
the expected rates relative to those of cyclohexene. The observed
relative substrate reactivity was cyclooctene > cyclohexene .
1-octene, consistent with attack of the alkene at an electrophilic
coordinated peroxide oxygen.38 The successful epoxidation of
stilbene and cyclooctene indicates the absence of significant
steric constraints at the active site.20,34,39
Ti-based heterogeneous epoxidation catalysts. Of particular note
is the high TOF reported for entry 20. This material is a highly
dispersed grafted species that provides good active site acces-
sibility. However, as for other catalysts in Table 2, the TOF of
the catalyst in entry 21 decreases as the surface density increases,
suggesting that site isolation may be incomplete.4 We note that
our reaction conditions are much more dilute than most in the
literature using Ti-based catalysts; thus, rates reported here on
material 3 underestimate those expected at the higher reactant
concentrations of comparative studies.4,20,42
Kinetic Analysis. Figure 3A shows catalytic turnovers to
epoxide (TON ) moles of epoxide/mole of Ti) in CHP and
excess cyclohexene as a function of reaction time using 3 as
catalyst at 60 °C. Figure 3B shows epoxide yields based on
peroxide (X ) moles of epoxide/initial mole of peroxide).
Neither plot shows an induction period, indicating that the
Table 2 also shows that cyclohexene epoxidation rates on
catalyst 3 resemble or exceed those reported on state-of-the-art
(36) Figueras, F.; Kochkar, H. Catal. Lett. 1999, 59, 79-81.
(37) Oldroyd, R. D.; Thomas, J. M.; Maschmeyer, T.; MacFaul, P. A.; Snelgrove,
D. W.; Ingold, K. U.; Wayner, D. D. M. Angew. Chem., Int. Ed. Engl.
1996, 35, 2787-2790.
(38) Sheldon, R. A. J. Mol. Catal. 1980, 7, 107-126.
(39) Dartt, C. B.; Davis, M. E. Appl. Catal., A 1996, 143, 53-73.
(40) Sheldon, R. A.; van Doorn, J. A. J. Catal. 1973, 31, 427-437.
(41) Krijnen, S.; Abbenhuis, H. C. L.; Hanssen, R. W. J. M.; Van Hooff, J. H.
C.; van Santen, R. A. Angew. Chem., Int. Ed. 1998, 37, 356-358.
(42) Corma, A.; Domine, M.; Gaona, J. A.; Jorda´, J. L.; Navarro, M. T.; Rey,
F.; Pe´rez-Parienta, J.; Tsuji, J.; McCulloch, B.; Nemeth, L. T. Chem.
Commun. 1998, 2211-2212.
9
J. AM. CHEM. SOC. VOL. 126, NO. 50, 2004 16483

A R T I C L E S
Notestein et al.
grafted species are the resting state of the active catalyst and
not merely precursors to active structures.43 The data in Figure
3B show a linear dependence of log(1 - X) on time at all
peroxide conversions, and the pseudo-first-order rate constants,
proportional to the slope of these lines, are unaffected by
increasing reaction time. These results indicate that catalyst
deactivation, leaching, or inhibition by products do not occur.
Consistent with this linear behavior, the deliberate addition of
an excess of coproduct alcohol at initial time did not influence
reaction rates or reactant conversions.
during reaction. The first-order dependence on hydroperoxide
was established above, and the dependence on alkene was
verified independently. At the reaction conditions given in
the Supporting Information, the initial TOF increased linearly
from 34 to 65 to 160 h-1 as the cyclohexene concentration
was raised from 18 to 48 to 138 mM (Supporting Information
Figure S4). The final concentration is approximately that of the
reaction system employed in this paper. The catalyst concentra-
tion [cat] is expressed as millimoles of Ti per reactor volume
(L). The rate was obtained from the integrated form of the rate
expression ln(1 - X) ) k[cat][A]t ) k2[A]t ) k1t. A graph of
ln(1 - X)[A]-1 versus time has a slope of k2 and a zero intercept.
A plot of k2 as a function of [cat] (Figure 3C) leads to a straight
line through the origin for single-site materials and a curve of
decaying slope for conventional grafted Ti catalysts.4,32,47
The lack of kinetic inhibition by alcohol products suggests a
strong preference for binding peroxide reactants instead of
abundant alcohol products on Ti active centers. First-order
behavior was observed previously on Ti-silsesquioxane com-
plexes41 and early Ti-SiO2 catalysts,40 but many initially active
epoxidation catalysts deactivate with reaction time,4,18,34 an effect
that has been ascribed to inhibition by alcohol products.
Grafted calixarene-TiIV catalysts reported here can be stored
in ambient air without detectable changes in epoxidation rates
and selectivities compared to freshly prepared materials. This
stability is in marked contrast to the sensitivity of conventional
Ti epoxidation catalysts to atmospheric moisture during storage,
which causes irreversible restructuring around Ti centers. As
with other epoxidation catalysts stable during storage, such as
grafted Ti(OSiPh3)4,2 traces of water during catalysis led to
lower selectivities and rate constants with increasing reaction
time, apparently as a result of acid-catalyzed epoxide hydrolysis,
which forms chelating diols that bind strongly to Ti centers.42
This situation is avoided by conducting the reaction in the
presence of molecular sieves,44,45 as we have performed here.
Experimental Evidence for Single-Site Behavior. Silica gel
surfaces are known to exhibit a distribution of acid sites, and
this can influence the number and type of ligands at a grafted
metal center.46 In contrast, we expect the bonding around the
Ti center in 3 to be largely controlled by the multidentate
calixarene ligand, which should force the Ti into a pseudotet-
rahedral coordination geometry, as in the single-crystal X-ray
diffraction structure of precursor 4b.11 To test this hypothesis,
we examined the epoxidation of cyclohexene using 3 as catalyst,
with a wide range of surface concentrations of Ti active sites
The value of k2 was determined for each catalyst using this
methodology, and the linear dependence shown by the data in
Figure 3C unequivocally shows that intrinsic rate constants
depend only on the number of Ti centers in the reactor and not
on the calixarene-TiIV surface coverage, an accepted hallmark
of single-site catalysts. Also, the use of intrinsic rate constants
instead of initial TOF ensures that these materials behave
identically at all conversions, rather than only during initial
contact with reactants. The rate constants k in Figure 3C are
11.1 ( 0.3 M-2 s-1 for TBHP and 25 ( 2 M-2 s-1 for CHP
over the entire range of calixarene-TiIV surface coverages
(0.03-0.25 nm-2).
This demonstrated single-site behavior differs markedly from
that reported on other Ti-based catalysts, for which epoxidation
turnover rates (per Ti center) either decrease monotonically with
increasing Ti concentration4,32,47 or sharply decrease above a
certain value,12,13,20 apparently because metal centers increas-
ingly reside at inactive or inaccessible surface sites or because
oligomers form with less active and selective Ti-O-Ti
connectivity. We propose that the steric bulk and geometric
constraints provided by the calixarene ligand prevent metal
center clustering or nonuniform coordination geometries even
near saturation coverage.
Identification of Catalytically Relevant Species. Control
experiments first sought to confirm the heterogeneous nature
of the catalysis observed and to determine the stability of 3
against leaching during reaction. The homogeneous precursor
4b44 (Table 2, entry 1) and its tert-butoxy analogue showed
>20-fold lower activity than the grafted material 3. The addition
of pure silica to a reaction mixture containing homogeneous
material 4b synthesizes 3 in situ, as determined by TGA and
UV-vis spectroscopy of recovered solids, and leads to a
concomitant increase in epoxidation rates (Supporting Informa-
tion Figure S5a). This illustrates the importance of Ti-O-Si
connectivity for epoxidation catalysis18,33 and demonstrates
synergy between grafted precursors and the support.
from 0.03 to 0.25 nm-2
.
Catalytic activity in a single-site material depends only on
the total concentration of Ti centers and not their surface
concentration. In Figure 3A, turnover frequencies for cyclo-
hexene epoxidation (TOF ) TON/time) extrapolated to initial
times show identical TOFs (slope) on catalysts with surface
densities between 0.06 and 0.25 nm-2. This behavior is usually
considered sufficient to prove single-site behavior,4 but here
we compare rate constants over the entire time course of reaction
to unequivocally prove the single-site behavior of 3 as catalyst.
In view of the linearity of the conversion data when plotted
as log(1 - X) and the large excess of cyclohexene reactants
used, we use a rate law give by r ) k[cat][P][A], where [P] is
the concentration of hydroperoxide (mol/L) and [A] is the
concentration of alkene (mol/L), which is essentially unchanged
In several experiments, a reaction mixture of catalyst 3-138,
CHP, and cyclohexene was filtered at 60 °C after reaction for
10 min (TON ≈ 30). The filtrate showed no detectable
epoxidation activity over 1 h (Figure S5b). Also, the addition
of pure silica to the filtrate, a procedure that leads to in situ
immobilization of 3 and to the appearance of catalytic activity
(43) Schofield, L. J.; Kerton, O. J.; McMorn, P.; Bethell, D.; Ellwood, S.;
Hutchings, G. J. J. Chem. Soc., Perkin Trans. 2 2002, 1475-1481.
(44) Massa, A.; D’Ambrosi, A.; Pronto, A.; Scettri, A. Tetrahedron Lett. 2001,
42, 1995-1998.
(45) Dusi, M.; Mallat, T.; Baiker, A. J. Catal. 1998, 173, 423-432.
(46) Lefort, L.; Chabanas, M.; Maury, O.; Meunier, D.; Cope´ret, C.; Thivolle-
Cazat, J.; Basset, J. M. J. Organomet. Chem. 2000, 593-594, 96-100.
(47) Ikeue, K.; Ikeda, S.; Watanabe, A.; Ohtani, B. Phys. Chem. Chem. Phys.
2004, 6, 2523-2528.
9
16484 J. AM. CHEM. SOC. VOL. 126, NO. 50, 2004

Metallocalixarenes as Organometallic Catalysts
A R T I C L E S
when homogeneous 4b is present in solution, did not lead to
detectable epoxidation products after an additional 1 h at 60
Ti-peroxide intermediate in heterogeneous epoxidation catalysts
remains unclear.48-51
°C.
The absence of leaching cannot be demonstrated unequivo-
cally from the observed noncatalytic character of the filtrate
because of the low epoxidation rates on homogeneous catalyst
4b and on its tert-butoxy analogue. Recycle experiments (Figure
S5c), in which the catalyst is washed with hot anhydrous octane,
dried, weighed, and used for another catalytic cycle, showed
very similar reaction rates and conversions in subsequent
catalytic runs (within 5%), confirming the absence of permanent
deactivation or leaching of sites. Also, addition of a second
aliquot of hydroperoxide after 100% conversion (Figure S5d)
led to an identical conversion rate, confirming that the catalyst
remains active in the presence of all reaction products and is
not, perhaps, reactivated by the recycling process.
Analysis of the solids recovered after filtration also did not
detect elemental or structural changes indicative of leaching of
calixarenes or of Ti-containing species. TGA and Ti ICP-MS
independently did not detect changes in the carbon or titanium
content within their respective accuracies (∼5%). UV-vis
spectroscopy (Figure 2A) and 13C CP/MAS NMR (Supporting
Information Figure S6) of the solids recovered after catalysis
showed that 3 was essentially unaltered during reaction. When
catalysis was deliberately performed under deactivating condi-
tions (without molecular sieves), the loss in activity was
paralleled by a loss in calixarene content by both UV-vis and
13C CP/MAS NMR of the solids (Figure S6). In the 13C CP/
MAS NMR experiment, aromatic calixarene resonances were
replaced by alkoxide resonances. These data, coupled with an
unchanged Ti content by ICP-MS, suggest that diol formation
from epoxide hydrolysis leads to the displacement of the
calixarene ligand with concomitant loss of catalytic activity.
The bound peroxide may be hydrogen-bonded with adjacent
calixarene phenols or surface silanols as proposed for epoxi-
dation in the presence of polar protic species,50 or the intermedi-
ate could be an η2 complex as proposed for the active site in
grafted complexes51 and observed in model compounds.52 Also,
while we cannot rule out insertion of the peroxide into the
Ti-OSi bond,4,18,49 we believe that insertion into one of the
Ti-OC bonds is more likely. Ti-OC bonds have been previ-
ously shown to be more kinetically labile than Ti-OSi bonds
in the presence of TBHP.5 Particular to our system, the three-
fold multiplicity of Ti-OC bonds relative to a single Ti-OSi
bond favors the proposed structure 7, and cleavage of a
Ti-OC bond will be further favored by release of conforma-
tional strain in the metallocalixarene. This strain is apparent in
the elliptical conformation of the calixarene macrocycle in the
structure of 4b as determined by single-crystal X-ray diffrac-
tion.11 Finally, 7 maintains the Ti-OSi connection required for
the observed heterogeneous catalysis.
Many factors, including steric constraints at the metal,18,33,53
diffusion to the active site,20,39,54 and support hydrophobic-
ity,36,42,55 influence epoxidation catalysis on Ti centers, but
alkene epoxidation rates are expected to increase with increasing
electron-withdrawing power of oxo ligands.40 In this context,
the main role of the calixarene ligand in 3 in epoxidation
catalysis is to withdraw electron density from the peroxide
oxygens in a reaction intermediate such as 7, rendering such
oxygens more electrophilic and prone to alkene attack.
Deliberate calixarene removal by combustion in dry air at
500 °C produces catalyst 6-146, which shows behavior es-
sentially similar to that of the grafted and calcined titanocene
catalyst reported by Maschmeyer et al.34 Under identical
conditions,
the initial epoxide TOF on 6-146 was 504 h-1 as compared to
416 h-1 on 3-146, but unlike the grafted calixarene catalyst,
epoxide production essentially stopped at an 80% yield of
epoxide. After this point, peroxide consumption continued and
a small fraction of the epoxide was decomposed (likely to ring-
opened product). At identical conditions, catalyst 3 continued
to >98% conversion with 100% selectivity. The relative increase
in initial TOF for 6-146 presumably reflects the removal of steric
limitations around the Ti center imposed by the calixarene
ligand. At high conversions, the absence of a calixarene ligand
results in strong inhibition by coproduct alcohol, as reported
previously.34
We suggest that calixarene ligands in 3 perform two
simultaneous functions by (i) introducing steric constraints that
prevent dimerization and formation of weaker Ti-O-Ti Lewis
acids18,34,56 and (ii) promoting Lewis acidity at the Ti center
(48) Bonino, F.; Damin, A.; Ricchiardi, G.; Ricci, M.; Spano`, G.; D’Aloisio,
R.; Zecchina, A.; Lamberti, C.; Prestipino, C.; Bordiga, S. J. Phys. Chem.
B 2004, 108, 3573-3583.
Taken together, these results indicate that material 3 is stable
with respect to ligand loss and does not leach into solution as
less active homogeneous species, such as titanium alkoxides,
which can potentially form during reaction. We conclude,
therefore, that the immobilization of active species 3 is
permanent under the conditions reported here and that it is
required for efficient and selective epoxidation catalysis.
(49) Sinclair, P. E.; Catlow, C. R. A. J. Phys. Chem. B 1999, 103, 1084-1095.
(50) Bellussi, G.; Carati, A.; Clerici, M. G.; Maddinelli, G.; Millini, R. J. Catal.
1992, 133, 220-230.
(51) Adam, W.; Corma, A.; Garc´ıa, H.; Weichold, O. J. Catal. 2000, 196, 339-
344.
(52) Boche, G.; Mo¨bus, K.; Harms, K.; Marsch, M. J. Am. Chem. Soc. 1996,
118, 2770-2771.
(53) Klunduk, M. C.; Maschmeyer, T.; Thomas, J. M.; Johnson, B. F. G. Chem.s
Eur. J. 1999, 5, 1481-1485.
(54) Blasco, T.; Corma, A.; Navarro, M. T.; Pariente, J. P. J. Catal. 1995, 156,
65-74.
Proposed Reaction Intermediate. We propose here structure
7 for the active species as consistent with catalytic and
characterization data, but we note that thebonding mode of the
(55) Mu¨ller, C. A.; Deck, R.; Mallat, T.; Baiker, A. Top. Catal. 2000, 11/12,
369-378.
(56) Sarsfield, M. J.; Ewart, S. W.; Tremblay, T. L.; Roszak, A. W.; Baird, M.
C. J. Chem. Soc., Dalton Trans. 1997, 3097-3104.
9
J. AM. CHEM. SOC. VOL. 126, NO. 50, 2004 16485

A R T I C L E S
Notestein et al.
relative to conventional alkoxide57 or Cp56 ligands. This stronger
Lewis acidity reflects the effective delocalization of electron
density away from the Ti center via Ti-O-Ph π-bonding;11
this mechanism is analogous to that invoked for the increased
Lewis acidity in early-transition-metal complexes with siloxy
ligands relative to those with alkoxy ligands.57-59 A similar
argument has also been invoked for the high activity and stability
of grafted transition-metal alkyls as polymerization catalysts.60
method for creating surface organometallic species since we
have shown that the calixarene-TiIV connectivity has been
maintained from the crystallographically characterized precursor
to the active catalytic species. We are currently exploiting the
diversity and flexibility made available by calixarene ligands
and the possibility of modifying the support properties for
additional improvements in epoxidation rates and selectivities
on grafted metallocalixarene materials.
Conclusions
Acknowledgment. We acknowledge the Exxon-Mobil Foun-
dation and the University of California at Berkeley, Department
of Chemical Engineering for financial support. J.M.N. is grateful
to the National Science Foundation for a graduate fellowship.
A.K. acknowledges the receipt of a Hellman Family Faculty
Award and a 3M Pre-Tenure Faculty Award. We acknowledge
Dr. Sonjong Hwang at the Caltech solid-state NMR facility for
his technical assistance.
We demonstrate here a novel route for the synthesis of
isolated Ti-oxo complexes via grafted metallocalixarenes and
their use as highly active and selective epoxidation catalysts.
The single-site nature of the observed electronic transitions and
catalysis on these materials confirms our premise that the steric
bulk of calixarene ligands prevents oligomerization events
leading to inactive and unselective Ti-O-Ti structures. The
coordination environment enforced by the cone calixarene ligand
leads to strong Lewis acidity and to faster epoxidation rates
compared with those in conventional Ti centers with alkoxy
and Cp ligands. The calixarene ligand provides steric constraints
around each Ti center, which leads to remarkable stability during
ambient storage and epoxidation catalysis, as well as to 100%
selectivity to epoxide products. We expect this to be a general
Supporting Information Available: Catalysis method and
substrate purification details, characteristic TGA of 3-146
(Figure S1), characteristic nitrogen physisorption data (Figure
S2), characteristic fluorescence emission spectra (Figure S3),
investigation of reaction order in alkene concentration (Figure
S4), silica addition (Figure S5a), filtration (Figure S5b), recycle
(Figure S5c), and second aliquot (Figure S5d) catalysis experi-
ments, and UV-vis (Figure S6a) and 13C NMR (Figure S6b)
comparison for fresh, active, and deactivated catalysts. This
material is available free of charge via the Internet at
http://pubs.acs.org.
(57) Coffindaffer, T. W.; Steffy, B. D.; Rothwell, I. P.; Folting, K.; Huffman,
J. C.; Streib, W. E. J. Am. Chem. Soc. 1989, 111, 4742-4749.
(58) LaPointe, R. E.; Wolczanski, P. T.; Van Duyne, G. D. Organometallics
1985, 4, 1810-1818.
(59) Rice, G. L.; Scott, S. L. Langmuir 1997, 13, 1545-1551.
(60) Ballard, D. G. H.; Heap, N.; Kilbourn, B. T.; Wyatt, R. J. Makromol. Chem.
1973, 170, 1-9.
JA0470259
9
16486 J. AM. CHEM. SOC. VOL. 126, NO. 50, 2004