Epoxidation with Ti(OSiMe3)4: A homogeneous model mimicking titania-silica mixed oxides?

Article abstract of DOI:10.1039/b300809f

Ti(OSiMe₃)₄ is an active homogeneous catalyst for the epoxidation of allylic alcohols under mild conditions using tert-butyl hydroperoxide as oxidant in an apolar solvent. It affords an average turnover frequency of up to 1470 h^-1 and 100% regioselectivity (geraniol) or 100% cis-selectivity (cyclohexenol). UV-Vis measurements indicate that the ability of substrates to coordinate to the catalyst determines the reaction rate and stereoselectivity. This is the likely explanation for the poor or completely missing reactivity of substrates possessing no OH function (cyclohexene, methoxycyclohexene) and also for the substrate-dependent effect of water in the system. Ti(OSiMe₃)₄ is suggested to be a good soluble homogeneous model catalyst mimicking the highly active isolated tetrahedral Ti sites in (silylated) titania-silica mixed oxides.

Full text of DOI:10.1039/b300809f

View Article Online / Journal Homepage / Table of Contents for this issue
Epoxidation with Ti(OSiMe3)4 : a homogeneous model mimicking
titania–silica mixed oxides?
Carsten Beck, Tamas Mallat and Alfons Baiker*
Institute for Chemical- and Bioengineering, ETH-Ho¨nggerberg, HCI, CH-8093, Zu¨rich
Switzerland. E-mail: baiker@tech.chem.ethz.ch; Fax: +41 1 632 11 63
Received (in Montpellier, France) 20th January 2003, Accepted 7th April 2003
First published as an Advance Article on the web 9th July 2003
Ti(OSiMe3)4 is an active homogeneous catalyst for the epoxidation of allylic alcohols under mild
conditions using tert-butyl hydroperoxide as oxidant in an apolar solvent. It affords an average turnover
frequency of up to 1470 hꢀ1 and 100% regioselectivity (geraniol) or 100% cis-selectivity (cyclohexenol).
UV-Vis measurements indicate that the ability of substrates to coordinate to the catalyst determines the
reaction rate and stereoselectivity. This is the likely explanation for the poor or completely missing
reactivity of substrates possessing no OH function (cyclohexene, methoxycyclohexene) and also for the
substrate-dependent effect of water in the system. Ti(OSiMe3)4 is suggested to be a good soluble
homogeneous model catalyst mimicking the highly active isolated tetrahedral Ti sites in (silylated)
titania–silica mixed oxides.
hydrogen peroxide can be used as oxidant.13–15 The role of
water in the formation of the active Ti sites and in the epoxi-
dation mechanism has been addressed by several research
groups.16–19
Various titanium silasesquioxane complexes have been
proposed as soluble models for the catalytically active cen-
ters in TS-1.20–23 Interestingly, tert-butyl hydroperoxide
(TBHP) was used as oxidant as none of these complexes
were active with aqueous H2O2 .24 Grafting a titanium silases-
quioxane onto a hydrophobic polysiloxane demonstrated that
Introduction
Sol-gel titania–silica mixed oxides, either as mesoporous aero-
gels1–4 or microporous xerogels,5,6 are powerful epoxidation
catalysts. Their activity is attributed to the high dispersion of
Ti in the silica matrix (site isolation) and to a high surface area.
A limitation to practical application is their strongly hydrophi-
lic character, which necessitates the use of organic peroxides as
the oxidant in non-aqueous medium. Silylation of the pre-
formed mixed oxides improves the catalytic performance
remarkably and diminishes the sensitivity towards water.7–9
Upon silylation, a more hydrophobic surface is obtained by
partial transformation of the surface silanol groups to trimethyl-
=
=
an ideal model for TS-1 should possess isolated Ti(OSi )
n
(n ¼ 3, 4) sites in a hydrophobic environment.24 Other lim-
itations of titanium silasesquioxane complexes as soluble,
homogeneous model catalysts are the presence of Ti–O–Ti
bonds in the structure and slightly distorted tetrahedron
bond angles.21
=
siloxy functions. In addition, the ( SiO) TiOH type active
=
sites are transformed into ( SiO) TiOSiMe ones (Scheme 1).
3
=
=
3
3
Another approach to synthesize hydrophobic mixed oxides
is the partial replacement of the tetraalkoxysilane precursor
by an alkyltrialkoxysilane in the sol-gel process. Introduction
of surface methyl or phenyl groups afforded excellent cata-
lysts for the epoxidation of olefins and allylic alcohols, even
in the presence of water.6,10–12
For comparison, in the microporous crystalline titanium
silicalite TS-1 the isolated Ti sites are located in hydrophobic
channels or cavities of the molecular sieve and aqueous
During the sol-gel synthesis of titania–silica aerogels various
active sites may form, ranging from isolated Ti atoms sur-
rounded by four siloxy groups (most active and selective) to
titania nanodomains possessing mainly Ti–O–Ti linkages
(barely active and selective).25 This structural variety may well
be tuned by the appropriate choice of catalyst precursors and
preparation conditions but the synthesis of a uniform structure
has not yet been achieved.4,26,27 The presence of structurally
different Ti sites in the amorphous silica matrix hampers the
unambiguous interpretation of the catalytic properties and
the development of a feasible mechanistic model. Hence, the
importance of using a homogeneous model of the isolated Ti
site in the silica matrix is obvious.
The aim of the present work was to test Ti(OSiMe3)4 as a
simple and commercially available model compound for a
stable and isolated tetrahedral Ti site surrounded by four
siloxy groups. This compound seems to be a particularly
attractive model of silylated titania–silica (Scheme 1). The cata-
lytic performance of Ti(OSiMe3)4 has been investigated in the
epoxidation of allylic alcohols, an allylic ether and cyclohex-
ene (Scheme 2). Reaction rates, chemo- and stereoselectivities,
and sensitivity to water are compared to the corresponding
values typical for solid Ti- and Si-containing epoxidation
catalysts.
Scheme 1 Schematic representation of an isolated Ti site in silylated
titania–silica (left) and of the homogeneous catalyst Ti(OSiMe3)4-
(right).
1284
New J. Chem., 2003, 27, 1284–1289
DOI: 10.1039/b300809f
This journal is # The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2003

View Article Online
column. The epoxides were identified by comparison with
authentic samples and by GC-MS. The internal standard
method was used for quantitative analysis. 100% yield could
not be achieved due to various side reactions, including
Lewis acid catalyzed and radical reactions. Taking the
epoxidation of 2-cyclohexen-1-ol with titania–silica as an exam-
ple, the following major products and by-products were
detected: cyclohexanol oxide (epoxidation), cyclohexan-1-ol-
3-one (rearranged epoxide), cyclohexenone, cyclohexanone
oxide and 1,4-cyclohexene diol (allylic oxidation). Besides,
the carbon balance indicated some oligomerization; these
heavy products could not be detected by GC. More details
of typical side reactions catalyzed by titania–silica have been
published recently.31 The same products as a result of Lewis
acid catalyzed and radical reactions were identified from epoxi-
dations with Ti(OSiMe3)4 .
Scheme 2 Substrates used in the epoxidation reactions (1–5) and in
the UV-Vis measurements (1s–3s).
Experimental
Materials
TBHP conversion was determined by iodometric titration
using a Metrohm 686 Titroprocessor. Two different types of
chemoselectivities are calculated: epoxide selectivity related
to the reactant consumed:
The following commercially available chemicals were used:
toluene (Fluka, 99.5%, abs.), dodecane (Merck, 99%), n-hex-
ane (Fluka, 99.7%, abs.), anisol (Fluka, 99%), geraniol (1,
Aldrich, 98%), 2-cyclohexen-1-ol (2, Fluka, 97%), tert-butyl
hydroperoxide (TBHP, Fluka, 5.5 N in nonane, additionally
SC Cð%Þ ¼ 100 ꢁ ½epoxideꢂ=ð½reactantꢂ ꢀ ½reactantꢂÞ
B
0
˚
dried over freshly activated 4 A molecular sieve), Ti(OSiMe3)4
and epoxide selectivity related to the peroxide consumed:
(ABCR, 99.5%, stored under argon), Si(OSiMe3)4 (Acros,
98%), 3,7-dimethyloctan-1-ol (1s, Aldrich, 99%), cyclohexanol
(2s, Fluka, 99%), cyclohexene (5, Merck, 99%).
STBHPð%Þ ¼ 100 ꢁ ½epoxideꢂ=ð½TBHPꢂ0 ꢀ ½TBHPꢂÞ:
Methoxycyclohexene (3, 97%)28 and 2-cycloocten-1-ol (4,
97%)29 were synthesized according to known procedures.
Methoxycyclohexane (3s, 99.5%) was prepared as follows: 10
ml anisol was hydrogenated at 100 bar and room temperature
with 75 mg catalyst (5 wt. % Rh/C, Engelhard) under vigorous
stirring. After 17 h, the catalyst was filtered off, the filtrate
distilled at 318 K (38 mbar) and the product identified with
GC-MS (m/z ¼ 114, 85, 82, 71, 67, 58, 55, 45, 41) and 1H-
NMR [d ¼ 1.27 (m, 5H), 1.53 (m, 1H), 1.74 (m, 2H), 1.93
(m, 2H), 3.14 (m, 1H), 3.34 (s, 3H)].
Epoxide yield and olefin conversion are related to the initial
amount of TBHP (limiting component). The reaction rate
is characterizꢀed1 by the turnover frequencies (mol epoxide ꢃ
mol Ticatalyst
ꢃ hꢀ1): the average turnover frequency
(TOF50%) denotes the rate of epoxide formation at 50% TBHP
conversion and the initial TOF0 is related to 2 min reaction
time. Selectivities, conversions and yields are calculated at
50% TBHP conversion, where t50% denotes the reaction time
needed for 50% TBHP conversion.
The amorphous mesoporous titania–silica aerogel Ae-1 (1
wt. % TiO2 , 1060 m2 gꢀ1 BET surface area, 3.1 cm3 gꢀ1 pore
volume) and the silylated aerogel Ae-1-sil (1 wt. % TiO2 , 750
m2 gꢀ1 BET surface area, 2.7 cm3 gꢀ1 pore volume) were pre-
pared on the basis of a former recipe.25 The high degree of
dispersion of Ti in the silica matrix was confirmed by UV-Vis
DRS, DRIFT, XPS25 and XANES30 methods.
Results and discussion
Effect of substrate structure on the epoxidation rate and
selectivity
The catalytic performance of Ti(OSiMe3)4 has been tested in
the epoxidation of some open chain and cyclic allylic alcohols
[geraniol (1), cyclohexenol (2), cyclooctenol (4)] and a cyclic
olefin (cyclohexene, 5). Epoxide formation at 333 K is shown
in Fig. 1 and the initial rates and selectivities are collected
in Table 1. Clearly, Ti(OSiMe3)4 is an active and selective
homogeneous epoxidation catalyst, providing high regio- and
diastereoselectivities in the epoxidation of allylic alcohols.
The critical role of the allylic OH group in epoxide
formation is demonstrated by the complete unreactivity of
methoxycyclohexene (3). No epoxide was detected even at
Methods
UV-Vis measurements were performed on a Perkin Elmer
Lambda 16 spectrometer and the spectra recorded above 210
nm. To solutions of Ti(OSiMe3)4 in dry n-hexane (c ¼ 0.356
mmol lꢀ1) were added 1s, 2s or 3s in 400-fold molar excess
and TBHP in 100-fold molar excess. The solutions were freshly
prepared and the same solutions but without Ti(OSiMe3)4
were used as reference. The experimental error is estimated
to be less than 10% in absorption area.
All epoxidation reactions were conducted under argon to
avoid the presence of oxygen and moisture. In the standard
epoxidation procedure the reactor was pre-dried at 473 K in
an argon flow for 1 h, than solvent, internal standard (dode-
cane) and 30 ml (0.067 mmol) Ti(OSiMe3)4 were added and
the epoxidation was carried out at 333 K. When the reaction
was catalyzed by an aerogel, 0.1 g (0.0125 mmol Ti) catalyst
was pre-dried in vivo in the reactor at 473 K for 1 h in an argon
flow. To the aerogel catalyst were then added solvent and 0.5
ml internal standard. This mixture was heated to the reaction
temperature. Olefin (10 mmol) was added and the reaction
started by adding 2.5 mmol TBHP in nonane (olefin : TBHP :
Ti(OSiMe3)4 ¼ 150 : 37.5 : 1). In some experiments up to
0.75 mmol H2O was added (H2O : Ti(OSiMe3)4 ¼ 11 : 1).
The total reaction volume was 10 ml.
The products were analyzed by an HP-6890 gas chromato-
graph equipped with a cool on-column inlet and an HP-FFAP
Fig. 1 Epoxidation of geraniol (1), 2-cyclohexen-1-ol (2), cyclo-
octenol (4) and cyclohexene (5); standard conditions.
New J. Chem., 2003, 27, 1284–1289
1285

View Article Online
Table 1 Epoxidations with Ti(OSiMe3)4 under dry conditionsa
of the substrates and TBHP, present in high excess, decreased
the intensity of the band at 210–215 nm and produced a new
broad band in the range of 240–280 nm. The changes are illu-
strated by the difference spectra in Fig. 2. The absorption
intensity of the Ti(OSiMe3)4 band at 210–215 nm decreases
in the following order upon complexation: TBHP > 3,7-di-
methyl-octane-1ol (1s) > cyclohexanol (2s) ꢆ methoxycyclo-
hexane (3s). The observed shift upon adding 3s was minor
and did not exceed the estimated experimental error. The com-
plexation order is attributed to steric hindrance in 2s relative to
1s and to the missing OH function in 3s. When the results are
transferred from the hydrogenated model substrates 1s–3s to
the substrates 1–3, this order corresponds well to the reactivity
order observed in epoxidations with Ti(OSiMe3)4 (Table 1).
Apparently, the ability of the substrate to coordinate to the
catalyst determines the rate of the epoxidation reaction,
assuming a similar epoxidation mechanism.
b
c
TOF50%
Substrate hꢀ1
/
t50%
min
/
% RS/
% STBHP % Yield DSd
% SC
=
C
1
2
3
4
5
1470
186
0
0.9
5
72
83.5
–
92
76
46
38
0
100
100
–
–
–
180
7
225
95
38
100
60
50
30
91.5
3.1
–
a
Standard reaction conditions, T ¼ 333 K. b Average TOF at 50% TBHP
c
d
conversion. Reaction time until 50% TBHP conversion. Regioselectiv-
ity for 2,3-epoxide for geraniol (1), diastereoselectivity for cis-epoxide for
cyclohexenol (2) and cyclooctenol (4).
363 K after 4 h reaction time. This shows the influence of
decreased nucleophilicity when interaction via the alcoholic
OH group is impossible, and also the effect of some steric hin-
drance by the –OMe group. For comparison, the epoxidation
of cyclohexene (5) was also strongly retarded by the missing
OH interaction. The initial rate was lower and the selectivities
were poor, compared to those achieved in the epoxidation of
cyclohexenol (2). Besides, epoxidation of cyclohexene was
always accompanied by a short initial period of a few minutes,
in which no epoxide was produced. To sum up, the epoxida-
tion activity (TOF50%) of Ti(OSiMe3)4 decreases in the order:
1 ꢄ 2 > 4 > 5 > 3 ( ¼ 0).
Catalyst stability—effect of water
As mentioned in the Introduction, the behavior of solid Ti- and
Si-containing catalysts towards water is remarkably different.
The hydrophylic sol-gel titania–silica mixed oxides hydrolyze
and deactivate in the presence of water, while TS-1 is com-
monly used for epoxidation in aqueous medium and water is
assumed to be a ligand in the active peroxo complex.37,38
The best known homogeneous catalyst, Ti(OPr-i)4 in the
presence of alkyltartrates (Sharpless epoxidation) is retarded
by water even at 253 K.39,40 Clearly, interaction of Ti(OSiMe3)4
with water is important for understanding the characteristics
of this catalyst.
The role of small amounts of water in the epoxidation of
various substrates is illustrated in Figs. 3–5. In general, the
faster the epoxidation under dry conditions, the smaller is
the effect of water on the reaction rate and (stereo)selectivity.
Using a H2O : Ti(OSiMe3)4 ¼ 11 : 1 molar ratio, in the epoxi-
dation of geraniol (1) the initial rate decreased by only 16%,
while the initial reaction rates were undetectably low in the
epoxidation of cyclooctenol (4) and cyclohexene (5) (see Fig.
3). Considering the average reaction rate for up to 4 h reaction
time, the slow epoxidation of 5 was retarded only slightly,
compared to the reaction under dry conditions. The retarding
effect of H2O on the epoxidation of 4 as a function of water :
catalyst ratio is shown in Fig. 4. Increasing water concentra-
tion led to a steady decrease of TOF [Fig. 4(a)].
A closely related catalyst, Ti(OSiPh3)4 , was found to be
almost inactive in the epoxidation of cyclohexene with
TBHP.22 The poor activity may be attributed to the increased
steric demand of OSiPh3 groups as compared to OSiMe3
groups, or to the higher stability of Ti–O–Si bonds in Ti(OSi-
Ph3)4 by electron orbital mixing, taking into account a prob-
able epoxidation mechanism involving Ti–O–Si bond cleavage.
=
Geraniol (1), possessing two isolated C C double bonds,
represents a suitable structure to gain information about
regioselectivity of the epoxidation reaction. Reaction at the
double bond remote to the OH functional group would be
favorable due to its enhanced nucleophilicity, whereas epoxi-
dation of the allylic double bond is directed by interaction of
the OH group with the Ti-hydroperoxy complex. In epoxida-
tions with Ti(OSiMe3)4 the 2,3-epoxide is formed exclusively
(Table 1).
Diastereoselectivity is another valuable source of informa-
tion on the hydroxy-directing effect in epoxidation reactions.
Epoxidation of cyclohexenol (2; dihedral angle of 139ꢅ)32
yielded solely the cis-epoxide (Table 1). Cyclooctenol (4) with
In the epoxidation of 1 and 2 the high stereoselectivities
remained unaffected in the presence of water. In contrast,
32
a dihedral angle of 199ꢅ was epoxidized mainly to the cis-
epoxide. It has been shown that diastereoselectivity in the
epoxidation of cyclic allylic alcohols depends on steric and
electronic factors, which are governed by the oxidizing agent
32–34
=
and the ring size (dihedral angle, C C–C–O).
UV-Vis analysis of catalyst-substrate-oxidant interactions
Differences in the complexation behavior of the open chain
allylic alcohol 1, the cyclic allylic alcohol 2 and the cyclic allylic
ether 3 were analyzed by UV-Vis spectroscopy. In these
measurements the saturated forms of olefins 1–3 (denoted as
1s–3s, Scheme 2) were used to mimic the catalyst-substrate
interaction but minimize the UV absorption above 210 nm.
Ti(OSiMe3)4 showed an intense absorption in the range of
210–215 nm, absent from the spectrum of the corresponding
silicon compound Si(OSiMe3)4 . This band was assigned to a
ligand-to-metal charge transfer (LMCT) transition of a four-
fold coordinated Ti center.35,36 Note that a change in the
LMCT band can either occur due to additional coordination
of the substrate directly to the Ti site or to H-bonding of the
substrate to the O atom of an –OSiMe3 ligand. Adsorption
Fig. 2 UV-Vis analysis of the interaction of Ti(OSiMe3)4 with TBHP
and the substrates 1 and 2, mimicked by the use of their saturated
forms: 3,7-dimethyloctan-1-ol (1s), cyclohexanol (2s) and methoxycy-
clohexene (3)s. The difference between the spectra obtained in the
presence and absence of additive is shown to emphasize the changes.
Ti(OSiMe3)4 : substrate molar ratio ¼ 1 : 400, Ti(OSiMe3)4 : TBHP ¼
1 : 100.
1286
New J. Chem., 2003, 27, 1284–1289

View Article Online
Fig. 3 Effect of water on the initial rate (TOF0) in the epoxidation of
geraniol (1), 2-cyclohexen-1-ol (2), cyclooctenol (4) and cyclohexene
(5). H2O : Ti(OSiMe3)4 molar ratio is 11; standard reaction conditions.
Fig. 5 Effect of water on the reaction rate (TOF) in the epoxidation
of geraniol (1) under otherwise standard reaction conditions. (L) Dry
conditions; (X) H2O added together with TBHP at zero time, H2O :
Ti(OSiMe3)4 molar ratio is 11; (`) Ti(OSiMe3)4 prehydrolyzed for
7.5 min at 333 K before addition of TBHP and 1, H2O : Ti(OSiMe3)4
molar ratio is 1.
water had a dramatic influence on the epoxidation of 4 as
shown in Fig. 4(b). Increasing the H2O : Ti(OSiMe3)4 molar
ratio from 0 to 11 successively lowered the fraction of cis-epox-
ide from 91% to 3.5%. Considering the initial selectivities, the
effect of water is even more striking: 100% cis-epoxide under
dry conditions after 2 min reaction time and 100% trans-epox-
ide with an eleven-fold excess of water after 7.5 min reaction
time. All these observations point towards detrimental struc-
tural changes induced by hydrolysis, resulting in a loss of
activity and selectivity.
7.5 min at 333 K (‘‘prehydrolysis’’) before the addition of
cyclooctenol (4), epoxidation was almost completely blocked
as illustrated in Fig. 4(a) (bottom line). The similar influence
of prehydrolysis of the catalyst in the absence of geraniol (1)
is shown in Fig. 5. When water is added at zero time together
with TBHP, the decrease in reaction rate was minor, compared
to the reaction under dry conditions. However, prehydrolysis
of the catalyst for 7.5 min at 333 K with only a small amount
of water [H2O : Ti(OSiMe3)4 ¼ 1] before adding geraniol and
TBHP lowered the reaction rate dramatically. One hundred
percent conversion was achieved only after 4 h reaction time,
though the regioselectivity to 2,3-epoxide was still excellent
(99%). Apparently, strong complexation of the allylic alcohol
substrate present in excess [olefin : Ti(OSiMe3)4 ¼ 150 : 1]
can retard hydrolysis of the catalyst. Note that different struc-
tures have been proposed for the ‘‘hydroxy-assisted’’ mechan-
ism of allylic alcohol epoxidation over Ti- and Si-containing
catalysts.41,42
In the above experiments water was added at zero time,
together with TBHP. When the reaction mixture in the pre-
sence of water [H2O : Ti(OSiMe3)4 ¼ 11 : 1] was stirred for
Comparison to other Ti- and Si-based catalysts
The performance of Ti(OSiMe3)4 , Ae-1-sil, Ae-1 and TS-1 is
compared in Table 2 for the epoxidation of 2-cyclohexene-1-
ol (2), a commonly used test reaction. TS-1 and Ae-1 are the
least active catalysts, though a reliable comparison is not pos-
sible due to the strikingly different reaction conditions. Silyla-
tion of the aerogel Ae-1 (to Ae-1-sil) increased the epoxide
yield but barely influenced the diastereoselectivities. The bene-
ficial influence of silylation of titania–silica mixed oxides has
generally been attributed to the reduced hydrophylic character
Table 2 Comparison of various Ti- and Si-based catalysts in the
epoxidation of 2-cyclohexen-1-ol (2)a
b
c
TOF0
hꢀ1
/
t50%
min
/
% Yield % cis:trans
Catalysts
Peroxide
TBHP
at t50%
at t50%
100 : 0d
76 : 24
79 : 21
90 : 10
Ti(OSiMe3)4
(0.067 mmol Ti)
Ae-1-sil
222
48
50
8
5
204
240
–
38
TBHP
41
(0.0125 mmol Ti)
Ae-1
TBHP
33
(0.0125 mmol Ti)
TS-1e
(0.11 mmol Ti)
Aq. H2O2
68
Fig. 4 Influence of water on the (a) reaction rate (TOF) and (b)
stereoselectivity in the epoxidation of cyclooctenol (4). The H2O :
Ti(OSiMe3)4 molar ratios are 0, 3.7, 7.5 and 11. Water is added at zero
time or else the catalyst is prehydrolyzed for 7.5 min in the absence of
4, under otherwise standard reaction conditions.
a
b
Standard reaction conditions, T ¼ 333 K. Initial TOF0 after 2 min reaction
time. c Reaction time until 50% TBHP conversion. d trans-epoxide not detected.
e
30% H2O2 , acetone, 8 h reflux.41
New J. Chem., 2003, 27, 1284–1289
1287

View Article Online
of the surface.7,9,25 Still, even Ae-1-sil is inferior to the homo-
geneous model catalyst Ti(OSiMe3)4 when considering either
the initial reaction rate per Ti site or the diastereoselectivity
to cis-cyclohexanol oxide. Ti(OSiMe3)4 shows the potential
of a completely silylated aerogel possessing isolated tetrahedral
Ti sites and a hydrophobic surface derived by protection of all
Ti–OH and Si–OH groups.
The aerogel Ae-1 and the homogeneous model Ti(OSiMe3)4
were compared also in two other reactions, in the epoxidations
of an open chain allylic alcohol (geraniol, 1) and a cyclic olefin
(cyclohexene, 5) (see Table 3). Ti(OSiMe3)4 was superior in the
epoxidation of 1 in which reaction the interaction of the allylic
OH group with the active site likely plays a crucial role in the
mechanism with both catalysts. In contrast, Ae-1 was far more
active in the epoxidation of 5. Note that epoxidation of cyclic
olefins is the best catalytic use of titania–silica mixed oxides.43
The low activity of Ti(OSiMe3)4 in the epoxidation of olefins
without an OH group (such as 3 and 5, Table 1) indicates that
the use of this compound as a homogeneous model of titania–
silica mixed oxides may be limited to the epoxidation of unsa-
turated alcohols.
geraniol can retard hydrolysis and good yields and excellent
stereoselectivities are obtained even in the presence of water.
We propose that Ti(OSiMe3)4 may be a suitable homoge-
neous model catalyst, mimicking the active sites in titania–
silica mixed oxides, particularly silylated titania–silica, in
the epoxidation of unsaturated alcohols. This soluble cata-
lyst possesses a tetrahedral Ti site isolated by four siloxy
groups, which seems to be structurally closely related to
the most active Ti sites in the mixed oxides. The sensitivity
to water is similar for the two catalysts. Application of
Ti(OSiMe3)4 as a homogeneous model catalyst can promote
theoretical calculation of the possible reaction mechanism in
epoxidations with titania–silica. A limitation of Ti(OSiMe3)4
as a homogeneous model catalyst is the partly different
‘‘ligand’’ environment of Ti, that is OSiMe3 groups in the
=
soluble catalyst instead of OSi(OSi ) groups in the mixed
3
=
oxide. This limitation is less important for silylated tita-
nia–silica mixed oxides with the general structure shown in
Scheme 1.
The sensitivity of Ti(OSiMe3)4 to water is moderate, com-
pared to other homogeneous Ti catalysts. For example, the
Ti catalyzed asymmetric epoxidation of allylic alcohols in the
presence of trace amounts of water (i.e., when the reaction is
carried out in the absence of a molecular sieve) affords low
conversion and enantioselectivity.39 Among the solid catalysts,
titania–silica mixed oxides are rapidly deactivated by water but
silylation diminishes this sensitivity.11 Hence, in this respect
Ti(OSiMe3)4 seems to be a good model for (silylated) titania–
silica.
References
1
2
3
4
H. D. Gesser and P. C. Goswami, Chem. Rev., 1989, 89, 765.
M. Schneider and A. Baiker, Catal Rev, 1995, 37, 515.
R. Hutter, T. Mallat and A. Baiker, J. Catal., 1995, 153, 177.
J. B. Miller, L. J. Mathers and E. I. Ko, J. Mater. Chem., 1995,
5, 1759.
5
M. A. Holland, D. M. Pickup, G. Mountjoy, E. S. C. Tsang,
G. W. Wallidge, R. J. Newport and M. E. Smith, J. Mater. Chem.,
2000, 10, 2495.
6
7
Y. Deng and W. F. Maier, J. Catal., 2001, 199, 115.
M. L. Pena, V. Dellarocca, F. Rey, A. Corma, S. Coluccia and
L. Marchese, Microporous Mesoporous Mater., 2001, 44, 345.
J. Bu and H. K. Rhee, Catal. Lett., 2000, 65, 141.
A. Corma, M. Domine, J. A. Gaona, J. L. Jorda, M. T. Navarro,
F. Rey, J. Perez-Pariente, J. Tsuji, B. McCulloch and L. T.
Nemeth, Chem. Commun., 1998, 2211.
The main limitation of Ti(OSiMe3)4 as a homogeneous
model is that it represents an ideal tetrahedral Ti site isolated
=
by four –O–SiMe3 ligands, instead of –O–Si(–O–Si )
=
8
9
3
‘‘ligands’’ corresponding to the silica matrix in titania–silica
(Scheme 1). It is very likely that this ligand effect also contri-
butes to the excellent activity and selectivity of Ti(OSiMe3)4
as compared to that of Ae-1-sil. A logical conclusion is that
Ti(OSiMe3)4 should be considered as a good homogeneous
model for silylated titania–silica in which comparison the dis-
torting effect of different ligands of the isolated Ti site is smal-
ler. Besides, it is conceivable that not only the improved
hydrophobicity but also the changes in the ligand environment
of the Ti site contributes to the better catalytic performance of
silylated titania–silica, as compared to the original mixed
oxide.
10 H. Kochkar and F. Figueras, J. Catal., 1997, 171, 420.
11 F. Figueras and H. Kochkar, Catal. Lett., 1999, 59, 79.
12 C. A. Muller, R. Deck, T. Mallat and A. Baiker, Top. Catal.,
¨
2000, 11/12, 369.
13 B. Notari, Adv. Catal., 1996, 41, 253.
14 B. Notari, Catal. Today, 1993, 18, 163.
15 C. B. Khouw, C. B. Dartt, J. A. Labinger and M. E. Davis,
J. Catal., 1994, 149, 195.
16 A. Zecchina, S. Bordiga, C. Lamberti, G. Ricchiardi, D. Scarano,
G. Petrini, G. Leofanti and M. Mantegazza, Catal. Today, 1996,
32, 97.
17 P. E. Sinclair, G. Sankar, C. R. A. Catlow, J. M. Thomas and
T. Maschmeyer, J. Phys. Chem. B, 1997, 101, 4232.
18 C. M. Zicovich-Wilson, R. Dovesi and A. Corma, J. Phys. Chem.
B, 1999, 103, 988.
Conclusions
19 G. Ricchiardi, A. de Man and J. Sauer, Phys. Chem. Chem. Phys.,
2000, 2, 2195.
20 H. C. L. Abbenhuis, S. Krijnen and R. A. van Santen, Chem.
Commun., 1997, 331.
21 M. Crocker, R. H. M. Herold, A. G. Orpen and M. T. A.
Overgaag, J. Chem. Soc., Dalton Trans., 1999, 3791.
22 J. M. Thomas, G. Sankar, M. C. Klunduk, M. P. Attfield, T.
Maschmeyer, B. F. G. Johnson and R. G. Bell, J. Phys. Chem.
B, 1999, 103, 8809.
Ti(OSiMe3)4 is a good homogeneous catalyst that is active and
highly stereoselective in the epoxidation of open chain and cyc-
lic allylic alcohols. Epoxidation of other substrates possessing
no OH function is slow or shows no reaction at all. UV-Vis
spectroscopy confirmed the importance of the substrate-cata-
lyst interaction via the OH functional group and the role of
steric restrictions in achieving a high rate. Though the catalyst
is sensitive to water, a strongly coordinating reactant such as
23 M. C. Klunduk, T. Maschmeyer, J. M. Thomas and B. F. G.
Johnson, Chem.-Eur. J., 1999, 5, 1481.
24 M. D. Skowronska-Ptasinska, M. L. W. Vortsenbosch, R. A.
Van Santen and H. C. L. Abbenhuis, Angew. Chem., Int. Ed.,
2002, 41, 637.
Table 3 Comparison of Ti(OSiMe3)4 and Ae-1 in the epoxidation of
geraniol (1) and cyclohexene (5)a
25 C. Beck, T. Mallat and A. Baiker, J. Catal., 2001, 204, 428.
26 G. Cerveau, R. J. P. Corriu and E. Framery, J. Mater. Chem.,
2000, 10, 1617.
27 K. Kosuge and P. S. Singh, J. Phys. Chem. B, 1999, 103, 3563.
28 J. Lessard, P. V. M. Tan, R. Martino and J. K. Saunders, Can.
J. Chem., 1977, 55, 1015.
29 H. Meier, C. Antony-Mayer, C. Schulz-Popitz and G. Zerban,
Liebigs Ann. Chem., 1987, 1087.
30 J. D. Grunwaldt, C. Beck, W. Stark, A. Hagen and A. Baiker,
Phys. Chem. Chem. Phys., 2002, 4, 3514.
Ti(OSiMe3)4
Ae-1
Substrate TOFb 50%/hꢀ1 t50%c /min TOFb 50%/hꢀ1 t50%c /min
1
5
1470
3.1
0.9
225
42.5
98
96
57
a
b
Standard reaction conditions, T ¼ 333 K. Average TOF at 50%
c
TBHP conversion. Reaction time till 50% TBHP conversion.
1288
New J. Chem., 2003, 27, 1284–1289

View Article Online
31 C. Beck, T. Mallat and A. Baiker, Catal. Lett., 2001, 75, 131.
32 W. Adam, C. M. Mitchell and C. R. Saha-Moller, Eur. J. Org.
Chem., 1999, 785.
33 T. Itoh, K. Jitsukawa, K. Kaneda and S. Teranishi, J. Am. Chem.
Soc., 1979, 101, 159.
34 W. Adam and A. K. Smerz, Tetrahedron, 1995, 51, 13 039.
35 C. K. Jørgensen, Modern Aspects of Ligand Field Theory, North-
Holland, Amsterdam, 1971.
38 P. E. Sinclair and C. R. A. Catlow, J. Phys. Chem. B, 1999, 103,
1084.
39 Y. Gao, R. M. Hanson, J. M. Klunder, S. Y. Ko, H. Masamune
and K. B. Sharpless, J. Am. Chem. Soc., 1987, 109, 5765.
40 M. G. Finn and K. B. Sharpless, J. Am. Chem. Soc., 1991,
113, 113.
41 R. Kumar, G. C. G. Pais, B. Pandey and P. Kumar, J. Chem.
Soc., Chem. Commun., 1995, 1315.
36 J. A. J. Duffy, J. Chem. Soc., Dalton Trans., 1983, 1475.
37 W. Adam, A. Corma, A. Martinez, C. M. Mitchell, T. I.
Reddy, M. Renz and A. K. Smerz, J. Mol. Catal. A, 1997,
117, 357.
42 W. Adam, A. Corma, T. I. Reddy and M. Renz, J. Org. Chem.,
1997, 62, 3631.
43 M. Dusi, T. Mallat and A. Baiker, Catal. Rev. Sci. Eng., 2000,
42, 213.
New J. Chem., 2003, 27, 1284–1289
1289