Synthesis of Hydrophobic TiO2-SiO2 Mixed Oxides for the Epoxidation of Cyclohexene

Article abstract of DOI:10.1006/jcat.1997.1820

A series of Ti-silica mixed oxides were prepared using sol-gel polycondensation of Ti and Si alkoxydes. The polarity of the surface was changed by substituting partly tetraethoxysilane by phenyl or methyl triethoxysilane. This type of solid adsorbs much less water but more n-hexane than the mixed oxide obtained by the normal procedure; thus it shows hydrophobic character. From UV-vis and IR spectra, titanium is incorporated in the lattice of silica. Drying with supercritical carbon dioxide at low temperature produces a mesoporous solid which was used as catalyst for the reaction of cyclohexene with hydrogen peroxide or tert-butyl hydroperoxide (TBHP). The selectivity of epoxidation with TBHP is very high and reaches >95% in acetonitrile as solvent. In acetonitrile as solvent, the rate of epoxidation of cyclohexene with TBHP decreases when the polarity of the surface decreases, but the reverse effect is observed when tert-butanol is the solvent. The reaction with hydrogen peroxide (30% in water) produces both cyclohex-2-en-1-ol and epoxide by two parallel paths. In presence of an aprotic solvent the oxygen balance overpasses 100%, which means that radical reactions involving atmospheric oxygen are involved. The selectivity of the allylic attack leading to cyclohex-2-en-1-ol is sensitive to the nature of the solvent and is minimized in tert-butanol. The epoxide is partly hydrolyzed to glycol, whatever the solvent and polarity of the surface. These catalysts epoxidize norbornene with high selectivity using hydrogen peroxide as oxidant.

Full text of DOI:10.1006/jcat.1997.1820

JOURNAL OF CATALYSIS 171, 420–430 (1997)
ARTICLE NO. CA971820
Synthesis of Hydrophobic TiO –SiO Mixed Oxides
2
2
for the Epoxidation of Cyclohexene
Hafedh Kochkar and Fran c¸ ois Figueras
Institut de Recherches sur la Catalyse du CNRS, 2 Avenue A, Einstein 69626, Villeurbanne, Cedex, France
Received November 28, 1996; revised May 26, 1997; accepted July 7, 1997
i
been obtained by the reaction of silica (4) with Ti(OPr )4
A series of Ti-silica mixed oxides were prepared using sol-gel or MCM-41 with a titanocene (5). These solids are not real
polycondensation of Ti and Si alkoxydes. The polarity of the sur- substitutes of TS-1, since they are active in epoxidation us-
face was changed by substituting partly tetraethoxysilane by phenyl
or methyl triethoxysilane. This type of solid adsorbs much less wa-
ter but more n-hexane than the mixed oxide obtained by the normal
procedure; thus it shows hydrophobic character. From UV-vis and
IR spectra, titanium is incorporated in the lattice of silica. Dry-
hydrophilicity attributed to the hydroxyl groups of silica:
ing with supercritical carbon dioxide at low temperature produces
a mesoporous solid which was used as catalyst for the reaction of
ing hydroperoxides but not diluted hydrogen peroxide. TiF4
anchored on silica has also been proposed and is indeed
active with concentrated hydrogen peroxide (6). It is com-
monly agreed that this behaviour could be related to the
water is strongly adsorbed and the organic substrate has a
limited access to the surface (7). Silanols are present at the
surface as defects, the number of which depends on the con-
cyclohexene with hydrogen peroxide or tert-butyl hydroperoxide
(TBHP). The selectivity of epoxidation with TBHP is very high
and reaches >95% in acetonitrile as solvent. In acetonitrile as sol- ditions of crystallization. Blasco et al. (8) recently reported
vent, the rate of epoxidation of cyclohexene with TBHP decreases the synthesis of an Al free Ti-BEA (F) zeolite, grown in a
when the polarity of the surface decreases, but the reverse effect is fluoride medium, showing a high crystallinity and adsorbing
observed when tert-butanol is the solvent. The reaction with hydro-
gen peroxide (30% in water) produces both cyclohex-2-en-1-ol and
epoxide by two parallel paths. In presence of an aprotic solvent the
small amounts of water (only 2 H2O/Ti). This hydropho-
bic solid is much more active and selective than a con-
ventional Ti-BEA (OH) when using hydrogen peroxide as
oxygen balance overpasses 100%, which means that radical reac-
oxidant.
tions involving atmospheric oxygen are involved. The selectivity of
Hydrophobic solids can also be obtained by sol-gel chem-
the allylic attack leading to cyclohex-2-en-1-ol is sensitive to the na-
istry. Indeed sol-gel is very attractive since it permits us to
prepare homogeneous mixed oxides of high specific area
ture of the solvent and is minimized in tert-butanol. The epoxide is
partly hydrolyzed to glycol, whatever the solvent and polarity of the
(
9). Moreover, it is long known (10) that it is possible to
surface. These catalysts epoxidize norbornene with high selectivity
using hydrogen peroxide as oxidant.
�c 1997 Academic Press
cocondensate chlorosilanes with alkylchlorosilanes with in-
corporation of phenyl groups at the surface. The C–Si bond
is chemically stable in acid media and it has been shown that
these phenyl groups are stable up to 623 K in air (11, 12).
The polycondensation of Ti alkoxide and alkyl or arylalkox-
INTRODUCTION
The development of redox zeolites in which a fraction of ides of silicium results in the substitution of hydroxyl by
Si atoms is replaced by transition metal cations has demon- phenyl or methyl groups and then in a hydrophobic solid.
strated the potential of these materials as catalysts for ox- This type of modification opens then a route to a control of
idation and epoxidation in liquid phase (1). The main in- the polarity of the surface. Klein and Maier (13) recently
convenience of these solids for many applications is the reported the synthesis of microporous amorphous Ti–Si
presence of microporosity which limits the accessibility to mixed oxides made from the polycondensation of titanium
substrates of small kinetic diameter. For instance, cyclohex- tetraisopropoxide, tetramethylethoxysilane, and methyltri-
ene cannot be epoxidised with titanosilicalite-1, which has ethoxysilane. These solids show pores of about 0.7 nm, but
the MFI structure (2).
relatively low activities in epoxidation with hydrogen per-
One of the hypotheses in this field is that site isolation oxide.
could yield an active solid, the porosityofwhich could be en-
Dutoit et al. (14) performed a systematic investigation of
gineered more easily. The first attempt in this direction was the different parameters of preparation of titania–silica gels
the Shell catalyst obtained by the surface reaction of tita- on their activity. These solids were found active for epoxi-
nium tetrachloride with silica (3). Active catalysts have also dation with organic peroxides only. With H2O2 as oxidant
420
0021-9517/97 $25.00
Copyright �c 1997 by Academic Press
All rights of reproduction in any form reserved.

HYDROPHOBIC TITANIA–SILICA GELS
421
the samples were inactive, owing either to severe inhibi-
Drying was done according to three different procedures
tion of the catalyst by water or to leaching of Ti from the previously described by Dutoit et al. (14): (a) conventional
solid. The method used for drying was a most critical point. drying in air at 393 K for 24 h; (b) high temperature su-
Drying at room temperature yielded microporous xerogels percritical drying using isopropanol at 533 K; and (c) low
of low activity. Drying in supercritical conditions with al- temperature extraction using supercritical CO2 at 313 K
cohols led to a macroporous solid, in which titania segre- and 24 MPa.
gated due to hydrolysis of Si–O–Ti bonds. The application
to these titania–silica catalysts of the techniques of super- (2) Physico Chemical Characterization
critical drying using CO2 produced a mesoporous solid in
The composition of the samples was determined by
which Ti was incorporated to the lattice and more selective
atomic absorption spectroscopy after dissolution of the
than Ti on silica for many epoxidations.
solids. The specific BET surface area and average pore ra-
The effective concentration of water at the surface of
dius were obtained from the isotherms of nitrogen adsorp-
the catalyst can also be controlled by the use of a proper
tion at 77 K, using a homemade automatic volumeter. The
solvent for the reaction. Indeed the comparison of TS-1 and
samples were first evacuated at 473 K for 16 H.
The FTIR spectroscopic study of the lattice was per-
Ti-BETA for the oxidation of n-hexene and cyclohexanol
by hydrogen peroxide has shown that the solvent must be
formed on a Bruker lFS-48 instrument. Sample wafers con-
chosen according to the polarity of the surface (15); for
sisted of 99 mg of dry KBr and 1 mg sample. The sam-
the epoxidation of n-hexene, the best solvent is methanol
ple cell was purged with a flow of dry nitrogen during the
measurements, 100 scans were accumulated for each spec-
(
protic) for TS-1 and acetonitrile (aprotic) for Ti-beta.
This paper is devoted to the preparation and charac-
�
1
trum in transmission at a spectral resolution of 4 cm . The
spectrum of dry KBr was taken for background substrac-
tion. UV-visible spectra were recorded on a Perkin Elmer
Lambda 9 spectrometer using a homemade cell equipped
with quartz windows.
terization of hydrophobic mixed oxides obtained by sub-
stitution of surface OH groups by phenyl groups using
sol-gel synthesis with mixtures of tetraethoxysilane, phenyl-
triethoxysilane, and titanium tetraisoproxide. The reaction
used to investigate the catalytic properties was the epoxida-
tion of cyclohexene since it has been used very commonly
as a model reaction. Both hydrogen peroxide and alkyl hy-
droperoxides were used as oxidizing agents.
The amount of water adsorbed by the different solids was
determined by microgravimetry, using Sartorius equipment
on samples previously dried at 473 K under a flow of nitro-
gen and then evacuated at the same temperature for 1 h.
Water adsorption was measured at 293 K. The isotherms of
adsorption show no saturation, and therefore, the different
samples were characterized by the amount adsorbed at a
relative partial pressure p/p0 = 0.5.
EXPERIMENTAL METHODS
(
1) Synthesis of the Titania–Silica Gels
The method is the sol-gel technique, i.e., polyconden-
Adsorption of hydrogen peroxide: this adsorption was
investigated by infrared spectroscopy, UV-vis spectroscopy,
and by EPR. The sample in powder form was treated with
hydrogen peroxide (30% aqueous solution) and then dried
at room temperature in a vacuum. The resulting solid was
analyzed by the different techniques, using a wafer in KBr
for infrared spectroscopy, or a powder for UV-vis and EPR.
sation of Ti and Si alkoxides, catalyzed by acids. In or-
der to change the polarity of the surface a mixture of
tetraethoxysilane (TEOS, Aldrich, 98% ) and phenyltri-
ethoxysilane (Aldrich, 98% ) was used. A typical prepara-
tion consists in a prehydrolysis at 343 K for 2 h of 0.166 mol
of tetraethoxysilane (or mixture of TEOS and phenyltri-
ethoxysilane) in a mixture of isopropanol (50 ml), water
(
6.25 ml), and HCl (0.3 ml of aqueous solution at 37% ). The
(
3) Catalytic Experiments
solution is then heated at 363 K and a solution (0.01 mol
in 11.4 ml) of titanium tetraisopropoxide (TiOP; Aldrich)
The oxidation reactions have been carried out at 348 K,
in isopropanol (Rh oˆ ne Poulenc, Normapur) is added. Fi- in a 20-ml round-bottom flask immersed in a thermostat-
nally water (6.25 ml) is added dropwise (time of addition: ted bath and equipped with a condenser, a thermometer,
3
0 min), and a transparent gel is obtained, which is dried in and a magnetic stirrer. In a typical experiment 63 mmol of
different conditions. alkene, 5 g of solvent, and 250 mg of catalyst were mixed
The ratio P of phenyltriethoxysilane to tetraethoxysilane in the flask and heated at the desired reaction tempera-
was changed from 0 to 0.6 in order to control the surface ture under stirring (1000 rpm). Then 3.125 mmol of diluted
properties. Phenyltriethoxysilane was also substituted in a hydrogen peroxide (30% aqueous solution) were instanta-
few cases by methyltriethoxysilane in order to study the in- neously added to the mixture (time zero) in order to permit
fluence of the size of the surface organic moiety. In that case a kinetic study of the reaction. Small samples were taken
the ratio methyltriethoxysilane/tetraethoxysilane is defined (1 �l) with a syringe at different reaction times through a
as Q.
septum fitted with the flask, diluted in the solvent, to which

4
22
KOCHKAR AND FIGUERAS
n-decane was added as the internal standard for quantita-
tive analysis.
The products were analyzed by gas chromatography
TABLE 2
Surface Areas and Porosities of Titania–Silica Mixed Oxides Pre-
pared Using Different Amounts of Alkyl or Aryltriethoxysilane
ATEOS) in the Reaction with Tetraethoxysilane and Titanium
(
SHIMADZU GC 14 B) equipped with a capillary col-
(
umn SUPELCOWAX 10 TM, 30-m length, 0.25 mm ID, Tetraisopropoxide
and 0.25 �m film thickness), and a flame ionization de-
tector (FID). The consumption of hydrogen peroxide was
Pore
%
ATEOS
Surface area radius
measured by iodometric titration, using a potentiometric
detector equipped with two similar platinum electrodes.
2
Sample
(mol% )
% Ti
(m /g)
(nm)
Xerogel
Aerogel P0
0
0
0.74
0.74
553
730
489
624
595
411
<1
3.4
3.5
3.7
<1
<1
RESULTS
Aerogel P10 10% phenyltriethoxysilane 0.99
Aerogel P20 10% methyltriethoxysilane 0.76
Xerogel
Xerogel
(
1) Characterization of the Mixed Oxides
0
3.68
20% phenyltriethoxysilane 3.86
The chemical analysis of the samples was performed be-
fore and after the chemical test in order to check the pos-
sible dissolution of titanium in the reaction medium. The
results reported in Table 1 show that no Ti was lost when
the catalytictest wasperformed with hydrogen peroxide. By
contrast, traces amounts of Fe, Ni, and Cr were observed
on the aerogels, but not on the xerogels. These metals are
characteristic of stainless steel, and introduced by the cor-
rosion of the autoclave by the acid contained in the original
gel. Fe is well known to promote Fenton activation of hy-
drogen peroxide. Its possible catalytic effect was checked
on Fe/SiO2 model catalysts obtained by the surface reac-
tion of Fe(acac)3 with Aerosil Degussa 200, followed by a
calcination of the sample.
All these gels are X-ray amorphous. The surface areas
and porosities of the gels are reported in Table 2. As gen-
erally reported the gels dried in air (xerogels) are micro-
porous solids of high surface area. The surface areas of
these materials are high in all cases and supercritical drying
shows the main advantage to increase the size of the pores,
but not the surface area. As expected from the method of
preparation the texture of the gels is not very dependent on
the amount of Ti. The substitution of surface OH groups by
phenyl or methyl groups does not dramatically affect the
surface area or porosity.
The UV-visspectra are reported in Figs. 1a–3a. Theyshow
a band at 220–230 nm, which is found in TS-1 at 210–215 nm
(
16) and is considered as characteristic of the incorporation
of titanium in the lattice of silica. A further argument for
this incorporation is the shift of this band to 260 nm when
pyridine is adsorbed on the sample (Fig. 4). In this case the
coordination of Ti is increased, which accounts for the shift
of the band. This change proves that Ti atoms are coordi-
natively unsaturated and therefore can have a coordination
of 4 or 5.
The samples obtained by supercritical drying using iso-
propanol show by contrast a UV band at 260 nm result-
ing from the formation of titanium oxide by hydrolysis of
Ti–O–Si bonds by the residual water present in the system,
as reported previously by Dutoit et al. (14).
TABLE 1
Chemical Composition of the Aerogels
Sample
% Ti
% C
Fe (ppm)
Ni (ppm)
Cr (ppm)
Ti-free silica
P0
P0
P0
P10
P20
Q10
Q20
0
147
102
25
18
63
27
0.74
0.77
0.86
0.99
0.86
0.76
0.86
8.8
After reaction using TBHP
After reaction using H2O2
13.59
75
11
21
SiO2 Degussa
Fe/SiO2-1
Fe/SiO2-2
Ti-HMS
TS-1
<5
375
933
<10
<10
<10
20
<10
<10
FIG. 1. UV-visible spectra of an hydrophilic TiO2–SiO2 xerogel
0.74% Ti): (a) original sample; (b) after treatment by H2O2 and drying in
vacuum.
0.87
0.8
<0.3
(

HYDROPHOBIC TITANIA–SILICA GELS
423
FIG. 4. UV-visible spectra of TiO2–SiO2 aerogel (P0, 0.74% Ti), (a) in
vacuum and (b) after adsorption of pyridine.
and in the mixed oxides (Figs. 6a and 7a). Then both UV-vis
and IR spectra suggest that Ti is effectively incorporated in
the lattice of silica in these gels and, thus, that site isolation
has been reached.
FIG. 2. UV-visible spectra of an hydrophobic TiO2–SiO2 xerogel ob-
tained with 10% phenyltriethoxysilane in the gel containing 0.83% Ti:
a) in vacuum; (b) after treatment by H2O2 and drying in vacuum.
(
The adsorption of water was performed in order to inves-
tigate the possible hydrophobicityofthe surface. The results
obtained for a series of samples are reported in Table 3.
In both cases of xerogel and aerogel the introduction of
The FTIR spectra in the range of lattice vibrations of two
characteristic xerogels are given in Figs. 5–7. They show the
�
1
presence of a band at 930–940 cm , attributed by many
authors to the deformation of SiO4 tetrahedra by the for-
mation ofSi–O–Tibonds(14, 17–23). Two recent arguments
support this assignment: the absence of this band in Ti-free
Si gels (13) and its shift upon isotopic exchange with 18O in
the case ofTi-MOR (23). Thisband appearsin TS-1(Fig. 5a)
FIG. 3. UV-visible spectra of TS-1 (0.8% Ti): (a) in vacuum; (b) after
FIG. 5. FTIR spectra of TS-1 in the range of lattice vibrations: (a)
treatment by H2O2 and drying in vacuum.
before and (b) after treatment with hydrogen peroxide.

4
24
KOCHKAR AND FIGUERAS
TABLE 3
Adsorption Properties of Titania–Silica Mixed Oxides
Containing 0.8% Ti
wt% adsorption of
%
ATEOS
Sample
(mol% )
Water
Hexane
Xerogel
Xerogel
Aerogel P0
Aerogel P10
Aerogel P20
0
11.6
4.3
15
7.6
10
20% phenyltriethoxysilane
0
10% phenyltriethoxysilane
20% phenyltriethoxysilane
5
8
6
3
14
alkyl or aryl groups decreases the adsorption of water by
the solid. The adsorption of water decreases and that of
n-hexane increases when the number of phenyl groups in-
creases. This is the behaviour expected for hydrophobic
solids. The 3% of water adsorbed by the gel obtained using
2
9
0% of phenyltriethoxysilane represents a ratio H2O/Ti =
, which is intermediate between the values 2.1 for an hy-
drophobic Ti-BEA (F) and 36.4 for a weakly hydrophilic
Ti-BEA (OH) reported by Blasco et al. (8).
FIG. 6. FTIR spectra in the range of lattice vibrations of an hydro-
phobic TiO2–SiO2 xerogel obtained with 10% phenyltriethoxysilane in the
gel, containing 0.83% Ti: (a) before and (b) after treatment with hydrogen
peroxide.
(
2) Adsorption of Hydrogen Peroxide
Former work has evidenced the formation of peroxo
species on TS-1, and it was therefore of interest to check
their formation in the present case. TS-1 was taken as ref-
erence since it has been investigated in detail. By the in-
teraction of hydrogen peroxide with the solid, the band at
�
1
9
60 cm in infrared decreases in intensity on TS-1 (Fig. 5b)
and on the hydrophobic xerogel (Fig. 6b), but it is less af-
fected in the case of the hydrophilic xerogel (Fig. 7b).
This interaction also changes the UV-vis spectra; with
TS-1, the band at 220–230 nm decreases in intensity and
is shifted to 250 nm, while a new band appears at about
3
80 nm. This behavior has been attributed to the formation
of a peroxo species by substitution of a OH species by OOH
from hydrogen peroxide (16). A weak band is observed at
the similar position with the hydrophobic xerogel, which
suggests that the same species is formed at the surface, but
not with the hydrophilic gel. The interaction of hydrogen
peroxide is then weaker in that case.
Upon interaction of the hydrophobic xerogel with hydro-
gen peroxide a paramagnetic species was observed by ESR
as reported in Fig. 8. Both hydrophilic and hydrophobic
samples show the same ESR signal observed at room tem-
perature after adsorption of hydrogen peroxide and vac-
uum drying at room temperature. However, the intensity is
much greater in the case of the hydrophobic surface.
(
3) Catalytic Experiments
FIG. 7. FTIR spectra in the range of lattice vibrations, of an hy-
drophilic TiO2–SiO2 xerogel (0.7% Ti): (a) before and (b) after treatment
The epoxidation of cyclohexene yields the epoxide,
with hydrogen peroxide.
the corresponding glycol, but also cyclohex-2-en-1-ol, the

HYDROPHOBIC TITANIA–SILICA GELS
425
FIG. 10. Selectivitiesofthe hydrophobicP0 aerogel, containing0.74%
Ti, as a function of conversion, in the same conditions as Fig. 9.
catalysts, an aerogel P0 and a Ti-HMS, prepared according
to the method reported by Tanev et al. (25), which contains
no impurities and has been calcined in air at 773 K. The
results show that in both cases the selectivity to cyclohex-2-
en-1-ol is high at low conversion and rather constant. The
selectivity to the epoxide decreases with time while that for
the glycol increases from very low values. It can then be con-
cluded that the epoxide and cyclohex-2-en-1-ol are primary
products of the reaction and that glycol is a secondary prod-
uct. The ratio cyclohex-2-en-1-ol/(epoxide + glycol) varies
FIG. 8. ESR spectrum at room temperature of a hydrophobic xerogel
0.83% Ti) after contact with hydrogen peroxide and vacuum drying at
room temperature.
(
corresponding ketone and epoxide. Allylic alcohols can be little; therefore the oxidations follow two parallel pathways:
formed by different schemes including rearrangement of a radical path leading to cyclohex-2-en-1-ol and a presum-
epoxides (24); therefore the kinetics of their formation was ably ionic path leading to the epoxide, as formerly proposed
investigated. The curves selectivity/conversion of cyclohex- by Sheldon and van Doorn (26) for the homogeneously
ene are reported in Figs. 9 and 10 for two representative catalyzed reaction (Scheme 1).
The catalytic properties can then be expressed by the ini-
tial reaction rate and the selectivity of attack at the double
bond, which includes the formation of epoxide and glycol
versus allylic oxidation, i.e. the sum of cyclohex-2-en-1-ol,
the corresponding ketone and epoxide.
The epoxide can be hydrolyzed to the glycol, and the se-
lectivity to the glycol corresponds then to the difference of
the selectivity for double bond attack and epoxide forma-
tion.
Several parameters can be analyzed including the polar-
ity of the surface, the influence of the oxidizing agent, and
FIG. 9. Selectivities of a Ti-HMS sample (0.87% Ti) as a function
of conversion. The reaction conditions are the standard ones, and 1,2-
dimethoxyethane is used as solvent.
SCHEME 1

4
26
KOCHKAR AND FIGUERAS
TABLE 4
Activities and Selectivities of Aerogels Obtained Using Hydro-
TABLE 6
Catalytic Properties of Different Aerogels Using H2O2 (30%)
gen Peroxide or tert-Butyl Hydroperoxide for the Oxidation of Cy- as Oxidant, tert-Butanol as Solvent, Urea as Basic Additive
clohexene in tert-Butanol as Solvent, Using a Reactor Flushed by (Urea)/(H2O2) = 0.01 (mol/mol), and Reactor Flushed by a N2 Flow
Nitrogen
P0
P10
P20
Ti-free
Ti-free
silica
�
1
� 1
)
silicaa
P10a
b
P10b
Rate (mmol � min � g
0.06
51
0.072
60
0.12
54
Selectivity for C == C attack
Rate (mmol � min� � g
1
� 1
)
0.032
26
25.6
60
83
15.4
0.15
79
31
98
100
30
0.13
85
85
51
69
43
0.68 Selectivity to epoxide (mol% )
45
65
29
8
41
86
36
6
42
97
41
3
Select for C == C attack
Select to epoxyde (mol)
Conversion of H2O2 or TBHP
Efficiency of H2O2 or TBHP
Yield of epoxyde
90
Conversion of H2O2 (% )
90
100
100
90
Epoxide yield
Amount of water adsorbed (wt% )
3
.3. Influence of the surface polarity. The polarity can
a
H2O2 (30% ).
TBHP in decane.
be changed by substituting OH groups for phenyl groups,
which corresponds as reported above to different amounts
of water adsorbed at the surface. The catalytic properties
of these aerogels are reported in Table 6. The solid pre-
pared with 20% phenyltriethoxysilane shows a higher ac-
tivity, but the same selectivity as the hydrophilic Ti-silica
sample (P0).
b
the effect of the solvent. In the following, the selectivity for
C == C attack represents the relative amount of cyclohexene
epoxide and the corresponding glycol to that of cyclohex-
2
-en-1-ol and the corresponding ketone and epoxide.
.1. Influence of the oxidizing agent. The results are re-
ported in Table 4 for two aerogels containing Ti (Si/Ti =
00) or not, using tert-butanol as solvent. It appears that
3
3
.4. Solvent effects. Protic and aprotic solvents of dif-
ferent polarity were used for this reaction in the presence
of hydrogen peroxide. The results are reported in Table 7. In
methanol (dielectric constant 32.7) no products of reaction
were formed. The distribution of products depends signif-
icantly on the solvent used when H2O2 is the oxidant. The
results reported in Tables 7 and 8 illustrate the fact that 1,2
dimethoxyethane is a very particular solvent, since in that
case the number of moles of oxygenated products surpasses
the amount of hydrogen peroxide engaged in the reaction.
1
the hydrophobic aerogel is far more selective to the epox-
ide when the organic hydroperoxide is used as reactant. The
Ti-free aerogel used as reference shows a surprisingly high
conversion, with both types of oxidizing agents, which we
attribute to the impurities, and a high selectivity for allylic
attack with hydrogen peroxide. In that case the hydrolysis
of the epoxide is minimized.
3
.2. Effect of basic additives. Basic additives are com- Pyridine addition has little effect on this process which is
monly added in order to improve the selectivity to the epox- also observed with Ti-HMS. This sample, prepared from the
ide. As reported in Table 5, activity decreases in the pres- same components as the gels, in the presence of a dodecy-
ence of basic additives such as the selectivity for double lamine, is calcined in air at 923 K and contains no impurities.
bond attack. By contrast the addition of bases blocks the In tert-butanol, by contrast, the oxygen balance is respected.
secondary reaction of hydration of the epoxide. The selec- The distribution of products with 1,2 dimethoxyethane as
tivity of C == C attack also decreases in presence of air.
solvent is similar to that obtained in the reaction of cyclo-
hexenyl hydroperoxide with cyclohexene.
TABLE 5
TABLE 7
Effect of the Additives (Ratio Additive/H2O2 = 0.01) and of the
Atmosphere of Reaction on the Catalytic Properties of Catalyst P10
in tert-Butanol as Solvent
Influence of the Solvent on the Oxidation of Cyclohexene
by Hydrogen Peroxide on a P10 Sample
1
,2 Dimethoxy-
Nitrogen atmosphere
ethane
Acetonitrile tert-Butanol
No
Iso-
Air additive Urea Bu4NOHa quinoline
� 1
� 1
)
Rate (mmol � min � g
0.29
68
0.08
55
0.15
79
Selectivity for C == C attack
Rate (mmol � min� � g
1
� 1
)
0.15
0.15
79
31
98
30
0.07
60
41
86
36
0.04
62
46
11
5
0.1
54
53
45
24
(mol% )
Selectivity epoxide
Conversion H O (mol% )
Efficiency of H O (mol% )
Epoxide yield
Dielectric constant
Sel. C == C attack (mol% ) 72
51
81
231
41
36
100
98
36
37.5
36
98
100
30
Selectivity to epoxide
Conversion H2O2 (% )
Epoxide yield
27
71
19
2
2
2
2
20
10.9
a
Tetrabutylammonium hydroxyde (40 wt% ).

HYDROPHOBIC TITANIA–SILICA GELS
427
TABLE 8
Distribution of the Products for the Reaction of Hydrogen Peroxide with Cyclohexene in a Reactor Open
to the Atmosphere in Different Solvents
Millimoles of product formed
Selectivities (mol% )
Initial
rate
No.
Sample
C == C attack
H2O2
1
2
3
4
5
6
7
8
9
P0
0.89
0.2
0.3
0.21
0.16
0.23
0.09
0.15
2.4
3.0
3.74
3.8
2.6
4.34
1.9
1.08
3.18
0.24
0.25
0.44
0.5
0.2
—
—
0.30
2.38
1.64
1.99
1.45
1.6
0.04
0.08
0.32
0.24
0.07
0.42
0.04
—
0.26
0.27
0.19
0.1
—
0.94
—
0.66
0.18
1.26
1.5
0.76
2
0.07
1.82
0.18
58
55
67.5
68.5
70.5
50
163
180
231
242
147
269
100
100
P0 + pyr
P10
P10 + pyr
Q10
1.1
Ti-HMS (0.87% Ti)
4.82
0.97
0.82
—
Q10 + pyr
67
72
P10
—
0.24
Q10
0.31
Note. Solvent: 1,2 dimethoxyethane in experiments 1–7 and 9, and tert-butanol in experiment 8. The oxidant is hydrogen
peroxide (30% aq) in experiments 1–8 and cyclohexenyl hydroperoxide in experiment 9.
The interaction between solvent effects and surface agents. Model catalysts prepared on silica were then used
polarity was also investigated in the case of tert-butyl- containing a few ppm of Fe. A calcination in air up to 773 K
hydroperoxide as oxidant, by comparing the initial rates was performed before the catalytic test in order to obtain
3+
of reaction on hydrophilic and hydrophobic catalysts. The Fe cations. The results are reported in Table 10.
selectivity to the epoxide and the selectivity for the perox-
3
.6. Influence of the structure of the substrate. Since the
ide are high in all the cases; therefore, the reaction is very
specific. The results reported in Table 9 show that acetoni-
trile is the best solvent for the reaction, but that a decrease
of the polarity has a beneficial influence when tert-butanol
is used and a negative effect when acetonitrile is the sol-
vent. A similar effect has been reported by Corma et al.
mechanism can change with the availability of allylic hy-
drogens in the substrate, cyclohexene, norbornene, and �-
pinene were compared. The results are reported in Table 11.
DISCUSSION
(
15) for reactions with hydrogen peroxide on hydrophobic
In agreement with former work (7, 9, 13, 14, 19), Ti can
be incorporated into the lattice of silica using sol-gel chem-
istry. The partial substitution of tetraethoxysilane by phenyl
or methyltriethoxysilane permits us to prepare solids with
hydrophobic character. The simultaneous control of the re-
tention of the structure and a reasonable porosity can be
reached by low temperature supercritical extraction of the
solvent using CO2.
TS-1 and more hydrophilic Ti-BETA.
3.5. Effect of the impurities on the solid. The chemi-
cal analysis shows that some Fe, Cr, and Ni impurities are
present on the aerogels, and it was then attempted to check
the catalytic effect of Fe, with the two types of oxidizing
TABLE 9
The coordination of titanium in these solids could be
tetrahedral from UV-vis spectra; however, ESR suggests a
different result. A similar anisotropic signal characterized
by gxx = 2.0030, gyy = 2.0099, and gzz = 2.0238 has been re-
ported for TS-1 by Geobaldo et al. (16) and attributed to
Effect of the Polarity of the Surface on the Oxidation of Cyclo-
hexene by TBHP in the Presence of Protic or Aprotic Solvents
Final
Selectivities (mol% )
�
Initial rate
conversion
(% )
a superoxide radical O2 adsorbed on Ti species. Giamello
(mmol � min� � g� 1
1
)
TBHP
Epoxide
Sample
et al. (27) recently reported similar results for zirconium
dioxide contacted with an aqueous solution of hydrogen
peroxide; the spectrum was assigned to the formation of a
Solvent tert-butanol
P0
P10
Q10
0.1
0.69
0.6
86
87
89
91
100
100
90
82
82
superoxide radical ion adsorbed on top of a Zr4 cation.
This signal was first reported by Shvets and Kazansky (28)
on model Ti/SiO catalysts. They proposed that the g fac-
+
Solvent acetonitrile
2
zz
P0
P10
Q10
2.92
1.38
2.14
91
94
97.6
91
100
100
97
95
95
�
tor varies with the symmetry of Ti : gzz = 2.020 for O2 ad-
sorbed on a Ti occupying a square pyramid and gzz = 2.026
�
for O2 adsorbed on tetrahedral Ti atoms which exist on

4
28
KOCHKAR AND FIGUERAS
TABLE 10
Catalytic Properties, in tert-Butanol as Solvent of Silica and Ti-Free Aerogels of Silica Containing
Trace Amounts of Impurities
Oxidant
Selectivity
Product selectivities
Allylic
Fe content
(ppm)
Initial rate
Conv.
(% )
(mmol � min� � g� 1
1
)
(mol% )
Epoxide
attack
Sample
With hydrogen peroxide
No catalyst
Ti-free SiO2 aerogel
SiO2 Degussa
Fe/SiO2-1
0
147
<5
375
933
0.0024
0.032
0.014
0.018
0.025
2.2
60
4.6
6.5
47
3.0
83
17.2
26
44
26
51
22
22
56
74
49
70
78
Fe/SiO2-2
88
With TBHP
Ti-free SiO2 aerogel
SiO2 Degussa
Fe/SiO2-2
147
<5
933
0.131
0.021
0.036
69
5.8
6.3
51
30
35
85
45
46
15
55
54
the silica surface at low Ti content. By this criterion, Ti in 10 s 1 can be calculated at 333 K from the work by Corma
�
this gel is closer to square pyramidal coordination than to et al. (30). This result is of the same magnitude as the TOF
a tetrahedral coordination as found in TS-1.
(3 mmol alkene/mol Ti s) observed here for the same re-
The catalytic activities of the gels can be compared to action at 348 K. Therefore the catalytic properties of these
literature data when transformed into turnover frequen- materials are representative of titanium silicates.
cies (TOF). These comparisons are only qualitative since
It can be pointed out that the low temperature drying
the reaction conditions, in particular the solvent, can affect results in the incorporation of trace amounts of Fe, Ni, and
the reactivity. Hutter et al. (29) reported a TOF equal to Cr ions, which are known to promote the Fenton reaction.
8
3 mmol alkene/s/mol Ti for the reaction of cyclohexene The contamination of the sample has little importance if
with cumene hydroperoxide at 333 K, while at 353 K, Klein TBHP is used as the oxidizing agent, since yields in the
�
1
et al. (7) reported a TOF of 19 s for the reaction with 96% range can be reached. With hydrogen peroxide, poor
�
1
TBHP in decane. The higher TOF (200 s ) found here in performances are obtained. However, several observations
acetonitrile at 348 K agrees with the result of Hutter et al. suggest that the low selectivity is not related to the presence
and is likely due in large part to the decrease of diffusional of some impurities:
limitations, due to the procedure of drying. For the oxida-
tion of cyclohexene by hydrogen peroxide a TOF of about
(
1) the residual activity of Fe free silica from Degussa
suggests that Fe is not responsible for the decomposition of
H2O2. Indeed the activity of the Fe-doped silicas is lower by
one order of magnitude than that of the TiO2–SiO2 mixed
TABLE 11
oxides. Moreover, the main loss of selectivity is not related
to a parallel reaction of hydrogen peroxide decomposition;
the selectivity towards H2O2 which is 60% in the absence
of Ti (Table 10) usually increases up to more than 95%
with the mixed oxide, i.e. that 95% of the oxygen is found
in the products of reaction.
Effect of the Substrate Structure on the Selectivity for Epoxida-
tion of Aerogels in 1,2 Dimethoxyethane as Solvent Using Hydrogen
Peroxide (30%) as Oxidant
Hydrogen
peroxide
Olefin
selectivities (% )
(
2) Ti-HMS prepared from high purity reactants and
Initial rate
Conv. Select. C == C
(mmol � min� 1 � g� 1
treated in glass reactors shows the same catalytic behaviour
as the mixed oxides.
Olefin
)
(% )
(% )
attack Epoxide
Aerogel Q10
Cyclohexene
Norbornene
0.16
0.76
0.16
100
97
40
147
100
97
70.5
100
52
55
95
32
From the kinetics of the reaction, it appears that two par-
allel paths exist and that the main cause of selectivity loss
is the occurrence of an allylic attack with the formation
of cyclohex-2-ene-1-ol, which is much higher in the case
of H2O2 than with TBHP. This allylic attack is favoured
in the presence of air, in an aprotic solvent, or with basic
additives, and unfavoured in tert-butanol as the solvent. The
�
-Pinene
Aerogel P10
Cyclohexene
Norbornene
0.3
1.04
0.14
100
97
71
231
100
100
68.5
100
26.4
50.5
95
21
�
-Pinene

HYDROPHOBIC TITANIA–SILICA GELS
429
observation of an oxygen balance above 100% is indica- bond reaches 80% , which indicates that direct epoxidation
tive of a radical mechanism induced by hydrogen peroxide, does occur, but the main product is glycol. By contrast,
but using atmospheric oxygen. A similar result has been Fe/SiO2 gives relatively high selectivities for the attack at
recently reported by Jorda et al. (31) for the oxidation of the allylic position, which reaches 78% on Fe/SiO2-2 in tert-
cyclohexene on Ti/SiO2 obtained from TiF4. The compari- butanol as the solvent.
son of cyclohexene, �-pinene, and norbornene also agrees
The epoxide when formed can be hydrolyzed to glycol
with the occurrence of an allylic attack. This attack is un- by an acid-catalyzed reaction. It can be remarked that on
favoured with norbornene, and a very good selectivity is the Ti-free aerogel, the reaction produces some epoxide
then obtained for epoxidation; �-pinene, which contains which is not converted to glycol since the selectivity for the
more allylic hydrogens, also shows a lower selectivity for epoxide (25.6% ) is close to the selectivity for the attack of
C == C attack and epoxidation.
the double bond 26% (Table 4). The introduction of Ti in
It can be remarked from Table 6 that the selectivity for the gel increases the activity, and the selectivity for attack
C == C attack increases on hydrophobic solids (58% on P0, at the double bond reaches 79% , but the selectivity for the
6
7% on P10, and 70% on Q10) using dimethoxyethane as epoxide increases only to 30% . This means that most of the
an aprotic solvent. On hydrophilic solids such as Ti-Beta, epoxide has been converted to glycol. Therefore we can
Corma et al. (15) suggested that water competes favourably conclude that the acid sites which catalyze the hydration
with nonprotic solvents and proposed the existence of a Ti of the epoxide are due to Ti species, as been concluded for
peroxo species stabilized by water (species 1). By contrast, TS-1 (33).
in tert-butanol as solvent the stabilizing ligand would be the
As usually observed in the literature (24, 34–36) this re-
alcohol. The lower donor properties of water with respect action can be inhibited by basic additives (Table 5). The se-
to alcohols would then account for a higher electrophilic lectivity for the epoxide is almost doubled by the addition
character of the hydrated species 1 and then of a higher of isoquinoleine to the reaction mixture. It is interesting to
reactivity.
notice that this addition decreases the activity, but it also
reduces the selectivity of the attack at the double bond.
This is in agreement with the effect of solvents and the
proposal that epoxidation catalysts are fairly strong Lewis
acids.
With tert-butanol as solvent the polarity of the surface
has a small influence on the activity (Table 6), but it can
be noticed that both reactions with hydrogen or organic
peroxides are faster on the solid prepared with phenyltri-
Species 1
This intermediate accounts for a higher conversion in ethoxysilane. A lower activity for hydration of the epox-
aprotic solvents, but also for the different selectivity for the ide is observed on an hydrophobic surface, which is indeed
allylic attack. If we measure the importance of the radical consistent with a lower concentration of water at the in-
reaction by the balance of oxygen in Table 8, we can con- terface. However, hydrophobicity has only a limited effect
clude that the contribution of this radical process is higher on the selectivity of epoxidation versus allylic attack. When
on hydrophobic solids, when dimethoxyethane is used as compared to TS-1 these mixed oxides appear then to show
the solvent. In acetonitrile the H2O2 efficiency is 98% , but rather different properties that are not simply related to hy-
here also allylic oxidation is observed with a selectivity of drophobicity, as proposed by others (7). The different be-
4
5% . This shows that the surface Ti complex decomposes haviour is more likely related to the different coordination
faster to a radical. An hydrophobic surface should adsorb of Ti. The different selectivities observed in the epoxida-
preferentially ether or acetonitrile (Lewis bases) to water. tions of norbornene and �-pinene with hydrogen peroxide
Then a complex could be formed in which the Lewis base confirm that the specific reactivity of these solids is related
could at least partially neutralize the charge on titanium to their ability to decompose TiOOH species into radicals.
and then reduce the activity for epoxidation and promote The possible applications are then limited to the epoxida-
a radical cleavage and allylic oxidation, as indeed observed tion with organic peroxides, or to those substrates which
in Table 8 by the higher importance of radical reactions cannot suffer allylic attack with H2O2.
when pyridine is added to the solid. As reported before by
In conclusion, if an organic hydroperoxide is used as the
Sheldon (26) and by Fusi (32) this radical oxidation yields oxidizing agent, these mixed gels are good catalysts, reach-
hydroperoxides which can be reacted with excess cyclohex- ing >95% yield for cyclohexene oxide. With hydrogen per-
ene and give epoxy-cyclohexane. The selectivity relative to oxide a selectivity of about 80% for the attack of the double
cyclohexene expected for this process is at most 50% .
bond can be reached with cyclohexene, by optimizing the
In tert-butanol as the solvent and using H2O2 as the ox- solvent, but the hydration of the resulting epoxide is fast
idant (Table 7) the selectivity for the attack at the double and can be slightly decreased but not suppressed, even on

4
30
KOCHKAR AND FIGUERAS
hydrophobic surfaces. This is consistent with the idea that 17. Boccuti, M. R., Rao, K. M., Zecchina, A., Leofanti, G., and Petrini,
G., Stud. Surf. Sci. Catal. 48, 133 (1989).
the site active for epoxidation is a strong Lewis site; the
selective inhibition of the secondary reaction is then practi-
cally impossible.
1
8. Aizawa, M., Nosaker, Y., and Fujii, N., J. Non. Crys. Solids 128, 77
1991).
(
1
9. Keshavaraja, A., Ramaswamy, V., Soni, H. S., Ramaswamy, A. V., and
Ratnasamy, P., J. Catal. 157, 501 (1995).
REFERENCES
20. Neumann, R., Chava, M., and Levin, M., J. Chem. Soc. Chem. Com-
mun., 1685 (1993).
1
2
. Sheldon, R. A., and Dakka, J., Catal. Today 19, 215 (1994).
. Tatsumi, T., Nakamura, M., Yuasa, K., and Tominaga, H., Catal. Lett.
21. Deo, G., Turek, A. M., Wachs, I. E., Huybrechts, D. R. C., and Jacobs,
P. A., Zeolites 13, 365 (1993).
1
0, 259 (1991).
22. Liu, Z., and Davis, R. J., J. Phys. Chem. 98, 1253 (1994).
23. Wu, P., Komatsu, T., and Yashima, T., J. Phys. Chem. 100, 10316
(1996).
24. Bartok, M., and Lang, K. L., in “Small Ring Heterocycles” (A.
Hassner, Ed.), Part 3, p. 71. Wiley, New York, 1985.
25. Tanev, P. T., Chibwe, M., and Pinnavia, T. J., Nature 368, 321 (1994).
26. Sheldon, R. A., and van Doorn, J. A., J. Catal. 31, 427 (1973).
27. Giamello, E., Rumori, P., Geobaldo, F., Fubini, B., and Paganini, M.
C., Appl. Magn. Reson. 10, 173 (1996).
28. Shvets, V. A., and Kazanski, V. B., J. Catal. 25, 123 (1970).
29. Hutter, R., Mallat, T., and Baiker, A., J. Catal. 153, 177 (1995).
30. Corma, A., Camblor, M. A., Esteve, P., Martinez, A., and Perez-
Pariente, J., J. Catal. 145, 151 (1994).
3
4
. Wulff, H. P., Shell Oil Co., U.S. Pat. 3,923,843 (1975).
. Fraile, J. M., Garc ´ı a, J. I., Mayoral, J. A., de M e` norval, L. C., and
Rachdi, F., J. Chem. Soc., Chem. Commun., 539 (1995).
. Maschmeyer, T., and Thomas, J. M., in “Proc. 11th Int. Cong. Catal.,
Baltimore, June 30–July 5, 1996.” [Poster Po-99]
5
6
. Jorda, E., Tuel, A., Teissier, R., and Kervenal, J., J. Chem. Soc. Chem.
Commun., 1775 (1995).
7
8
. Klein, S., Thorimbert, S., and Maier, W. F., J. Catal. 163, 476 (1996).
. Blasco, T., Camblor, M. A., Corma, A., Esteve, P., Martinez, A., Prieto,
C., and Valencia, S., J. Chem. Soc. Chem. Commun., 2367 (1996).
. Toba, M., Mizukami, F., Niwa, S-I., Sano, T., Maeda, K., Annila, A.,
and Komppa, V., J. Mol. Catal. 91, 277 (1994).
9
1
1
0. Brown, J. F., J. Polymer Science, Part C1 83, 1 (1963).
1. Guyot, A., and Bartholin, M., Compt. Rend. Acad. Sci. 264, 1232
31. Jorda, E., personal communication.
32. Fusi, A., Ugo, R., and Zanderighi, G. M., J. Catal. 34, 175 (1974).
33. Auroux, A., Gervasini, A., Jorda, E., and Tuel, A., in “Zeolites and
Related Microporous Materials: State of the Art 1994” (J. Weitkamp,
H. G. Karge, H. Pfeifer, and W. H o¨ lderich, Eds.), Garmisch-
Partenkirschen 1994, Stud. Surf. Sci. Catal. 84, 653 (1994).
34. Bellussi, G., Carati, A., Clerici, M. G., and Maddinelli, G., J. Catal. 133,
220 (1992).
(
1967).
1
1
2. Danforth, J. D., and Roberts, J. H., J. Catal. 10, 252 (1968).
3. Klein, S., and Maier, W. F., Angew. Chem. Int. Ed. Engl. 35, 2230
(
1996).
4. Dutoit, D. C. M., Schneider, M., and Baiker, A., J. Catal. 153, 165
1995).
1
(
1
1
5. Corma, A., Esteve, P., and Martinez, A., J. Catal. 161, 11 (1996).
6. Geobaldo, F., Bordiga, S., Zecchina, A., and Giamello, E., Catal. Lett.
35. Clerici, M. G., and Ingallina, P., J. Catal. 140, 71 (1993).
36. Sato, T., Dakka, J., and Sheldon, R. A., “Zeolites Related Microporous
Materials,” Stud. Surf. Sci. Catal. 84, 1853 (1994).
1
6, 109 (1992).