Applied Catalysis A: General 300 (2006) 91–99

www.elsevier.com/locate/apcata

Comparison of liquid-phase olefin epoxidation over MoOx inserted within

mesoporous silica (MCM-41, SBA-15) and grafted onto silica
Paul Célestin Bakala, Emmanuel Briot *, Laurent Salles, Jean-Marie Brégeault
Systèmes Interfaciaux à l’Echelle Nanométrique, Université Pierre et Marie Curie, CNRS-UMR 7142,
Case 196, 4 place Jussieu, F-75252 Paris Cedex 05, France
Received 26 July 2005; received in revised form 13 September 2005; accepted 22 September 2005
Available online 19 December 2005

Abstract
Molybdenum(VI)-containing siliceous-MCM-41 and -SBA-15 mesoporous molecular sieves and silica-supported molybdenum(VI) complexes
have been prepared according to an original pathway by the reaction between low-cost Mo(VI) peroxo species and silica precursors
(tetraethylorthosilicate), silica beads (pure SiO2) or precipitated silica in an acidic and peroxidic medium (peroxo route). Hydrogen peroxide
is added to avoid the formation of iso-(or hetero-)polyoxometalates in the direct one-pot synthesis of mesoporous solids, or in the case of pure SiO2.
The resulting calcined materials have been characterized by chemical analysis, X-ray powder diffraction, nitrogen sorption isotherms, TEM and
EDX analysis, UV–vis diffuse reflectance spectroscopy, Raman spectrometry and catalysis tests. The peroxo routes lead to inserted and/or
supported molybdenum oxide zones which are restricted in size due to the porous structure of the silica-based materials and to the involment of
low-condensed and even mononuclear oxoperoxo species. This allows to support MoOx species in a dispersion which cannot be achieved with the
systems using heptamolybdate [Mo7O24]6
salts or heteropolyoxometalates, which favor the anchoring of MoO3 clusters. Catalysis tests show that
the very dispersive structure of the MoOx-siliceous-MCM-41, -SBA-15 and MoOx/SiO2 promotes the generation of fairly active oxidation catalysts
for liquid-phase olefin epoxidation (cyclooctene, R-(+)-limonene from ambient to 40 8C) using anhydrous tert-butylhydroperoxide (TBHP) as
oxidant. Under specific conditions, leaching of redox-active species has been minimized (less than 2 ppm) especially with MoOx/SiO2 which is
proposed as the simplest catalyst for these oxidations. MCM-41 and SBA-15 do not show clear advantages over silica in epoxidation reactions with
TBHP.
# 2005 Elsevier B.V. All rights reserved.

Keywords: Mesoporous solids; Silica; Oxoperoxo species; Epoxidation; Hydrogen peroxide

1. Introduction achieved with other synthesis routes, inducing crystallized

In the field of catalytic oxidations, great efforts have been
made to convert homogeneous to selective heterogeneous
processes, including the use of clean and safe oxidants such as
O2, H2O2 and alkylhydroperoxides [1,2].
Since the early 1990s and the advent of mesoporous silica
[3–7], the development of new solid catalysts and the
heterogenization of homogeneous systems for oxidation
processes have become a very attractive topic. Indeed, ‘‘chimie
douce’’ syntheses of mesoporous silica lead to amorphous
materials, allowing the most elegant pathway for the
incorporation of transition metal ions that could not be
* Corresponding author.
E-mail address: briot@ccr.jussieu.fr (E. Briot).
0926-860X/$ – see front matter # 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2005.09.038
products. The lack of industrial development of catalytic
processes involving redox mesoporous silica can be explained
as follows: (i) the presence of expensive inorganic complexes
such as metal alkoxides or quaternary ammonium ions (organic
template), (ii) lengthy hydrothermal synthesis, (iii) template
removal, implying long thermal treatments and very often the
irreversible loss of the organic template, although mild routes
would allow the reuse of this material, (iv) oxidations in the
liquid phase with organic hydroperoxides or even H2O2 are very
often subject to leaching phenomena, and the question of the
true nature of the catalytic reaction (homogeneous or
heterogeneous) is a serious one, (v) poor attrition strength
for large-scale reactor and then large pressure drop when placed
in a fixed-bed reactor. It is worth considering the catalytic
properties of redox mesoporous materials in comparison with
standard amorphous silica or alumina with supported active
92 P. Célestin Bakala et al. / Applied Catalysis A: General 300 (2006) 91–99
sites, the latter being synthesized with good control and very
cheaply on the industrial scale.
Among transition metal ions incorporated into meso- or
microporous silica, molybdenum is one of the most studied (with
Ti), due to its interesting catalytic properties in many reactions.
On the one hand, molybdenum is used in the petroleum chemistry
(hydrocracking, hydrodesulfurization, etc.), which is the subject
most studied. On the other hand, it has been incorporated into
mesoporous molecular sieves (M41S, SBA-n, silicalites, etc.) by
wetness, dry impregnation or thermal spreading [8–11], grafting
[12–20] or insertion [21–31]. These materials have been tested
for several reactions such as 2-methylfuran condensation with
acetone [8], polymerization of methylcyclopentadiene [11],
peroxidative bromination of phenol red [13], oxidation of the 2,6-
di-tert-butyl phenol [15], methanol [17] and methane [21]
oxidations to acetaldehyde, decomposition of NOx [24],
oxidative dehydrogenation of ethanol [28], cyclohexane and
cyclohexanol oxidation with H2O2 [29], propylene oxidation
[31] and olefin (mainly cyclooctene) epoxidation with H2O2 [30]
or tert-butyl hydroperoxide (TBHP) [12,14,18–20,22,25,27,30].
However, the choice of cyclooctene, which is not only about the
easiest common olefin to epoxidize, but also can form epoxides
from dioxygen and other oxidants which do not usually transfer
single oxygen atoms, clearly indicates that the investigation of
metal-containing mesoporous materials is still at the prospecting
stage. Moreover, leaching of active species in the presence of
nucleophilic agents (TBHP, H2O2, H2O, etc.) has been, up to now,
poorly investigated if at all. Among epoxidation tests, few
authors [12,18–20,25,27,30] have rigorously studied the ques-
tion of leaching and the homogeneous and/or the heterogeneous
nature of the catalytic process, as recently recommended by
Sheldon et al. [32,33]. Note that even the famous titanium
silicalite (TS-1) releases Ti(IV) in liquid-phase processes [34].
In the present paper, we report the direct (one-pot)
synthesis of MCM-41 and SBA-15 mesoporous silica
chemically modified by insertion of molybdenum, using
the S+X
I+ pathway in an acidic and peroxidic medium, in the
so-called peroxo route (S+ = CTMA+, cetyltrimethylammo-
nium, X
= Cl
and I+ represents the protonated silanol
species) [25,26]. All materials were characterized by
chemical analysis, X-ray powder diffraction, UV–vis diffuse
reflectance spectroscopy, Raman spectrometry and nitrogen
sorption. The catalytic activity of the calcined mesoporous
solids (Mo-MCM-41, Mo-SBA-15) was tested for the
epoxidation of olefins with TBHP and compared with the
activity of previously calcined amorphous silica chemically
modified by wet impregnation by Mo(VI) peroxo species
(Mo–SiO2) to avoid the formation of [SiMo12O40]4
. It is well
known that molybdate uptake by SiO2 is relatively low over
the entire pH range, except for a small increase at pH < 2
related to the formation of [SiMo12O40]4
ions that can
partially desorb into aqueous solutions [35]. It should be
noted that in many cases, catalytic activity has been only
compared with that of zeolites and mesoporous silica;
comparisons with amorphous silica are generally ignored!
Particular attention was paid to the study of the leaching of
active species and catalyst regeneration.
2. Experimental
2.1. Materials
For the mesoporous solid syntheses, tetraethylorthosilicate
(TEOS) (purum, 99%), cetyltrimethylammonium chloride
(CTMACl) (25 wt.% in aqueous solution) and MoO3 (purum,
99.5%) were purchased from Fluka. Pluronic P123, poly(-
ethylene glycol)-block-poly(propylene glycol)-block-poly(-
ethylene glycol), average Mn ca. 5800, ca. 30 wt.% ethylene
glycol (triblock copolymer, EO20PO70EO20) was purchased
from Aldrich. 30 wt.% aqueous H2O2 (stabilizer: stannate) was
a commercial product (Prolabo). A 37 wt.% HCl was purchased
from Carlo Erba. Pure native silica bead samples were provided
by Procatalyse and Rhodia (263 m2 g
1, 1.52 cm3 g
1) and
precipitated silica was obtained from a sodium silicate solution;
the solid was filtered, washed with distilled water and dried
over P4O10 (143 m2 g
1, 0.84 cm3 g
1).
For the catalysis tests, cyclooctene (95%) and decane
(>98%) were purchased from Fluka, hexane (99%), undecane
(99%), R-(+)-limonene (97%) and TBHP (ca. 6 M in decane)
from Aldrich. The TBHP was dehydrated with anhydrous
MgSO4 (24 h) and stored at 4 8C over 4 Å molecular sieve.
Acetonitrile was used as received (SDS).
2.2. Syntheses
The syntheses (peroxo route) of redox mesoporous materials
were adapted from the procedures recently described for W- or
Mo-MCM-41 [23,25,36] and Mo-SBA-15 [26].
In a typical synthesis of Mo-MCM-41 [(Si/Mo)initial (mol/
mol) = 50], MoO3 (0.242 g) was first reacted with 30 wt.%
aqueous hydrogen peroxide (1.7 mL) and H2O (1.7 mL) to
generate a yellow solution of soluble low-condensed peroxo
species (solution A), mainly [MoO(O2)2(H2O)2] and
[O{MoO(O2)2(H2O)2}]2
, after 1 h stirring at 60 8C. In
solution B, TEOS (18.5 mL) was mixed with a solution
containing the templating agent CTMACl (13.1 mL) and HCl
(61.5 mL) in H2O (145 mL). After a few minutes and the
beginning of mesoporous silica formation, solution A was
added to solution B. The synthesis gel composition was: 1SiO2/
0.12CTMACl/8.9HCl/0.2H2O2/0.02MoO3/136H2O. The reac-
tion mixture was aged for 2 h at room temperature with
moderate stirring. The pale yellow precipitate was then suction-
filtered, washed with distilled water and dried over P4O10 at
room temperature. The as-synthesized materials were calcined
in air (180 cm3 min
1, 1 K min
1) from ambient to 820 K
(samples maintained at the final temperature for 6 h) to remove
the template and to favor the anchoring of molybdenum species
in the mesoporous silica.
The synthesis of Mo-SBA-15 [(Si/Mo)initial (mol/mol) = 50],
similar to Mo-MCM-41, was performed as follows—solution
A: 2.5 mL H2O, 2.5 mL H2O2, 0.118 g MoO3; solution B:
9.1 mL TEOS, 137 mL H2O, 20 mL HCl, 1.8 mL CTMACl,
4 g Pluronic P123. The synthesis gel composition was then:
1SiO2/0.017P123/0.033CTMACl/5.9HCl/0.6H2O2/0.02MoO3/
210H2O. The reaction mixture was aged for 4 h at room
 P. Célestin Bakala et al. / Applied Catalysis A: General 300 (2006) 91–99
temperature with moderate stirring, the rest of the synthesis
being identical to that for Mo-MCM-41.
Before impregnation, SiO2 (beads or precipitated silica)
were calcined in air (180 cm3 min
1, 1 K min
1) from
ambient to 820 K (samples maintained at the final tem-
perature for 6 h). Wet impregnation was performed by
suspending 0.95 g of the silica in a peroxidic solution of
Mo(VI) initially containing 0.06 g MoO3, 3 mL H2O and
0.5 mL H2O2 to generate the peroxo complexes, the pH of the
solution being below the isoelectric point of silica (i.e. 1.5).
The suspension was stirred for 4 h at room temperature. The
yellow solid was filtered off on a fritted glass disk, then as
such quickly dried over P4O10 (without addition of alk-
ylammonium or phosphonium as counter-ions; the [BBSi–O–
Mo(O2)]
surface species are decomposed with H2O2 and
water in excess). With the assumption of 2–3 OH/nm2 on
SiO2 [37], the final material can contain, at best, 0.25–1 Mo
per OH [23]. Molybdenum-containing beads and precipitated
silica are denoted Mo–SiO2-bead and Mo–SiO2-prec,
respectively.
2.3. Apparatus
X-Ray diffraction patterns were recorded on a Bruker D8
diffractometer (Cu Ka radiation) with 0.028 (2u) step size and
6 s step time in the range 0.3 < 2u < 108 for Mo-SBA-15, and
2 < 2u < 108 for Mo-MCM-41. Adsorption and desorption
isotherms for nitrogen were obtained at 77 K using a
Micromeritics ASAP 2010. The samples were outgassed at
393 K and 0.3 Pa for 4 h before measurements. UV–vis diffuse
reflectance spectroscopy was performed on a Varian Cary 5E
spectrophotometer equipped with an integration sphere.
Polytetrafluoroethylene was used as reference. Raman spectra
were recorded on a Kaiser Hololab 5000R (excitation line,
785 nm; power source, 5–50 mW, 100 scans, 10 s per scan,
0.3 cm
1 resolution). Elemental analyses were carried out at
the Service Central d’Analyse (CNRS-Lyon) by inductive
coupling plasma atomic emission spectroscopy (ICP-AES)
after alkaline fusion with Li2B4O7.
2.4. Catalysis tests
In a typical test, with stirring at 40 8C, 100 mg of the catalyst
were placed in a round-bottomed flask or in a home-made
reactor (avoiding direct contact between the catalyst and the
magnetic stirring bar) with 1.5 mL of anhydrous TBHP (ca.
6 M in decane), ca. 3 mmol of substrate (i.e. cyclooctene or
limonene), internal standard (undecane) and solvent to obtain a
total volume of ca. 10 mL to give the following ratios (mol/
mol): substrate/Mo in the 60–160 range and TBHP/substrate ca.
2. TBHP consumption was evaluated by iodometric titration.
The progress of the reaction was monitored by GC (Delsi 30 gas
chromatograph equipped with a 0.25 mm diameter, 50 m OV-
1701 capillary column and a flame ionization detector linked to
a Delsi Enica 10 electronic integrator). By-products were
analysed with a Thermoquest Trace MS. All the catalysis tests
were performed at least twice.
93
3. Results and discussion
3.1. Physicochemical characterizations
3.1.1. X-ray analysis
The X-ray diffraction patterns of Mo-MCM-41 and Mo-
SBA-15 (Fig. 1) display one broad peak at very low angle
(2u = 2.88 for Mo-MCM-41 and 2u = 1.28 for Mo-SBA-15)
which can be assigned to the (1 0 0) reflection [5,7,26],
indicating that the nearly amorphous materials are character-
ized by a short-range order. Although it is known that the
hydrothermal treatment enhances the maturation of the solid,
other diffraction peaks (i.e. (1 1 0), (2 0 0)) are present on
Mo-MCM-41 diffractograms. This indicates that the room
temperature synthesis is responsible for the better crystal-
lographic response for Mo-MCM-41, whereas Mo-SBA-15 is
not so well organized. However, the presence of peroxo
species prohibits the use of hydrothermal treatments during
the early stage of the synthesis, this step being the key to
obtaining a good dispersion of Mo(VI) in the silica matrix.
Based on hexagonal symmetry, the unit cell parameter pffiffiffia0 can
be calculated using the formula: a0 ¼ 2d1 0 0 = 3 (see
Table 1).
3.1.2. Comments on the molybdenum content
The experimental (Si/Mo) molar ratios (Table 1) match
approximately the initial composition in the gel for Mo-MCM-
41 (31 for 50) and for SBA-15 (59 for 50), indicating that the
acidic/peroxo route is a reliable method for monitoring
molybdenum insertion within the mesoporous materials. For
Mo–SiO2-bead, only part of the molybdenum initially
introduced during the synthesis ((Si/Mo)th = 50) can be grafted
onto the solid ((Si/Mo)exp = 84). The pore size profile and the
surface area of the silica beads were expected to yield a lower
Mo content by grafting the peroxo complexes, compared to
mesoporous materials. Moreover, with silica beads, diffusion
hindrance clearly appeared during the impregnation procedure,
since the beads were not uniformly yellow colored. For Mo–
SiO2-prec, no diffusion hindrance was encountered, even when
the Si/Mo molar ratio was as low as 25. To ascertain whether the
molybdenum dispersion was high, we performed additional
energy dispersive X-ray (EDX) analyses of the calcined
Fig. 1. X-ray diffraction patterns of calcined samples: (a) Mo-MCM-41 and (b)
Mo-SBA-15.
94 P. Célestin Bakala et al. / Applied Catalysis A: General 300 (2006) 91–99

Table 1
Textural properties of Mo-MCM-41, Mo-SBA-15 and Mo–SiO2
Mo (Si/Mo)exp a0 Vta Vmeso Vmicro Smeso Smicro Sext dp(4V/S) wb
(wt.%) (mol/mol) (Å) (cm3 g
1) (cm3 g
1) (cm3 g
1) (m2 g
1) (m2 g
1) (m2 g
1) (Å) (Å)
Mo-MCM-41 4.21 31 37 0.66 0.64 0.00 1058 0 14 24 13
Mo-SBA-15 2.35 59 85 0.83 0.72 0.09 792 266 16 36 49
Mo–SiO2-bead 1.80 84 – 1.21 – – 220c 220d –
Mo–SiO2-prec 5.5 25 – 0.65 – – 103c 252d –
a
Vt is determined at P/P0 ca. 0.97.
b
w ¼ a0
dp ð4Vmeso =Smeso Þ.
c
Calculated with the BET equation.
d
Calculated with 4V/S, V being the total volume (interpreted as intergranular porosity).

samples. These results clearly indicate a ‘‘quasi-homogeneous
distribution’’ of the Mo sites [26].
3.1.3. Nitrogen sorption at 77 K and molecular
accessibility
The evaluation of specific area, mean pore size diameter and
especially pore size distribution of mesoporous materials are still
subject of controversy [38]. In our case, Mo-MCM-41 samples
exhibit reversible, hysteresis-free type IVisotherms, according to
IUPAC nomenclature, whereas Mo-SBA-15 displays a type IV
isotherm with a hysteresis loop near P/P0 = 0.5, the Mo–SiO2
(bead and prec) isotherms being characteristic of materials with
wide mesopores and low-specific surface area compared to
mesoporous solids (Fig. 2) [39]. It is known that the BET model is
somewhat inaccurate for evaluating the specific surface areas
of mesoporous materials, leading to over-estimation [40] in
some cases and only allowing the total specific surface area
estimation. The as-plot method, based on the comparison of
the adsorption isotherms of the studied sample and of a ref-
erence sample (LiChrospher Si-1000 silica, SBET = 25 m2 g
1,
V = 9.12 cm3 g
1 STP), which is not porous and has the same
chemical composition (i.e. SiO2), allows a detailed evaluation of
mesopore, micropore and external specific surface areas, and
mesopore and micropore volumes [39]. Recently, Jaroniec et al.
[41] gave the standard reduced nitrogen adsorption isotherm data
for LiChrospher Si-1000 silica, with relative pressures ranging
from 5.5  10
7 to 0.998, which constitutes an improvement on
the data published by Gregg and Sing [39] and Aristov et al. [42],
especially at very low-relative pressure. A fit of these data in three
pressure ranges allows a rapid conversion of adsorption data (Vads
STP, P/P0) to reduced data (Vads STP, as). The as-plots of Mo-
MCM-41 and Mo-SBA-15 are presented on Fig. 3; area and
volume data are displayed in Table 1. The mean pore size
diameter (dp) has been calculated as 4V/S (V is the mesopore
volume (cm3 g
1) and S is the mesopore specific area (cm2 g
1),
both calculated by the as-plot method). Data reported in Table 1
indicate that Mo-MCM-41 and Mo-SBA-15 are mainly
mesoporous, and that SBA-15 displays some microporosity
with a significant contribution of the micropore area to the total
surface area. Mean pore diameters are very similar for Mo-
containing mesoporous silica, but the amorphous walls are
thicker for Mo-SBA-15. As expected, either BET or as-plot
models can be applied to the Mo–SiO2 (bead and prec) isotherms
to obtain the specific surface area (see Table 1). The pore
diameter of this material is calculated, assuming that the total

Fig. 2. Nitrogen sorption isotherms obtained at 77 K: (a) Mo-MCM-41, (b) Mo-

SBA-15 and (c) Mo–SiO2-bead. Fig. 3. as-Plot of (a) Mo-MCM-41 and (b) Mo-SBA-15.
 P. Célestin Bakala et al. / Applied Catalysis A: General 300 (2006) 91–99 95

pore volume corresponds to the volume of intergranular pores. groups may contribute to reduce leaching compared to MoO3
For Mo–SiO2-bead and Mo–SiO2-prec, the decrease in pore deposited or anchored on silica.
volume and specific surface area is consistent with the grafting of
Mo species on the silica surface at the molecular level, without 3.2. Catalysis tests
formation of clusters.
Tests performed with commercial TBHP (5–6 mol L
1 in
3.1.4. UV–vis diffuse reflectance spectroscopy decane, ca. 4 wt.% H2O) did not give reproducible results, the
The UV–vis spectra of Mo-containing silicas (Mo-MCM- cyclooctene conversion fluctuating between 40 and 70% (after
41, Mo-SBA-15 and Mo–SiO2) are similar and display broad 24 h) with Mo-MCM-41 in hexane. Anhydrous TBHP is
absorption bands of high-energy maxima (Fig. 4). The therefore recommended for the catalysis tests.
absorption edge is clearly shifted towards higher wavelengths The results of cyclooctene epoxidation over Mo-containing
for MoO3, indicating the presence of oligomeric species MCM-41, SBA-15 and SiO2 are listed in Table 2. Blank
[43,44]. These results are also consistent with the presence of experiments, performed without any catalyst or with purely
low-nuclearity molybdenum oxo species for most of the siliceous supports, show a very low-cyclooctene conversion,
dispersed Mo(VI), inserted into the mesoporous silica matrix even after 4 h reaction at 40 8C (<5% in acetonitrile, 13% in
(Mo-MCM-41 and Mo-SBA-15) or grafted (Mo–SiO2-bead and hexane and 15% in decane). In contrast, Mo-containing MCM-
Mo–SiO2-prec) on silica. 41, SBA-15 and SiO2 catalysts are active at 40 8C. Preliminary
catalysis tests, performed with Mo-MCM-41 in hexane as
3.1.5. Raman spectroscopic studies solvent, showed that 3 h and 8 mL of solvent are optimal
UV–vis results correlate with data obtained by Raman conditions. After 3 h, leaching of active species seems to be
spectrometry (Fig. 5). Mo-MCM-41, Mo-SBA-15 and Mo– more pronounced (solution is pale yellow after removal of the
SiO2 (bead or prec) display similar Raman spectra: bands at catalyst, due to Mo-peroxo species). Similarly, with 5.5 mL of
506 and 615 cm
1 (vibration of cyclic trisiloxanes) are solvent, conversion reaches 91% and the solution is yellow.
assigned to the SiO2 matrix [36,45] as well as the shoulder at However, it should be noticed that the previous catalysis tests
920 cm
1 (ñSi
O), and the Raman signal at 955 cm
1 is were performed in a round-bottomed flask and contact between
attributed to the symmetric stretching mode of the terminal the solid and the magnetic stirring bar enhances the attrition of
Mo O bond, with a possible contribution of Si–O stretching
of SiOH  OMo ‘‘defective sites’’. Comparison of the spectra
of Mo-containing silica and a mixture of MoO3/SiO2 clearly
indicates that crystalline molybdenum oxide (MoO3) is not
formed in the calcined product via the peroxo route.
Moreover, characteristic bands of [SiMo12O40]4
(i.e. three
maxima at 990, 970 and 910 cm
1, a broad band at 635 cm
1,
and peaks at 252 and 220 cm
1) [46,47] are missing,
indicating that the peroxo route avoids the formation of
polyoxometalates [23]. So far, spectroscopic data are
consistent with the presence of low-nuclearity MoOx species
in mesoporous silica or on silica; these species can coordinate
–O–O–t-Bu by reaction with anhydrous t-BuOOH. High-
incorporation levels with a nearly homogeneous distribution
of the dopants are obtained, avoiding the formation of isolated
MoO3 clusters. Anchoring of the molybdenum(VI) by BBSi–O

Fig. 4. UV–vis diffuse reflectance spectra of calcined samples: (a) Mo-MCM- Fig. 5. Raman spectra of precatalysts: (a) Mo-MCM-41, (b) Mo-SBA-15, (c)
41, (b) Mo-SBA-15, (c) Mo–SiO2, (d) MoO3 and (e) SiO2. Mo–SiO2-bead, (d) MoO3/SiO2, (e) H4[SiMo12O40] and (f) Mo–SiO2-prec.
96 P. Célestin Bakala et al. / Applied Catalysis A: General 300 (2006) 91–99

Table 2
Cyclooctene epoxidation over Mo-catalysts, after 3 h reaction at 40 8C
Mo-MCM-41 (68)a Mo-SBA-15 (122)a Mo–SiO2-bead (160) a
Acetonitrile Hexane Decane Acetonitrile Hexane Decane Acetonitrile Hexane Decane
Conversion (%) 66 90 97 85 95 >99 51 68 69
Selectivity (%) >99 97 95 93 98 93 50 94 87
(Si/Mo)expb (mol/mol) 38 42 35 134 236 65 420 98 89
Moc (ppm) 95 155 60 180 250 30 185 35 14
Mo loss (mol%) 18 26 10 56 75 9 80 14 6
In all cases, the TBHP selectivity is above 90%.
a
Cyclooctene/Mo (mol/mol).
b
Si/Mo molar ratio after catalysis test.
c
Mo wt.% in the catalytic solution after the first catalysis run.

the solid, leading to the leaching of both Mo and Si, i.e. the
erosion of the solid support. All catalysis tests were then
performed in a reactor avoiding direct contact between the
catalyst and the magnetic stirring bar. The evolution of the
cyclooctene conversion as a function of time, presented in
Fig. 6, shows that cyclooctene oxide is produced very quickly,
whatever the solvent (hexane or decane) or the catalyst (Mo-
MCM-41 or Mo-SBA-15), resulting in the high-epoxidation
activity of Mo catalytic species.
Cyclooctene conversion increases in the order: acetonitrile
(polar) < hexane < decane (apolar). This result can be
explained by considering the deleterious effect of the presence
of adventitious water in the solvents. The more polar the
solvent, the more water is soluble. Acetonitrile contains more
water than decane, and water must be removed to make all these
catalysts efficient.
Comparison of Mo-MCM-41, Mo-SBA-15 and Mo–SiO2-
bead shows that Mo-SBA-15 is one of the most effective
catalysts, with the lowest mesopore specific surface area.
Moreover, Mo-SBA-15 has fewer active sites per gram than Mo-
MCM-41 and Mo–SiO2-bead. Mo species can be homoge-
neously inserted into the amorphous walls of mesoporous
materials (vide supra), therefore the thick walls of Mo-SBA-15
(49 Å) contain more inaccessible Mo sites than the thin walls
(13 Å) of Mo-MCM-41, whereas with Mo–SiO2-bead where
molybdenum is only dispersed on the surface, there is a clear
evidence of diffusion problems. The size of the beads is
important: the use of smaller particles with the same
molybdenum content leads to a significant increase in the yield
of epoxide, and grinding the silica beads accelerates the reaction
(vide infra). It should already be mentioned that Mo–SiO2-bead,
although less active than Mo-containing mesoporous materials,
displays an interesting activity towards cyclooctene epoxidation
with decane, its preparation being easier and less expensive than
that of the corresponding mesoporous ones.
The heterogeneous nature of the catalytic process is
demonstrated by the steady-state conversion observed 24 h
after catalyst removal. Filtration was performed at the tem-
perature of the experiment. Molybdenum species, possibly
present in the solution, due to leaching of the solid catalyst, are
ineffective for epoxidation and the solid can be recycled after
standard thermal treatments.
Questions about mechanical stability, regeneration of the
catalyst and leaching of active species are of prime importance
for a possible development of these materials. As reported in
Table 2, decane is the most valuable solvent since the wt.% of Mo
loss in the solution is quite low, compared to hexane or
acetonitrile. The Mo loss corresponds to 10%, 9% and 6% of the
initial Mo-loading for Mo-MCM-41, Mo-SBA-15 and Mo–SiO2-
bead, respectively. These results, obtained for a first run, are very
promising, and the loss of Mo becomes negligible after the
second run for the three catalysts (see Table 3). Indeed, all results
indicate that the loss of Mo takes place mainly during the first run,
while cyclooctene conversion is approximately constant,

Fig. 7. Cyclooctene epoxidation conversion after 3 h reaction at 40 8C in

Fig. 6. Cyclooctene conversion as a function of time, at 40 8C: (solid line) decane, as a function of calcination temperature and the delay between
hexane, (dashed line) decane, (*) SBA-15 and () MCM-41. synthesis and calcination, and between calcination and catalysis test.
 P. Célestin Bakala et al. / Applied Catalysis A: General 300 (2006) 91–99
Table 3
Regeneration of Mo-catalysts for cyclooctene epoxidation with TBHP in 10 mL hexane or decane (3 h at 40 8C)
Run Mo-MCM-41
Decane Hexane
Conversion Mo (Si/Mo) Conversion
(%) (ppm) (mol/mol) (%)
1 >99 65 – 98
2 >99 14 – 97
3 90 <2 67 94
indicating that Mo species lost during the first run are probably
not highly active for epoxidation. Then, a steady-state conversion
and leaching (i.e. very low wt.% Mo in solution and (Si/Mo)exp
unchanged) are obtained for the three catalysts after the third run.
The catalytic activity of Mo-MCM-41 towards cyclooctene
epoxidation has been studied by varying the calcination
temperatures and the delays between syntheses and calcina-
tions, and between calcinations and catalysis tests. Some of
these results are presented in Fig. 7. Obviously, the calcinations
must be done at least at 550 8C, to remove the major part of the
template from the mesopores, and to allow the substrate to
reach the active sites located inside the channels of MCM-41 or
on the external surface. Concerning the stability of Mo-MCM-
41, the results of Fig. 7 suggest that both as-synthesized and
calcined samples can easily be kept in a simple desiccator
(containing P4O10). Indeed, all catalysts are effective; this can
be correlated with the UV–vis spectra of freshly prepared and
old native and/or calcined samples (Fig. 8) and Raman spectra
(not shown), which do not indicate any drastic change in the
environment of molybdenum species.
The calcination program, which is presumed to form and
stabilize the active molybdenum species linked to the silica
surface, is also of prime importance as regards the leaching of
Mo(VI) into the solution. For example, calcination of Mo-
MCM-41 at 60 8C h
1 gives 90 ppm of Mo(VI) in solution after
3 h epoxidation of cyclooctene in decane, whereas at
24 8C h
1, only 11 ppm of Mo(VI) is present in the solution.
Mo-containing silicas display interesting catalytic activity
towards cyclooctene epoxidation, with low leaching of active
species under specific conditions. However, cyclooctene is well
known as a non-demanding substrate for epoxidation. The
Fig. 8. UV–vis diffuse reflectance spectra of Mo-MCM-41. Days since synth-
esis/days since calcinations: (a) 0/0, (b) 15/30 and (c) 30/15.
97
Mo-SBA-15, decane Mo–SiO2-bead, decane
Conversion Mo Conversion Mo
(%) (ppm) (%) (ppm)
Mo (Si/Mo)
(ppm) (mol/mol)
148 – 95 100 80 8
<1 – 75 <2 79 <2
4 69 61 <2 57 1
regioselective epoxidation of R-(+)-limonene is more difficult,
and side-reactions can develop even at room temperature on the
substrate and/or the epoxides [48,49].
Table 4 shows that limonene can be epoxidized with all the
Mo-containing siliceous catalysts with good regioselectivity for
monoepoxides (cis and trans, see Scheme 1), the by-products
being mainly diepoxides and diols. As for cyclooctene
epoxidation, Mo-containing mesoporous materials are more
efficient than Mo–SiO2-bead (Table 4, columns 2–6). However,
the grinding of the silica beads or the use of precipitated silica
as the non-porous silica source leads to very active catalysts
(Table 4, columns 7 and 8), which can epoxidize limonene with
good selectivity, the beads having probably an unfavorable
effect on substrate and/or product diffusion, as previously
mentioned for cyclooctene epoxidation. Concerning the
leaching of active species, Mo-containing mesoporous materi-
als seems better than Mo–SiO2-bead. However, Mo–SiO2-prec
(Table 4, column 8) seems to be by far the best catalyst for
limonene epoxidation, since good conversion and selectivity
can be obtained and very low leaching of Mo(VI) is observed,
especially after the second run, the third run (not shown) giving
approximately the same results as the second run. It must be
noted that the impregnation of the precipitated silica by the
peroxidic Mo(VI) solution needs to scrupulously respect the
two following points: (i) too much H2O2 leads to very few
anchored oxo–peroxo species on the silica surface and then to a
very low-Mo(VI) content after thermal treatment, (ii) the native
impregnated sample must not be washed before calcination; the
calcination process is important to obtain a strong interaction
between Mo(VI) and silanol groups.
Limonene can also be epoxidized by the expensive
‘‘methyltrioxorhenium/H2O2/H2O’’ system (MTO/H2O2)
(methyltrioxorhenium) in two-phase catalysis [48,49] or by
oxoperoxophosphatotungstate complexes in phase transfer
catalysis [50] (see Table 5). Obviously, comparison of het-
erogeneous catalysts and complexes (Tables 4 and 5) show
that siliceous materials are less active than Re(VII) or W(VI)
Scheme 1. Main and byproducts resulting from the epoxidation of R-(+)-
limonene.
98 P. Célestin Bakala et al. / Applied Catalysis A: General 300 (2006) 91–99

Table 4
Limonene epoxidation over Mo-catalysts, after 24 h reaction at room temperature
Mo-MCM-41 (68)a Mo-SBA-15 Mo–SiO2-bead Mo–SiO2-bead Mo–SiO2-prec
(122)a (160)a (160)a ‘‘crushed beads’’ (55)a
Pentane Hexane Pentane Pentane Pentane Pentane
T (8C) 20 20 40 20 20 20 20
Conversion (%) 51 46 78 44 40 94 66 (71)b
Selectivity (%) for monoepoxide 87 87 91 80 83 86 84 (92)b
a/b ratioc 2.5 2.9 2.2 3.6 3.0 3.4 2.5 (1.8)b
Selectivity (%) for diepoxide 9 6 7 14 13 7 10 (5)b
Selectivity (%) for diols 4 5 2 6 4 7 6 (3)b
Mod (ppm) 18 16 ye 17 38 ye 27 (<2)b
Mo loss (mol%) 3 3 – 5 15 - 3 (0.2)b
In all cases, the TBHP selectivity is above 90%.
a
Limonene/Mo (mol/mol).
b
Data of the second catalysis run.
c
Cis/trans ratio of the monoepoxides.
d
Mo wt.% in the catalytic solution after the first catalysis run.
e
The solution is yellow, indicating a large amount of leached Mo species.

Table 5
Limonene epoxidation over MTO in biphase CH2Cl2/H2O–H2O2 system and over oxoperoxophosphatotungstate complexes in phase transfer catalysis
Catalyst T (8C) t (h) Conversion (%) Selectivity (%) for monoepoxides a/b ratio
MTO 4 2 98 86 1
PWn 20 0.5 >99 98 1
Mo–SiO2-prec 20 24 66 84 2.5

complexes for limonene epoxidation, especially for the
conversion, the selectivity for monoepoxides being approxi-
mately the same with all catalysts (i.e. >80%). The ratio
between cis and trans diastereoisomers of the limonene
epoxides clearly shows that Mo-containing silica favors the
cis isomer, whereas an equimolar mixture is obtained for
Re(VII) and W(VI) complexes, in favor of the heterogeneous
nature of the epoxidation processes on Mo-containing silica. Of
all the Mo-containing catalysts, that made with precipitated
silica presents very good catalytic activity, very low leaching of
active species, and the possibility of easy recycling.
drous TBHP, not as high as that of Re(VII) or W(VI)
complexes, but the ease and the low cost of their synthesis, as
compared to mesoporous catalysts, make them good candidates
for oxidation reactions; they should be considered for other uses
of molybdenum(VI) silica catalysts.
To the best of our knowledge, SiO2–MoO3 binary condensed
phase diagrams are not known. The so-called ‘‘peroxo route’’
with low-nuclearity peroxo metal precursors appears as an
elegant method for combining these oxides which have a priori
no strong affinity, and to obtain fair molybdenum content and
high dispersity.

4. Conclusion Acknowledgements

A methodology, developed for the insertion of MoOx into
mesoporous materials (MCM-41 and SBA-15) has been
successfully extended to the preparation of MoOx on silica
by wet impregnation. The high dispersion of the peroxo species
on supports promotes the generation of active oxidation
catalysts, although one run is maybe necessary to obtain a
steady-state catalytic activity and stable catalysts. The
preparation methods and thermal treatments of as-synthesized
materials, which determine the resistance to attrition and to
leaching of active species in the presence of a nucleophilic
reagent such as TBHP, remain to be optimized. Demanding
olefins such as limonene can also be epoxidized with Mo-
containing mesoporous materials or with Mo–SiO2 prepared
from beads, but with fair conversion.
Mo–SiO2 materials prepared from precipitated silica display
interesting catalytic activity towards epoxidation with anhy-
We thank Dr. John Lomas for constructive discussions and for
correcting the manuscript, Dr. J.-Y. Piquemal for UV–vis spec-
troscopy and Dr. F. Villain for Raman spectrometry facilities.
References
[1] R. Noyori, M. Aoki, K. Sato, Chem. Commun. (2003) 1977.
[2] J.-M. Brégeault, J. Chem. Soc., Dalton Trans. (2003) 3289.
[3] T. Yanagisawa, T. Shimizu, K. Kuroda, C. Kato, Bull. Chem. Soc. Jpn. 63
(1990) 988.
[4] T. Yanagisawa, T. Shimizu, K. Kuroda, C. Kato, Bull. Chem. Soc. Jpn. 63
(1990) 1535.
[5] J.S. Beck, J.C. Vartuli, J.W. Roth, M.E. Leonowicz, C.T. Kresge, K.D.
Schmitt, C.T.W. Chu, D.H. Olson, E.W. Sheppard, S.B. McCullen, J.-B.
Higgins, J.L. Schlenker, J. Am. Chem. Soc. 114 (1992) 10834.
[6] R. Landau, G.A. Sullivan, D. Brown, Chem. Technol. 6 (1979) 602.
[7] C.T. Kresge, M.E. Leonowicz, J.W. Roth, J.C. Vartuli, J.S. Beck, Nature
359 (1992) 710.
 P. Célestin Bakala et al. / Applied Catalysis A: General 300 (2006) 91–99 99

[8] T. Li, S. Cheng, J.-F. Lee, L.Y. Jang, J. Mol. Catal. A: Chem. 198 (2003) 139. [28] P.S. Raghavan, M.P. Vinod, J. Mol. Catal. A: Chem. 135 (1998) 47.
[9] Z. Li, L. Gao, S. Zheng, Appl. Catal. A: Gen. 236 (2002) 163. [29] R.K. Rana, B. Viswanathan, Catal. Lett. 52 (1998) 25.
[10] S. Che, Y. Sakamoto, H. Yoshitake, O. Terasaki, T. Tatsumi, J. Phys. [30] D. Juwiler, J. Blum, R. Neumann, Chem. Commun. (1998) 1123.
Chem. B 105 (2001) 10565. [31] D.-H. Cho, T.-S. Chang, S.-K. Ryu, Y.K. Lee, Catal. Lett. 64 (2000) 227.
[11] M. Pillinger, I.S. Gonçalves, A.D. Lopes, P. Ferreira, J. Rocha, G. Zhang, M. [32] R.A. Sheldon, M. Wallau, I.W.C.E. Arends, U. Schuchardt, Acc. Chem.
Schäfer, O. Nuyken, F.E. Kühn, Phys. Chem. Chem. Phys. 4 (2002) 696. Res. 31 (1998) 485.
[12] D. Nunes, A.A. Valente, M. Pillinger, A.C. Fernandes, C.C. Romão, J. [33] I.W.C.E. Arends, R.A. Sheldon, Appl. Catal. A: Gen. 212 (2001) 175.
Rocha, I.S. Gonçalves, J. Mater. Chem. 12 (2002) 1735. [34] L. Davies, P. McMorn, G.J. Hutchings, D. Bethell, P.C. Bulman Page, F.
[13] M.S. Morey, J.D. Bryan, S. Schwarz, G.D. Stucky, Chem. Mater. 12 (2000) King, F.E. Hancock, Chem. Commun. (2000) 1807.
3435. [35] H. Jeziorowski, H. Knözinger, J. Phys. Chem. 83 (1979) 1166.
[14] I.S. Gonçalves, A.M. Santos, C.C. Romão, A.D. Lopes, J.E. Rodriguez- [36] E. Briot, J.-Y. Piquemal, M. Vennat, J.-M. Brégeault, G. Chottard, J.-M.
Borges, M. Pillinger, P. Ferreira, J. Rocha, F.E. Kühn, J. Organomet. Manoli, J. Mater. Chem. 10 (2000) 953.
Chem. 626 (2001) 1. [37] L. Marchese, E. Gianotti, V. Dellarocca, T. Maschmeyer, F. Rey, S.
[15] K.K. Kang, W.S. Ahn, J. Mol. Catal. A: Chem. 159 (2000) 403. Coluccia, J.-M. Thomas, Phys. Chem. Chem. Phys. 1 (1999) 585.
[16] P. Ferreira, I.S. Gonçalves, F.E. Kühn, M. Pillinger, J. Rocha, A. Thurs- [38] J. Choma, M. Jaroniec, W. Burakiewicz-Mortka, M. Kloske, Appl. Surf.
field, W.-M. Xue, G. Zhang, J. Mater. Chem. 10 (2000) 1395. Sci. 196 (2002) 216.
[17] I.J. Shannon, T. Maschmeyer, R.D. Oldroyd, G. Sankar, J.M. Thomas, H. [39] S.J. Gregg, K.S.W. Sing, Adsorption, Surface Area and Porosity, Aca-
Pernot, J.-P. Balikdjian, M. Che, J. Chem. Soc., Faraday Trans. 94 (1998) demic Press, London, 1982.
1495. [40] A. Sayari, M. Kruk, M. Jaroniec, Catal. Lett. 49 (1997) 147.
[18] Q. Yang, C. Copéret, C. Li, J.-M. Basset, N. J. Chem. 27 (2003) 319. [41] M. Jaroniec, M. Kruk, J.P. Olivier, Langmuir 15 (1999) 5410.
[19] M. Jia, W.R. Thiel, Chem. Commun. (2002) 2392. [42] B.G. Aristov, A.V. Kisilev, Kolloidn. Z. 27 (1965) 299.
[20] P. Ferreira, I.S. Gonçalves, F.E. Kühn, A.D. Lopes, M.A. Martins, M. [43] R.S. Weber, J. Catal. 151 (1995) 470.
Pillinger, A. Pina, J. Rocha, C.C. Romão, A.M. Santos, T.M. Santos, A.A. [44] E. Iglesia, D.G. Barton, S.L. Soled, S. Miseo, J.E. Baumgartner, W.E.
Valente, Eur. J. Inorg. Chem. (2000) 2263. Gates, G.A. Fuentes, G.D. Meitzner, in: J.W. Hightower, W.N. Delgass,
[21] L.-X. Dai, Y.-H. Teng, K. Tabata, E. Suzuki, T. Tatsumi, Microp. Mesop. E. Iglesia, A.T. Bell (Eds.), Studies in Surface Science and Catalysis,
Mater. 44–45 (2001) 573. Proceedings of the 11th International Congress on Catalysis—40th Anni-
[22] M. Masteri-Farahani, F. Farzaneh, M. Ghandi, J. Mol. Catal. A: Chem. 192 versary, vol. 101, Elsevier, Amsterdam, 1996, p. 533.
(2003) 103. [45] C.Y. Chen, H.-X. Li, M.E. Davies, Micropor. Mater. 2 (1993) 17.
[23] J.-Y. Piquemal, E. Briot, G. Chottard, P. Tougne, J.-M. Manoli, J.-M. [46] C. Mestl, T.K.K. Srinivasan, Catal. Rev. Sci. Eng. 40 (1998) 451.
Brégeault, Micropor. Mesopor. Mater. 58 (2003) 279. [47] C. Rocchiccioli-Deltcheff, R. Thouvenot, R. Franck, Spectrochim. Acta A
[24] S. Higashimoto, R. Tsumura, S.G. Zhang, M. Matsuoka, H. Yamashita, C. 32 (1976) 587.
Louis, M. Che, M. Anpo, Chem. Lett. (2000) 408. [48] A. Deloffre, S. Halut, L. Salles, J.-M. Brégeault, J.R. Gregorio, B. Denise,
[25] J.-Y. Piquemal, E. Briot, M. Vennat, J.-M. Brégeault, G. Chottard, J.-M. H. Rudler, J. Chem. Soc., Dalton Trans. (1999) 2897.
Manoli, Chem. Commun. (1999) 1195. [49] H. Rudler, J.R. Gregorio, B. Denise, J.-M. Brégeault, A. Deloffre, J. Mol.
[26] E. Briot, J.-Y. Piquemal, J.-M. Brégeault, N. J. Chem. 26 (2002) 1443. Catal. A: Chem. 133 (1998) 255.
[27] U. Arnold, R. Serpa da Cruz, D. Mandelli, U. Schuchardt, J. Mol. Catal. A: [50] L. Salles, J.-M. Brégeault, R. Thouvenot, C.R. Acad. Sci. Ser. IIc Chim./
Chem. 165 (2001) 149. Chem. 3 (2000) 183.