The coordinatively saturated vanadium MIL-47 as a low leaching heterogeneous catalyst in the oxidation of cyclohexene

Article abstract of DOI:10.1016/j.jcat.2011.09.014

A Metal Organic Framework, containing coordinatively saturated V ^+IV sites linked together by terephthalic linkers (V-MIL-47), is evaluated as a catalyst in the epoxidation of cyclohexene. Different solvents and conditions are tested and compared. If the oxidant TBHP is dissolved in water, a significant leaching of V-species into the solution is observed, and also radical pathways are prominently operative leading to the formation of an adduct between the peroxide and cyclohexene. If, however, the oxidant is dissolved in decane, leaching is negligible and the structural integrity of the V-MIL-47 is maintained during successive runs. The selectivity toward the epoxide is very high in these circumstances. Extensive computational modeling is performed to show that several reaction cycles are possible. EPR and NMR measurements confirm that at least two parallel catalytic cycles are co-existing: one with V^+IV sites and one with pre-oxidized V ^+V sites, and this is in complete agreement with the theoretical predictions.

Full text of DOI:10.1016/j.jcat.2011.09.014

Journal of Catalysis 285 (2012) 196–207
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
The coordinatively saturated vanadium MIL-47 as a low leaching
heterogeneous catalyst in the oxidation of cyclohexene
Karen Leus a,1, Matthias Vandichel , Ying-Ya Liu , Ilke Muylaert , Jan Musschoot , Steven Pyl ,
b,1
a
a
c
d
c
c
d
c
e
Henk Vrielinck , Freddy Callens , Guy B. Marin , Christophe Detavernier , Paul V. Wiper ,
Yaroslav Z. Khimyak , Michel Waroquier , Veronique Van Speybroeck , Pascal Van Der Voort
Department of Inorganic and Physical Chemistry, Center for Ordered Materials, Organometallics and Catalysis, Ghent University, Krijgslaan 281-S3, 9000 Ghent, Belgium
Center for Molecular Modeling, Ghent University, Technologiepark 903, 9052 Zwijnaarde, Belgium
Department of Solid State Sciences, Ghent University, Krijgslaan 281-S1, 9000 Ghent, Belgium
e
b
b,
⇑
a,
⇑
a
b
c
d
e
Laboratory for Chemical Technology, Ghent University, Krijgslaan 281-S5, 9000 Ghent, Belgium
Department of Chemistry, University of Liverpool, Liverpool L69 3BX, United Kingdom
a r t i c l e i n f o
a b s t r a c t
A Metal Organic Framework, containing coordinatively saturated V+IV sites linked together by tere-
phthalic linkers (V-MIL-47), is evaluated as a catalyst in the epoxidation of cyclohexene. Different sol-
vents and conditions are tested and compared. If the oxidant TBHP is dissolved in water, a significant
leaching of V-species into the solution is observed, and also radical pathways are prominently operative
leading to the formation of an adduct between the peroxide and cyclohexene. If, however, the oxidant is
dissolved in decane, leaching is negligible and the structural integrity of the V-MIL-47 is maintained dur-
ing successive runs. The selectivity toward the epoxide is very high in these circumstances. Extensive
computational modeling is performed to show that several reaction cycles are possible. EPR and NMR
Article history:
Received 9 August 2011
Revised 12 September 2011
Accepted 13 September 2011
Available online 19 October 2011
Keywords:
Metal Organic Frameworks
Vanadium
Oxidation catalysis
DFT-D
+
IV
measurements confirm that at least two parallel catalytic cycles are co-existing: one with V
sites
+
V
and one with pre-oxidized V sites, and this is in complete agreement with the theoretical predictions.
Epoxidation
Ó 2011 Elsevier Inc. All rights reserved.
1
. Introduction
Metal Organic Frameworks (MOFs) have received an enormous
reactions [12–18], enantioselective reactions [19–21], photocataly-
sis [22], carbonyl cyanosilylation [23–25], hydrodesulfurization
[26], and esterification [27].
amount of attention over the last decade, as they are considered as
the newest generation of porous materials [1,2]. This type of mate-
rials is constructed by linking organic ligands with metal ions or
metal clusters to form infinite network structures. The porosity,
structure, and functionality of the MOF can be tuned by varying
the linkers (bridging ligands) and/or the metal centers. MOFs are
promising materials for applications in gas adsorption [3,4], sepa-
ration [5], and catalysis [6–8]. By introducing catalytically active
metal-connecting points or functional groups into the highly por-
ous MOF structure, MOFs exhibit clear advantages in heteroge-
neous catalysis. Moreover, unlike traditional immobilized
catalysts [9–11], the active catalytic sites in MOFs are well pro-
tected by the coordination bonds that prevent leaching to a higher
extent. Although the research in this field is relatively new, MOFs
have been studied as catalysts in hydrogenation and oxidation
Catalytic studies on the oxidation of cyclohexene in MOFs have
already been published by several groups. For example, the group
of Rosseinsky [28] has reported a post-modification procedure of
2
the IRMOF-3 framework with VO(acac) . They carried out a preli-
minary catalytic investigation on this modified MOF structure for
the oxidation of cyclohexene with tert-butylhydroperoxide (TBHP)
as the oxidant. Volkmer and co-workers [14] have designed and
synthesized a MOF containing unsaturated cobalt sites, namely
MFU-1. MFU-1 showed a cyclohexene conversion of 22.5% after
22 h of reaction (theoretically maximum 50% conversion), with
tert-butyl-2-cyclohexenyl-1-peroxide as the main product (66%
selectivity). Garcia et al. recently published a review on the use
of MOFs as oxidation catalysts [29]. Recent studies by us [30] have
focused on the catalytic performance of a saturated vanadium
IV
MOF, namely MIL-47 V O(CO
–C
2 6
4 2
H –CO ) [31].
As can be seen fromFig. 1, MIL-47has a three-dimensional frame-
+
IV
work, where each V center is coordinated to four oxygen atoms
from four carboxylate groups, and to two oxygen atoms on the O–
V–O axis, thus forming a saturated octahedral coordination node.
The nodes are further connected by sharing the carboxylate linkers
and thus grow into a three-dimensional framework. 1D rhombus
⇑
Corresponding authors. Fax: +32 9 264 49 83 (P. Van Der Voort), +32 9 264 65
0 (V. Van Speybroeck).
E-mail addresses: Veronique.VanSpeybroeck@UGent.be (V. Van Speybroeck),
6
Pascal.Vandervoort@ugent.be (P. Van Der Voort).
1
Both authors contributed equally to this work.
0
021-9517/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2011.09.014

K. Leus et al. / Journal of Catalysis 285 (2012) 196–207
197
2
. Experimental: materials and methods
2.1. General procedures
All chemicals for the synthesis of MIL-47 and the catalytic tests
were purchased from Sigma Aldrich (unless otherwise noted) and
used without further purification. An ultra-fast GC equipped with
a flame ionization detector (FID) and a 5% diphenyl/95% poly-
dimethylsiloxane column with 10 m length and 0.10 mm internal
diameter was used to follow the conversions of the products dur-
ing the catalytic tests. Helium was used as carrier gas, and the flow
rate was programmed as 0.8 mL/min. The reaction products were
identified with a TRACE GC ꢀ GC (Thermo, Interscience), coupled
to a TEMPUS TOF-MS detector (Thermo, Interscience). The first
column consists of a dimethyl polysiloxane package and has a
length of 50 m, with an internal diameter of 0.25 mm, whereas
the second column has a length of 2 m with an internal diameter
of 0.15 mm. The package of the latter is a 50% phenyl polysilphen-
ylene-siloxane. Helium was used as carrier gas with a constant
flow (1.8 mL/min).
Fig. 1. Perspective view of the structure of MIL-47.
X-ray fluorescence (XRF) measurements were performed with
an ARTAX (Bruker) Peltier cooled silicon drift detector, using a
Mo X-ray source. The detector was placed above the surface of
the solution. Spectra were accumulated during 200 s. The system
was calibrated by measuring the XRF spectra of different aque-
channels are along the a axis with sizes of about 10.5 ꢀ 11.0 Å
(
Van der Waals radii excluded) for the channel windows. The struc-
ture shows good thermal stability (stable up to 350 °C in air).
ous-based solutions of VO(acac)
2
and Co(NO
3
)
2 2
ꢁ6H O with known
Recently, we published the first catalytic evaluation of V-MIL-
V/Co ratios. Nitrogen adsorption experiments were carried out at
ꢂ196 °C using a Belsorp-mini II gas analyzer. Prior to analysis,
the samples were dried overnight under vacuum at 120 °C to re-
move adsorbed water. X-ray powder diffraction (XRPD) patterns
were collected on a ARL X’TRA X-ray diffractometer with Cu Ka
radiation of 0.15418 nm wavelength and a solid-state detector.
Electron paramagnetic resonance (EPR) spectra of dry powders
were recorded at room temperature with a Bruker ESP300E X-band
4
7 in the liquid phase oxidation of cyclohexene, using TBHP in
water as the oxidant and chloroform as the solvent [30]. In general,
the liquid phase oxidation of cyclohexene can produce one or more
of following products as shown in Scheme 1.
In this paper, we have extended our earlier work to a more in-
depth mechanistic study. We examine the catalytic performance of
MIL-47 in water and water-free media. In each medium, two
catalyst loadings will be considered: 0.11 mmol and 0.42 mmol
V-centers in the total reaction mixture. The catalytic activity of
(
9.77 GHz) spectrometer, equipped with a HP5350B frequency
counter and a ER035 M gaussmeter. Standard NMR sample tubes
4 mm inner diameter) were filled to a height of approximately
mm, and for comparison of intensities, the spectra were divided
2
the MIL-47 is compared to the homogeneous catalyst VO(acac) .
(
5
In order to fully unravel possible reaction mechanisms, which are
operative in the MIL-47, an extensive ab initio computational study
has been conducted. The nature of the active site is investigated by
studying a variety of possible reaction cycles that might occur in
the cyclohexene oxidation. Some active sites start from vanadium
in oxidation state (+IV) where others produce first vanadium in
oxidation state (+V). EPR and NMR measurements were also per-
formed to determine the presence and activity of vanadium in both
oxidation states.
by the powder mass. For all spectra, the magnetic field was modu-
lated at 100 kHz with an amplitude of 0.2 mT, and the microwave
power was set to 2 mW, avoiding saturation.
2.2. Synthesis of MIL-47
MIL-47 was synthesized according to a synthesis route de-
scribed in literature [31]. Typically, 1.37 g VCl
phthalic acid were mixed together with 15.7 mL of deionised
O. The resulting mixture was transferred into a Teflon-lined
stainless autoclave for 4 days at 200 °C. The as-synthesized MIL-
3
and 0.36 g tere-
H
2
4
7 was filtered, washed with acetone, and calcined for 21 h
3
0 min at 300 °C to remove free terephthalic acid from the pores.
Typical synthesis yield is approximately 15%.
2.3. General procedure for the cyclohexene oxidation
During a typical catalytic test, a 100-mL round-bottom flask
was charged with 30 mL of chloroform (anhydrous) used as sol-
vent, 5 mL of cyclohexene, and 6.2 mL of 1,2,4-trichlorobenzene
(
used as internal standard for the GC analysis). The oxidants used
in this paper were tert-butylhydroperoxide (TBHP) in water (70%)
and TBHP 5.5 M in decane. The molar ratio cyclohexene/oxidant
was 1/2. All the catalytic tests were performed at a temperature
of 50 °C and with an Ar-containing balloon on top of the condenser.
Blank reactions at this temperature showed no catalytic conversion
of cyclohexene.
Scheme 1. Oxidation of cyclohexene 1 toward the main reaction products: (a)
epoxidation to cyclohexene oxide 2 and consecutive ring opening to cyclohexane-
1
,2-diol 3, (b) radical pathway to tert-butyl-2-cyclohexenyl-1-peroxide 4 and (c)
allylic oxidation to 2-cyclohexene-1-one 5.

1
98
K. Leus et al. / Journal of Catalysis 285 (2012) 196–207
The reaction mixture was stirred under an inert argon atmo-
sphere until a plateau was reached in the cyclohexene conversion.
Aliquots were gradually taken out of the mixture, diluted with
time. (3) In the initial stages, the cyclohexene conversion is almost
linear in all cases. There are, however, differences in the slope of
these linear sections: the 0.42 mmol V catalytic systems start
clearly the fastest (approximately 20% conversion per hour for
TBHP in water and 7% conversion per hour of TBHP in decane)
whereas the 0.11 mmol V systems start much slower (ꢃ4% conver-
sion per hour for the TBHP in water and even slower for the TBHP
in decane). (4) In the systems that use TBHP in water, the cyclohex-
ene oxide attains a maximum, after which the concentration grad-
ually returns to zero. This is partly due to the transformation of the
epoxide 2 into the diol 3 resulting from the reaction with water
and terephthalic acid (as shown in Scheme 1). (5) The byproduct
5
00 ll ethylacetate, and subsequently analyzed by GC-FID and
GC ꢀ GC TOF-MS. After a catalytic run, the catalyst was recovered
by filtration on a combined nylon-membrane filter, washed with
acetone, and vacuum dried overnight. The filtrate was analyzed
by XRF to determine the leached vanadium.
3
. Experimental results
3
.1. Characterization
4, formed via radicalar pathways, grows systematically and ex-
ceeds the concentration of the epoxide after sufficient long reac-
tion times. The crossing of those two yields occurs at relative
short reaction times for the TBHP in water systems and is heavily
delayed by using the TBHP in decane. This is due to the particular
role of water in the reaction medium in stabilizing the radical
intermediates (see computational discussion, Section 3.2). (6) The
formation of the ketone 5 is marginal in all cases. (7) During the
first 7 h of reaction, the selectivity toward the cyclohexene oxide
is almost 100% in the 0.11 mmol V-sites with TBHP in water
system. After that, the product distribution changes. The cyclohex-
ene oxide reaches a maximum of 38% after 25 h of reaction. After
that, the cyclohexane-1,2-diol 3 starts to be produced, and the
tert-butyl-2-cyclohexenyl-1-peroxide 4 becomes dominant (see
Table 1).
The faster kinetics observed in Fig. 3c compared to Fig. 3a are
mainly due to the higher V-loading. But in both cases, also a signif-
icant amount of leached V is observed. Our earlier work [30]
showed that after 1 h of catalysis, 12.8% of the V is leached out
of the structure with 0.11 mmol of V-sites.
In order to reduce this leaching, the catalytic performance of
MIL-47 in the oxidation of cyclohexene was analyzed with TBHP/
decane as oxidant (see Fig. 3b and d). When the catalytic perfor-
mance of MIL-47 in the water-based media is compared to the
water-free media, then it is clear that the cyclohexene conversion
is approximately 30% after 7 h of reaction with TBHP/decane as
oxidant, while the total cyclohexene conversion in the water dis-
solved TBHP system is much higher in the initial stage (55% after
In this paper, we will compare the catalytic performance of MIL-
7 in water and water-free media. A thorough study of the catalytic
performance of MIL-47 is presented, together with a spectroscopic
study and a computational model.
4
3
.1.1. Characterization of the catalyst
The MIL-47 had an average Langmuir surface area of 1225 m /g
2
and a pore volume of 0.38 ml/g. Only batches of MIL-47 that di-
verted less than 5% from these values were used in this work.
2
The value of 1225 m /g is consistent with literature [31].
The XRPD pattern of the synthesized MIL-47 is presented in
Fig. 2 and compared to the XRPD pattern, generated from the ori-
ginal CIF-file [31]. The experimental XRPD pattern of MIL-47
matches perfectly with the simulated pattern, confirming the
phase purity of the synthesized MIL-47. Some differences in reflec-
tion intensities between the simulated and experimental pattern
are due to the variation in crystal orientation of the micro crystals.
3
.1.2. The catalytic performance of MIL-47 in the oxidation
of cyclohexene
In Fig. 3, the catalytic performance of MIL-47 is shown using
respectively TBHP/water (left) and TBHP/decane (right) as oxidant.
In this figure, two catalyst loadings (0.11 mmol and 0.42 mmol of
V-centers) are considered. The reader should pay attention to the
differences in the timescale of the x-axes, as all experiments were
performed until a linear plateau in the conversion was reached or
until a maximum time of 82 h.
Several general trends can be deducted from Fig. 3. (1) Conver-
gence regimes for cyclohexene conversion are reached in all cases,
provided sufficient large reaction time is taken into account. (2)
The conversion tends toward 80% at the end of the reaction. The
lower conversion in Fig. 3c is probably due to a too short reaction
7
h of reaction). Furthermore, it is interesting to note that, as the
conversion is lower, the selectivity toward the cyclohexene oxide
is remarkably higher. In Table 1, the selectivity toward the cyclo-
hexene oxide with TBHP/water and TBHP/decane as oxidant is
shown for each V-loading, together with the TON and TOF of
2
MIL-47 and VO(acac) . This Table 1 shows that the selectivity in
TBHP/decane is significantly higher compared to the water-based
equivalent.
The presence of water during the catalytic test clearly affects
both the rate of the epoxidation reaction and the stabilization of
radical species toward the formation of the adduct 4. More details
about the reaction mechanisms for both systems are given in the
computational sections (vide infra). The former effect might be
partly due to vanadium leaching. Only a negligible amount (<3%)
of vanadium was leached out of the structure when TBHP/decane
was used as oxidant after 7 h of reaction, whereas with TBHP/
water, after the same time, a V-leaching of almost 50% was
observed.
2
3.1.3. Catalytic performance of VO(acac) in the oxidation
of cyclohexene
Next, the catalytic performance of MIL-47 in TBHP/water and
TBHP/decane as oxidant was compared with the homogeneous cat-
2
alyst VO(acac) in identical conditions and V-loadings. In Fig. 4, the
cyclohexene conversion and yield of cyclohexene oxide 2, cyclo-
hexane-1,2-diol 3, and the tert-butyl-2-cyclohexenyl-1-peroxide
ꢄ
Fig. 2. XRPD patterns of MIL-47 (experimental and simulated [31]) ( is due to the
background of the sample holder at an angle of 32.9°).

K. Leus et al. / Journal of Catalysis 285 (2012) 196–207
199
Fig. 3. Cyclohexene (1) conversion (j) and the yield of cyclohexene oxide (2) (d), tert-butyl-2-cyclohexenyl-1-peroxide (4) (N) and cyclohexane-1,2-diol (3) (s) for MIL-47 in
TBHP/water (Left) with 0.11 mmol V-sites (a) and 0.42 mmol V-sites (c); and for MIL-47 in TBHP/decane (Right) with 0.11 mmol V-sites (b) and 0.42 mmol V-sites (d). No
cyclohexane-1,2-diol (s) is formed when TBHP in decane is used. The production of 2-cyclohexene-1-one is in all cases less than 5%, for the clarity of the figure this is not
shown.
Table 1
Selectivity toward the cyclohexene oxide with TBHP/water and TBHP/decane as oxidant for each V-loading and the TON
and TOF of MIL-47 and VO(acac) .
2
TONa
TOF (h )b
ꢂ1
Selectivityc
V-loading (mmol)
Oxidant
Catalyst
0.11
0.42
0.11
0.42
0.11
0.42
0.11
0.42
TBHP/water
TBHP/water
TBHP/decane
TBHP/decane
TBHP/water
TBHP/water
TBHP/decane
TBHP/decane
MIL-47
MIL-47
MIL-47
MIL-47
VO(acac)
VO(acac)
VO(acac)
VO(acac)
150
67
42
28
169
62
30
29
37
8
44
57
35
48
70
43
63
83
70
25
87
78
2
2
2
2
112
74
a
b
c
TON was calculated after 7 h.
TOF was calculated after 30 min.
The selectivity toward the cyclohexene oxide calculated at 40% cyclohexene conversion.
4
is shown for the VO(acac)
2
catalyst respectively in the water and
some differences. For the VO(acac)2, the cyclohexene conversion
is much faster compared to the MIL-47. After 7 h of reaction, a
cyclohexene conversion of 65% is observed for the 0.42 mmol V-
sites compared to 30% for the microporous MOF. Likewise, with
the lower V content, it can be seen that the cyclohexene conversion
water-free medium for total V-loadings of 0.11 and 0.42 mmol. The
reader is again reminded that the x-axes have different timescales.
The homogeneous catalyst shows a similar behavior as MIL-47
in the water-based oxidant. Firstly, the cyclohexene conversions
reach a plateau after approximately the same time (80% after
2
is much slower compared to the VO(acac) catalyst. Nevertheless,
7
0 h for 0.11 mmol and 55% after 7 h of catalysis for 0.42 mmol
the selectivity of both catalysts toward cyclohexene oxide is more
or less the same as can be seen from Table 1.
V-sites). Secondly, for both V-loadings, the cyclohexene oxide 2 at-
tains a maximum after 20 h for the 0.11 mmol V and 2 h for the
0
.42 mmol V when the cyclohexane-1,2-diol 3 starts to be pro-
3.1.4. Stability and regenerability of the catalyst
duced and the tert-butyl-2-cyclohexenyl-1-peroxide 4 becomes
the main product. However, comparison of the homogeneous cat-
alyst with the MIL-47 for TBHP/decane as oxidant clearly shows
The crystalline structure of the MOF was examined by X-ray dif-
fraction to verify the stability of the MOF after a catalytic run. In
Fig. 5, the XRPD pattern of MIL-47 is shown before and after a

2
00
K. Leus et al. / Journal of Catalysis 285 (2012) 196–207
Fig. 4. Cyclohexene (1) conversion (j) and the yield of cyclohexene oxide (2) (d), tert-butyl-2-cyclohexenyl-1-peroxide (4) (N) and cyclohexane-1,2-diol (3) (s) for
VO(acac) in TBHP/water (left) with 0.11 mmol V-sites (a) and 0.42 mmol V-sites (c); and for VO(acac) in TBHP/decane (right) with 0.11 mmol V-sites (b) and 0.42 mmol V-
2
2
sites (d). No cyclohexane-1,2-diol (s) is formed when TBHP/decane is used. The production of 2-cyclohexene-1-one is in all cases less than 5%, for the clarity of the figure this
is not shown.
catalytic run, using TBHP/water or TBHP/decane as oxidant at a
temperature of 50 °C. Furthermore, the XRPD pattern of the organic
linker in MIL-47 (terephthalic acid or BDC, benzene dicarboxylic
acid) is shown. As can be seen from this figure, the XRPD pattern
of the MIL-47 after a first catalytic run with TBHP in water as oxi-
dant (Fig. 5c) changes significantly compared to the original XRPD
pattern (Fig. 5a). Especially, some new diffraction peaks, which can
be assigned to the BDC linkers, appear after a first catalytic run.
This observation indicates that the structure partially decomposes
in the TBHP/water system. This also explains the significant V-
leaching that has been observed. Subsequently, MIL-47 was regen-
erated by a treatment in a tubular furnace under a nitrogen flow at
To test the regenerability of the catalyst, four consecutive cata-
lytic runs have been performed in TBHP/decane as oxidant. During
each catalytic cycle, the cyclohexene conversion was monitored
during at least 6 h. The XRPD patterns of the original MIL-47 and
after each run are shown in Fig. 6. As can be seen from this figure,
MIL-47 preserves its crystalline structure during these additional
runs. The lower signal to noise ratio in run IV is due to the tiny
amount of sample that was left for analysis.
In Table 2, the TON and TOF values are shown for MIL-47 in the
four consecutive runs. Furthermore, the percentage of leached V in
the filtrate is given after each catalytic test. Only in the first cycle, a
negligible amount of leached V was observed (<3%), whereas in the
following cycles, no leaching was detected. The TON and TOF drop
after the first run but remain relatively constant during the next
additional cycles. This observation can be explained by the fact
that after each run, still some organic compounds of the previous
catalytic test clog the pores. This was supported by nitrogen
adsorption measurements that showed that the surface area of
1
50 °C for 4 h to remove the organic compounds inside the pores.
After this extra step, the structure of MIL-47 was regained as
shown in Fig. 5d.
Comparison of the XRPD pattern of MIL-47 before and after a
catalytic run, using TBHP in decane (Fig. 5a and b) as oxidant, how-
ever, clearly demonstrates that in a water-free medium, the struc-
tural integrity of the MOF is entirely preserved. No peaks of free
BDC can be observed, and no additional treatment is required to re-
move these free linkers.
2
MIL-47 after a first run dropped from 1225 m /g to approximately
2
500 m /g; this will surely lead to a decrease of the reaction rate in
the additional runs. We have shown previously [30] that even in

K. Leus et al. / Journal of Catalysis 285 (2012) 196–207
201
Table 2
TOF, TON and the percentage of leached V for the consecutive catalytic runs with
TBHP/decane as oxidant.
TONa
TOFb (hꢂ1
)
Leaching (%)
Run I
23
14
15.3
12.7
7.8
4.8
2.9
3
<3
0
0
Run II
Run III
Run IV
0
a
TOF was calculated after 38 min;
TON was calculated after 6 h.
b
conditions of severe leaching (TBHP in water), still these leached
out species are relatively inert for the epoxidation mechanism as
the building of epoxide stops immediately after the hot filtration
of the catalysts; whereas the radical formation of the adduct 4
continues.
3.1.5. EPR and NMR results
In Fig. 7, the results of room temperature EPR measurements on
dry powders of MIL-47 are shown as a function of catalytic reaction
time for MIL-47 in TBHP/decane. The inset displays the EPR spectra
of the MIL-47 powder and the ML-47 powder after 3 h of catalytic
performance. Both spectra appear as broad lines and can be well-
fitted by single Lorentzians. The position, which is practically the
same in the two spectra, is compatible with a signal assignation
+
IV
51
to V
(g = 1.96) [32]. The absence of V hyperfine structure is
+IV
attributed to spin–spin interactions between the close-by V ions
in the MIL-47 structure. The width of the spectra, however, is a
point of significant difference between the spectra, which will be
further addressed in a future investigation.
Fig. 5. XRPD pattern of (a) MIL-47, (b) MIL-47 after catalysis with TBHP in decane as
In the present study, we are mainly interested in the evolution
of the total spectral intensity. This was evaluated by double inte-
gration of the field-modulated spectra. In order to avoid possible
influences of broad background signals, the total intensities were
also calculated from the area of the best-fit Lorentzian to the singly
integrated spectra. Intensities were normalized to that of the cal-
cined MIL-47 powder. As seen in Fig. 7, the two methods lead
essentially to the same conclusions. Within the investigated range
of catalytic reaction times, the spectra experience an intensity drop
oxidant, (c) MIL-47 after catalysis with TBHP in water as oxidant, (d) MIL-47 after
ꢄ
regeneration and (e) terephthalic acid ( is due to the background of the sample
holder at an angle of 32.9°).
Fig. 7. Catalytic reaction time evolution of the intensity of the EPR spectrum for the
MIL-47 in TBHP/decane system. EPR spectra were recorded on dry powders and
intensities were evaluated by single Lorentzian fitting to the singly integrated
spectra and by double integration. All intensities were normalized to that of the
calcined MIL-47. Inset-EPR spectra for calcined MIL-47 (black) and after 3 h of
catalytic reaction in the MIL-47/TBHP/decane system (red). For better comparison
of the line widths of the EPR spectra, the spectrum of calcined MIL-47 divided by 2
in intensity is shown in blue. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)
Fig. 6. XRPD pattern of the original MIL-47 before catalysis and the XRPD pattern of
MIL-47 after the first, second, third and fourth catalytic run ( is due to the
background of the sample holder at an angle of 32.9°).
ꢄ

2
02
K. Leus et al. / Journal of Catalysis 285 (2012) 196–207
by 20–30%, suggesting that approximately this percentage of the
+
IV
+V
V
centers observable with EPR is oxidized to V . The major part
of the intensity decrease is observed in the first few hours (1–3 h).
In the later stages, the EPR intensity seems to stabilize or at least to
change far more slowly.
The structural changes around vanadium sites upon calcination
of the initial framework are confirmed by 51V solid-state NMR. Due
to the paramagnetism of V in the calcined material, a very broad
featureless spectrum was recorded. After catalysis, clear presence
+
V
of V in the species is detected and the spectrum profile is remi-
+
V
niscent of that reported for other materials containing V species.
The spectral profile is consistent with reported complexes contain-
+
V
ing V species [33]. The changes in the environment of vanadium
1
centers after catalysis are reflected in the H–13C CP/MAS spectra
as much broader peaks are observed for the calcined solids (spectra
are shown in the ESI).
3.2. Discussion: theoretical results on reactive pathways
for epoxidation
This computational section aims at: (i) defining the active sites
for catalysis, (ii) introducing a finite cluster representing the active
site within the MIL-47, (iii) proposing plausible reaction pathways
for the epoxidation of cyclohexene, and (iv) elucidating the role of
the solvent (water versus decane) on the product selectivity in the
cyclohexene conversion, as observed in experiment.
Fig. 8. (a) selected cluster (in balls) in the MIL-47 crystal structure, (b) cluster cut
out from the crystal as indicated in (a) is now terminated with hydrogens and is
chosen large enough to represent the active site in MIL-47, (c) schematical
representation of the extended cluster as displayed in (c), (d) bidentate ligand
represented by L. Geometry following the unconstrained model for (c) and (d).
3.2.1. Selection of the cluster model
In an undamaged saturated MIL-47 framework, the coordina-
9
8.5° when the outer atoms are fixed, this angle increases by 12°
tion sphere around the metal ion is completely blocked by the or-
ganic linkers, leaving no free positions available for substrate
chemisorptions. In our earlier theoretical modeling work on the
in the completely unconstrained model. The other geometrical
parameters are not substantially different, even the distance be-
tween the two V-centers remains almost unaltered for both cluster
models (3.80 versus 3.81 Å, compared to 3.42 Å in the MIL-47 crys-
tal). As will be seen later, the deviancy in spatial orientation of the
linkers leads to important changes in the kinetics of the reactions
with non-negligible impact on the preferential pathways.
The cluster model is schematically shown in Fig. 8c, where L
represents a bidentate ligand (Fig. 8d) and hence shows a twofold
coordination with V. Notice that the ligand L in the cluster model
contains also a V atom, but this site is not regarded as catalytically
active during the simulations. The upper vanadium atom as shown
in Fig. 8c is taken as the active catalytic center. Both vanadium
atoms have oxidation state +IV in the starting cluster model.
2
homogeneous VO(acac) catalyst, we found that first vacant coor-
dination sites should be created to activate the catalyst [30]. This
is in accordance with earlier literature data [34]. For the MIL-47,
free coordination positions should also be created by ligand re-
moval or by folding one terephthalic acid ligand away from the
vanadium center or by removing bridging oxygen between the
neighboring vanadium centers from the O–V–O axis. In view of
these structural defects in the periodic structure, we opted to sim-
ulate the reaction pathways in a cluster model. The numerical algo-
rithms to search for transition states are better established in the
computational programs used for the cluster calculations com-
pared with their periodic equivalents. Of course the cluster should
be chosen large enough to represent the real active environment
within the MIL-47 topology. In view of this, we propose the ex-
tended cluster model as displayed in Fig. 8. This model contains
two vanadium atoms, bridged by two terephthalate linkers. On
the vanadyl chain, only two V@O bonds are maintained, and the
continuation to further terephthalate linkers is interrupted and re-
placed by terminating OH-groups. Starting from this cluster, one
can search for the optimized geometry by relaxing all atoms or
by fixing the outer carboxyl oxygen atoms at their crystal positions
to prevent unphysical deformation of the cluster, while relaxing all
other atoms during the geometry optimization. When taking into
account dispersion interactions, the constrained geometry is fa-
vored by about 42 kJ/mol. Moreover, the obtained result is a realis-
tic representation of the actual molecular environment around the
3.2.2. Possible reaction pathways toward formation
of cyclohexene oxide
The reaction cycle starts with an activation step, in which TBHP
coordinates with the vanadium center to form an alkylperoxo spe-
cies 7 (as shown in Scheme 2). The two most probable reaction
routes toward epoxidation starting from this complex are dis-
played in Scheme 2. First, a direct pathway to epoxidation leads
to the formation of cyclohexene oxide 2 and brings the catalyst
to a less active complex 8. Secondly, a radical mechanism is plau-
sible (indicated as ‘‘radical’’ in Scheme 2). In this route, +V vana-
dium complexes 9 are formed by homolytic cleavage of the
peroxy linkage that can then be further activated with TBHP. The
+
V
generated V -activated complexes 10 and 12 can again epoxidize
cyclohexene. Both direct and radical routes are completed by a
regeneration step that closes the cycle.
+
IV
active site in the MIL-47. The MIL-47 (with V ) is one of the MOF
structures that only shows a small degree of flexibility, and thus,
this model should be well suited. Although for completeness, we
also performed all calculations in the model in which all atoms
are relaxed, allowing as such too much conformational flexibility.
In particular, the angle between the two terephthalate planes con-
taining the two terephthalic linkers changes largely during such
computational protocol. While this particular angle amounts to
Apart from the complexes given in Scheme 2, other possible V+
(OOtBu) complexes can be formed, which also lead to cyclohexene
epoxidation. All these plausible pathways have been investigated
in this work, but they turned out to be not competitive with re-
spect to those taken up in Scheme 2. We disregard them from fur-
ther discussion.
V

K. Leus et al. / Journal of Catalysis 285 (2012) 196–207
203
Scheme 2. Two competitive cyclohexene epoxidation pathways for MIL-47 (direct and radical).
3
.2.2.1. Reaction kinetics: direct versus radical pathway. For each
To give an idea on the values of the physisorption energy for the
various species, the complete energy diagram is taken up in Fig. 9
for the direct path 7 ? 8.
reaction path energy barriers, Gibbs-free energies at 323 K and
reaction rates have been computed in the two models for geometry
optimization (unconstrained versus constrained) as outlined be-
fore. The results are given in Table 3. All energetics have been cal-
culated within the B3LYP/6-311 + g(3df,2p)-D3//B3LYP/6-31 + g(d)
level of theory. The B3LYP/6-311 + g(3df,2p) energies were further
refined by including van der Waals interactions at the B3LYP-D3 le-
vel of theory with the Orca software package [35]. For the con-
strained model, we used the partial Hessian vibrational analysis
The physisorption energy of cyclohexene is of about ꢂ28 kJ/mol,
while ꢂ70 kJ/mol for the epoxide. However, the position of the
cyclohexene/epoxide in the adsorbed complex and the way these
molecules are coordinated with the V-center is not always un-
iquely determined, and therefore, the reaction kinetics is per-
formed starting from the gas-phase species in this study.
The direct epoxidation reaction 7 ? 8 is irreversible as can eas-
ily be seen from the energy diagram of Fig. 9. Similar large reaction
energies are observed for the other epoxidation reactions. There-
fore, only the rate constants for the forward reactions are taken
up in Table 3. For convenience of the reader, the backward reaction
barriers and Gibbs-free energies are also tabulated in Table S1.
The transition state of the direct epoxidation reaction 7 ? 8 is
given in Fig. 11a. The cyclohexene molecule is located in the free
spot that extends to a whole hemisphere in front of the two
remaining linkers and the t-butoxy group, stressing once more that
the epoxidation reaction can only take place if some ligands are
removed.
(
PHVA) method for the frequency calculations making use of the
in-house developed software module TAMkin [36,37]. The outer
oxygen atoms, which were used to fix the cluster, were given an
infinite mass [38–42]. All structures for starting geometries were
built using ZEOBUILDER [37,43].
The kinetics for all epoxidation reactions are obtained using the
bimolecular approach in which the two reactants are treated in gas
phase. In this case, bimolecular transition state theory should be ap-
plied as outlined in Supporting Information. More details are also gi-
venbyVanSpeybroecketal. [44]onthe computationalproceduresfor
discriminating between intrinsic and apparent reaction kinetics.

2
04
K. Leus et al. / Journal of Catalysis 285 (2012) 196–207
Table 3
Energy barriers and Gibbs-free energy barriers for the epoxidation reactions (bimolecular) and the radical decomposition (unimolecular). Also the reaction energies are reported,
as well the reaction Gibbs-free energies at 323 K.
Unimolecular
Bimolecular
z
z
unimol
323;intr
ꢂ1
z
z
bimol
323;app
3
D
E
D
G
k
D
E
D
G
k
D
E
0,r
DG
323,r
(kJ/mol)
0
;intr
323;intr
0;app
323;app
ꢂ1 ꢂ1
(kJ/mol)
(
kJ/mol)
(kJ/mol)
(s
)
(kJ/mol)
(kJ/mol)
(m mol
s )
Unconstrained model
Direct
–
–
–
ꢂ2.8
46.3
–
5.71E+03
–
ꢂ157.7
ꢂ174.8
ꢂ34.7
7
? 8
Radical
37.1
32.9
3.22E+07
–
28.5
7
? 9
1
1
0 ? 11
2 ? 11
–
–
–
–
–
–
35.1
26.8
89.6
82.9
5.73Eꢂ04
6.97Eꢂ03
ꢂ219.4
ꢂ191.9
ꢂ213.3
ꢂ187.2
Constrained model
Direct
27.4
38.0
40.3
35.2
2.02E+06
1.37E+07
ꢂ1.0
48.9
–
2.19E+03
–
ꢂ157.3
ꢂ171.2
ꢂ37.3
7
? 8
Radical
–
23.8
7
? 9
1
1
0 ? 11
2 ? 11
–
–
–
–
–
–
29.1
41.9
90.2
107.0
4.69Eꢂ04
8.87Eꢂ07
ꢂ226.5
ꢂ189.8
ꢂ220.3
ꢂ177.8
The constrained and unconstrained models predict nearly the
same reaction barriers and entropic contributions (see Table 3)
as the geometries resulting from the models only differ in the spa-
tial orientation of the two remaining linkers. The energy diagram
of Fig. 9 learns that the reaction is strongly exergonic and thus
irreversible.
The direct route 7 ? 8 is highly favored on basis of the energy
barriers and Gibbs-free energies of activation. The reactions
1
0 ? 11 and 12 ? 11 are higher activated (90.2 kJ/mol and
1
tion
07.0 kJ/mol versus 48.9 kJ/mol for the Gibbs-free energy of activa-
z
DG
following the constrained model Table 3) but are
3
23;app
thermodynamically driven as the product 11 is very stable (Gibbs
reaction-free energies of ꢂ220.3 and ꢂ177.8 kJ/mol). The epoxida-
tion reactions 10 ? 11 and 12 ? 11 are initiated by a radical
decomposition 7 ? 9 thereby bringing the active V-site into an oxi-
dation state +V followed by a reaction of the originally inactive V-
complex 9 with TBHP to an active complex 10 or 12. The radical
decomposition (7 ? 9) is modeled as an unimolecular process with
a reaction barrier of 37–38 kJ/mol or a Gibbs-free energy barrier of
3
3–35 kJ/mol at 323 K, which thus slightly decreases with temper-
ature. This reaction will certainly take place, and in addition, it is
also thermodynamically favored as the reaction becomes more
and more exergonic with temperature due to entropic effects
(
see Fig. 9). Some transformations taken up in Scheme 2 are mod-
eled as equilibration steps and the values are taken up in Table 4.
Starting from complex 9, an equilibration step occurs in which the
+
V
TBHP coordinates with species 9, to form another V species 10.
This step is thermodynamically controlled toward species 10 at
lower temperatures. Species 10 can further transform to species
1
2, in which TBHP covalently bonds to the vanadium center. The
equilibrium of this step lies in the direction of V-cluster 12. From
+
V
species 10 and 12 that are both V species, the epoxidation of
cyclohexene can be initiated.
Overall, the whole reaction cycle starting from 7 to 11 is very
exothermic and also irreversible as the backward reactions have
very high activation barriers (up to 300 kJ/mol (see Table S1)). In
this ‘‘radical’’ pathway, which allows transformation of oxidation
state from +IV to +V for vanadium, one of the epoxidation reactions
10 ? 11 and 12 ? 11 becomes the rate determining step.
They are higher activated but eventually become also important
as this reaction cycle is thermodynamically controlled.
Fig. 9. Energy diagram for the direct epoxidation reaction 7 ? 8, the radical
reaction pathway 7 ? 9 and the subsequent epoxidation reactions 10 ? 11 and
1
2 ? 11. Energies (at 0 K) are given in blue, Gibbs-free energies at 323 K are given
in red. All energies in kJ/mol. The V-cluster appears as an activated complex in the
reactants and as an inactivated complex in the products. In radical reaction 7 ? 9
the oxidation number of vanadium changes from +IV to +V. Both energy (at 0 K) and
Gibbs-free energy (at 323 K) profiles are drawn relative with respect to the first
level in each scheme. Numerical results are obtained using the constrained cluster
model. (For interpretation of the references to color in this figure legend, the reader
is referred to the web version of this article.)
The reaction 12 ? 11 requires some special interest as the re-
sults are more sensitive to the particular model used for the kinet-
ics calculations, i.e., constrained versus unconstrained. In contrast

K. Leus et al. / Journal of Catalysis 285 (2012) 196–207
205
Fig. 10. Energy profiles for the reaction between cyclohexene and the tert-butylperoxy radical. Energies (at 0 K) are given in blue, Gibbs-free energies at 323 K are given in
red. All energies in kJ/mol. (a) Pathway a leading to formation of cyclohexene oxide. (b) Pathway b leading to formation of the adduct 4. Electronic energies at 0 K are indicated
in red. Energies in kJ/mol. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
to the direct pathway 7 ? 8 with a V+IV active complex, steric hin-
py of activation. For T = 323 K, this results to a Gibbs-free energy of
activation of about 80 kJ/mol, which is in excellent agreement with
the theoretical predictions for the pathways 10 ? 11 and 12 ? 11,
which proceed through a vanadium in oxidation state +V.
drance lies now on the origin of the significant increase of the bar-
+
V
rier in the constrained cluster model. In this V complex (Fig. 11d),
the position of the cyclohexene in the transition state is slightly
+
IV
different with that in the V complex (Fig. 11a). In addition, the
free spot for the coordination of cyclohexene is more limited in
the V+ complex, and in the constrained model, one of the linkers
causes a steric repulsive effect on the cyclohexene, inducing an
asymmetric spatial position with respect to the oxygen of the V-
complex. This unphysical situation causes too high barriers. Realis-
tically, the values will lie in between the values of the two models
and overall the conclusions on the most preferable pathway to
epoxidation are not altered. The two geometries of the transition
state are visualized in Fig. 11c and d where we can easily see that
in the unconstrained case, the position of the linker is more relaxed
allowing the cyclohexene to take the most favorable position in the
transition state.
Summarizing, there are two pathways which are competitive
and which form a closed cycle of epoxidation and regeneration
and wherein the peroxide is systematically used to bring the inac-
tive vanadium complex (species 6) into an active complex with
oxidation state +IV (species 7) or an active complex with oxidation
state +V (species 10 and 12). The two epoxidation pathways start
from a complex with another oxidation state for vanadium, but a
switch from one cycle to the other is possible at any time. Once
V
+
V
+IV
V
complexes are formed, a return to the V -cycle is possible
+V
+IV
via the equilibration step 12 ? 6, which we will call the V –V
recycling step for further reference. This step is energetically not
favored ( 0,r = 132–135 kJ/mol in Table 4) but becomes possible
D
E
at temperatures of 323 K due to entropic contributions, which
bring the free energy of reaction to reasonable values 67–77 kJ/
mol. These barriers are of the same order as the barriers for epox-
idation and are thus competitive with these pathways.
Thus far, no solvent effects have been taken into account. In the
experimental part, the catalytic activity of the MIL-47 has been
investigated with TBHP dissolved in water or in decane. One of
the main differences was the much larger initiation time for cyclo-
hexene conversion with decane as solvent. The proposed reaction
scheme may serve to explain some of the experimental findings.
The reaction cycle can only start once free coordination positions
are created, either by folding a ligand away or by removing a
The regeneration reactions with TBHP go smoothly as they are
mostly thermodynamically driven (see Table 4).
It is now interesting to couple our theoretical results to the
available experimental data. The quantitative values of our calcula-
tions can indirectly be validated with the results of a kinetic study
2
on the epoxidation of cyclohexene with VO(acac) . Gould et al. [45]
suggested that vanadium is probably converted to the +V oxidation
state and remains in that state throughout. This is in full agree-
ment with the situation encountered in the epoxidation reactions
1
0 ? 11 and 12 ? 11. The experiment reports values of 53.1 kJ/
mol for the enthalpy of activation and ꢂ82.8 J/K/mol for the entro-

2
06
K. Leus et al. / Journal of Catalysis 285 (2012) 196–207
Fig. 11. Transition states of the three epoxidation reactions: (a) direct pathway 7 ? 8, (b) pathway 10 ? 11, (c) and (d) pathway 12 ? 11. Use of the unconstrained model in
a)–(c); use of the constrained model in (d).
(
Table 4
in case of decane, as TBHP takes also an active role in the activation
period. These findings are compatible with the observation that the
measured experimental diffraction pattern does not change after
reaction with TBHP/decane + cyclohexene. The bulk structure re-
mains better conserved during the reaction cycle, whereas in case
of water, more vanadium/terephthalic acid bonds are broken. The
diffraction pattern of the catalyst with TBHP/water after catalysis
Thermodynamical parameters for the reactions modeled as equilibrium steps.
DE
0,r
represents the reaction energy corrected with zero-point energies and dispersion,
323,r the reaction-free energy at 323 K. The equilibrium constant K323 at 323 K is
obtained after a fitting procedure between 273 and 373 K.
D
G
D
E
0,r
D
G
323,r
K
323
[K323]
(
kJ/mol)
(kJ/mol)
Unconstrained model
(
see Fig. 5c) shows a broadening of the peaks that might be as-
6
8
1
9
1
1
? 7
ꢂ59.8
ꢂ27.5
6.7
ꢂ64.7
ꢂ27.5
132.2
ꢂ40.8
ꢂ7.4
5.0
ꢂ4.8
ꢂ26.1
67.6
3.96E+06
1.55E+01
1.53Eꢂ01
1.60Eꢂ01
1.67E+04
4.41Eꢂ10
–
–
cribed to further structure deterioration.
? 7 (regeneration)
1 ? 12 (regeneration)
? 10
Another feature observed experimentally is the formation of
side products, such as diols, adducts, and cyclohexenones, which
are manifestly present after a short reaction time on a MIL-47 cat-
alyst with a V-loading of 0.42 mmol and with TBHP dissolved in
water. The experiment with TBHP in decane as oxidant does not re-
veal any relevant by-product formation. The formation of the tert-
butyl-2-cyclohexenyl-1-peroxide 4 must be ascribed to radical
pathways between the substrate and the oxidant TBHP and was re-
cently further investigated for unsaturated Co-MOFs [14,46]. Freely
diffusing peroxy radicals can be generated by a variety of mecha-
nisms in which the oxidation state of vanadium changes. It is well
known that radicals are better stabilized in water than in decane
and thus such radical pathways are more prominent present in
such polarisable media. Also the radicals tBuOꢅ and tBuOOꢅ,
appearing in some pathways of Scheme 2, lie on the origin of the
formation of the side products, and in particular that of the adduct
4 as schematically shown in Fig. 10 (pathway b). An enhanced con-
centration of radicals leads to more termination reactions and
hence a larger production of adducts. The increasing concentration
of the of tert-butylperoxy radical tBuOOꢅ requires some attention,
as it is also able to directly epoxidize cyclohexene following the so-
called Twigg mechanism as reported by Van Sickle et al. [47] (path-
way a in Fig. 10). We examine the two competitive pathways. In
the first, a radical addition reaction takes place creating an
intermediate cyclohexenyl radical 13 followed by cleavage of the
–
m
–
3
ꢂ1
mol
0 ? 12
2 ? 6 (V+V–V+IV recycling)
m
ꢂ3
mol
Constrained model
6
8
1
9
1
1
? 7
ꢂ60.3
ꢂ27.9
4.6
ꢂ52.7
ꢂ36.7
134.6
ꢂ42.3
ꢂ10.9
ꢂ4.3
6.1
7.02E+06
5.73E+01
4.94E+00
2.74Eꢂ03
7.54E+06
1.21Eꢂ11
–
–
? 7 (regeneration)
1 ? 12 (regeneration)
? 10
–
m
–
3
ꢂ1
mol
0 ? 12
ꢂ42.5
2 ? 6 (V+V–V+IV recycling)
77.2
m
ꢂ3
mol
terephthalic acid from vanadium. The structure 6 in scheme 2 can
be created by cleaving a terephthalate–vanadium bond and replac-
ing the ligand by a hydroxyl group. Such hydroxyl groups can eas-
ily be introduced in a water-based medium and thus water helps in
creating free coordination sites, enhancing the initial rate of cyclo-
hexene conversion. In case of decane as solvent, such sites of type 6
are not formed as this solvent has no OH-group present. In this
case, the activation should be induced by TBHP itself. This can
either lead to the cleavage of a vanadium bond and an oxygen of
the carboxylate group to form complex 7 immediately but also
other species such as VO(L)(OtBu) could be formed, which can by
ligand exchange with TBHP form the active species 7. Therefore,
one can anticipate that the initiation time for epoxidation is larger

K. Leus et al. / Journal of Catalysis 285 (2012) 196–207
[2] H. Li, M. Eddaoudi, M. O’Keeffe, O.M. Yaghi, Nature 402 (1999) 276.
207
tert-butylperoxy radical and formation of cyclohexene oxide. The
[
3] N.L. Rosi, J. Eckert, M. Eddaoudi, D.T. Vodak, J. Kim, M. O’Keeffe, O.M. Yaghi,
Science 300 (2003) 1127.
energy profile is displayed in Fig. 10 (pathway a). The radical addi-
tion reaction shows a high free-energy barrier, while the reaction is
strongly endergonic, indicating that the backward reaction (disso-
ciation) goes much faster than the forward reaction (association).
In pathway b, a hydrogen abstraction occurs from the allylic posi-
tion of cyclohexene forming the 3-cyclohexenyl radical, which can
rapidly recombine with another tert-butylperoxy radical yielding
the adduct 4. The hydrogen abstraction also requires a high energy
barrier (only 8 kJ/mol lower than in pathway a). The subsequent
radical recombination reaction goes quite quickly since the reac-
tion energy is strongly negative and heavily thermodynamically
driven. These theoretical findings are in complete agreement with
the suggestions made in a recent experimental work of Tonigold
et al. [46] where pathway b has been proposed as the major path-
way. Concluding, without excluding the direct epoxidation of
cyclohexene by the tert-butylperoxy radical tBuOOꢅ, the radical
pathway will favor the adduct formation.
[
4] L.J. Murray, M. Dinca, J.R. Long, Chem. Soc. Rev. 38 (2009) 1294.
[5] L. Pan, B. Parker, X.Y. Huang, D.H. Olson, J. Lee, J. Li, J. Am. Chem. Soc. 128
2006) 4180.
6] J. Perles, M. Iglesias, M.A. Martin-Luengo, M.A. Monge, C. Ruiz-Valero, N.
Snejko, Chem. Mater. 17 (2005) 5837.
(
[
[7] D. Farrusseng, S. Aguado, C. Pinel, Angew. Chem. Int. Ed. 48 (2009) 7502.
8] A. Corma, H. Garcia, F.X.L. Xamena, Chem. Rev. 110 (2010) 4606.
9] P. Van der Voort, M.G. White, E.F. Vansant, Langmuir 14 (1998) 106.
[
[
[
10] G. Catana, R.R. Rao, B.M. Weckhuysen, P. Van Der Voort, E. Vansant, R.A.
Schoonheydt, J. Phys. Chem. B 102 (1998) 8005.
11] P. Van Der Voort, M. Baltes, E.F. Vansant, Catal. Today 68 (2001) 119.
12] A. Henschel, K. Gedrich, R. Kraehnert, S. Kaskel, Chem. Commun. (2008) 4192.
13] N.V. Maksimchuk, K.A. Kovalenko, S.S. Arzumanov, Y.A. Chesalov, M.S.
Melgunov, A.G. Stepanov, V.P. Fedin, O.A. Kholdeeva, Inorg. Chem. 49 (2010)
[
[
[
2920.
[
14] M. Tonigold, Y. Lu, B. Bredenkötter, B. Rieger, S. Bahnmüller, J. Hitzbleck, G.
Langstein, D. Volkmer, Angew. Chem. Int. Ed. 48 (2009) 7546.
[15] N.V. Maksimchuk, M.N. Timofeeva, M.S. Melgunov, A.N. Shmakov, Y.A.
Chesalov, D.N. Dybtsev, V.P. Fedin, O.A. Kholdeeva, J. Catal. 257 (2008) 315.
16] H. Garcia, A. Dhakshinamoorthy, M. Alvaro, Chem. – A Eur. J. 17 (2011) 6256–
262.
17] H. Garcia, A. Dhakshinamoorthy, M. Alvaro, Chemcatchem 2 (2010) 1438–
443.
18] F.X.L.I. Xamena, O. Casanova, R.G. Tailleur, H. Garcia, A. Corma, J. Catal. 255
2008) 220–227.
[19] J.S. Seo, D. Whang, H. Lee, S.I. Jun, J. Oh, Y.J. Jeon, K. Kim, Nature 404 (2000)
82.
[
[
[
6
1
4
. Conclusions
(
This study reveals that V-MIL-47 can be a highly selective cata-
9
lyst in the epoxidation of cyclohexene using TBHP as the oxidant.
Water should be avoided as the solvent for the peroxide, as it en-
hances strongly the leaching of the V-centers and as it accelerates
an unwanted radical side reaction, forming an adduct between the
peroxide and cyclohexene.
[
[
20] L.Q. Ma, C. Abney, W.B. Lin, Chem. Soc. Rev. 38 (2009) 1248.
21] C.D. Wu, A. Hu, L. Zhang, W.B. Lin, J. Am. Chem. Soc. 127 (2005) 8940.
[22] L.L. Wen, F. Wang, J. Feng, K.L. Lv, C.G. Wang, D.F. Li, Cryst. Growth Des. 9
(2009) 3581.
23] P. Phuengphai, S. Youngme, P. Gamez, J. Reedijk, Dalton T. 39 (2010) 7936.
24] S. Neogi, M.K. Sharma, P.K. Bharadwaj, J. Mol. Catal. A – Chem. 299 (2009) 1.
25] S. Horike, M. Dinca, K. Tamaki, J.R. Long, J. Am. Chem. Soc. 130 (2008) 5854.
[26] F. Gandara, E.G. Puebla, M. Iglesias, D.M. Proserpio, N. Snejko, M.A. Monge,
Chem. Mater. 21 (2009) 655.
27] J. Juan-Alcaniz, E.V. Ramos Fernandez, U. Lafont, J. Gascon, F. Kapteijn, J. Catal.
69 (2010) 229.
[28] M.J. Ingleson, J.P. Barrio, J.B. Guilbaud, Y.Z. Khimyak, M.J. Rosseinsky, Chem.
Commun. (2008) 2680.
29] A. Dhakshinamoorthy, M. Alvaro, H. Garcia, Catal. Sci. Technol. 1 (2011) 856–
67.
[30] K. Leus, I. Muylaert, M. Vandichel, G.B. Marin, M. Waroquier, V. Van
Speybroeck, P. Van der Voort, Chem. Commun. 46 (2010) 5085.
31] K. Barthelet, J. Marrot, D. Riou, G. Ferey, Angew. Chem. Int. Ed. 41 (2001) 281.
32] A. Brückner, Top Catal. 38 (1–3) (2006) 133.
[
[
[
When decane is used to dissolve the peroxide, the MIL-47 is a
highly selective catalyst toward the epoxide, especially in the first
linear regime of conversion. The leaching of V-centers is negligible
in that case and the structural integrity of the MOF is preserved
during successive runs. Computational studies show that several
catalytic pathways co-exist and compete with each other, but
every catalytic cycle starts with the breaking of two V-terephthalic
bonds to coordinate with the peroxide. EPR and NMR studies con-
[
2
[
8
+
IV
+V
firm that approximately 20% of the V sites are oxidized to V in
the first minutes of the catalytic reaction and remain relatively
constant afterward.
[
[
[33] A. Fenn, M. Wächtler, T. Gutmann, H. Breitzke, A. Buchholz, I. Lippold, W. Plass,
G. Buntkowsky, Solid State Nucl. Mag. 36 (2009) 192.
[
34] (a) E.P. Talsi, V.D. Chinakov, V.P. Babenko, K.I. Zamaraev, J. Mol. Catal. 81
Acknowledgments
(
(
1993) 235;
b) H. Mimoun, M. Mignard, P. Brechot, L. Saussine, J. Am. Chem. Soc. 108
K.L. is grateful to the Long Term Structural Methusalem Grant
Nr. 01M00409 Funding by the Flemish Government. M.V. thanks
the research board of Ghent University (BOF). I.M. thanks the
Institute for the Promotion of Innovation through Science and
Technology in Flanders (IWT Vlaanderen). Furthermore, this
research is co-funded by the Ghent University, GOA Grant Nr.
(1986) 3711.
[
[
35] ORCA. 2.6.35ed. <http://www.thch.uni-bonn.de/tc/orca/>.
36] A. Ghysels, T. Verstraelen, K. Hemelsoet, M. Waroquier, V. Van Speybroeck, J.
Chem. Inf. Model. 50 (2010) 1736.
[37] CMM Code. <http://molmod.ugent.be/code/wiki>.
38] A. Ghysels, D. Van Neck, V. Van Speybroeck, T. Verstraelen, M. Waroquier, J.
Chem. Phys. 126 (2007) 224102.
[
[
39] A. Ghysels, D Van Neck, M. Waroquier, J. Chem. Phys. 127 (2007) 164108.
0
1G00710, BELSPO in the frame of IAP 6/27 and the European
[40] A. Ghysels, V. Van Speybroeck, T. Verstraelen, D. Van Neck, M. Waroquier, J.
Chem. Theory Comput. 4 (2008) 614.
Research Council (FP7(2007-2013) ERC Grant Nr. 240483).
Computational resources (Stevin Supercomputer Infrastructure)
and services were provided by Ghent University.
[
41] A. Ghysels, V. Van Speybroeck, E. Pauwels, D. Van Neck, B.R. Brooks, M.
Waroquier, J. Chem. Theory Comput. 5 (2009) 1203.
[42] A. Ghysels, V. Van Speybroeck, E. Pauwels, S. Catak, B.R. Brooks, D. Van Neck,
M. Waroquier, J. Comput. Chem. 31 (2010) 994.
[
43] T. Verstraelen, V. Van Speybroeck, M. Waroquier, J. Chem. Inf. Model. 48 (2008)
1530.
44] V. Van Speybroeck, J. Van der Mynsbrugge, M. Vandichel, K. Hemelsoet, D.
Lesthaeghe, A. Ghysels, G.B. Marin, M. Waroquier, J. Am. Chem. Soc. 133 (4)
Appendix A. Supplementary material
[
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2011.09.014.
(
2011) 888.
[
[
45] E.S. Gould, R.R. Hiatt, K.C. Irwin, J. Am. Chem. Soc. 90 (1980) 4573.
46] M. Tonigold, Y. Lu, A. Mavrandonakis, A. Puls, R. Staudt, J. Mollmer, J. Sauer, D.
Volkmer, Chem. Eur. J. 17 (2011) 8671.
References
[
47] D.E. Van Sickle, F.R. Mayo, R.M. Arluck, J. Am. Chem. Soc. 87 (21) (1965) 4824.
[
1] B.F. Hoskins, R.J. Robson, J. Am. Chem. Soc. 112 (1990) 1546.