Aerobic oxidation of cycloalkenes catalyzed by iron metal organic framework containing N-hydroxyphthalimide

Article abstract of DOI:10.1016/j.jcat.2012.02.015

Iron metal organic framework [Fe(BTC)] loaded with N-hydroxyphthalimide (NHPI) promotes the aerobic oxidation of (cyclo)alkenes to give variable percentages of allylic oxidation products and the corresponding epoxide, dependidng on the nature of the substrate. In the case of cyclopentene and cyclohexene, aerobic oxidation catalyzed by NHPI/Fe(BTC) renders their corresponding unsaturated cyclic alcohol and ketone with 97% selectivity in 5 h at 6% and 12% conversion, respectively. Under the same experimental conditions, cyclooctene exhibited 95% selectivity toward the formation of cyclooctene oxide with 2% of cyclooctenol/one at 4 h. Cycloheptene as susbstrate exhibits an intermediate behavior, and the aerobic oxidation catalyzed by NHPI/Fe(BTC) leads to the formation of cycloheptenol/cycloheptenone with 77% selectivity, accompanied by 23% of cycloheptene oxide at 4 h. Further experiments with non-symmetric olefins exhibited also a mixture of products including epoxides and allyic products. A mechanism to explain these experimental results has been proposed.

Full text of DOI:10.1016/j.jcat.2012.02.015

Journal of Catalysis 289 (2012) 259–265
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Aerobic oxidation of cycloalkenes catalyzed by iron metal organic framework
containing N-hydroxyphthalimide
⇑
Amarajothi Dhakshinamoorthy, Mercedes Alvaro, Hermenegildo Garcia
Instituto Universitario de Tecnología Química CSIC-UPV and Departamento de Química, Universidad Politécnica de Valencia, Av. De los Naranjos s/n, 46022 Valencia, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
Iron metal organic framework [Fe(BTC)] loaded with N-hydroxyphthalimide (NHPI) promotes the aerobic
oxidation of (cyclo)alkenes to give variable percentages of allylic oxidation products and the correspond-
ing epoxide, dependidng on the nature of the substrate. In the case of cyclopentene and cyclohexene, aer-
obic oxidation catalyzed by NHPI/Fe(BTC) renders their corresponding unsaturated cyclic alcohol and
ketone with 97% selectivity in 5 h at 6% and 12% conversion, respectively. Under the same experimental
conditions, cyclooctene exhibited 95% selectivity toward the formation of cyclooctene oxide with 2% of
cyclooctenol/one at 4 h. Cycloheptene as susbstrate exhibits an intermediate behavior, and the aerobic
oxidation catalyzed by NHPI/Fe(BTC) leads to the formation of cycloheptenol/cycloheptenone with 77%
selectivity, accompanied by 23% of cycloheptene oxide at 4 h. Further experiments with non-symmetric
olefins exhibited also a mixture of products including epoxides and allyic products. A mechanism to
explain these experimental results has been proposed.
Received 27 November 2011
Revised 19 February 2012
Accepted 23 February 2012
Available online 30 March 2012
Keywords:
Aerobic oxidation
Epoxidation of cycloalkenes
Metal organic frameworks as solid
Catalysts
N-hydroxyphthalimide
Ó 2012 Elsevier Inc. All rights reserved.
1
. Introduction
Oxygen is the ideal oxidizing agent from the environmental
Metal organic frameworks (MOFs), also called as porous coordi-
nation polymers (PCPs) [8], are attracting much current interest as
heterogeneous catalysts because they combine a large surface area
and porosity with a high percentage of active sites [9–12]. While
MOFs are micro-/mesoporous crystalline solids that have many
similarities in their use as solid catalysts with zeolites and related
periodic structured silicas and related materials, MOFs have the
advantage of a much larger percentage of transition metal content
that are in crystallographic well-defined positions with strictly reg-
ular surroundings. Thus, MOFs can be considered as solid catalysts
that contain a single type of site, this being reflected in the possibil-
ity to achieve the largest possible turn over numbers. In addition,
MOFs are very versatile in the structure of organic linker and nature
of the transition metal that can be present in their composition.
In the present case, the MOF that has been used, Fe(BTC) (iron
content 25%) (BTC: 1,3,5-benzenetricarboxylate) contains Fe(III) as
metal and tripodal BTC as ligand. Although the crystal structure of
a Fe(BTC) corresponding to MIL-100 has been determined by
X-ray diffraction methods, the exact crystal structure of the com-
mercial Fe(BTC) sample used in the present work provided by BASF
is still unresolved. Commercial BASF Fe(BTC) exhibits XRD with
broad diffraction peaks frequently found in PCPs rather than in
point of view since it is free, forms water as the only by-product,
and contains the maximum oxidizing weight percentage [1]. While
there has been in recent years a considerable progress on the aer-
obic oxidation of alcohols, particularly using noble metal nanopar-
ticles as catalysts [2], there are still many oxygenation reactions in
which a general catalyst has yet to be found. Although ethylene
oxide is produced by the reaction of ethylene and oxygen using a
silver catalyst [3], an analogous process for aerobic epoxidation
of other (cyclo)alkenes has not been yet possible. Limited progress
in this field has been achieved by using supported gold catalysts in
combination with alkanes that can form in situ hydroperoxides
that then will act as oxygen transfer intermediates in the formation
of the epoxide [4]. Also recently, Iglesia and Haruta have reported
the epoxidation of propene using supported gold catalysts but the
conversions were less than 1% in order to keep high the selectivity
toward the target epoxide [5,6]. The target for propene epoxidation
will be to achieve conversions higher than 5% with very high selec-
tivity toward epoxide. In general, selective epoxidation of alkenes
using molecular oxygen instead of peracids or peroxides still re-
mains a challenge in catalysis and green chemistry. Although epox-
idation of alkenes having allylic positions is the main challenge,
allylic oxidations [7] are also of industrial relevance in the fra-
grance industry and the synthesis of fine chemicals.
2
À1
MOFs, a high surface area (840 m g ) and average pore diameter
(21.7 Å) which are the characteristic features of MOFs, but the crys-
tal cell is not yet known.
We have shown that commercially available BASF Fe(BTC) could
be conveniently used as heterogeneous catalysts in various
reactions [13–17]. In addition, it has been demonstrated that
MIL-100(Fe) exhibited high selectivity and conversion in p-xylene
⇑
Corresponding author. Fax: +34 9638 7807.
E-mail address: hgarcia@qim.upv.es (H. Garcia).
0
021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2012.02.015

2
60
A. Dhakshinamoorthy et al. / Journal of Catalysis 289 (2012) 259–265
acylation with benzoyl chloride [18]. Oxidation of cyclohexene to
cyclohexene oxide, 2-cyclohexenone, and 2-cyclohexenol has been
reported using Co [19] and V [20] containing MOFs as catalysts in
the presence of tert-butylhydroperoxide (TBHP) as oxidant. In con-
trast, allylic oxidation of cyclohexene with molecular oxygen has
also been reported using copper-based MOFs [21]. Recently, a
new heterogeneous catalyst for epoxidation of cyclohexene, cyclo-
octene, and styrene was reported using IRMOF-3 post-functional-
external purging method. Blank experiment was performed under
identical conditions but under nitrogen atmosphere. Nitrogen
adsorption isotherms were measured at 77 K on a Micromeritics
ASAP 2000 volumetric adsorption analyser. Before the measure-
ments, the samples were out gassed for 4 h at 473 K. The specific
surface area was calculated by applying the BET model to the nitro-
gen adsorption data. Chemical composition was determined by
atomic absorption in a Varian SpectrAA-10 Plus and elemental anal-
ysis in a Fisons EA1108CHN-S. IR spectra of Fe(BTC) and NHPI/
Fe(BTC) were recorded within KBr pellets in a Nicolet 710 FT spec-
trophotometer. FT-Raman spectra of Fe(BTC) and NHPI/Fe(BTC)
were recorded with a Bio-Rad FT-Raman II spectrophotometer.
Nd:YAG laser was used for excitation along with a germanium
detector cooled at liquid nitrogen temperature using high-quality
quartz tubes as cells.
ized with
a manganese(II) acetylacetonate complex using
molecular oxygen under mild reaction conditions [22].
Although MOFs have been used as heterogeneous catalysts for
many organic reactions, and in particular epoxidation of olefins
using TBHP as oxidant, developing a protocol where molecular
oxygen is used as oxidant for olefin epoxidation is a challenging
and attractive methodology. In the present work, we report that
an iron containing MOF including N-hydroxyphthalimide (NHPI
2
0 wt.% loading) as cocatalyst is able to affect the aerobic allylic
2.3. Experimental procedure
oxidation of cyclopentene and cyclohexene and, more remarkably,
the epoxidation of cyclooctene at conversions higher than 10%
with almost complete selectivity toward the corresponding epox-
ide. Cycloheptene exhibits an intermediate behavior giving mix-
tures of allylic oxidation and epoxidation. Scheme 1 summarizes
the catalytic activity observed for the aerobic oxidation of cycloalk-
enes promoted by NHPI containing Fe(BTC). Besides activity and
selectivity, one additional advantage of our catalyst is that it con-
tains a base metal rather than precious metals.
In a typical reaction, 0.5 mL of cycloalkene was placed into the
flask (10 mL capacity) together with the appropriate amount of cat-
alyst (75 mg). These mixtures were stirred at 100 °C under molecu-
lar oxygen atmosphere for the required time. For cyclopentene and
cyclohexene, the reactions were carried out in closed reinforced
glass vessels that can stand pressure. After the reaction time, the
catalyst was filtered and extracted with acetonitrile, and the reac-
tion mixture analyzed by GC. Nitrobenzene was always used as an
external standard for the determination of conversion level. The
products were identified by GC and GC–MS. A preparative scale
experiment using 5 mL of cyclooctene was carried out under the
same conditions. When the conversion was 5%, the reaction mixture
was filtered to remove the catalysts, concentrated under vacuum,
and submitted to flash chromatography to isolate cyclooctene oxide
in 4.5% yield. The reusability of the catalyst was tested for the oxida-
tion of cyclooctene. After the reaction time, the reaction mixture
was diluted with acetonitrile, filtered, dried at 70 °C for 2 h, and re-
used directly without further treatment for a subsequent run with
fresh cyclooctene. The mass balance in all the runs was higher than
2
. Experimental methods
2.1. Materials
Basolite F300 [Fe(BTC)] and other MOFs used in the present
study were purchased from Sigma Aldrich and used as received.
NHPI, various substrates and solvents used in the present study
were purchased from Sigma Aldrich.
2.2. Characterization techniques
9
0%. The absence of active species leached from the solid to the solu-
tion was assured by the Sheldon’s hot filtration test. In this case, the
reaction was initiated under the typical oxidation conditions indi-
cated above and the solid catalyst filtered at the reaction tempera-
ture at a cyclooctene conversion of 4%. Then, the clear solution
was allowed to react further at 100 °C in the absence of any solid.
No further cyclooctene conversion was observed after filtration of
the catalyst.
The percentage conversion, purity, and relative yields of the final
products were determined using Hewlett Packard 5890 series II gas
chromatograph with FID detector and high purity helium as carrier
gas. The products were identified by comparing their retention
times in GC with authentic samples and with GC–MS Hewlett Pack-
ard 6890 series. Powder XRD diffraction patterns were recorded in
the refraction mode using a Philips X’Pert diffractometer using the
A series of three control experiments was performed using
Cu K
a radiation (k = 1.54178 Å) as the incident beam, PW3050/60
6
0 mg of Fe(BTC) with 12 mg of NHPI or 145 mg of iron nitrate nona-
(
2h) as Goniometer, PW 1774 spinner as sample stage, PW 3011
hydrate with 30 mg of NHPI or 115 mg of Fe(acac) with 23 mg of
3
as detector, incident mask fixed with 10 mm. PW3123/10 for Cu
was used as a monochromator. PW3373/00 Cu LFF was used as X-
ray tube with power scanning of 45 kV and 40 mA current. The sam-
ple powder was loaded into a holder and levelled with a glass slide
before mounting it on the sample chamber. The specimens were
NHPI for the unsupported NHPI-catalyzed aerobic oxidation of
cyclooctene at 100 °C adopting the above-mentioned reaction
procedure.
2.4. Experimental procedure for silylation
À1
scanned between 2° and 70° with the scan rate of 0.02 s . ESR spec-
tra were recorded using a Bruker EMX, with the typical settings: fre-
quency 9.80 GHz, sweep width 30.6 G, time constant 80 ms,
modulation frequency 100 kHz, modulation width 0.2G, microwave
power 200 mW. Benzonitrile (2 mL) containing 75 mg of NHPI/
Fe(BTC) was stirred under oxygen atmosphere at 120 °C for 2 h.
The air in the ESR tube was replaced by nitrogen gas by means of
The reaction was performed as indicated in the manuscript un-
der the same reaction conditions. After 4 h, unreacted cyclooctene
was removed, and 1 mL of N,O-(trimethylsilyl)trifluoroacetamide
containing trimethylchlorosilane was added and stirred at 80 °C
for 8 h. This reaction mixture was analyzed by GC–MS. A control
using commercial octanedioic acid showed that over 95% can be
silylated with this procedure.
O
NHPI/Fe(BTC)
HO
O
+
+
2.5. Preparation of NHPI/Fe(BTC)
n
O2
n
n
n
n= 0,1,2,3
NHPI (200 mg, 1.22 mmol) was dissolved in 7 mL of dichloro-
Scheme 1. Aerobic oxidation of cycloalkenes catalyzed by NHPI/Fe(BTC).
methane and 3 mL of methanol. To this solution, 1 g of [Fe(BTC)]

A. Dhakshinamoorthy et al. / Journal of Catalysis 289 (2012) 259–265
261
(
4
3.8 mmol) was added and the resulting suspension was heated at
0 °C for 4 h. After the required time, the mixture was cooled and
tricarboxylate linker in Fe(BTC) precluded to observe any major
spectroscopic changes after NHPI inclusion.
the solvent was removed by rotary evaporation. The resulting
NHPI/Fe(BTC) was dried at 80 °C for 2 h and used. The NHPI/Fe wt.
and molar ratios were 1/1.06 and 1/3.1, respectively. As an average,
Considering the molecular size of NHPI (around 6.5 Å of kinetic
diameter) and pore dimension presumed for Fe(BTC), it is reason-
able to assume that the NHPI will be accommodated inside Fe(BTC)
intracrystalline space. One evidence in supporting this internal
location of NHPI inside the MOF is the significant reduction of
BET surface area of Fe(BTC) after incorporation of NHPI that goes
the number of NHPI molecules per surface area unit is
À2
1
.4
lmol m
.
2
from 840 to 24 m /g (see experimental section). The paramagnetic
3
. Results and discussion
property of Fe(BTC) precludes the study of NHPI/Fe(BTC) by both in
1
solid and in liquid state H NMR spectroscopy. The complete lack of
3.1. Characterization of Fe(BTC)
solubility of NHPI in cycloalkenes makes unlikely the migration of
NHPI outside the pores under the reaction conditions. To provide
further evidence that no NHPI migrates to the solution due to its
remarkable insolubility, we performed a control in which NHPI/
Fe(BTC) was extracted with cyclooctene at 100 °C for 5 h and then
the solid filtered while hot. This presumed extracted NHPI/Fe(BTC)
was used as catalyst for cyclooctene oxidation observing the same
performance as the one that was not submitted to solid liquid
extraction.
Surface area, pore volume, and pore size distribution of the solid
samples (200 mg) were calculated by the BET method by carrying
out liquid nitrogen adsorption experiment at 77 K in a Micromeri-
tics Flowsorb apparatus. Prior to this experiment, the solid was
treated at 200 °C for 2 h to desorb the water molecules. This treat-
ment is common for most of the gas adsorption studies in MOFs
23] due to the thermal unstability of the organic linker, and the
fact that the hydration of the adsorbed water can be performed
at mild temperatures.
[
2
3.3. Aerobic oxidation of cyclooctene
BET surface area (m /g)
840
3
Total pore volume
Average pore diameter
0.457 cm /g
21.7 Å
The main objective of the present study is to oxidize cyclic, ali-
phatic olefins to their epoxides with high selectivity in the pres-
ence of oxygen. To achieve this target, we have made use of
NHPI/Fe(BTC) (1:5) as catalyst whose catalytic activity in styrene
oxidation has already been reported elsewhere [24]. The reactivity
of styrene versus aerobic oxidation is different to that of aliphatic
Chemical composition was determined by atomic absorption in
a Varian SpectrAA-10 Plus and elemental analysis in a Fisons
EA1108CHN-S. The iron content in Fe(BTC) was measured to be
(
cyclo)alkenes due to the reactivity of the benzylic positions, this
explaining the formation of benzaldehyde as the major product un-
der various reaction conditions.
2
5%.
The EPR spectra shown in Fig. 1 demonstrate the existence of
NHPI in combination with Co salts [25] has been extensively
used as homogeneous catalyst for the functionalization of hydro-
carbons [26,27] in the presence of acetic acid as solvent. In these
systems, the role of Co salts is to generate the first N-oxyl radical
from NHPI by electron transfer. This radical generated from NHPI
initiates a radical chain mechanism with alkene and oxygen lead-
ing to the products. It would be anticipated that Fe(III) present in
Fe(BTC) can play the role of Co salts as promoter of NHPI, leading
to the formation of NHPI-derived radical that will initiate hydro-
carbon oxidations. Here, we wish to report the results obtained
with NHPI/Fe(BTC) as catalyst for the aerobic oxidation of olefins
in the absence of any solvent, particularly acetic acid that is not
compatible with Fe(BTC).
Fe(III). The values for parameter G corresponding to the main sig-
nal on the spectra are in the range of 2.01. In order to quantify the
signal from Fe(BTC) spectra, we have used Cu(SO
4
)
2
Á12H
2
O as ref-
erence material. The spectra of known weights of both solids were
recorded under the same conditions and the area of the signals
determined. These values have led to estimate the percentage of
Fe(III) in Fe(BTC) as 28%.
3.2. Characterization of NHPI/Fe(BTC)
N-hydroxyphthalimide was incorporated onto Fe(BTC) by
adsorption from a dichloromethane solution as described in the
experimental section, and the resulting NHPI/Fe(BTC) material
was characterized by IR and Raman spectra. As it can be seen in
Figs. 2 and 3, the presence of NHPI at the small loading used in this
work and the high intensity of the IR and Raman bands due to the
Initially, epoxidation of cyclooctene resulted in no conversion
with oxygen in the absence or presence of Fe(BTC) as catalyst.
However, NHPI adsorbed onto Fe(BTC) [NHPI/Fe(BTC)] resulted in
an active heterogeneous catalyst in the presence of oxygen at
80000
60000
40000
20000
0
2
00
150
00
1
(
b)
-20000
-40000
-60000
-80000
5
0
(a)
0
750
1500 2250 3000 3750
-1
0
1000 2000 3000 4000 5000 6000 7000
Gauss
Wavenumber/cm
Fig. 2. FTIR spectra of (a) Fe(BTC) and (b) NHPI/Fe(BTC) recorded at room
Fig. 1. EPR spectrum recorded for Fe(BTC) in solid state at room temperature.
temperature.

2
62
A. Dhakshinamoorthy et al. / Journal of Catalysis 289 (2012) 259–265
on the general inverse relationship between alkene conversion and
3
2
1
000
000
000
0
(b)
epoxide selectivity, the selectivity of epoxide decreased gradually
as the conversion increases. The occurrence of cyclooctene allylic
oxidation resulting in the formation 2-cyclooctenol and 2-cyclooc-
tenone is observed at higher conversions (Table 1). Increasing the
temperature to 110 and 120 °C increases cyclooctene conversion
to 33% and 48% with a selectivity toward epoxide still over 80%.
In order to confirm the analytical data obtained by GC analysis,
we performed the aerobic oxidation of cyclooctene (5 mL) and pro-
ceeded to the isolation of cyclooctene oxide. When the cyclooctene
conversion was 5%, this large-scale reaction allowed us to isolate
(a)
0
1000 2000 3000 4000 5000
-1
Raman Shift/cm
1
cyclooctene oxide with high purity as confirmed by H NMR with
Fig. 3. Raman spectra of (a) Fe(BTC) and NHPI/Fe(BTC).
4.5% yield. These results are outstanding and without equivalent
for any aerobic oxidation of neat cycloalkenes in the liquid phase.
For instance, van Sickle and coworkers reported that treatment of
cyclooctene with oxygen forms the corresponding cyclooctene
oxide with 39.9% selectivity [28]. Also, recent reports from Her-
1
00 °C, and 3.6% of cyclooctene was converted to cyclooctene oxide
with 99% selectivity in 2 h in the absence of any cosolvent. Table 1
lists the results obtained for the epoxidation of cyclooctene by oxy-
gen using NHPI/Fe(BTC) as solid catalysts.
mans’ group about autooxidation of a-pinene have led to complex
mixtures in which the epoxide is accompanied by many other
products [29,30]. To evaluate the performance of NHPI/Fe(BTC) as
catalyst for aerobic oxidation of cyclooctene, turnover numbers
Increasing the time, gradually increases cyclooctene conversion
to reach in 6 h 13% conversion with 90% selectivity toward the for-
mation of cyclooctene oxide (Fig. 1). The formation of a very low
amount of hydroperoxides and peroxides in the reaction mixture
was confirmed by adding triphenylphosphine to the reaction mix-
ture and observing the formation of very little amount of triphen-
ylphosphine oxide as well as a slight increase in the yield of ol/one.
This test indicated that the combined selectivity of hydroperoxides
and peroxides is 0.27% at 9% conversion of cyclooctene. Addition-
ally, the formation of trace amounts (<0.2%) of octanedioic acid de-
rived from the oxidative cleavage of C@C double bond was
determined by silylation of the residue after concentration of the
reaction mixture with N,O-(trimethylsilyl)trifluoroacetamide con-
taining trimethylchlorosilane and observing only detectable
amounts of corresponding octanedioc acid disilyl derivative. Con-
trol experiments performing the same derivatisation protocol with
authentic samples of octanedioic acid showed that complete silyla-
tion occurs under our experimental conditions. As expected based
(
TONs) for epoxide formation both with respect to NHPI and Fe
were calculated. The data (see footnote ‘‘b’’ in Table 1) indicated
that about four molecules of epoxides are formed per NHPI mole-
cule. To put this TON into context in a related precedent using a
Co(II)-containing MOF as catalyst and t-butylhydropeoxide (TBHP)
as oxidant reagent, the estimated TON for epoxide formation was
around 5 [19] that is similar as the one achieved here using oxygen
as oxidant. Similarly, V-MIL-47 exhibits for cyclohexene oxidation
À1
using TBHP as reagent a turnover frequency of 43 h [20], while
NHPI/Fe(BTC) inspite of the induction period (see Fig. 4) and the
À1
use of oxygen exhibits a turnover frequency of 5 h
.
Based on the presented evidence (reduction of BET surface area
upon incorporation of NHPI and the lack of solubility of NHPI in
cycloalkenes), we believe that the loaded NHPI on Fe(BTC) should
be located in and around of Fe(BTC) crystals, and it will not under-
go extraction or relocation under the reaction conditions.
In order to determine the prevailing advantages of NHPI/Fe(BTC)
in cyclooctene epoxidation, a control experiment was performed
using a physical mixture of Fe(BTC) and NHPI in 5:1 ratio as solid cat-
alyst, whereby a 18% conversion with 72% of epoxide selectivity
accompanied by 22% enol/enone was measured at 4 h. The observed
results clearly show that NHPI/Fe(BTC) exhibited higher selectivity
of epoxide compared with the NHPI and Fe(BTC) mechanical mix-
ture. This better performance of NHPI/Fe(BTC) over its mixture could
be attributed to the following two reasons: (i) confinement of the
radical intermediates generated during the course of the reaction in-
side Fe(BTC) pores leading to preferential in-cage recombination
resulting in higher epoxide selectivity as compared to the situation
Table 1
a
Aerobic oxidation of cyclooctene using NHPI/Fe(BTC) as heterogeneous catalyst.
Run Catalyst
Time (h) Con. (%) Selectivity (%)
Epoxide Enol + enone
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
–
2
2
5
1
2
3
4
5
4
4
5
7
5
5
4
4
4
–
–
–
–
–
–
–
–
–
1
1
1
2
9
2
3
12
16
9
–
22
9
Fe(BTC)
Silica/NHPI
NHPI/Fe(BTC)
NHPI/Fe(BTC)
NHPI/Fe(BTC)
NHPI/Fe(BTC)
NHPI/Fe(BTC)
NHPI/Fe(BTC)
NHPI/Fe(BTC)
NHPI/Fe(BTC)
NHPI/Fe(BTC)
NHPI/Fe(BTC)
NHPI/Fe(BTC)g
NHPI+Fe(BTC)
0.8
3.6
6.6
9.2
13
98 (1)
97 (2)
97 (2)
95 (3)
87 (4)
95
93
88
84
85
b
8.6c
d
1
1
1
1
1
1
1
1
9.0
29
33
48
<0.4
18
27
4
e
e
(a)
f
12
8
–
72 (6)
52 (8)
64 (5)
NHPI+Fe(NO
NHPI+Fe(acac)
3
)
3
Á9H
2
O
3
31
a
(b)
Reaction conditions: cyclooctene (0.5 mL), NHPI/Fe(BTC) (75 mg), 100 °C, oxy-
gen purged through balloon. The values in brackets correspond to peroxides
4
3
determined by reaction with PPh .
TON is calculated by mmol of cyclooctene converted to epoxide/mmol of NHPI.
b
0
0
1
2
3
4
5
6
TON is 4 and 1 based on NHPI and Fe, respectively, at 4 h.
First reuse.
Second reuse.
Reaction temperature 110 °C.
Reaction temperature 120 °C.
With 50 mg of TEMPO.
c
d
e
f
Time (h)
Fig. 4. Time conversion plot and leaching test as evidence for heterogeneity in the
catalysis of epoxidation of cyclooctene with oxygen (a) in the presence of NHPI/
Fe(BTC) or (b) after filtration of the solid at about 4% conversion.
g

A. Dhakshinamoorthy et al. / Journal of Catalysis 289 (2012) 259–265
263
where no confinement occurs, and (ii) control of the reactivity and
3.5. Reusability
transition states formed inside the pores of NHPI/Fe(BTC) due to ste-
ric restrictions inside the cage. On the other hand, it has been re-
ported that Fe/MgO in combination with NHPI can also catalyze
CAH oxidation in benzylic compounds [31].
The reusability of NHPI/Fe(BTC) for the epoxidation of cyclooc-
tene using molecular oxygen was investigated under identical con-
ditions as described in Table 1. After the required time, the reaction
mixture was stirred with acetonitrile and the catalyst filtered and
dried. The recovered solid was used for another consecutive run
without further treatment. Only minor changes in the conversion
and no noticeable variations in the epoxide selectivity were ob-
served for cyclooctene oxidation upon two consecutive reuses.
After three consecutive uses, the stability of MOF crystal structure
was confirmed by comparison of the XRD patterns of the fresh and
two times reused NHPI/Fe(BTC) (Fig. 6). In addition, Fig. 6 shows
that inclusion of NHPI on Fe(BTC) does not change the diffraction
pattern of the PCPs. The results of these reusability tests are in
agreement with the previous experiment in which extraction with
cyclooctene of NHPI was not observed.
In order to have some insights onto the reaction mechanism, a
control experiment for aerobic oxidation of cyclooctene under the
condition described in Table 1 was performed with TEMPO as rad-
ical quencher. Interestingly, TEMPO completely quenches the reac-
tion (see Table 1, entry 14). This indicates that the reaction
proceeds through carbon-centered radicals. The induction period
observed in the time-conversion plot shown in Fig. 4 can be inter-
preted as the time needed for the generation of these C-centered
radicals.
To provide a qualitative assessment of the catalytic perfor-
mance of NHPI/Fe(BTC) with respect to analogous, unsupported
catalysts, we have also performed two additional experiments with
iron(III) salts in combination with NHPI (5:1) having the same total
iron amount as that of NHPI/Fe(BTC) catalyst. Fig. 5 summarizes
the results obtained using unsupported NHPI compared with the
performance of NHPI loaded on Fe(BTC). The reaction with iron ni-
trate nonahydrate with NHPI resulted in 27% conversion with 52%
of epoxide and 9% of enol/enone in 4 h. In addition, some uniden-
tified products were also formed which reduce significantly the
selectivity of desired epoxide. On the other hand, iron acetylaceto-
nate with NHPI exhibited 4% conversion with 64% epoxide and 31%
enol/enone selectivities in 4 h. These control experiments clearly
demonstrate the prevailing advantages of NHPI/Fe(BTC) in achiev-
ing higher selectivity of epoxide, reducing the formation of other
side by-products by performing the reaction in the confined envi-
ronment. In other words, selectivity of the aerobic oxidation will be
modified by the fact that the process taken place is mostly in a re-
stricted reaction cavity where the concentration of substrate, oxy-
gen, and products is different from that in the liquid phase. This
diffusion limitation typically increases selectivity by minimizing
secondary reactions of the primary products and also by favouring
preferential reaction pathways. Similar rationalization as the one
provided here based on the confinement effect in NHPI/Fe(BTC)
has been already proposed to explain the synergistic effects ob-
served for the encapsulation of metal phthalocyanine complexes
on MIL-101 in the aerobic oxidation of tetralin [32].
3.6. Aerobic oxidation cycloalkenes
In view of the excellent results achieved for the selective epox-
idation of cyclooctene in the absence of solvent, the scope of the
NHPI/Fe(BTC) aerobic oxidation was studied for other cycloalkenes
(
Table 2). Thus, when NHPI/Fe(BTC) was used as catalyst for aero-
3
.4. Hot filtration test
bic oxidation of cyclopentene and cyclohexene under identical
conditions as those used for cyclooctene, three main products
namely the epoxide, 2-cycloalkanol, and 2-cycloalkanone were
identified at short reaction times. At 5 h, the selectivity toward
To verify whether the catalysis of NHPI/Fe(BTC) is truly hetero-
geneous or the catalytic activity is due, at least in part, to some lea-
ched iron species present in the liquid phase, the reaction was
carried out under optimized conditions (footnote ‘‘a’’ in Table 1
for reaction conditions) and the NHPI/Fe(BTC) solid catalyst was
filtered out of the reaction mixture at 4% formation of cyclooctene
oxide under the reaction temperature using nylon filter. After
removal of the NHPI/Fe(BTC) catalyst, the solution in the absence
of solid was again stirred. After 4 h, no further product formation
was observed in the absence of solid (see Fig. 4). Hence, it can be
concluded that the catalysis occurs due to the presence of the solid
phase and the process is truly heterogeneous.
2
-cycloalkanone increased significantly, while the percentage of
epoxide present in the reaction mixture was very low.
The aerobic oxidation was also inhibited by the presence of
TEMPO indicating that the reaction mechanism for the three
products is a C-centered radical autooxidation. Interestingly, in
the case of cycloheptene, the epoxide selectivity is maintained
fairly constant over the course of the reaction to a value about
2
0%. In contrast, we observe that for cyclopentene and cyclohex-
ene, the selectivity of epoxide decreases significantly with reaction
time (see Table 2, entries 1, 2 and 3, 4). For cycloheptene, 23% of
epoxide was observed in 13% conversion at 4 h, while the epoxide
selectivity is 3% for cyclopentene and cyclohexene at 5 h. This
500
400
300
200
100
0
*
(
c)
b)
a)
(
(
0
10 20 30 40 50 60 70
2
θ
Fig. 5. Comparison of conversion and epoxide selectivity achieved for the aerobic
oxidation of cyclooctene with different catalysts: (1) NHPI/Fe(BTC); (2) NHPI+-
Fig. 6. Powder XRD pattern of (a) Fe(BTC), (b) NHPI/Fe(BTC), and (c) two times
reused NHPI/Fe(BTC) (see footnote ‘‘a’’ in Table 1 for reaction conditions). The
asterisk indicates the peak due to the sample cell holder.
Fe(BTC); (3) Fe(NO
)
3 3
Á9H
2 3
O+NHPI; and (4) Fe(acac) +NHPI.

2
64
A. Dhakshinamoorthy et al. / Journal of Catalysis 289 (2012) 259–265
Table 2
a
Aerobic oxidation of cycloalkenes using NHPI/Fe(BTC) as heterogeneous catalysts .
Run
Cycloalkene
Time (h)
Con. (%)b
Selectivity (%)b
Epoxide
9
Enol
Enone
64
1c
1
1.1
27
2c
3
5
2
6
9
3
20
5
33
92
47
c
4c
5
6
5
5
1
12
1.5
1.5
3
–
20
38
40
30
59
60
50
Scheme 2. Proposed mechanism for the aerobic oxidation of C@C double bonds
catalyzed by NHPI/Fe(BTC). This mechanism is favored for highly reactive C@C
double bonds.
d
7
8
9
1
1
2
3
4
3
5
6
21
22
23
–
32
28
27
35
39
47
50
50
65
44
10
13
2000000
1000000
0
e
0
1
2
f
3.3
17
a
Reaction conditions: cycloalkene (0.5 mL), NHPI/Fe(BTC) (75 mg), 100 °C, oxy-
gen purged with balloon.
b
Determined by GC. TON values calculated based on NHPI for cyclopentene (5 h),
cyclohexene (5 h) and cycloheptene (4 h) are 13, 25, 24, respectively.
Reaction was carried out with 5 bar of oxygen in a sealed reactor.
Reaction was carried out with 5 bar of oxygen and 50 mg of TEMPO in a sealed
reactor.
-
1000000
2000000
3
c
-
d
485
3490
3495
3500
3505
e
f
With 50 mg of TEMPO.
With 10 mg of b-carotene.
Magnetic Field /Gauss
Fig. 7. ESR spectrum of phthalimide N-oxyl radical generated in situ by exposing a
NHPI/Fe(BTC) suspension in benzonitrile to oxygen at 120 °C.
decrease in selectivity could indicate that in the case of cyclopen-
tene and cyclohexene, the corresponding epoxides are not stable
under the reaction conditions decomposing during the course of
the reaction.
difference in reactivity of the C@C double bond as a function of
the ring size previously observed is also responsible for the higher
tendency of cyclooctene to undergo epoxidation rather than allylic
oxidation, while the less reactive smaller cycloalkenes containing
less reactive C@C double bond prefer allylic oxidation.
With respect to the reaction mechanism, we observe that the
aerobic oxidation of cycloheptene in the presence of radical
quenchers, namely TEMPO and b-carotene, also retards the per-
centage conversion. Comparison of the results shown in Tables 1
and 2 establishes two different reactivity patterns for the aerobic
oxidation promoted by NHPI/Fe(BTC), that is, either epoxidation
or allylic oxidation. The later predominates for cyclopentene and
cyclohexene, while cycloheptene presents a transition toward the
high epoxide selectivity found for cyclooctene at low or moderate
conversions. In fact, in the case of cyclooctene, formation of cyclo-
octene oxide dominates with very high selectivity even up to con-
versions close to 50%. Previous reports in the literature have found
that the order of reactivity of cyclic olefins increases in the order,
cyclohexene < cycloheptene < cyclooctene [33,34]. The conversion
and selectivity toward epoxides are mainly dependent on cycloal-
kene conformation, the bond angle strain, torsional strain, ring
size, and electronic nature of cyclic olefin which as it has been ob-
served in earlier precedents [33,34]. We propose that this
3.7. Scope of NHPI/Fe(BTC)
To further extend the scope of NHPI/Fe(BTC), 1-methylcyclo-
hexene (0.5 mL) was also used as substrate. 16% selectivity to
epoxide at 6% conversion was observed for the reaction carried
out at 80 °C with 25 mg of catalyst. Increasing the catalyst amount
to 75 mg at 100 °C resulted in 22% conversion with 13% selectivity
to epoxide. In addition to the epoxide, we also observed the forma-
tion of different ol/one isomers of 1-methylcyclohexene at various
allylic positions. Similarly, (S)-a-pinene (0.5 mL) undergoes oxida-
tion in the presence of NHPI/Fe(BTC) giving the corresponding
epoxide with a maximum selectivity of 31% at 5% conversion. In
this case, we observed the formation of products derived from
epoxide rearrangement (see Table 3). This data show the general
Table 3
a
Aerobic oxidation of various alkenes catalyzed by NHPI/Fe(BTC) .
Run
Substrate (mL)
NHPI/Fe(BTC) (mg)
Temp. (°C)
Time (h)
Con. (%)
Selectivity (%)b
Epoxide
Ol/one
80c
Others
1
2
3
4
5
1-Methylcyclohexene (0.5)
1-Methylcyclohexene (0.5)
(S)-Pinene (0.5)
(S)-Pinene (1)
(S)-Limonene (0.5)
25
75
25
25
75
80
100
80
100
100
2.5
2
2
2.2
2
6
22
5
11
15
16
13
31
34
4
6
4
6
8
c
81
c
65
60
c
d
c
33
59
a
b
c
All the reactions were performed in oxygen atmosphere.
Percentage selectivity was determined by GC and products were identified by GC–MS.
Various isomers of ol/one products at different allylic positions.
Epoxide selectivity corresponds to cis and trans.
d

A. Dhakshinamoorthy et al. / Journal of Catalysis 289 (2012) 259–265
265
catalytic activity of NHPI/Fe(BTC) for aerobic oxidation and the
Acknowledgments
selectivity toward the epoxide depending on the substrate and
the conversion. Finally, (S)-limonene (0.5 mL) resulted in 15% con-
version with 33% selectivity of cis- and trans-epoxide with 75 mg
of catalyst at 100 °C in 2 h.
Financial support by the Spanish DGI (CTQ-2010-18671 and
CTQ2007-67805) and European Commission Integrated Project
(MACADEMIA) is gratefully acknowledged.
Besides oxidation of cyclic alkenes, we were also interested to oxi-
dation of aliphatic olefins. It was found that 1-octene undergoes in the
presence of NHPI/Fe(BTC) under the optimal conditions indicated in
footnote of Table 1 1% conversion, mostly in allylic positions.
References
[
[
[
1] G. ten Brink, I.W.C.E. Arends, R.A. Sheldon, Science 287 (2000) 1636–1639.
2] A. Corma, H. Garcia, Chem. Soc. Rev. 37 (2008) 2096–2126.
3] M. Olus Özbek, I. Önal, R.A.V. Santen, ChemCatChem 3 (2011) 150–153.
3.8. Mechanism
[4] C. Aprile, A. Corma, M.E. Domine, H. Garcia, C. Mitchell, J. Catal. 264 (2009) 44.
[
[
5] M. Ojeda, E. Iglesia, Chem. Commun. (2009) 352–354.
6] J. Huang, T. Akita, J. Faye, T. Fujitani, T. Takei, M. Haruta, Angew. Chem. Int. Ed.
Concerning the reaction mechanism and in view of the product
48 (2009) 7862–7866.
distribution and the precedents in the use of NHPI as organocata-
lyst for aerobic oxidations [35], it is reasonable to consider that
all the products arise from autooxidation. The generation of the
first carbon-centered radical will arise from phthalimide N-oxyl
radical derived from NHPI. Evidence for this has been obtained
by EPR. Allylic oxidation will arise from the interception of allylic
radicals by oxygen and will predominate for 5 and 6 membered
cycloalkenes. Concerning the formation of cyclooctene oxide, this
compound will result of the higher reactivity of C@C bond for this
large cycloalkene. In this case, the initial 2-cyclooctene peroxyl
radical would add to the C@C double bond of another cyclooctene
leading eventually to the epoxide as the product by elimination of
[7] N.V. Maksimchuk, K.A. Kovalenko, V.P. Fedin, O.A. Kholdeeva, Adv. Synth. Catal.
352 (2010) 2943–2948.
[
[
8] S. Kitagawa, R. Kitaura, S.-I. Noro, Angew. Chem. Int. Ed. 43 (2004) 2334–2375.
9] A. Corma, H. Garcia, F.X. Llabres i Xamena, Chem. Rev. 110 (2010) 4606.
[
10] D. Farrusseng, S. Aguado, C. Pinel, Angew. Chem. Int. Ed. 48 (2009) 7502.
[11] J.Y. Lee, O.K. Farha, J. Roberts, K.A. Scheidt, S.T. Nguyen, J.T. Hupp, Chem. Soc.
Rev. 38 (2009) 1450.
12] G. Ferey, Chem. Soc. Rev. 37 (2009) 191.
13] A. Dhakshinamoorthy, M. Alvaro, H. Garcia, J. Catal. 267 (2009) 1.
[
[
[14] A. Dhakshinamoorthy, M. Alvaro, H. Garcia, Adv. Synth. Catal. 352 (2010) 711–
17.
15] A. Dhakshinamoorthy, M. Alvaro, H. Garcia, Chem. Eur. J. 16 (2010) 8530–
536.
[16] A. Dhakshinamoorthy, M. Alvaro, H. Garcia, Chem. Commun. 46 (2010) 6476–
478.
7
[
8
6
[
[
17] A. Dhakshinamoorthy, M. Alvaro, H. Garcia, ChemCatChem 2 (2010) 1438.
18] L. Kurfirtova, Y.-K. Seo, Y.K. Hwang, J.-S. Chang, J. Cejka, Catal. Today 179
(2012) 85–90.
19] M. Tonigold, Y. Lu, B. Bredenkotter, B. Rieger, S. Bahnmuller, J. Hitzbleck, G.
Langstein, D. Volkmer, Angew. Chem. Int. Ed. 48 (2009) 7546.
20] K. Leus, I. Muylaert, M. Vandichel, G.B. Marin, M. Waroquier, V.V. Speybroeck,
P.V.D. Voort, Chem. Commun. 46 (2010) 5085.
2
-cyclooctene oxyl radical (Scheme 2). This mechanism is in agree-
ment with those previously reported in the literature [28–30].
C-centered radicals will be generated from NHPI after formation
of the corresponding N-oxyl radical. To confirm the formation of
phthalimide N-oxyl radicals promoted by NHPI/Fe(BTC) in the
presence of oxygen, electron spin resonance (ESR) spectra were re-
corded [17]. A control test did not detect the presence of any rad-
ical in powder NHPI/Fe(BTC). When a benzonitrile solution of
NHPI/Fe(BTC) was exposed to dioxygen at 120 °C for 2 h, an ESR
spectrum attributable to phthalimide N-oxyl radicals (Fig. 7) was
recorded as a triplet signal. The g-value and hfs constant observed
for NHPI/Fe(BTC) were coincident with those reported for phthali-
[
[
[21] D. Jiang, T. Mallat, D.M. Meier, A. Urakawa, A. Baiker, J. Catal. 270 (2010) 26–
3.
22] S. Bhattacharjee, D.-A. Yang, W.-S. Ahn, Chem. Commun. 47 (2011) 3637–
639.
3
[
3
[23] P. Horcajada, S. Surble, C. Serre, D.-Y. Hong, Y.-K. Seo, J.-S. Chang, J.-M.
Greneche, I. Margiolaki, G. Ferey, Chem. Commun. (2007) 2820–2822.
24] A. Dhakshinamoorthy, M. Alvaro, H. Garcia, ACS Catal. 1 (2011) 836–840.
25] Y. Ishii, S. Sakaguchi, in: J.E. Backvall (Ed.), Modern Oxidation Methods, Wiley,
2004 (Chapter 5, and the other chapters described in this book).
[
[
[
[
26] Y. Ishii, S. Sakaguchi, T. Iwahama, Adv. Synth. Catal. 343 (2001) 393–427.
27] I. Hermans, J.V. Deun, K. Houthoofd, J. Peeters, P.A. Jacobs, J. Catal. 251 (2007)
mide N-oxyl in solution (g = 2.0073, a
. Conclusion
We have shown that Fe(BTC) containing NHPI could be used as
N
= 0.423 mT) [36].
204–212.
4
[28] D.E. Van Sickle, F.R. Mayo, R.M. Arluck, J. Am. Chem. Soc. 87 (1965) 4824.
[
[
[
29] U. Neuenschwander, F. Guignard, I. Hermans, ChemSusChem 3 (2010) 75.
30] U. Neuenschwander, I. Hermans, Phys. Chem. Chem. Phys. 12 (2010) 10542.
31] S.-H. Cho, M.-S. Cheong, K.-D. Jung, C.-S. Kim, S.-H. Han, Appl. Catal. A: Gen.
267 (2004) 241–244.
32] E. Kockrick, T. Lescouet, E.V. Kudrik, A.B. Sorokin, D. Farrusseng, Chem.
Commun. 47 (2011) 1562–1564.
33] T.-J. Wang, Y.-Y. Yan, M.-Y. Huang, Y.-Y. Jiang, React. Funct. Polym. 29 (1996)
145–149.
34] K.M. Jinka, J. Sebastian, R.V. Jasra, J. Mol. Catal. A: Chem. 274 (2007) 33–41.
35] R.A. Sheldon, I.W.C.E. Arends, Adv. Synth. Catal. 346 (2004) 1051.
36] A. Mackor, T.A.J.W. Wajer, T.J. de Boer, Tetrahedron 24 (1968) 1623–1631.
heterogeneous catalyst for the aerobic oxidation of cycloalkenes.
Two different types of products, namely, epoxidation and allylic
oxidation have been observed in the absence of cosolvents at mod-
erate conversions depending on the ring size. Very high selectivity
for the aerobic epoxidation was achieved with cyclooctene, while
decreasing the ring size favors allylic oxidation. In general, the
present study highlights the use of commercially available Fe(BTC)
as heterogeneous catalysts in aerobic oxidation of cycloalkenes and
no additional solvent was used.
[
[
[
[
[