Effects of porphyrin composition on the activity and selectivity of the iron(III) porphyrin catalysts for the epoxidation of cyclooctene by hydrogen peroxide

Article abstract of DOI:10.1016/j.molcata.2007.03.030

A detailed investigation was carried out of the effects of porphyrin composition on the activity and selectivity of iron(III) porphyrin catalysts used for the epoxidation of cyclooctene by hydrogen peroxide. Under conditions where the formation of μ-oxo-dimers can be avoided, the mechanism of cyclooctene epoxidation and hydrogen peroxide decomposition are identical for all of the porphyrin catalysts investigated. It is observed that as the electron-withdrawing character of the phenyl groups is enhanced through halogenation, the tendency of iron(III) porphyrin chloride to dissociate decreases, but the Lewis acidity of the resulting iron(III) porphyrin cation increases. Increasing the Lewis acidity of the Fe(III) center also enhances the rate of heterolytic versus homolytic cleavage of the O-O bond of coordinated hydrogen peroxide, as well as the rate hydrogen peroxide consumption for olefin epoxidation versus peroxide decomposition. We have also shown that solvent composition affects both the extent of hydrogen peroxide coordination to the iron(III) porphyrin cation, thus affecting its activity, and the reactivity of hydrogen peroxide with the iron(IV) pi-radical cation, thus affecting the selectivity with which hydrogen peroxide is used for epoxidation.

Full text of DOI:10.1016/j.molcata.2007.03.030

Journal of Molecular Catalysis A: Chemical 272 (2007) 108–117
Effects of porphyrin composition on the activity and selectivity
of the iron(III) porphyrin catalysts for the epoxidation
of cyclooctene by hydrogen peroxide
Ned A. Stephenson, Alexis T. Bell∗
Chemical Sciences Division, Lawrence Berkeley Laboratory, Department of Chemical Engineering, University of California,
201 Gilman Hall, Berkeley, CA 94720-1462, United States
Received 30 January 2007; received in revised form 8 March 2007; accepted 9 March 2007
Available online 15 March 2007
Abstract
A detailed investigation was carried out of the effects of porphyrin composition on the activity and selectivity of iron(III) porphyrin catalysts used
for the epoxidation of cyclooctene by hydrogen peroxide. Under conditions where the formation of ␮-oxo-dimers can be avoided, the mechanism
of cyclooctene epoxidation and hydrogen peroxide decomposition are identical for all of the porphyrin catalysts investigated. It is observed that
as the electron-withdrawing character of the phenyl groups is enhanced through halogenation, the tendency of iron(III) porphyrin chloride to
dissociate decreases, but the Lewis acidity of the resulting iron(III) porphyrin cation increases. Increasing the Lewis acidity of the Fe(III) center
also enhances the rate of heterolytic versus homolytic cleavage of the O–O bond of coordinated hydrogen peroxide, as well as the rate hydrogen
peroxide consumption for olefin epoxidation versus peroxide decomposition. We have also shown that solvent composition affects both the extent
of hydrogen peroxide coordination to the iron(III) porphyrin cation, thus affecting its activity, and the reactivity of hydrogen peroxide with the
iron(IV) pi-radical cation, thus affecting the selectivity with which hydrogen peroxide is used for epoxidation.
© 2007 Elsevier B.V. All rights reserved.
Keywords: Porphyrin; Peroxide; Epoxidation; Mechanism; Cyclooctene
1. Introduction
groups attached to the meso-position of the porphyrin ring,
whereas third-generation porphyrins have halogen atoms at the
␤-pyrrole positions on the porphyrin ring [10,11]. Halogenation
of the phenyl groups has resulted in catalysts that are more
selective for epoxidation over peroxide decomposition [5,9] and
are less susceptible to degradation of the porphyrin ring [4,5].
Third-generation porphyrins, which contain halogen atoms at
the ␤-pyrrole positions cause severe distortion of the porphyrin
ring [9] and, as a result, reactivity studies have revealed that
these catalysts are less robust [5] and less efficient [11] than
second-generation iron porphyrins.
The selectivity of hydrogen peroxide for olefin epoxidation
versus peroxide decomposition is determined by two com-
peting reactions—heterolytic and homolytic cleavage of the
oxygen–oxygen bond of hydrogen peroxide coordinated to the
iron(III) cation [3,15]. Heterolytic cleavage of this bond leads
to the formation of an iron(IV)-oxo pi-radical cation, which is
an intermediate in hydrocarbon oxidation, whereas homolytic
cleavage leads exclusively to peroxide decomposition [3,15]. It
has also been reported that the high valent, iron-oxo pi-radical
Work carried out by Groves et al. at the end of the 1970s
showed that iron(III) [tetraphenyl]porphyrin chloride,
(TPP)FeCl, catalyzes the epoxidation of olefins by iodosyl-
benzene [1]. However, attempts to use this catalyst for olefin
epoxidation with hydrogen peroxide as the oxidant have not
been as successful, since (TPP)FeCl is not very selective
[2–5] and is highly susceptible to degradation [4,5]. These
observations have motivated the synthesis and evaluation of a
large variety of porphyrin analogues in an attempt to improve
the catalytic activity and robustness of porphyrin catalysts for
use with hydrogen peroxide; reviews of this work are given in
Refs. [6–9]. Most notable are the so-called second and third
generation catalysts [5,10–14]. Second-generation analogues of
(TPP)FeCl have substituents, usually halogens, on the phenyl
∗
Corresponding author. Tel.: +1 510 642 1536; fax: +1 510 642 4778.
E-mail address: alexbell@berkeley.edu (A.T. Bell).
1381-1169/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2007.03.030

N.A. Stephenson, A.T. Bell / Journal of Molecular Catalysis A: Chemical 272 (2007) 108–117
109
Fig. 1. The proposed mechanism for iron(III) porphyrin catalyzed epoxidation of cyclooctene by H2O2 in a methanol containing solvent. Note that the methanol
ligand is not shown after reaction (2) for the sake of clarity.
cationic species can react to form an intermediate that leads to
further decomposition of hydrogen peroxide [15–17]. Studies by
Nam et al. have shown that the composition of second generation
porphyrins affects both the selectivity for homolytic versus het-
erolytic cleavage of the O–O bond of coordinated H2O2 [3], and
the selectivity of the iron(IV)-oxo pi-radical cation for hydro-
carbon oxidation versus peroxide degradation [17]. This work
demonstrated qualitatively that the addition of electronegative
substituents improves the selective utilization of hydrogen per-
oxide for hydrocarbon oxidation over peroxide decomposition,
but did not provide a detailed analysis of the effects of porphyrin
composition on the mechanism and kinetics of the elementary
processes involved in olefin epoxidation and peroxide decom-
position.
2. Experimental
2.1. Reagents
Deuterium oxide, 99.9%, was obtained from Cambridge
Isotope Laboratories, Inc. cis-Cyclooctene, 95%, was obtained
from Alfa-Aesar. Iron(III) [tetrakis(pentafluorophenyl)] por-
phyrin chloride (F20TPP)FeCl and dodecane, 99+%, were
obtained from Aldrich. Omni Solv dichloromethane, 99.96%,
methanol, 99.98%, ACS grade chloroform, 99.8%, and hydro-
gen peroxide, 30%, were all obtained from EMD Chemicals.
Iron(III) [tetraphenyl]porphyrin chloride (TPP)FeCl, iron(III)
[tetrakis(4-fluorophenyl)] porphyrin chloride (2-F4TPP)FeCl,
iron(III) [tetrakis(2,6-difluorophenyl)] porphyrin chloride
In previous studies of olefin epoxidation by hydrogen per-
oxide catalyzed by (F20TPP)FeCl, the kinetics of epoxide
formation and hydrogen peroxide consumption are described
very accurately for a wide range of reaction conditions by rate
expressions derived from the reaction mechanism shown in
Fig. 1 [15,18–20]. This has enabled us to determine the effects
of solvent composition and axial ligand composition on the rate
and equilibrium coefficients associated with each of the ele-
mentary reactions appearing in the mechanism. The purpose
of the work reported here is, first, to determine the effects of
porphyrin composition on the mechanism shown in Fig. 1 and,
second, to quantify the effects of the halogenation of the phenyl
groups attached to the porphyrin ring on the individual rate and
equilibrium parameters.
(2,6-F8TPP)FeCl,
iron(III)
[tetrakis(3,5-difluorophenyl)]
porphyrin chloride (3,5-F8TPP)FeCl, and iron(III) tetrakis(2,6-
dichlorophenyl)] porphyrin chloride (2,6-Cl8TPP)FeCl, and
the ␮-oxo dimer of iron(III) [tetrakis(3,5-difluorophenyl)]
porphyrin chloride (3,5-F8TPP)Fe–O–Fe(3,5-F8TPP) were
obtained from Frontier Scientific. The structures and compo-
sitions of the iron porphyrins used in this study are presented
in Fig. 2, together with their full name and the symbol used to
identify the porphyrin.
2.2. Reactions
Prior to initiating a reaction, cyclooctene, solvent, and
catalyst were mixed in a 5 mL reaction vial. The solvent

110
N.A. Stephenson, A.T. Bell / Journal of Molecular Catalysis A: Chemical 272 (2007) 108–117
2.4. Hydrogen peroxide consumption and analysis
The hydrogen peroxide concentration was measured as a
function of time using 1H NMR as described in Ref. [21]. These
analyses were carried out using a 400 MHz VMX spectrome-
ter. Samples were prepared in 5 mL reaction vials, as described
above, with the exception that the catalyst was not added to the
vial. A portion of this reaction mixture was transferred to a pre-
cision NMR tube. Chloroform was used as the internal standard.
The reaction was initiated by adding a concentrated solution of
porphyrin catalyst to the NMR tube prior to taking scans. A
NMR spectrum was taken every 32 s by averaging the data from
four scans. The area of the hydrogen peroxide peak, located at
∼10 ppm, was then integrated and compared to the area of an
internal standard. 1H NMR experiments were run in triplicate to
verify the repeatability of the experiment.
Analysis of hydrogen peroxide in various solvents was car-
ried out using a 400 MHz VMX spectrometer. 1H NMR spectra
were taken of the mixtures of hydrogen peroxide (14 mM) and
solvent, which were prepared by adding 5 ␮L of standardized
hydrogen peroxide solution to 3 mL of solvent.
Fig. 2. Structure and terminology used for various iron(III) porphyrin catalysts.
2.5. Porphyrin analysis
consisted of a mixture of methanol and dichloromethane;
the methanol concentration was 5.6 M unless specified oth-
erwise. The cyclooctene concentration was varied depending
on the experiment. The total reaction volume was held con-
stant for all reactions at 3.35 mL by changing the concentration
of dichloromethane. Reactions were initiated by addition of
hydrogen peroxide (1–10 ␮L). To facilitate product analysis,
10 ␮L of dodecane was added as a standard to the reaction
mixture. All reactions were performed at room temperature.
Reactions using (TPP)FeCl as the catalyst were conducted
under nitrogen using deoxygenated solvents since it was
observed that the presence of oxygen promoted competitive
reactions involving free-radicals. Evidence for radical oxida-
tion was not observed for the other catalysts in the presence of
air.
Degradation of the porphyrin species was quantified by
UV–vis spectroscopy using a Varian Cary 400 Bio UV–vis spec-
trometer. Scans of the reacting system were taken as a function
of reaction time. The UV–vis spectra were then integrated to find
the area under the peaks between 280 and 700 nm. An extinction
coefficient was calculated based on a linear plot of integrated
area versus porphyrin concentration. The repeatability of the
UV–vis scans was shown to be better than 2%. Negligible
error was associated with taking repeated scans; all error was
associated with the sample preparation.
Analyses of the porphyrin species were carried out using a
400 MHz VMX spectrometer to corroborate the exchange of the
chloride ligand for a molecule of methanol. Additional experi-
mental details for these 1H NMR experiments can be found in
Ref. [18].
2.3. Product analysis
3. Analysis of reaction kinetics
The concentration of cyclooctene epoxide as a function of
time was determined by gas chromatography. Analysis was car-
ried out using a HP 6890 Series gas chromatograph equipped
with an Agilent DB Wax (30 m × 0.32 mm × 0.5 ␮m) capillary
column and a FID detector. The temperature program for anal-
ysis was as follows: 2.5 min at 50 ◦C, ramp at 20 ◦C per min
to 175 ◦C, and 1.5 min at 175 ◦C. The split/splitless inlet was
operated at 200 ◦C and 14.2 psi with a split ratio of 5. Dode-
cane and cyclooctene epoxide were eluted at 5.5 and 8.2 min,
respectively. The time required to analyze a single sample via gas
chromatography was on the order of the time for the completion
of the reaction. Therefore, multiple reactions were performed in
order to obtain data points as a function of time and to verify
reproducibility. The precision of measuring cyclooctene epox-
ide via gas chromatography was determined to be within 1%
by analysis of a calibrated standard.
As noted in Section 1, our previous studies have shown that
the kinetics of cyclooctene epoxidation by H2O2 catalyzed by
(F20TPP)FeCl are well described by rate expressions derived
from the mechanism presented in Fig. 1 [15,18–20]. For the cat-
alyst to become active (F20TPP)FeCl must first dissociate into
cations and anions (reaction (1)). Hydrogen peroxide then coor-
dinates to the iron(III) cation (reaction (2)). The oxygen–oxygen
bond of the iron(III)–hydrogen peroxide intermediate is then
cleaved either heterolytically (reaction (3)) or homolytically
(reaction (4)). Heterolytic cleavage results in the formation of
iron(IV) pi-radical cation species which can either react further
with hydrogen peroxide (reaction (6)) or epoxidize an olefin
(reaction (7)). Homolytic cleavage results exclusively in perox-
ide decomposition via an iron(IV)-hydroxo intermediate. The
mechanism presented in Fig. 1 can be used to derive Eqs. (1)–(4).

N.A. Stephenson, A.T. Bell / Journal of Molecular Catalysis A: Chemical 272 (2007) 108–117
111
Here, kobs is the observed rate coefficient for hydrogen peroxide
consumption, k3 and k4 are the rate coefficients for Reactions
(3) and (4) (see Fig. 1), and [FeIII–MeOH] is the concentration
of iron(III) methanol-coordinated porphyrin species.
d[H2O2]
= −kobs[H2O2]
(1)
dt
k3K2[FeIII–MeOH][MeOH]
kobs
=
(2)
(3)
Y
∞
[C8–O]
[H2O2] = [H2O2]0 −
Y
∞
[C8–O]
k3 [MeOH]
∞
∼
=
Y
∞
=
× 100%
× 100% (4)
[H2O2]0
k3 [MeOH] + 2k4
The value of kobs was determined by measuring the initial
rate of cyclooctene oxide production. This value was deter-
mined using Eq. (1), where kobs is defined according to Eq. (2).
Only methanol-coordinated porphyrin species, represented by
FeIII–MeOH in Eq. (2), are active catalytically [18,19]. The con-
centration of hydrogen peroxide is related to the concentration of
cyclooctene oxide (C8–O) via Eq. (3) using the final yield (Y )
∞
of cyclooctene oxide. The assumption that the yield of epoxide is
independent of the concentration of hydrogen peroxide was ver-
ified for each porphyrin catalyst. Since the yield and observed
rate constants are independent of the concentration of hydro-
gen peroxide, the yield can be approximated according to Eq.
(4) when reaction (6) does not contribute significantly to hydro-
gen peroxide consumption; here the yield is approximated as the
ratio of the rate of peroxide consumption by heterolytic cleavage
of the oxygen–oxygen bond of hydrogen peroxide to the sum of
the rates of peroxide consumption by heterolytic and homolytic
cleavage of the oxygen–oxygen bond of hydrogen peroxide. A
factor of 2 is placed before the rate of homolytic cleavage, since
two molecules of hydrogen peroxide are consumed via that reac-
tion path (reactions (4) and (5)). It then follows that the values
of k3K2 and k4K2 can be determined from kobs and Y . Values
∞
for all measured rate parameters are reported in Table 1.
The validity of determining the rate parameters exclusively
from gas chromatography data was investigated by measuring
the actual rates of hydrogen peroxide consumption via 1H NMR.
Rates of hydrogen peroxide consumption were measured both
in the presence and absence of substrate. The individual rate
parameters were determined by writing rate expressions for the
mechanism shown in Fig. 1 and fitting the rate parameters to both
the gas chromatography and 1H NMR data. The values obtained
for k3K2 and k4K2 using this method were within 10% of those
obtained using only the rate of cyclooctene oxide production as
measured by gas chromatography. The rates of peroxide con-
sumption were measured via 1H NMR for all porphyrins except
(F20TPP)Fe, for which the rate of peroxide consumption was
too rapid to allow for quantification by 1H NMR.
The relative rates of reactions (6) and (7) were determined
by measuring the yield of cyclooctene oxide as a function of the
ratio of hydrogen peroxide to cyclooctene and then fitting the
observed data by adjusting the ratio k6/k7. The fitted values of
k6/k7 are given as the final entry in Table 1.

112
N.A. Stephenson, A.T. Bell / Journal of Molecular Catalysis A: Chemical 272 (2007) 108–117
We have shown previously that the rate parameter k5 can
be determined by fitting hydrogen peroxide consumption rate
data, determined via 1H NMR, in the absence of olefin [15,22].
For k5 to be determined by this technique, the rate of hydrogen
peroxide consumption must be slow enough that the concentra-
tion of hydrogen peroxide can be determined temporally by 1H
NMR. In addition, k5 must be less than ∼300 M−1 s−1 or the
contribution of reaction (5) to the rate of peroxide decomposi-
tion becomes indistinguishable from the overall rate of peroxide
decomposition. In the present study, k5 could not be determined
for (F20TPP)Cl because the rate of reaction was too rapid to be
1
quantified by H NMR, and values of k5 for (TPP)FeCl, (2,6-
F8TPP)FeCl, and (2,6-Cl8TPP)FeCl could not be determined
because k5 was sufficiently large that the contribution of reaction
(5) could not be distinguished from the overall rate of peroxide
decomposition.
Fig. 3. Observed rate constant for (F20TPP)Fe vs. the concentration of
methanol-coordinated iron porphyrin as determined from K1 = 1.3 × 10−5. The
cyclooctene and hydrogen peroxide concentrations were 0.72 M and 16 ␮M,
respectively, in a dichloromethane/methanol (3:1) solvent mixture.
4. Results and discussion
4.1. Aprotic solvent composition
In our previous studies of cis-cyclooctene epoxidation cat-
alyzed by (F20TPP)FeCl, a mixture of acetonitrile and alcohol
was used as the solvent [15,19]. This solvent mixture could
not be used for the present work, though, since all of the
other porphyrins listed in Table 1 have limited solubilities in
acetonitrile and methanol; however, all of these porphyrins
are soluble in dichloromethane. Therefore, it was necessary to
use a dichloromethane/methanol mixture instead of an acetoni-
trile/methanol mixture as the solvent. To determine whether the
composition of the aprotic solvent affects the rate parameters,
the experimental studies of (F20TPP)FeCl reported previously
[15,19], which had been carried out in an acetonitrile/methanol
mixture, were repeated using a dichloromethane/methanol
mixture. The concentration of oxidant, substrate, catalyst,
and methanol were the same for experiments using both
dichloromethane and acetonitrile, while equal volumes of apro-
tic solvents were used to maintain a constant reaction volume.
Initial studies of selectivity and reactivity under similar
reaction conditions (0.72 M cyclooctene, 15 mM H2O2, 75 ␮M
(F20TPP)FeCl, 5.6 M methanol, 3.35 mL total volume) revealed
that the selectivity for heterolytic cleavage is independent of
the nature of the aprotic solvent, but that the rate of peroxide
consumption is much faster in the presence of dichloromethane
than in the presence of acetonitrile. As shown in Table 1, the
observed rate constant for peroxide consumption was 1.4 min−1
for the reaction carried out in dichloromethane/methanol and
0.25 min−1 for the reaction carried out in acetonitrile/methanol.
Additional experiments, similar to those carried out in acetoni-
trile/methanol, confirmed that the mechanism presented in Fig. 1
still applies when dichloromethane/methanol is used as the sol-
vent.
constant and yield of epoxide are independent of the ini-
tial concentration of hydrogen peroxide, verifying the pseudo
first-order relationship expressed in Eq. (1). The equilibrium
constant for reaction (1) was determined to be 1.3 × 10−5 via
1H NMR. This value is slightly higher than that of 1.0 × 10−5
obtained for the acetonitrile/methanol system. To test the valid-
ity of this value, the observed rate constant was determined
experimentally for varying porphyrin concentrations in the
dichloromethane/methanol system. As shown in Fig. 3 and sug-
gested by Eq. (2), kobs is linearly related to the concentration
of methanol-coordinated iron porphyrin as determined from the
equilibrium constant K1. In addition, the slope of the data in
Fig. 3 predicts a value of 80 M−2 s−1 for k3K2, which is simi-
lar to that obtained from the initial rate studies (82 M−2 s−1 as
shown in Table 1).
As shown in Table 1, the rate parameters k3K2 and k4K2
are almost an order of magnitude greater when the epoxida-
tion of cyclooctene is carried out in dichloromethane/methanol
than when this reaction is carried out in acetonitrile/methanol.
However, the ratios of k3/k4 are identical, indicating that k3 and
k4 are independent of the composition of the aprotic solvent.
Therefore, it follows that the difference in the observed rates
of reaction is due primarily to a difference in the value of K2
in dichloromethane versus acetonitrile. Two possibilities may
explain the dependence of K2 on the solvent composition. The
first possibility is that hydrogen peroxide may compete with
the solvent for coordination to the iron(III) cation. The second
possibility is that solvent interactions with hydrogen peroxide
may stabilize hydrogen peroxide in solution, and thereby, reduce
its tendency to coordinate to iron porphyrin cations. The first of
these options seems unlikely since hydrogen peroxide molecules
are much more nucleophilic than any of the solvent molecules
considered and are, therefore, expected to interact to a greater
extent with the Lewis acidic iron(III) cation. Consistent with
this, the observed rate of reaction is independent of the hydro-
gen peroxide concentration. If solvent molecules competed with
Rate parameters for the reactions in dichloromethane/
methanol, based on initial rate data, were determined and are
compared in Table 1 to those obtained for reactions in ace-
tonitrile/methanol. Several observations can be made. First,
as expected from the proposed mechanism, the observed rate

N.A. Stephenson, A.T. Bell / Journal of Molecular Catalysis A: Chemical 272 (2007) 108–117
113
hydrogen peroxide for coordination to the iron(III) cation, then
the observed rate of reaction would be expected to increase with
increasing peroxide concentration.
The second explanation for the solvent dependence of K2
seems more reasonable since hydrogen peroxide is expected
to interact to varying degrees with different solvent molecules.
Methanol, present in both sets of experiments, contains both a
hydroxyl group and two lone pairs of electrons that can bond to
hydrogen peroxide molecules. Acetonitrile also has a lone pair
of electrons which can form a hydrogen bond with hydrogen
peroxide molecules. On the other hand, dichloromethane does
not have the ability to hydrogen bond with hydrogen perox-
ide. Hydrogen bonding between hydrogen peroxide and solvent
molecules would affect the electronic properties of the hydrogen
peroxide molecules and could consequently change the extent of
coordination of hydrogen peroxide to the iron(III) cations. Evi-
dence for the effects of the solvent on the electronic properties
1
of hydrogen peroxide was obtained from H NMR. For these
Fig. 4. The observed rate constant for (F20TPP)Fe as a function of the methanol
concentration. Cyclooctene, hydrogen peroxide, and porphyrin concentrations
were 0.72 M, 16 mM, and 75 ␮M, respectively. Dichloromethane concentration
varied to maintain constant volume.
experiments, spectra were taken of hydrogen peroxide (14 mM)
in each solvent. The observed shifts of the hydrogen peroxide
proton peak were 7.4, 9.3, and 11.3 ppm for dichloromethane,
acetonitrile, and methanol, respectively. The greater extent of
deshielding, as characterized by the downfield shift, indicates
that the nucleophilicity of the hydrogen peroxide molecule
decreases in the order of CH2Cl2 > CH3CN > CH3OH. A less
nucleophilic hydrogen peroxide molecule would be expected to
interact to a lesser extent with the Lewis acidic iron(III) cation.
Therefore, the increased nucleophilicity of the hydrogen per-
oxide molecule in the presence of dichloromethane/methanol
relative to acetonitrile/methanol explains the observed increase
in the apparent equilibrium constant for reaction (2).
Experiments were also carried out to determine the effects
of the methanol concentration on K2. For these experiments,
the total solvent volume was held constant at 3.0 mL while the
relative amounts of methanol and dichloromethane were var-
ied. As shown in Fig. 4, the observed rate constant increased
to a maximum and then decreased. As seen in Eq. (2), kobs is
dependent on the concentration of methanol-coordinated por-
phyrin species, the concentration of methanol, and the yield,
all of which increase with increasing methanol concentration.
Inspection of Eqs. (2) and (4) leads to the expectation that kobs
should increase over the entire range of methanol concentra-
tion shown in Fig. 4. The observation that kobs increases to a
maximum and then decreases suggests that another parameter
in Eq. (2) is affected by the relative concentrations of methanol
and dichloromethane. As discussed in the previous paragraph,
one possibility is that the equilibrium constant K2 is affected
by the solvent composition. As the methanol concentration
increases, the extent of hydrogen bonding between hydrogen
peroxide and methanol increases, decreasing the nucleophilic-
ity of hydrogen peroxide and thereby reducing the value of K2.
Thus, the explicit dependence on methanol concentration result-
ing from Eqs. (2) and (4), which should lead to a monotonic
increase in kobs, is offset by the decrease in K2 with increas-
ing methanol concentration. The combined effects lead to a
maximum observed rate constant near a methanol concentration
of 10 M.
There is reason to believe that the decrease of K2 with increas-
ing methanol concentration becomes significant only above
6 M. Fig. 5 shows that when kobs is plotted versus the product
of the methanol concentration and divided by the methanol-
coordinated porphyrin concentration to the final yield a straight
line is obtained with a slope equal k3K2. Linearity of the plot over
the range of 0–6 M methanol indicates that K2 is not affected sig-
nificantly at low methanol concentrations. Note that the lower
data points fall slightly below the best-fit line in Fig. 5; this
is likely a result of underestimation of the observed rate con-
stant due to excessive porphyrin degradation at low methanol
concentrations.
The composition of the aprotic solvent was also observed to
affect the competition between reactions (6) and (7) for the pi-
radicalcationspecies. Theratioof k6/k7 wasdeterminedtobe0.3
Fig. 5. Observed rate constant for (F20TPP)Fe obeys Eq. (2) at low methanol
concentrations. Cyclooctene, hydrogen peroxide, and porphyrin concentrations
were 0.72 M, 16 mM, and 75 ␮M, respectively. Dichloromethane concentration
varied to maintain constant volume.

114
N.A. Stephenson, A.T. Bell / Journal of Molecular Catalysis A: Chemical 272 (2007) 108–117
in acetonitrile/methanol and 0.5 in dichloromethane/methanol.
This observation can also be explained by the relative nucle-
ophilicity of hydrogen peroxide in each of the solvent systems.
The increase in the nucleophilicity of hydrogen peroxide
in dichloromethane/methanol relative to acetonitrile/methanol
makes it more reactive with the Lewis acidic pi-radical cation
iron porphyrin species, thus increasing k6/k7.
chloride anions dissociate completely upon the formation of
either species:
[Fe–MeOH+][Cl−]
K1 =
(5)
[Fe–Cl][MeOH]
As evidenced by a larger value of K1, chloride dissocia-
tion from (TPP)FeCl becomes easier if halogen substituents
are added to the ortho-positions of the phenyl groups, and
chlorine substituents have a larger effect than do fluorine sub-
stituents. Previous research has noted a similar affect for the
binding of N-methyl-imidizole to ortho-substituted porphyrins
[14]. Those results and the present study suggest that size effects
are more important than through-bond electronic effects. The
increase in K1 with increasing size of the ortho-halogen sub-
stituents is likely due to increased overlap of the electron clouds
of the ortho-halogens with the pi-electron system of the por-
phyrin ring, resulting in electron donation toward the porphyrin
ring. Increased electron donation back to the iron cation would
be expected to decrease the Lewis acidity of the iron cation
and hence weaken the iron–chloride bond [14]. The Fe3+/Fe2+
reduction potential can be used as an indicator of the relative
electronegativity of iron porphyrins; increased electronegativity
of the porphyrin complex leads to a shift in the reduction poten-
tial to more positive values [3]. As seen in the first row of Table 1,
the reduction potential of (2,6-Cl8TPP)FeCl is more negative
than (2,6-F8TPP)FeCl, indicating that (2,6-Cl8TPP)FeCl has a
lower Lewis acidity and, hence, a weaker iron–chloride bond.
While ortho-halogenation of tetra-aryl porphyrins promotes
dissociation of the iron(III) porphyrin chloride salt, fluorine sub-
stituents on the meta- and para-positions decrease the value
of K1. Electronegative groups in the meta- and para-positions
should result exclusively in electron withdrawal from the por-
phyrin ring, increasing the Lewis acidity of the iron cation, and
thus increase the strength of the iron–chloride bond. As expected
from increased electron-withdrawal, two fluorine substituents in
the meta-positions make dissociation of the chloride ligand more
difficult than does one fluorine substituent in the para-position.
(TPP)FeCl and (F20TPP)FeCl dissociate to a similar extent.
However, the values reported for the Fe3+/2+ reduction potential
in Table 1 are very different. This inconsistency can be explained
by the fact that the reduction potential values taken from litera-
ture [23,24] were measured in different solvents. The reduction
potential for (TPP)FeCl was measured in dichloromethane,
while the reduction potentials of the other porphyrins were
measured in a 1:1 mixture of dichloromethane and acetonitrile.
Reduction potentials of porphyrins are known to be influenced
by the solvent, and based on studies comparing solvent donor
number and reduction potential of (TPP)FeX porphyrins, the
Fe3+/2+ reduction potential is expected to be approximately
+0.1 V greater in acetonitrile than in dichloromethane [27–29].
This suggests that the reduction potentials for (TPP)FeCl and
(F20TPP)FeCl should be similar in the same solvent, thus
explaining the similar extents of dissociation, which very likely
result from offsetting effects between the electron donation of
fluorinesubstituentsintheortho-positionandtheelectronaccep-
tance of fluorine substituents in the meta- and para-positions.
4.2. Effects of phenyl group halogenation on the kinetics of
cyclooctene epoxidation by iron porphyrins
Five other iron porphyrins (see Fig. 2) were studied in addi-
tion to (F20TPP)FeCl. Numerous experiments were carried out
in which porphyrin, hydrogen peroxide, andsubstrate concentra-
tion were varied in order to determine whether the epoxidation of
cyclooctene using these catalysts follows the mechanism shown
in Fig. 1. As will be discussed later, several of the porphyrin
catalysts formed ␮-oxo dimers as the porphyrin concentra-
tion was increased, resulting in a decreased yield of epoxide
with increasing porphyrin concentration. To determine indi-
vidual rate parameters independent of ␮-oxo dimer formation
for (TPP)FeCl, kinetic data were obtained at a low porphyrin
concentration (75 ␮M). However, even at low porphyrin con-
centrations, ␮-oxo dimerization occurred to a significant extent
for (3,5-F8TPP)FeCl and (4-F4TPP)FeCl. Therefore, individ-
ual rate parameters were not determined for these catalysts.
The epoxide yield for (2,6-Cl8TPP)FeCl also decreased with
increasing porphyrin concentration. Since it is well known
that (2,6-Cl8TPP)FeCl does not form ␮-oxo dimers [25,26],
another reaction pathway involving multiple equivalents of (2,6-
Cl8TPP)FeCl must contribute to the consumption of hydrogen
peroxide at high porphyrin concentrations. To minimize the
contribution of any such pathway, kinetic parameters were
determined for (2,6-Cl8TPP)FeCl at a low porphyrin concentra-
tion (25 ␮M). Additional details of the individual experiments
for (TPP)FeCl, (2,6-Cl8TPP)FeCl, (2,6-F8TPP)FeCl, and (4-
F4TPP)FeCl are provided as supplemental information. The
experimental data for (F20TPP)FeCl and (2,6-F8TPP)FeCl are
consistent with the mechanism shown in Fig. 1 and the kinetics
described by Eqs. (1)–(4) over the entire range of experimen-
tal conditions, while the data collected for (TPP)FeCl and
(2,6-Cl8TPP)FeCl are consistent with the mechanism at low
porphyrin concentrations.
4.2.1. Influence of phenyl group composition on the
dissociation of iron(III) porphyrin chloride
Using Eq. (5), the equilibrium constant, K1, for the dis-
sociation of the chloride ligand can be determined via 1H
NMR [18]. Methanol-coordinated and chloride-coordinated
porphyrin species are represented by [Fe–MeOH] and [Fe–Cl],
respectively. The value of K1 (see Fig. 1) in a 1:3 methanol:
dichloromethane mixture is given for each porphyrin catalyst in
Table 1. For porphyrin catalysts resulting in the formation of
␮-oxo dimers, the concentration of chloride anions in solution
was calculated as the sum of the concentration of the methanol-
coordinated porphyrin species and two times the concentration
of the ␮-oxo dimer species, based on the assumption that the

N.A. Stephenson, A.T. Bell / Journal of Molecular Catalysis A: Chemical 272 (2007) 108–117
115
4.2.2. Effect of iron(III) porphyrin dissociation on the
Lewis acidity of iron(III) porphyrin cations
tivity and the electronic properties of the porphyrin is much
more complex. In particular, there is no difference in selectivity
between (F20TPP)FeCl and (2,6-F8TPP)FeCl even though the
former porphyrin is much more electronegative and exhibits a
higher reduction potential. Therefore, the ratio of heterolytic to
homolytic cleavage must also be affected by some other factor
such as, for example, the conformation of the porphyrin ring.
The size and position of the halogen substituents on the phenyl
rings likely affects the angle between the phenyl plane and the
porphyrin plane as well as the planarity of the porphyrin ring
[33], a structural change which would be expected to affect
the electronic properties of hydrogen peroxide coordinated to
the iron cation. Since a molecule of alcohol is involved in the
heterolytic cleavage of the oxygen–oxygen bond of hydrogen
peroxide [19], changing the conformation of the porphyrin may
also change the interaction of the coordinated hydrogen perox-
ide with the alcohol molecule facilitating heterolytic cleavage of
the peroxide O–O, thus influencing the rate constant for reaction
(3).
The absolute values for k3K2 and k4K2 are also shown in
Table1. Theextentofinteractionbetweentheiron(III)cationand
hydrogen peroxide is expected to increase as the iron(III) cation
becomes more Lewis acidic, thus increasing K2. The meta- and
para-fluorines of (F20TPP)Fe+ are strongly electron withdraw-
ing, which would be expected to increase K2 and would explain
in part the large values of k3K2 and k4K2 for (F20TPP)Fe rela-
tive to the other porphyrins. The orientation of the phenyl rings
relative to the porphyrin plane and the degree of planarity of the
porphyrin plane may also affect the extent of coordination of the
hydrogen peroxide molecule to the iron(III) cation. However, the
inability to isolate the hydrogen peroxide-coordinated porphyrin
species makes it impossible to decouple the rate parameter K2
from k3 and k4. Theoretical studies are needed to better under-
stand the effects of the nature of the solvent, the electronic nature
of the porphyrin, and the physical structure of the porphyrin
on the coordination of hydrogen peroxide to the iron(III) por-
phyrincationandthesubsequentcleavageoftheoxygen–oxygen
bond. This would allow the rate parameter K2 to be decou-
pled from the rate parameters for homolytic and heterolytic
cleavage.
It is evident from the reduction potentials reported in
the first and second rows of Table 1 that the iron cation
becomes more electronegative after the chloride anion disso-
ciates. It is also noted that the reduction potential
increases in the order of (TPP)Fe+ < (2,6-Cl8TPP)Fe+ < (2,6-
F8TPP)Fe+ < F20TPP)Fe+ once the chloride ligand is removed,
indicating that the ortho-halogen substituents act as electron-
donors when the chloride anion is coordinated and as
electron-acceptors when chloride is dissociated. The change in
function by the ortho-halogens may be a result of the struc-
tural conformation of the porphyrin, since porphyrin reduction
potentials are affected by a combination of electronic and struc-
tural factors [30]. Chloride-coordination to the iron(III) cation
causes the iron atom to be displaced out of the porphyrin plane,
which affects the conformation and electronic nature of the por-
phyrin catalyst [31]. This structural change could increase the
interaction between the electron clouds of the ortho-halogens
and the pi-electron cloud of the porphyrin ring. Dissociation of
the chloride ligand may allow the iron(III) cation to recede back
into the porphyrin ring, which could reduce the extent of electron
cloud overlap such that electron-withdrawing inductive effects
dominate. Consistent with this idea, local density function
calculations of free-base tetraphenyl porphyrins indicate that
the ortho-halogens of both (2,6-Cl8TPP)2− and (2,6-F8TPP)2−
exhibit electron-withdrawing effects, albeit to a lesser extent
than para- or meta-substituents [32].
4.2.3. Effects of phenyl group composition on the relative
rates of heterolytic versus homolytic cleavage of O–O bonds
of the coordinated H2O2
Once a molecule of hydrogen peroxide has coordinated
to the iron(III) porphyrin cation, the oxygen–oxygen bond
of the coordinated hydrogen peroxide molecule can cleave
either homolytically to create a one-electron oxidized iron(IV)-
hydroxo species or heterolytically to create a two-electron
oxidized iron(IV) pi-radical cationic species (see Fig. 1) [3,15].
Homolytic cleavage leads exclusively to peroxide decomposi-
tion, while the iron(IV) pi-radical cation created via heterolytic
cleavage can either decompose hydrogen peroxide or epoxi-
dize an olefin. Nam et al. have presented a qualitative model
that shows that the cleavage of the oxygen–oxygen bond of the
coordinated hydrogen peroxide molecule proceeds via both het-
erolytic and homolytic cleavage and that the ratio of heterolytic
to homolytic cleavage increases as the electron-withdrawing
ability of the porphyrin ligand increases [3].
As discussed above, all halogen substituents on the phenyl
rings act as electron-withdrawing groups after dissociation of
the chloride ligand. This pattern is observed by looking at the
Fe3+/Fe2+ reduction potential of the iron(III) porphyrin cation.
A positive shift in the reduction potential is indicative of a
more electron-deficient porphyrin complex and, hence, should
increase the oxidation potential of the porphyrin macrocycle
[3,30]. While the data in Table 1 do show, in general, that
the selectivity towards heterolytic cleavage (k3/k4) increases
with the reduction potential, the relationship between selec-
4.2.4. Effects of porphyrin composition on the selectivity of
iron(IV) pi-radical cation species for cyclooctene
epoxidation
H2O2 selectivity for epoxidation is determined by com-
petition between hydrogen peroxide and olefin for reaction
with the iron(IV) pi-radical cation, Reactions (6) and (7). As
shown previously, increasing electron-donation from the axial
ligand to the iron(IV) pi-radical cation species decreases the
selectivity towards epoxidation [19]. Similarly, decreasing the
electronegativity of the porphyrin ligand, as indicated by the
reduction potential, decreases the rate of epoxidation (reaction
(6)) relative to the rate of hydrogen peroxide decomposition
(reaction (7)). This finding is consistent with the findings of
Nam et al. [17]. Two reasons have been suggested to explain
the increased selectivity towards epoxidation that occurs as
electron density is withdrawn from the porphyrin ring. First,

116
N.A. Stephenson, A.T. Bell / Journal of Molecular Catalysis A: Chemical 272 (2007) 108–117
iron(IV) pi-radical cations of electron-deficient porphyrins may
exhibit a greater propensity for electron-transfer to an olefin than
for hydrogen abstraction from a peroxide molecule [16]. Sec-
ond, peroxide decomposition may be diffusion controlled while
epoxidation is not [16]. Thus, increasing the electronegativity
of the pi-radical cation would increase the rate of epoxidation
but not the rate of peroxide decomposition, since it is diffusion
controlled.
(TPP)FeCl dissolved in methanol/dichloromethane confirmed
that all three porphyrins do indeed form ␮-oxo dimers [35,36].
It was also noted that the extent of dimerization increased in the
order of (3,5-F8TPP)FeCl > (4-F4TPP)FeCl > (TPP)FeCl, indi-
cating that dimerization occurs more readily for electronegative
porphyrin species. The formation of ␮-oxo dimers was minimal
for low concentrations of (TPP)FeCl. Therefore, it was possible
to make a meaningful comparison of the reactivity and selectiv-
ity of (TPP)FeCl to other iron porphyrin catalysts by obtaining
kinetic data at low porphyrin concentrations. The details of these
experiments and the subsequent data analysis are provided as
supplemental information. On the other hand, the dimerization
of (3,5-F8TPP)FeCl and (4-F4TPP)FeCl occurred to a signif-
icant extent, even at low porphyrin concentrations, making it
impossible to carry out a similar study of kinetic activity for
those two catalysts.
4.2.5. Porphyrin degradation
Degradation of the porphyrin catalyst is thought to occur
via attack of the porphyrin ring by hydroxyl radicals [15,34].
Electron-withdrawing substituents increase the reduction poten-
tial of the porphyrin, thus protecting the catalyst from oxidative
destruction [30]. Therefore, substitution of halogens for hydro-
gens on the phenyl ring should increase the robustness of the
porphyrin catalyst. As seen in Table 1, an increase in the reduc-
tion potential does indeed reduce the catalyst’s susceptibility to
degradation.
4.2.7. Catalytic activity of the μ-oxo dimmer
To understand the effects of ␮-oxo dimers upon catalytic
activity, a sample of the (3,5-F8TPP) ␮-oxo dimer was obtained
and studied independently of the monomeric form. Analysis
by H NMR confirmed that only ␮-oxo dimer species were
4.2.6. Formation of μ-oxo dimmers
1
The mechanism shown in Fig. 1 suggests that the yield of
epoxide should be independent of the porphyrin concentration.
While this expectation was borne out for (F20TPP)Fe and (2,6-
F8TPP)Fe, reactions carried out using (TPP)Fe, (3,5-F8TPP)Fe,
and (4-F4TPP)Fe as catalyst showed a decrease in yield with
increasing porphyrin concentration; see Fig. 6 for example. The
results seen in this figure suggest that an additional mechanism,
involving multiple equivalents of the porphyrin catalyst con-
tributed to hydrogen peroxide decomposition when (TPP)Fe,
(3,5-F8TPP)Fe, or (4-F4TPP)Fe was used as the catalyst. A
possible explanation is that the porphyrins lacking halogen sub-
stituents at the ortho-positions of the phenyl groups form ␮-oxo
dimers, which are effective for catalyzing the decomposition
of hydrogen peroxide but not for catalyzing the epoxidation of
olefins.
present when dissolved in dichloromethane (see Fig. 7). Upon
addition of methanol (5.6 M) to the sample, a second por-
phyrin species, characterized by a ␤-pyrrole peak at 82 ppm,
was observed. This suggests that the formation of the ␮-oxo
dimer species is reversible. The peak location of the non-
dimeric species is indicative of an iron porphyrin with a strongly
bound axial ligand [35–38], most likely a hydroxo ligand.
Hydroxo-coordinated porphyrins are well known to form ␮-oxo
dimers readily [36,39–41]. Also, as seen in Fig. 7, methanol-
coordinated species, characterized by a peak near 55 ppm, are
not formed by the interaction of the ␮-oxo dimer with methanol.
As shown previously, iron porphyrins with strongly bound axial
ligands do not react with hydrogen peroxide [18,19,22,42,43].
Therefore, any observed reaction must result of reaction from
the hydrogen peroxide with the ␮-oxo dimer species. Reac-
The presence of a ␤-pyrrole resonance near 14 ppm in the
1H NMR spectra of (3,5-F8TPP)FeCl, (4-F4TPP)FeCl, and
Fig. 6. Dependence of epoxide yield on total porphyrin concentration
cyclooctene, hydrogen peroxide, and methanol concentrations were 3.8 M,
5.9 mM, and 5.6 M, respectively.
Fig. 7. 1H NMR spectra of (3,5-F8TPP)Fe–O–Fe(3,5-F8TPP) before and after
the addition of methanol (5.6 M).

N.A. Stephenson, A.T. Bell / Journal of Molecular Catalysis A: Chemical 272 (2007) 108–117
117
tivity studies of the ␮-oxo dimer of (3,5-F8TPP)Fe in the
References
presence of methanol revealed that the dimer decomposes
hydrogen peroxide rapidly but is inefficient for the epoxida-
tion of olefin. Therefore, the formation of the ␮-oxo dimer
when the monomeric (3,5-F8TPP)FeCl is dissolved in methanol
will result in reduced yields of epoxide. This is consistent
with the proposal that the reduction in yield with increas-
ing porphyrin concentration of (TPP)FeCl, (3,5-F8TPP)FeCl,
and (4-F4TPP)FeCl is due to the formation of ␮-oxo
dimers.
[1] J.T. Groves, T.E. Nemo, R.S. Myers, J. Am. Chem. Soc. 101 (1979)
1032–1033.
[2] W. Nam, S.Y. Oh, Y.J. Sun, J. Kim, W.K. Kim, S.K. Woo, W. Shin, J. Org.
Chem. 68 (2003) 7903–7906.
[3] W. Nam, H. Han, S. Oh, Y.J. Lee, M. Choi, S. Han, C. Kim, S. Woo, W.
Shin, J. Am. Chem. Soc. 122 (2000) 8677–8684.
[4] I.D. Cunningham, T.N. Danks, J.N. Hay, I. Hamerton, S. Gunathilagan, C.
Jankczak, J. Mol. Catal. A: Chem. 185 (2002) 25–31.
[5] E. Porhiel, A. Bondon, J. Leroy, Tetraherdon Lett. 39 (1998) 4829–4830.
[6] D. Mansuy, Coord. Chem. Rev. 125 (1993) 129–141.
[7] D. Mansuy, Pure Appl. Chem. 59 (1987) 759–770.
5. Conclusions
[8] P.A. Adams, Preox. Chem. Biol. 2 (1991) 171–200.
[9] D. Dolphin, T.G. Traylor, L.Y. Xie, Acc. Chem. Res. 30 (1997) 251–259.
[10] R.A. Sheldon, Metalloporphyrins in Catalytic Oxidations, Marcel Dekker,
New York, 1994.
The results of this study have shown that the composition of
iron(III) porphyrins has a strong influence on the activity and
selectivity of these materials for olefin epoxidation by hydro-
gen peroxide versus decomposition of hydrogen peroxide. The
kinetics of olefin epoxidation and hydrogen peroxide decompo-
sition are well described by the mechanism presented in Fig. 1
for all of the catalysts investigated (see Table 1) provided that
conditions are avoided under which the porphyrin forms ␮-oxo
dimers. We find that the position and number of halogen atoms
present on the phenyl groups bonded to the porphyrin affect
the properties of the porphyrins in a number of ways. Increasing
the electron-withdrawing character of the phenyl groups inhibits
the dissociation of iron(III) porphyrin chloride (reaction (1)) but
increases the Lewis acidity of the resulting iron(III) porphyrin
cation and the extent of its coordination to hydrogen perox-
ide (reaction (2)). Higher electronegativity of the phenyl groups
attached to the porphyrin ring also causes an increase in the
ratio of the rate coefficients for reactions (3) and (4), as well
as a decrease in the ratio of the rate coefficients for reactions
(6) and (7). A rise the ratio of k3 to k4 is associated with an
increase in the rate of heterolytic relative to homolytic cleavage
of the O–O bond of coordinated hydrogen peroxide. Finally,
we have shown, for the first time, that the extent of hydro-
gen peroxide coordination to the iron(III) cation in reaction
(2) is affected significantly by the intermolecular interactions
between the solvent and hydrogen peroxide molecules. As a
result of these combined effects, the most electronegative por-
phyrin (F20TPP)FeCl, dissolved in a methanol/dichloromethane
solvent mixture, is found to be the most active and selective
catalyst for the epoxidation of cyclooctene via hydrogen perox-
ide.
[11] Z. Gross, L. Simkhovich, Tetrahedron Lett. 39 (1998) 8171–8174.
[12] A.A. Guedes, J.R.L. Smith, O.R. Nascimento, D.F.C. Guedes, M.D. Assis,
J. Braz. Chem. Soc. 16 (2005) 835–843.
[13] E. Baciocchi, T. Boshci, L. Cassioli, C. Galli, A. Lapi, P. Tagliatesta,
Tetrahedron Lett. 38 (1997) 7283–7286.
[14] R. Koerner, J.L. Wright, X.D. Ding, M.J.M. Nesset, K. Aubrecht, R.A.
Watson, R.A. Barber, L.M. Mink, A.R. Tipton, C.J. Norvell, K. Skidmore,
U. Simonis, F.A. Walker, Inorg. Chem. 37 (1998) 733–745.
[15] N.A. Stephenson, A.T. Bell, J. Am. Chem. Soc. 127 (2005) 8635–8643.
[16] T.D. Traylor, S. Tsuchiya, Y.S. Byun, C. Kim, J. Am. Chem. Soc. 115
(1993) 2775–2781.
[17] M.G. Yeong, W. Nam, Inorg. Chem. 38 (1999) 914–920.
[18] N.A. Stephenson, A.T. Bell, Inorg. Chem. 45 (2006) 5591–5599.
[19] N.A. Stephenson, A.T. Bell, Inorg. Chem. 45 (2006) 2758–2766.
[20] N.A. Stephenson, A.T. Bell, J. Mol. Catal. A 258 (2006) 231–235.
[21] N.A. Stephenson, A.T. Bell, Anal. Bioanal. Chem. 381 (2005) 1289–1293.
[22] N.A. Stephenson, A.T. Bell, Inorg. Chem. 46 (2007) 2278–2285.
[23] M.H. Lim, Y.J. Lee, Y.M. Goh, W. Nam, C. Kim, Bull. Chem. Soc. Jpn. 72
(1999) 707–713.
[24] J.E. Lyons, P.E. Ellis, H.K. Myers, J. Catal. 155 (1995) 59–73.
[25] B. Meunier, Biomimetic Oxidations Catalyzed by Transition Metal Com-
plexes, Imperial College Press, London, 1999.
[26] P.S. Traylor, D. Dolphin, T.G. Traylor, J. Chem. Soc. Chem. Commun.
(1984) 279–280.
[27] K.M. Kadish, J. Electroanal. Chem. 108 (1984) 261–274.
[28] P. O’Brien, D.A. Sweigart, Inorg. Chem. 24 (1985) 1405–1409.
[29] L.A. Bottomley, K.M. Kadish, Inorg. Chem. 20 (1981) 1348–1357.
[30] A. Ghosh, J. Am. Chem. Soc. 117 (1995) 4691–4699.
[31] M.S. Liao, S. Scheiner, J. Comput. Chem. 23 (2002) 1391–1403.
[32] A. Ghosh, J. Mol. Struct. (Theo.) 388 (1996) 359–363.
[33] A. Rosa, G. Ricciardi, E.J. Baerends, J. Phys. Chem. A 110 (2006)
5180–5190.
[34] I.D. Cunningham, T.N. Danks, J.N. Hay, I. Hamerton, S. Gunathilagan,
Tetrahedron 57 (2001) 6847–6853.
[35] E.R. Birnbaum, J.A. Hodge, M.W. Grinstaff, W.P. Schaefer, L. Henling,
J.A. Labinger, J.E. Bercaw, H.B. Gray, Inorg. Chem. 34 (1995) 3625–3632.
[36] R.J. Cheng, L. Latos-Grazynski, A.L. Balch, Inorg. Chem. 21 (1982)
2412–2418.
Acknowledgment
[37] T.C. Woon, A. Shirazi, T.C. Bruice, Inorg. Chem. 25 (1986) 3845–3846.
[38] J.T. Groves, R. Quinn, T.J. McMurry, M. Nakamura, G. Lang, B. Boso, J.
Am. Chem. Soc. 107 (1985) 354–360.
[39] J.W. Buchler, Angew. Chem. Int. Ed. 17 (1978) 407–423.
[40] L. Fielding, G.R. Eaton, S.S. Eaton, Inorg. Chem. 24 (1985) 2309–2312.
[41] K.M. More, G.R. Eaton, S.S. Eaton, Inorg. Chem. 24 (1985) 3698–3702.
[42] W. Nam, M.H. Lim, S.Y. Oh, J.H. Lee, H.J. Lee, S.K. Woo, C. Kim, W.
Shin, Angew. Chem. Int. Ed. 39 (2000) 3646–3649.
This work was supported by the Director, Office of Basic
Energy Sciences, Chemical Sciences Division of the U.S.
Department of Energy under Contract DE-AC02-05CH11231.
Appendix A. Supplementary data
[43] W. Nam, H.J. Lee, S.Y. Oh, C. Kim, H.G. Jang, J. Inorg. Biochem. 80
(2000) 219–225.
Supplementary data associated with this article can be found,
in the online version, at doi:10.1016/j.molcata.2007.03.030.