Effects of methanol on the thermodynamics of iron(III) [tetrakis(pentafluorophenyl)]porphyrin chloride dissociation and the creation of catalytically active species for the epoxidation of cyclooctene

Article abstract of DOI:10.1021/ic0521067

In a previous study, the authors showed that iron(III) [tetrakis(pentafluorophenyl)]porphyrin chloride [(F₂₀TPP)FeCl] is catalytically inactive for cyclooctene epoxidation by hydrogen peroxide in acetonitrile but is catalytically active if the solvent contains methanol. It was suggested that the precursor to the active species is (F₂₀TPP) Fe(OCH₃) in methanol-containing solvents. The present study was aimed at evaluating this hypothesis. (F₂₀TPP)Fe(OCH₃) was synthesized and characterized by ¹H NMR but was found to be inactive in both acetonitrile and methanol. Further investigation of the interactions of (F₂₀TPP)FeCl with methanol in acetonitrile/methanol mixtures was then carried out using NMR. Two species, characterized by ¹H NMR resonances at 82 and 65 ppm, were observed. The first resonance is attributed to the β-pyrrole protons on molecularly dissolved (F₂₀TPP)FeCl, whereas the second is attributed to β-pyrrole protons of [(F ₂₀TPP)Fe]⁺ cations that are stabilized by coordination with a molecule of methanol, viz., [(F₂₀TPP)Fe(CH₃OH) ⁺. The relative concentration of [(F₂₀TPP)Fe(CH ₃OH)]⁺ increases as the fraction of methanol in the solvent increases, suggesting that methanol facilitates the dissociation of (F₂₀TPP)FeCl into cations and anions. A thermodynamic model of the dissociation is proposed and found to describe successfully the experimental observation over a range of solvent compositions, porphyrin concentrations, and temperatures. UV-visible spectroscopy was also used to validate the developed model. In addition, the observed rate constant for cyclooctene epoxidation was found to be proportional to the concentration of [(F₂₀TPP)Fe(CH ₃OH)]⁺ calculated using the thermodynamic model, suggesting that this intermediate is a precursor to the species that catalyzes olefin epoxidation. The catalytic activity of [(F₂₀TPP)Fe(CH ₃OH)]⁺ was further confirmed through experiments in which (F₂₀TPP)Fe-(OCH₃) dissolved in methanol was reacted with HCl(aq). This reaction produced a product with an NMR peak at 65 ppm attributable to [(F₂₀TPP)Fe(CH₃OH)]⁺, and this mixture was found to have activity for cyclooctene epoxidation similar to that of (F₂₀TPP)FeCl dissolved in methanol.

Full text of DOI:10.1021/ic0521067

Inorg. Chem. 2006, 45, 5591−5599
Effects of Methanol on the Thermodynamics of Iron(III)
[Tetrakis(pentafluorophenyl)]porphyrin Chloride Dissociation and the
Creation of Catalytically Active Species for the Epoxidation of
Cyclooctene
Ned A. Stephenson and Alexis T. Bell*
Chemical Sciences DiVision, Lawrence Berkeley Laboratory, and
Department of Chemical Engineering, UniVersity of California, Berkeley, California 94720-1462
Received December 8, 2005
In a previous study, the authors showed that iron(III) [tetrakis(pentafluorophenyl)]porphyrin chloride [(F20TPP)FeCl]
is catalytically inactive for cyclooctene epoxidation by hydrogen peroxide in acetonitrile but is catalytically active if
the solvent contains methanol. It was suggested that the precursor to the active species is (F20TPP)Fe(OCH3) in
methanol-containing solvents. The present study was aimed at evaluating this hypothesis. (F20TPP)Fe(OCH3) was
1
synthesized and characterized by H NMR but was found to be inactive in both acetonitrile and methanol. Further
investigation of the interactions of (F20TPP)FeCl with methanol in acetonitrile/methanol mixtures was then carried
out using NMR. Two species, characterized by 1H NMR resonances at 82 and 65 ppm, were observed. The first
resonance is attributed to the
â-pyrrole protons on molecularly dissolved (F20TPP)FeCl, whereas the second is
attributed to
â
-pyrrole protons of [(F20TPP)Fe]+ cations that are stabilized by coordination with a molecule of methanol,
viz., [(F20TPP)Fe(CH3OH)]+. The relative concentration of [(F20TPP)Fe(CH3OH)]+ increases as the fraction of methanol
in the solvent increases, suggesting that methanol facilitates the dissociation of (F20TPP)FeCl into cations and
anions. A thermodynamic model of the dissociation is proposed and found to describe successfully the experimental
observation over a range of solvent compositions, porphyrin concentrations, and temperatures. UV−visible
spectroscopy was also used to validate the developed model. In addition, the observed rate constant for cyclooctene
epoxidation was found to be proportional to the concentration of [(F20TPP)Fe(CH3OH)]+ calculated using the
thermodynamic model, suggesting that this intermediate is a precursor to the species that catalyzes olefin epoxidation.
The catalytic activity of [(F20TPP)Fe(CH3OH)]+ was further confirmed through experiments in which (F20TPP)Fe-
(OCH3) dissolved in methanol was reacted with HCl(aq). This reaction produced a product with an NMR peak at
65 ppm attributable to [(F20TPP)Fe(CH3OH)]+, and this mixture was found to have activity for cyclooctene epoxidation
similar to that of (F20TPP)FeCl dissolved in methanol.
Introduction
compensated by a strongly bound ligand, such as Cl-, the
porphyrin is totally inactive when dissolved in an aprotic
solvent, such as acetonitrile, but becomes active in a mixture
of protic and aprotic solvents.12-21 On the other hand, when
Iron(III) porphyrins are highly active catalysts for the
oxidation of various organic substrates by either hydrogen
peroxide or organic peroxides, and as a result, considerable
attention has been devoted to understanding the effects of
the porphyrin and solvent compositions on the mechanism
and kinetics of such reactions.1-22 These studies have
revealed that when the iron(III) porphyrin cation is charge-
(2) Montanari, F.; Casella, L. Metalloporphyrins Catalyzed Oxidations;
Kluwer Academic Publishers: Boston, 1994.
(3) Meunier, B. Biomimetic Oxidations Catalyzed by Transition Metal
Complexes; Imperial College Press: London, 1999.
(4) Bruice, T. C. Acc. Chem. Res. 1991, 24, 243-249.
(5) Traylor, T. G.; Kim, C.; Fann, W. P.; Perrin, C. L. Tetrahedron 1998,
54, 7977-7986.
(6) Fujii, H. Chem. Lett. 1994, 1491-1494.
(7) Almarsson, O.; Bruice, T. C. J. Am. Chem. Soc. 1995, 117, 4533-
4544.
* To whom correspondence should be addressed. E-mail:
bell@cchem.berkeley.edu.
(1) Sheldon, R. A. Metalloporphyrins in Catalytic Oxidations; Marcel
Dekker: New York, 1994.
10.1021/ic0521067 CCC: $33.50
Published on Web 06/13/2006
© 2006 American Chemical Society
Inorganic Chemistry, Vol. 45, No. 14, 2006 5591

Stephenson and Bell
the iron(III) porphyrin is charge-compensated by a weakly
bound anion, such as CF3SO3 , the porphyrin is active even
TPP)Fe(OCH3), or some other porphyrin species represents
the precursor to the catalytically active form of (F20TPP)Fe.
To this end, we synthesized (F20TPP)Fe(OCH3) and exam-
ined its ability to catalyze the epoxidation of cyclooctene
by hydrogen peroxide. Because (F20TPP)Fe(OCH3) was
found to be catalytically inactive, the nature of the active
species formed upon dissolution of (F20TPP)FeCl in a
mixture of methanol and acetonitrile was investigated in more
detail. The results of these studies led to the conclusion that
the precursor to the active intermediate for epoxidation is
likely to be methanol-coordinated (F20TPP)Fe cations, viz.,
[(F20TPP)Fe(CH3OH)]+. The thermodynamics of forming
these cations is discussed in detail.
-
in aprotic solvents.20,21 While it is recognized that the catalytic
activity of iron(III) porphyrins is affected by the composition
of the axial ligand, the exact nature of the catalytically active
form of the porphyrin is not well understood.17,19-22 More-
over, the relative strength of coordination of iron(III)
porphyrin cations by different anions continues to be a
subject of ongoing discussion.23-25
In a recent investigation of the mechanism and kinetics
of cyclooctene epoxidation by hydrogen peroxide catalyzed
by iron(III) [tetrakis(pentafluorophenyl)]porphyrin chloride
[(F20TPP)FeCl], we observed that (F20TPP)FeCl was inactive
in acetonitrile but became active when a mixture of methanol
and acetonitrile was used as the solvent.26 On the basis of
several observations, it was suggested that activation of the
porphyrin complex involves the replacement of the chloride
anion by a methoxide anion. In particular, the appearance
of the latter species was suggested by the appearance of a
new â-pyrrole proton NMR peak when (F20TPP)FeCl was
dissolved in a mixture of methanol and acetonitrile. Fur-
thermore, a nonlinear relationship was observed between the
rate constant and the total porphyrin concentration. This rela-
tionship was found to be consistent with a proposed equilib-
rium relationship between the chloride- and methoxide-
coordinated forms of the porphyrin, as given by reaction
1.
Experimental Section
Reagents. Non-UV-grade acetonitrile (99.99%), methanol
(99.98%), hydrogen peroxide (30%), and ethylene glycol (98%)
were obtained from EMD Chemicals. Iron(III) [tetrakis(pentafluo-
rophenyl)]porphyrin chloride [(F20TPP)FeCl], tetrabutylammonium
chloride (g97%), and dodecane (99+%) were obtained from Sigma-
Aldrich. cis-Cyclooctene (95%) was obtained from Alfa-Aesar.
Deuterium oxide (99.9%) and deuterated chloroform (99.8%) were
obtained from Cambridge Isotope Laboratories, Inc. Sodium
methoxide (30% in methanol) was obtained from Fluka. Benzene
(99.9%) was obtained from Fisher Scientific. All solvents were used
as purchased without further purification.
Synthesis of Methoxide-Coordinated Porphyrin. Methoxide-
coordinated iron(III) [tetrakis(pentafluorophenyl)]porphyrin was
synthesized using a procedure analogous to that reported for the
synthesis of methoxide-coordinated iron(III) tetraphenylporphyrin.27
(F20TPP)FeCl (100 mg) was dissolved in a 50-mL mixture of
benzene and methanol (5:2 by volume) in a 250-mL round-bottomed
flask. Sodium methoxide in methanol (0.67 mL of a 30% w/w
solution) was added to this mixture, causing it to change from dark
green to dark red. The solution was stirred at room temperature
for 1 h and then reduced to a small volume under vacuum. The
crystals that precipitated were then separated by filtration and
washed with methanol (0 °C). The resulting crystals were dried at
44 °C under 0.72 bar of vacuum. The synthesis resulted in a 97%
yield with a purity of approximately 98% based upon NMR
spectroscopy.
The present investigation was undertaken to establish
whether the methoxide-coordinated porphyrin cation, (F20-
(8) He, G. X.; Bruice, T. C. J. Am. Chem. Soc. 1991, 113, 2747-
2753.
(9) Traylor, T. G.; Fann, W. P.; Bandyopadhyay, D. J. Am. Chem. Soc.
1989, 111, 8009-8010.
(10) Traylor, T. G.; Xu, F. J. Am. Chem. Soc. 1987, 109, 6201-6202.
(11) Nam, W.; Lim, M. H.; Moon, S. K.; Kim, C. J. Am. Chem. Soc. 2000,
122, 10805-10809.
NMR Characterization of Porphyrin Samples. Paramagnetic
1H NMR spectra of the porphyrin species were obtained using a
400-MHz VMX spectrometer. A 90° pulse was used. The spectral
width was increased to 140 ppm, and the spectra were centered at
60 ppm. Because of the high concentration of methanol protons in
the sample, it was necessary to use a low gain. A line-broadening
factor of 75 Hz was used to increase the signal-to-noise ratio. On
the basis of the line-width at half-maximum, the relaxation constant
for the paramagnetic porphyrin peaks was estimated to be a few
milliseconds; therefore, the time delay between scans, D1, was set
to 25 ms. D1 was set to 4 s when studying the axial ligand proton
peaks in the 0-10 ppm range. The number of data points was set
to 4000. The number of scans required to obtain a suitable signal-
to-noise ratio was dependent on the relative intensities of the two
(12) Traylor, T. G.; Tsuchiya, S.; Byun, Y. S.; Kim, C. J. Am. Chem. Soc.
1993, 115, 2775-2781.
(13) Traylor, T. G.; Kim, C.; Richards, J. L.; Xu, F.; Perrin, C. L. J. Am.
Chem. Soc. 1995, 117, 3468-3474.
(14) Nam, W.; Lim, M. H.; Lee, H. J.; Kim, C. J. Am. Chem. Soc. 2000,
122, 6641-6647.
(15) Traylor, T. G.; Xu, F. J. Am. Chem. Soc. 1990, 112, 178-186.
(16) Traylor, T. G.; Ciccone, J. P. J. Am. Chem. Soc. 1989, 111, 8413-
8420.
(17) Nam, W.; Lee, H. J.; Oh, S. Y.; Kim, C.; Jang, H. G. J. Inorg. Biochem.
2000, 80, 219-225.
(18) Lee, K. A.; Nam, W. J. Am. Chem. Soc. 1997, 119, 1916-1922.
(19) Nam, W.; Oh, S. Y.; Sun, Y. J.; Kim, J.; Kim, W. K.; Woo, S. K.;
Shin, W. J. Org. Chem. 2003, 68, 7903-7906.
(20) Nam, W.; Jin, S. W.; Lim, M. H.; Ryu, J. Y.; Kim, C. Inorg. Chem.
2002, 41, 3647-3652.
(21) Nam, W.; Lim, M. H.; Oh, S. Y.; Lee, J. H.; Lee, H. J.; Woo, S. K.;
Kim, C.; Shin, W. Angew. Chem., Int. Ed. 2000, 39, 3646-3649.
(22) Goh, Y. M.; Nam, W. Inorg. Chem. 1999, 38, 914-920.
(23) Hoshino, A.; Nakamura, M. Chem. Lett. 2004, 33, 1234-1235.
(24) Song, B.; Park, B.; Han, C. Bull. Korean Chem. Soc. 2002, 23, 119-
122.
(26) Stephenson, N. A.; Bell, A. T. J. Am. Chem. Soc. 2005, 127, 8635-
8643.
(27) Otsuka, T.; Ohya, T.; Satu, M. Inorg. Chem. 1984, 23, 1777-
(25) Reed, C. A.; Guiset, F. J. Am. Chem. Soc. 1996, 118, 3281-3282.
1779.
5592 Inorganic Chemistry, Vol. 45, No. 14, 2006

Thermodynamics of (F20TPP)FeCl Dissociation
peaks and ranged from 512 to 32 000. The total time to acquire a
spectrum ranged from 1 min to 1 h.
packaged by the manufacturer without further purification or drying.
Reactions were conducted under air because test reactions under
nitrogen with deoxygenated solvents and substrate showed no
difference. The presence of water was considered to be inconse-
quential. Solvents were not rigorously dried because the amount
of water present in the added aqueous hydrogen peroxide solution
is on the order of that present in the solvents. Furthermore, the
addition of water in microliter aliquots was observed to not affect
the kinetics.
Samples were prepared for NMR analysis by dissolving mea-
sured amounts of (F20TPP)FeCl in 400 µL of solvent in precision
NMR tubes. A capillary filled with either D2O or CDCl3 was also
placed in the NMR tube to serve as the locking solvent. For
variable-temperature experiments, the temperature was monitored
by a thermocouple located 2 mm from the tip of the sample tube
and by an ethylene glycol-based NMR thermometer.28 For the
ethylene glycol thermometer, a second capillary with ethylene glycol
was placed in the NMR tube; the presence of two capillaries
required a reduction of the solvent volume to 300 µL. To verify
that the presence of two capillary tubes in the NMR tube does not
affect the readings, separate measurements were also taken using
an NMR tube filled only with ethylene glycol and a single capillary
filled with the lock solvent; no difference in the temperature was
observed between the two methods. Values from both the thermo-
couple and the NMR thermometer were always within 2 °C of each
other; the NMR thermometer temperatures were used for all
calculations. Several minutes were allowed for thermal equilibration
at each temperature before acquiring a spectrum. In addition,
duplicate spectra were taken to verify temperature stability; the
difference between the two spectra was always negligible. Tem-
perature stability was also tested by observing a constant-temper-
ature reading from the NMR thermometer. Analysis of NMR spectra
and peak integration was done in Origin 7.0.
Results and Discussion
Characterization of the Methoxide-Coordinated Spe-
1
cies. The paramagnetic H NMR spectrum of porphyrin
produced by reaction of (F20TPP)FeCl with sodium meth-
oxide showed resonances at 81 and 4.9 ppm when dissolved
in CDCl3. The first of these features is very close to the
position reported previously for â-pyrrole protons of por-
phyrins ligated by strongly bound axial ligands such as
chloride, methoxide, and hydroxide anions,29-32 and the
second feature is characteristic of protons associated with
methoxy groups. The ratio of the peak areas for the peaks at
4.9 and 81 ppm is 0.38 and is very close to the 3:8 ratio
expected for (F20TPP)Fe(OCH3). Both hydroxo and meth-
oxide ligands are considered strong ligands as evidenced by
1H NMR (â-pyrrole shifts near 80 ppm).29-32 If hydroxo-
coordinated species were formed, they would readily convert
to the µ-oxo species.30 A small peak was observed at 14 ppm,
suggesting the presence of µ-oxo species,29 but this species
accounts for less than 2% of the total porphyrin species.
Therefore, the hydroxo-coordinated species very likely are
not formed to any appreciable extent. Thus, on the basis of
NMR analysis, we conclude that virtually pure (F20TPP)Fe-
(OCH3) was prepared.
UV-Visible Spectroscopy. Coordination of methanol to
(F20TPP)FeCl was studied by absorbance experiments using a
Varian Cary 400 Bio UV-visible spectrometer. (F20TPP)FeCl
(25 µM) was dissolved in acetonitrile, and methanol was then
added in increments ranging from 5 to 50 µL. A UV-visible
spectrum was taken after each methanol addition. Reactions with
hydrogen peroxide were also studied in situ by taking scans
approximately every 1 min over the course of 20-30 min after the
addition of hydrogen peroxide. Epoxidation reactions were studied
by combining the catalyst (25 mM final concentration), the solvent
(3.26 mL), cyclooctene (66 µL), and hydrogen peroxide (1 µL
of a 30% w/w aqueous solution) in a 10-mm quartz spectropho-
tometer cell.
Reactivity of the Methoxide-Coordinated Species. The
activity of the (F20TPP)Fe(OCH3)-catalyzed cyclooctene
epoxidation was investigated using hydrogen peroxide as the
oxidant and methanol as the solvent. Upon dissolution of
(F20TPP)Fe(OCH3), the solution turned dark orange. How-
ever, no epoxidation occurred over the course of several
Cyclooctene Reactions. To investigate the epoxidation of
cyclooctene, we combined the catalyst (75 µM final concen-
tration, unless stated otherwise), the solvent (3.0 mL), cyclo-
octene (330 µL) and dodecane (10 µL), as an internal standard, in
a 5-mL glass vial with magnetic stirring. Hydrogen peroxide (5.0
µL of a 30% w/w aqueous solution) was then added via a gastight
syringe to initiate the reactions. At proper times, aliquots (0.5 µL)
from the reaction mixture were sampled and injected into an
HP6890 series gas chromatograph fitted with an Agilent DB Wax
(30 m × 0.32 mm × 0.5 µm) capillary column and a flame
ionization detector. The temperature program for analysis was as
follows: 2.5 min at 50 °C, ramp at 20 °C/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. Dodecane and cyclooctene epoxide
were eluted at 5.5 and 8.2 min, respectively. Data were obtained
in triplicate to ensure precision, and the repeatability of the data
was within (2%. The accuracy of measuring the epoxide concen-
tration via gas chromatography was determined to be (1% by
analysis of a calibrated standard. All chemicals were used as
1
hours. The reaction mixture was analyzed via H NMR
1
spectroscopy after a period of 3.5 h. As evidenced by a H
NMR peak at 10.3 ppm,33 hydrogen peroxide still remained
in the solution. In situ UV-visible spectra of the reaction
mixture showed a monotonic decrease in the intensity of the
entire spectrum but no change in the peak shape as a function
of time after hydrogen peroxide was added to the reaction
mixture. As evidenced by complete bleaching of the por-
(29) Birnbaum, E. R.; Hodge, J. A.; Grinstaff, M. W.; Schaefer, W. P.;
Henling, L.; Labinger, J. A.; Bercaw, J. E.; Gray, H. B. Inorg. Chem.
1995, 34, 3625-3632.
(30) Cheng, R. J.; Latos-Grazynksi, L.; Balch, A. L. Inorg. Chem. 1982,
21, 2412-2418.
(31) Woon, T. C.; Shirazi, A.; Bruice, T. C. Inorg. Chem. 1986, 25, 3845-
3846.
(32) Groves, J. T.; Quinn, R.; McMurry, T. J.; Nakamura, M.; Lang, G.;
Boso, B. J. Am. Chem. Soc. 1985, 107, 354-360.
(33) Stephenson, N. A.; Bell, A. T. Anal. Bioanal. Chem. 2005, 381, 1289-
1293.
(28) Amman, C.; Meier, P.; Merbach, A. E. J. Magn. Reson. 1982, 46,
319-321.
Inorganic Chemistry, Vol. 45, No. 14, 2006 5593

Stephenson and Bell
acetonitrile and pure methanol suggests that the species
responsible for this peak are the same in both solvents and
are not affected significantly by the solvent composition.
Assignment of this peak to molecularly dissolved (F20TPP)-
FeCl is fully consistent with previous studies, which show a
1
â-pyrrole H NMR peak near 80 ppm for iron tetraphen-
ylporphyrins with tightly bound axial ligands.29-32 The exact
position of the peak depends on the composition of the
solvent, but this effect is limited to less than 5 ppm. The
peak near 82 ppm shifts upfield while the peak near 65 ppm
shifts slightly downfield as the methanol concentration is
increased.
As the strength of the interaction of the ligand with the
porphyrin decreases, the position of the â-pyrrole peak shifts
upfield progressively from 80 ppm to as far as -62
ppm.23,25,32 We propose that the peak at 65 ppm is attributable
to [(F20TPP)Fe]+ cations, each of which is weakly coordi-
nated to a molecule of methanol, i.e., [(F20TPP)Fe(CH3-
OH)]+. The formation of such species requires the dissocia-
tion of (F20TPP)FeCl into cations and anions. The endo-
thermicity of this process is offset by the energy released
via the coordination of methanol to the [(F20TPP)Fe]+ cation
and by the solvation of both the resulting [(F20TPP)Fe(CH3-
OH)]+ cation and the Cl- anion. Therefore, in the presence
of methanol, (F20TPP)FeCl is thought to dissociate via
reaction 2.
Figure 1. 1H NMR spectra of (F20TPP)FeCl dissolved in methanol/
acetonitrile solutions. The total porphyrin concentration for each spectrum
was ∼15 mM.
phyrin, the reactions ended because of complete destruction
of the porphyrin and not because of complete consumption
of the oxidant.
The inability of (F20TPP)Fe(OCH3) to catalyze the epoxi-
dation of cyclooctene suggests that iron(IV) π-radical cation
porphyrin species, the proposed active intermediates for
epoxidation,26 are not formed and, hence, that heterolytic
cleavage of the oxygen-oxygen bond in the iron(III)
hydroperoxide species does not occur when hydrogen
peroxide interacts with (F20TPP)Fe(OCH3). Evidence was
also not seen for homolytic cleavage of iron(III) hydroper-
oxide species because the occurrence of this process would
have produced iron(IV) oxo species, which would have
turned the solution bright red and resulted in a peak near
540 nm in the UV-visible spectrum of the solution.34 Sim-
ilar to the reaction of hydrogen peroxide with the chloride-
coordinated species in aprotic solvents, only a small portion
of the hydrogen peroxide was consumed over the course of
several hours, with the only consequence being porphyrin
degradation. The absence of either hydrogen peroxide
decomposition or cyclooctene epoxidation indicates that
some species other than (F20TPP)Fe(OCH3) must be the
precursor to the intermediate responsible for olefin epoxi-
dation when the chloride-coordinated porphyrin is dissolved
in methanol.
The position of the â-pyrrole 1H NMR peak at 65 ppm is
consistent with the proposed attribution of this peak to [(F20-
TPP)Fe(CH3OH)]+ rather than to [(F20TPP)Fe]+. As noted
above the interaction of [(F20TPP)Fe]+ cations with very
weakly bound ligands shifts the position of the â-pyrrole
peak to -62 ppm.25 The observation of a new peak at 65
ppm is characteristic of a weakly bound ligand similar to a
triflate anion, which shows a â-pyrrole resonance at 57 ppm.
This leads to the suggestion that methanol coordinated to
[(F20TPP)Fe]+ cations affects the shielding of the â-pyrrole
hydrogen atoms on the porphyrin ring in much the same
manner as a weakly bound anionic ligand.
An interesting question is why (F20TPP)Fe(OCH3) does
not dissociate in a manner similar to that of (F20TPP)FeCl
when dissolved in methanol, viz., reaction 3. We attribute
this to the fact that methoxide is a much stronger base than
chloride. As a result, the methoxide anions would deprotonate
any [(F20TPP)Fe(CH3OH)]+ cations present in solution;
therefore, reaction 3 does not occur.
Identification of the Precursor to the Active Intermedi-
ate. When (F20TPP)FeCl is dissolved in acetonitrile, a dark-
brown solution is obtained; the solution turns emerald green
1
as methanol is substituted for the aprotic solvent. H NMR
spectra of these solutions for different concentrations of
methanol are shown in Figure 1. In pure acetonitrile, a single
peak is observed at 82 ppm, and as the mole fraction of
methanol in the solvent increases, a new peak appears at 65
ppm and grows in intensity. At the same time, the peak at
82 ppm decreases in intensity. However, even in pure
methanol (24.7 M), a small peak is still observed near 82
ppm. The observation of the peak near 82 ppm in both pure
(34) Lee, K. A.; Nam, W. Bull. Korean Chem. Soc. 1996, 17, 669-671.
5594 Inorganic Chemistry, Vol. 45, No. 14, 2006

Thermodynamics of (F20TPP)FeCl Dissociation
[Fe-CH3OH+][Cl-]
[Fe-Cl][CH3OH]
K )
(1)
[Fe-CH3OH+]2
K )
(2)
(3)
[Fe-Cl][CH3OH]
[Fe-Cl]0 ) [Fe-CH3OH+] + [Fe-Cl]
[Fe-CH3OH+] ) {-K[CH3OH] +
(K[CH OH])2 + 4K[CH OH][Fe-Cl] }/2 (4)
x
3
3
0
Several experiments were designed to test the validity of
our proposed model. In a previous publication, we had used
eq 4 to determine the value of K by varying the total
porphyrin concentration.26 In that experiment, we discovered
that the observed rate constant for the epoxidation of
cyclooctene did not increase linearly with the total porphyrin
concentration. Using eq 4 to calculate [(F20TPP)Fe(CH3-
OH)]+, we were able to show that the observed rate constant
increases linearly with [(F20TPP)Fe(CH3OH)]+. (Note: the
previous publication refers to this species as the methoxide-
ligated porphyrin species.) A value of K ) 1.2 × 10-5 was
obtained in that study.26
If the species responsible for the NMR peak at 65 ppm
are due to methanol-coordinated porphyrin cations, the
species responsible for the peak at 82 ppm are due to
chloride-coordinated species, and the equilibrium relationship
given by eq 1 is correct, then the equilibrium constant
determined from an analysis of the paramagnetic NMR data
should agree with the value previously obtained by analysis
of the reaction kinetics. The equilibrium constant K was
determined by fitting eq 4 to the NMR data obtained in an
experiment in which the total porphyrin concentration was
varied, and the results are shown in Figure 3. [Fe-Cl] was
determined from the area of the peak at 82 ppm, and [Fe-
CH3OH+] was determined from the area of the peak at 65
ppm. The value of K obtained in this way is 1.0 × 10-5,
which is in excellent agreement with the value previously
obtained from kinetics.26 It is also notable that the concentra-
tions of porphyrin used for each of the two sets of
experiments are noticeably different. The total porphyrin
concentration used for the measurements of epoxidation
kinetics ranged from 37 µM to 0.9 mM, whereas the NMR
experiments reported here were carried out using a total
porphyrin concentration ranging from 1.8 to 15 mM.
Therefore, as shown by two methods, eq 4 fits the observed
data over a wide range of porphyrin concentrations for a
given methanol concentration.
Figure 2. 1H NMR spectra of the methoxide-ligated porphyrin: (a) in
acetonitrile with HCl; (b) in methanol with HCl; (c) in methanol without
HCl. [(F20TPP)Fe(OCH3)] ∼ 11 mM and [HCl] ∼ 25 mM.
To test the arguments given above, we introduced free
protons in the form of hydrochloric acid to a methanol
solution of (F20TPP)Fe(OCH3). A total of 2-3 equiv of HCl-
(aq) were added per 1 equiv of porphyrin. The solution
immediately changed from dark orange to emerald green,
and [(F20TPP)Fe(CH3OH)]+ cations were formed, as evi-
denced by a decrease in the intensity of the â-pyrrole peak
near 80 ppm and the formation of a â-pyrrole peak at 67
ppm (see Figure 2). Furthermore, as discussed below, the
species created by the addition of hydrochloric acid to (F20-
TPP)Fe(OCH3) in methanol exhibited similar catalytic activ-
ity for cyclooctene epoxidation as (F20TPP)FeCl dissolved
in pure methanol, providing further support for the conclusion
that dissociation of the porphyrin salt into [(F20TPP)Fe(CH3-
OH)]+ and Cl- is a necessary condition for catalysis to occur.
As evidenced by no change in the position of the â-pyrrole
resonance, formation of [(F20TPP)Fe(CH3OH)]+ or any other
weakly coordinated cationic porphyrin species was not
evidenced when HCl(aq) was added to (F20TPP)Fe(OCH3)
dissolved in acetonitrile. We suggest that this is further
evidence that acetonitrile is not capable of stabilizing the
ionic products of reaction 2.
Thermodynamics of (F20TPP)FeCl Dissociation. An
equilibrium relationship (eq 1) can be written between the
species shown in reaction 2. Conservation of charge requires
that the concentrations of both species on the right side of
the equilibrated reaction be the same. Therefore, eq 1 can
be rewritten as eq 2. If we also assume that the only
porphyrin species are the chloride-coordinated species and
the methanol-coordinated porphyrin cations, then a balance
of porphyrin species results in eq 3, where [Fe-Cl]0 is the
concentration of all porphyrin species in solution. Combining
eqs 2 and 3 leads to a quadratic relationship between the
concentration of the methanol-coordinated porphyrin cations
and the methanol concentration (eq 4). This relationship also
predicts that the concentration of methanol-coordinated
porphyrin cations will be a nonlinear function of the total
porphyrin concentration, [Fe-Cl]0.
To validate the model further, eq 4 was used to fit the
equilibrium constant to NMR data taken at different methanol
concentrations. Data fits similar to the one shown in Figure
3 were obtained in each case. Figure 4 shows that the
equilibrium constant increases exponentially with the metha-
nol concentration, resulting in a sigmoidal relationship
between the concentration of the methanol-coordinated
species and the methanol concentration (see Figure 5).
Inorganic Chemistry, Vol. 45, No. 14, 2006 5595

Stephenson and Bell
Figure 6. Plot of ln(K) vs [CH3OH]. All data are for room temperature at
∼23 °C.
Figure 3. Plot of the concentration of [(F20TPP)Fe(CH3OH)]+ vs the total
porphyrin concentration. The solid curve represents a fit of eq 4 to the data
for a value of K ) 1.0 × 10-5 when the methanol concentration is 6.175
M.
Figure 7. Plot of ln(K) vs 1/RT. All data are for a methanol concentration
of 14.8 M.
methanol concentration, is found to be -0.20 kcal‚L/mol2.
Figure 4. Plot of the equilibrium constant determined experimentally vs
the methanol concentration.
K ) e-∆G°/RT
(5)
(6)
∆H° ) ∆H0° + R[CH3OH]
K ) K0e-∆H°/RT
(7)
(8)
K ) K1e-R[CH OH]/RT
3
Additional NMR experiments were conducted at a fixed
methanol concentration while varying the temperature be-
tween 0 and 40 °C. A methanol concentration of 14.8 M
was chosen for these experiments. This choice of conditions
resulted in the peaks at 65 and 82 ppm having roughly equal
areas, allowing the number of scans required for a reasonable
signal-to-noise ratio to be a minimum. As expected for
paramagnetic species, the peak positions of both species
showed Curie-type behavior, shifting linearly with the inverse
of temperature. The observed equilibrium constant was
determined for each temperature and plotted as a function
of 1/RT, as seen in Figure 7. Fitting of the data shown in
Figures 6 and 7 results in ∆H0° ) 6.0 ( 0.3 kcal/mol, R )
-0.20 ( 0.01 kcal‚L/mol2, and ∆S° ) -5.8 ( 0.6 cal/mol‚
K, where the equilibrium constant is expressed according to
eq 9.
Figure 5. Plot of the concentration of [(F20TPP)Fe(CH3OH)]+ vs the
concentration of methanol in the solvent. Data are shown for a porphyrin
concentration of ∼15 mM.
Because the equilibrium constant can be expressed in terms
of the Gibbs free energy (eq 5), we hypothesized that the
enthalpy change is linearly proportional to the methanol
concentration, as shown in eq 6. Upon substitution and
rearrangement, eq 5 can be rewritten as eq 7, where K0 is
equal to e∆S°/R. This equation can be simplified to eq 8, where
K1 is equal to K0e-∆H °/RT. As shown in Figure 6, the value
0
0
3
of R, as determined from the slope of a plot of ln(K) vs
K ) e-(∆H °+R[CH OH])/RTe∆S°/R
(9)
5596 Inorganic Chemistry, Vol. 45, No. 14, 2006

Thermodynamics of (F20TPP)FeCl Dissociation
To show that the experimentally determined thermody-
namic parameters are valid over a wide range of data, two
additional analyses were conducted. The variable-temperature
experiment was repeated using a sample with a methanol
concentration of 9.9 M; analysis of the plot of ln(K) vs 1/RT,
not shown, gives R ) -0.20 kcal‚L/mol2 and ∆S ) -5.8
cal/mol‚K, values that are identical with those obtained for
a methanol concentration of 14.8 M. In addition, the value
of ∆S° was determined by using the equilibrium constants
determined at room temperature for methanol concentrations
between 6.2 and 24.7 M. ∆G° was calculated as a function
of the methanol concentration using eq 5. ∆S° was then
determined by subtracting ∆H°, as calculated from eq 6, from
∆G° and dividing by the temperature. ∆S° determined in
this way is equal to -5.9 ( 0.8 cal/mol‚K and is independent
of the methanol concentration.
On the basis of the above analysis, the value of ∆H0° is
positive, indicating that the dissociation of (F20TPP)FeCl is
endothermic. This indicates that the enthalpies of solvation
of the cations and anions formed upon dissociation of this
porphyrin are not sufficiently negative to offset the positive
enthalpy required to form the cation and anion pair. The value
of R is negative as well, indicating that the value of ∆H°
decreases with an increase in the methanol fraction in the
solvent mixture. As discussed above, the decrease in ∆H°
with an increase in the methanol content of the solvent is
due to increased solvation of the chloride anions, as well as
to stabilization of the [(F20TPP)Fe]+ cations through coor-
dination with methanol. Nevertheless, ∆H° remains positive
even in pure methanol. It is also found that ∆S° is negative,
which indicates an increase in order when (F20TPP)FeCl
dissociates. Because the number of species before and after
dissociation (see reaction 2) does not change, the negative
sign is likely due to an increase in the solvent ordering around
the cations and anions formed upon dissociation of (F20TPP)-
FeCl, relative to the undissociated porphyrin.
As a further test of the validity of the thermodynamic
model described above for reaction 2, an experiment was
carried out in which chloride anions were added to the
mixture using tetrabutylammonium chloride as the source.
The experiment was conducted in a solvent mixture contain-
ing equal volume amounts of methanol and acetonitrile;
chloride anions, 0.72 equiv relative to the concentration of
the porphyrin species, were added to the solution. Consistent
with the model for reaction 2, the addition of chloride anions
shifted the equilibrium to the left. This particular salt was
used because the tetrabutylammonium cation was not
expected to affect reaction 2. Prior to chloride addition,
approximately 30% of the porphyrin species was coordinated
to methanol, while only 13% was methanol-coordinated after
the addition of the chloride source. The model for the
equilibrium constant presented above predicts that 12.5% of
the porphyrin species should be coordinated to methanol after
the addition of 0.72 equiv of chloride.
Figure 8. (a) UV-visible spectra obtained by methanol addition to (F20-
TPP)FeCl dissolved in acetonitrile. Arrows indicate the direction of the
absorbance change with an increase in the methanol concentration. (b)
Comparison of the concentration of [(F20TPP)Fe(CH3OH)]+ as determined
from the spectra in part a to the thermodynamic model. The initial total
porphyrin concentration was 25 µM.
required to obtain data, NMR spectroscopy experiments were
conducted using much higher porphyrin concentrations
relative to the concentration used for kinetic studies (10 mM
vs 75 µM). Additional experiments using UV-visible
spectroscopy were conducted to show that the thermody-
namic model presented above is applicable at lower por-
phyrin concentrations. Because of absorbance limitations, a
porphyrin concentration of 25 µM was used for these
experiments. (F20TPP)FeCl was dissolved in acetonitrile, and
microliter quantities of methanol were added incrementally
to the solution. The resulting spectra are shown in Figure
8a; note that the systematic decrease in the absorbance is
due to dilution of the porphyrin concentration as a result of
methanol addition. Addition of methanol results in a decrease
in the absorbance of the Soret peak at 407 nm and the
formation and subsequent increase of a new Soret peak at
390 nm; new features are also formed at 460 and 569 nm
(see Figure 8a). Clear isosbestic points are seen at 376, 396,
484, 551, and 589 nm. We attribute these changes in the
UV-visible spectra to an exchange of the chloride ligand
for a methanol solvent molecule. The change in absorbance
at 569 nm was used to determine the concentration of [(F20-
TPP)Fe(CH3OH)]+ as a function of the methanol concentra-
tion; this peak was chosen because it was the best resolved
of the newly formed peaks. As shown in Figure 8b, the
change in the concentration of [(F20TPP)Fe(CH3OH)]+ with
Application of the Model to Lower Porphyrin Con-
centrations. In efforts to minimize the instrument time
Inorganic Chemistry, Vol. 45, No. 14, 2006 5597

Stephenson and Bell
Figure 9. Proposed mechanism for (F20TPP)FeCl-catalyzed epoxidation of cyclooctene using hydrogen peroxide as the oxidant.
the methanol concentration determined from the UV-visible
spectra is in excellent agreement with that predicted by the
thermodynamic model. Therefore, the model developed via
NMR spectroscopy at high porphyrin concentrations is valid
for porphyrin concentrations over 3 orders of magnitude (10
µM to 10 mM).
cations, and Y∞ is the yield of epoxide determined at
the point of complete consumption of H2O2, as defined by
eq 11. The yield is used to relate the rate of epoxide
formation to the overall rate of hydrogen peroxide consump-
tion.
k3K2[Fe-CH3OH+][CH3OH]
Implications for the Kinetics of Cyclooctene Epoxida-
tion. As noted in the Introduction, we have previously
investigated the kinetics of cyclooctene epoxidation by H2O2
catalyzed by (F20TPP)FeCl dissolved in mixtures of aceto-
nitrile and methanol.26,35 The kinetics of this reaction are well
described over a wide range of conditions by rate expressions
derived from the mechanism shown in Figure 9.26,35 In
reaction 1, (F20TPP)FeCl dissociates to form [(F20TPP)Fe-
(CH3OH)]+ and Cl-. The first of these species then coor-
dinates H2O2 reversibly in reaction 2. The heterolytic and
homolytic cleavages of the O-O bond of H2O2 occur in
reactions 3 and 4, the first of which involves the direct
participation of methanol. The iron(IV) π-radical cation
formed via reaction 3 can then react with cyclooctene
to produce cyclooctene epoxide (reaction 7) or with H2O2
to produce free radicals and the products of peroxide
decomposition (reactions 6 and 5). When the concentration
of cyclooctene to H2O2 is in excess, the rate of reaction 7
greatly exceeds that of reaction 6. Under these condi-
tions, the rate coefficient (kobs) for the consumption of H2O2,
the limiting reagent, is given by eq 10.26,35 [Fe-CH3OH+]
is the concentration of methanol-coordinated porphyrin
kobs
)
(10)
Y∞
[C8-O]∞
k3[CH3OH]
Y∞ )
× 100% =
× 100% (11)
[H2O2]0
k3[CH3OH] + 2k4
In our initial studies of cylclooctene epoxidation by H2O2
catalyzed by (F20TPP)FeCl dissolved in methanol/acetonitrile
mixtures, the value of kobs was found to increase nonlinearly
with the concentration of (F20TPP)FeCl for a fixed solvent
composition.26 The question to be addressed is whether kobs
is linearly proportional to the concentration of [(F20TPP)-
Fe(CH3OH)]+, as indicated by eq 10. Figure 10a shows that
kobs indeed increases linearly with the concentration of [(F20-
TPP)Fe(CH3OH)]+, as determined from eq 4. When the
concentration of (F20TPP)FeCl is held constant but the
concentration of methanol is varied, kobs is predicted to
increase linearly with the ratio of [Fe-CH3OH+][CH3OH]
to Y∞. Figure 10b shows such a plot and further demonstrates
the agreement between theory and experiment. Therefore,
the equilibrium relationship based on reaction 2 is fully
consistent, not only with the NMR and UV-visible observa-
tions but also with the kinetics for the epoxidation of
cyclooctene.26,35
(35) Stephenson, N. A.; Bell, A. T. Inorg. Chem. 2006, in press.
5598 Inorganic Chemistry, Vol. 45, No. 14, 2006

Thermodynamics of (F20TPP)FeCl Dissociation
further support the hypothesis that [(F20TPP)Fe(CH3OH)]+
is a critical intermediate in the epoxidation of olefins by
hydrogen peroxide in methanol-containing solutions of (F20-
TPP)FeCl.
Conclusions
We have shown that (F20TPP)FeCl is dissolved molecu-
larly in acetonitrile but dissociatively in mixtures of aceto-
nitrile and methanol. In the latter case, the iron(III) porphyrin
cation is stabilized by a molecule of methanol as [(F20TPP)-
Fe(CH3OH)]+. A similar cation is not formed, though, when
(F20TPP)Fe(OCH3) is dissolved in pure methanol. The lower
degree of dissociation for (F20TPP)Fe(OCH3) relative to (F20-
TPP)FeCl in methanol is due to the much higher basicity of
methoxide anions relative to chloride anions. However, [(F20-
TPP)Fe(CH3OH)]+ cations can be formed from (F20TPP)-
Fe(OCH3) dissolved in methanol when HCl(aq) is added to
the solution. The thermodynamic parameters that determine
the equilibrium constant for the dissociation of (F20TPP)-
FeCl in methanol-containing solutions, ∆H° and ∆S°, have
been evaluated, and the proposed model for (F20TPP)FeCl
dissociation has been shown to be valid over a range of
temperatures, methanol concentrations, and porphyrin con-
centrations. We have also shown that [(F20TPP)Fe(CH3OH)]+
cations formed via dissociation of (F20TPP)FeCl in methanol-
containing solutions are necessary intermediates in the
epoxidation of cyclooctene by H2O2. Furthermore, we have
shown that the observed rate constant for the consumption
of H2O2, kobs, is linearly proportional to the concentration of
[(F20TPP)Fe(CH3OH)] +.
Figure 10. (a) Plot of kobs vs the concentration of [(F20TPP)Fe(CH3OH)]+
for experiments with a constant concentration of CH3OH (∼6.3 M) and
varying concentrations of (F20TPP)FeCl. (b) Plot of kobs vs the concentration
of [(F20TPP)Fe(CH3OH)]+ times the concentration of CH3OH divided by
the yield of cyclooctene epoxide. In this case, the concentration of [(F20-
TPP)Fe(CH3OH)]+ was varied by changing the concentration of CH3OH
while holding the concentration of (F20TPP)FeCl constant at ∼75 µM.
Acknowledgment. The authors thank Rudi Nunlist for
his help in optimizing the parameters used for the paramag-
netic NMR experiments and John Prausnitz and John
Newman for their helpful discussions concerning the ther-
modynamics of porphyrin dissociation. The authors also
thank the reviewers of this work for their constructive
comments and suggestions, which have led to an improve-
ment in the presentation of our work. 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.
Because the reaction of (F20TPP)Fe(OCH3) with HCl(aq)
in methanol produces [(F20TPP)Fe(CH3OH)]+, an experiment
was carried out to determine whether [(F20TPP)Fe(CH3OH)]+
produced in this way would catalyze the epoxidation of
cyclooctene. For this experiment, 40 equiv of HCl(aq) was
added per 1 equiv of (F20TPP)Fe(OCH3) dissolved in
methanol. A value for kobs of 0.4 min-1 was obtained, which
is similar to that obtained when (F20TPP)FeCl is dissolved
in methanol, 0.6 min-1. The yield of cyclooctene epoxide
was also virtually the same in both cases, 88%. These results
IC0521067
Inorganic Chemistry, Vol. 45, No. 14, 2006 5599