Mechanistic Studies of the Formation and Decay of Diiron(III) Peroxo Complexes in the Reaction of Diiron(II) Precursors with Dioxygen

Article abstract of DOI:10.1021/ic951242g

Mechanistic studies of the reactions of three analogous alkoxo-bridged diiron(II) complexes with O₂ have been carried out. The compounds, which differ primarily in the steric accessibility of dioxygen to the diiron(II) center, form metastable μ-peroxo intermediates when studied at low temperature. At ambient temperatures, these intermediates decay to form (μ-oxo)polyiron(III) products. The effect of ligand steric constraints on the O₂ reactivity was investigated. When access to the diiron center was unimpeded, the reaction was first-order with respect to both [Fe₂^II] and [O₂] and the activation parameters for O₂ addition were similar to those for O₂ reacting with the dioxygen transport protein hemerythrin. When the binding site was occluded, however, reduced order with respect to [O₂] was observed and a two-step mechanism was required to explain the kinetic results. Decay of all three peroxide intermediates involves a bimolecular event, implying the formation of tetranuclear species in the transition state.

Full text of DOI:10.1021/ic951242g

2590
Inorg. Chem. 1996, 35, 2590-2601
Mechanistic Studies of the Formation and Decay of Diiron(III) Peroxo Complexes in the
Reaction of Diiron(II) Precursors with Dioxygen
†
‡
‡
‡
Andrew L. Feig, Michael Becker, Siegfried Schindler, Rudi van Eldik, and
Stephen J. Lippard*,†
Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, and
Institute for Inorganic Chemistry, University of Erlangen-N u¨ rnberg, Egerlandstrasse 1,
91058 Erlangen, Germany
X
ReceiVed September 26, 1995
Mechanistic studies of the reactions of three analogous alkoxo-bridged diiron(II) complexes with O2 have been
carried out. The compounds, which differ primarily in the steric accessibility of dioxygen to the diiron(II) center,
form metastable µ-peroxo intermediates when studied at low temperature. At ambient temperatures, these
intermediates decay to form (µ-oxo)polyiron(III) products. The effect of ligand steric constraints on the O2 reactivity
was investigated. When access to the diiron center was unimpeded, the reaction was first-order with respect to
II
both [Fe 2] and [O2] and the activation parameters for O2 addition were similar to those for O2 reacting with the
dioxygen transport protein hemerythrin. When the binding site was occluded, however, reduced order with respect
to [O2] was observed and a two-step mechanism was required to explain the kinetic results. Decay of all three
peroxide intermediates involves a bimolecular event, implying the formation of tetranuclear species in the transition
state.
Introduction
Exposure of the apo form of this altered enzyme to iron(II) and
dioxygen converts the tyrosine residue to dihydroxyphenylala-
nine (DOPA).
Reduced dinuclear iron centers occur in several metallopro-
teins, including hemerythrin (Hr),1 the R2 protein of ribo-
nucleotide reductase (RNR) from Escherichia coli,4 and the
hydroxylase (H) component of soluble methane monooxygenase
-3
Reactions of the reduced forms of RNR and sMMO with
-7
1
4-21
dioxygen occur by multistep processes.
An early step in
both cases is two-electron oxidation of the diiron(II) center. This
process is similar to the dioxygen-binding reaction of deoxyHr,
but the resultant peroxo species in MMO and RNR differ from
8
-10
(
sMMO).
In each of these systems, the diiron center is
directly involved in a reaction with dioxygen that is integral to
the function of the protein.1
1,12
Of major interest is determining
1
22,23
the η -hydroperoxide coordination mode found in oxyHr.
the features of these proteins which differentiate their modes
of reactivity. Hr binds dioxygen, but subsequent reactions
leading to cleavage of the O-O bond are disfavored. In RNR
and MMO, four-electron reduction of dioxygen occurs with
chemical utilization of the oxidizing equivalents. In RNR, the
dioxygen reaction oxidizes Tyr-122 to a functionally important
tyrosyl radical whereas, in sMMO, a substrate alkane is oxidized
to an alcohol. A similarity between the RNR and MMO
In both MMO and RNR, the peroxo species is unstable and
decomposes to less well defined intermediates. In the case of
MMO, a diiron(III) peroxo intermediate (Hperoxo) decays to an
intermediate designated Q which has been postulated to be either
a diferryl species or a diiron(III) center bound to two oxyl
1
0,15,16,20
radicals.
Further spectroscopic and mechanistic work
is required to define the chemical nature of this transient. In
RNR, a diferric center bridged by a single radical ligand has
been proposed to be the intermediate in the oxidation
1
3
activities can be observed in the F208Y mutant R2 protein.
1
4,17-19
†
mechanism.
Massachusetts Institute of Technology.
University of Erlangen-N u¨ rnberg.
Abstract published in AdVance ACS Abstracts, April 1, 1996.
‡
X
(13) A° berg, A.; Orm o¨ , M.; Nordlund, P.; Sj o¨ berg, B.-M. Biochemistry 1993,
32, 9845-9850.
(
1) Kurtz, D. M., Jr.; Shriver, D. F.; Klotz, I. M. Coord. Chem. ReV. 1977,
24, 145-178.
(14) Bollinger, J. M., Jr.; Edmondson, D. E.; Huynh, B. H.; Filley, J.;
Norton, J. R.; Stubbe, J. Science 1991, 253, 292-298.
(15) Lee, S.-K.; Nesheim, J. C.; Lipscomb, J. D. J. Biol. Chem. 1993, 268,
21569-21577.
(
2) Sanders-Loehr, J. In Iron Carriers and Iron Proteins; Loehr, T. M.,
Ed.; VCH: New York, 1989; Vol. 5; pp 373-466.
(
(
3) Stenkamp, R. E. Chem. ReV. 1994, 94, 714-726.
4) Nordlund, P.; Sj o¨ berg, B.-M.; Eklund, H. Nature 1990, 345, 593-
(16) Lee, S.-K.; Fox, B. G.; Froland, W. A.; Lipscomb, J. D.; M u¨ nck, E.
J. Am. Chem. Soc. 1993, 115, 6450-6451.
598.
(
5) Stubbe, J. In AdVances in Enzymology and Related Areas of Molecular
Biology; Meister, A., Ed.; John Wiley and Sons: New York, 1990;
Vol. 62; pp 349-420.
(17) Bollinger, J. M., Jr.; Tong, W. H.; Ravi, N.; Huynh, B. H.; Edmondson,
D. E.; Stubbe, J. J. Am. Chem. Soc. 1994, 116, 8015-8023.
(18) Bollinger, J. M., Jr.; Tong, W. H.; Ravi, N.; Huynh, B. H.; Edmondson,
D. E.; Stubbe, J. J. Am. Chem. Soc. 1994, 116, 8024-8032.
(19) Ravi, N.; Bollinger, J. M., Jr.; Huynh, B. H.; Edmondson, D. E.;
Stubbe, J. J. Am. Chem. Soc. 1994, 116, 8007-8014.
(20) Liu, K. E.; Valentine, A. M.; Wang, D.; Huynh, B. H.; Edmondson,
D. E.; Salifoglou, A.; Lippard, S. J. J. Am. Chem. Soc. 1995, 117,
10174-10185.
(
6) Fontecave, M.; Nordlund, P.; Eklund, H.; Reichard, P. In AdVances
in Enzymology and Related Areas of Molecular Biology; Meister, A.,
Ed.; John Wiley and Sons: New York, 1992; Vol. 65; pp 147-183.
7) Nordlund, P.; Eklund, H. J. Mol. Biol. 1993, 232, 123-164.
8) Rosenzweig, A. C.; Frederick, C. A.; Lippard, S. J.; Nordlund, P.
Nature 1993, 366, 537-543.
(
(
(
9) Lipscomb, J. D. Annu. ReV. Microbiol. 1994, 48, 371-399.
(21) Liu, K. E.; Valentine, A. M.; Qiu, D.; Edmondson, D. E.; Appelman,
E. H.; Spiro, T. G.; Lippard, S. J. J. Am. Chem. Soc. 1995, 117, 4997-
4998.
(
10) Liu, K. E.; Lippard, S. J. In AdVances in Inorganic Chemistry; Sykes,
A. G., Ed.; Academic Press, Inc.: San Diego, CA, 1995; Vol. 42, pp
2
63-289.
(22) Stenkamp, R. E.; Sieker, L. C.; Jensen, L. H.; McCallum, J. D.;
Sanders-Loehr, J. Proc. Natl. Acad. Sci., U.S.A. 1985, 82, 713-716.
(23) Shiemke, A. K.; Loehr, T. M.; Sanders-Loehr, J. J. Am. Chem. Soc.
1986, 108, 2437-2443.
(
11) Que, L., Jr. In Bioinorganic Catalysis; Reedijk, J., Ed.; Marcel
Dekker: New York, 1993; pp 347-393.
12) Feig, A. L.; Lippard, S. J. Chem. ReV. 1994, 94, 759-805.
(
0
020-1669/96/1335-2590$12.00/0 © 1996 American Chemical Society

Reaction of Diiron(II) Precursors with Dioxygen
Inorganic Chemistry, Vol. 35, No. 9, 1996 2591
Scheme 1
Several model compounds have been synthesized for the
of the reaction (Scheme 1). When first reported by other
24-35
reduced cores of these enzymes.
Although one comes very
laboratories, it was noted that stable peroxide adducts formed
close,34 none is a perfect structural mimic for a protein core,
and none exhibits catalytic monooxygenase activity. Several
of these complexes react cleanly with dioxygen, however, to
afford well-characterized products, and we have therefore
undertaken a systematic investigation of the kinetics of the
dioxygen reaction. Elsewhere we have described our studies
with [Fe2(O2CH)4(BIPhMe)4] and [Fe2(OH)(O2CCH3)2(Me3-
following exposure to dioxygen at low temperature.
28,30
Spec-
troscopic studies of these complexes revealed the presence of a
symmetrically bound peroxide, presumably coordinated in a
µ-1,2 fashion. In the case of compound 1a, dioxygen binding
2
8
was reversible. Once dioxygen was bound to either 2a or
3
0,36,37
3a, however, it could not be released.
Thus, although
the chemical differences among 1a-3a are subtle, consisting
primarily of changes in the steric properties around the diiron
center, the effects of these differences on the reactions with
dioxygen are profound. Upon warming, all three of the
metastable diferric peroxo complexes decompose, but at tem-
+
51,62
TACN)2] .
In the present article, we report results for three
closely related compounds, each containing a (µ-alkoxo)(µ-
carboxylato)diiron(II) core coordinated by N,N,N′,N′-tetrasub-
stituted-1,3-diamino-2-hydroxypropane (HPTR) ligands. The
dioxygen reactivity of these compounds is of particular interest
mechanistically because of the ability to isolate different phases
3
0
peratures which vary considerably.
We are interested in
understanding these differences, especially the structural features
which influence the reactivity of the diiron(II) compounds and
their activated dioxygen complexes. From this information, it
should be possible to predict the chemical properties of the
metastable diiron(III) peroxo species. A portion of this work
(
24) Hartman, J. R.; Rardin, R. L.; Chaudhuri, P.; Pohl, K.; Wieghardt,
K.; Nuber, B.; Weiss, J.; Papaefthymiou, G. C.; Frankel, R. B.;
Lippard, S. J. J. Am. Chem. Soc. 1987, 109, 7387-7396.
(
(
(
(
(
(
(
(
(
25) Borovik, A. S.; Que, L., Jr. J. Am. Chem. Soc. 1988, 110, 2345-
3
8
2347.
has been communicated previously.
26) Kitajima, N.; Fukui, H.; Moro-oka, Y.; Mizutani, Y.; Kitagawa, T. J.
Am. Chem. Soc. 1990, 112, 6402-6403.
Experimental Section
27) Tolman, W. B.; Liu, S.; Bentsen, J. G.; Lippard, S. J. J. Am. Chem.
Soc. 1991, 113, 152-164.
General Considerations. All solvents were dried and distilled prior
39
to use. For most studies, propionitrile (Fluka) was prepared by
28) Hayashi, Y.; Suzuki, M.; Uehara, A.; Mizutani, Y.; Kitagawa, T. Chem.
Lett. 1992, 91-94.
2
distillation from CaH . Significant levels of residual water were present
after this treatment, as measured by Karl Fischer analysis (Galbraith
Laboratories). Rigorously dry solvent, used to determine the effects
29) Stassinopoulos, A.; Schulte, G.; Papaefthymiou, G. C.; Caradonna, J.
P. J. Am. Chem. Soc. 1991, 113, 8686-8697.
30) Dong, Y.; Menage, S.; Brennan, B. A.; Elgren, T. E.; Jang, H. G.;
Pearce, L. L.; Que, L., Jr. J. Am. Chem. Soc. 1993, 115, 1851-1859.
31) Hagen, K. S.; Lachicotte, R.; Kitaygorodskiy, A.; Elbouadili, A. Angew.
Chem. 1993, 105, 1404-1406.
of small amounts of added water, was obtained by predrying over CaH
decanting the solvent into a distillation flask containing P (5 g/L),
and refluxing overnight under N . The propionitrile was collected by
distillation and then refluxed over CaH (5 g/L). After 12 h, the solvent
2
,
2 5
O
2
2
32) Zang, Y.; Elgren, T. E.; Dong, Y.; Que, L., Jr. J. Am. Chem. Soc.
1993, 115, 811-813.
was collected by distillation and transferred to a glovebox. Karl Fischer
analysis of propionitrile treated in this manner contained <34 ppm of
residual water. Dioxygen was purified by passing the gas stream
through a 1 m column of Drierite. Anaerobic manipulations were
carried out in a nitrogen-filled glovebox (Vacuum Atmospheres) or by
33) Kitajima, N.; Tamura, N.; Amagai, H.; Fukui, H.; Moro-Oka, Y.;
Mizutani, Y.; Kitagawa, T.; Methur, R.; Heerwegh, K.; Reed, C. A.;
Randall, C. R.; Que, L., Jr.; Tatsumi, K. J. Am. Chem. Soc. 1994,
116, 9071-9085.
(
(
(
(
34) Herold, S.; Pence, L. E.; Lippard, S. J. J. Am. Chem. Soc. 1995, 117,
1
8
using standard Schlenk techniques. Labeled dioxygen (≈98%
2
O )
6134-6135.
35) Coucouvanis, D.; Reynolds, R. A., III; Dunham, W. R. J. Am. Chem.
Soc. 1995, 117, 7570-7571.
was obtained from Isotech (Miamisburg, OH) and used without further
purification.
36) Menage, S.; Brennan, B. A.; Juarez-Garcia, C.; M u¨ nck, E.; Que, L.,
Jr. J. Am. Chem. Soc. 1990, 112, 6423-6425.
(38) Feig, A.; Lippard, S. J. J. Am. Chem. Soc. 1994, 116, 8410-8411.
(39) Perrin, D. D.; Armarego, W. L. F. Purification of Laboratory
Chemicals, 3rd ed.; Butterworth-Heinemann: Oxford, U.K., 1988.
37) Brennan, B. A.; Chen, Q.; Juarez-Garcia, C.; True, A. E.; O’Connor,
C. J.; Que, L., Jr. Inorg. Chem. 1991, 30, 1937-1943.

2
592 Inorganic Chemistry, Vol. 35, No. 9, 1996
Feig et al.
Ligand Syntheses. The compound N,N,N′,N′-tetrakis[(6-methylpy-
rid-2-yl)methyl]-2-hydroxy-1,3-diaminopropane (HPTMP) (1),28 was
synthesized by a new method described below. The other ligands,
N,N,N′,N′-tetrakis(2-pyridylmethyl)-2-hydroxy-1,3-diaminopropane
(HPTP) (2) and N,N,N′,N′-tetrakis[(N-ethyl-2-benzimidazolyl)methyl]-
2
-hydroxy-1,3-diaminopropane (Et-HPTB) (3), were prepared according
3
0
to literature procedures.
6
-Methyl-2-(hydroxymethyl)pyridine. To a solution of 6-methyl-
-pyridinecarboxaldehyde (25.0 g, 0.207 mol) in ethanol (100 mL) at
°C was added 3.0 g of NaBH . The solution was stirred for 1 h,
O was added. The solution was neutralized
by addition of 1.0 M aqueous HCl, and the product was extracted into
Et O. The solution was dried over MgSO and the solvent removed
under reduced pressure, yielding the product alcohol as a colorless oil
2
0
4
after which 100 mL of H
2
2
4
1
(
1
2
23.2 g, 0.189 mol, 91.5%). H NMR (300 MHz, CDCl
3
): δ 7.56 (t,
H, J ) 7.5 Hz), 7.05 (apparent t, 2 H, J ) 7.8 Hz), 4.71 (s, 2 H),
.54 (s, 3 H).
-Methyl-2-(bromomethyl)pyridine. The above alcohol was dis-
solved in 70 mL of 48% HBr. Concentrated H SO (50 mL) was added
slowly, and the solution was refluxed overnight. The reaction mixture
was poured into 100 mL of H O and then neutralized with saturated
aq Na CO The product was extracted into Et O and dried over
MgSO . The solvent was removed under reduced pressure, yielding a
pink solid (29.0 g, 0.156 mol, 82.5%). H NMR (250 MHz, CDCl
6
2
4
2
2
3
.
2
Figure 1. Schematic drawing of the Toepler pump used in the isotope-
labeling studies. Open circles represent manual control valves. Shaded
circles represent solenoids controlled by the electronic relay, and A,
B, and C are contacts used by the relay circuit. Circles with arrows
are one-way valves in the directions indicated.
4
1
3
):
δ 7.57 (t, 1 H, J ) 7.8 Hz), 7.23 (d, 1 H, J ) 7.8 Hz), 7.05 (d, 1 H,
J ) 7.8 Hz), 4.51 (s, 2 H), 2.55 (s, 3H). EI MS: m/z 185, 187 (isotope
+
ratio 1:1, M ).
N,N,N′,N′-Tetrakis[(6-methylpyrid-2-yl)methyl]-2-hydroxy-1,3-
diaminopropane. To a solution of 1,3-diamino-2-propanol (0.902 g,
reaction flasks and cooled to -77 °C. The samples were oxygenated
with either 16
O
or
18
O
, and both flasks were purged of excess O
by
2
2
2
1
2
0 mmol) in H O (25 mL) were added (6-methylpyrid-2-yl)methyl
18
repeated vacuum/argon cycling. The
2
O sample was then transferred
bromide (8.60 g, 46 mmol) and NaOH (4.5 g, 113 mmol). The solution
was refluxed overnight, and upon cooling, a biphasic mixture was
by cannula to the flask containing the unlabeled material. The head
space above the reaction mixture was evacuated by using a modified
Toepler pump (Figure 1) while the sample was kept cold. This process
involves preevacuation of the pump by opening valves 1, 2, 3, and 5.
Valves 1 and 3 were then closed, and valve 4 was opened. With the
valves in this setting, gas from the reaction flask was pumped into the
collection tube, the pumping speed being controlled by partially opening
obtained. The organic component was extracted three times into CH
2
-
Cl , the extract was dried over MgSO , and the solvent was removed
2
4
under reduced pressure. The product was purified by silica column
chromatography (EtOAc/MeOH), yielding a viscous brown oil (3.73
1
g, 9.03 mmol). H NMR (250 MHz, CDCl
3
): δ 7.45 (apparent t, 4 H,
J ) 7.7 Hz), 7.17 (d, 4 H, J ) 7.8 Hz), 6.93 (d, 4 H, J ) 7.5 Hz), 3.90
m, 1 H), 3.80 (dd, 4 H, J ) 6.0 and 37.5 Hz), 2.45-2.62 (m, 8 H),
.46 (s, 12 H). This spectrum matches that previously reported for
valve 6. This procedure was repeated until no O
2
was detected in the
(
collection tube. The reaction flask was then isolated from the pump
by closing valve 4, and the solution was stirred at room temperature
under vacuum for 2 h, at which time the decay of the peroxo species
was over, on the basis of the complete discharge of the blue color.
The pump was pre-evacuated as just described, and valve 4 was opened.
Gas collected in the receiving tube connected to the outlet of the Toepler
pump was then analyzed by mass spectrometry (Hewlett Packard 5971
mass spectrometer).
Stopped-Flow Kinetics Experiments. All ambient-pressure kinetics
experiments were carried out by using a Canterbury SF-41 stopped-
flow instrument (Hi-Tech) and a specially assembled fiber-optics
spectrometer (components from Oriel Corp.). It consisted of a visible
2
28
material prepared by an alternative route. The overall three step yield
was 68.2%.
Preparation of Iron(II) Compounds. The three complexes used
in this study, [Fe
BPh (2a), and [Fe
by methods previously reported
2
(HPTMP)(OBz)](BPh
4
)
2
(1a), [Fe
(3a), were synthesized
and purified by multiple (usually
2
(HPTP)(OBz)]-
(
)
4 2
2
(Et-HPTB)(OBz)](BF )
4 2
2
8,30
three) recrystallizations from acetonitrile by vapor diffusion of Et
under a nitrogen atmosphere.
2
O
Optical Absorption Spectra. Electronic spectral properties of
2
8,30
compounds 1b-3b were previously reported.
kinetic work, we recollected these spectra using a Hewlett Packard
452A diode array instrument fitted with a specially modified Dewar
Prior to beginning
light source (quartz-halogen lamp, 100 W) irradiating a monochro-
8
1
mator ( /
8
m, holographic grating 1200 lines/mm, blazed at 200 nm)
flask with quartz windows. Absorptions due to the metastable dioxygen
adducts were observed at -77 °C. Values for the extinction coefficients
were determined on the basis of a Beer’s law analysis of parameters
obtained by curve fitting of the results of the kinetic experiments carried
out in dry solvent. The values matched those measured in static
oxygenation experiments. Extinction coefficients were assumed not
to vary with temperature.
connected to a fused-silica fiber-optics bundle. The monochromatic
light was focused on the stopped-flow cell. A second fiber-optics
bundle collected the transmitted light and irradiated a photomultiplier
tube (Oriel 77341). Data were collected through software developed
locally (Labview2 environment, National Instruments, Apple Macintosh
IIfx computer) at a sampling rate such that approximately 1000 points
were acquired over the course of the reaction with a linear time base.
The effect of pressure on the formation of the peroxide intermediate
Manometry. Manometric experiments were carried out in propi-
2
7
40
onitrile as described previously, but with minor modifications.
Because of the temperature sensitivity of the peroxide intermediates,
dioxygen uptake for the first phase of the reaction was measured with
the reaction flask cooled to -77 °C by immersion in a dry ice/acetone
bath. After equilibrium was achieved, decomposition of the peroxide
was measured by warming the reaction vessel to 22 °C. Dioxygen
consumption was measured manometrically at least twice and is
reported as the mean of the values.
3b was measured as described elsewhere. The preferred solvent for
these studies was propionitrile because of its low freezing (-95 °C)
and high boiling (97 °C) points and was used unless noted otherwise.
The formation of intermediates 1b-3b was studied under pseudo-first-
order conditions with excess of O2. Concentrations of the ferrous
complexes varied from 0.02 to 0.50 mM, and the dioxygen concentra-
tion at saturation (20 °C) was 8.8 mM.41 Saturation concentrations in
other solvents were calculated from the Ostwald coefficients (L) and
Dioxygen Isotope Exchange Studies. To determine whether oxygen
atom exchange occurred during the decomposition of 3b to 3c, 50 mg
samples of 3a dissolved in propionitrile were prepared in separate
(
40) van Eldik, R.; Gaede, W.; Wieland, S.; Kraft, J.; Spitzer, M.; Palmer,
D. A. ReV. Sci. Instrum. 1993, 64, 1355-1357.

Reaction of Diiron(II) Precursors with Dioxygen
Inorganic Chemistry, Vol. 35, No. 9, 1996 2593
eq 1, where X is the mole fraction, R is the gas constant, T is the
-
1
RT
X )
+ 1
(1)
(
P LV°
)
(
g)
(l)
temperature, P(g) is the partial pressure, and V°(l) is the molar volume.42
Dioxygen concentrations were varied over the 0.8-4.4 mM range at
room temperature. A linear correction for the contraction of propi-
onitrile as a function of temperature was previously measured41 and
is given in eqs 2 and 3. This correction was applied to the dioxygen
F ) 0.8011(4) - 0.0009801(90)T (T in °C)
(2)
(3)
[
O2]
22°C
[
O ] )
FT
2
T
F22°C
concentration whenever the rate constant depended upon this value.
Once it was determined that the decay phase of the reaction was not
[O ]-dependent, the reaction conditions for these runs were varied
2
beyond the pseudo-order limits, employing complex concentrations as
high as 1.0 mM.
Rate constants were derived from optical traces obtained at 600 nm,
a wavelength at which both the formation and the decomposition of
the peroxide intermediate can be observed. The data were analyzed
with Kaleidagraph (Abelbeck Software) running on a Macintosh IIfx
computer and fit to model-dependent equations by nonlinear least-
squares regression. The formation and decay phases of the reaction
were studied independently by varying the temperature and time base.
Traces used for curve fitting were all recorded to at least 3 half-lives.
For the formation of the peroxo species, optical density data were fit
to eq 4, a generalized first-order growth expression, where A
t
is the
A ) A + (∆A)(1 - exp(-kobst))
(4)
t
∞
absorbance at time t, A is the absorbance at infinite time, and ∆A is
∞
the change in absorbance during the reaction. Errors provided for the
kinetic constants are based on the standard deviation from the mean
for multiple runs under identical conditions. Errors for the activation
parameters are based on the propagated errors from the linear fit of
the Eyring plot or plot of ln(k) vs pressure. Beer’s law analysis of the
fitting parameters allowed the determination of the extinction coef-
ficients at 600 nm for 1b-3b. These values were used to convert the
optical density to concentration prior to the curve fitting of decay traces
Figure 2. (A) Optical spectra of 3b and 3c in EtCN. (B) Sample kinetic
traces for the growth (top and right axes) and decay (left and bottom
axes) of 3b. Experimental traces are overlaid by a least-squares fit of
the data to the equations given in the text. Growth phase conditions:
-
55 °C, 0.40 mM 3a, 4.52 mM O
2
. Decay phase conditions: 55 °C,
0
.125 mM 3a, 4.21 mM O
2
.
over a 6 min period. The residual ketyl was then flushed from the
system with a small volume of dry THF and then dry EtCN. The
experiments were run immediately thereafter. Solvents for these
experiments were freshly distilled as discussed above. Oxygenated
water was added to the oxygenated propionitrile solution just prior to
reaction with the dry, anaerobic iron solution.
to eq 5, where [B]
t ∞
is the concentration of B at time t, [B] is the final
concentration of B, and ∆[B] is the change in the concentration of B
during the reaction. Under certain conditions for compound 1a, both
the formation and decay could be observed with the same time base.
Such data could be fit to the sum of a first-order growth and a second-
order decay (eq 6). Omission of the initial growth phase and refitting
Results
1
Oxygenation of the Diferrous Complexes 1a-3a. The
reaction of compounds 1a-3a with dioxygen at low temperature
leads to the formation of 1b-3b, diferric species presumed to
contain a symmetrically bridging peroxide ligand. Previous
spectroscopic studies of these complexes are consistent with
[
B] ) [B] +
(5)
t
∞
1
[B]
+
kobst
∆
1
[
B] ) [B] + (∆[B])(1 - exp(k1,obs)) +
(6)
t ∞
1
[B]
+
k2,obs
t
1
1
2
2
either a µ-η :η or a µ-η :η geometry, although the former is
∆
-
1
preferred on the basis of the 875-900 cm O-O stretching
the data to a simple second-order decay yielded identical rate constants.
Under conditions of incomplete separation of the first and second phases
of the reaction, kinetic results for the decay reaction are reported as
fits of eq 5 to the decay phase of the experimental trace. This analysis
allowed accurate fitting of the traces with fewer independent variables.
Reactions performed under strictly anhydrous conditions were used
to provide zero points prior to studies of the effect of added water on
the oxidation rates. Procedures for these experiments were slightly
altered to ensure a minimum amount of residual water. Prior to these
experiments, THF solutions of sodium benzophenone ketyl were passed
through the flow lines until the blue color showed no loss in intensity
3
0
frequencies observed in the Raman spectra. This structure
has now been confirmed crystallographically in two related
4
3
complexes.
The reaction was followed by monitoring a
peroxide-to-iron charge transfer band, a broad feature centered
near 600 nm for all three complexes (Figure 2A). When studied
under conditions of excess dioxygen, simple first-order kinetic
traces were observed for the buildup of the peroxide intermedi-
ates (Figure 2B).
(
43) While this work was being reviewed, we learned of two X-ray
structural analyses of related (µ-1,2-peroxo)diiron(III) complexes:
Ookubo, T.; Sugimoto, H.; Nagayama, T.; Masuda, H.; Sato, T.;
Tanaka, K.; Maeda, Y.; Okawa, H.; Hayashi, Y.; Uehara, A.; Suzuki,
M. J. Am. Chem. Soc. 1996, 118, 701-702. Dong, Y.; Yan, S.; Young,
V. G. Jr.; Que, L. Jr. Angew. Chem. 1996, in press.
(
41) Karlin, K. D.; Wei, N.; Jung, B.; Kaderli, S.; Niklaus, P.; Zuberb u¨ hler,
A. D. J. Am. Chem. Soc. 1993, 115, 9506-9514.
42) Oxygen and Ozone; Battino, R., Ed.; Pergamon Press: Oxford, U.K.,
(
1
981; Vol. 7.

2
594 Inorganic Chemistry, Vol. 35, No. 9, 1996
Feig et al.
(9)
Table 1. Manometric Studies of the Dioxygen Uptake of 1a-3aa
k0
1
a y z 1a′
k-0
compd
peroxide formn reacn
peroxide decay reacn
1
2
3
a
a
a
n.d.b
0.87 ( 0.10
0.80 ( 0.10
0.60 ( 0.10
0.45 ( 0.10
0.40 ( 0.10
k ′
8 1b
1
9
1a′ + O2
(10)
(11)
c
a
In equivalents based on the amount of diiron(II) sample used in
the experiment. T ) -77 °C unless stated otherwise for the formation
and 22 °C for the decay reaction. Previously measured to be 1.0.
d[B] k k ′[1a][O ]
0
1
2
)
dt
k-0 + k ′[O ]
1 2
b
30
c
Measured at -25 °C.
k0
1
a + O y z 1a·O
(12)
(13)
2
2
k-0
k ′
1
1
a·O2
′K
1 + K [O ]
9
8 1b
d[B]
dt
k
1
0
[1a][O
]
2
)
(14)
0
2
corresponding to these reaction schemes are provided in eqs
1
1 and 14.
Reactions following the simple one-step addition (eq 7) can
be differentiated from those undergoing two-step mechanisms
-
1
-1
(
eqs 9 and 10 and eqs 12 and 13) by plotting (kobs) vs [O2] ,
-
1
-1
as indicated by eqs 15-17, respectively. The (kobs) vs [O2]
1
1
Figure 3. Plot of ln(V ) versus ln([3a]) (0.05-0.50 mM) used to
0
)
(15)
determine the reaction order with respect to complex for the formation
of 3b. The slope of the linear regression corresponds to the reaction
order.
kobs k [O ]
1 2
1
1
1
)
+
(16)
(17)
kobs k0 K k ′[O ]
0
1
2
Rate constants were derived from time-dependent absorbance
changes by using a nonlinear least-squares fitting procedure
described in the Experimental Section. Extinction coefficients
at 600 nm for 1b-3b of 1550 ( 58, 1559 ( 41, and 1421 (
1
kobs
1
1
)
+
k ′ K k ′[O ]
-
1
-1
1
0 1 2
2
2 M cm , respectively, derived from the fitting parameters
agree with those obtained by a Beer’s law analysis of the low-
temperature spectra (see Experimental Section). These values
were used to convert molar absorbances to concentrations for
the determination of initial rates and for curve fitting of the
second-order decay phase of the reaction, discussed below.
Manometric analysis at low temperature verified the previously
reported 1:1 stoichiometry for the reactions forming 2b and 3b
plot for the simpler reaction (eq 7) has a zero intercept at all
temperatures whereas, for the other two cases, the intercept of
the plot corresponds to (k0) or (k1′) . Since the rate constants
-1
-1
are temperature-dependent, so too will be the intercept. Plots
-
1
-1
of (kobs) vs [O2] for 1a-3a are shown in Figure 4 (see also
Tables S5-S7). These experiments clearly show that the
dioxygen reactivity of 1a differs from those of 2a and 3a.
Compound 1a must react by one of the two-step mechanisms.
The results shown in Figure 4 cannot differentiate these two
possibilities, however.
2
8,30
(Table 1).
Experimental Rate Laws for the Formation of 1b-3b.
Kinetic runs were made by varying the complex concentrations
over the range 0.05-0.50 mM. A plot of ln(V0) versus
ln([3a]) is depicted in Figure 3 (see also Table S1). (Note:
Throughout, “S” preceding the number of a table or a figure
indicates Supporting Information.) The reaction was first-
order (0.9-1.0 ( 0.1) with respect to complex in all cases.
When the [O2] was varied, however, different behavior was
observed for 1a compared to the other two compounds.
Clean first-order behavior with respect to dioxygen concen-
tration (Tables S2-S4) was exhibited by 2a and 3a. The
mechanism and rate law for the formation of 2b and 3b are
therefore as described by eqs 7 and 8, respectively. A par-
Activation Parameters for the Oxygenation of 1a-3a. The
temperature dependence of the reactions of 1a-3a with O2 was
investigated between -75 and +20 °C. An Eyring plot (Figure
5
and Tables S8-S10) of these data provided the activation
parameters listed in Table 2. The results for the three
compounds fall into two categories, divided as previously noted
for the experimental rate laws. Compounds 2a and 3a have
q
-1
almost identical activation parameters, ∆H ) 16 kJ mol and
S ≈ -120 J mol K .
q
-1
-1 20,44-47
∆
The rate constant for oxygenation of 1a has a dramatically
different temperature dependence, reinforcing the contrast
between the reaction mechanisms for this compound compared
to the other two. The enthalpic contribution to the activation
barrier is almost 3 times greater than those for 2a and 3a. The
k1
A + O2
d[B]
9
8 B
(7)
(8)
)
k [2a,3a][O ]
1 2
(44) Petrou, A. L.; Armstrong, F. A.; Sykes, A. G.; Harrington, P. C.;
dt
Wilkins, R. G. Biochim. Biophys. Acta 1981, 670, 377-384.
(
(
(
45) Zhang, M.; van Eldik, R.; Espenson, J. H.; Bakac, A. Inorg. Chem.
tial order with respect to dioxygen, 0.65 ( 0.10, was mea-
sured for the reaction of 1a, indicating that the mechanism
is more complicated. Two mechanisms can explain this
reduced order in O2, as shown in eqs 9-14. Rate laws
1994, 33, 130-133.
46) Becker, M.; Schindler, S.; van Eldik, R. Inorg. Chem. 1994, 33, 5370-
5371.
47) Projahn, H.-D.; Dreher, C.; van Eldik, R. J. Am. Chem. Soc. 1990,
112, 17-22.

Reaction of Diiron(II) Precursors with Dioxygen
Inorganic Chemistry, Vol. 35, No. 9, 1996 2595
Figure 5. Eyring plot for the reaction of 0.16-0.25 mM 1a (diamonds),
2
a (circles), and 3a (triangles) with ≈4.4 mM dioxygen in propionitrile
based on the observed second-order rate constants between -70 and
+
20 °C.
binding of dioxygen in this step has a very large enthalpy of
activation for an associative reaction, however, and is more in
keeping with a bond-breaking step.
Effect of Pressure on the Rate of Oxygenation. The effect
of pressure on the rate of 3b formation was studied in
propionitrile between 10 and 140 MPa (Figure 6 and Table S11).
4
8
From eq 18, we computed the volume of activation to be
q
∂
( ∂
(ln k)
P
∆V
)
-
(18)
)
T
RT
3
-1
-
12.8 ( 0.9 cm mol at 20 °C, a value consistent with the
simple addition mechanism discussed above. Efforts to measure
the volumes of activation for the formation of 1b and 2b were
not successful because of technical limitations of the high-
pressure stopped-flow system, which was unable to access
sufficiently low temperatures to study the reaction efficiently.
Effect of Water on the Rate of Oxygenation. When 1a
crystallized from MeOH/H2O, water coordinated to one of the
2
9
iron atoms. We therefore studied the effect of added water
on the rate of formation of the peroxide intermediates. Reaction
of 3a was investigated, because its greater thermal stability
allowed work at higher temperatures and therefore higher water
concentrations. At temperatures of -30 °C and below, ice
crystals formed in aqueous propionitrile solutions at the higher
end of the water concentration range, resulting in light scattering
and the inability to collect useful data. The addition of up to
Figure 4. Double-reciprocal analysis for the formation of 1b (A), 2b
1
% (v/v) water (0.5 M) at 20 °C had no effect on the reaction
(B), and 3b (C) based on eqs 11 and 12. Data were collected between
-
70 and +20 °C and at dioxygen concentrations between 0.1 and 0.48
rate (Figure S1A and Table S12), indicating that if water was
ligated, as it was in the solid state, it was readily displaced by
dioxygen. An interesting observation during this study was that
the extinction coefficient of the peroxide complex increased
linearly by up to 60% over the range of water concentrations
studied (Figure S1B).
Decomposition of the Diferric Peroxo Species 1b-3b. The
complexes formed initially from the reaction of dioxygen and
compounds 1a-3a are all metastable. When kept sufficiently
cold, they survive indefinitely. The three compounds have quite
mM. Data at each temperature were fit to a linear least-squares
regression (solid lines).
Eyring plot in Figure 5 assumes that the interaction of 1a with
O2 is an elementary reaction with a single barrier. The situation
has already been shown to be more complicated, however. By
use of the rate constants derived from the slopes and intercepts
of Figure 4A at each temperature, the effective activation barriers
for the two phases of the reaction can be separated. These
values have different meanings, however, depending on which
model is used to define the mechanism. In the case of eqs 9-11,
the activation parameters derived from the intercepts in Figure
different decomposition rates and temperature sensitivities,
28,30
however.
The most stable adduct is 3b, which survives for
minutes to hours at room temperature depending on the
concentration and is completely stable below -35 °C. The least
stable is 2b. An initial report indicated that this peroxide adduct
4
A correspond to k0, the forward reaction rate constant in a
ligand rearrangement equilibrium, whereas the model in eqs 12-
4 indicates that this value corresponds to the reversible addition
1
30
did not even form at room temperature. The statement is,
of dioxygen to the ferrous complex. The entropy of activation
derived from these intercepts is negative, a result which supports
the interpretation of the mechanism outlined in eqs 12-14. The
(
48) van Eldik, R.; Merbach, A. E. Comments Inorg. Chem. 1992, 12, 341-
378.

2
596 Inorganic Chemistry, Vol. 35, No. 9, 1996
Feig et al.
Table 2. Activation Parameters for Formation and Decay of Peroxide Intermediates 1b-3b and Other Relevant Protein and Model Systems
q
3
∆Hqb (kJ mol-1
39.6 ( 2.4
∆Sqb (J mol-1 K-1
)
∆
Hq a (kJ mol-1
42.2 ( 1.6
)
∆Sq a (J mol-1 K-1
Oxygenation Reactions
-63 ( 6
)
∆V (cm mol-1)
)
1
a f 1b
a f 2b
-114 ( 9
-65 ( 6
3
9.2 ( 1.6
2
3
16.5 ( 0.4
15.4 ( 0.6
-114 ( 2
-121 ( 3
a f 3b
-12.8 ( 0.9
-21 ( 1c
I
II
LCu f LCu (O
LCo f LCo (O
Hr f oxyHr
2
)
II
II
d
f
2
)
+0.4 ( 0.5
+13.3 ( 1.1
e
16.8
-46
Mb f oxyMb
+5.2 ( 0.5g
Peroxide Decomposition Reactions
1b f 1c
2b f 2c
3b f 3c
113.3 ( 7.7
51.9 ( 2.7
80.6 ( 3.9
187 ( 27
-47 ( 11
74 ( 14
370 ( 6
0.2 ( 7.0
-8 ( 12
73 ( 24
8
MMOh
peroxo f Q
Q f decay
111
75
147
8
a
Activation parameters in this column are derived from Eyring plots of observed rate constants (Figures 6 and S2). b Activation parameters for
1
a in this column correspond to values derived from the slope (top value) and intercept (bottom value) of the double-reciprocal analysis in Figure
5
A. Based on the model in eq 17; the top value relates to k
1
′ and the bottom value k
1
′K
0
. For 3b, the activation parameters correspond to k
3
(top
′ (bottom value) derived from Figure S3. Data taken from ref 47. d Data taken from ref 46. From Themiste zostericola in Tris
c
e
vlaue) and K
3
k
2
4
5
f
57
g
48
buffer, pH 8.2, I ) 0.1 M.
From T. zostericola in Tris buffer, pH 8.5, I ) 0.10 M.
From sperm whale in Tris buffer, pH 8.5, I ) 0.1 M.
h
From Methylococcus capsulatus (Bath) in MOPS buffer, pH 7.0, I ) 25 mM in the presence of protein B.20
decay reaction as a disproportionation that yields 1 equiv of
dioxygen and either 2 equiv of a dinuclear ferric complex or 1
equiv of a tetranuclear species (Scheme 1). Mixed-isotope
1
6
18
16
18
studies using O2 and O2 showed that no O- O formed
during decomposition.
Nature of the Final Oxidation Products. The studies
reported here use the optical absorbance of the peroxo inter-
mediate to analyze its decay, and therefore do not reflect
subsequent reactions. In Scheme 1, we depict the products 1c-
3
c as tetranuclear species, the structure of 3c having been
1
1
crystallographically determined. Moreover, an X-ray study
of the dinuclear ferric system [Fe2O(HPTB)(OBz)](NO3)2, where
HPTB is identical to Et-HPTB but lacks the N-ethyl groups,
similarly revealed aggregation of two dinuclear units to form a
49
tetranuclear species. We therefore assign the structures of 1c-
c as indicated in the scheme, but recognize that the initial
Figure 6. Plot showing the effect of pressure on the reaction of 0.20
mM 3a with dioxygen between 10 and 140 MPa.
3
product of the decomposition may be dinuclear.
strictly speaking, incorrect, but the adduct decomposes so rapidly
at this temperature that its blue color cannot be visually detected.
At -77 °C, 2b is quite stable. Compound 1b was reported
correctly to have stability properties between these two ex-
tremes.28 Addition of polar aprotic cosolvents such as DMSO
significantly stabilizes acetonitrile solutions of 2b and 3b, but
a detailed investigation of these effects was beyond the scope
of the present study.
Rate Laws for the Decay of the Peroxide Intermediates
1b-3b. A series of experiments systematically varying the
concentrations of dioxygen and diiron(II) complexes was
undertaken in a manner analogous to that described above for
the formation of the intermediates, except that the temperatures
used for these studies were higher to promote faster decomposi-
tion. The traces for this phase of the reaction fit to second-
order decay equations, providing the first indication that the
process was bimolecular. As shown in Figure 7 (see also Tables
S13-S15), the reactions all have a second-order dependence
on the concentration of the peroxide intermediate.
30
We first consider the stoichiometry of the decomposition
reaction. As noted above, the diiron(II) compounds react in a
1
:1 ratio with dioxygen to form peroxo intermediates. The
original reports on compounds 2b and 3b noted that iron(III)
decomposition products formed upon warming and that, at best,
substoichiometric oxygen atom transfer occurred when good O
The reaction order with respect to dioxygen was a little more
difficult to extract from the data. The effect of varying the
dioxygen concentration is shown in Figure 8 (see also Tables
S16-S18). The simplest case is the decomposition of 3b, in
which the reaction is clearly independent of the amount of
dioxygen added. This result derives from the large difference
between the rates of formation, which is O2 dependent, and
decay of 3b. By the time the decay begins, the reaction of 3a
with O2 to form 3b is complete. Therefore, the formation of
the peroxide adduct has no observable effect on the kinetics of
3
0
atom acceptors such as triphenylphosphine were present. By
carrying out more detailed manometric studies, we were able
to establish the stoichiometry shown in Scheme 1. Dioxygen
uptake was first monitored at low temperature, after which the
reaction vessel was allowed to warm to room temperature. The
net number of moles of dioxygen consumed by all three
compounds after completion of the decomposition reaction,
however, was half the amount taken up during the initial low-
temperature reaction step (Table 1). In other words, half of
the dioxygen initially consumed to form the peroxide adduct
was released following decomposition. We therefore write the
(
49) Chen, Q.; Lynch, J. B.; Gomez-Romero, P.; Ben-Hussein, A.; Jameson,
G. B.; O’Connor, C. J.; Que, L., Jr. Inorg. Chem. 1988, 27, 2673-
2681.

Reaction of Diiron(II) Precursors with Dioxygen
Inorganic Chemistry, Vol. 35, No. 9, 1996 2597
vary from 52 kJ mol- for 2b to 113 kJ mol-1 for 1b, a more
significant deviation than was observed for the oxygenation
phase. A surprising feature is the dramatic difference in
1
-
1
-1
activation entropy of 250 J mol
K
for the three reactions.
Although the enthalpic variation is large, the large entropy
changes clearly determine the relative stabilities of these
intermediates. This feature is discussed in detail below.
In a final kinetic experiment, we investigated whether the
rate law analogous to that outlined in eq 19 applies to the decay
of all three compounds. To obtain this information, the complex
concentration was varied at different temperatures. For the
simplest case of eq 19, a single elementary reaction applies and
the activation parameters are invariant. More complex behavior
q
q
is indicated by changes in either ∆H or ∆S , in which case a
double-reciprocal analysis analogous to eq 16 or 17 must be
used. The activation parameters for the decomposition of 1b
and 2b were invariant over a concentration range of 0.12-1.00
mM. The same could not be said for 3b, however. Assuming
that a pre-equilibrium step was involved, similar to the behavior
0
Figure 7. Reaction order plot showing the ln(V ) for peroxide
intermediate decomposition versus ln([1b,2b,3b]), the concentration at
which second-order decomposition is observed to begin: 1b, diamonds;
2
b, circles; 3b, triangles.
-
1
-1
of the oxygenation of 1a, (kobs) was plotted against [3b]
and a linear correlation was found at each temperature (Figure
S3 and Table S22.). This more complex system can therefore
be described by the reaction scheme and rate laws given in eqs
2
1-24.
k3
B y z B′
(21)
(22)
(23)
k-3
k ′
2
9
B + B′
8 C
2
d[B]
dt
k k
3 2
′[3b]
-
)
k-3 + k ′[3b]
2
Figure 8. Reaction order plot showing the effect of [O ] (0.2-4.4
2
1
1
k-3
+
mM) on the observed second-order rate constant for the decay of
)
(24)
peroxide intermediates 1b (5 °C, diamonds), 2b (22 °C, circles), and
k
obs
k
3
k k ′[3b]
3
2
3
b (55 °C, triangles); 0.15-0.75 mM.
Discussion
the decay phase. For 1b and 2b, the separation of the two
phases is less distinct. Since the kinetics are followed by using
the optical band of the intermediate, no suitable absorptions
being present in the spectra of the reaction products, a partial
negative reaction order with respect to dioxygen is measured.
This result derives from the formation of the peroxide discussed
above and does not relate to the actual decomposition step. A
minimal rate law describing this decay reaction is given by eq
The Role of Steric Factors in the Oxygenation Reaction.
The three model compounds studied here are structurally quite
similar (Scheme 1). The donor atoms of the dinucleating ligand
are identical, and each complex has an additional benzoate
bridge. Each iron atom is five-coordinate with an open binding
site for dioxygen. The four pendant donor nitrogen atoms derive
from slightly different residues, pyridyl, 6-methylpyridyl, and
N-ethylbenzimidazolyl, all of comparable basicity. Despite these
similarities, the kinetic behaviors of these three compounds in
their reactions with dioxygen are substantially different.
Three parameters can be used to define the dioxygen-binding
pockets in 1a-3a, diameter, depth, and capping potential. The
diameter of the pocket is determined by the [Fe2(OR)(O2CPh)]2
core and is identical for the three complexes. The depth of the
pocket is defined by the nature of the four pendant bases,
specifically the size of the aromatic rings which create it.
Compound 3a has a deeper pocket than 1a and 2a because of
the fused rings of the benzimidazole moieties. Ligands having
quinoline rings have been employed to provide extra pocket
depth to pyridine-like bases in copper(I)-dioxygen reaction
1
9. By combination of the two phases of the reaction, the rate
law in eq 20 can be written, the integrated form of which is
d[B]
dt
2
-
)
) k [B]
(19)
(20)
2
+
d[B]
dt
2
k [A][O ] - k [B]
1
2
2
shown in eq 6 above. One could add an additional term,
k-1[B], assuming a back-reaction to re-form the diiron(II)
-
starting material. In these intermediate regimes, however,
satisfactory fits were obtained without such a term, showing
that, from a kinetic standpoint, it was insignificant under the
conditions studied.
Activation Parameters for the Decomposition of 1b-3b.
The temperature dependence of the decomposition reaction was
studied in propionitrile over a wide range, 80 °C for 2b and,
because of the slower decomposition rates, 50 °C for 1b and
4
1
chemistry. Finally, partial capping of the pocket is effected
by the methyl groups of the HPTMP ligand (Figure 9).
These structural features, rather than minor electronic dif-
ferences, most likely account for the kinetic behavior of 1a-
3a. The vacant sites on the iron atoms lie within the pockets
just described. The pocket dimensions of 3a provide little to
no obstruction to approach of a small ligand like O2. The effect
3b (Figure S2 and Tables S19-S21). The activation enthalpies

2
598 Inorganic Chemistry, Vol. 35, No. 9, 1996
Feig et al.
binding site is blocked by partial methyl group capping. The
shift from bridging to terminal carboxylate coordination does
not directly afford a passageway for dioxygen-binding, however,
unless, in the process, additional ligand rearrangement occurs.
This motion contrasts with the carboxylate shift reactions
postulated in the oxidation of [Fe2(OH)(O2CCH3)2(Me3TACN)2]-
(
ClO4), where the shift directly provides a dioxygen binding
5
1
site. Dissociation of a pyridyl arm is the other alternative.
The steric barrier to dioxygen approach to the metal center
provided by the four methyl groups would be relieved by partial
or complete dissociation of one of these arms. Iron(II) is a rather
labile metal, and the dissociation rates for any single arm are
5
4
likely to be quite rapid. The complex as a whole is stabilized
by the chelate effect, which implies that, statistically, it is
unlikely for more than one arm to dissociate at a given time.
If either of the foregoing proposals is correct, the same type
of behavior should be occurring for 2a and 3a. The difference
is that, for these compounds, the dioxygen-binding step is not
affected and the kinetics of adduct formation do not reflect the
process. This point is important, for it implies inner-sphere
coordination prior to, or concomitant with, the rate-determining
step in the oxidation to form peroxide adducts. For a series of
capped cyclidene iron(II) complexes, complicated kinetic changes
accompanied the O2 binding process when the cavity was
5
5,56
constricted to a size which prevented O2 access.
It was
proposed that, under such conditions, autoxidation proceeded
by an outer-sphere electron transfer mechanism. Compound
1
-
a is unreactive with dioxygen over a period of hours below
77 °C in propionitrile. Brief exposure to higher temperatures
followed by cooling results in the formation of stable 1b. An
outer-sphere mechanism for autoxidation therefore appears not
to be accessible in the present systems, or if it is, it must occur
on a time scale longer than that investigated.
Figure 9. Space-filling models of 1b-3b showing the dioxygen-
binding site from two perspective views.
of capping by HPTMP is far more obstructive, however. There
is a void space below these convergent groups which could
accommodate a bridging peroxide ligand once it gets there
The scenario in eqs 12-14 offers a second possible mech-
anism to account for the kinetic results. In this case, an iron-
dioxygen adduct other than the µ-1,2-peroxo species 1b must
exist as an intermediate. Such a species could contain a terminal
superoxide ligand coordinated in a monodentate fashion to one
iron in a mixed-valent core. This adduct might form in the
case of 1a oxidation because of steric constraints. For reasons
discussed above, access to the vacant binding site is inhibited.
Terminal coordination might anchor the dioxygen molecule in
place as additional ligand motion ensues to allow access to the
second iron atom and peroxide formation.
(Figure 9). It is the process of accessing this compartment that
differentiates the dioxygen reactions of these three systems. The
decay of the peroxide intermediates is affected by a different
parameter. Peroxo species 3b decays much more slowly than
1
b or 2b, and the temperature dependence of the reaction
indicates that a dissociative step is important at lower temper-
atures. In this reaction, the relevant structural feature is the
depth of the binding pocket (Figure 9). The fused rings of the
benzimidazole groups of 3b are substantially larger than the
pyridyl arms of 1b and 2b, resulting in a deeper well and greater
protection of the bound peroxide from bimolecular decomposi-
tion.
These two possible mechanisms can in principle be differenti-
ated by the activation parameters derived from the slopes and
intercepts of the double-reciprocal plots. In the case of eqs
9-11, these parameters correspond, respectively, to a composite
Comparison of the Mechanism of Formation of 1b to That
of 2b and 3b. The temperature-dependent results most clearly
illustrate the contrast in mechanism for the reactions of dioxygen
with 1a compared to either 2a or 3a. Both the diminished order
in dioxygen and the behavior of 1a shown in Figure 4 can be
explained by the two-step mechanisms in eqs 9-14. In the
first scenario, eqs 9-11, a ligand dissociation equilibrium is
invoked. There are two chemically reasonable possibilities for
the 1a h 1a′ equilibrium. One is a carboxylate shift,50 a process
previously implicated in the oxidation chemistry of coordina-
tively saturated diiron(II) complexes51 and known to accompany
constant (K0k1′) and a unimolecular ligand rearrangement rate
constant (k0). In the second case, eqs 12-14, the parameters
represent (K0k1′) and a rate constant for the unimolecular
q
rearrangement of 1a·O2 to 1b. The ∆S value is negative in
both plots (Table 2 and Figure S4), which implies a highly
ordered transition state. This result favors a bimolecular process,
and for this reason, we slightly prefer the latter model. The
large enthalpy of activation implies that bond stretching or
breaking is probably involved upon approach to this transition
state, which can arise from the need to alleviate the steric
constraint imposed by the partial capping of the binding site.
redox state changes of the diiron centers in both sMMO
hydroxylase and RNR.8
,52,53
(52) Rosenzweig, A. C.; Nordlund, P.; Takahara, P. M.; Frederick, C. A.;
The steric difference between 1a
Lippard, S. J. Chem. Biol. 1995, 2, 409-418.
and the other two compounds is that the putative dioxygen-
(
(
53) A° berg, A. Ph.D. Thesis, Stockholm University, 1993.
54) Porterfield, W. W. Inorganic ChemistrysA Unified Approach;
Academic: New York, 1993.
(50) Rardin, R. L.; Tolman, W. B.; Lippard, S. J. New J. Chem. 1991, 15,
4
17-430.
(55) Sauer-Masarwa, A.; Dickerson, L. D.; Herron, N.; Busch, D. H. Coord.
Chem. ReV. 1993, 128, 117-137.
(56) Busch, D. H.; Alcock, N. W. Chem. ReV. 1994, 94, 585-623.
(
51) Feig, A. L.; Masschelein, A.; Bakac, A.; Lippard, S. J., Manuscript in
preparation.

Reaction of Diiron(II) Precursors with Dioxygen
Inorganic Chemistry, Vol. 35, No. 9, 1996 2599
Interpretation of the Activation Parameters for Adduct
Formation and Comparisons to Other Systems. The activa-
tion parameters of a reaction reflect the transition state, allowing
deductions about the reaction coordinate and comparisons to
other known reactions to be made. Theoretically, the Eyring
equation is valid only for elementary reactions. When more
complicated reactions are subjected to such an analysis, negative
curvature in the plot can be encountered, indicating that two
separate steps are rate-limiting at the temperature extremes. In
practice, however, it is often possible to study complex rate
equations by this methodology so long as curvature is absent
over a given temperature range. The values then correspond
to the energy barrier for the rate-determining elementary step
in that range. The Eyring plot shown in Figure 5 exemplifies
such behavior for the reaction of 1a with dioxygen. Over the
temperature range studied, a single step dominates the activation
barrier and a linear plot results. The reactions of 2a and 3a are
simple second-order reactions, studied under pseudo-first-order
conditions, and linearity is predicted.
The activation enthalpy for the reactions of 2a and 3a with
dioxygen is remarkably similar to that of deoxyHr formation.
It is possible that this agreement is merely a coincidence, but it
may represent the energy barrier to reach the transition state in
the interaction of dioxygen with a reduced diiron(II) center
having an accessible binding site to form a peroxodiiron(III)
species. In Hr, one iron atom has such a vacant site. Since
the structures of the peroxide adducts differ for Hr and the model
compounds, the implication is that the binding of dioxygen and
the associated two-electron oxidation of the iron center comprise
the rate-determining step. Structural rearrangements to form a
similar to that encountered in an acid-base titration. The
changes in optical density therefore do not relate to a binding
event.
Two Possible Peroxo Intermediate Decay Mechanisms.
The results for decay of the peroxide intermediates provide some
interesting clues about the behavior of these metastable species.
This reaction is of primary interest to those seeking to
understand non-heme iron oxo-transfer chemistry. It is this step
in the overall transformation which diverts small-molecule
model compounds from mimicking the chemistry of the sMMO
and R2 active sites. As already discussed, the model compounds
form peroxide intermediates similar to those found in the native
protein systems, and they do so with analogous activation
parameters provided that the diiron center is sterically accessible.
Yet these same models cannot activate substrates or perform
catalysis. The reason for this behavior is apparent from the
kinetics of the decay reactions, which are second-order with
respect to complex. In order for a catalyst to generate a highly
energetic intermediate with enough oxidizing potential, for
example, to break a C-H bond, all lower energy pathways
leading to autoxidation must be avoided. The bimolecular reac-
tion uncovered here is one such pathway, knowledge of which
should facilitate the development of strategies to avoid it.
We now consider the nature of the peroxide decomposition
step. In work with two other model compounds, [Fe2(OH)(O2-
CCH3)2(Me3TACN)2](ClO4) and [Fe2(O2CH)4(BIPhMe)2], we
proposed a (µ4-peroxo)diiron(II)diiron(III) transition state.51
Such an activated complex would probably arise by a different
mechanism for peroxide intermediates 2b and 3b for two
reasons. One relates to the apparent irreversibility of their
formation. Once dioxygen reacts with the diiron(II) compounds,
it is impossible to reverse this process at low temperature, even
(µ-1,2-peroxo)diiron(III) intermediate in the case of the models,
or the terminal hydroperoxide with a hydrogen-bond to the oxo-
bridge in Hr, might be subsequent, kinetically silent, steps.
-
3
in vacuum down to 10 Torr. Even warming a solution of 2b
or 3b under vacuum does not regenerate 2a or 3a. The second
reason is based on the relative rates of the formation and decay
of the peroxo intermediates, especially 3b. The half-lives of
the two phases of the reaction differ by 4-5 orders of magnitude
depending on complex concentration. By the time 3b begins
to decompose, no 3a is available as a reactant. For 1a, however,
the reaction with dioxygen is reversible, with a P1/2 of 6 Torr
The measured value of -12.8 cm mol-1 for the activation
volume of the reaction is another parameter which confirms,
along with the negative value of the activation entropy, the
highly structured nature of the transition state. Strongly negative
3
3
-1
volumes of activation (≈-20 cm mol ) were also reported
for the reaction of copper(I) complexes with O2.4
6,57
These
values were interpreted in terms of bond formation accompanied
2
8
at -35 °C in CH2Cl2. Although this binding is reasonably
strong, it is possible that the second-order decomposition results
from reversal of the oxygenation step followed by reaction of
another equivalent of the peroxo intermediate with the resultant
II
by intramolecular electron transfer to produce the Cu -O2
species. By contrast, volumes of activation for the binding of
dioxygen to macrocyclic cobalt(II) complexes were close to
zero and interpreted in terms of a substitution-controlled
binding process.45 In the case of the dioxygen carrier proteins
myoglobin, hemocyanin, and hemerythrin, the volumes of
activation for the binding of dioxygen differ significantly from
those reported above due to influence of the protein environ-
1
a. Given the similarity of the three compounds, however, it
is likely that the decomposition reactions follow similar
pathways, with the only perturbations arising from different
steric constraints.
Since a mechanism in which an equivalent of the peroxide
adduct reacts with the diiron(II) starting material seems unlikely
under our conditions (excess O2 in solution), two possibilities
remain for the decomposition reaction. These possibilities are
presented in Scheme 2. Both fit the kinetic data, resulting in
second-order decay reactions with the loss of an equivalent of
dioxygen. The first pathway (A) is analogous to the decay of
4
7,58
ment.
Solvent Effects. Since water binds to 1a,28 it could poten-
tially affect the kinetics of the oxygenation reaction by blocking
a binding site. Extensive studies with added water, however,
showed no significant effect on the kinetic parameters for
oxidation of 3a in the 2.0-500 mM range. Exploratory work
in acetonitrile and CHCl3 similarly failed to reveal any dramatic
changes. The absolute rates were slightly slower, a factor of
secondary and tertiary alkyl peroxide radicals, which proceeds
59-61
through a tetraoxide intermediate.
Following decomposi-
2
-3, in CHCl3, consistent with a passive role for solvent in
tion, an equivalent of alcohol, an equivalent of ketone, and a
molecule of dioxygen are formed. The other mechanism (B)
involves more conventional nucleophilic attack and dispropor-
these reactions. A significant effect on the extinction coefficient
of 3b was observed as a function of water concentration. The
linear dependence of this effect indicated that it did not arise
from water coordination, which would have a hyperbolic form
16
18
tionation. Mechanism A was evaluated by the O2/ O2-
labeling experiment discussed above. Since a new O-O
bond forms in the decomposition step, the central two
(
57) Goldstein, S.; Czapski, G.; van Eldik, R.; Cohen, H.; Meyerstein, D.
J. Phys. Chem. 1991, 95, 1282-1285.
(59) Bartlett, P. D.; Guaraldi, G. J. Am. Chem. Soc. 1967, 89, 4799-7801.
(60) Bartlett, P. D.; G u¨ nther, P. J. Am. Chem. Soc. 1966, 88, 3288-3294.
(61) Traylor, T. G.; Bartlett, P. D. Tetrahedron Lett. 1960, 30-36.
(58) Projahn, H.-D.; Schindler, S.; van Eldik, R.; Fortier, D. G.; Andrew,
C. R.; Sykes, A. G. Inorg. Chem. 1995, 34, 5935-5941.

2
600 Inorganic Chemistry, Vol. 35, No. 9, 1996
Feig et al.
Scheme 2
oxygen atoms of the tetraoxide depart as O2. In the O2-labeling
experiment, where 3b and 3b* were mixed, degassed, and
allowed to decompose under vacuum, no 16O- O was recov-
ered. Mechanism A can therefore be excluded, provided that
an immeasurable but kinetically competent quantity of the
diiron(II) starting material did not form. In mechanism B, the
dioxygen molecule released following decomposition is one
taken up during the first phase of the reaction. No O-O bonds
are formed during the reaction. Because dioxygen is released
after a rate-determining, second-order process in mechanism B,
it is impossible to determine whether this process occurs
immediately or occurs in a subsequent step. Both possibilities
are depicted in Scheme 2B.
reaction observed here is prevented from occurring in the protein
because the peroxodiiron(III) species is secluded deep within a
folded polypeptide. The measured activation parameters for the
bimolecular reaction, however, provide a measure of the
instability of the protein intermediates even though the decay
mechanisms differ. A model compound will decompose by the
lowest energy pathway, regardless of whether it is unimolecular
or bimolecular. If the transition state energies for the two
pathways are comparable, then both reactions will occur in
parallel. The kinetic traces in such cases would appear biphasic,
the sum of first-order and second-order decays. Therefore, the
activation enthalpies presented here for the bimolecular reaction
provide a lower limit for a comparable, unobserved unimolecular
mechanism.
18
The bottom pathway in Scheme 2B would provide a common
intermediate for the decay pathway presented here and the one
A reaction coordinate diagram reflecting the experimentally
determined activation parameters is presented in Figure 10. Not
all of the information necessary for constructing this diagram
was available, so approximations were made. For 1a, the value
proposed previously.5
1,62
A (µ4-peroxo)tetrairon adduct has been
63
crystallographically characterized. In this molecule, there are
II
III
only Fe(III) centers instead of the Fe 2Fe 2 transition state
postulated here. As mentioned above, loss of dioxygen from
28
of P1/2 for dioxygen binding is known (6 Torr at -35 °C)
-
1
2
b and 3b was not observed even under vacuum. The energy
and corresponds to a ∆G value of -9.6 kJ mol . Although
barrier for the back-reaction thus appears to be greater than that
for the decomposition. During the bimolecular collision,
however, the kinetic energy of the two molecules might lead to
a species with sufficient vibrational potential energy that O2
dissociation is observed. Once O2 is released, newly formed
diiron(II) material would be able to react rapidly with a molecule
of the peroxo intermediate to lead to product formation.
Relation of the Kinetic Parameters to the Chemical
Properties of the Compounds. The stability of the peroxide
adducts 1b-3b is a property of considerable interest, since these
species are analogs of the biological precursors to high-energy
intermediates which oxygenate substrates at the active sites of
dinuclear non-heme iron enzymes. The bimolecular decay
the temperature dependence of the binding has not been reported,
-
1
-1
a crude approximation of 80 J mol
K
can be made based
6
4
on the measured values reported for Hr. The ∆G values for
the decomposition of the peroxide intermediates were chosen
arbitrarily, but since the reactions are irreversible, the importance
of the c f b activation barriers is minimal. The reactions of
2a and 3a with dioxygen cannot be reversed by the application
of vacuum. One may therefore conservatively set an upper limit
on dioxygen binding at P1/2 e 0.05 Torr. This inequality leads
to an upper bound of ∆G < -19 kJ mol for 2a and 3a, the
value used in preparing Figure 10. The depth of the wells
containing 2b and 3b could be substantially greater than actually
shown. Microcalorimetry experiments would be useful to
determine these energies. Even by using the maximum values
for ∆G it can be seen that formation of 2b is unlikely to be
reversible at accessible temperatures. The potential for observ-
-
1
(
62) Liu, K. E.; Feig, A. L.; Goldberg, D. P.; Watton, S. P.; Lippard, S. J.
In The ActiVation of Dioxygen and Homogeneous Catalytic Oxidation;
Barton, D. H. R., Ed.; Plenum: New York, 1993; pp 301-320.
63) Micklitz, W.; Bott, S. G.; Bentsen, J. G.; Lippard, S. J. J. Am. Chem.
Soc. 1989, 111, 372-374.
(
(64) De Waal, D. J. A.; Wilkins, R. G. J. Biol. Chem. 1976, 251, 2339-
2343.

Reaction of Diiron(II) Precursors with Dioxygen
Inorganic Chemistry, Vol. 35, No. 9, 1996 2601
of Hr, it has been proposed that the peroxide adduct is stabilized
by hydrogen bonding and that this interaction somehow prevents
its decomposition by pathways normally observed for sMMO
and RNR. In the present model systems, however, the ground
state of the reversible binder is higher than that of the
intermediates which undergo only irreversible decomposition.
By formation of a dioxygen adduct which is too stable, the
barrier to reversibility cannot be overcome and only forward
pathways leading to the cleavage of the O-O bond are observed.
Other systems such as the cyclidenes and capped porphyrins
also bind dioxygen reversibly. It is possible that such a ground
state effect operates in these cases too. The diiron centers in
RNR and sMMO have vacant coordination sites on both iron
atoms and the formation of a 1,2-bridging peroxide has been
proposed in each case. Such a peroxide intermediate may be
too low in energy to undergo O2 release and therefore is
committed to a pathway leading to dioxygen activation.
‡
Figure 10. Reaction coordinate diagram showing ∆G for the reactions
of 1a-3a with dioxygen at +20 and -70 °C.
ing reversibility with 3b is somewhat greater, but again the ∆G
value for 3b formation could be substantially more favorable
than actually depicted.
Summary and Conclusions
The dioxygen uptake and subsequent decomposition reactions
for three diiron(II) model compounds have been studied by
stopped-flow spectroscopy. The mechanistic results show that
two of the models accurately mimic the oxygenation chemistry
of Hr. When sufficient ligand steric constraints prevent efficient
access to the diiron(II) center by the incoming dioxygen
molecule, a two-step mechanism is required to explain the
kinetic data. This process could involve a pre-equilibrium
involving a ligand dissociation process or, more likely, the
intermediate reversible formation of a dioxygen adduct other
than the symmetrically coordinated peroxide. This species could
be a terminal superoxide adduct. The key steric factor affecting
the addition of dioxygen to the diferrous compounds was
identified as capping of the binding site (compound 1a).
Thermal decomposition of the peroxide intermediates occurs
in all three cases by a nonbiomimetic second-order rate-
determining step. In this reaction, an equivalent of dioxygen
is evolved such that the stoichiometry of the overall transforma-
tion is 0.5 equiv of O2 per dinuclear center, similar to that
The formation of 1b and the decay of 3b are not elementary
steps, although linear behavior in an Eyring plot was observed.
The shallow wells shown as part of the transitions from 1a to
1b and from 3b to 3c (Figure 10) represent the intervening steps
discussed above. The heights of the net barriers for the reactions
have been measured, but the two component steps contributing
to the overall transformations have not been delineated.
The variation in the activation parameters among the three
peroxide intermediates reveals the primary difference in their
q
stabilities to be entropic. For 1b and 3b decomposition, ∆S is
large and favorable. Such a finding is unusual for a bimolecular
event where the transition state is usually more ordered than
the reactants. There are two potential explanations for this
q
behavior. One is that ∆S reflects solvent reorganization. The
reactants are divalent cations and will be highly solvated. On
approaching the transition state, charge increases to +4. This
process changes the surface-to-volume ratio of the charged
species and could release many solvent molecules, contributing
to a large positive increase in the entropy of activation. Addition
of DMSO has been reported to stabilize the peroxo adducts,
observed for reactions of other non-heme diiron model com-
51
pounds with dioxygen.
In order for efficient biomimetic
30
especially 2b. Altering the nature of this solvation might exert
very significant effects on the entropic term. The second, more
likely, possibility is that the transition state involves release of
oxygenation of substrates to occur, this second-order decay
reaction must be subverted. The trends observed for these three
compounds reveal that, whereas capping is important for the
O2-binding step, the depth of the coordination pocket plays a
more significant role in the decomposition reactions. The
stability of the µ-1,2-peroxide intermediates studied here
indicates that such species are not likely to be the species directly
responsible for oxygen transfer in the enzymes. This statement
implies either that O-O bond cleavage occurs prior to C-H
activation or that the dioxygen adduct isomerizes to a more
activated form prior to its interaction with a substrate in
monooxygenase reactions.
1
equiv of dioxygen (Scheme 2B, bottom pathway). Such a
mechanism would offset the effect of a bimolecular collision,
and there would be no change in the net number of species.
In all three cases, the activation enthalpy for decomposition
of the peroxide adduct is a factor of 3-5 greater than that for
its formation. The relative values of the activation entropy for
the forward (k2) and reverse (k-1) reactions therefore dictate
the relative decomposition rates. In addition, ∆G for converting
diferrous material to the peroxide adduct plays a key role in
determining the reversibility of the reaction. This value affects
the depth of the well in which the intermediate resides (Figure
Acknowledgment. This work was supported by grants from
the National Institute for General Medical Science, the AKZO
Corp., and the Deutsche Forschungsgemeinschaft. We wish to
thank V. L. Pecoraro and A. Bakac for interesting discussions.
10) and therefore the partition ratio between the forward reaction
and back reaction. The relatively high energy of 1b most likely
reflects steric interactions between dioxygen and the methyl
groups in the pocket it occupies (Figure 10).
Destabilization of the peroxide intermediate in this manner
leads to an interesting possibility which goes against current
dogma for rationalizing reversible dioxygen binding. In the case
Supporting Information Available: Tables S1-S22, containing
kinetic constants used in preparing the figures, and Figures S1-S4 (26
pages). Ordering information is given on any current masthead page.
IC951242G