Ultrafast and ultraslow oxygen atom transfer reactions between late metal centers

Article abstract of DOI:10.1021/ja065713h

Oxotrimesityliridium(V), (mes)₃Ir=O (mes = 2,4,6- trimethylphenyl), and trimesityliridium(III), (mes)₃Ir, undergo extremely rapid degenerate intermetal oxygen atom transfer at room temperature. At low temperatures, the two complexes conproportionate to form (mes) ₃Ir-O-Ir(mes)₃, the 2,6-dimethylphenyl analogue of which has been characterized crystallographically. Variable-temperature NMR measurements of the rate of dissociation of the μ-oxo dimer combined with measurements of the conproportionation equilibrium by low-temperature optical spectroscopy indicate that oxygen atom exchange between iridium(V) and iridium(III) occurs with a rate constant, extrapolated to 20 °C, of 5 × 10⁷ M^-1 s^-1. The oxotris(imido) osmium(VIII) complex (ArN)₃Os=O (Ar = 2,6-diisopropylphenyl) also undergoes degenerate intermetal atom transfer to its deoxy partner, (ArN) ₃Os. However, despite the fact that its metal-oxygen bond strength and reactivity toward triphenylphosphine are nearly identical to those of (mes)₃Ir=O, the osmium complex (ArN)₃Os=O transfers its oxygen atom 12 orders of magnitude more slowly to (ArN)₃Os than (mes)₃Ir=O does to (mes)₃Ir (k_OsOs = 1.8 × 10^-5 M^-1 s^-1 at 20 °C). Iridium-osmium cross-exchange takes place at an intermediate rate, in quantitative agreement with a Marcus-type cross relation. The enormous difference between the iridium-iridium and osmium-osmium exchange rates can be rationalized by an analogue of the inner-sphere reorganization energy. Both Ir(III) and Ir(V) are pyramidal and can form pyramidal iridium(IV) with little energetic cost in an orbitally allowed linear approach. Conversely, pyramidalization of the planar tris(imido)osmium(VI) fragment requires placing a pair of electrons in an antibonding orbital. The unique propensity of (mes)₃Ir=O to undergo intermetal oxygen atom transfer allows it to serve as an activator of dioxygen in cocatalyzed oxidations, for example, acting with osmium tetroxide to catalyze the aerobic dihydroxylation of monosubstituted olefins and selective oxidation of allyl and benzyl alcohols.

Full text of DOI:10.1021/ja065713h

Published on Web 12/23/2006
Ultrafast and Ultraslow Oxygen Atom Transfer Reactions
between Late Metal Centers
Kevin C. Fortner, David S. Laitar, John Muldoon, Lihung Pu, Sonja B. Braun-Sand,
Olaf Wiest, and Seth N. Brown*
Contribution from the Department of Chemistry and Biochemistry, 251 Nieuwland Science Hall,
UniVersity of Notre Dame, Notre Dame, Indiana 46556-5670
Received August 7, 2006; E-mail: Seth.N.Brown.114@nd.edu
Abstract: Oxotrimesityliridium(V), (mes)3IrdO (mes ) 2,4,6-trimethylphenyl), and trimesityliridium(III),
(mes)3Ir, undergo extremely rapid degenerate intermetal oxygen atom transfer at room temperature. At
low temperatures, the two complexes conproportionate to form (mes)3IrsOsIr(mes)3, the 2,6-dimethylphenyl
analogue of which has been characterized crystallographically. Variable-temperature NMR measurements
of the rate of dissociation of the µ-oxo dimer combined with measurements of the conproportionation
equilibrium by low-temperature optical spectroscopy indicate that oxygen atom exchange between iridium-
(V) and iridium(III) occurs with a rate constant, extrapolated to 20 °C, of 5 × 107 M-1 s-1. The oxotris-
(imido)osmium(VIII) complex (ArN)3OsdO (Ar ) 2,6-diisopropylphenyl) also undergoes degenerate
intermetal atom transfer to its deoxy partner, (ArN)3Os. However, despite the fact that its metal-oxygen
bond strength and reactivity toward triphenylphosphine are nearly identical to those of (mes)3IrdO, the
osmium complex (ArN)3OsdO transfers its oxygen atom 12 orders of magnitude more slowly to (ArN)3Os
than (mes)3IrdO does to (mes)3Ir (kOsOs ) 1.8 × 10-5 M-1 s-1 at 20 °C). Iridium-osmium cross-exchange
takes place at an intermediate rate, in quantitative agreement with a Marcus-type cross relation. The
enormous difference between the iridium-iridium and osmium-osmium exchange rates can be rationalized
by an analogue of the inner-sphere reorganization energy. Both Ir(III) and Ir(V) are pyramidal and can
form pyramidal iridium(IV) with little energetic cost in an orbitally allowed linear approach. Conversely,
pyramidalization of the planar tris(imido)osmium(VI) fragment requires placing a pair of electrons in an
antibonding orbital. The unique propensity of (mes)3IrdO to undergo intermetal oxygen atom transfer allows
it to serve as an activator of dioxygen in cocatalyzed oxidations, for example, acting with osmium tetroxide
to catalyze the aerobic dihydroxylation of monosubstituted olefins and selective oxidation of allyl and benzyl
alcohols.
Introduction
out even more rarely, and so there is little systematic under-
standing of the factors that affect the rates of intermetal oxygen
Oxygen atom transfer reactions (eq 1) are prototypical
examples of two-electron inner-sphere redox reactions, with
important applications in catalysis in organic synthesis1 and by
metalloenzymes.2 In general, the net transfer of an oxygen atom
requires two steps, with the formation of an intermediate where
the oxygen atom is shared between the two redox centers. The
thermodynamic requirements of such reactions have been
extensively reviewed.3 Many rate measurements of oxygen atom
transfer reactions between transition metal and main group
species have been made, although in only a few cases has it
been possible to analyze both steps.4 Kinetic studies of reactions
between two different transition metal species have been carried
atom transfer. This situation is in stark contrast to the under-
standing of outer-sphere single-electron transfer rates provided
by Marcus theory, which relates the facility of such reactions
to their thermodynamic driving force (∆G°), the degree to which
the bonding environments of the metals must be altered to
accommodate the redox change (the inner-sphere reorganization
energy, λi), and the degree to which the solvent and counterions
must reorganize to accommodate the redox change (the outer-
sphere reorganization energy, λo).5
MdO + X / MsOsX / M + OdX
(1)
(1) Wiberg, K. B., Ed. Oxidation in Organic Chemistry; Academic Press: New
York, 1985.
Our desire to better understand the factors that govern
intermetal oxygen atom transfer reactions was piqued by our
earlier observation6 that Wilkinson’s iridium(V) oxo complex
(2) (a) Spiro, T. G., Ed. Molybdenum Enzymes; Wiley: New York, 1985. (b)
Hille, R. Chem. ReV. 1996, 96, 2757-2816. (c) Holm, R. H.; Kennepohl,
P.; Solomon, E. I. Chem. ReV. 1996, 96, 2239-2314.
(3) (a) Holm, R. H. Chem. ReV. 1987, 87, 1401-1449. (b) Holm, R. H.;
Donahue, J. P. Polyhedron 1993, 12, 571-589.
(5) Lappin, A. G. Redox Mechanisms in Inorganic Chemistry. Ellis Horwood:
Chichester, 1994; Chapter 2.
(6) Jacobi, B. G.; Laitar, D. S.; Pu, L.; Wargocki, M. F.; DiPasquale, A. G.;
Fortner, K. C.; Schuck, S. M.; Brown, S. N. Inorg. Chem. 2002, 41, 4815-
4823.
(4) (a) Moyer, B. A.; Sipe, B. K.; Meyer, T. J. Inorg. Chem. 1981, 20, 1475-
1480. (b) Seymore, S. B.; Brown, S. N. Inorg. Chem. 2000, 39, 325-332.
(c) Bhattacharyya, S.; Chakraborty, I.; Dirghangi, B. K.; Chakravorty, A.
Inorg. Chem. 2001, 40, 286-293.
9
588
J. AM. CHEM. SOC. 2007, 129, 588-600
10.1021/ja065713h CCC: $37.00 © 2007 American Chemical Society

Ultrafast/Ultraslow O Transfer between Late Metal Centers
A R T I C L E S
(mes)3IrdO (mes ) 2,4,6-trimethylphenyl, mesityl)7 undergoes
degenerate intermetal oxygen atom transfer with trimesityliri-
dium(III)8 on the NMR time scale at room temperature. Here
we report a quantitative study of this exceptionally fast oxygen
atom transfer reaction. In particular, through measurements of
the self-exchange rates and the rates and equilibria in cross-
exchange reactions with the Os(VI)/Os(VIII)dO couple (ArN)3Os/
(ArN)3OsO (Ar ) 2,6-diisopropylphenyl),9 we can for the first
time apply the Marcus cross-relation to oxygen atom transfer
reactions.
NMR measurements of the CD2Cl2 solutions after reduction with
phosphine. Absorbances were corrected for temperature using the
volume contraction of the CHFCl2 with decreasing temperature (dF/dt
) -0.00242 g mL-1 K-1 and F at 293 K ) 1.3783 g mL-1).12
Three runs, with total iridium concentrations ranging from 0.51 to
1.18 mM and mole fractions of Ir(III) at room temperature ranging
from 0.30 to 0.94, were analyzed over the temperature range -40 to
-140 °C. Concentrations of the three species (mes)3Ir, (mes)3IrsOs
Ir(mes)3, and (mes)3IrdO were calculated using the measured total
iridium concentration and the measured ratio of oxygen to iridium and
the conproportionation Keq, which was calculated using ∆H° and ∆S°
as adjustable parameters. Theoretical values for the absorbance at 513
nm were calculated using these concentrations, the measured extinction
coefficients for (mes)3Ir (920 M-1 cm-1) and (mes)3IrdO (1020 M-1
cm-1), and the extinction coefficient for (mes)3IrsOsIr(mes)3 as a third
adjustable parameter. The three adjustable parameters were allowed to
vary, and the sum of [(Acalcd - Aobsd)2]/Acalcd was minimized using Solver
in Microsoft Excel.13 Optimized values for the parameters were
∆H° ) -5.14 ( 0.13 kcal/mol, ∆S° ) -15.2 ( 0.7 cal/mol K, and
ꢀ513 of (mes)3IrsOsIr(mes)3 ) 25 400 ( 300 M-1 cm-1, with estimated
errors calculated as described in the literature.14
Experimental Section
General Methods. Except as noted, all weighings were carried out
on the benchtop and all other procedures and kinetics measurements
were carried out in an inert atmosphere using standard glove box and
vacuum line techniques. Dichloromethane and CHFCl2 (Freon-21,
Aldrich) were dried over 4 Å molecular sieves, followed by CaH2.
Benzene was dried over Na. Triphenylphosphine, tri-o-tolylphosphine,
and other reagents used to synthesize the metal complexes were
commercially available and were used without further purification. The
quinuclidine adduct of osmium tetroxide, (C7H13N)OsO4, was isolated
as described by Griffith.10 The iridium oxo complex (mes)3IrdO6,7 and
the osmium complexes (ArN)3Os and (ArN)3OsdO (Ar ) 2,6-
diisopropylphenyl)9b were prepared using literature procedures.
NMR spectra were measured on a Varian VXR-300 or VXR-500
FT-NMR spectrometer. Temperatures below room temperature were
Preparation of (2,6-Me2C6H3)3IrdO. This compound has been
described briefly in the literature.7 It was prepared by the same
procedure used to prepare (mes)3IrdO,6 substituting 2,6-dimethylphe-
nylmagnesium bromide (Aldrich) for mesitylmagnesium bromide. After
purification by repeated chromatography on silica gel, eluting with 2%
Et2O/hexane, followed by crystallization from acetonitrile/water, (2,6-
1
Me2C6H3)3IrdO was isolated as dark blue crystals in 3.3% yield. H
1
calibrated using the separation between the H NMR peaks of neat
NMR (C6D6): δ 2.50 (s, 18H, CH3); 6.49 (t, 7 Hz, 3H, p-ArH), 6.95
(d, 7 Hz, 6H, m-ArH). 13C{1H} NMR (C6D6): δ 28.94 (CH3), 127.80
(para), 130.43 (meta), 132.56 (ortho), 147.58 (ipso). UV-vis (CH2-
Cl2): λmax ) 616 nm, ꢀ ) 2040 M-1 cm-1. IR (evaporated film, cm-1):
3049 (w), 2958 (w), 2922 (w), 2849 (w), 1442 (s), 1374 (w), 1234
(w), 1166 (w), 1025 (w), 844 (m), 763 (m), 667 (w). Anal. Calcd for
C24H27IrO: C, 55.04; H, 5.21. Found: C, 55.26; H, 5.11.
CH3OH with 0.03 vol % added aqueous HCl.11 Low-temperature optical
spectra were measured using an Ocean Optics spectrophotometer
equipped with a fiber optic immersion probe with a 0.2 cm path length.
Measurement of Conproportionation of (mes)3Ir and (mes)3Ird
O by Visible Spectroscopy. In a typical run, 5.0 mg of OdIr(mes)3
and 0.3 mg of P(o-tol)3 (0.13 equiv) were weighed into an NMR tube.
Dry CD2Cl2 (0.5 mL) was vacuum transferred into the tube, which
was then sealed with a Teflon valve. As the reaction proceeds, the P(o-
tol)3 is oxidized to OdP(o-tol)3, generating a mixture of OdIr(mes)3
and Ir(mes)3. When it had been determined by NMR that all of the
P(o-tol)3 had reacted, the tube was taken into a glove box, poured into
a 10 mL volumetric flask, and diluted to 10 mL with CHFCl2 precooled
to -90 °C. The solution was poured into a 25 mL two-neck flask into
which were inserted a thermocouple and a UV/vis immersion probe
with a 2 mm path length. The temperature of the solution was lowered
to -140 °C in a cold well in the glove box using liquid nitrogen as the
coolant. The solution was allowed to warm up slowly with continuous
swirling to maintain a uniform temperature throughout the solution,
and the optical spectrum from 450 to 900 nm was recorded between
-140 °C and -30 °C.
X-ray Crystallography of (2,6-Me2C6H3)3IrsOsIr(2,6-Me2C6H3)3‚
CD2Cl2. A solution of (2,6-Me2C6H3)3IrdO in CD2Cl2 was treated with
a substoichiometric amount of P(o-tol)3. Very dark brown crystals of
the µ-oxo complex grew when a sealed NMR tube containing this
solution was allowed to stand in a -20° freezer for several weeks. A
0.4 × 0.3 × 0.3 mm3 crystal was removed from the tube and placed in
inert oil and transferred to the tip of a glass fiber in the cold N2 stream
of a Bruker Apex CCD diffractometer (T ) -173 °C). Data were
reduced, correcting for absorption and decay, using the program
SADABS. The crystal was cubic, and its space group was determined
to be Pa3h from the systematic absences. The structure was solved using
Patterson methods, which located the iridium atom (on the crystal-
lographic 3h axis) and oxygen atom (on the inversion center). The single
crystallographically unique aryl group and the dichloromethane mol-
ecule (whose carbon atom was disordered about an inversion center,
with the two chlorine atoms located on the 3h axis) were located on
difference Fourier maps. Hydrogens were placed in calculated positions.
Final full-matrix least-squares refinement on F2 converged at R1 )
0.0156 for 2024 reflections with Fo > 4σ(Fo), R1 ) 0.0177 for all
2186 unique reflections (wR2 ) 0.0385, 0.0396, respectively). All
calculations used SHELXTL (Bruker Analytical X-ray Systems), with
scattering factors and anomalous dispersion terms taken from the
literature.15
The optical spectra of pure OdIr(mes)3 and Ir(mes)3 were measured
separately and used to determine the concentrations [Ir(mes)3]0 and [Od
Ir(mes)3]0 in the flask used for the equilibrium constant measurement
at -30 °C, at which temperature [(mes)3IrsOsIr(mes)3] is negligible.
This was done by fitting the entire optical spectrum from 450 to 850
nm of the solution at -30 °C to a linear combination of the spectra of
pure OdIr(mes)3 and Ir(mes)3. The Ir(III)/Ir(V) ratios determined by
1
this method agreed within a few percent with those measured by H
(7) Hay-Motherwell, R. S.; Wilkinson, G.; Hussain-Bates, B.; Hursthouse,
M. B. Polyhedron 1993, 12, 2009-2012.
Measurement of (mes)3IrsOsIr(mes)3 Dissociation by 1H NMR
Line shape Analysis. Solutions of (mes)3IrdO in CD2Cl2 were partially
(8) Hay-Motherwell, R. S.; Wilkinson, G.; Hussain-Bates, B.; Hursthouse,
M. B. J. Chem. Soc., Dalton Trans. 1992, 3477-3482.
(9) (a) Anhaus, J. T.; Kee, T. P.; Schofield, M. H.; Schrock, R. R. J. Am.
Chem. Soc. 1990, 112, 1642-1643. (b) Schofield, M. H.; Kee, T. P.;
Anhaus, J. T.; Schrock, R. R.; Johnson, K. H.; Davis, W. M. Inorg. Chem.
1991, 30, 3595-3604.
(12) Riddick, J. A.; Burger, W. B.; Sakano, T. K. Organic SolVents: Physical
Properties and Methods of Purification, 4th ed.; John Wiley & Sons: New
York, 1986; pp 490, 560.
(13) Harris, D. C. J. Chem. Educ. 1998, 75, 119-121.
(14) de Levie, R. J. Chem. Educ. 1999, 76, 1594-1598.
(15) International Tables for Crystallography; Kluwer Academic Publishers:
Dordrecht, The Netherlands, 1992; Vol. C.
(10) Cleare, M. J.; Hydes, P. C.; Griffith, W. P.; Wright, M. J. J. Chem. Soc.,
Dalton Trans. 1977, 941-944.
(11) Gordon, A. J.; Ford, R. A. The Chemist’s Companion; John Wiley &
Sons: New York, 1972; p 303.
9
J. AM. CHEM. SOC. VOL. 129, NO. 3, 2007 589

A R T I C L E S
Fortner et al
reduced by allowing the solution to stand at room temperature in a
sealed NMR tube for several days or by the addition of substoichio-
metric amounts of PPh3. The composition of the solutions was analyzed
by integration of the spectra at -110 °C (where the species give separate
resonances) or by measurement of the chemical shifts at room
temperature (where Ir(V) and Ir(III) give rise to weighted-average
resonances). 1H NMR spectra were recorded between -80° and -110
°C on a Varian VXR-500 NMR spectrometer. The line shape of the
para-methyl region was analyzed using the program gNMR16 to
generate calculated lineshapes and superimpose them on the observed
spectra, with optimization of the fit as a function of kdiss performed
manually. The spectra were generally simulated assuming that the ratio
of the concentrations of the two major iridium-containing species
((mes)3IrsOsIr(mes)3 and either (mes)3IrdO or (mes)3Ir) was as
determined in the -110 °C spectrum and the concentration of the minor
species (Ir(mes)3 if deoxygenation was <50% complete, (mes)3IrdO
if deoxygenation was >50% complete) was negligible. Achieving
satisfactory agreement of the chemical shift of the coalesced peaks in
the higher-temperature spectra did sometimes require including small
amounts of the minor species in the calculation, but never more than
5% of the total iridium concentration. Values of kdiss from different
runs varied by <15% from the mean values, with no systematic
deviations based on whether the solution was generated by autogenous
reduction or by addition of PPh3, or whether (mes)3IrdO or (mes)3Ir
was the major species exchanging with (mes)3IrsOsIr(mes)3. Activa-
tion parameters were calculated by plotting ln(kdiss/T) vs 1/T in the range
170-193 K.
Kinetics of Reduction of (ArN)3OsdO with PPh3. Solutions of
(ArN)3OsdO were prepared in the drybox in dichloromethane (room
temperature) or 1,2-dichlorobenzene (variable-temperature) in 1 cm
quartz cells fitted with septum caps. The cells were inserted into the
multicell transport block of a Beckman DU-7500 diode array spectro-
photometer. The temperature was regulated by a thermostatted water/
ethylene glycol mixture circulated through the cell block and was
measured by a thermocouple inserted in the cell block. Solutions were
allowed to equilibrate for 10 min at the desired temperature, and
reactions were initiated by injecting 20-50 equiv of PPh3 dissolved in
a small volume of solvent. Reactions were monitored by measuring
the decrease in absorbance at 610 nm. Pseudo-first-order rate constants
were obtained from the slopes of plots of ln(A - Af) vs time, which
were linear over >4 half-lives. The observed pseudo-first-order rate
constants were found to depend linearly on [PPh3] in the range 0.002-
0.010 M in CH2Cl2 at room temperature. Activation parameters were
calculated from reactions measured in 1,2-dichlorobenzene by plotting
ln(kOsP/T) vs 1/T over the range 288.6-330.6 K.
(9.6, 10.2), 715 (26.2, 27.3), 716 (60.7, 59.6), 717 (84.4, 83.4), 718
(88.9, 88.9), 719 (74.8, 75.6), 720 (100, 100.1), 721 (34.8, 35.8), 722
(6.7, 6.9).
Kinetics of Exchange Between Os(NArD)3 and OdOs(NAr)3.
Reaction mixtures were prepared by weighing in the drybox 6-15 mg
each of OdOs(NAr)3 and deuterium-enriched Os(NArD)3 (Ar ) 2,6-
diisopropylphenyl) into an NMR tube sealed to a ground glass joint.
The solids were dissolved in dry C6D6, and the tube was flame-sealed
under a vacuum. 1H NMR spectra were taken every day or two as the
mixture was allowed to react at 20 °C, and the integrals for the protons
in the para positions of the aromatic imido groups for each compound
(OdOs(NAr)3, δ 6.89 ppm, Os(NAr)3, δ 7.35 ppm) were measured.
The fraction of osmium(VI) in the mixture was determined from the
peaks for the (CH3)2CHsAr septets (OdOs(NAr)3, δ 3.48 ppm, Os-
(NAr)3, δ 3.86 ppm) and remained constant at about 0.5 throughout
the course of each reaction.
Kinetics were analyzed according to standard McKay exchange
kinetics, with kobsd determined from the slope of the plot of ln{[Os-
(VI)]/[total Os] - [unlabeled Os(VI)]/[total unlabeled Os]} vs time.18
These plots curved significantly after ∼1 half-life, apparently due to
autocatalysis. The kobsd values determined over the first half-life were
reproducible and were proportional to the total osmium concentration.
The bimolecular rate constant was averaged over those observed in
seven independent runs.
For two of these runs, the EI mass spectra of two mixtures of Od
Os(NAr)3 and deuterium-labeled Os(NAr)3 dissolved in C6D6, which
had undergone complete isotopic scrambling as determined by 1H NMR,
were acquired. The isotope envelope of the Os(NAr)3 parent ion (m/z
) 714-722) was fit using a combination of the d0, d1, d2, and d3
isotopomers using a locally written program.19
Measurement of Keq for (mes)3IrdO + Os(NAr)3 / Ir(mes)3 +
OdOs(NAr)3. OdIr(mes)3 (1.75 mg) and Os(NAr)3 (Ar ) 2,6-
diisopropylphenyl) (10.0 mg) were mixed in an NMR tube sealed to a
ground glass joint and dissolved in CD2Cl2 in the drybox. The tube
was attached to a needle valve, evacuated at low temperature, and flame-
sealed. The solution was allowed to react at room temperature for a
few minutes until the system reached equilibrium. The tube was inserted
1
into an NMR probe, and the H NMR spectrum was recorded. The
sample was cooled in the probe of the NMR spectrometer, and spectra
were recorded at temperatures from room temperature to -30 °C at 5°
intervals, allowing sufficient time at each temperature for the solution
to reach equilibrium. The ratio [OdOs(NAr)3]/[Os(NAr)3] was deter-
mined at each temperature by measuring the relative integrals of the
isopropyl methine resonances. The ratio [Ir(mes)3]/[OdIr(mes)3] was
determined from the chemical shift of the aromatic H compared to its
position in the spectra of authentic samples of (mes)3Ir and (mes)3Ird
O at the same temperature. (Formation of (mes)3IrsOsIr(mes)3 is
negligible at these temperatures and concentrations.)
Kinetics of (mes)3IrdO + Os(NAr)3 f Ir(mes)3 + OdOs(NAr)3.
Freshly prepared solutions of OdIr(mes)3 and Os(NAr)3 in dry,
deoxygenated CD2Cl2 were mixed under nitrogen in an NMR tube
sealed to a ground glass joint and cooled to -78 °C. The NMR tube
was evacuated and flame-sealed, and the solution was kept at -78 °C
until it was inserted in the NMR probe, which was precooled to the
desired temperature (-30 to -70 °C). The reaction was monitored by
measuring the change in the chemical shift of the mesityl aromatic
and ortho-CH3 resonances over time. Since chemical exchange between
OdIr(mes)3 and Ir(mes)3 is rapid on the NMR time scale at these
temperatures, a single set of mesityl resonances is seen for mixtures of
OdIr(mes)3, Ir(mes)3, and (mes)3IrsOsIr(mes)3, and the peak positions
shift upfield as the iridium is deoxygenated. In order to avoid
precipitation of the sparingly soluble µ-oxo complex, the reactions were
conducted using dilute solutions of iridium (∼ 1.5 mg/0.6 mL) and
Preparation of (ArDN)3Os (ArD ) 4-Deuterio-2,6-diisopropyl-
phenyl). A sample consisting of 10 g of 2,6-diisopropylaniline (Acros)
was washed twice with 1 mL of D2O (Cambridge Isotope Laboratories).
The aniline was then placed into a 500 mL glass bomb with 100 g of
CH3OD (Aldrich) and 8 mL of concentrated DCl in D2O (Aldrich).
The bomb was sealed, and the mixture was heated for 5 d at 60 °C.
The solvent was removed in vacuo and the solid was dissolved in water
and rendered basic with sodium carbonate. The solution was extracted
three times with ether, and the organic phases were collected. The ether
solution was dried with magnesium sulfate and concentrated in vacuo
to give 2,6-diisopropylaniline enriched with deuterium at the para
position. Using the deuterium-labeled aniline as a starting material,
labeled ArDNCO17 and Os(NArD)3 were prepared using literature
9b
procedures for the corresponding protio compounds. Mass spectral
analysis of the parent ion envelope Os(NArD)3 was consistent with 82%
1
deuteration of the aryl ligands; H NMR indicated that only the para
position of the aryl groups contained deuterium. EI-MS (parent ion
only; observed, calculated relative intensities in parentheses): m/z 714
(16) Budzelaar, P. H. M. gNMR, v. 3.5.6.; Cherwell Scientific Publishing: 1996.
(17) Knoelker, H.; Braxmeier, T.; Schlechtingen, G. Angew. Chem., Int. Ed.
Engl. 1995, 34, 2497-2500.
(18) Espenson. J. H. Chemical Kinetics and Reaction Mechanisms; McGraw-
Hill Inc.: New York, 1995.
(19) Brown, S. N.; Mayer, J. M. Inorg. Chem. 1992, 31, 4091-4100.
9
590 J. AM. CHEM. SOC. VOL. 129, NO. 3, 2007

Ultrafast/Ultraslow O Transfer between Late Metal Centers
A R T I C L E S
were only monitored for the first 10-20% of reaction. Over this range
of reaction, the chemical shift changes are linear with time. Second-
order rate constants for this initial portion of the reaction were calculated
from the slope of these curves, ∂(δ)/∂t, according to eq 2:
O (3.1 mg, 5.5 µmol) and 1-octene (4.3 µL, 0.027 mmol). After shaking,
0.5 mL of the (DHQD)2PHAL/osmium solution was added to the NMR
tube. Upon completion of the reaction (determined by NMR), the
reaction mixture was diluted with Et2O (8 mL), washed with saturated
aqueous sodium sulfite to remove the metals, and dried over MgSO4.
After the solvent was removed on the rotary evaporator, the diol was
filtered through a small plug of silica gel using EtOAc/hexane (1:1) to
give pure 1,2-octanediol. The two enantiomers were separated by chiral
GC (J&W Scientific 112-2532, Cyclodex B; injection temperature 250
°C, detector temperature 300 °C, oven temperature 75 °C for 10 min
followed by 2 °C/min to 145 °C for 5 min; retention times ) 42.6 and
42.9 min).
Representative Procedure for the Oxidation of Benzylic and
Allylic Alcohols: Oxidation of Benzyl Alcohol. Into a 2.5 mL screw-
capped vial was added a solution of (mes)3IrdO (4.5 mg, 7.9 µmol, 1
mol %) and benzyl alcohol (82 µL, 0.79 mmol) in 0.4 mL of THF. In
an NMR tube, the osmium reagent was generated by adding allyl ethyl
ether (12.0 µL, 0.106 mmol, 13.2 mol %) to a suspension of
(quinuclidine)OsO4 (5.8 mg, 0.016 mmol, 2 mol %) in 0.1 mL of THF.
After allowing this osmium-containing solution to stand for 40 s after
addition of the alkene, the solution containing the substrate and iridium
was added to the NMR tube. Oxygen gas, presaturated with THF by
passing it through a gas dispersion tube in a 100 mL two-neck flask
filled with THF, was admitted to the screw-cap NMR tube with a
septum cap using a needle. Oxygen was bubbled through the reaction
mixture at a moderate rate (∼3 bubbles/s) and vented through a mineral
oil bubbler. The reaction was complete in 3 days.
∂(δ)/∂t
[Os]
1 + K[Ir]
kIrOs
)
‚
(2)
δIr(III) + 2K[Ir]δIrOIr - (1 + K[Ir])δIrdO
where K is the equilibrium constant for the conproportionation of
(mes)3Ir and (mes)3IrdO calculated from the thermodynamic parameters
measured by optical spectroscopy in CHFCl2 and δIr(III), δIrOIr, δIrdO
are the chemical shifts of the resonances for pure (mes)3Ir, (mes)3Irs
OsIr(mes)3, and (mes)3IrdO, respectively, measured at (or in the case
of (mes)3IrsOsIr(mes)3, extrapolated to) the temperature of the kinetics
run. Concentrations of iridium and osmium were corrected for thermal
contraction of the solvent.12 Since the chemical shifts for the µ-oxo
complex can only be measured over a very narrow range of temperatures
(-100 to -110 °C), this extrapolation is likely to be rather imprecise.
However, because the chemical shifts of the µ-oxo complex are close
to the average of the shifts of the iridium(III) and iridium(V) complexes,
and because the mole fraction of iridium(IV) is small at these
concentrations and temperatures, the calculated rate constants are
relatively insensitive to the treatment of the µ-oxo complex, with
calculated rate constants differing by no more than 10% from those
where the formation of the µ-oxo complex is completely neglected.
Results from the shifts of the aromatic hydrogens agreed with those
measured from the o-CH3 resonances, and second-order rate constants
calculated from eq 2 were independent of [Os(NAr3)] in the range
0.007-0.016 M. Activation parameters were obtained from a plot of
ln(k/T) vs 1/T.
Results
Conproportionation of (mes)3IrdO and (mes)3Ir. Earlier
we reported that solutions containing mixtures of Wilkinson’s
homoleptic trimesityliridium(III), (mes)3Ir (mes ) 2,4,6-tri-
methylphenyl),8 and oxotrimesityliridium(V), (mes)3IrdO,7
display only a single set of mesityl resonances in the 1H NMR
at room temperature.6 This is due to fast intermolecular oxygen
atom transfer between iridium(V) and iridium(III) leading to
chemical exchange of the two types of mesityl group (eq 3).
Given the differences in chemical shift between (mes)3IrdO
and (mes)3Ir and the fact that only fairly sharp coalesced signals
are seen at room temperature, this exchange must take place
DFT Calculations on Iridium and Osmium Complexes. Calcula-
tions were performed on model complexes with the mesityl groups in
the iridium complexes and the 2,6-iPr2C6H3 groups in the osmium
complexes replaced by methyl groups. All structures were optimized
using the hybrid B3LYP method, which has been successfully used in
a number of studies involving transition metal complexes.20
A
6-311+G** basis set was used for the carbon, hydrogen, and oxygen
atoms, while an SDD basis set was used on the iridium and osmium
atoms. Stationary points were characterized by harmonic frequency
analysis. Thermochemical corrections to the reported energies were
calculated from the harmonic frequency analysis for a temperature of
293 K and a pressure of 1 atm, using a scaling factor of 0.9614.21 All
calculations were performed using the G9822 and G0323 series of
programs. The Kohn-Sham orbital24 representations were created in
Molden using a contour value of 0.05.
with kIrIr > 4 ×105 M-1 s-1
.
(mes)3IrdO + Ir(mes)3 / (mes)3Ir + OdIr(mes)3 (3)
Aerobic Dihydroxylations with (mes)3IrdO/(C7H13N)OsO4. Rep-
resentative Procedure: Dihydroxylation of 1-Octene. Into a screw-
capped NMR tube in the air were added (C7H13N)OsO4 (2.0 mg, 5.5
µmol, 20 mol %), (mes)3IrdO (3.1 mg, 5.5 µmol, 20 mol %), 1-octene
(4.3 µL, 0.027 mmol), and dimethyl terephthalate as an internal standard
(1.0 mg, 5.2 µmol) and 0.5 mL of THF-d8/2% D2O. The NMR tube
was rapidly shaken. The reaction was monitored by NMR and was
complete in a week.
Enantioselective Aerobic Dihydroxylation of 1-Octene with
(mes)3IrdO/OsO4/(DHQD)2PHAL. Into a vial were added OsO4 (2.5
mg, 9.8 µmol), (DHQD)2PHAL (Aldrich, 27 mg, 0.035 mmol), and 1
mL of THF-d8. Into a screw-capped NMR tube were added (mes)3Ird
A logical intermediate in this oxygen atom transfer reaction
is the iridium(IV) µ-oxo complex (mes)3IrsOsIr(mes)3. Results
of the earlier study6 suggest that this conproportionation product
is not present in significant concentrations at ambient temper-
atures. In particular, (mes)3IrdO is deoxygenated by oxygen
atom acceptors such as triphenylphosphine or triphenylarsine
in reactions which show clean pseudo-first-order kinetics (in
the presence of excess reducing agent) over the entire course
of the reaction. Substantial amounts of conproportionation would
imply that the reaction would slow as (mes)3Ir bound to the
oxo complex (mes)3IrdO, except in the unlikely event that the
µ-oxo complex were more readily deoxygenated than the
terminal oxo complex.
Conproportionation is important at lower temperatures,
however, as indicated by color changes when mixtures of
(mes)3IrdO and (mes)3Ir are cooled. Solutions of neither blue-
green (mes)3IrdO nor yellow-brown (mes)3Ir are thermochro-
mic, and optical spectra taken at room temperature of green
solutions containing mixtures of the two compounds are simple
(20) (a) Koch, W.; Holthausen, M. C. A Chemist’s Guide to Density Functional
Theory; Wiley-VCH: Weinheim, 2000. (b) Cundari, T. R. Computational
Organometallic Chemistry; Marcel-Dekker: New York, 2001. (c) Niu, S.;
Hall, M. B. Chem. ReV. 2000, 100, 353-405.
(21) Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502-16513.
(22) Frisch, M. J., et al. Gaussian 98; Gaussian, Inc.: Wallingford, CT, 1998.
(23) Frisch, M. J., et al. Gaussian 03, revision C.01; Gaussian Inc.: Wallingford,
CT, 2004.
(24) For a discussion of the relationship of Kohn-Sham and Hartree-Fock
orbitals, see: Stowasser, R.; Hoffmann, R. J. Am. Chem. Soc. 1999, 121,
3414-3420.
9
J. AM. CHEM. SOC. VOL. 129, NO. 3, 2007 591

A R T I C L E S
Fortner et al
Table 2. Selected Bond Lengths (Å) and Angles (deg) for
(2,6-Me2C6H3)3IrsOsIr(C6H3-2,6-Me2)3‚CD2Cl2, Compared with
Corresponding Parameters in (mes)3Ira and (mes)3IrdOb
Ar3Ir
sOsIrAr3
(mes)3Ira
Ar ) 2,6-Me2C6H3
(mes)3Ird
Ob
IrsC, Å
2.00[4] av
2.01
2.007(2)
2.04
2.015[19] av
2.05
IrsO, Å
n/a
1.9142(1)
1.725(9)
1.92
1.74
CsIrsC, deg
CsIrsO, deg
IrsOsIr, deg
107.5[17] av
106.5
n/a
106.43(5)
94.6
112.36(4)
121.5
180.0
109[6] av
101.2
110[6] av
116.7
n/a
n/a
170.5
a Reference 8. b Reference 7. c Numbers in parentheses represent esti-
mated standard deviations in the last digits of single crystallographically
measured values; values in brackets represent standard deviations in the
average of the three independent measurements in the molecule. Values in
italics are the results of DFT calculations on the (CH3)3Ir analogues of the
triaryliridium complexes.
Figure 1. SHELXTL plot (50% ellipsoids) of (2,6-Me2C6H3)3Ir-O-Ir-
(C6H3-2,6-Me2)3.
Table 1. Crystallographic Details for
(2,6-Me2C6H3)3Ir-O-Ir(C6H3-2,6-Me2)3‚CD2Cl2
empirical formula
FW
C49H54D2Cl2Ir2O
1118.31
temperature (K)
λ
100(2)
0.710 73 Å (Mo KR)
cubic
Pa3h
51 216
2186
16.2538(3)
crystal class
space group
total data collected
no. of indep reflns.
a (Å)
V (Å3)
4294.03(14)
Z
4
Figure 2. Optical spectra of a mixture of (mes)3Ir and (mes)3IrdO as a
function of temperature (20:1 CHFCl2/CD2Cl2, total [Ir] ) 1.2 × 10-3 M,
0.2 mm path length). Temperatures are indicated on the left. Inset:
Absorbance at 513 nm as a function of temperature. The solid line is the
fit calculated on the basis of the thermodynamic parameters for conpro-
portionation given in the text.
calcd F (g/cm3)
1.730
cryst size (mm3)
0.4 × 0.3 × 0.3
µ (mm-1
)
6.352
no. of refined parameters
R indices [I > 2σ(I)]a
R indices (all data)a
GOF on F2
104
R1 ) 0.0156, wR2 ) 0.0385
R1 ) 0.0177, wR2 ) 0.0396
1.082
aryl groups surrounding each iridium have opposing chirality.
A linear µ-oxo bridge was also seen in the isoelectronic
neopentylrhodium(IV) complex (Me3CCH2)3RhsOsRh(CH2-
CMe3)3.25 The geometry of the triaryliridium fragment (IrsC
distance of 2.007(2) Å, CsIrsC angles of 106.43(5)°) is
remarkably similar to what is found in both trimesityliridium-
(III) (2.00 Å av, 107.5° av)8 and oxotrimesityliridium(V) (2.015
Å av, 108.9° av)7 (Table 2).
The change of color from green to brown on cooling solutions
of partially reduced (mes)3IrdO can be analyzed quantitatively
by low-temperature optical spectroscopy. As the temperature
of solutions of (mes)3IrdO/(mes)3Ir mixtures are lowered below
about -60 °C, the absorbance of the solution increases at all
wavelengths longer than 450 nm (Figure 2). The µ-oxo complex
has a particularly strong absorption at 513 nm (ꢀ ) 25 400 M-1
cm-1) and also shows two other intense peaks at 560 and 632
nm, as well as shoulders at longer wavelengths (675, 736 nm).
Similar behavior is observed in CH2Cl2 or in mixtures of
2
2
2
a R1 ) Σ||Fo| - |Fc||/Σ|Fo|; wR2 ) (Σ[w(Fo - Fc )2]/Σw(Fo )2)1/2
.
superpositions of the optical spectra of the two constituents.
However, when dichloromethane solutions are cooled to well
below room temperature, a color change from green to dark
brown takes place. If the solutions are allowed to stand at low
temperature, a dark brown microcrystalline precipitate forms.
Both the precipitation and the color change are reversed on
warming the sample.
By allowing a partially reduced sample of the 2,6-dimethyl-
phenyl analogue (2,6-Me2C6H3)3IrdO7 to stand at -20 °C,
we were able to grow large single crystals of the sparingly
soluble dark brown compound, which was confirmed by X-ray
crystallography to be the µ-oxo complex (Me2C6H3)3IrsOsIr-
(C6H3Me2)3 (Figure 1). The compound crystallizes in the cubic
space group Pa3h (Table 1), with the complex sitting on a
crystallographic threefold axis with the bridging oxygen atom
on the inversion center. The molecule therefore has crystallo-
graphically required S6 symmetry. In particular, the µ-oxo bridge
is strictly linear, and the three-bladed propellers formed by the
(25) Hay-Motherwell, R. S.; Wilkinson, G.; Hussain-Bates, B.; Hursthouse,
M. B. Polyhedron 1990, 9, 2071-2080.
9
592 J. AM. CHEM. SOC. VOL. 129, NO. 3, 2007

Ultrafast/Ultraslow O Transfer between Late Metal Centers
A R T I C L E S
dichloromethane and CHFCl2 (Freon-21), although the Freon
mixture is preferred for quantitative studies because it remains
fluid at the extremely low temperatures (∼140 K) needed to
shift the equilibrium strongly toward the µ-oxo complex at the
submillimolar concentrations required for optical spectroscopy.
The rise in absorbance at 513 nm as the temperature is lowered
is directly proportional to the concentration of the limiting
reagent of eq 4, consistent with a 1:1 stoichiometry of (mes)3Ird
O and (mes)3Ir. Runs conducted at different iridium concentra-
tions give consistent values for the Keq of eq 4, and analysis of
the temperature-dependence of the absorbance from 135 to 220
K (inset, Figure 2) allows the determination of the thermody-
namic parameters ∆H°4 ) -5.14 ( 0.13 kcal/mol and ∆S°4 )
-15.2 ( 0.7 cal/mol‚K.
(mes)3IrdO + Ir(mes)3 / (mes)3IrsOsIr(mes)3 (4)
Kinetics of Dissociation of (mes)3IrsOsIr(mes)3. 1H NMR
spectra of equilibrium mixtures of (mes)3IrdO, (mes)3IrsOs
Ir(mes)3, and (mes)3Ir recorded above about -80 °C show only
a single set of mesityl resonances due to the rapid exchange of
mesityl environments among the three species. Both aromatic
hydrogens appear as a single resonance, as do both ortho-methyl
groups. However, these resonances do show some broadening
as the temperature is lowered to -80 °C, due to hindered IrsC
bond rotation in the three species. The iridium(III) complex was
previously reported to exhibit a sizable barrier to IrsC bond
rotation (∆G‡ ) 13.0 kcal/mol, extrapolated to 163 K)6,8 and
shows sharp, separate signals for the aromatic and o-methyl
peaks even at -20 °C. In contrast, the oxo complex (mes)3Ird
O shows a much faster IrsC bond rotation. Even at -110 °C,
the aromatic resonance, though slightly broadened, is still
coalesced. However, at this low temperature the o-CH3 reso-
nances of (mes)3IrdO have decoalesced sufficiently to determine
their chemical shifts (Figure S1), and this allows line shape
analysis over the range -110 to -80 °C, giving thermodynamic
parameters for IrsC bond rotation in (mes)3IrdO (Figure S2)
of ∆H‡ ) 5.73 ( 0.04 kcal/mol, ∆S‡ ) -9.80 ( 0.24 cal/mol‚
K, and ∆G‡ (163 K) ) 7.33 kcal/mol.
At temperatures above about -80 °C, the para-methyl
resonance, which is insensitive to bond rotation but does change
chemical shift with iridium oxidation state, remains a single
sharp resonance in Ir(V)/Ir(IV)/Ir(III) mixtures. At temperatures
below -80 °C, this resonance broadens and decoalesces as the
rate of intermolecular exchange slows (Figure 3). The aromatic
and ortho-methyl resonances likewise broaden and separate,
although the detailed lineshapes of these resonances are
complicated by the effects of intramolecular exchange due to
IrsC bond rotation. The low-temperature limiting spectra show
only the µ-oxo complex (mes)3IrsOsIr(mes)3 and one of the
monomeric iridium species, (mes)3Ir or (mes)3IrdO, depending
on whether more or less than half of the oxygen has been
removed. At these concentrations and this low temperature, the
position of the equilibrium strongly favors conproportionation.
Note that the aromatic and ortho-methyl resonances of (mes)3Irs
OsIr(mes)3 are well-separated, but still slightly broadened, at
-110°, indicating that the barrier to IrsC bond rotation in the
µ-oxodiiridium(IV) complex (krot ) 40 s-1, ∆G‡ ) 8.1 kcal/
mol at 163 K) is intermediate between that of Ir(III) and Ir(V).
This correlation of rotation barrier with the electronic unsat-
uration at iridium, rather than with steric hindrance, substantiates
Figure 3. Variable-temperature 500-MHz 1H NMR spectra of a (mes)3-
IrsOsIr(mes)3/(mes)3IrdO mixture in CD2Cl2 over the range -70 to -110
°C. The ortho-CH3 groups of (mes)3IrdO at -110 °C appear as a broad,
partially decoalesced peak from 1.8-2.6 ppm (see Figure S1).
our earlier contention that agostic interactions between iridium
and the ortho-methyl groups are significant in (mes)3Ir.6
The line shape of the para-methyl peaks can be analyzed in
the temperature range -80 to -102.5 °C to determine the rate
of dissociation of (mes)3IrsOsIr(mes)3 to give (mes)3IrdO and
(mes)3Ir (Figure 3; note that at these concentrations and
temperatures conproportionation is highly favorable, so only one
of (mes)3IrdO or (mes)3Ir will be present in appreciable
concentration, depending on whether the average oxidation state
is greater than or less than +4). The rate constants kdiss obtained
by line shape analysis are independent within experimental error
of whether the observed exchange is between Ir(IV) and Ir(V)
or between Ir(IV) and Ir(III), and are independent of whether
the mixture of oxidation states is generated by the slow
spontaneous reduction of (mes)3IrdO in CD2Cl2 or by deliberate
addition of substoichiometric PPh3 (a process that has been
shown to generate (mes)3Ir cleanly6), substantiating a mechanism
involving simple, unassisted dissociation of the µ-oxo dimer
without participation of other species such as OPPh3 (if present).
An Eyring plot (Figure S3) gives the activation parameters for
the dissociation of the µ-oxo complex as ∆H‡ ) 10.04 (
diss
0.14 kcal/mol, ∆S‡diss ) +8.6 ( 0.8 cal/mol‚K. Combining these
activation parameters with the thermodynamic parameters for
the net reaction of eq 4 allows one to compute activation
parameters for the forward direction of eq 4 (neglecting the
difference between the 95% CHFCl2/5% CD2Cl2 used for
9
J. AM. CHEM. SOC. VOL. 129, NO. 3, 2007 593

A R T I C L E S
Fortner et al
Scheme 1. Preparation of Labeled Os(NArD)3
equilibrium measurements and the CD2Cl2 used for kinetics):
∆H‡IrIr ) 4.90 ( 0.19 kcal/mol, ∆S‡IrIr ) -6.6 ( 1.1 cal/mol‚
K, ∆G‡IrIr (293 K) ) 6.83 ( 0.36 kcal/mol. Using these data to
extrapolate the rate constant for bimolecular degenerate atom
transfer to 293 K, where conproportionation is the rate-
determining step in complete oxygen atom transfer, gives
kIrIr ) 5 × 107 M-1 s-1
.
Oxygen Atom Cross-Exchange Reactivity of (ArN)3Osd
O. In order to better understand the features that make (mes)3Ird
O so kinetically reactive toward oxygen atom self-exchange,
we sought a system that might share a number of the unusual
features of the (mes)3IrdO/(mes)3Ir redox couple. The tris-
(imido) osmium complexes (ArN)3OsdO and (ArN)3Os (Ar )
2,6-diisopropylphenyl), described by Schrock and co-workers,9
appeared to be an ideal comparison: Both pairs of complexes
involve low-coordinate, late transition metal oxo complexes,
and in both instances the reduced species are stable as three-
coordinate monomers which do not readily bind additional
ligands.
The osmium oxo complex (ArN)3OsdO also appears to be
similar to (mes)3IrdO in its reactivity toward main group
oxygen atom acceptors. The osmium oxo complex undergoes
clean reduction by triphenylphosphine to form OPPh3 and
(ArN)3Os in a reaction that is first-order in both reagents (eq
5). The activation parameters for this reduction (289-331 K,
1,2-dichlorobenzene, Figure S4), ∆H‡ ) 7.7 ( 0.3 kcal/mol,
and ∆S‡ ) -30.3 ( 1.1 cal/mol‚K, show the large negative
activation entropy typical of bimolecular oxygen atom transfer
reactions.4b At room temperature, the rate constant kOsP is 1.30-
(7) M-1 s-1 (25 °C, CH2Cl2), only three times smaller than the
corresponding rate constant for iridium6 (kIrP ) 3.95 M-1 s-1).
equilibrium mixture of (mes)3IrsOsIr(mes)3 and (mes)3Ir. The
measured initial rates of reaction are proportional to both [Ir]
and [Os], and the variation of the observed second-order rate
constants kIrOs with temperature (Figure S6) allows one to
calculate activation parameters for the cross-exchange reaction
of eq 6 (in the forward direction): ∆H‡IrOs ) 11.3 ( 0.5 kcal/
mol and ∆S‡
) -15.3 ( 2.4 cal/mol‚K.
IrOs
Oxygen Atom Self-Exchange of (ArN)3Os/(ArN)3OsdO.
Degenerate intermetal oxygen atom transfer between (ArN)3Os
and (ArN)3OsdO is slow, with mixtures of the two osmium
complexes showing sharp, separate, unperturbed NMR reso-
nances for both complexes even at +100 °C in toluene-d8. In
order to measure exchange between the two species on the
chemical time scale, we prepared the tris(imido)osmium(VI)
complex labeled in the para position with deuterium, (ArDN)3-
Os (Scheme 1).
kOsP
Upon mixing Os(NArD)3 with unlabeled (ArN)3OsdO, growth
of the para resonance of the osmium(VI) complex in the H
1
(ArN)3OsdO + PPh3
8 (ArN)3Os + OPPh3
(5)
NMR is exceedingly slow, requiring weeks to achieve equilib-
rium at room temperature. Kinetics of isotope exchange were
measured in benzene, since (ArN)3OsdO decomposes in
dichloromethane at a significant rate on the extremely slow time
scale of the reaction. The reactions exhibited a moderate degree
of autocatalysis, with the apparent rate constants increasing after
the first half-life or so. The origin of the autocatalysis is
unknown, but it is not due to production of trace amounts of
free aniline, since spiking the mixture with 2,6-diisopropylaniline
does not accelerate exchange. However, reaction rates measured
in the first half-life or so are reproducible, and redistribution of
the deuterium conforms to the pseudo-first-order decay expected
for exchange kinetics, regardless of the rate law.18 The fact that
the observed rate constant is proportional to the total osmium
concentration (Figure S7) indicates that the reaction is second-
order, and from seven runs at 20 °C, the average rate constant
Even more remarkably, the osmium and iridium complexes
have nearly equivalent thermodynamic propensities toward
oxygen atom transfer. Within minutes at room temperature,
intermetal oxygen atom transfer takes place between (mes)3Ird
O and (ArN)3Os to give an equilibrium mixture of both oxo
complexes and both reduced species (eq 6), with KIrOs ) 0.6 in
CD2Cl2 at 20 °C. The position of equilibrium is only slightly
temperature-dependent, with a van’t Hoff plot between 243 and
288 K giving ∆H°IrOs ) -0.86 ( 0.03 kcal/mol and ∆S°IrOs
)
-3.89 ( 0.10 cal/mol‚K (Figure S5). In this temperature range
the formation of the iridium(IV) µ-oxo complex is negligble,
and there is no evidence by NMR for any interaction between
(ArN)3OsdO and (ArN)3Os at any temperature. Since the
reaction shown in eq 6 corresponds to breaking an IrdO bond
and forming an OsdO bond, the small ∆H° indicates that the
bond dissociation enthalpies of the two bonds are nearly equal,
with the OsdO bond having the greater BDE by 0.86 kcal/
mol.
kOsOs is (1.8 ( 0.2) × 10-5 M-1 s-1
.
The NMR analysis of isotope exchange measures the rate at
which the imido groups exchange environments between
osmium(VI) and osmium(VIII) but cannot distinguish between
oxygen atom transfer between (ArN)3OsdO and (ArN)3Os or
direct imido group exchange between osmium centers. These
two processes can be distinguished by mass spectrometry, since
oxygen atom exchange retains the integrity of a tris(imido)-
osmium unit, while imido exchange severs the correlation any
individual imido group once had with its fellows in the Os-
(NAr)3 fragment, and must therefore result in a completely
(mes)3IrdO + (ArN)3Os / (mes)3Ir + (ArN)3OsdO
(6)
While the rate of cross-exchange (eq 6) is rapid at room
temperature, rates can be measured between -70 and -30 °C
by observing the growth of the resonances of (ArN)3OsdO as
well as the changes in the positions of the mesityl resonances,
which shift upfield as (mes)3IrdO is deoxygenated to an
9
594 J. AM. CHEM. SOC. VOL. 129, NO. 3, 2007

Ultrafast/Ultraslow O Transfer between Late Metal Centers
A R T I C L E S
Figure 4. Mass spectrometric analysis of the isotopic composition of Os(NAr)3 after equilibration of deuterium-enriched Os(NArD)3 and unlabeled Od
Os(NAr)3 (middle right). Isotope patterns of the parent ions of the 82% deuterated Os(NArD)3 and natural-abundance Os(NAr)3 are shown on the lower left.
Isotope patterns calculated for the equilibrium mixture given the measured overall composition and assuming either an oxygen atom transfer mechanism
(upper right) or an individual imido group transfer mechanism (lower right) are shown for comparison with the observed pattern.
statistical distribution of imido groups (Figure 4). The experi-
mental data are highly nonstatistical and clearly exclude any
significant amount of exchange of individual imido groups.
Agreement with the groupwise exchange model is excellent,
consistent with assignment of the observed exchange to an
oxygen atom transfer process.
of catalytic (C7H13N)OsO4 and (mes)3IrdO to give 1,2-diols
(eq 7). Diols are the sole product of oxidation of aliphatic
alkenes, and functional group compatibility is good. Arylalkenes
give lower yields of diols (75% for styrene to 1,2-phenyle-
thanediol), presumably because of overoxidation of the benzylic
alcohol in the product (see below).
Application of Intermetal Oxygen Atom Transfer to
Cocatalyzed Air Oxidations. Oxotrimesityliridium(V) is inert
to all common organic functional groups, reacting only with
very easily oxidized substrates such as phosphines and arsines.6
Thus, despite the fact that (mes)3Ir activates O2 readily to form
the oxo compound, its use as a catalyst in air oxidations is
limited. However, the great kinetic facility of (mes)3IrdO in
intermetal oxygen atom transfer reactions suggests that it could
be used in tandem with other oxometal reagents to allow greater
scope for catalyzed air oxidations.
The goal of cocatalyzed air oxidation has been achieved using
quinuclidine-ligated OsO4 as a cocatalyst with (mes)3IrdO in
two different reactions. First, this reagent combination mediates
the well-known dihydroxylation of olefins26 using air and water
as stoichiometric reagents. Air has been used as the terminal
oxidant in dihydroxylations under organic/basic aqueous bi-
phasic conditions,27 but no turnover is seen in organic solvents.
However, in the presence of catalytic (mes)3IrdO, a variety of
monosubstituted alkenes react with air and water in the presence
Unfortunately, the rate of dihydroxylation is very low (5
turnovers/week), and the reaction is limited to monosubstituted
alkenes. More highly substituted alkenes react readily with
osmium to form glycolates, but these glycolates do not react
with iridium at appreciable rates. Monitoring the reaction over
time (Figure S8) indicates that the reaction slows down
considerably as it progresses. This is not due to catalyst
degradation, as the (mes)3IrdO and the osmium glycolate
concentrations, as monitored by 1H NMR, are unchanged over
the course of the reaction. Instead the slowdown is accounted
for, quantitatively, by inhibition by the diol product.
1-Octene can be dihydroxylated enantioselectively with
catalytic OsO4 and (mes)3IrdO using (DHQD)2PHAL.28 While
the reaction occurs with a modest 45 ((2)% ee, this is identical
(26) Schro¨der, M. Chem. ReV. 1980, 80, 187-213.
(27) (a) Do¨bler, C.; Mehltretter, G.; Beller, M. Angew. Chem., Int. Ed. 1999,
38, 3026-3028. (b) Do¨bler, C.; Mehltretter, G. M.; Sundermeier, U.; Beller,
M. J. Am. Chem. Soc. 2000, 122, 10289-10297. (c) Do¨bler, C.; Mehltretter,
G. M.; Sundermeier, U.; Beller, M. J. Organomet. Chem. 2001, 621,
70-76.
(28) (a) Kolb, H. C.; Van, Nieuwenhze, M. S.; Sharpless, K. B. Chem. ReV.
1994, 94, 2483-2547. (b) Johnson, R. A.; Sharpless, B. K. In Catalytic
Asymmetric Synthesis, 2nd ed.; Ojima, I., Ed.; Wiley-VCH: New York,
2000; Chapter 6.
9
J. AM. CHEM. SOC. VOL. 129, NO. 3, 2007 595

A R T I C L E S
Fortner et al
to that shown by stoichiometric OsO4/(DHQD)2PHAL under
the same conditions. The lower enantioselectivity than achieved
under the optimized Sharpless conditions is attributed to the
While these are not classical oxygen atom transfer reactions,
since they involve only net one-electron transfer, they are related
in that they involve electron transfer through an oxygen bridge.
Rates range from 0.1 M-1 s-1 (Cr) to 240 M-1 s-1 (Ti).
Given these moderate rates, the extreme facility of oxygen
atom transfer between (mes)3IrdO and (mes)3Ir, with kIrIr ) 5
× 107 M-1 s-1 at 20 °C, is remarkable. Although the dearth of
quantitative measurements makes comparison difficult, we know
of no other examples where a metal oxo complex and its deoxy
form are in rapid exchange by NMR at any temperature; (mes)3Ir
and (mes)3IrdO are in rapid exchange by NMR even at -80
°C. Even highly exothermic oxygen atom transfer reactions, such
as the ∼70-kcal/mol downhill34 oxidation of PPh3 by [Ru(bpy)2-
(py)(O)]2+ (kRuP ) 1.75 × 105 M-1 s-1),4a proceed at slower
rates than this degenerate intermetal atom transfer. While there
are examples of electron transfer and one-electron atom transfer
reactions35 between transition metal centers that proceed at
comparable (or faster) rates, the only two-electron atom transfer
self-exchange reaction that approaches this rate is the degenerate
transfer of I+ between [Cp2RuI]+ and Cp2Ru (k ≈ 7 × 105 M-1
s-1, 20 °C, CD3CN),36 and the rapid rate in this case is
contingent on the heavy halogen, with the chlorine analogue
t
solvent (THF compared to BuOH/H2O). The equivalence of
the selectivity under stoichiometric and catalytic conditions
demonstrates that the (mes)3IrdO does not participate directly
in the osmylation of the olefin and acts solely as a reoxidant
for the osmium.
Oxotrimesityliridium(V) can also cocatalyze the room-tem-
perature air oxidation of benzylic and allylic alcohols to the
corresponding aldehydes and ketones. Here the cooxidant is
(C7H13N)OsO4 prereduced with a sacrificial alkene such as vinyl
acetate or allyl ethyl ether, as was previously described using
copper salts as the oxygen-activating component.29 Because
alcohol oxidation is not inhibited by the product carbonyl
compounds in the way that alkene oxidation is inhibited by the
product diols, much higher turnover numbers can be achieved.
Thus, complete oxidation of a variety of allyl and benzyl
alcohols at atmospheric pressure is routinely achieved in 3 days
in THF using 2 mol % of alkene-activated (C7H13N)OsO4 and
1 mol % of OdIr(mes)3. The scope (primary and secondary
benzylic and allylic alcohols, toleration of other oxidizable
functional groups such as alkenes), selectivity (no overoxidation
to the carboxylic acid, no oxidation of aliphatic alcohols), and
limitations (1,2- and 1,3-diols prevent reaction) are the same
as those discussed previously for the osmium/copper-catalyzed
oxidation.29 Rates are generally somewhat slower than those
for the copper-catalyzed process, which requires only about 18
h to go to completion using similar catalyst loadings.
undergoing atom transfer at only 16.5 M-1 s-1 37
.
One obvious reason for the rapidity of the oxygen atom
transfer process compared to other examples in the literature is
that intrinsic rates are often obscured by rearrangements of the
ancillary ligands that accompany the redox chemistry. The
dithiocarbamate ligands in the molybdenum(IV) complex
(dtc)2MoO must rearrange from trans to cis in order to form
the molybdenum(V) µ-oxo complex (eq 8), and chloride transfer
is required in addition to oxygen transfer in the porphyrinato-
chromium or -titanium reactions. Often, the complex in the
lower oxidation state binds a ligand which must dissociate to
open a site to accommodate the bridging oxygen.
In contrast, the (mes)3IrO/(mes)3Ir and (ArN)3OsO/(ArN)3Os
redox couples examined here are well-suited to studying oxygen
atom transfer free from complications of ligand dissociation or
rearrangement. Both compounds exist in the lower oxidation
state as stable, three-coordinate monomers. Oxygen atom
transfer can take place near thermoneutrality among these
complexes (Table 3), not only in the degenerate reactions but
also in the cross-exchange reaction (KIrOs ) 0.6 at 293 K). Even
conproportionation of (mes)3IrO with (mes)3Ir to form the µ-oxo
complex (mes)3IrsOsIr(mes)3 is not strongly favored at room
temperature (∆G° ) -0.69 kcal/mol at 293 K).
Discussion
Factors Governing Rates of Intermetal Oxygen Atom
Transfer. Oxygen atom transfer between transition metal centers
is generally accepted to proceed via a µ-oxo intermediate (cf.
eq 1). In many cases conproportionation is favored and µ-oxo
complexes are the final observed products.30 Quantitative studies
of the rates of intermetal oxygen atom transfer reactions are
surprisingly rare. One well-characterized example is the mo-
lybdenum(VI)/molybdenum(IV) redox couple (dtc)2MoO2/(dtc)2-
MoO (dtc ) N,N-diethyldithiocarbamate), where oxygen atom
transfer proceeds through an observable molybdenum(V) in-
termediate, (dtc)2Mo(O)sOsMo(O)(dtc)2 (eq 8). In fact, con-
Comparison of the iridium and osmium systems demonstrates
that ancillary ligand dissociation or rearrangement is not the
only factor limiting oxygen atom transfer rates between metals.
Despite the thermodynamic and structural similarities between
the iridium and osmium systems, atom transfer in the two
systems takes place at vastly differing rates. (ArN)3OsdO
transfers its oxygen atom to (ArN)3Os with a rate constant of
proportionation in this system is thermodynamically favored (Keq
) 500 M-1 at 298 K) and proceeds at a moderate rate (kMoMo
) 1470 M-1 s-1).31 Woo and co-workers have also examined
near-degenerate oxygen/chlorine exchange in porphyrinato
chromium32 and titanium complexes33 (TTP)MdO/(OEP)MCl.
(34) Estimated from the phosphorus-oxygen bond strength of 133 kcal/mol
(ref 3b) and the electrochemical data in Moyer, B. A.; Meyer, T. J. Inorg.
Chem. 1981, 20, 436-444.
(35) (a) Song, J.-S.; Bullock, R. M.; Creutz, C. J. Am. Chem. Soc. 1991, 113,
9862-9864. (b) Creutz, C.; Song, J.-S.; Bullock, R. M. Pure Appl. Chem.
1995, 67, 47-54.
(36) (a) Smith, T. P.; Iverson, D. J.; Droege, M. W.; Kwan, K. S.; Taube, H.
Inorg. Chem. 1987, 26, 2882-2884. (b) Kirchner, K.; Dodgen, H. W.;
Wherland, S.; Hunt, J. P. Inorg. Chem. 1990, 29, 2381-2385.
(37) Shea, T. M.; Deraniyagala, S. P.; Studebaker, D. B.; Westmoreland, T. D.
Inorg. Chem. 1996, 35, 7699-7703.
(29) Muldoon, J.; Brown, S. N. Org. Lett. 2002, 4, 1043-1045.
(30) Woo, L. K. Chem. ReV. 1993, 93, 1125-1136.
(31) Matsuda, T.; Tanaka, K.; Tanaka, T. Inorg. Chem. 1979, 18, 454-457.
(32) Woo, L. K.; Goll, J. G.; Berreau, L. M.; Weaving, R. J. Am. Chem. Soc.
1992, 114, 7411-7415.
(33) Hays, J. A.; Day, C. L.; Young, V. G.; Woo, L. K. Inorg. Chem. 1996, 35,
7601-7607.
9
596 J. AM. CHEM. SOC. VOL. 129, NO. 3, 2007

Ultrafast/Ultraslow O Transfer between Late Metal Centers
A R T I C L E S
Table 3. Summary of Experimental and Calculated Thermodynamic and Kinetic Parameters for Oxygen Atom Transfer between R3Ir/
R3IrdO and (R′N)3Os/(R′N)3OsdOa
R3IrdO + IrR3 f
R3IrsOsIrR3
-5.14(13)
-17.7
R3IrdO + Os(NR′)3 f
(R′N)3OsdO + Os(NR′)3 f
R3Ir + OdOs(NR′)3
-0.86(3)
17.7
(R′N)3Os + OdOs(NR′)3
∆H° (kcal/mol)
0
∆S° (cal/mol‚K)
-15.2(7)
-29.1
-3.89(10)
-5.2
0.28(4)
0
0
∆G° (kcal/mol, 293 K)
-0.69(24)
-9.1
19.4
∆H‡ (kcal/mol)
4.90(19)
-6.6(11)
6.8(4)
11.3(5)
b
b
∆S‡ (cal/mol‚K)
-15.3(24)
∆G‡ (kcal/mol, 293 K)
15.8(9)
23.50(6)
k (293 K, M-1 s-1
)
5 × 107c
1.0 × 101c
1.8(2) × 10-5
a Experimental values are given in roman type for R ) mes, R′ ) 2,6-iPr2C6H3, with calculated values in italic type for R ) R′ ) CH3. All ∆G° values
are at 293 K. b Not measured. c Extrapolated to 293 K.
calculations indicate that the former orbital is the HOMO and
the latter is the LUMO of the complex, with a sizable energy
gap of 2.5 eV between them. The a2u orbital interacts with the
oxygen 2p orbital and is strongly Ir-O σ* in character. The
a1g combination, in contrast, can interact only with the 2s orbital
on oxygen, and this orbital is much lower-lying than the 2p.38
The energy match with iridium is correspondingly poor, so the
a1g orbital is only weakly antibonding, and remains low in
energy.
All of these orbitals lie along the z-axis. Thus, as (mes)3Ir
approaches OdIr(mes)3, the nonbonding pair of electrons in its
a1 orbital can interact with the a1 LUMO of the oxo complex,
ending up in the low-lying a1g orbital. Productive dissociation
follows the microscopic reverse pathway, with the roles of the
iridium atoms interchanged. Only a modest barrier is expected
to this allowed reaction, consistent with the observed high rates.
Calculations in the gas phase do not find any barrier to
conproportionation and overestimate the stability of the µ-oxo
complex (Table 3). Solvation might partially account for the
differences between the calculations and observations, as the
smaller accessible surface of the dimer with respect to the two
monomers would make it (and presumably the transition state)
less enthalpically (though more entropically) favorable in
solution than in the gas phase. The smaller size of the methyl
groups used in the calculation is also expected to exaggerate
the calculated stability of the µ-oxo complex compared to the
mesityl complex. For these reasons, and because of the inherent
difficulties in calculations involving third-row transition metals,
it is not expected that the computed thermodynamic values
would be quantitatively accurate. Nevertheless, the agreement
between the experimental and computed geometries (Table 2)
indicates that the results from electronic structure calculations
can be used in the qualitative arguments outlined above.
Note that the bonding here differs significantly from that in
the tetrahedral monooxo complexes (silox)3MdO (M ) Nb,
Ta), whose stereoelectronic requirements for oxygen atom
transfer have been examined by Wolczanski and co-workers.39
These early metal oxo complexes are d0, so their LUMOs are
of E symmetry, which requires a bent approach to allow oxygen
atom transfer. This was experimentally demonstrated in the
Figure 5. Frontier orbitals of (CH3)3Ir, (CH3)3IrdO, and (CH3)3IrsOs
Ir(CH3)3. (a) (CH3)3Ir HOMO, E ) -5.65 eV. (b) (CH3)3IrdO LUMO,
E ) -2.61 eV. (c) (CH3)3IrsOsIr(CH3)3 HOMO, E ) -5.30 eV. (d)
(CH3)3IrsOsIr(CH3)3 LUMO, E ) -2.80 eV.
only 1.8 × 10-5 M-1 s-1, 12 orders of magnitude slower than
the analogous reaction of iridium.
The facility of oxygen transfer between iridium centers can
be illuminated by an analysis of the bonding in the iridium-
(III), (IV), and (V) complexes. The effective C3V symmetry of
the (mes)3Ir fragment implies that the iridium d orbitals split
into two pairs of e symmetry and a lone orbital of a1 symmetry.
In order for d6 (mes)3Ir to be diamagnetic, one e set and the a1
2
orbital (derived from the Ir dz ) must be occupied. This is
confirmed by DFT calculations (on (CH3)3Ir), which indicate
that the a1 orbital is the HOMO (Figure 5). Analogous symmetry
arguments applied to diamagnetic d4 (mes)3IrdO indicate that
the a1 orbital in this case must be empty; DFT calculations
indicate that this Ir-O σ* orbital is the LUMO of the complex.
Interaction of the hydrocarbyl ligands with these orbitals is
minimal because they lie approximately in the nodal cone
2
of dz .
In a linear µ-oxo dimer of approximately D3d symmetry, the
2
two dz orbitals will combine to form orbitals of a1g and a2u
t
symmetry. Again, the observed diamagnetism of the dimer, with
10 d electrons, requires that exactly 1 of these 2 orbitals is filled
(since all the other d orbitals occur as degenerate pairs). DFT
microscopic reverse direction, in that bulky Bu3PO, which is
(39) (a) Veige, A. S.; Slaughter, L. M.; Wolczanski, P. T.; Matsunaga, N.;
Decker, S. A.; Cundari, T. R. J. Am. Chem. Soc. 2001, 123, 6419-6420.
(b) Veige, A. S.; Slaughter, L. M.; Lobkovsky, E. B.; Wolczanski, P. T.;
Matsunaga, N.; Decker, S. A.; Cundari, T. R. Inorg. Chem. 2003, 42, 6204-
6224.
(38) For O, ꢀp - ꢀs ) 16.5 eV: Allen, L. C. J. Am. Chem. Soc. 1989, 111,
9003-9014.
9
J. AM. CHEM. SOC. VOL. 129, NO. 3, 2007 597

A R T I C L E S
Fortner et al
effect of the λi term on electron-transfer rates has been
extensively explored.41
Viewed from this perspective, the (mes)3Ir fragment supports
exceptionally rapid oxygen atom transfer because it has an
exceptionally small inner-sphere reorganization energy. Indeed,
crystallographic results on all three oxidation states reveal a
(mes)3Ir fragment whose IrsC bond lengths and angles are
identical within experimental error (Table 2). The only signifi-
cant bonding change on oxygen transfer takes place along the
IrsOsIr axis (as it must in order to achieve atom transfer). In
contrast, the reduced osmium complex, Os(NAr)3, is planar,
while that fragment is pyramidal in OdOs(NAr)3. Because of
the dπ-pπ repulsions on pyramidalization, this bonding change
is expected to be energetically costly, and thus Os(NAr)3 is
expected to have a substantial inner-sphere reorganization energy
and correspondingly slow rates of oxygen atom transfer.
Figure 6. Qualitative orbital interactions in (RN)3Os-O-Os(NR)3.
sterically inhibited from attaining a bent geometry, is not
deoxygenated, while Me3PO is deoxygenated readily. But in
d4 (mes)3IrdO the lowest e level is filled, and the symmetry-
induced barrier to a linear approach is lifted because of the a1
symmetry of both the (mes)3IrdO LUMO and (mes)3Ir HOMO.
The bonding situation in the µ-oxo complex (mes)3IrsOsIr-
(mes)3 also differs from that of the phosphine oxide complex
(Ph3PO)Ir(mes)3 that is the presumed intermediate in the
oxidation of Ph3P by OdIr(mes)3. In the phosphine oxide
complex, the electronic differences between iridium and phos-
phorus favor localization of the HOMO on iridium rather than
phosphorus (in contrast to its delocalization over both iridiums
If the bonding changes on atom transfer can be considered
analogous to inner-sphere reorganization energy in electron-
transfer reactions, then the analogy between the present study
and explorations of the Marcus cross-relation is evident. This
is the first example where self-exchange and cross-exchange
kinetics of oxygen atom transfer have been measured in a system
where the thermodynamics of cross-exchange are known. The
cross relation has been applied to other one-electron inner-sphere
reactions, such as hydrogen atom transfer,42 as well as to proton
transfer,43 but this appears to be the first example of any
multielectron atom transfer reaction to have been examined in
this way. For example, while rates have been measured for the
two-electron pseudo-self-exchange reactions between nitrido-
metal and chlorometal porphyrins (both Mn(V)/Mn(III)44 and
Cr(V)/Cr(III)45) and the corresponding Mn/Cr cross-exchange,46
the cross-reactions are irreversible and the driving force is
unknown. A similar situation is found in the two-electron
halogen transfer reactions between Cp2RuX+ and Cp2Os.36a,37
2
in (mes)3IrsOsIr(mes)3). This filled dz orbital would interact
unfavorably with the filled orbitals on the oxygen, and in fact
the donor complex (Ph3P)Ir(mes)3 is observed to have a
sawhorse geometry, with a substantial barrier to phosphine
binding on the threefold axis. This destabilization of the pseudo-
tetrahedral geometry of (Ph3PO)Ir(mes)3 as initially formed by
atom transfer has been suggested as one factor contributing to
the modest rate of oxidation of PPh3 by (mes)3IrdO.6 Oxidation
of (mes)3Ir, in contrast, can form the stable linear µ-oxo complex
directly.
While Os(NAr)3 is a three-coordinate, late-metal complex like
(mes)3Ir, the osmium complex is planar rather than pyramidal.9
When the Os(NAr)3 fragment pyramidalizes to accept the
The Marcus cross-relation, if applied to the iridium-osmium
2
oxygen atom, the dz orbital overlaps with the a1 combination
oxygen atom transfer reaction studied here, predicts that ∆G‡
IrOs
of nitrogen p orbitals (Figure 6). In fact, all the d orbitals are
π-antibonding with respect to a pyramidal M(NR)3 fragment.40
Since a µ-oxo complex of the form (ArN)3Os-O-Os(NAr)3
must accommodate a pair of d electrons somewhere, this pair
of electrons must be π* in character and the µ-oxo complex is
expected to be a high-energy, effectively 20-electron intermedi-
ate. This is consistent with the large barrier to degenerate oxygen
atom transfer observed in the osmium system. Computationally,
all attempts to find a µ-oxo complex (MeN)3Os-O-Os(NMe)3
as a stable minimum or even a first-order saddle point failed; a
scan of the Os-O distance from 3 Å to 2.35 Å with all other
coordinates fully optimized leads to a monotonic rise in energy
of 10 kcal/mol (Figure S9).
Another way of thinking about these electronic influences
on the rate of oxygen atom transfer is to consider them as an
inner-sphere reorganization energy. Barriers to single electron
transfer in the Marcus framework are principally ascribed to
the effects of driving force and reorganization energy λ. The
reorganization energy can be further partitioned into an outer-
sphere term, λo, involving rearrangement of solvent and coun-
terions, and an inner-sphere term, λi, which encompasses the
changes in bond lengths and angles upon electron transfer. The
≈ /2(∆G‡ + ∆G‡
+ ∆G°IrOs).47 The activation barrier
1
IrIr
OsOs
for cross-exchange calculated in this way is 15.3 ( 0.4 kcal/
mol, in excellent agreement with the observed barrier of 15.8
( 0.9 kcal/mol. While it is far too early to decide whether the
Marcus-style analysis of oxygen atom transfer will be generally
applicable, its success in this case is part of a growing number
of studies that suggest that many of the same principles that
govern outer-sphere electron transfer are useful in understanding
the rates of atom transfer. In particular, effects of inner-sphere
reorganization, whose importance in atom transfer is dramati-
cally illustrated by the system described here, have been applied
to explain rates of hydrogen atom transfer,42 and proton-transfer
reactions can in many cases be rationalized by inner-sphere
reorganization energy, as in cases where slow proton transfers
(41) Stanbury, D. M. AdV. Chem. Ser. 1997, 253, 165-182.
(42) Roth, J. P.; Yoder, J. C.; Won, T.-J.; Mayer, J. M. Science 2001, 294,
2524-2526.
(43) Kristja´nsdo´ttir, S. S.; Norton, J. R. J. Am. Chem. Soc. 1991, 113, 4366-
4367.
(44) Woo, L. K.; Goll, J. G.; Czapla, D. J.; Hays, J. A. J. Am. Chem. Soc.
1991, 113, 8478-8484.
(45) Neely, F. L.; Bottomley, L. A. Inorg. Chem. 1997, 36, 5432-5434.
(46) Bottomley, L. A.; Neely, F. L. Inorg. Chem. 1997, 36, 5435-5439.
(47) This neglects the quadratic term (∆G°IrOs)2/8(∆G‡IrIr + ∆G‡OsOs), which is
justifiable given the low driving force for the cross-reaction (it amounts to
only 3 × 10-4 kcal/mol).
(40) Morrison, D. L.; Wigley, D. E. Inorg. Chem. 1995, 34, 2610-2616.
9
598 J. AM. CHEM. SOC. VOL. 129, NO. 3, 2007

Ultrafast/Ultraslow O Transfer between Late Metal Centers
A R T I C L E S
Scheme 2. Cocatalysis of Aerobic Oxidations by (mes)3IrdO
from carbon48 or from metal hydrides49 have been attributed to
the need to change the geometry and bonding at the center bound
to the proton. One is accustomed to thinking of inner-sphere
reactions in terms of the electronic details of how bonds can be
made and broken. The electronic contrast between the iridium
system, where transfer of the oxygen atom in a linear fashion
is orbitally allowed, and the osmium system, where it is not, is
in line with these expectations. Yet the orbital effects can be
viewed as simply another way of expressing inner-sphere
reorganization: the (mes)3Ir fragment is unchanged in the three
oxidation states because the bonding changes to oxygen do not
affect the metal-carbon bonds, and the electronic difficulties
in achieving a µ-oxo complex of osmium are reflected in the
energetic cost of pyramidalizing the Os(NAr)3 fragment.
The reorganization energy is certainly not the only factor in
atom transfer reactions. Clearly, a key distinction between
oxygen atom transfer and both outer-sphere electron transfer
and one-electron inner-sphere processes such as hydrogen atom
transfer is the presence of an intermediate in oxygen atom
transfer reactions. If this intermediate is high in energy, oxygen
atom transfer is slow regardless of the degree of inner-sphere
reorganization. For example, main group oxygen atom transfer
partners such as phosphines and phosphine oxides have structur-
ally similar R3P fragments, yet direct atom transfer between
main group partners is usually vanishingly slow because of the
marked instability of a putative µ-oxo intermediate such as
R3PsOsPR3. Similarly, the fact that (mes)3IrdO and (ArN)3Osd
O both oxidize triphenylphosphine at very similar rates con-
tradicts simple predictions based on reorganization energy,
which would imply much slower rates of reaction with osmium.
The instability of a pseudo-tetrahedral intermediate (mes)3Ir-
(OPPh3) (the phosphine adduct (mes)3Ir(PPh3) is over 12 kcal/
mol lower in free energy in its saddle-shaped geometry than as
a tetrahedron6) may contribute to the barrier for phosphine
oxidation by iridium in this case.
oxygen atom rapidly to other metal centers raises the intriguing
possibility of a general scheme for cocatalyzed air oxidation
(Scheme 2). Many reagents can oxidize organic substrates but
are kinetically unable to be reoxidized by dioxygen. If they were
able to accept an oxygen atom from (mes)3IrdO, then the
addition of catalytic amounts of oxotrimesityliridium could
transform an otherwise stoichiometric organic oxidation to a
catalytic one using dioxygen as the stoichiometric oxidant, with
potential advantages in cost and environmental impact. Fur-
thermore, since the iridium would interact only with the oxidant
and not with the substrate, the selectivity of the oxidation would
be that of the stoichiometric reagent and would be unaffected
by the iridium.
Use of oxotrimesityliridium(V) as a cocatalyst in air oxida-
tions has been demonstrated in two systems: with amine-ligated
osmium tetroxide in the dihydroxylation of terminal alkenes,26
and with osmium(VI) esters in the oxidation of allyl and benzyl
alcohols to aldehydes.29 Both reactions require stoichiometric
osmium in the absence of iridium but can use air as the
stoichiometric oxidant in the presence of catalytic (mes)3IrdO.
The “innocence” of the iridium (the fact that it is not involved
in substrate oxidation) is demonstrated in the first reaction by
the observation that the enantiomeric excess of 1,2-octanediol
using (DHQD)2PHAL as a chiral ligand is identical in the
stoichiometric oxidation and cocatalytic oxidations.
Osmium/iridium-cocatalyzed dihydroxylations are strongly
inhibited by the 1,2-diol product (Figure S8), which limits the
number of turnovers (∼5 per week) and the scope of the reaction
(more substituted alkenes, whose diols bind more tightly, do
not turn over). The product inhibition supports a mechanism in
which reversible loss of diol occurs from the observed osmium-
(VI) diolate resting state preceding rate-determining oxygen
atom transfer from iridium to the osmium complex produced
by loss of diol. Assignment of this unobserved osmium species
as “OsO3” (or a ligated form thereof) is plausible, given the
stability of isoelectronic Os(NAr)3 and its proven ability to
accept an oxygen atom from (mes)3IrdO. Little can be said
about the reoxidation step in the alcohol oxidation; it is unknown
even whether reoxidation takes place at the Os(IV) or Os(V)
oxidation state.
The thermodynamic instability of the µ-oxo osmium com-
plexes (ArN)3OssOsMX3 is an adequate qualitatiVe explana-
tion for the slower rate of intermetal oxygen atom transfer
involving osmium. The kinetic effects in intermetal oxygen atom
transfer, however, are not negligible: for iridium, essentially
thermoneutral formation of the µ-oxo intermediate nevertheless
has a 6.8 kcal/mol barrier. The quantitatiVe agreement between
the observed and calculated cross-reaction rates suggests that
not only the thermodynamic effects on the stability of the
intermediate but also the kinetic effects in its formation scale
linearly on progressing from iridium to osmium.
Use of (mes)3IrdO as an “Innocent” Dioxygen Activator.
Trimesityliridium(III) undergoes extremely rapid oxygen atom
transfer with oxotrimesityliridium(V), but this reaction is
degenerate and therefore cannot be harnessed to achieve net
chemical change. The ability of the (mes)3Ir fragment to mediate
intermetal oxygen atom transfer extends to nondegenerate
intermetal atom transfer reactions as well, as demonstrated by
the ability of (mes)3IrdO to oxidize Os(NAr)3 at respectable
rates (kIrOs ) 10 M-1 s-1, extrapolated to 293 K).
In neither of these particular reactions does use of (mes)3Ird
O represent a practical advance as an oxygen activator, since
dioxygen use in both dihydroxylations50 and alcohol oxidations51
using osmium is cocatalyzed equally effectively with much
cheaper copper salts, and indeed under certain conditions aerobic
oxidations of alkenes27 and alcohols52 can take place without a
cocatalyst (although the selectivity in the latter reaction is
different from that observed here). Still, the fact that (mes)3Ird
O can be used as an efficient activator of dioxygen in catalytic
Combining the ability of (mes)3Ir to activate O2 to form
(mes)3IrdO6,7 with the ability of (mes)3IrdO to transfer its
(50) (a) Michaelson, R. C.; Austin, R. G. U.S. Patent 4,390,739 (June 28, 1983).
(b) Austin, R. G.; Michaelson, R. C.; Myers, R. S. U.S. Patent 4,824,969
(April 25, 1989).
(48) (a) Bernasconi, C. F. Acc. Chem. Res. 1992, 25, 9-16. (b) Costentin, C.;
Save´ant, J.-M. J. Am. Chem. Soc. 2004, 126, 14787-14795.
(49) (a) Kramarz, K. W.; Norton, J. R. Prog. Inorg. Chem. 1994, 42, 1-65. (b)
Jordan, R. F.; Norton, J. R. J. Am. Chem. Soc. 1982, 104, 1255-1263.
(51) Coleman, K. S.; Coppe, M.; Thomas, C.; Osborn, J. A. Tetrahedron Lett.
1999, 40, 3723-3726.
(52) Do¨bler, C.; Mehltretter, G. M.; Sundermeier, U.; Eckert, M.; Militzer, H.-
C.; Beller, M. Tetrahedron Lett. 2001, 42, 8447-8449.
9
J. AM. CHEM. SOC. VOL. 129, NO. 3, 2007 599

A R T I C L E S
Fortner et al
reactions has potentially important consequences. First, the well-
defined nature of the (mes)3IrdO/(mes)3Ir redox couple and its
inertness to most reactions other than intermetal oxygen atom
transfer may make it a suitable cooxidant in delicate systems
where copper salts would be unsuitable. For example, many
late-metal complexes have been prepared which activate hy-
drocarbons, but linking this reactivity with dioxygen reduction
to achieve catalytic selective hydrocarbon oxidation remains a
challenge.53 In these and other organometallic systems, the
inertness of its supporting ligands and its reluctance to undergo
outer-sphere electron transfer may make trimesityliridium a more
attractive cooxidant than, say, copper salts. Second, the analysis
presented here highlights some of the reasons for the reactivity
of the (mes)3IrdO/(mes)3Ir couple toward oxygen atom transfer,
notably the importance of minimizing inner-sphere reorganiza-
tion energy. The use of these principles may aid in the
development of new, perhaps cheaper or more reactive, oxygen
atom transfer agents for use in catalysis.
is structurally identical in the III, IV, and V oxidation states).
The ability of (mes)3Ir to activate dioxygen and then transfer
an oxygen atom to another metal has been demonstrated in
catalytic cycles involving osmium complexes as the substrate-
activating species to achieve alkene and alcohol oxidation.
Acknowledgment. We thank Dr. Alicia Beatty for her
assistance with the X-ray crystallography, Dr. Bill Boggess for
his assistance with the mass spectra, and Dr. Doug Miller for
his assistance with the low-temperature optical spectroscopy
experiments. The Center for Research Computing at the
University of Notre Dame provided computing resources for
the electronic structure calculations. Financial support from the
Camille and Henry Dreyfus Foundation (New Professor Award),
DuPont (Young Professor Award), and the Dow Chemical
Company (Innovation Recognition Program) is gratefully ac-
knowledged. L.P. thanks the Bayer Corporation for a postdoc-
toral fellowship, administered by the Notre Dame Center for
Environmental Science and Technology. S.B.S. thanks the Clare
Boothe Luce Foundation and the National Science Foundation
for graduate fellowships.
Conclusions
Oxotrimesityliridium(V), (mes)3IrdO, undergoes degenerate
oxygen atom transfer with trimesityliridium(III), (mes)3Ir, at
unprecedentedly fast rates (kIrIr ) 5 × 107 M-1 s-1, extrapolated
to 293 K). In contrast, the osmium(VIII) oxo complex (ArN)3Osd
O, with an essentially identical metal-oxygen bond strength,
transfers its oxygen to (ArN)3Os 12 orders of magnitude more
slowly (kOsOs ) 1.8 × 10-5 M-1 s-1). Cross-reaction takes place
at intermediate rates, in quantitative agreement with a Marcus-
type analysis. The unusual facility of the iridium oxygen atom
transfer can be explained by orbital alignment (both the HOMO
of (mes)3Ir and the LUMO of (mes)3IrdO have σ symmetry,
making atom transfer allowed in a linear geometry) and by low
inner-sphere reorganization energy (the triaryliridium geometry
Supporting Information Available: Complete refs 22 and 23;
variable-temperature NMR spectra of (mes)3IrdO; Eyring plots
of bond rotation in (mes)3IrdO, dissociation of (mes)3IrsOs
Ir(mes)3, deoxygenation of (ArN)3OsdO by PPh3, and oxygen
atom transfer from (mes)3IrdO to Os(NAr)3; van’t Hoff plot
of oxygen atom transfer from (mes)3IrdO to Os(NAr)3; kinetic
traces for oxygen atom transfer between (ArN)3OsO and
(ArDN)3Os and for (mes)3IrO/(quinuclidine)OsO4-cocatalyzed
air oxidation of octene; Cartesian coordinates, energies, and
thermal corrections of all optimized structures; and a plot of
energy vs distance scan for (CH3N)3OssOsOs(NCH3)3 (PDF
format); and crystallographic data for (Me2C6H3)3IrsOsIr-
(C6H3Me2)3‚CD2Cl2 in CIF format. This material is available
free of charge via the Internet at http://pubs.acs.org.
(53) (a) Rostovtsev, V. V.; Henling, L. M.; Labinger, J. A.; Bercaw, J. E. Inorg.
Chem. 2002, 41, 3608-3619. (b) Ackerman, L. J.; Sadighi, J. P.; Kurtz,
D. M.; Labinger, J. A.; Bercaw, J. E. Organometallics 2003, 22, 3884-
3890. (c) Vedernikov, A. N.; Binfield, S. A.; Zavalij, P. Y.; Khusnutdinova,
J. R. J. Am. Chem. Soc. 2006, 128, 82-83.
JA065713H
9
600 J. AM. CHEM. SOC. VOL. 129, NO. 3, 2007