Mechanisms of water oxidation catalyzed by the cis,cis-[(bpy) 2Ru(OH2)]2O4+ ion

Article abstract of DOI:10.1021/ja030594g

The cis,cis-[(bpy)₂Ru^III(OH₂)] ₂O⁴⁺ μ-oxo dimeric coordination complex is an efficient catalyst for water oxidation by strong oxidants that proceeds via intermediary formation of cis,cis-[(bpy)₂Ru^V(O)]₂O ⁴⁺ (hereafter, {5,5}). Repetitive mass spectrometric measurement of the isotopic distribution of O₂ formed in reactions catalyzed by ¹⁸O-labeled catalyst established the existence of two reaction pathways characterized by products containing either one atom each from a ruthenyl O and solvent H₂O or both O atoms from solvent molecules. The apparent activation parameters for μ-oxo ion-catalyzed water oxidation by Ce⁴⁺ and for {5,5} decay were nearly identical, with ΔH? = 7.6 (±1.2) kcal/mol, ΔS? = -43 (±4) cal/deg mol (23 °C) and ΔH? = 7.9 (±1.1) kcal/mol, ΔS? = -44 (±4) cal/deg mol, respectively, in 0.5 M CF₃SO₃H. An apparent solvent deuterium kinetic isotope effect (KIE) of 1.7 was measured for O₂ evolution at 23 °C; the corresponding KIE for {5,5} decay was 1.6. The ³²O₂/ ³⁴O₂ isotope distribution was also insensitive to solvent deuteration. On the basis of these results and previously established chemical properties of this class of compounds, mechanisms are proposed that feature as critical reaction steps H₂O addition to the complex to form covalent hydrates. For the first pathway, the elements of H₂O are added as OH and H to the adjacent terminal ruthenyl O atoms, and for the second pathway, OH is added to a bipyridine ring and H is added to one of the ruthenyl O atoms.

Full text of DOI:10.1021/ja030594g

Published on Web 07/14/2004
Mechanisms of Water Oxidation Catalyzed by the
4+
2 2 2
cis,cis-[(bpy) Ru(OH )] O
Ion
†
‡
†
,‡
Hiroshi Yamada, William F. Siems, Tohru Koike, and James K. Hurst*
Contribution from the Department of Chemistry, National Defense Academy,
1
-10-20 Hashirimizu, Yokosuka, Kanagawa 239, Japan, and Department of Chemistry,
Washington State UniVersity, Pullman, Washington 99164-4630
Received October 23, 2003; Revised Manuscript Received May 10, 2004; E-mail: hurst@wsu.edu
III
O4+ µ-oxo dimeric coordination complex is an efficient catalyst for
Abstract: The cis,cis-[(bpy)
2 2
Ru (OH )]
2
V
4+
water oxidation by strong oxidants that proceeds via intermediary formation of cis,cis-[(bpy)
2
Ru (O)]
formed in
reactions catalyzed by O-labeled catalyst established the existence of two reaction pathways characterized
by products containing either one atom each from a ruthenyl O and solvent H O or both O atoms from
solvent molecules. The apparent activation parameters for µ-oxo ion-catalyzed water oxidation by Ce
2
O
(
hereafter, {5,5}). Repetitive mass spectrometric measurement of the isotopic distribution of O
2
18
2
4+
q
q
and for {5,5} decay were nearly identical, with ∆H ) 7.6 ((1.2) kcal/mol, ∆S ) -43 ((4) cal/deg mol (23
C) and ∆H ) 7.9 ((1.1) kcal/mol, ∆S ) -44 ((4) cal/deg mol, respectively, in 0.5 M CF
q
q
°
3
SO
3
H. An
apparent solvent deuterium kinetic isotope effect (KIE) of 1.7 was measured for O
2
evolution at 23 °C; the
corresponding KIE for {5,5} decay was 1.6. The 32
34
2 2
O / O isotope distribution was also insensitive to solvent
deuteration. On the basis of these results and previously established chemical properties of this class of
compounds, mechanisms are proposed that feature as critical reaction steps H O addition to the complex
to form covalent hydrates. For the first pathway, the elements of H O are added as OH and H to the
2
2
adjacent terminal ruthenyl O atoms, and for the second pathway, OH is added to a bipyridine ring and H
is added to one of the ruthenyl O atoms.
Scheme 1. Detectable Redox States of the µ-Oxo Ion8a
Introduction
Almost 20 years ago, T. J. Meyer’s group reported that cis,cis-
[
(bpy)2Ru(OH2)]2O4+ was a remarkably effective catalyst for
1
water oxidation by strong oxidants. This report stimulated
considerable global interest in the underlying chemical reaction
mechanisms, which have been viewed as relevant to biological
reactions involving metal dinuclear and tetranuclear µ-oxo
reaction centers and potentially important to development of
practical devices for large-scale energy utilization. However,
a
The reduction potentials shown are for 0.5 M CF3SO3H and are
7
referenced against NHE. Decay of {5,5} may lead to either H2O2 or O2 as
2
the immediate product.
3
despite subsequent intensive investigation of reactions catalyzed
In earlier redox titrimetric and electrochemical studies, we
established that, in dilute trifluoromethanesulfonic (triflic) acid,
the [(bpy)2Ru(OH2)]2O ion undergoes progressive one- and
by this ion and related complexes of the general formula cis,cis-
n+ 4-6
4+
[
L2Ru(OH2)]2O ,
the reaction mechanisms remain obscure.
(
In the general formula, L represents a functionalized congener
two-electron oxidations with equivalent loss of protons to give
ultimately the corresponding 4 e -oxidized product (designated
-
of 2,2′-bipyridine or a related diimine.)
{
5,5} in Scheme 1) containing ruthenyl oxo atoms in place of
†
7
National Defense Academy.
Washington State University.
the original water molecules. As discussed herein, the {5,5}
ion undergoes first-order decay to the {4,4} ion with rate
parameters that approximate those for O2 formation measured
‡
(
(
(
1) Gilbert, J. A.; Eggleston, D. S.; Murphy, W. S., Jr.; Geselowitz, D. A.;
Gersten, S. W.; Hodgson, D. J.; Meyer, T. J. J. Am. Chem. Soc. 1985, 107,
3
855-3864.
9
under steady-state conditions, indicating that the {5,5} ion is
2) See, for example: (a) Hoganson, C. W.; Babcock, G. T. Science 1997,
77, 1953-1956. (b) Pecoraro, V. L.; Baldwin, M. J.; Caudle, M. T.; Hsieh,
2
either the O2-evolving species or its immediate precursor. In
contrast, decay of the lower oxidation states (Scheme 1) is too
slow for them to be directly involved in O2 formation. A variety
of mechanisms have been proposed for this reaction that focus
almost universally upon the ruthenyl group. As summarized in
W.-Y.; Law, N. A. Pure Appl. Chem. 1998, 70, 925-929.
3) (a) Energy Resources through Photochemistry and Catalysis; Gr a¨ tzel, M.
Ed.; Academic Press: New York, 1983. (b) Photochemical ConVersion
and Storage of Solar Energy; Connolly, J. S., Ed.; Academic Press: New
York, 1981.
(
(
4) Extensive compilations of earlier papers are given in refs 5 and 6.
5) Chronister, C. W.; Binstead, R. A.; Ni, J.; Meyer, T. J. Inorg. Chem. 1997,
3
6, 3814-3815.
(
6) Lei, Y.; Hurst, J. K. Inorg. Chem. 1994, 33, 4460-4467.
(7) Yamada, H.; Hurst, J. K. J. Am. Chem. Soc. 2000, 122, 5303-5311.
9786
9
J. AM. CHEM. SOC. 2004, 126, 9786-9795
10.1021/ja030594g CCC: $27.50 © 2004 American Chemical Society

Mechanisms of Water Oxidation
A R T I C L E S
Scheme 2. Representative Pathways for O-O Bond Formation Involving the Diruthenyl Unit of the {5,5} Iona
a
Only key intermediates are shown.
Scheme 2, they include reductive elimination of ruthenyl oxo
atoms derived from coordinated aqua ligands in higher oxidation
In the present study, we describe mass spectrometric methods
that allow one to follow in real time the course of evolution of
1
8
states of the µ-oxo ion, forming either O2 directly or H2O2 as
O2 in the {5,5} ion. Using O-isotopically enriched catalyst,
we show that the isotopic distribution of O2 obtained is
quantitatively accountable by invoking two reaction pathways,
both of which involve incorporation of solvent oxygen atoms
in the reaction product. These studies eliminate from further
consideration pathways that involve either (1) internal re-
arrangement within a single complex ion with reductive
elimination of O2 or (2) bimolecular reactions between two
complex ions with direct O-O bond formation between their
coordinated oxygen-containing ligands. Furthermore, the mea-
sured temperature and deuterium solvent isotope dependencies
of O2 evolution rates appear inconsistent with a simple bi-
molecular reaction between H2O and the {5,5} ion, but rather
a reaction intermediate;1
0,11
analogous bimolecular reactions
involving two catalyst ions;12 and reactions between solvent H2O
and ruthenyl oxo atoms, leading to incorporation of either one
or two solvent oxygen atoms in the final O2 product.
1
2-14
(
8) The electrons involved in forming the µ-oxo bridge are extensively
delocalized over the Ru-O-Ru unit (see, for example: Weaver, T. R.;
Meyer, T. J.; Adeyemi, S. A.; Brown, G. M.; Eckberg, R. P.; Hatfield, W.
E.; Johnson, E. C.; Murray, R. W.; Untereker, D. J. Am. Chem. Soc. 1975,
9
7, 3039-3048). Thus, for example, the notation {3,4} signifies that the
Ru-O-Ru unit contains 13 π-symmetry electrons (with four originating
in O2 ), but is not meant to imply the existence of discrete valence-localized
-
4
5
ruthenium d and d ions.
(
9) Lei, Y.; Hurst, J. K. Inorg. Chim. Acta 1994, 226, 179-185.
(
(
(
(
10) Rotzinger, F. P.; Munavalli, S.; Comte, P.; Hurst, J. K.; Gr a¨ tzel, M.; Frank,
A. J. J. Am. Chem. Soc. 1987, 109, 6619-6626.
11) Comte, P.; Nazeruddin, F. P.; Rotzinger, F. P.; Frank, A. J.; Gr a¨ tzel, M. J.
Mol. Catal. 1989, 52, 63-84.
12) Binstead, R. A.; Chronister, C. W.; Ni, J.; Hartshorn, C. M.; Meyer, T. J.
J. Am. Chem. Soc. 2000, 122, 8464-8473.
(14) Meyer, T. J. In Oxygen Complexes and Oxygen ActiVation by Transition
Metals; Martell, A. E.; Sawyer, D. T., Eds.; Plenum Press: New York,
1988; pp 33-48.
13) Hurst, J. K.; Zhou, J.; Lei, Y. Inorg. Chem. 1992, 31, 1010-1017.
J. AM. CHEM. SOC.
9
VOL. 126, NO. 31, 2004 9787

A R T I C L E S
Yamada et al.
7
suggest that O2 formation is preceded by rate-limiting uni-
molecular activation of the complex ion, perhaps by internal
electronic rearrangement. A novel reaction mechanism is
proposed that features formation of a bipyridyl ligand π-cation
with reaction of solvent at that site; this mechanism may have
precedents in water oxidation and other ligand reactions reported
electrolysis cell attached to a potentiostat as previously described. To
conserve the dimeric coordination complex, the electrolysis cell was
modified for experiments measuring the temperature dependence of
5,5} decay. In this case, a thin 5.9 mm i.d. porous Vycor tube
Bioanalytical Systems) filled with carbon fibers was used in place of
the larger standard 8.2 mm i.d. electrode, allowing preparation of the
5,5} ion at a slower flow rate. The effluent from the cell was run
{
(
{
1
5-22
for a broader range of ruthenium bipyridyl complexes.
directly into a 1 mm or 2 mm optical cell and the UV-vis spectra
were repetitively recorded following stoppage of the flow. Oxygen
evolution experiments were conducted using reaction chambers com-
posed of 6-10 mL reservoirs encased within outer glass chambers
through which thermostated fluids could be circulated from a constant-
temperature bath. Access to the reaction chamber was provided by two
2 mm bore capillary glass tubes through which purging gases or reagents
could be introduced and a larger ∼12 mm tube whose diameter was
appropriate to receive a YSI model 5331 oxygen probe. This probe
was fitted with O-rings to achieve an airtight seal within the tube. Rates
of O2 evolution were made on continuously stirred argon-purged
samples with the probe located within the gaseous headspace region
of the reaction chamber. In a typical run, reaction was initiated by
syringe addition of 0.25 mL of 0.30 M Ce4 through a septum to 6.75
mL of acidic solutions containing 0.10 mM of the {3,3} ion, and the
response of the electrode was recorded as a function of time using a
strip-chart recorder. The electrode was calibrated by measuring the
amplitude of the pen deflection when air originally present in the
chamber was replaced by Ar. For experiments measuring the mass
Experimental Section
Materials. The µ-oxo-bridged dimeric ruthenium coordination
4
+
complex cis,cis-[(bpy)
2
Ru(OH
2
)]
2
O
was prepared as the perchlorate
1
salt following well-established synthetic procedures and was recrystal-
lized repeatedly to remove minor impurities. The final product was
7
isolated as fine dark blue crystalline needles that were ∼1 mm in length.
Concentrations of reagent solutions of the {3,3} ion were determined
4
-1
-1 7
spectrophotometrically using ꢀ636 ) 2.2 × 10 M cm . Purity was
established by spectroscopically monitoring one-electron oxidative
titrations of the {3,3} ion in 0.5-1.0 M trifluoromethanesulfonic (triflic)
acid with Ce . These reactions occurred with [Ce4+]/[{3,3}] stoi-
chiometries of 1.0. Five sharp isosbestic points (at 533, 404, 340, 294,
and 261 nm) were observed in the optical spectra over the course of
the titration, and changes in the absorption maxima of {3,3} at 636
nm and {3,4} at 448 nm were proportional to the amount of added
4+
+
4+
Ce . The visible isosbestic points are particularly good criteria of purity
because the impurities formed during preparation absorb strongly in
these regions. Reagent solutions of the complex containing 18O-
isotopically enriched aqua ligands were prepared by incubating the {3,3}
distributions of gases formed in the reaction chamber, the O electrode
2
was replaced by a Lucite plug in which was mounted one end of a
1
8
4+
ion (typically, at 7.5 mM) in 97.2% [ O]-H
been added sufficient triflic anhydride to give 0.5 M triflic acid. After
90 min at room temperature, the complex was oxidatively titrated to
the {3,4} state with a slight stoichiometric excess (∼5%) of concentrated
Ce . Under these conditions, the reaction half-times for water exchange
at the cis-aqua position were t1/2 ≈ 2 min for the {3,3} ion and t1/2
0 d for the {3,4} ion; exchange of the bridging µ-oxo atom with
2
O (ICON) to which had
glass capillary column. In a typical run, 0.9 mL of 0.25 M Ce was
added to 8.1 mL of 0.5 M triflic acid containing 0.04-0.8 mM of the
18
∼
O-isotopically enriched complex ion prepared as described in the
preceding section. Headspace gases were continuously withdrawn at
∼1 mL/min into a VG7070EHF double-focusing mass spectrometer
operated in the EI mode at 70 eV with a 10-50 m/z continuum scan.
The purging gas used in these experiments was He. A 10 mL syringe
containing He was attached via a Hamilton three-way valve to one of
the sidearms of the reaction cell. During runs, the valve was opened to
the reaction chamber, allowing gases withdrawn by the mass spec-
trometer to be replaced by He, thereby maintaining normal atmospheric
pressure in the chamber. Ceric ion was introduced via syringe transfer
through the other sidearm of the reaction cell to initiate the reaction.
Resonance Raman (RR) spectra were acquired at room temperature by
using 90° scattered light from sealed samples in glass capillaries. The
Raman spectrometer, which uses a McPherson 2061 spectrograph
coupled to a Princeton Instruments LN-1100PB CCD detector, has been
4+
>
1
8
8
O did not occur in either oxidation state.7,13
The O-
18
solvent H
2
isotopically enriched {3,4} ions were kept in H
2
18O until immediately
before their use in isotope tracer experiments, which was always within
a few hours of their preparation. The final enrichments were calculated
1
6
by assuming that the O atoms introduced upon addition of triflic
16
2
anhydride equilibrated with solvent H O during the incubation period
and that the isotopic composition at the cis-aqua positions were identical
to the final solvent composition. As calculated, the enrichments for
these preparations were 89-91%. Solutions of the analogous complex
18
18
18 4+
2 2
)] O ,
containing O in the bridging position, that is, [(bpy)
2
Ru( OH
in O-enriched H O using the same
procedures as normal {3,3} ion. In this case, however, the limited
availability of H
18O precluded purification of the product. Triflic acid
99% CF SO H, Alfa Aesar) was vacuum-distilled and stored at 10 °C
18
23,24
were prepared from cis-Ru(bpy)
2
Cl
2
2
previously described.
1
3
Data Analyses. To avoid complications associated with the slow
secondary conversion of {4,4} to {3,4}, decay of the {5,5} ion was
analyzed from changes in optical absorption at wavelengths corre-
sponding to the isosbestic points of the {4,4} and {3,4} ions (470 nm
2
(
3
3
as 0.5-1.0 M aqueous solutions. Other chemicals were reagent grade
and used as received from commercial suppliers; water was purified
using a Milli-Q ion exchange/reverse osmosis system.
Analytical Methods. The kinetics of decay of the {5,5} ion were
monitored spectrophotometrically by using a computer-interfaced HP
in 0.5 M triflic acid/H
these wavelengths, the reactions exhibited simple first-order behavior
over 3 to 4 half-lives. Determination of O evolution rates required
2 2
O and 468 nm in 0.5 M triflic acid/D O). At
2
knowledge of the headspace volume in the reaction cell, which was
estimated from the difference in weight of the cell when empty and
8
452A diode array instrument. For most experiments, the {5,5} ion
was prepared electrochemically by using a carbon fiber columnar flow
filled with H
were estimated to introduce a 10-15% uncertainty in the O
which were calculated in units of nmol O formed/s. Prior studies had
established that these rates were linearly dependent upon [{5,5}].
The turnover number (kcat), which is a first-order rate constant, was
2
O. Variations in the effective volume for a given run
2
rates,
(
15) Creutz, C.; Sutin, N. Proc. Natl. Acad. Sci. U.S.A. 1975, 72, 2858-2862.
16) Nord, G.; Pedersen, B.; Bjergbakke, E. J. Am. Chem. Soc. 1983, 105, 1913-
2
(
7
,9
1
919.
(17) Ghosh, P. K.; Brunschwig, B. S.; Chou, M.; Creutz, C.; Sutin, N. J. Am.
Chem. Soc. 1984, 106, 4772-4783.
2
determined by dividing the O formation rate by the amount of catalyst
(
18) Ledney, M.; Dutta, P. K. J. Am. Chem. Soc. 1995, 117, 7687-7695.
19) Berg-Brennan, C.; Subramanian, P.; Absi, M.; Stern, C.; Hupp, J. T. Inorg.
Chem. 1996, 35, 3719-3722.
(
present in the cell.
(
20) Zhang, H.-T.; Yan, S. G.; Subramanian, P.; Skeens-Jones, L. M.; Stern,
C.; Hupp, J. T. J. Electroanal. Chem. 1996, 414, 23-29.
21) Rollick, K. L.; Kochi, J. K. J. Org. Chem. 1982, 47, 435-444.
22) Sag u¨ e´ s, J. A. A.; Gillard, R. D.; Lancashire, R. J.; Williams, P. A. J. Chem.
Soc., Dalton Trans. 1979, 193-198.
(23) Yamada, H.; Koike, T.; Hurst, J. K. J. Am. Chem. Soc. 2001, 123, 12775-
12780.
(24) Nakamura, N.; Mo e¨ nne-Loccoz, P.; Tanizawa, K.; Mure, M.; Suzuki, S.;
Klinman, J. P.; Sanders-Loehr, J. Biochemistry 1997, 36, 11479-11486.
(
(
9788 J. AM. CHEM. SOC.
9
VOL. 126, NO. 31, 2004

Mechanisms of Water Oxidation
Isotopic distributions of 32 , 34 , and 36
O O O formed in the catalyzed
2 2 2
A R T I C L E S
reaction were corrected for background counts by subtracting the
averaged intensities at these masses for 10-20 scans immediately prior
4
+
to adding Ce . Typically, these corrections to the data amounted to
10% for 34O
, but were 30-80% for the other ions, with the relative
magnitude of the background correction decreasing progressively as
the reaction proceeded and O accumulated in the reaction cell. The
2
reasons for the relative large corrections for O differed,
<
2
2
3
2
36
O
2
and
32
however. The background intensity for
of the presence of air in the spectrometer, whereas the amount of
O was relatively large because
2
36
O
2
produced in the catalyzed reaction was exceedingly small and did not
rise appreciably above the low background intensities recorded at this
mass. The relative yields of the three isotopes in individual scans were
calculated from the background-corrected mass intensities; as described
in the Results section, plots of these data vs time were made and
extrapolated to t ) 0 to obtain the isotope distribution from the catalyzed
reaction at the start of the reaction. This procedure corrects for the
small amount of isotopic dilution that occurs by incorporation of
3
4
2
coordinated O into the O product and its subsequent replacement in
3
2
the Ru coordination sphere by O from the solvent. The refined data
were analyzed by recognizing that there exist three possible types of
reaction pathways, defined by whether the O
2
formed contains zero,
one, or two O atoms obtained from the coordinated ruthenyl O atoms.
1
8
2
From the estimated isotopic composition of the O-enriched [(bpy) -
4
+
Ru(O)]
the reaction solution, one can calculate the O
expected for each of these types of pathways. These distributions can
2
O
ions used in the reaction and the known composition of
2
isotopic distribution
32
34
then be used with the experimentally determined yields of
2 2
O , O ,
3
6
and
2
O to evaluate the relative contributions of each of the pathways
in the overall reaction. In practice, this evaluation made use of the
following equations:
3
2
%
%
( O ) formed ) a x + b y + c z
2 1 1 1
Figure 1. Mass spectrometric analysis of evolved gases from µ-oxo ion-
34
( O ) formed ) a x + b y + c z
catalyzed water oxidation. (a) Changes in peak intensities of O isotopes at
2
2
2
2
2
4
+
ambient temperature following addition of Ce to give a final solution
1
8
18
4+
composition of 0.86 mM 89.4% O-enriched cis,cis-[(bpy)2Ru( O)]2O
where x, y, and z represent the fractional contributions of each of the
plus 25 mM Ce4 in 7.8% O-enriched H2O (0.5 M CF3SO3H). As
indicated, the upper and lower solid lines refer to intensities at 32 and 36
amu, and the dashed line refers to the intensity at 34 amu. (b) The 10-50
m/z spectrum for a scan at 75 s after initiation of the reaction. Peak
+
18
32
34
pathways, a
for pathway x under the prevailing conditions, and the pairs b
,c are analogously defined for pathways y and z. Simultaneous
1
and a
2
are the percent formation of
O
2
and
O
2
predicted
1
,b and
2
c
1
2
+
+
solution of these equations subject to the constraint, x + y + z ) 1,
gives unique values for x, y, and z.
assignments (relative intensities) are: m/z ) 14, N (0.6); m/z ) 16, O
16 16 18
+
+
+
(6.6); m/z ) 17, OH (8.3); m/z ) 18, OH2 (35.7); m/z ) 20, OH2 ;
+
32
+
34
+
m/z ) 28, N2 (26.2); m/z ) 32, O2 (99.6); m/z ) 34, O2 (100); m/z
+
+
+
Results
) 36, 36O2 (8.0); m/z ) 40, Ar (0.5); m/z ) 44, CO2 (1.3).
18O-Isotopic Labeling Studies. Use of closed reaction cells
configured as described in the Experimental Section allowed
rapid repetitive determination of the mass distributions of
gaseous products evolved from triflic acid solutions containing
gas formation was observed within the reaction cell, which was
reflected in a large immediate increase in the intensities at 32
and 34 amu, with a much smaller increase occurring at 36 amu.
Typically, data were collected for a period of time corresponding
to about one turnover of the catalyst, as determined by
measuring the rate of O2 formation with a Clark electrode. For
the reaction conditions given in Figure 1, the single turnover
time was ∼125 s. The slow decrease observed in the rate of
4+
the catalyst and the strong oxidizing Ce
ion. In these
experiments, [(bpy)2Ru(OH2)]2O4 ions that were isotopically
+
enriched in their cis-aqua positions were first oxidized to their
{
3,4} oxidation state while in 18O-enriched H2O to minimize
3
4
32
any subsequent water exchange upon addition to isotopically
dilute solvent.23 Then, following brief bubbling with He to
remove air, the reaction was initiated by syringe-addition of
formation of O2 relative to that of O2 over the course of the
16
reaction can therefore be attributed to substitution of H2 O at
the isotopically labeled cis-aqua position during cycling of the
catalyst.
4+
excess Ce , causing near-quantitative oxidation of the catalyst
to its {5,5} oxidation state. Throughout this procedure and
during the initial stages of the reaction, headspace gases were
scanned every 2.0 to 2.5 s over the range of 10-50 amu. The
recorded intensities at 32, 34, and 36 amu of a typical run are
displayed in Figure 1a. Prior to addition of the oxidant,
intensities at 34 and 36 amu levels are negligible. The intensity
at 32 amu is measurable, but unchanging; it represents the
Control experiments were performed in which Ce4 was
injected into identical solutions that contained no catalyst; in
these cases, no increase in intensities as 32, 34, and 36 amu
were detected. The mass intensity at 28 amu (corresponding to
N2) also underwent no significant change during the course of
injecting oxidant and subsequent O2 formation in the catalyzed
reaction. Barely detectable increases were observed in the mass
intensity at 44 amu (corresponding to natural abundance CO2),
+
4+
background level of O2 in the instrument. Upon addition of Ce ,
J. AM. CHEM. SOC.
9
VOL. 126, NO. 31, 2004 9789

A R T I C L E S
Yamada et al.
Table 1. Relative Contributions of Pathways Calculated from O2
a
Isotopic Distributions
[
Ru
2
O] (mM)
conditionsb
32O c
0.54 0.46 0.006 <0.01 0.49
0.49 0.495 0.015 0.01 0.44
2
34O c
2
36O c
2
class id class iie class iiif
0
0
0
0
0
0
0
0
0
0
.037
.10
.10
.25
.27
.29
.29
.80
.86
.74
28 °C
0.51
0.55
0.55
0.69
0.64
0.47
0.43
0.85
0.69
0.55
28 °C
28 °C
10 °C
28 °C
28 °C
47 °C
28 °C
ambient
ambient,
0.49 0.50 0.008 <0.01 0.45
0.37 0.62 0.011 <0.01 0.31
0.40 0.58 0.022
0.50 0.48 0.020
0.01 0.35
0.05 0.48
0.59 0.40 0.015 <0.01 0.57
0.21 0.71 0.080 <0.01 0.15
0.30 0.64 0.058 -0.01 0.32
0.36 0.60 0.037 <0.01 0.45
92% D2O
a
Calculated as indicated in the Experimental Section. b For all reactions,
+
c
[
Ce4 ] ) 25 mM, [CF3SO3H] ) 0.5 M. Fractional yields extrapolated to
d
e
t ) 0. Both O atoms from {5,5} terminal ruthenyl (RudO) groups.
f
Both O atoms from solvent H2O. One O atom from RudO and one O
Figure 2. Decay of 0.25 mM electrochemically prepared {5,5} at 13 °C
in 0.5 M CF3SO3H. The solid line is the exponential fit to the data with k
atom from H2O.
-
3
-1
)
1.42 × 10
s .
which appeared to become more pronounced with time of
exposure to Ce . A representative mass spectrum, taken 75 s
4
+
measured first-order rate constant in 0.5 M triflic acid at ambient
4+
-3 -1
following addition of Ce , is shown in Figure 1b. Collectively,
these results indicate that the observed increase in O2 mass
intensities following oxidant addition arise solely from catalyzed
water oxidation, that reagent manipulations introduce no detect-
able adventitious air, and that oxidative degradation of the
temperature (∼23 °C) was kH O ) 2.3 ((0.2) × 10
s
and
2
was invariant with reactant concentration over the range [{5,5}]
) 0.2-0.7 mM. The rate constant measured in these studies is
4-fold lower than the value we previously reported based upon
7
a more limited data set. The corresponding reaction in 0.5 M
catalyst is minimal despite being exposed to a large (30-fold to
2
triflic acid, 98.9% D2O, at ambient temperature was kD O ) 1.4
2
50-fold) excess of Ce4 ion.
+
-3 -1
((0.1) × 10 s , giving a solvent kinetic isotope effect (KIE)
of kH O/kD O ) 1.6.
Another property of the {5,5} ion revealed by these studies
2
2
As described in the Experimental Section, the rates of O2
evolution were measured by using a calibrated Clark electrode
suspended in the headspace of a closed reaction cell. Reaction
is that exchange between the coordinated ruthenyl oxygen and
solvent H2O must be slow relative to catalytic evolution of O2.
Otherwise, one would have observed a rapid change in the
4
+
32
34
was initiated by addition of a 100-fold excess Ce to a 0.5 M
relative rates of O2 and O2 formation following addition of
4+
triflic acid solution containing the {3,3} ion; this amount of
Ce as this process occurred in parallel with water oxidation.
In any event, the measured O2-isotope distributions were
corrected for possible minor changes that might have arisen by
O-atom exchange, as well as any influence of oxidative
modification of the catalyst, by the manner in which the data
were analyzed. Specifically, the relative intensities at 32, 34,
and 36 amu measured were plotted against time; in the early
stages of the reaction, that is, the first 20-40 mass spectral
scans, these changes were linear. Extrapolation to t ) 0, the
4
+
Ce is sufficient to quantitatively oxidize the complex to its
5,5} oxidation state. Under these conditions, initial rates of
O2 formation are linearly dependent upon the catalyst concentra-
{
7,9
tion. No O2 was detected when the catalyst was omitted from
the reaction medium. First-order rate constants (kcat) were
determined by dividing the measured evolution rates by the
amount of catalyst present in solution. In H2O containing 0.5
M triflic acid, the value obtained at 23 °C was kcat(H2O) ) 5.2
-
3 -1
point at which Ce4 was introduced, gave values for the isotope
distributions of O2 formed at the start of the reaction, a point at
which the isotopic distributions in the catalyst and solution were
accurately known. The extrapolated values for the various
reaction conditions investigated are listed in Table 1.
+
((1.0) × 10 s ; the corresponding value in 99.8% D O was
2
-
3 -1
kcat(D2O) ) 3.1 ((0.1) × 10 s , giving an apparent KIE for
catalyzed O2 evolution of 1.7.
The coordinated H2O molecules in the {3,4} oxidation state
of the µ-oxo ion are strongly acidic and undergo sequential one
+
and two-H deprotonations, that is:
Solvent Deuterium Isotope Effects. We have previously
estimated the kinetics of decay of electrochemically prepared
7
{
5,5} ion using both RR and optical spectrometric detection.
These methods are experimentally challenging because, in the
former case, the {5,5} ion undergoes photoreduction in the
7
relatively intense laser beam used to obtain the RR spectra and,
in the latter case, the optical spectra of the {5,5} and {4,4}
ions are very similar and decay of the {4,4} ion to {3,4} is
sufficiently rapid to interfere with the {5,5} f {4,4} transfor-
mation (Scheme 1). In reinvestigating the {5,5} decay kinetics,
we have obtained reproducible data by monitoring the relatively
small optical changes at the isosbestic points for the {3,4} and
{
4,4} visible bands, which were at 470 nm in H2O and 468 nm
with reported ionization constants of pK1a = 0.4 and pK2a =
3.2 at room temperature. In 99% D2O, the fully protonated
1
in D2O under the prevailing medium conditions. A representative
kinetic profile for {5,5} decay is given in Figure 2. The
{3,4} ion exhibits a symmetric visible absorption band at 442
9790 J. AM. CHEM. SOC.
9
VOL. 126, NO. 31, 2004

Mechanisms of Water Oxidation
A R T I C L E S
Figure 3. Normalized resonance Raman spectra of 1 mM {3,3} in 0.5 M
CF3SO3H, H2O (solid line), and 99.7% D2O (dashed line). Spectra are the
average of 10 scans with 10 s accumulation times at 647 nm. The intense
Figure 4. Temperature dependence of first-order rate constants for O2
evolution (0) and {5,5} decay (O). Conditions: 100 µM {3,3} plus 11
4+
mM Ce (O2 evolution) or 0.25 mM {5,5} (decay) in 0.5 M CF3SO3H.
-
1
band at ∼370 cm is attributable to the Ru-O-Ru symmetric stretching
Solid lines are linear least-squares fits to the data.
7
mode (νs(Ru-O-Ru)). The solvent isotope-dependent shift is evident in
the H2O-D2O difference spectrum (inset).
nm; a corresponding symmetric band appears at 494 nm in the
conjugate base. Spectrophotometric titration with 0.05-3.47 M
triflic acid in D2O gave a value for pK1a(D2O) = 0.69. Thus,
the equilibrium isotope ratio for the first ionization constant is
K1a(H2O)/K1a(D2O) = 1.9.
The symmetric stretching frequency (νs) of the bridging Ru-
O-Ru unit in the {3,3} and {3,4} ions appeared at slightly lower
-1
energies (∆ν ) 2-5 cm ) in the room-temperature RR spectra
of acidic D2O solutions than in the corresponding spectra in
H2O. Representative spectra are given in Figure 3. These results
are very similar to data reported earlier on frozen solutions at
9
0 K.13 Confirmation of this effect in the liquid phase is
important, however, because the solvent-induced shifts are small
and the low temperatures and long acquisition times required
to obtain high quality spectra on the instrumentation used in
the earlier studies could have introduced artifacts associated with
imposition of the solid matrix or photodecomposition in the
exciting beam.
Figure 5. Normalized resonance Raman spectra of 0.18 mM {3,4} in 0.5
M CF3SO3H. Spectra are averages of 10 6-s accumulations taken with 40
Temperature Dependence of O2 Evolution and {5,5}
Decay. The values for kcat increased progressively with tem-
1
8
mW excitation at 488 nm. Solid lines: spectra of {3,4} containing O in
4
+
the µ-oxo bridge before and after complete reaction with 7.2 mM Ce
;
-
3
-1
-2 -1
perature from 3.7 × 10
s
to 4.5 × 10
s
over the
16
dashed line: spectra of {3,4} containing O in the µ-oxo bridge. The intense
-
1
18
-1
temperature range of 10-60 °C. A plot of ln(kcat/T) vs T
Figure 4) was nearly linear and gave ∆H ) 7.6 ((1.2) kcal/
mol and ∆S ) -43 ((4) cal/deg mol (23 °C) as apparent
activation parameters. Direct optical spectrophotometric mea-
surement of the decay of electrochemically prepared {5,5} to
O-isotope dependent band at ∼390 cm is attributable to (νs(Ru-O-
-
1
q
Ru)), and the weaker band at ∼780 cm is attributable to its first overtone
7 18 16
(
(
2νs(Ru-O-Ru)). The inset shows the O- O difference spectrum (solid
q
18
line) and the difference spectrum obtained for the O-substituted ion before
and after 10 cycles of the catalyst (dashed line).
-
1
{
4,4} over the temperature range of 13-41 °C gave similar
As shown in Figure 5, this band appears at 389 cm in the
q
q
16
-1
values of ∆H ) 7.9 ((1.1) kcal/mol and ∆S ) -44 ((4)
solution spectra when the bridge contains O and at 382 cm
1
8
18
16
cal/deg mol (Figure 4).
when the bridge contains O; the large O- O difference
spectrum provides a sensitive measure of isotopic substitution
at that position. However, no difference peaks appear in the
spectra of the {3,4} ion recorded before and after >10 catalytic
Discussion
Relative Contributions of the Pathways. The mechanistic
model used here to analyze the measured isotopic distributions
of O2 is predicated upon the assumption that exchange does
not occur in the µ-oxo bridge during catalytic turnover. We have
established this point in previous studies by examining the
position of the prominent Ru-O-Ru symmetric stretching
3
+
4+
cycles of water oxidation using either Co or Ce as oxidants
7
,13
(Figure 5).
This clearly demonstrates the absence of O-
substitution within the bridge following repetitive cycling of
the catalyst, not only justifying the analytical model but also
allowing one to eliminate from consideration pathways such as
those shown in Scheme 2, pathway B, that involve formation
7
,13
mode (νs(Ru-O-Ru)) in the RR spectrum of the {3,4} ion.
J. AM. CHEM. SOC.
9
VOL. 126, NO. 31, 2004 9791

A R T I C L E S
Yamada et al.
of symmetric di-µ-oxo bridged intermediates within a single
complex ion.25 Specifically, if this were the mechanism, one
would predict substantial substitution within the bridge after
only a single turnover. These experiments do not exclude similar
di-µ-oxo-bridged intermediates that might be formed by bridging
between terminal cis-oxo ligands of two complex ions, as
recently discussed by Meyer and associates.1 However, as
discussed in the succeeding paragraph, pathways of the latter
type are effectively excluded by the isotope distribution patterns
observed in these studies.
Scheme 3. Kinetic Models for First-Order Decay of the {5,5} Ion
2
Scheme 4. Hypothetical Intermediates for O-O Bond Formation
by the kiii Pathway
Listed in Table 1 are the results of calculations that allow
one to distinguish between three classes of reaction pathways
based upon the molecular origins of O atoms in the O2 products,
namely: (i) both atoms derived from the ruthenyl oxo atoms,
(ii) both atoms derived from solvent H2O, or (iii) one atom from
each source. The results of these calculations (Table 1) clearly
indicate that contribution of pathways of class i are negligible
under all reaction conditions investigated. The small amount
of O2 formation and {5,5} decay are independent of how the
intermediate partitions. For the latter model, the observed
temperature dependence of the isotopic yields should also appear
in the overall rate of O2 formation, which is equal to the sum
of the decay rates by the various pathways. This model could
be excluded if the activation parameters for the two steps
predicted significant curvature in the Eyring plots. The differ-
3
6
of O2 that is formed is attributable almost entirely to reaction
18
4+
18
between [(bpy)2Ru( O)]2O and H2 O by class iii pathways.
Consequently, it can be concluded that all unimolecular and
bimolecular reactions that involve direct O-O bond formation
between cis-ruthenyl oxygens are negligible and can be excluded
from further consideration. This includes pathways A, B, and
D shown in Scheme 2, as well as additional similar pathways
discussed by Meyer and co-workers.12 Furthermore, relative
contributions of the other two pathways are large and demon-
strably temperature-dependent (Table 1), indicating that solvent
is an obligatory participant in O2 formation that acts by at least
two distinct mechanisms.26 These results are totally consistent
with earlier data that we reported for reactions of the same
32
ences in activation parameters between reactions forming O2
34
and O2 calculated from the data in Table 1 assuming this model
q
q
are ∆(∆H ) = 5.3 kcal/mol and ∆(∆S ) = 17 eu. Combining
these data with the activation parameters obtained from Figure
q
q
4
-
∆
, one calculates values of ∆H = 10.2 kcal/mol and ∆S =
q
34 eu for the class ii pathway and ∆H = 5.0 kcal/mol and
S = -51 eu for the class iii pathway. Eyring plots made
q
using these parameters showed the expected curvature, but the
magnitude of deviation from linearity was within the uncertainty
of the experimental data. Thus, the temperature dependence of
rates and relative yields cannot be used in this instance to
distinguish between the two reaction models.
Reaction Mechanisms. Class iii Pathway. We have previ-
ously suggested that this pathway involves nucleophilic attack
of H2O upon an electrophilic ruthenyl O atom, as illustrated by
the structure in Scheme 4. Hydrogen bonding to the adjacent
bridging O atom may facilitate O-O bond formation both by
properly orienting and increasing the nucleophilicity of the water
molecule. The solvent dependence of νs(Ru-O-Ru) in the
complex ion with Co3 at ambient temperature, but differ from
+
13
data reported by Geselowitz and Meyer, who found substantially
3
6
4+
higher relative yields of O2 in reactions carried out using Ce
2
7
as oxidant. The basis for these differences is uncertain, but
13,27
the earlier data
were considerably less accurate because they
involved relatively infrequent bolus addition of accumulated
gases into the mass spectrometer.
Temperature Dependence and Reaction Models. The
existence of at least two reaction pathways is suggested from
the temperature dependence of the O2 isotopic distributions
(Table 1). In contrast, Eyring plots for catalyzed rates of O2
evoluton and {5,5} decay are linear within experimental
uncertainty (Figure 4). This temperature dependence is consistent
with two simple reaction models, in which either a common
intermediate is formed in the rate-limiting step or there are two
independent concurrent pathways for reaction between {5,5}
{
3,3} and {3,4} ions (Figure 3) provides indirect evidence for
this type of interaction. Although the D2O-induced shifts are
small, they are comparable to shifts measured in biological
M-O-M centers, for example, the Fe-O-Fe unit in hem-
2
8
erythrin, where H-bonding to the µ-oxo bridge has been
q
and solvent whose activation enthalpies (∆H ) are fortuitously
2
9
confirmed by X-ray crystallography.
similar (Scheme 3). For the former model, the relative yields
of isotopic products are governed by the rates of decomposition
of the intermediate by the two pathways, but the overall rates
Experimental data obtained for oxidation of O-H and C-H
3
0
31
32,33
bonds in peroxides, hydroquinones, alcohols,
and
34,35
arenes by monomeric ruthenyl polypyridyl compounds (e.g.,
(bpy)2(py)Ru O] ) strongly suggest that these reactions
IV
2+
[
(
(
(
25) Mechanisms of this type have been previously suggested within the context
of biological water oxidation, based primarily upon the observation that
2
2
certain di-µ-oxo copper complexes transform reversibly into µ-η -η -bound
molecule upon changing the solvent composition of the medium (Halfen,
(28) Shiemke, A. D.; Loehr, T. J.; Sanders-Loehr, J. J. Am. Chem. Soc. 1986,
108, 2437-2443.
(29) Stenkamp, R. E. Chem. ReV. 1994, 94, 715-726.
(30) Gilbert, J.; Roecker, L.: Meyer, T. J. Inorg. Chem. 1987, 26, 1126-1132.
(31) Binstead, R. A.; McGuire, M. E.; Dovletoglou, A.; Seok, W. K.; Roecker,
L. E.; Meyer, T. J. J. Am. Chem. Soc. 1992, 114, 173-186.
(32) Thompson, M. S.; Meyer, T. J. J. Am. Chem. Soc. 1982, 104, 4106-4115.
(33) Roecker, L.; Meyer, T. J. J. Am. Chem. Soc. 1987, 109, 746-754.
(34) Bryant, J. R.; Mayer, J. M. J. Am. Chem. Soc. 2003, 125, 10351-10361.
(35) Bryant, J. R.; Matsuo, T.; Mayer, J. M. Inorg. Chem. 2004, 43, 1587-
1592.
O
2
J. A.; Mahapatra, S.; Wilkinson, E. C.; Kaderli, S.; Young, V. G., Jr.; Que,
L.; Zuberbhhler, A. D.; Tolman, W. B. Science 1996, 271, 1397-1400).
26) It is perhaps useful to note that these conclusions were already evident
from the raw data; that is, before the analyses described in the Experimental
Section were performed. Since the complex was ∼90% enriched in cis-
1
8
18
[
O]H
2 2
O and the solvent typically contained <3% [ O]H O, the measured
36 32 34
relative yields of
2 2 2
O , O , and O must reflect fairly closely the relative
contributions of pathways i-iii, respectively.
27) Geselowitz, D. A.; Meyer, T. J. Inorg. Chem. 1990, 29, 3894-3896.
9792 J. AM. CHEM. SOC.
9
VOL. 126, NO. 31, 2004

Mechanisms of Water Oxidation
A R T I C L E S
proceed by hydride33 or hydrogen atom abstraction
(
30,33,34,36
proportional to the {5,5} concentration (unpublished observa-
tions). Although the signal is as yet unidentified, its spectro-
scopic features are consistent with those expected for a ligand
radical cation, suggesting the possible existence of an intra-
molecular redox equilibrium of the type:
31
alternatively, proton-coupled electron transfer ) mechanisms.
The unusually strong O-H bond in H2O poses severe energetic
constraints on these types of mechanisms for water oxidation
by the {5,5} ion. If a similar mechanism is operative, then
H-atom abstraction must also be coupled to formation of a
peroxo O-O bond because intermediary formation of a discrete
hydroxyl radical, even if H-bonded to the µ-oxo atom, is
energetically prohibitive.37 Direct formation of a peroxo-bound
3
8
+
H + {5,5} a
IV
V
5+
[
(bpy)(bpyπ•+)Ru (OH)-O-Ru (O)(bpy) ]
2
{
4,4} ion by a concerted mechanism such as that shown in
in which the {5,5} ion is the predominant species. The π-cation
Scheme 4 may also be slightly endergonic. From the relevant
two-electron potentials, one calculates ∆E° ) -0.19 V for the
reaction, {5,5} + 2H2O f {4,4}+ H2O2; the analogous reaction
to form a peroxo anion stabilized by coordination to a highly
charged ruthenium center in the {4,4} ion should have similar
thermodynamic values. The thermodynamic driving force for
the overall reaction, 2{5,5} + 2H2O f 2{4,4} + O2, is large,
with ∆E° ) 0.71 V; for this mechanism, the overall reaction
would therefore be driven to completion by the highly exergonic
reaction between the intermediate or H2O2 and an additional
radical is presumably stabilized by protonation of the proximal
terminal ruthenyl oxygen and by the -O-Ru (O)(bpy)2 unit,
V
which, at least in lower oxidation states, appears by several
1,23,39,40
criteria
to be strongly σ-donating relative to simple
monomeric (bpy)2RuX2 analogues and might thereby signifi-
cantly lower the HOMO of coordinated bipyridine. The “push-
pull” asymmetry generated by electronic redistribution within
this complex could enhance reactivity by strengthening the
H-bond between the µ-oxo atom and the reacting solvent
molecule (Scheme 4), thereby increasing its nucleophilicity
without proportionately decreasing the electrophilicity of an
{
5,5} ion.
V
A more serious mechanistic issue arises from the absence of
adjacent Ru terminal oxo atom. Additionally, the higher
an appreciable KIE when the reactions were run in D2O. On
basis of the nature of the postulated reaction intermediate, one
would expect relatively large KIEs resulting from substantial
stretching of the O-H bond in the transition state if the reaction
involved rate-limiting bimolecular interaction between H2O and
the {5,5} ion. Indeed, one characteristic feature of the oxidation
reactions of the monomeric ruthenyl analogues, for example,
reduction potential of the relatively unstable {4,5}-bipyridyl
π-cation radical could be crucial in lowering the thermodynamic
barrier to forming a peroxide-bound {4,4} intermediate.
Class ii Pathway. The existence of a pathway by which O2
is formed catalytically by two solvent molecules was already
1
8
13,27
evident from earlier O-labeling experiments.
Devising a
plausible mechanism to account for this reaction, however, has
been challenging. This task is not diminished by the recognition
IV
2+
[
(bpy)2(py)Ru O] , is their pronounced primary KIEs, which
30-34
23
typically exceed values of 20.
In contrast, the apparent KIE
from this (Figure 1) and other recent studies that the catalyst
for reaction of the {5,5} ion was only 1.6. This value includes
contributions from both class ii and class iii pathways. The very
small change in the 32O2/ O2 distribution measured in D2O
under these conditions (Table 1) indicates that the pathways
leading to formation of each of these products have similar KIEs.
If the reaction between solvent and {5,5} is direct (Scheme 3),
it follows from the fact that the overall rate of O2 formation is
the sum of the concurrent pathways that neither pathway can
have a large KIE. For this reaction model, one calculates from
the data (Table 1) that KIE = 2.0 for the class iii pathway and
KIE = 1.1 for the class ii pathway. If the reaction proceeds by
rate-limiting formation of a common intermediate, one can
conclude from the solvent isotope dependence on the O2/ O2
distribution only that the two pathways for O2 formation have
very similar KIEs. The challenge then remains to identify an
intermediate whose rate of formation is not solvent isotope-
sensitive.
does not undergo aqua ligand exchange with solvent during
catalysis. We had previously suggested that reactive intermedi-
ates with hydroperoxide anions bound to each of the Ru atoms
13
34
might form, for example, by addition of a second H O molecule
2
to the covalent hydrate in Scheme 4 to give a {3,3}-dihydro-
peroxy complex, as in, for example:
4
+
[(bpy) Ru(OOH)ORu(OH)(bpy) ] + H O f
2
2
2
2+
+
[
(bpy) Ru(OOH)] O + 2H (1)
2 2
Subsequent intramolecular rearrangement to yield O2 or H2O2
might proceed via a bridging tetroxide intermediate. Meyer has
3
2
34
1
4
suggested a similar possibility involving a bridging ozonide
2
-
dianion (O3 ) that could conceivably also be formed by
rearrangement of the covalent hydrate shown in Scheme 4 (e.g.,
by analogy to the formation of hydrogen trioxide (H2O3) from
•
•
41-43
the combination of perhydroxyl (HO2 ) and hydroxyl (OH )).
One candidate for a common reactive intermediate is an EPR-
detectable species that appears at 80 K as a minor component
in both chemically and electrochemically prepared solutions of
the {5,5} ion.6 The intensity of this cryogenic signal is
Subsequent attack by solvent at the central O atom could then
give O2 in which both O atoms are obtained from solvent.
14
These intermediates are plausible in the sense that H2O3 is a
,7
44-47
well-established chemical entity
and that experimental
(
36) Mayer, J. M. Acc. Chem. Res. 1998, 31, 441-450.
(38) See, for example: (a) DeArmond, M. K.; Hanck, K. W.; Wertz, D. W.
Coord. Chem. ReV. 1985, 64, 65-81. (b) Kaim, W. Coord. Chem. ReV.
1987, 76, 187-235.
(
37) A lower limit on the one-electron potential for {5,5} reduction can be
estimated from the measured {5,5} f {4,4} potential (E° {5,5}/{4,4} )
1
.59 V) and the apparent lack of {4,5} formation during redox titrations
(39) Dobson, J. C.; Sullivan, B. P.; Doppelt, P.; Meyer, T. J. Inorg. Chem. 1988,
27, 3863-3866.
7
of the µ-oxo ion. Assuming that accumulation of {4,5} is less than 10%
of the total dimer concentration, the comproportionation constant (K ) for
j 10 , from which one
c
(40) Doppelt, P.; Meyer, T. J. Inorg. Chem. 1987, 26, 2027-2034.
(41) Czapski, G.; Bielski, B. H. J. J. Phys. Chem. 1963, 67, 2180-2184.
(42) Bielski, B. H. J.; Schwarz, H. A. J. Phys. Chem. 1968, 72, 3836-3841.
(43) Bielski, B. H. J. J. Phys. Chem. 1968, 72, 3836-3841.
(44) Deglise, X.; Gigu e` re, P. A. Can. J. Chem. 1971, 49, 2242-2247.
(45) Cerkovnik, J.; Plesni cˇ ar, B. J. Am. Chem. Soc. 1993, 115, 12169-12170.
-2
the reaction, {5,5} + {4,4} h 2{4,5} is K
c
•
-
calculates E° {5,5}/{4,5} j 1.5 V. Using E°(OH /OH ) ) 2.02 V (CRC
Handbook of Chemistry and Physics, 83rd ed.; Lide, D. R., Ed.; CRC
Press: Boca Raton, FL, 2002-2003), it follows that ∆E° j -1.3 V for
•
the reaction {5,5} + H
2
O f {4,5} + OH .
J. AM. CHEM. SOC.
9
VOL. 126, NO. 31, 2004 9793

A R T I C L E S
Yamada et al.
evidence also exists for the formation of hydrogen tetroxide
it is difficult to rationalize by this type of mechanism why
monomeric analogues of the type, cis-L2Ru(OH2)2 , and
44
2+
(H2O4); the hypothetical oxo-polyoxo dibridged intermediates
represent ruthenium-coordinated analogues of the hydrogen
polyoxides. Both polyoxides are unstable; unimolecular decom-
position of H2O3 gives O2 ( ∆) as a product, and oxidized water
dinuclear ions not containing the (H2O)RuORu(OH2) core, for
4
+
example, [(bpy)2(py)RuORu(H2O)(bpy)2] , appear incapable
1
40,54
of catalyzing water oxidation.
“Spontaneous” oxidation of Ru(bpy)33 and similar strongly
+
species are also produced upon decomposition of the less well-
characterized H2O4. Examination of structural models suggests
that there are no large steric restrictions to formation of mixed
dibridged complexes of the catalyst containing the analogous
dianions. However, reactions involving formation of either
oxidizing group 8 polypyridyl complexes in alkaline solutions
1
5-18
has been reported by several groups.
In general, these
reactions involve predominantly oxidation of the bipyridine
ligands, accompanied by copious evolution of CO and the
2
7
intermediate may be untenable on energetic grounds. Specif-
appearance of numerous ruthenium(II)-containing products
ically, the reaction {4,4} + H2O2 + H2O f {3,3} + H2O3 can
be used as a rough guide for the energetics of the reaction
bearing modified ligands. However, Ledney and Dutta have
3
+
reported that, when the Ru(bpy)3 ions are entrapped within
zeolite cages, complex decomposition is minimal and water
4
+
[
(bpy)2Ru(OOH)ORu(OH)(bpy)2] f [((bpy)2Ru2)(µ-O)(µ-1,3-
2+
+
18
O3)] + 2H ; on the basis of a recent estimate for the enthalpy
oxidation becomes particularly efficient. These researchers
4
8
of formation for H2O3 of ∆Hf° ) -23 kcal/mol, with the
assumption that entropy contributions approximately cancel, one
calculates ∆E° ∼ -0.7 V for the former reaction. Similarly,
one obtains ∆E° ∼ -0.5 V as an estimate of the energetics of
have argued that this occlusion of the catalyst effectively blocks
the bimolecular ligand decomposition pathways which dominate
the chemistry in homogeneous solution, allowing expression of
the water oxidation pathway. For the same reason, these
reactions apparently do not require formation of the µ-oxo dimer
(or other multinuclear complex ion intermediates) that have been
7
reaction 1 using previously determined redox potentials for the
reaction {4,4} + 2H2O f {3,3} + H2O2. Formation of both
dibridged intermediates should be considerably less favorable
than these estimates because in both cases they require release
of protons to the highly acidic medium and are accompanied
by reduction of the overall electrostatic charge on the complex
from 4+ to 2+, thereby decreasing coordinate bond stabilization.
Consistent with this view, recent ab initio calculations suggest
that pathways for unimolecular decomposition of H2O3 involve
either acid catalysis or participation of solvent as a bifunctional
-
proposed to be the actual catalysts of OH oxidation by
3
+ 55
Ru(bpy)3
.
The mechanism proposed by Ledney and Dutta
1
8
-
for this reaction involves initial nucleophilic addition of OH
to an electrophilic carbon atom on one of the bipyridine rings,
forming a relatively stable ligand-centered radical complex of
Ru(II). Subsequent one-electron oxidation of the metal center
yields a Ru(III)-ligand radical species that is primed for reaction
-
2+
with a second OH to give H O and Ru(bpy)
3
2
2
as final
4
6,49-51
catalyst in concerted proton donor-acceptor reactions;
products. Support for this reaction model includes the kinetics
2
+
3+
in these studies, no low energy pathways involving nucleophilic
attack of the O atom of a water molecule on the central O atom
of H2O3 have been identified. If analogous mechanisms are
applied to decomposition of the hypothetical [((bpy)2Ru)2(µ-
of Ru(bpy) formation, which are first-order in [Ru(bpy)3
]
3
-
and [OH ] under most reaction conditions in homogeneous
solution, as well as demonstrations that Ru(bpy)3 and similar
M(III) group 8 complexes readily undergo analogous nucleo-
15
3+
O)(µ-1,3-O3)]2 intermediate, O2, rather than O2, would be
the isotopic product of the 18O-labeled ruthenium complexes.
Thus, these putative low energy pathways cannot account for
the appearance of 32O2 in the product gases (Figure 1). The
activation barrier for reaction 1 must also be relatively high
because the reactant ruthenium complex contains no ruthenyl
group to initiate H-atom abstraction from H2O.
+
34
32
philic addition reactions to give ring-modified polypyridine
19,21
ligands,
2
among which is apparent addition of H O to give
2
2
covalent hydrates, and that reaction between radiolytically
generated OH and Ru(bpy) or various M(bpy) ions gave
products with spectroscopic signatures that are very similar
to optically detectable intermediates appearing in the Ru(bpy)3
•
3+
2+
3
3
1
5
3+
-
-
OH reaction.
Formation of O2 from two H2O’s might involve expansion
of a Ru(V) coordination sphere by addition of H2O, which then
undergoes subsequent reaction with a second solvent molecule.
Some support for this notion is obtained from the kinetics of
water exchange on monomeric Ru(III) centers, which suggest
A bipyridine ligand-based mechanism for class ii water
oxidation by {5,5} that incorporates key features of the
3+
Ru(bpy)3 reactions is given in Scheme 5. The reaction is
initiated by addition of the elements of water to form a covalent
-
+
hydrate in which OH is added to a bipyridine ligand and H
5
2
that an associative interchange mechanism is operative, and
discussions that encompass a wider chemical literature of seven-
protonates a terminal ruthenyl O atom. This reaction would also
be considerably facilitated by internal electronic rearrangement
5
3
coordinate species. However, if this mechanism were opera-
tive, one might expect that the intermediate would also provide
a pathway for facile exchange of the cis-ruthenyl oxygen atoms
in {5,5} with solvent H2O, which is not observed. Furthermore,
•+
that generated a {4,5}-bpy radical cation. Alternatively,
covalent hydrate formation might be promoted by H-atom
abstraction at a RudO center in a concerted reaction similar to
that proposed for the class iii reaction (Scheme 4). In this case,
the OH segment is positioned to add to the ring six-position to
form a {4,5} intermediate containing a neutral ligand radical
(
46) Plesni cˇ ar, B.; Tuttle, T.; Cerkovnik, J.; Koller, J.; Cremer, D. J. Am. Chem.
Soc. 2003, 125, 11553-11564.
(
(
(
(
(
47) Engdahl, E.; Nelander, B. Science 2002, 295, 482-483.
(
Scheme 5). In this manner, the reaction could avoid intermedi-
48) Lay, T. H.; Bozzelli, J. W. J. Phys. Chem. A 1997, 101, 9505-9510.
49) Koller, J.; Plesni cˇ ar, B. J. Am. Chem. Soc. 1996, 118, 2470-2472.
50) Mckay, D. J.; Wright, J. S. J. Am. Chem. Soc. 1998, 120, 1003-1013.
51) Xu, X.; Muller, R. P.; Goddard, W. A., III. Proc. Natl. Acad. Sci. U.S.A.
•
+
ary formation of the {4,5}-bpy radical cation. The neutral
radical also represents an alternative candidate for the cryogenic
EPR signal detected in frozen solutions of the {5,5} ion.6
,7,18
2
002, 99, 3376-3381.
(
52) Rappaport, I.; Helm, L.; Merbach, A. E.; Bernhard, P.; Lundi, A. Inorg.
Chem. 1988, 27, 873-879.
(53) Serpone, N.; Ponterini, G.; Jamieson, M. A.; Bolletta, F.; Maestra, M. Coord.
Chem. ReV. 1983, 50, 209-302.
(54) Collin, J. P.; Sauvage, J. P. Inorg. Chem. 1986, 25, 135-141.
(55) Lay, P. A.; Sasse, W. H. F. Inorg. Chem. 1985, 24, 4707-4710.
9794 J. AM. CHEM. SOC.
9
VOL. 126, NO. 31, 2004

Mechanisms of Water Oxidation
A R T I C L E S
Scheme 5. Hypothetical Intermediates for O-O Bond Formation
by the kii Pathway
{3,3} ion. In this context, one should note that identification of
the actual site(s) of OH attachment to the ring is not intended
in Scheme 5. Addition at other sites might well occur and could
significantly affect catalysis. For example, addition at the
heterocyclic N or bridging C positions would be expected to
minimize competing pathways leading to oxidative degradation
of the ligand.
Additional Comments. The mechanism presented here is a
minimal scheme that best accommodates the existing data and
is subject to refinement as detectable intermediates become
better characterized. For example, the rate constants for first-
order decay of {5,5} measured by optical spectroscopy vary
by as much as 4-fold under different experimental conditions
(cf., our reported values at 23 °C in this work and ref 7), which
might ultimately be attributable to the appearance of strongly
absorbing intermediates. Furthermore, the mass spectral kinetic
profiles obtained immediately following addition of Ce4
suggest that there is a very short induction period prior to
+
3
2
34
formation of O2 that is not observed for O2. This behavior
is consistent with accumulation of a reactive intermediate in
the class ii pathway that is not required for the class iii pathway,
a circumstance that would require expansion of the simple
Specifically, internal ligand-to-metal electron transfer could be
fairly slow, as has been demonstrated for the [(NH3)5Copy-
•
3+
56
OH ] radical adduct, allowing accumulation of the ligand
radical to EPR-detectable levels during catalyst turnover.
Subsequent addition of a second H2O molecule to the modified
bipyridine ring to form a {4,4}-coordinated bipyridine diol
34
mechanisms given in Scheme 4. Similarly, the increase in O2/
O2 ratio observed at the highest {5,5} concentrations (Table
, data for 28 °C) might be indicative of minor contribution
from a bimolecular pathway involving, for instance, reaction
between an intermediate species and {5,5}. Identification of the
anomalous signal in the low-temperature EPR spectra of the
complexes, currently in progress, is expected to elucidate many
of these mechanistic issues.
3
2
1
(
Scheme 5) is suggested by analogy to similar reactions
3+
-
proposed in the overall reactions between Ru(bpy)3 and OH
that lead to ligand decomposition.17 This reaction would
be facilitated in the dinuclear µ-oxo ion by the pendant
+
[
(bpy)2Ru(dO)O] moiety, which can function as an electron
II
sink for the unpaired electron. Unlike monomeric Ru (bpy)2-
bpy(OH)2) analogues, the bipyridine diol in the dinuclear ion
2+
(
Acknowledgment. We thank Joseph T. Hupp (Northwestern
University) and Carl C. Wamser (Portland State University) for
helpful discussions and Pierre Mo e¨ nne-Loccoz (OGI School of
Science & Engineering, Oregon Health & Science University,
Portland, OR) for technical assistance in acquiring the resonance
Raman spectra. This work is supported financially by the
Division of Chemical Sciences, Office of Basic Energy Sciences,
U.S. Department of Energy, under Grant DE-FG03-99ER14943.
is coordinated to a two-electron oxidizing center which is nearly
as strongly oxidizing as the original {5,5} ion. Additional
internal electron transfer could drive oxidation of the OH
5
7
substituents to form, for example, an unstable endoperoxide
that undergoes further bond rearrangement to give O2 and the
(
56) Hoffman, M. Z.; Kimmel, D. W.; Simic, M. G. Inorg. Chem. 1979, 18,
479-2485.
57) Wasserman, H. H.; Scheffer, J. R. J. Am. Chem. Soc. 1967, 89, 3073-
075.
2
(
3
JA030594G
J. AM. CHEM. SOC.
9
VOL. 126, NO. 31, 2004 9795