Electronic modification of the [RuII(tpy)(bpy)(OH 2)]2+ scaffold: Effects on catalytic water oxidation

Article abstract of DOI:10.1021/ja106108y

The mechanistic details of the Ce(IV)-driven oxidation of water mediated by a series of structurally related catalysts formulated as [Ru(tpy)(L)(OH ₂)]²⁺ [L = 2,2′-bipyridine (bpy), 1; 4,4′-dimethoxy-2,2′-bipyridine (bpy-OMe), 2; 4,4′-dicarboxy-2, 2′-bipyridine (bpy-CO₂H), 3; tpy = 2,2′6″,2″- terpyridine] is reported. Cyclic voltammetry shows that each of these complexes undergo three successive (proton-coupled) electron-transfer reactions to generate the [Ru^V(tpy)(L)O]³⁺ ([Ru^V=O] ³⁺) motif; the relative positions of each of these redox couples reflects the nature of the electron-donating or withdrawing character of the substituents on the bpy ligands. The first two (proton-coupled) electron-transfer reaction steps (k₁ and k₂) were determined by stopped-flow spectroscopic techniques to be faster for 3 than 1 and 2. The addition of one (or more) equivalents of the terminal electron-acceptor, (NH₄)₂[Ce(NO₃)₆] (CAN), to the [Ru^IV(tpy)(L)O]²⁺ ([Ru^IV=O] ²⁺) forms of each of the catalysts, however, leads to divergent reaction pathways. The addition of 1 eq of CAN to the [Ru^IV=O] ²⁺ form of 2 generates [Ru^V=O]³⁺ (k₃ = 3.7 M^-1 s^-1), which, in turn, undergoes slow O-O bond formation with the substrate (k_O-O = 3 × 10^-5 s ^-1). The minimal (or negligible) thermodynamic driving force for the reaction between the [Ru^IV=O]²⁺ form of 1 or 3 and 1 eq of CAN results in slow reactivity, but the rate-determining step is assigned as the liberation of dioxygen from the [Ru^IV-OO]²⁺ level under catalytic conditions for each complex. Complex 2, however, passes through the [Ru^V-OO]³⁺ level prior to the rapid loss of dioxygen. Evidence for a competing reaction pathway is provided for 3, where the [Ru ^V=O]³⁺ and [Ru^III-OH]²⁺ redox levels can be generated by disproportionation of the [Ru^IV=O]²⁺ form of the catalyst (k_d = 1.2 M^-1 s^-1). An auxiliary reaction pathway involving the abstraction of an O-atom from CAN is also implicated during catalysis. The variability of reactivity for 1-3, including the position of the RDS and potential for O-atom transfer from the terminal oxidant, is confirmed to be intimately sensitive to electron density at the metal site through extensive kinetic and isotopic labeling experiments. This study outlines the need to strike a balance between the reactivity of the [Ru=O]^z unit and the accessibility of higher redox levels in pursuit of robust and reactive water oxidation catalysts.

Full text of DOI:10.1021/ja106108y

Published on Web 10/26/2010
Electronic Modification of the [RuII(tpy)(bpy)(OH2)]2+ Scaffold:
Effects on Catalytic Water Oxidation
Derek J. Wasylenko, Chelladurai Ganesamoorthy, Matthew A. Henderson,
Bryan D. Koivisto, Hans D. Osthoff, and Curtis P. Berlinguette*
Department of Chemistry and Institute for Sustainable Energy, EnVironment & Economy,
UniVersity of Calgary, 2500 UniVersity DriVe N.W., Calgary, Canada T2N-1N4
Received July 9, 2010; E-mail: cberling@ucalgary.ca
Abstract: The mechanistic details of the Ce(IV)-driven oxidation of water mediated by a series of structurally
related catalysts formulated as [Ru(tpy)(L)(OH2)]2+ [L ) 2,2′-bipyridine (bpy), 1; 4,4′-dimethoxy-2,2′-bipyridine
(bpy-OMe), 2; 4,4′-dicarboxy-2,2′-bipyridine (bpy-CO2H), 3; tpy ) 2,2′;6′′,2′′-terpyridine] is reported. Cyclic
voltammetry shows that each of these complexes undergo three successive (proton-coupled) electron-
transfer reactions to generate the [RuV(tpy)(L)O]3+ ([RuV)O]3+) motif; the relative positions of each of these
redox couples reflects the nature of the electron-donating or withdrawing character of the substituents on
the bpy ligands. The first two (proton-coupled) electron-transfer reaction steps (k1 and k2) were determined
by stopped-flow spectroscopic techniques to be faster for 3 than 1 and 2. The addition of one (or more)
equivalents of the terminal electron-acceptor, (NH4)2[Ce(NO3)6] (CAN), to the [RuIV(tpy)(L)O]2+ ([RuIV)O]2+
)
forms of each of the catalysts, however, leads to divergent reaction pathways. The addition of 1 eq of CAN
to the [RuIV)O]2+ form of 2 generates [RuV)O]3+ (k3 ) 3.7 M-1 s-1), which, in turn, undergoes slow O-O
bond formation with the substrate (kO-O ) 3 × 10-5 s-1). The minimal (or negligible) thermodynamic driving
force for the reaction between the [RuIV)O]2+ form of 1 or 3 and 1 eq of CAN results in slow reactivity, but
the rate-determining step is assigned as the liberation of dioxygen from the [RuIV-OO]2+ level under catalytic
conditions for each complex. Complex 2, however, passes through the [RuV-OO]3+ level prior to the rapid
loss of dioxygen. Evidence for a competing reaction pathway is provided for 3, where the [RuV)O]3+ and
[RuIII-OH]2+ redox levels can be generated by disproportionation of the [RuIV)O]2+ form of the catalyst (kd
) 1.2 M-1 s-1). An auxiliary reaction pathway involving the abstraction of an O-atom from CAN is also
implicated during catalysis. The variability of reactivity for 1-3, including the position of the RDS and potential
for O-atom transfer from the terminal oxidant, is confirmed to be intimately sensitive to electron density at
the metal site through extensive kinetic and isotopic labeling experiments. This study outlines the need to
strike a balance between the reactivity of the [RudO]z unit and the accessibility of higher redox levels in
pursuit of robust and reactive water oxidation catalysts.
as a Mn4O4 cluster,9-12 the majority of synthetic water oxidation
catalysts have been designed to contain multiple metal centers.6,13
Introduction
The conversion of solar energy to chemical energy is a
promising sustainable option for high-density energy storage
and transportation.1-5 In this context, there is a long-standing
effort to develop molecular (photo)electrocatalysts to negotiate
the proton-coupled electron-transfer (PCET) and thermodynamic
(E° ) 1.23 V at pH ) 0) demands of extracting H2 and O2
fuels from water.2,6-8 Drawing inspiration from the fact that
the oxygen-evolving complex (OEC), the site of biological water
oxidation catalysis within photosystem II, has been identified
The first reported synthetic water oxidation catalyst was the
“blue dimer”, cis,cis-[(bpy)2(H2O)RuIIIORuIII(OH2)(bpy)2]4+ (I;
bpy ) 2,2′-bipyridine).14 Since that discovery almost three
decades ago, a relatively small sample of compounds has
demonstrated competence in mediating dioxygen formation from
water. This list includes derivatives of the “blue dimer”,15
(9) Ferreira, K. N.; Iverson, T. M.; Maghlaoui, K.; Barber, J.; Iwata, S.
Science 2004, 303, 1831–1838.
(10) Stull, J. A.; Stich, T. A.; Service, R. J.; Debus, R. J.; Mandal, S. K.;
Armstrong, W. H.; Britt, R. D. J. Am. Chem. Soc. 2010, 132, 446–
447.
(1) Nocera, D. G. ChemSusChem 2009, 2, 387–390.
(2) Eisenberg, R.; Nocera, D. G. Inorg. Chem. 2005, 44, 6799–6801 and
references therein.
(11) Loll, B.; Kern, J.; Saenger, W.; Zouni, A.; Biesiadka, J. Nature 2005,
438, 1040–1044.
(3) Lewis, N. S.; Nocera, D. G. Proc. Natl. Acad. Sci. U.S.A. 2006, 103,
15729–15735.
(12) Yano, J.; Kern, J.; Sauer, K.; Latimer, M. J.; Pushkar, Y.; Biesiadka,
J.; Loll, B.; Saenger, W.; Messinger, J.; Zouni, A.; Yachandra, V. K.
Science 2006, 314, 821–825.
(4) Armaroli, N.; Balzani, V. Angew. Chem., Int. Ed. 2006, 46, 52–66.
(5) Lewis, N. S. Science 2007, 315, 798–801.
(6) Eisenberg, R.; Gray, H. B. Inorg. Chem. 2008, 47, 1697–1699 and
references therein.
(13) Yagi, M.; Kaneko, M. Chem. ReV. 2001, 101, 21–35.
(14) Gersten, S. W.; Samuels, G. J.; Meyer, T. J. J. Am. Chem. Soc. 1982,
104, 4029–4030.
(7) Huynh, M. H. V.; Meyer, T. J. Chem. ReV. 2007, 107, 5004–5064.
(8) Chang, C. J.; Chang, M. C. Y.; Damrauer, N. H.; Nocera, D. G.
Biochim. Biophys. Acta 2004, 1655, 13–28.
(15) Rotzinger, F. P.; Munavalli, S.; Comte, P.; Hurst, J. K.; Gra¨tzel, M.;
Pern, F. J.; Frank, A. J. J. Am. Chem. Soc. 1987, 109, 6619–6626.
9
16094 J. AM. CHEM. SOC. 2010, 132, 16094–16106
10.1021/ja106108y  2010 American Chemical Society

Electronic Modification of Water Oxidation Catalysts
A R T I C L E S
bimetallic complexes of different geometries containing Ru and
Mn active sites,16-21 and polyoxometallic complexes containing
Ru or Co.22-24
related compounds15), the completely inorganic Ru-substituted
polytungstate cluster [{Ru4O4(OH)2(H2O)4}(γ-SiW10O36)2]10-
(II),22,23,41,42 the dimeric complexes {[Ru(tpy)(OH2)]2(µ-bpp)}3+
(III; bpp ) bis(2-pyridyl)-3,5-pyrazolate)43 and [(tpy)(OH2)-
MnIII(O)2MnIV(OH2)(tpy)]3+ (tpy ) 2,2′;6′,2′′-terpyridine),44 and
others,13 there remain many advantages to the study of single-
site Ru(II) complexes; e.g., synthetic accessibility; relatively
well-defined spectroscopic and electrochemical properties for
in situ studies; lower susceptibility to annation and lower
computational time required for modeling studies.29,37,45 Meyer
et al. have taken advantage of these attributes to quantitatively
describe a catalytic pathway negotiated by [RuII(tpy)(bpm)-
(OH2)]2+ (IV; bpm ) 2,2′-bipyrimidine), [RuII(tpy)(bpz)-
(OH2)]2+ (V; bpz ) 2,2′-pipyrazine), and related compo-
unds.25,29,31,46 The importance of this elegant work lies in the
fact that it is one of the most well-defined pathways determined
for a homogeneous water oxidation catalyst to date. A particu-
larly compelling outcome of this analysis is the possibility that
two catalytic pathways (i.e., Ru(II)/Ru(V) vs Ru(III)/Ru(V)) are
operative depending on the conditions employed. Expanding
on these studies, they have also provided experimental evidence
to explain in explicit detail the critical O-O bond formation
process.47
The landscape of homogeneous water oxidation catalysis has
changed substantially in light of the recent discoveries that
single-site complexes are capable of facilitating dioxygen
formation from water.25 The rapidly growing catalogue of
mononuclear water oxidation catalysts now includes a family
of cyclometalated Ir complexes reported by Bernhard et al.,26
and a cyclopentadienyl Ir derivative documented by Crabtree
and Brudvig.27 There is also an extensive suite of polypyridyl
Ru catalysts that have been independently reported by the
research programs headed by Thummel,17,28 Meyer,25,29-31
Sun32 and Sakai.33 A related set of complexes have also served
as useful platforms for reducing water directly to dihydrogen,34
oxidizing water with the assistance of light,24,35,36 and unraveling
the intricate reaction details with theory.37
Despite these significant breakthroughs, there remains interest
in gaining a precise understanding of the mechanistic details of
water oxidation to further the collective pursuit of highly active
catalysts that are not susceptible to deactivation. While there
has been a wealth of knowledge gained by the extensive
mechanistic and/or computational studies on I14,38-40 (and
Building on these collective observations, our group is
expanding the development of the water oxidation catalyst,
[RuII(tpy)(bpy)(OH2)]2+ (1), to unravel the myriad catalytic
pathways that may be possible during the oxidation of water.48
We consider this complex, which was first reported to be a water
oxidation catalyst in convincing fashion by Sakai et al.,33 to
have broad utility for furthering the development of homoge-
neous water oxidation catalysts because of the inherent tun-
ability, stability in solution, and well-defined electrochemical
and spectroscopic handles. By installing electron-donating and
-withdrawing groups about the periphery of the polypyridyl
ligands, for example, we have been able to directly correlate
how catalytic activity and catalyst deactivation are affected by
electron density at the metal.48
(16) Limburg, J.; Vrettos, J. S.; Liable-Sands, L. M.; Rheingold, A. L.;
Crabtree, R. H.; Brudvig, G. W. Science 1999, 283, 1524–1527.
(17) Zong, R.; Thummel, R. J. Am. Chem. Soc. 2005, 127, 12802–12803.
(18) Muckerman, J. T.; Polyansky, D. E.; Wada, T.; Tanaka, K.; Fujita, E.
Inorg. Chem. 2008, 47, 1787–1802.
(19) Wada, T.; Tsuge, K.; Tanaka, K. Angew. Chem., Int. Ed. 2000, 39,
1479–1482.
(20) Sens, C.; Romero, I.; Rodriguez, M.; Llobet, A.; Parella, T.; Benet-
Buchholz, J. J. Am. Chem. Soc. 2004, 126, 7798–7799.
(21) Deng, Z. P.; Tseng, H. W.; Zong, R. F.; Wang, D.; Thummel, R. Inorg.
Chem. 2008, 47, 1835–1848.
(22) Geletii, Y. V.; Botar, B.; Kögerler, P.; Hillesheim, D. A.; Musaev,
D. G.; Hill, C. L. Angew. Chem., Int. Ed. 2008, 47, 3896–3899.
(23) Sartorel, A.; Carraro, M.; Scorrano, G.; De Zorzi, R.; Geremia, S.;
McDaniel, N. D.; Bernhard, S.; Bonchio, M. J. Am. Chem. Soc. 2008,
130, 5006–5007.
(24) Yin, Q.; Tan, J. M.; Besson, C.; Geletii, Y. V.; Musaev, D. G.;
Kuznetsov, A. E.; Luo, Z.; Hardcastle, K. I.; Hill, C. L. Science 2010,
328, 342–345.
In this work, we further evaluate the mechanistic details
associated with water splitting using a suite of Ru catalysts with
electron-donating and -withdrawing substituents at the activating
positions of the scaffold [RuII(tpy)(bpy-R)(OH2)]2+ [R ) sH
(1); sOMe (2); and -CO2H (3); bpy-OMe ) 4,4′-dimethoxy-
2,2′-bipyridine; bpy-CO2H ) 4,4′-dicarboxy-2,2′-bipyridine;
Figure 1]. (Notation used in this article to describe redox and
protonation levels: [RuIIsOH2]2+ ) [RuII(tpy)(bpy-R)(OH2)]2+;
(25) Concepcion, J. J.; Jurss, J. W.; Brennaman, M. K.; Hoertz, P. G.;
Patrocinio, A. O. T.; Murakami Iha, N. Y.; Templeton, J. L.; Meyer,
T. J. Acc. Chem. Res. 2009, 42, 1954–1965.
(26) McDaniel, N. D.; Coughlin, F. J.; Tinker, L. L.; Bernhard, S. J. Am.
Chem. Soc. 2008, 130, 210–217.
(27) Hull, J. F.; Balcells, D.; Blakemore, J. D.; Incarvito, C. D.; Eisenstein,
O.; Brudvig, G. W.; Crabtree, R. H. J. Am. Chem. Soc. 2009, 131,
8730–8731.
[RuIIIsOH2]3+ ) [RuIII(tpy)(bpy-R)(OH2)]3+; [RuIVdO]2+
)
(28) Tseng, H.; Zong, R.; Muckerman, J. T.; Thummel, R. Inorg. Chem.
2008, 47, 11763–11773.
(29) Concepcion, J. J.; Tsai, M.-K.; Muckerman, J. T.; Meyer, T. J. J. Am.
Chem. Soc. 2010, 132, 1545–1557.
(40) Binstead, R. A.; Chronister, C. W.; Ni, J.; Hartshorn, C. M.; Meyer,
T. J. J. Am. Chem. Soc. 2000, 122, 8464–8473.
(30) Concepcion, J. J.; Jurss, J. W.; Norris, M. R.; Chen, Z.; Templeton,
J. L.; Meyer, T. J. Inorg. Chem. 2010, 49, 1277–1279.
(31) Concepcion, J. J.; Jurss, J. W.; Templeton, J. L.; Meyer, T. J. J. Am.
Chem. Soc. 2008, 130, 16462–16463.
(41) Geletii, Y. V.; Huang, Z.; Hou, Y.; Musaev, D. G.; Lian, T.; Hill,
C. L. J. Am. Chem. Soc. 2009, 131, 7522–7523.
(42) Sartorel, A.; Miro, P.; Salvadori, E.; Romain, S.; Carraro, M.; Scorrano,
G.; Di Valentin, M.; Llobet, A.; Bo, C.; Bonchio, M. J. Am. Chem.
Soc. 2009, 131, 16051–16053.
(32) Duan, L. L.; Fischer, A.; Xu, Y. H.; Sun, L. C. J. Am. Chem. Soc.
2009, 131, 10397–10399.
(33) Masaoka, S.; Sakai, K. Chem. Lett. 2009, 38, 182–183.
(34) Karunadasa, H. I.; Chang, C. J.; Long, J. R. Nature 2010, 464, 1329–
1333.
(43) Bozoglian, F.; Romain, S.; Ertem, M. Z.; Todorova, T. K.; Sens, C.;
Mola, J.; Rodriguez, M.; Romero, I.; Benet-Buchholz, J.; Fontrodona,
X.; Cramer, C. J.; Gagliardi, L.; Llobet, A. J. Am. Chem. Soc. 2009,
131, 15176–15187.
(35) Duan, L.; Xu, Y.; Zhang, P.; Wang, M.; Sun, L. Inorg. Chem. 2010,
49, 209–215.
(44) Limburg, J.; Vrettos, J. S.; Chen, H.; de Paula, J. C.; Crabtree, R. H.;
Brudvig, G. W. J. Am. Chem. Soc. 2001, 123, 423–430.
(45) Meyer, T. J.; Huynh, M. H. V. Inorg. Chem. 2003, 42, 8140–8160.
(46) Chen, Z.; Concepcion, J. J.; Jurss, J. W.; Meyer, T. J. J. Am. Chem.
Soc. 2009, 131, 15580–15581.
(36) Besson, C.; Huang, Z.; Geletii, Y. V.; Lense, S.; Hardcastle, K. I.;
Musaev, D. G.; Lian, T.; Proust, A.; Hill, C. L. Chem. Commun. 2010,
46, 2784–2786.
(37) Wang, L.-P.; Wu, Q.; Van Voorhis, T. Inorg. Chem. 2010, 49, 4543–
4553.
(47) Chen, Z.; Concepcion, J. J.; Hu, X.; Yang, W.; Hoertz, P. G.; Meyer,
T. J. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 7225–7229.
(48) Wasylenko, D. J.; Ganesamoorthy, C.; Koivisto, B. D.; Henderson,
M. A.; Berlinguette, C. P. Inorg. Chem. 2010, 49, 2202–2209.
(38) Yang, X.; Baik, M.-H. J. Am. Chem. Soc. 2004, 126, 13222–13223.
(39) Yamada, H.; Siems, W. F.; Koike, T.; Hurst, J. K. J. Am. Chem. Soc.
2004, 126, 9786–9795.
9
J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010 16095

A R T I C L E S
Wasylenko et al.
otherwise stated, rate constants were determined by fitting absor-
bance versus time traces using the kinetics-fitting algorithm within
the Varian Cary WinUV Kinetics Application [Version 3.00(182)]
software package. An average of 10 trials was used to determine
reported values of k1-k2 until standard deviations of less than 10%
were achieved.
Dioxygen evolution data were recorded using a custom-built
apparatus consisting of a 10-mL round-bottom flask equipped with
a septum and a threaded side arm for insertion of the probe; the
total working volume is 16.8 mL. In a typical experiment, the flask
was charged with a solution of (NH4)2[Ce(NO3)6] (CAN) in 3.0
mL of the acid solution (13 mM) and the headspace was then purged
with N2(g) for ca. 20 min until a stable reading was obtained. Note
that the CAN solution was not purged to maintain dioxygen
saturation of the reaction solution so that a rapid response to
dioxygen evolution could be measured. A deaerated solution
containing the catalyst was then injected through a rubber septum
resulting in a catalyst concentration of 7 × 10-5 M. The solution
was stirred in a temperature modulated oil bath at 30 ( 2 °C for
the duration of the experiment. Dioxygen evolution was monitored
every 10 s with an optical probe (Ocean Optics FOXY-OR125-
AFMG) and a multifrequency phase fluorimeter (Ocean Optics
MFPF-100) calibrating using N2 purging as a zero point and air as
20.9%. Raw data from the sensor were collected by the TauTheta
Host Program and then converted into the appropriate calibrated
O2 sensor readings in “%O2” by the OOISensors application.
Separate solutions of 1 and CAN in 0.1 M HNO3 were prepared
and taken up in separate syringes for the ESI-MS experiments and
directed to the capillary for ionization (combined flow rate ) 50
µL/min). All mass spectra were recorded on a Waters Micromass
Q-Tof Micro mass spectrometer or Agilent Technologies 6520
Accurate-Mass Q-TOF LC/MS spectrometer. Instrument parameters
were optimized to maximize observed spectra: capillary voltage
(3200 V); source temperature (100 °C); desolvation temperature
(300 °C); desolvation flow rate (250 L/h); cone voltage (20 V);
collision voltage (1 V); quadrupole ion energy (2 V).
Figure 1. Compounds under investigation in this study (counterion )
ClO4-).
[RuIV(tpy)(bpy-R)O]2+; [RuVdO]3+ ) [RuV(tpy)(bpy-R)O]3+.)
The location of the substituents on the pyridyl ligands was
selected by virtue of our prior work, which showed that the
catalytic behavior of derivatives of 1 is inherently more sensitive
to substituents positioned on the bpy ligand rather than the tpy
ligand because one of the pyridyl rings of the bpy ligand resides
trans to the Ru-O bond.48 The overarching goal of this study
is to detail specifically how electronic density at the active site
affects the catalytic performance while holding the balance of
the structure at parity. While it is shown that each of the catalysts
adhere in large part to the acid-base mechanism previously
documented by Meyer et al., there are some distinctive
differences in reactivity among the series that are highlighted
by extensive kinetic and isotopic labeling experiments. This
study offers clarity on how a balance must be established in
terms of making the higher redox levels of the metal ion more
accessible (e.g., by using an electron-rich ligand environment)
and the reactivity of the [RudO] unit (e.g., electron deficient
ligands). This work also outlines how an auxiliary reaction
pathway involving the abstraction of an O-atom from a common
terminal oxidant figures into the analysis of water oxidation
catalysts.
The H218O labeling studies were performed using water contain-
ing 9.87% 18OH2 on a weight basis prepared from 98.7% 18OH2
purchased from Cambridge Isotopes Laboratories. The reactions
were performed in the same apparatus as used for the dioxygen
evolution studies. In a typical experiment, a solution of CAN (0.11
M) in HNO3 solutions was degassed by thoroughly purging the
solution and headspace with dinitrogen until a stable dioxygen
reading was obtained (ca. 30 min). A deaerated solution of the
catalyst dissolved in H2O was then injected via syringe into the
stirring CAN solution to achieve catalyst concentrations of ∼3 ×
10-4 M; the additional 16OH2 content was taken into account. After
the dioxygen reading had stabilized indicating the stoichiometric
consumption of CAN, several 10-20 µL samples of the headspace
gases were directly injected into a Varian 210 GC/MS Ion Trap
containing a Molsieve 5A gas separation column and an ion trap
set to focus on ions within the m/z 20-80 range. Traces of
individual ions were determined by extracting the desired m/z value
from the spectrum; relative concentrations of isotopes were
determined by integrating the area under the signal of the appropri-
ate extracted m/z value. The introduction of atmospheric dioxygen
into the GC/MS apparatus was accounted for by using the measured
O2/N2 ratio of pure N2 (extracted from the reaction vessel without
the catalyst) to correct for the 32O2 signal during the catalytic runs.
Mixing ratios of NO2 were monitored by optical absorption at 405
nm using cavity ring-down spectroscopy (CRDS); the spectrometer
has been described in detail elsewhere.50 Each experiment involved
the initial sampling of the headspace of a reaction flask containing
a 1.0 M HNO3 solution of CAN (2.0 mL). After a baseline response
was obtained, a sparged solution containing the catalyst was added
Experimental Section
Preparation of Compounds. Ligands 2,2′;6′,2′′-terpyridine (tpy),
4,4′-dimethoxy-2,2′-bipyridine (bpy-OMe), and 2,2′-bipyridine
(bpy) and metal complex precursor RuCl3 ·3H2O were purchased
from Aldrich and Pressure Chemical Company, respectively, and
used without further purification. The ligand 4,4′-dicarboxy-2,2′-
bipyridine (bpy-CO2H) was synthesized according to a published
procedure.49 Compounds [Ru(tpy)(bpy)(OH2)](ClO4)2 (1), [Ru(tpy)-
(bpy-OMe)(OH2)](ClO4)2 (2), and [Ru(tpy)(bpy-CO2H)(OH2)]-
(ClO4)2 (3) were prepared and purified as previously reported.48
Acid solutions were prepared from concentrated HNO3 (70%, ACS
grade) and distilled, deionized water and were measured to have a
pH of 1.07 and 0.05 for 0.1 M HNO3, 1.0 M HNO3, respectively.
Physical Methods. Electrochemical measurements were per-
formed under anaerobic conditions and recorded with a Princeton
Applied Research VersaStat 3 potentiostat, a glassy carbon working
electrode (diameter ) 3 mm), and a [Ag]/[AgCl] reference electrode
(3 M NaCl; 210 mV vs NHE). All potentials reported herein are
referenced to a normal hydrogen electrode (NHE). Electrochemistry
experiments carried out at variable pH values were carried out in
appropriate buffered solutions with I ) 0.1 M ([analyte] ) ∼1
mM), except at pH 0, where 1.0 M HNO3 was used ([analyte] )
∼1 mM). Electronic spectroscopy data were collected on a Varian
Cary 5000 UV-vis spectrophotometer. Kinetics measurements were
performed by stopped-flow methods using a Hi-Tech Scientific
SFA-20 coupled to the Cary 5000 spectrometer, and absorbance
versus time traces were collected at appropriate wavelengths. Unless
(49) Hoertz, P. G.; Staniszewski, A.; Marton, A.; Higgins, G. T.; Incarvito,
C. D.; Rheingold, A. L.; Meyer, G. J. J. Am. Chem. Soc. 2006, 128,
8234–8245.
(50) Paul, D.; Osthoff, H. D. Anal. Chem. 2010, 82, 6695–6703.
9
16096 J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010

Electronic Modification of Water Oxidation Catalysts
A R T I C L E S
Figure 2. Summary of (possible) competing reaction pathways for 1-3 describing relevant (proton-coupled) electron-transfer steps, O-O bond formation,
exclusion of dioxygen, and substrate binding (L ) coordinating ligand; e.g., Cl-, MeCN; R ) -H, 1; -OMe, 2; -CO2H, 3). Rate-determining steps (RDSs)
for 1 and 3 (RDS), and 2 (RDS′) at pH 1 are indicated.51
to the sample chamber. All concentrations and conditions were the
same as those for the dioxygen evolution experiments described
above.
M HNO3 (i.e., k4); alternatively in 1.0 M HNO3, the one-electron
oxidation of [RuIVsOO]2+ (i.e., k5) occurs followed by the rapid
reductive elimination of dioxygen and subsequent coordination
of water. The rate-determining step (RDS) in 0.1 M HNO3 is
assigned as kO2 within the “Ru(II)/Ru(V) cycle”, while k5
represents the RDS in the “Ru(III)/Ru(V) cycle” in 1 M HNO3.
While our initial set of measurements suggested that 1-3
do, in large part, follow this same “acid-base” mechanism (i.e.,
nucleophilic attack of [RuVdO]3+ by water), there are certain
aspects of reactivity that diverge from that of IV. We have
rationalized that this discrepancy arises not only because of a
difference in the position of the RDS but also because of
supplementary reaction cycles, in certain cases, that include
disproportionation pathways and oxygen abstraction from CAN
(denoted by red arrows in Figure 2).53 Our experiments
described herein shine light on the fact that the accessibility of
each of these pathways is affected not only by the electron
density at the metal but also by reaction conditions. The
following sections detail our interpretation of the reactivity of
1-3, which balances the thermodynamic demands of accessing
the Ru(V) redox level, the electrophilicity of the high-valent
[RudO] fragment, the acid medium, and the presence of CAN.
Structural Identity of the Active Site. There has been a
general lack of solidarity in the literature over the past three
decades over whether 1 is a bona fide water oxidation
catalyst.13,28,31 This controversy is presumably rooted in the fact
that water oxidation is not observed upon generation of
[RuIVdO]2+ in solution,54 and the [RuIVdO]2+ form of 1 may
not be capable of undergoing electron transfer with CAN (E1/2,
Ce(IV)/Ce(III) ) 1.45 V vs NHE; Figure S1) due to the
Results
This study was designed to develop a detailed description of
the reaction landscape for a structurally related series of water
oxidation catalysts 1-3 that contain variable electron density
at the metal center. In this vein, each of the catalysts was
evaluated by spectrophotometrically monitoring the Ce(IV)-
driven oxidation of water (eq 1) in an acidic medium. We also
tracked this reaction with an optical probe positioned in the
reaction headspace as a secondary means of quantifying the
catalytic activity. ESI-MS and UV-vis spectroscopy were used
to help structurally identify catalytic intermediates, rate constants
for key (proton-coupled) electron-transfer steps were determined
by stopped-flow methods, and the source of O-atoms in
dioxygen liberated during the reactions was verified by 18OH2-
labeling experiments.
4Ce(IV) + 2H2O catalyst8 4Ce(III) + O2(g) + 4H+
(1)
Our mechanistic studies were designed to examine whether
1 adheres to the “acid-base” mechanism (denoted by black
arrows in Figure 2) aligned with the scheme documented by
Meyer et al., for IV.31,46 The reaction pathway mediated by IV
begins with two consecutive PCET steps to form [RuIVdO]2+
.
A subsequent oxidation process generates the highly electro-
philic [RuVdO]3+ fragment that interacts with water to undergo
the requisite OsO bond formation step, kO-O, to form
[RuIIIsOOH]2+, which collapses to [RuIVsOO]2+ following
another PCET step, k4.52 The dioxygen ligand of this species
can then be displaced by water to regenerate the catalyst in 0.1
inaccessible [RuVdO]3+/[RuIVdO]2+ redox couple (E1/2
≈
1.60-1.80 V30,48). Moreover, the identification of the [RuVdO]3+
(53) kOAT represents an “O-atom transfer” (OAT) step that refers strictly
to the abstraction of an O-atom from CAN by the Ru catalyst, and
does not necessarily imply a static formal charge of the O-atom before
and after the reaction.
(54) Takeuchi, K. J.; Thompson, M. S.; Pipes, D. W.; Meyer, T. J. Inorg.
Chem. 1984, 23, 1845–1851.
(51) We do not rule out the possibility of [RuII-OOH]+ being formed
during the kO-O′ step.
(52) Specific bonding arrangement of peroxo Ru(IV) complexes is discussed
in a later section.
9
J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010 16097

A R T I C L E S
Wasylenko et al.
Table 1. Summary of Electrochemical Data and
evidenced by the respective cyclic voltammograms for 1-3 in
0.1 M HNO3 (Table 1; Figures S2-S3). An oxidative sweep
for 1 produces two closely spaced single-electron redox
processes at +1.04 and +1.23 V, along with a third signal at
∼+1.80 V, which occurs with the onset of a rapid increase in
peak current due to the concomitant catalytic oxidation of water.
The close proximity of the first two signals is a consequence of
(proton-coupled) electron-transfer processes stabilizing the
higher redox levels of the complexes (e.g., the first two oxidation
steps generate [RuIIIsOH2]3+ and [RuIVdO]2+, respectively).
The reversible [RuIIIsOH2]3+/[RuIIsOH2]2+ wave occurs at
1.04, 0.91, and 1.16 V for 1-3, respectively, in accordance with
the variable electron-donating character of the substituents. The
variance of the respective [RuIVdO]2+/[RuIIIsOH2]2+ processes
for 1 and 2 is much less pronounced (i.e., ca. 30 mV), which
reflects a diminished level of π-backbonding in stabilizing the
Ru(III) redox level. Two distinct oxidation processes are not
observed in the voltammograms for 3, which is ascribed to slow
reaction kinetics at the electrode surface.57 A more pronounced
Spectrophotometrically Determined pKa Values for 1-3
E1/2 (V vs NHE)a
pKad
Complex Ru(III)/Ru(II)b Ru(IV)/Ru(III)b Ru(V)/Ru(IV)b [RuII-OH2]2+ [RuIII-OH2]3+
1
2
3
1.04
0.91
1.16
1.23
1.24
sc
1.80
1.77
1.89
10.5
11.2
10.4
1.7e
3.2
1.2
a Samples (1 mM) dissolved in 0.1 M HNO3 were measured at 50
mV/s and referenced to a Ag/AgCl (3 M NaCl) electrode at 25 ( 2 °C;
Ru(V)/Ru(IV) value determined by square wave voltammetry.
b Assignment of proton-coupled oxidation waves are indicated: Ru(II) )
[RuIIsOH2]2+; Ru(III) ) [RuIIIsOH2]3+; Ru(IV) ) [RuIVdO]2+; Ru(V)
d
) [RuVdO]3+
.
c Not observed. [Ru] ) 5.0 × 10-5 M in buffered
phosphate solution. e Previously reported.54
species prior to the onset of catalysis at the electrode can be
ambiguous in certain conditions. There has been specula-
tion that the decomposition of Ru coordination complexes
into catalytically active nanoparticles broadly formulated as
RuO2•xH2O is the source of dioxygen formation. We have ruled
out this possibility by a series of control experiments,48 the full
recovery of the catalyst following the addition of 30 equiv of
CAN to 1-3 (vide infra), and the fact that RuO2 forms
catalytically inactive RuO4 at the Ce/Ru molar ratio relevant to
our experiments.55,56
variance (ca. 120 mV) is observed for the [RuVdO]3+
/
[RuIVdO]2+ couple for the series (the E1/2 for this process was
resolved by square-wave voltammetry; Figure S4).
The Pourbaix diagrams for 1,54 2, and 3 reveal that the
[RuIIIsOH]2+/[RuIIsOH2]2+ and [RuIVdO]2+/[RuIIIsOH]2+ re-
dox processes each change by approximately 59 mV per pH
decade over a large pH range (Figure 3). Deviations from
Nernstian behavior are observed in regions where the dissocia-
tion of a proton can be affected; i.e., at relevant pKa values.
The peak separations of the Ru(III)/Ru(II) and Ru(IV)/Ru(III)
redox couples for 2 are approximately 200 mV over the pH
3.1-11.2 range. At pH < 3.1, the oxidation of 2 is not associated
with the loss of a proton based on the pH-independent behavior
of the [RuIII-OH2]3+/[RuII-OH2]2+ redox couple. Consequently,
the subsequent oxidation step is accompanied with the loss of
two protons (i.e., [RuIVdO]2+/[RuIII-OH2]3+) in strongly acidic
media. This assignment is corroborated by the slope of -118
mV/pH below pH 3.1 in Figure 3, as well as independent
spectrophotometric measurements (Figure S7). Compound 2 is
deprotonated at pH > 11.2 resulting in the pH-independent
oxidation of the Ru(II) center until pH 12; the first oxidation
wave at pH > 12 is assigned as a two-electron, one-proton event
(i.e., [RuIVdO]2+/[RuIIsOH]1+). The [RuVdO]3+/[RuIVdO]2+
redox couple remains relatively constant at ∼1.77 V vs NHE
over the pH 0-13 range.
While there now appears to be general agreement that 1 is,
indeed, a water oxidation catalyst,25,33,48 the nature of the active
site remains an open question. Given that dioxygen formation
28
has been observed in reactions catalyzed by [Ru(tpy)(bpy)Cl]+
and seven-coordinate Ru(II) compounds with strained ligand
environments,32 we considered the possibility that 1-3 could
undergo an expansion of the primary coordination sphere to
retain a Ru-anion bond. A series of experiments carried out in
our laboratories48 and those of Sakai et al.33 and Meyer et al.,29
however, have verified that the catalytically active forms of 1-3
are the [Ru-OH2]2+ species. In fact, 1-3 each exist in a
dynamic equilibrium with coordinating ligands (e.g., MeCN,
Cl-, SO42-) that effectively suppress the catalytic activity of
these Ru complexes.33,48 Consequently, these coordinating
species were avoided in this study to avoid complicating our
interpretation of the mechanistic behavior. Each of the samples
were prepared in bulk and analyzed for purity by H NMR
spectroscopy, elemental analyses, and UV-vis spectroscopy
prior to analysis.
Electrochemical Behavior. The electron-donating and -with-
drawing characters of the bpy-OMe and bpy-CO2H ligands are
1
Figure 3. Plot of the pH-dependence on the electrochemical behavior of (a) 2 and (b) 3. The red solid lines indicate trends in data; diamonds, circles, and
triangles correspond to the Ru(III)/Ru(II), Ru(IV)/Ru(III), and Ru(V)/Ru(IV) redox couples, respectively. The silent Ru(IV)/Ru(III) wave at low pH for 3
is ascribed to slow kinetics at the electrode. (The Pourbaix diagram for 1 is reported elsewhere.54
)
9
16098 J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010

Electronic Modification of Water Oxidation Catalysts
A R T I C L E S
Table 2. Electron-Transfer Rate Constants in 0.1 M HNO3 at 25 (
2 °C
(Figures S6-S8). A detailed listing of the spectroscopic features
in Figure 4 is provided in Table S2.
Compounda
The relative stabilities of the lowest redox level (i.e., Ru(II))
for each of the catalysts in solution (monitored by tracking the
bleaching of the MLCT band) are progressively higher for 2,
1, and 3, respectively. This behavior is more pronounced in
stronger acids (e.g., 1 M HNO3), where 2 is rapidly oxidized to
Ru(III) via the series of reactions described in eqs 2-4.59
Complex 1 undergoes this same decomposition at a slower rate
corresponding to a pseudo-first-order rate coefficient of 1.8 ×
10-3 s-1, while 3 is stable in solution over a period of several
weeks. Consequently, all of our electron-transfer rates and
catalytic measurements were carried out in 0.1 M HNO3 to (i)
suppress the spontaneous decomposition of 1 and 2 in solution
at lower pH values and (ii) to evade the formation of complex
hydroxides when CAN is dissolved in aqueous media at higher
pH levels. We also note that determination of rate constants
for subsequent reaction steps (e.g., k3, kO-O) at pH 0 is impeded
by the strong absorbance of 1 M HNO3 at <350 nm masking
the absorbance region of interest.
Reaction
Step
Reaction
Description
2
1
3
k1 (M-1 s-1
k2 (M-1 s-1
)
)
[RuIIsOH2]2+ f [RuIIIsOH2]3+
+
1.7 × 105 4.4 × 104
b
rapid
H+ (CeIV f CeIII
)
[RuIIIsOH2]3+ f [RuIVdOH2]3+
H+ (CeIV f CeIII
+
4.6 × 103 6.6 × 103 3.3 × 105
)
a [Catalyst] ) 5 × 10-5 M. b Time scale too short to be resolved
with our instrument (i.e., >106 M-1 s-1).
The Pourbaix diagram for 3 displays some minor deviations
relative to that of 1 and 2. For instance, a smaller difference
(ca. 90 mV) is observed between the first two successive PCET
events over the 1.2 < pH < 10.5 range. The first oxidation wave
is pH-independent at 10.5 > pH < 1.2. The [RuIVdO]2+
/
[RuIIIsOH]2+ couple was not observed at pH < 2.8 using various
electrochemical conditions and techniques (e.g., differential
pulse voltammetry, square-wave voltammetry). While we
considered the possibility that the Ru(III) level is not stable and
undergoes disproportionation,31,58 our titration experiments do
not support this scenario; consequently, this silent redox couple
is attributed to slow reaction kinetics at the electrode.57 The
[RuVdO]3+/[RuIVdO]2+ couple for 3 was observed at ca. 1.9
V vs NHE over a significant pH region.
-
2[RuIIsOH2]2+ + NO3 + 3H+ h 2[RuIIIsOH2]3+
+
HNO2 + H2O (2)
-
[RuIIsOH2]2+ + NO3 + H+ h [RuIVdO]2+ + HNO2 +
The E1/2 vs pH plots facilitate the extrapolation of the pKa
values of the aqua ligand for 1-3, which are shown to track
with electron density at the metal center; i.e., pKa increases for
3, 1, and 2, respectively (Figure 3). These values were
independently verified by spectrophotometric titrations of the
respective complexes in buffered solutions (Table 2 and Figure
S5).
H2O (3)
[RuIIsOH2]2+ + [RuIVdO]2+ + 2H+ f 2[RuIIIsOH2]3+
(4)
The electron-withdrawing character imparted by the sCO2H
substituents is clearly demonstrated by the instability or reactiv-
ity of the [RuIVdO]2+ form of 3 in solution. The [RuIVdO]2+
species that is generated in solution undergoes a slow decom-
position to species with spectral features indicative of divalent
and trivalent Ru. The decay of the [RuIVdO]2+ complex was
found to be second-order in [3], thereby indicating a bimolecular
decomposition pathway. On this basis, we attribute this process
to a disproportionation reaction step (kd) that forms [RuVdO]3+
and [RuIIIsOH]2+ in the acid medium (kd ) 1.2 M-1 s-1; Figure
S9); the [RuVdO]3+ species, in turn, is presumed to go on to
react with water to form [RuIIIsOOH]2+ via kO-O. The implica-
tion of this result is that a Ru(V) species is generated after the
addition of only 2 equiv of CAN to 3, thus providing access to
the kO-O pathway. Moreover, evidence for the kO-O′ step, which
involves dioxygen bond formation at a Ru(IV) center, is
provided by the onset of a signal at λmax ≈ 490 nm, which we
Spectroscopic Signatures for Well-Defined Redox Levels of
1-3. The absorbance profiles of the three different valence forms
found early in the catalytic cycle (i.e., [RuIIsOH2]2+
,
[RuIIIsOH2]3+, and [RuIVdO]2+ in Figure 2) for 1-3 are
provided in Figure 4. The lowest redox level for all three
catalysts exhibits the signature metal-to-ligand charge-transfer
(MLCT) band between 470 and 500 nm along with higher-
energy ligand-based π-π* transitions. The oxidation of each
of these complexes results in a bleaching of this broad visible
band leaving prominent spectral features in the UV region. The
respective Ru(III) and Ru(IV) species were each generated by
adding stoichiometric equivalents of CAN to 5.0 × 10-5
M
solutions of the respective complexes in 0.1 M HNO3; clear
isosbestic points (Table S1) were observed in titration experi-
ments with 1-3 and signal the successive stoichiometric
conversion of [RuIIsOH2]2+ to [RuIIIsOH2]3+ and [RuIVdO]2+
Figure 4. Electronic absorption spectra of the [RuIIsOH2]2+, [RuIIIsOH2]3+ and [RuIVdO]2+ redox levels for (a) 1, (b) 2, and (c) 3. The Ru(III) and Ru(IV)
redox levels were generated by stoichiometric titrations of the complex (5.0 × 10-5 M) with CAN in 0.1 M HNO3.
9
J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010 16099

A R T I C L E S
Wasylenko et al.
assign as the formation of the [RuIIsO2H2]2+ (or [RuIIsOOH2]+)
species. (A second disproportionation reaction generating the
[RuIIsOH2]2+ species according to eq 5 was also considered
but is not supported by other titration experiments; vide infra.)
Both of these reaction pathways deviate from the acid-base
mechanism and not only show a thermodynamically favorable
route for the formation of a [RuVdO]3+ species in these titration
experiments and but also provide evidence that nucleophilic
attack of an electron-deficient [RuIVdO]2+ by water can occur.
feature at 320 nm later in the reaction (Figure S10), occurs on
a much slower time scale than that of k3; i.e., k3 and kO-O are
3.7 M-1 s-1 and 3 × 10-5 s-1, respectively. This [RuIIIsOOH]2+
species does not disproportionate to [RuIVsOO]2+ and
[RuIIsH2O2]2+ on a reasonable time scale (as was observed for
IV).60 We note that the measured kO-O value is slower than
kcat, suggesting that (i) the second process that is occurring is
not actually kO-O or (ii) auxiliary reactions lead to an under-
estimation or misrepresentation of kO-O (vide infra).
2[RuIIIsOH2]3+ h [RuIIsOH2]2+ + [RuIVdO]2+ + 2H+
d[RuIVdO]
-
) k2[RuIVdO][CAN]
) kO-O[RuIIIsOOH]
(8)
(9)
(5)
dt
Monitoring the spectral changes of 1 reveals some minor
spectral changes that we assign as this same disproportionation
process at a slow rate of ∼0.3 M-1 s-1; there is also evidence
that this process can also occur at an even slower rate for 2.
Our spectroscopic data do not support the occurrence of kO-O
for 1 or 2.
Rates of Electron Transfer and O-O Bond Formation.
Intermolecular electron-transfer steps k1-k3 and dioxygen bond
formation step kO-O were determined (where applicable) by
stopped-flow methods in 0.1 M HNO3 by adding equimolar
quantities of CAN to the Ru(II) or Ru(III) forms of each of the
three catalysts (Table 2). The first two electron-transfer steps
can be determined by monitoring the rates of the reactions
according to eqs 6-7.29 The first electron-transfer step k1 is
rapid for each of the catalysts; k2 is slower but still rapid, and
follows the trend 3 > 1 > 2. Both of these electron-transfer events
are relatively fast for all of the catalysts and do not represent
RDSs within the respective catalytic cycles.
d[RuIIIsOOH]
dt
A - Ao
A∞ - Ao
′
ln
) -kO-O
t
(10)
(
)
The addition of 1 equiv of CAN to the [RuIVdO]2+ form of
1 results only in a slightly lower intensity of the absorbance
profile (Figure S11). Two possible interpretations of these data
include the following: (i) the electron-transfer reaction is slow
or nonexistent; or (ii) the [RuIVsOO]2+ species is generated at
a faster rate than the [RuVdO]3+ level. The slow reaction step
k3 for 1 relative to 2 is supported by the greater thermodynamic
demands associated with oxidizing the Ru(IV) center. The rapid
generation of the [RuIVsOO]2+ species is consistent with a slow
kd step followed by relatively faster kO-O/k4 steps and/or a fast
oxygen-atom transfer (OAT) step, kOAT. This scenario implies
that the [RuIIIsOH]2+ species that is formed by kd would be
quickly regenerated to the [RuIVdO]2+ species; i.e., the dominant
species in solution under these conditions would be [RuIVdO]2+
and [RuIVsOO]2+, thus rendering minor spectroscopic changes.
In the case of 3, the addition of 1 equiv of CAN to the
[RuIVdO]2+ form leads to a successive decrease and increase
in absorbance at 320 nm. While this behavior is expected if we
are tracking only k3 and kO-O, there are two additional pathways
that diverge from the [RuIVdO]2+ species that may be operative
based on the observed spectral changes (e.g., kO-O′ and kd).
Because electron-transfer step k3 is expected to be very slow
under these conditions due to the energetic demands of oxidizing
a Ru(IV) center, the spectral changes are attributed to the
formation of the [RuIII-OOH]2+ species by means of the
independent kd/kO-O and kO-O′/k3′ pathways. The former is
supported by the spectral trends at 320 nm, which would
correspond to kd ≈ 2 M-1 s-1 and kO-O ) 1 × 10-4 s-1 (the
latter is expected to be faster than 2, because of the greater
susceptibility to nucleophilic attack by water, but is again slower
than kcat, underscoring that competing pathways may lead to a
false representation of this value). The kO-O step is supported
by the gradual buildup of a species characterized by λ ≈ 490
nm that we assign as the [RuII-O2H2]2+ complex ([RuII-OOH]+
is not ruled out), followed by a rapid decrease concomitant with
the onset of a signal at 690 nm, which we assign as
[RuIII-OOH]2+ (Figure S12). The disproportionation reaction
d[RuIIsOH2]
-
) k1[RuIIsOH2][CAN]
) k2[RuIIIsOH2][CAN]
(6)
(7)
dt
d[RuIIIsOH2]
dt
-
We set out to extract the value of k3 by tracking the rate of
CAN or [RuIVdO]2+ decay (and/or the onset of a distinctive
[RuVdO]3+ species) following the addition of 1 equiv of CAN
to the respective [RuIVdO]2+ forms of the catalysts (eq 8). A
subsequent (slower) kO-O step can then be extracted according
to eq 9 using the corresponding integrated rate law described
in eq 10 (A ) absorbance at time t; Ao ) absorbance at t ) 0;
A∞ ) absorbance at t∞). This procedure was shown to be
particularly effective for the analysis of single-site catalysts by
Meyer et al.29 but produced a disparate and often ambiguous
set of results for each of the catalysts in this study. For instance,
the combination of 1 equiv of CAN with the [RuIVdO]2+ form
of 1 did not produce distinct spectral changes that are consistent
with an electron-transfer step and/or dioxygen formation, while
trends corresponding to the formation of the ostensible
[RuVdO]3+ and [RuIIIsOOH]2+ species do appear in the case
of 2. We therefore assign the reaction of CAN with [RuIVdO]2+
as electron-transfer step k3 for 2 in accordance with the
acid-base mechanism. The subsequent kO-O step for 2, which
is determined by monitoring the onset of a new species with a
(58) Masllorens, E.; Rodriguez, M.; Romero, I.; Roglans, A.; Parella, T.;
Benet-Buchholz, J.; Poyatos, M.; Llobet, A. J. Am. Chem. Soc. 2006,
128, 5306–5307.
(55) Kiwi, J.; Gra¨tzel, M.; Blondeel, G. J. Chem. Soc., Dalton Trans. 1983,
2215–2216.
(59) Moyer, B. A.; Meyer, T. J. J. Am. Chem. Soc. 1979, 101, 1326–1328.
(60) A direct comparison of kO-O values of IV to 2 is made difficult by the
disparate rates of disproportionation, affecting the curvature of the
absorbance versus time traces.
(56) Mills, A. J. Chem. Soc., Dalton Trans. 1982, 1213–1216.
(57) Dovletoglou, A.; Adeyemi, S. A.; Meyer, T. J. Inorg. Chem. 1996,
35, 4120–4127.
9
16100 J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010

Electronic Modification of Water Oxidation Catalysts
A R T I C L E S
Table 3. Spectrophotometric Determination of Key Rate Constants
for 1-3 at 25 ( 2 °Ca
unlikely based on thermodynamic considerations, and that kO-O
is presumably faster for 3 than 1 (yet kcat follows an opposite
trend), we assign the RDS as kO2 for both catalysts, which is
analogous to that of IV. This result indicates that, under
conditions of excess CAN, the k3 step is enhanced due to a
combination of Nernstian behavior (e.g., Ered is 1.43 and 1.54
V with 3 and 30 equiv of CAN in 0.1 M HNO3, respectively)
and the [RuIII-OOH]2+ species being siphoned off by the excess
oxidant in solution.
0.1 M HNO3
Complex rate expression
1 M HNO3
b
kcat
b
kcat
rate expression
1
2
3
kcat[1]
6.6 × 10-4 s-1 (1100) kcat[1][CAN] 22 M-1 s-1 (210)
kcat[2][CAN] 1.0 M-1 s-1 (460)
kcat[2][CAN] 180 M-1 s-1 (3)
kcat[3]
4.3 × 10-4 s-1 (1600) kcat[3][CAN] 9.2 M-1 s-1 (500)
a [Ru] ) (1.0-9.0) × 10-5 M; [CAN] ) (3.0-27) × 10-4 M; kcat
b
determined by monitoring the decay of CAN at 360 nm in the dark. t1/2
indicated in parentheses in units of s; second-order t1/2 values correspond to
ln(2)/kcat[CAN]i.
The extraction of kcat for 2, on the other hand, reveals behavior
that is non-zero order in [CAN]61 and follows the expression
rate ) kcat[2][CAN], where kcat ) 1.0 M-1 s-1 (Figures 5b and
S13). This rate expression is aligned with the RDS being k5 (k3
is faster), which underscores how electronic parameters can
affect the position of the RDS (the RDS is kO2 for 1, 3, IV,29,31
and V29,31 in 0.1 M HNO3). The higher kcat value for 2 compared
to 1 and 3 also indicates that there is favorable dioxygen release
from a Ru(V) level relative to a Ru(IV) center and/or the
electron-donating groups help to labilize the (end-on37 or side-
on29) peroxo ligand. Note that the slope in Figure 5b does not
pass through the origin, which suggests the possibility of
described in eq 5 is ruled out on the basis that the CAN present
in solution in this experiment would quickly regenerate the
[RuIII-OH2]3+ that is formed during the kd step. We caution
that the credibility of our measured “kO-O” value is compromised
by these competitive pathways in solution.
Spectrophotometric Determination of kcat for 1-3. The
consumption of CAN was monitored in 0.1 and 1.0 M HNO3
at 360 nm to determine the observed catalytic rates (kcat) for
the reaction described in eq 1; relevant data are collected in
Table 3. A typical reaction involves adding 30 equiv of CAN
to a (1-9.0) × 10-5 M solution of the catalyst.
competing reaction pathways (e.g., decomposition of CAN, kOAT
;
vide infra).
The spectrophotometric determination of kcat for 1 and 3
reveals a dominant pathway that follows a zero-order depen-
dence in [CAN] over the initial stages of the reaction. (Under
catalytic conditions for 3 in 0.1 M HNO3, a visible precipitate
forms in the reaction flask; this process is concentration
dependent.) A linear relationship between initial reaction rates
(kinitial) and [Ru] indicates that the catalytic behavior follows
the rate expression rate ) kcat[Ru] (determined by the slopes in
Figure 5a and 5c), which implies that the RDS is not associated
with an oxidation or OAT step. Considering that the oxidation
of the purported [RuIV-OO]2+ complexes for 1 and 3 are
A particularly telling outcome of these experiments is the
spectral profiles that are converged upon at the end of the
catalytic runs (Figure 6). The dominant species under catalytic
conditions for 1 is consistent with a Ru(IV) complex (e.g.,
[RuIVdO]2+ or [RuIVsOO]2+), which is aligned with the
significant energetic barriers associated with k3 and kO2. The
time-dependent spectra for 2 do not converge on the [RuIVdO]2+
complex and support the dominant species in solution being a
[RuIVsOO]2+ or Ru(V) species. The spectral difference may
also reflect the stabilization of different bonding motifs of the
Figure 5. Spectrophotometric determination of kcat (by monitoring the consumption of CAN at 360 nm) for (a) 1, (b) 2, and (c) 3 in 0.1 M HNO3 at 25 (
2 °C by tracking the initial rates (∆t ) 10 min) of CAN consumption as a function of [Ru] following the addition of 30 equiv of CAN to the catalyst (see
text for details).
Figure 6. Spectral changes as a function of time following the addition of 30 equiv of CAN in 0.1 M HNO3 for (a) 1 (5.0 × 10-5 M), (b) 2 (2.5 × 10-5
M), and (c) 3 (2.5 × 10-5 M). Spectra are measured in 60-min time intervals; arrows indicate trends in data. The spectra for the [RuIVdO]2+ forms of each
of the catalysts are provided as a benchmark.
9
J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010 16101

A R T I C L E S
Wasylenko et al.
Table 4. Observed Rates of Dioxygen Evolution and Catalyst
Efficiencies in HNO3 at 30 ( 2 °Ca
0.1 M HNO3
1.0 M HNO3
kcat-O2
(× 10-4 s-1
Eff.
(%)c
kcat-O2
(× 10-4 s-1
b
)
b
)
Complex
1
2
3
0.61
1.5
-
80
117
50d
10
18
6.9
d
a Measured using a fluorescent optical probe in the headspace of the
reaction vessel with 200 equiv of CAN; [Ru] ) 7.0 ((0.03) × 10-5 M;
[CAN] ) 1.4 ((0.05) × 10-2 M; initial turnover frequencies are
provided in Table S3. b Pseudo-first-order rate constant obtained by
exponential fit of dioxygen evolution traces ((5%). c Catalyst efficiency
is defined as the consumption of CAN resulting in stoichiometric
equivalents of dioxygen (max 20 h; (5%). d A precipitate forms over
the course of the reaction leading to zero-order behavior in [Ru].
Figure 7. Dioxygen evolution traces for 1-3 in (a) 0.1 M HNO3, (b) 1 M
HNO3 ([Ru] ) 7.0 × 10-5 M; Vacid ) 3 mL; [CAN] ) 13 mM).
[RuIVsOO]2+ complex late in the catalytic cycle relative to 1
(i.e., end-on versus side-on;29,37 eq 11). In the case of 3, the
spectral changes are distinctive in that a signal at 486 nm begins
to emerge over time. This is an intriguing result because this
band is a signature of a Ru(II) MLCT transition, but the Ru(II)
complexes specified in Figure 2 are all susceptible to rapid
oxidation in the presence of CAN. Consequently, we assign the
onset of this signal to the stabilization of the Ru(II) resonance
form of the side-on peroxo/dixoygen complex (eq 12) by the
electron-withdrawing groups. Taking these results into collective
consideration indicates that the generic assignment of kO2 as
the RDS may be complicated by the precise bonding arrange-
ment of the Ru(IV) species late in the catalytic cycle; this
specific feature currently lacks clarity in the literature.29,37
(0.61 × 10-4 s-1); both follow the expression rate ) kobs-O2[Ru]
(zero-order in [CAN]) under pseudo-first-order conditions.
Stoichiometric conversion of oxidizing equivalents to dioxygen
are observed only for 2 over a 20-h period due in large part to
the faster catalytic rates of 2. While complications arise for
measurements with 3 because the catalyst precipitates over the
course of the reaction, catalysis persists following zero-order
reaction behavior at ∼3 × 10-9 M-1 s-1. This behavior is
ascribed to either slow dissolution of a catalytically active form
of 3 or dioxygen formation at a nanoparticulate decomposition
product. Over longer time periods, 1 and 3 may be able to
mediate the complete consumption of CAN, though complica-
tions are likely to arise due to the instability of CAN in solutions
over a period of days. In 1.0 M HNO3, the rates of dioxygen
evolution by all three catalysts are enhanced by more than an
order of magnitude and follow the trend 2 > 1 > 3 (Figure 7b).
For each of the catalysts, the amount of dioxygen evolved was
expected to correspond to the amount of oxidant introduced into
the reaction flask. In 0.1 M HNO3, however, the amount of
dioxygen formed is less than stoichiometric for 1 and 3 and
greater for 2. This provides strong evidence of an auxiliary
reaction pathway involving a secondary O-atom source (i.e.,
kOAT). This behavior also supports that the position of the RDS
can be affected by pH (i.e., kO2 at pH 1 and k5 at pH 0 for 1
and 3).29,46
We observe a significant increase in the rate of consumption
of CAN in 1.0 M HNO3 for all of the catalysts, where each
follow the expression rate ) kcat[Ru][CAN] (Table 3). The kcat
value is particularly rapid for 2 (180 M-1 s-1), while 1 is only
slightly faster than that of 3. (There is no observable precipitate
for 3 under these conditions.) We attribute this trend to k5 being
rate-determining for all three catalysts under these conditions
(consistent with that of IV31), so that the kcat is dictated by the
accessibility of the Ru(V) level; i.e., 2 > 1 > 3. The disparate
rates at pH 0 and 1 highlight how reaction conditions can affect
catalytic activity given that k3 and k5 are 3.7 and 1.0 M-1 s-1 in
0.1 M HNO3, respectively, but kcat in 1.0 M HNO3 is 180 M-1
Electrospray Ionization Mass Spectrometric Analysis. Elec-
trospray-ionization mass spectrometry (ESI-MS) is a soft
ionization technique that has demonstrated the facility to probe
reactions under catalytic conditions.62-64 Generally, polar
solvents function best for purposes of ESI-MS, and most
instruments are well equipped to evaporate solvents with boiling
points as high as water. While we are cognizant that neutral
species are invisible to ESI-MS, viable intermediates in the
catalytic pathways of 1-3 are anticipated to bear a charge
enabling the ESI-MS study of native reactions at optimal
concentrations or loadings. Also fortuitous is the fact that many
of the catalytic intermediates in this study are characterized by
unique empirical formulas that appear in a distinct region of
the mass spectrum; thus, we are not faced, in most cases, with
the tedious issue of delineating the overlapping spectroscopic
s-1
.
Determination of kobs-O2 by Monitoring Dioxygen Evolution.
Time-dependent dioxygen evolution measurements were mea-
sured for each of the complexes using a fluorescent probe in
the headspace of the reaction vessel (Figure 7a and Table 4).
In contrast to the previous experiments, these trials require a
large excess of CAN (i.e., g 200 equiv) to generate a significant
quantity of dioxygen in the headspace for a satisfactory signal-
to-noise level.
In 0.1 M HNO3, the catalytic rate of dioxygen evolution
(kobs-O2) for 2 (1.5 × 10-4 s-1) is approximately twice that of 1
(62) Henderson, W.; McIndoe, J. S. Mass Spectrometry of Inorganic and
Organometallic Compounds; John Wiley and Sons, Ltd.: Hoboken,
NJ, 2005.
(61) We note that this same behavior was observed to a lesser extent
for 1 and 3; i.e., increasing concentrations of CAN leads to a slight
enhancement of reaction rates. The implication of this result is that
the rate law expression may be better described as a mixture of
first- and second-order kinetic behavior; e.g., rate ) k[Ru] +
k′[Ru][CAN].
(63) Fenn, J. B.; Mann, M.; Meng, C. K. A. I.; Wong, S. F.; Whitehouse,
C. M. Science 1989, 246, 64–71.
(64) Chisholm, D. M.; Oliver, A. G.; McIndoe, J. S. Dalton Trans. 2009,
39, 364–373.
9
16102 J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010

Electronic Modification of Water Oxidation Catalysts
A R T I C L E S
Figure 9. Positive ion ESI-MS/MS of {[RuIV(tpy)(bpy)O][CeIV(NO3)5]}+
(K) in Figure 8 highlighting that [Ru(tpy)(bpy)O2]+ (R) is a product of the
gas-phase fragmentation process. Remaining species are assigned as
[Ru(tpy)O]+ (S); [Ru(tpy)(N2C10H7)O]+ (T); {[Ru(tpy)(bpy)O](NO3)}+ (U);
{[Ru(tpy)(bpy)O][Ce(NO3)3]}+ (V); {[Ru(tpy)(bpy)O2][Ce(NO3)3]}+ (W);
and {[Ru(tpy)(bpy)O2][Ce(NO3)4]}+ (X).
of the other anions present in solution. This observation provided
a hint that perhaps an auxiliary reaction pathway involving
[Ce(NO3)5]- was operative, where the Ce(IV) ion plays a role
in activating one of the N-O bonds of the NO3-. It is also this
special interaction that we hold responsible for the higher rates
of electron transfer between the catalyst and CAN and/or the
stabilization of the highly oxidizing [RuVdO]3+/[RuIVdO]2+
centers.
The mass spectrum recorded under catalytic conditions (i.e,
1/CAN 1:16; t ≈ 10 min after mixing; Figure 8e) provides a
snapshot of the various intermediates that are generated in a
genuine catalytic water oxidation reaction. ESI-MS has been
used to characterize species at the end of a catalytic run,32 but
we are not aware of any other water oxidation catalyst that has
been studied in situ with this particular technique. The spectrum
under these conditions bears a close resemblance to those
recorded after the addition of 3-4 equiv of CAN; however,
there are new signals that are instructive to this analysis. For
instance, we observe a signal corresponding to the dioxygen
complex, [Ru(tpy)(bpy)O2]+ (R), which liberates dioxygen under
MS/MS conditions (Figures 9 and S14). The intensity of R is
only ∼1% of the mass spectrum, but it does provide rare
structural evidence of an intermediate that occurs late in the
catalytic cycle depicted in Figure 2. We rationalize this signal
to be a consequence of a direct O-O interaction between the
respective [RudO] and [Ce(NO3)] constituents. Similar experi-
ments were conducted on 2, and the solution-phase ESI-MS
and MS/MS of the bimetallic {[Ru(tpy)(bpy-OMe)(O)]-
[Ce(NO3)5]}+ are shown in the Supporting Information (Figures
S15 and S16). We note that dioxygen complexes of Ru and Ir
have been previously observed by ESI-MS in gas phase
reactions,65,66 but the observation of such a species from a
solution phase reaction has limited precedent.
Figure 8. Positive and negative ion electrospray mass spectrum of a
solution of 1 recorded immediately after the addition of (a) 1, (b) 2, (c) 3,
(d) 4, and (e) 16 equiv (i.e., catalytic conditions) of CAN in 0.1 M HNO3.
Insets: Expanded m/z regions showing experimental and calculated isotope
patterns of selected signals. Assignment of signals not described in figure:
[Ru(tpy)(bpy)]2+ (A); [Ru(tpy)(bpy)OH]+ (D); {[Ru(tpy)(bpy)](NO3)}+ (E);
{[Ru(tpy)(bpy)OH](ClO4)}+ (G1); {[Ru(tpy)(bpy)O][Fe(NO3)3]}+ (H);
[Fe(NO3)4]- (L); [Ce(NO3)3O]- (M); and [Ce(NO3)3(ClO4)]- (O). The
presence of [Fe(NO3)4]+ is ascribed to slight dissolution of the steel tubing
during these experiments.
signals of chemically similar compounds (e.g., [RuIVdO]2+
versus [RuIVsOO]2+).
Figure 8 shows the mass spectra of 1 measured shortly (t )
6 s) after the addition of 1, 2, 3, 4, and 16 equiv of CAN in 0.1
M HNO3. The addition of 1-2 equiv of CAN (Figure 8a-b)
reveals a solution with a mixture of the expected cationic Ru(III)
and Ru(IV) metal species that reach a charge balance with the
various anions present in solution (residual Ru(II) remains in
solution due to slow reaction times or incomplete mixing). The
spectrum obtained after the addition of 3 equiv of CAN to 1
does not reveal any signals corresponding to a [RuVdO]3+
species in experiments recorded 6 s, 1 min (Figure 8c), 60 min,
and 2 d after mixing the reagents. The addition of another
equivalent of CAN (Figure 8d) leads only to a minor change in
speciation. An interesting feature in both spectra is the over-
representation of the complex ion pair {[Ru(tpy)(bpy)O]-
[Ce(NO3)5]}+ (K). Because ClO4-, NO3-, and [Ce(NO3)4]- are
present in greater quantities than [Ce(NO3)5]- at this stage of
the reaction, we view this as an indication that the [RuIVdO]2+
fragment may have a higher affinity for [Ce(NO3)5]- than any
18OH2-Labeling Studies. Compelling evidence that there are
competing pathways to the acid-base pathway is provided by
isotopic labeling experiments in 18OH2-labeled water (Table 5).
Tracking the relative concentrations of the gases that form in
the headspace under catalytic conditions at pH 1 revealed that
1 and 2 produced a higher level of 16Od16O and lower levels
of 16Od18O and 18Od18O than would otherwise be expected if
only the acid-base mechanism is operative (reliable data for 3
could not be collected because of the formation of the
(65) Molina-Svendsen, H.; Bojesen, G.; McKenzie, C. J. Inorg. Chem. 1998,
37, 1981–1983.
(66) Thewissen, S.; Plattner, D.; Debruin, B. Int. J. Mass Spectrom. 2006,
249-250, 446–450.
9
J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010 16103

A R T I C L E S
Wasylenko et al.
Table 5. Relative Abundance of 16Od16O, 16Od18O, and 18Od18O
Generated during the Oxidation of 18OH2-Labeled Water by 1 and
findings by Sakai and our program show that a Ru-Cl bond is
not required for catalysis and the presence of coordinating
ligands (e.g., MeCN, Cl-, SO42-) actually suppresses catalytic
activity.33,48 Thus, the starting point in the proposed catalytic
cycles for 1-3 is the [RuII-OH2]2+ form of the catalyst. Our
spectrophotometric determination of relevant pKa values and
Pourbaix diagrams indicate that the first oxidation of 1 by CAN
is an electron-transfer process in a strongly acidic medium, while
the second oxidation process is accompanied by the loss of two
protons to generate [RuIVdO]2+ in solution. At higher pH levels,
these two steps would formally be consecutive PCET processes.
This scenario differs slightly from that of IV31 because of the
higher relative stability of the Ru(III) redox level for 1.
The assignment of the latter stages of the catalytic cycle
becomes significantly more arduous, particularly because reac-
tion conditions play a critical role in the reactivity. For example,
even when all other experimental parameters are held at parity,
kcat values are measured in the presence of 30 equiv of CAN,
while k1-k3 and kO-O are measured in the presence of 1-3 equiv
of CAN. These disparate conditions appear to evince different
reaction pathways. Evidence to support this includes the fact
that the reaction of 1 equiv of CAN with [RuIVdO]2+ does not
produce spectroscopically distinct trends that correspond to the
successive formation of [RuVdO]3+ and [RuIIIsOOH]2+, yet
dioxygen evolution is observed with a kcat value of 6.6 × 10-4
s-1 under catalytic conditions. We therefore contend that, at a
1/CAN ratio of 1:1, the electron-transfer reaction between
[RuIVdO]2+ and CAN does not occur (the oxidation of
[RuIVdO]2+ by CAN is a formally endothermic process; ∆G
) +32 kJ/mol) or is slower than kO-O; it is difficult to resolve
these two possibilities spectrophotometrically. Reaction step kd
may also be responsible for the slow decay in the titration
experiments but is expected to be too slow to be relevant under
catalytic conditions.
Under conditions of excess CAN, the acid-base reaction
steps all become accessible due to equilibrium considerations
(including the Nernstian effect on the Ce(IV)/Ce(III) potential)
to ultimately form [RuIVsOO]2+, which precedes the RDS, kO2
(in 0.1 M HNO3). Tracking the catalytic mixture (i.e., 1/CAN
1:30) converges on a spectral signature consistent with a Ru(IV)
species (Figure 6a). These data suggest that [RuIVdO]2+ and/
or [RuIVsOO]2+ are the dominant forms of the catalyst under
catalytic conditions, which would be expected given the
demands associated with the oxidation of [RuIVdO]2+ and the
exclusion of dioxygen from a Ru(IV) center. The spectropho-
tometric determination of kcat following the addition of 30 equiv
of CAN to 1 also follows the expression rate ) kcat[1] to support
kO2 being the RDS.
a
2 in 0.1 M HNO3
relative abundance
observedb
theoreticalc
% deviationd
1
2
16Od16O
16Od18O
18Od18O
0.90(3)
0.09(3)
0.004(1)
0.837
0.156
0.0072
8
-41
-45
16Od16O
16Od18O
18Od18O
0.862(1)
0.133(1)
0.055(1)
0.843
0.150
0.0067
2
-11
-18
a [Ru] ) (2.9-3.4) × 10-4 M; [CAN] ) 0.08-0.09 M; total volume
) 2.9-3.5 mL. b Relative ratios of dioxygen signals; standard deviations
of no less than three trials are indicated in parentheses. c Probability of
10% 18OH2-labeled water producing 32O2, 34O2, and 36O2 is 0.81, 0.18,
d
and 0.01, respectively; values in table account for dilution factors.
%
deviation ) (observed - theoretical)/theoretical ×100%; theoretical
values assume both O-atoms of dioxygen are derived from water.
precipitate). Given that the only other source of oxygen present
in the respective reaction flasks is a stoichiometric quantity of
-
-
ClO4 counterions and a significant excess of NO3 (from the
acid and CAN), we infer that the anions are involved in the
O-O bond-forming step. (Note that control experiments exclude
18O being incorporated into CAN on a relevant time scale and
the possibility that the O-atoms of bpy-OMe are introduced into
the dioxygen product during the analysis of 2.) It is primarily
on these grounds that we invoke the OAT pathway kOAT in
Figure 2. The results in Table 5 support the notion that an
O-atom is derived from a species other than water. In the case
of 2, for example, the isotopic distribution is consistent with
10-12% of the evolved dioxygen containing an O-atom from
CAN (or another source other than water) via an OAT pathway.
(Note that the relative ratios of isotopomers preclude atmo-
spheric dioxygen being a source of the large % deviations in
Table 5.)
In order to corroborate our theory of a reaction step involving
OAT, we set out to verify the exclusion of NO2 during catalysis
(see kOAT in Figure 2). While GC-MS did not prove to be an
effective method for detecting this particular species at modest
concentrations, we found laser cavity ring-down spectroscopy to
be a useful tool in this regard: NO2 was detected immediately
following the addition of catalyst to a solution of CAN in 1.0 M
HNO3 (Figure S17). Moreover, the maximum rate of NO2 evolution
tracks the maximum rate of dioxygen evolution (a short delay in
signal response is due to transit of the analyte to the sensor within
the sample chamber). In combination with the 18OH2 labeling
studies, the detection of NO2 provides strong evidence of the
existence of the O-atom transfer pathway defined in Figure 2.
A characterization of the products that form in the headspace
of the reaction flask under catalytic conditions reveals a ratio
that is not in full agreement with both O-atoms in dioxygen
being derived from water. Indeed, the isotopic distribution of
16Od16O, 16Od18O, 18Od18O in the headspace of the reaction
flask is consistent with a competing reaction pathway involving
the abstraction of an O-atom from a NO3- anion by the catalyst.
The detection of NO2 by laser cavity ring-down spectroscopy
further corroborates the kOAT pathway described in Figure 1.
Chemical insight into this auxiliary pathway is provided by the
ESI mass spectrum of 1 recorded under catalytic conditions,
which reveal a prominent signal corresponding to the complex
ion, {[Ru(tpy)(bpy)O][Ce(NO3)5]}+. The intensity of this signal,
despite the prevalence of other anions in solution, hints at an
affinity between [RuIVdO]2+ and the Ce(IV) salt; this statement
Discussion
This study builds on the mechanistic studies of homogeneous
water oxidation catalysts that have recently appeared in the
literature by establishing specifically how electron density at
the metal site affects the catalytic activity of mononuclear
octahedral polypyridyl Ru(II) complexes related to 1. Our
mechanistic studies, augmented by the use of ESI-MS as a probe
under catalytically relevant conditions, unveils a complicated
reaction landscape that expands on the prevailing acid-base
mechanism documented by Meyer et al.31 A full description of
the mechanistic details for 1-3 is addressed in sequence below.
Description of Catalytic Cycle for 1. Our data indicate that
the catalytically active forms of the Ru catalysts in this
investigation are the aqua-derived species, [Ru-OHx]. Previous
9
16104 J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010

Electronic Modification of Water Oxidation Catalysts
A R T I C L E S
is bolstered by titration experiments. Further support for an
interaction between the [RudO]2+ unit and [Ce(NO3)5]- is that
fragmentation of {[Ru(tpy)(bpy)O][Ce(NO3)5]}+ by MS/MS
results in the dioxygen species, [RuIIIsOO]+. Our interpretation
of this entire collection of data is that the Ce(IV) ion plays a
role in activating a NsO bond of a NO3-; i.e., the O-atom is
not derived from a free NO3- ion. This is supported by myriad
control experiments and isotopic labeling experiments indicating
a suppression of the OAT pathway in other acids (e.g., triflic
acid).
under catalytic conditions, the evolution of NO2 during catalysis,
and the greater than stoichiometric amounts of dioxygen formed
during the reaction (Table 4). This same alternative pathway is
available for both 1 and 3, but may be less likely to occur because
kO-O is presumably faster for these two catalysts relative to 2 due
to the electrophilicity of the [RuVdO]3+ fragment.
Description of Catalytic Cycle for 3. The first two (proton-
coupled) electron-transfer steps were found to be the fastest for
3 among the series, an observation that is attributed to a
combination of thermodynamic driving force and the electro-
static interaction between the cationic [Ru-OHn]z unit and
[CeIV(NO3)5]-. These well-defined steps are followed by the
reactivity of the [RuIVdO]2+ form that is not entirely consistent
with the behavior of 1 or 2. The [RuIVdO]2+ form not only
spontaneously decays in solution to a [RuIIIsOH]2+ species due
to an apparent disproportionation pathway, but spectral features
arise that are consistent with the formation of [RuIIsO2H2]2+
via kO-O′. We believe this is the first evidence of O-O bond
formation at a Ru(IV) site for any single-site catalyst. We ascribe
the formation of a [RuIIIsOH]2+ species to a slow kd step; the
concomitant formation of [RuVdO]3+ is followed by a rapid
kO-O step to form [RuIII-OOH]2+. The observation of dioxygen
formation in electrochemical experiments where the potential
is held at 1.7 V (i.e., below the Ru(V)/Ru(IV) potential) lends
further support to at least one of these pathways occurring in
solution.
Description of Catalytic Cycle for 2. The first two oxidation
processes for 2 are rapid and furnish the [RuIVdO]2+ species.
The third oxidation process is assigned as the generation of the
ostensible [RuVdO]3+ species, which is then poised for nucleo-
philic attack by an incoming water substrate. The rate of this
latter reaction was determined by the onset of a signal at 320
nm shortly after the addition of 1 equiv of CAN to [RuIVdO]2+
(kO-O ) 3.3 × 10-5 s-1). The lower magnitude of kO-O relative
to IV (kO-O ) 9.6 × 10-3 s-1) is aligned with a lower
electrophilicity of the [RudO] fragment due to the electron-
donating groups, but we caution that competing pathways in
solution may falsely represent the kO-O value. In the case of
IV, the [RuIIIsOOH]2+ is found to disproportionate to
[RuIIsH2O2]2+ and [RuIVsOO]2+ due to the instability of the
Ru(III) level. This decay is not observed for 2 because of the
stabilization of the Ru(III) level by the electron-rich sOMe
groups.
The observation of k3 for 2, but neither 1 nor 3, is rationalized
in part by the lower thermodynamic demands associated with
oxidizing the [RuIVdO]2+ form of 2. Because the same is true
for k5, this oxidation step late in the catalytic cycle becomes
accessible for 2; this step becomes rate-determining and follows
the rate expression rate ) kcat[2][CAN] (where kcat ) 1.0 M-1
s-1). The faster kcat value for 2 relative to 1 and 3 indicates that
the exclusion of dioxygen from a Ru(V) species is faster than
that from a Ru(IV) level; the difference may also be ascribed
to EDGs enhancing cleavage of the peroxo ligand from the Ru
complex.37 The time-dependent spectra recorded for 2 following
the addition of 30 equiv of CAN collapse to an absorption
envelope that does not match exactly the [RuIVdO]2+ species
but does bear features that would be expected for either
[RuVdO]3+ or [RuIVsOO]2+. This provides further evidence
that significant reaction barriers in the catalytic cycle involve
dioxygen formation and access to the Ru(V) level. The lower
TONs observed for 2 relative to the other catalysts when a large
excess of CAN is used can be rationalized by the buildup of
the Ru(V) level preceding the slow kO-O step; i.e., the RusNbpy
bond trans to the oxo ligand is susceptible to cleavage, and
increasingly so at higher redox levels, such that catalyst
degradation is faster for 2.48 On these same grounds, the
[RuIV-OO]2+ form of 2 may also be a point of departure in the
catalytic cycle.
Tracking the spectral changes at 322 nm following the
addition of 1 equiv of CAN to the [RuIVdO]2+ form of 3 reveals
the onset of one species at a rate of ∼2 M-1 s-1, followed by
the formation of a second species at a rate of 1 × 10-4 s-1
;
these trends are consistent with successive kd and kO-O steps.67
A separate process is also occurring that is consistent with the
formation of a [RuIIsO2H2]2+ complex (kO-O′) based on the
onset of a shoulder at 483 nm, followed by the formation of
[RuIIIsOOH]2+ (k3′). We do not rule out the possibility of a
slow kd process and comparable kO-O step, followed by a rapid
k4 step leading to [RuIVsOO]2+, which would be in resonance
with a Ru(II) dioxygen complex (eq 12). This latter species
would be stabilized by the electron-withdrawing groups and is
consistent with the feature at 486 nm.
Under catalytic conditions (i.e., 30 equiv of CAN), 3 is found
to follow the same rate expression as that for 1. Because kO-O
is presumably faster for 3 than 1, the RDS is assigned as kO2.
Under steady-state conditions, the emergence of a signal at 486
nm is observed, which we assign as the stabilization of the Ru(II)
resonance form in eq 12. Computational studies are underway
to unravel the nature of this interaction for the series.
Interaction of CAN with the Catalyst. This study raises the
issue that the interaction between the catalyst and CAN may
need to be given consideration in studies of this type. Making
it somewhat difficult to pinpoint this behavior is the fact that
the nature of CAN in solution has not been unambiguously
defined in the literature. Based on our ESI-MS experiments,
CAN appears to exist predominantly as [Ce(NO3)5]- in a HNO3
medium (Figure S18). Noting that CAN exists as [Ce(NO3)6]2-
in the solid state,68 this dianionic form does not appear to be
the dominant species in solution (we do not rule out the
possibility that [Ce(NO3)6]2- is the dominant species in solution
and undergoes charge reduction in the spectrometer). There is
Taking into account that the purported kO-O step is slower
relative to kcat, we suggest that 2 partially bypasses the acid-base
mechanism under catalytic conditions to proceed through an OAT
process with CAN via kOAT. This hypothesis is rationalized by the
fact that the Ru(V) redox level is more accessible to electron
transfer with CAN, but the [RuVdO]3+ fragment may not be
sufficiently susceptible to nucleophilic attack by water because of
the electron-donating substituents. Offering support for this hy-
pothesis is the relative ratios of the isotopically labeled dioxygen
species formed in the reaction (Table 5), which shows clear
evidence that an O-atom is derived from a source other than water
(67) These data assume appropriate second-and first-order kinetics.
(68) Dvorkin, A. A.; Krasnova, N. F.; Simonov, Y. A.; Abashkin, V. M.;
Yakshin, V. V.; Malinovskii, T. I. Kristallografiya 1984, 29, 471.
9
J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010 16105

A R T I C L E S
Wasylenko et al.
also evidence for [Ce(NO3)O]- in solution (that presumably
exists as [Ce(NO3)3(OH)] prior to charge reduction in the MS
instrument), which is more prevalent at higher pH values, our
interpretation of the data in acidic media only considers the
interaction of the catalyst with [Ce(NO3)5]-. The anionic nature
of this ion presumably helps to facilitate intermolecular interac-
tions with the positively charged catalysts in the various valent
forms. Isotopic labeling experiments show that O-atom exchange
with NO3- in the absence of CAN with or without the catalyst
at 0.1 and 1.0 M HNO3 does not occur on a relevant time scale
to influence the data presented in Table 5.
other acids (e.g., HOTf) indicates a suppression of the OAT
pathway. This result is attributed to the displacement of the NO3-
ligands about the Ce(IV) ion by the respective acid anion (e.g.,
OTf-), which effectively inhibits an interaction between the
activated [Ce(NO3)5]- complex and the [RudO] unit. These
collective observations further underscore the importance of
understanding the chemical behavior of the terminal oxidant when
assessing the mechanistic behavior of water oxidation catalysts.
Conclusions
Our experiments show that the NO3- anions affect the reactivity
of these catalysts in two ways: (i) they can oxidize Ru(II) to Ru(III)
in strongly acidic media (eqs 2-4), and (ii) they can be a source
of oxygen in the critical O-O bond-forming step by way of a
reaction with the high-valent [RudO] unit. Because control
experiments rule out dioxygen formation between the Ru com-
plexes and free NO3- anions, and because the amount of dioxygen
produced in the case of 2 testifies to NO3- being at least one source
of extrinsic oxygen, we believe the high positive charge of the
Ce(IV) ion, in tandem with the high valent [RudO] unit, is critical
to activating an NsO bond of the NO3- anion. Our kinetic studies
do not show clear evidence for the [RuIVdO]2+ forms of 1 nor 2
being sufficiently electrophilic to abstract an O-atom from
[CeIV(NO3)5]- (or [CeIII(NO3)4]-) in solution. On this basis, we
propose that it is the [RuVdO]3+ form of the catalyst that abstracts
an O-atom from [Ce(NO3)5]- (i.e., kOAT in Figure 2). This proposal
invokes the production of [CeIV(NO3)4], which presumably is
regenerated to [CeIV(NO3)5]- in the acid medium to enable
subsequent electron-transfer processes and can therefore render
catalytic efficiencies in excess of 100%. The possibility that kOAT
involves ClO4- is not wholly dismissed, but we postulate that
[CeIV(NO3)5]- is the predominantly reactive anion on the grounds
that ClO4- is not present in sufficiently high concentrations and
the relative pKa values of HNO3 and HClO4. Further support is
provided by ESI-MS studies, which show an apparent propensity
of the high-valent [RudO] forms of either 1 and 2 to form ion
pairs in solution with [Ce(NO3)5]-, despite the much higher
concentration of NO3-, and in certain cases [Ce(NO3)4]-, in solution
under catalytic conditions. The gas-phase fragmentation of this
complex ion pair renders a [Ru-OO] complex, which lends some
credence to there being a bonding interaction between the catalyst
and Ce salt. A related set of catalytic experiments involving 1 in
This study provides evidence that catalytic water oxidation
driven by CAN provides access to reaction pathways that
diverge from the prevailing “acid-base” mechanism for single-
site catalysts. The preference for each of these pathways is
dependent on reaction conditions and electron density at the
metal site. The position of the RDS is shown to be sensitive to
electron density at the Ru center and shows that higher catalytic
activity is attained when the [RuV-OO]2+ is made accessible
(i.e., RDS is kO2′ rather than kO2). However, too much density
at the metal site can compromise the critical bond formation
step, thus highlighting the inherent limitation for achieving high
catalytic rates with catalysts of this type. This study also
demonstrates that a minor auxiliary reaction pathway may be
operative that enables the incorporation of an O-atom from a
source other than water (e.g., CAN) into the dioxygen product.
These findings provide important insight into the design and
study of homogeneous water oxidation catalysts.
Acknowledgment. The authors thank Dr. Scott McIndoe
(University of Victoria) for access to ESI-MS instrumentation.
This work was financially supported by the Canadian Natural
Science and Engineering Research Council (NSERC), Canada
Research Chairs, Canada Foundation for Innovation, and Alberta
Ingenuity.
Supporting Information Available: Experimental details,
dioxygen and NO2 evolution traces, and spectrophotometric and
electrochemical data. This material is available free of charge
via the Internet at http://pubs.acs.org.
JA106108Y
9
16106 J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010