Cyclometalated iridium(III) aquo complexes: Efficient and tunable catalysts for the homogeneous oxidation of water

Article abstract of DOI:10.1021/ja074478f

A series of bis-phenylpyridine, bis-aquo iridium(III) complexes is herein shown to robustly and efficiently catalyze the oxidation of water to dioxygen in the presence of a sacrificial oxidant. Through substitution on the cyclometalating ligands of these complexes, it is shown that a broad range of oxidation potentials can be achieved within this class of catalyst. Parallel, dynamic monitoring of oxygen evolution, made possible by equipping reaction vessels with pressure-voltage transducers, facilitates correlation of these complexes' ionization potentials with their respective activity toward water oxidation. The importance of these catalysts lies in (A) their ability to oxidize water in a purely aqueous medium, (B) their simplicity of design, (C) their durability, and (D) the ease with which they can be tuned to accommodate the electrochemical needs of photosensitizers in hypothetical photochemical water oxidation and full artificial photosynthetic schemes.

Full text of DOI:10.1021/ja074478f

Published on Web 12/07/2007
Cyclometalated Iridium(III) Aquo Complexes: Efficient and
Tunable Catalysts for the Homogeneous Oxidation of Water
Neal D. McDaniel, Frederick J. Coughlin, Leonard L. Tinker, and Stefan Bernhard*
Contribution from the Department of Chemistry, Princeton UniVersity,
Princeton, New Jersey 08544
Received June 19, 2007; E-mail: bern@princeton.edu
Abstract: A series of bis-phenylpyridine, bis-aquo iridium(III) complexes is herein shown to robustly and
efficiently catalyze the oxidation of water to dioxygen in the presence of a sacrificial oxidant. Through
substitution on the cyclometalating ligands of these complexes, it is shown that a broad range of oxidation
potentials can be achieved within this class of catalyst. Parallel, dynamic monitoring of oxygen evolution,
made possible by equipping reaction vessels with pressure-voltage transducers, facilitates correlation of
these complexes’ ionization potentials with their respective activity toward water oxidation. The importance
of these catalysts lies in (A) their ability to oxidize water in a purely aqueous medium, (B) their simplicity
of design, (C) their durability, and (D) the ease with which they can be tuned to accommodate the
electrochemical needs of photosensitizers in hypothetical photochemical water oxidation and full artificial
photosynthetic schemes.
Introduction
facilitate direct investigation of the catalyst. By far the most
widely used oxidant for this purpose is the ceric ion, which,
Anticipation of increased global energy consumption has
fostered widespread desire to efficiently collect and store solar
energy in the form of dihydrogen and other fuels. Artificial
photosynthesis is one strategy for accomplishing this task. In
short, it is the goal of photosynthetic systems to photolytically
convert a common redox product (in this case water) into its
fuel and oxidant parent materials (H2 and O2), thereby storing
solar energy for later use. The development of a good artificial
photosynthetic system represents a complex challenge that
becomes more practicable when divided into its photophysical
and electrochemical components. The electrochemical aspect
of this endeavor is typically further broken down into its
oxidative and reductive half-reactions, which can be studied
independently. While much progress has been made in the field
of water reduction to evolve dihydrogen gas, the complementary
oxidative half-reaction has proven far more difficult to achieve.1-4
The challenge of water oxidation primarily lies in the implicit
complexity of mediating four highly energetic charge transfers
to catalytically obtain only the desired dioxygen product, despite
the reaction’s harshly oxidative environment.
despite its 1.72 volt oxidative potential vs NHE, is only capable
of oxidizing water very slowly under intense irradiation.5
Previously reported water oxidation catalysts for homoge-
neous and microheterogenous water splitting systems generally
fall into two categories. First, there is the oxo- and otherwise
bridged, bi-, and multinuclear transition metal complex system,
first pioneered by Meyer et al. in the early 1980s.6-22 Addition-
ally, there is a body of work that deals with turning metal oxides
with moderate band gaps into colloidal suspensions of nano-
scopic electrodes.23-29 Of particular interest to the current work
(5) Evans, M. G.; Uri, N. Nature 1950, 4223, 602-603.
(6) Gersten, S. W.; Samuels, G. J.; Meyer, T. J. J. Am. Chem. Soc. 1982, 104,
4029-4030.
(7) Gilbert, J. A.; Eggleston, D. S.; Murphy, W. R., Jr.; Geselowitz, D. A.;
Gersten, S. W.; Meyer, T. J. J. Am. Chem. Soc. 1985, 107, 3855-3864.
(8) Gilbert, J. A.; Geselowitz, D. A.; Meyer, T. J. J. Am. Chem. Soc. 1986,
108, 1493-1501.
(9) Ramaraj, R.; Kira, A.; Kaneko, M. Angew. Chem. 1986, 98, 824-825.
(10) Rotzinger, F. P.; Munavalli, S.; Comte, P.; Hurst, J. K.; Gra¨tzel, M.; Pern,
F.; Frank, A. J. J. Am. Chem. Soc. 1987, 109, 6619-6626.
(11) Wiedghardt, K. Angew. Chem. 1989, 101, 1179-1198.
(12) Yagi, M.; Tokita, S.; Nagoshi, K.; Ogino, I.; Kaneko, M. J. Chem. Soc.,
Faraday Trans. 1996, 92, 2457.
(13) Chronister, C. W.; Binstead, R. A.; Ni, J.; Meyer, T. J. Inorg. Chem. 1997,
36, 3814-3815.
The first step in homogeneous, photodriven water oxidation
is activation of the oxidation catalyst through interaction with
the energetic, oxidized state of a photosensitizer molecule, such
as [Ru(bpy)3]3+ or [Ir(ppy)2(bpy)]2+ (bpy ) 2,2′-bipyridine, ppy
) 2-phenylpyridine). However, to avoid complications like
photosensitizer decomposition, a single-electron, sacrificial
oxidant is typically employed in place of the photosensitizer to
(14) Limburg, J.; Brudvig, G. W.; Crabtree, R. H. J. Am. Chem. Soc. 1997,
119, 2761-2762.
(15) Lebeau, E. L.; Adeyemi, S, A.; Meyer, T. J. Inorg. Chem. 1998, 37, 6476-
6484.
(16) Wada, T.; Tsuge, K.; Tanaka, K. Inorg. Chem. 2001, 40, 329-337.
(17) Sens, C.; Romero, I.; Rodriguez, M.; Llobet, A.; Parella, T.; Benet-
Buchholz, J. J. Am. Chem. Soc. 2004, 126, 7798-7799.
(18) Shimazaki, Y.; Nagano, T.; Takesue, H.; Ye, B.; Tani, F.; Naruta, Y. Angew.
Chem., Int. Ed. 2004, 43, 98-100.
(19) Zong, R. Z.; Thummel, R. P. J. Am. Chem. Soc. 2005, 127, 12802-12803.
(20) Baffert, C.; Romain, S.; Richardot, A.; Lepreˆtre, J.; Lefebvre, B.; Deronzier,
A.; Collomb, M. J. Am. Chem. Soc. 2005, 127, 13694-13704.
(21) Narita, K.; Kuwabara, T.; Sone, K.; Shimizu, K.; Yagi, M. J. Phys. Chem.
B 2006, 110, 23107-23114.
(22) Liu, F.; Cardolaccia, T.; Hornstein, B. J.; Schoonover, J. R.; Meyer, T. J.
J. Am. Chem. Soc. 2007, 129, 2446-2447.
(1) Meyer, T. J. Acc. Chem. Res. 1989, 22, 163-170.
(2) Ru¨ttinger, W.; Dismukes, G. C. Chem. ReV. 1997, 97, 1-24.
(3) Hurst, J. K. Coord. Chem. ReV. 2005, 249, 313-328.
(4) Alstrum-Acevedo, J. H.; Brennaman, M. K.; Meyer, T. J. Inorg. Chem.
2005, 44, 6802-6826.
9
210
J. AM. CHEM. SOC. 2008, 130, 210-217
10.1021/ja074478f CCC: $40.75 © 2008 American Chemical Society

Cyclometalated Iridium(III) Aquo Complexes
A R T I C L E S
Figure 2. Structure of the described oxidation catalyst, [Ir(5-R1,4′-R2,2-
phenylpyridine)2(OH2)2]+.
Figure 1. (A) Ruthenium dimer generated by Gra¨tzel et al. in 1986,10 and
(B) three diruthenium complexes synthesized for water oxidation by
Thummel in 2005.19
electron withdrawing and electron-donating substituents, which
serve to tune the HOMO energies of the respective iridium
complexes, allowing us to control their oxidation potentials. The
preparation of the parent (unsubstituted) species, first reported
by Schmid et al. in 1994,30 is straightforward and is herein
proven versatile enough to generate aquo complexes with a wide
range of cyclometalating ligands.
The kinetics of these water-oxidizing dark reactions are
monitored via pressure transducers, which are mounted into the
headspace of the reaction vessels. The strength of this approach
is its simplicity, which renders analysis and quantization
relatively unproblematic while providing a detailed picture of
how the several reactions proceed, as well as mechanistic
insights for the complicated chemistry that ensues.31 Dynamic
monitoring of these O2-evolving reactions is crucial to under-
standing these systems and may aid ongoing efforts to pair such
catalysts with appropriate water reduction analogues in a
complete photosynthetic system.
is the study performed by Gra¨tzel and co-workers10 in which a
bis(aquo), bis(5,5′-dicarboxylated-2,2′-dipyridyl) ruthenium com-
plex was oxidized to form an oxo-bridged binuclear catalyst
for water oxidation, very similar to the mixed valence, blue
ruthenium dimer originally developed by Meyer et al.6 Not only
is Gra¨tzel’s study one of the few to report dynamic monitoring
of oxygen evolution, with a maximum O2 evolution rate of
300 µL per hour, but they also report a turnover number of 75
when using Co(III) as an electron acceptor. More recently,
Thummel and Zong described another class of diruthenium
polypyridyl complexes that are active for water oxidation, with
a reported maximum turnover number of 3200 for the electroni-
cally optimized case.19 The active structures identified in the
Gra¨tzel and Thummel examples are depicted in A and B of
Figure 1, respectively.
Notably, the primary polypyridyl ligand in Thummel’s
complex confines the two ruthenium molecules to a well-defined
geometry, and electrochemical tunability may be possible by
virtue of the substituents on the attached pyridine groups.
Despite this system’s enhanced catalytic activity, however, it
also has its disadvantages, such as an increased complexity of
designsespecially the primary ligand’s synthesis. In short, there
is no existing synthetic catalyst for homogeneous water oxida-
tion that is simultaneously simple, robust, and effective.
An alternative, cyclometalated iridium-based compound
(structure shown in Figure 2) is presented as a synthetically
accessible, robust, and efficient solution to some of the problems
with existing catalysts. It is likewise demonstrated that the ligand
structure of this species can be easily modified with various
Experimental Section
General. 1H NMR spectra were recorded on a Bruker BioSpin
Avance-500 MHz spectrometer at room temperature. Mass spectra (MS)
were obtained using a Hewlett-Packard Electrospray (ESI) MS engine.
Gas chromatography (GC) was performed with an argon carrier gas
through a 12′ × 1/4′′ column packed with 13×, 80/100 mesh molecular
sieve (Alltech). This was incorporated into a Perkin-Elmer Model 3920
gas chromatograph with a thermal conductivity detector.
Materials. The cyclometalating ligands and cyclometalated iridium
dimers were prepared as reported by Lowry et al.32
Synthesis of [Ir(ppy)2(H2O)2]+ and Analogues. The aquo com-
plexes of cyclometalated iridium(III) dimers were prepared in a similar
fashion to the synthesis employed by Schmid (see Scheme 1).30 In place
of Schmid’s methanol/dichloromethane solvent system, a 9:1 mixture
of ethanol/water was used to achieve the desired conversion in yields
ranging from 44 to 76%. Specifically, the appropriate cyclometalated
iridium chloride-bridged dimers were added to 9:1 mixtures of ethanol
(23) Shafirovich, V. Y.; Shilov, A. E. Kinet. Katal. 1979, 20, 1156-1162.
(24) Harriman, A.; Richoux, M.; Christensen, P. A.; Mosseri, S.; Neta, P. J.
Chem. Soc., Faraday Trans. 1 1987, 83, 3001-3014.
(25) Nahor, G. S.; Neta, P. J. Phys. Chem. 1988, 92, 4499-4504.
(26) Nahor, G. S.; Hapiot, P.; Neta, P.; Harriman, A. J. Phys. Chem. 1991, 95,
616-621.
(27) Hara, M.; Waraksa, C. C.; Lean, J. T.; Lewis, B. A.; Mallouk, T. E. J.
Phys. Chem. A 2000, 104, 5275-5280.
(30) Schmid, B.; Garces, F. O.; Watts, R. J. Inorg. Chem. 1994, 33, 9-14.
(31) Tinker, L. L.; McDaniel, N. D.; Curtin, P. N.; Smith, C. K.; Ireland, M. J.;
Bernhard, S. Chem.-Eur. J. 2007, 13, 8726-8732.
(32) Lowry, M. S.; Hudson, W. R.; Pascal, R. A.; Bernhard, S. J. Am. Chem.
Soc. 2004, 126, 14129-14135.
(28) Chen, G.; Delafuente, D. A.; Sarangapani, S.; Mallouk, T. E. Catal. Today
2001, 67, 341-355.
(29) Hara, M.; Lean, J. T.; Mallouk, T. E. Chem. Mater. 2001, 13, 4668-4675.
9
J. AM. CHEM. SOC. VOL. 130, NO. 1, 2008 211

A R T I C L E S
McDaniel et al.
Scheme 1. General Procedure for Cleaving the Chloride-Bridged Dimers to Procure the Aquo Compounds 1-5
Table 1. Yields of Purified 1-5 Resulting from Cleavage of the
Respective Chloride-Bridged Dimers with Silver
Trifluoromethanesulfonate
Computational Studies. Hybrid density functional calculations
(B3LYP/LANL2DZ) were performed using Gaussian 03.33 Whereas
the default parameter was used for gradient convergence, the threshold
for wave function convergence was relaxed to CONVER)7, due to
the presence of the heavy metal. Geometry optimizations for both singlet
and doublet geometries were carried out under the constraint of C2
symmetry.
Results and Discussion
Dioxygen-generating reactions were initiated by injecting
aqueous solutions of cerium(IV) ammonium nitrate (CAN) into
larger volumes of degassed aqueous solutions containing the
aquo-iridium complex. Despite generation of O2 from water and
consequential discharge of H+, the pH of these 10 mL reactions
remains roughly constant at 0.7 ( 0.05 throughout the reactions’
duration. Because the CAN was added as a standardized aqueous
solution, the dominating species of ceric ion was presumably
the aquo/hydroxo species,5 and the constant pH is then explained
by liberation of OH- upon reduction of Ce(IV) to Ce(III). Gas
evolution was monitored by pressure transducer and later
confirmed to be dioxygen by GC analysis. Control experiments
involved allowing solutions of cerium(IV) to stand under
identical reaction conditions as experienced by the catalyzed
reactions and did not produce measurable oxygen during the
typical 30-hour reaction window.
(Pharmco, >99.5%) and water to make 0.01 M slurries. A total of 2.2
molar equiv of silver trifluoromethanesulfonate (Acros, >99%) was
added to the slurries, after which the reaction mixtures were refluxed
for 24 h. The product solutions were then concentrated under rotary
evaporation, after which the product was extracted from the solids using
dichloromethane (EM, 99.5%). After filtering the remaining silver salts,
the solvent was once again removed under vacuum. Products were
recrystallized by vapor diffusion from chloroform/pentane. Yields of
purified products are reported in Table 1, whereas characterizations
1
are described in the Supporting Information, including H NMR, 13C
To ensure that the formed dioxygen originates from water
rather than from the nitrate of CAN, ceric trifluoromethane-
sulfonate (OTf-) was employed as an oxidant. It was found
that the use of Ce(OTf)4 in place of CAN also leads to oxygen
production, confirming that water is being oxidized in these
reactions. Notably, substituting Ce(OTf)4 for CAN reduces the
rate of the reaction by an order of magnitude. Likewise, adding
an equivalent molar concentration (roughly 0.6 molar) of
trifluoromethanesulfonic acid to aqueous CAN has a similar
detrimental kinetic effect. Meanwhile, addition of an equivalent
molar concentration of fluoroboric acid to aqueous CAN has
negligible adverse effect on the rate of oxygen evolution,
suggesting that the OTf- counterion somehow impedes the
reaction. Additions of known ligating solvents such as aceto-
nitrile and dimethylsulfoxide were found to similarly disrupt
oxygen evolution. One final control was run in which the
coordinating waters in 1 were replaced by an inert bipyridine
moiety, the synthesis of which has been previously described.32
To aqueous CAN was added the mock catalyst [Ir(ppy)2(bpy)]+,
structure shown in Figure 3. No oxygen evolved from this
mixture, suggesting that aquo coordination is essential. With
this in mind, it is thus likely that free coordination sites on the
iridium catalyst are a necessary condition for these reactions,
precluding the use of strongly coordinating solvents or coun-
terions in water oxidation reactions.
NMR, low-resolution MS, and elemental analysis.
Electrochemical Analyses. All cyclic voltammograms were recorded
on a CH-Instruments Electrochemical Analyzer Model 600C. Studies
in water were performed under neutral conditions using a 25 mm2
indium tin oxide working electrode, a platinum coil counter electrode,
and an Acumet SCE reference electrode. Studies in acetonitrile used a
1 mm2 platinum working electrode, a platinum coil counter electrode,
and a silver wire as a pseudo-reference electrode. Ferrocene (Aldrich)
was then employed as an internal standard in acetonitrile, and its
standard oxidation half-wave potential was taken to be 610 mV against
the normal hydrogen electrode. Solutions for voltammetry were prepared
with 0.1 M supporting electrolyte-sodium tetrafluoroborate (Aldrich,
98%) for water-based studies, and tetra-n-butylammonium hexafluo-
rophosphate (Fluka, electrochemical grade) for analyses in acetonitrile.
Solutions of 1-5 were prepared at 250 µM concentrations in either
solvent, with the exception of 3, which was only slightly soluble in
water. Solutions were degassed with N2 bubbling for 10 min before
each voltammogram.
Dynamic Oxygen Evolution Measurements. All oxygen evolving
reactions were contained by 40 mL EPA vials (VWR, Traceclean). The
caps of these vials were modified to house pressure transducers (Omega
PX-138-015D5V) for dynamic monitoring of the headspace pressure
above each reaction in parallel. The caps were also equipped with sep-
tum seals, which facilitated degassing and reagent injections. All
reactions were carried out under dark conditions at 25 °C, under argon
headspace. The exact setup and procedure used to monitor these
reactions is detailed in the Supporting Information. The output from
the pressure transducers was scaled by GC analysis of the reactions’
headspace.
(33) Frisch, M. J.; et al. Gaussian 03, revision C.02; Gaussian,Inc.: Wallingford,
CT, 2004.
9
212 J. AM. CHEM. SOC. VOL. 130, NO. 1, 2008

Cyclometalated Iridium(III) Aquo Complexes
A R T I C L E S
Figure 3. Structure of [Ir(ppy)2(bpy)]+, used as a surrogate catalyst in a
control experiment to confirm the importance of free coordination sites.
Figure 5. Cyclic voltammograms from a 250 µM solution of 1 in water,
as well as the solvent baseline (top half). Scan rates in water were 0.1 V/s.
The voltammograms on the bottom half are derived from a 250 µM solution
of 1 in acetonitrile, at scan rates of 0.1 and 10 V/s. Supporting electrolyte
was 100 mM NaBF4 for water studies and 100 mM N(n-Bu)4PF6 for
acetonitrile CVs.
it is clear that an in situ alteration in the catalyst must be
responsible for the early changes in rate, suggesting that perhaps
multiple active species are present throughout the course of the
reaction. The specific nature of this change is currently under
investigation.
Substitution on the Phenylpyridine. To explore the elec-
trochemical tunability of this catalyst, several structurally altered
derivatives were prepared for comparison to the parent complex.
Static DFT calculations (B3LYP/LANL2DZ) aided in the ligand
selection process for this demonstration. These calculations are
discussed in further detail below. Ultimately, several complexes
were chosen and prepared from substituted phenylpyridines and
then analyzed for their electrochemical properties and ability
to catalyze water oxidation. In particular, substitutions were
made at the 4′ position on 5-methyl-2-phenylpyridine, and the
resulting ligands were then used to form the respective cyclo-
metalated iridium bis(aquo) complexes. These substituents
ranged from electron-withdrawing groups like fluorine to
electron-donating groups such as phenyl and are identified in
Figure 2.
Cyclic voltammograms (CVs) in both water and acetonitrile
were used to evaluate the oxidation potentials associated with
these various species. For a number of reasons, aqueous
voltammetry proved problematic with these compounds. Several
working electrodes were employed, including gold, platinum,
glassy carbon, and indium tin oxide surfaces. Most of these
materials gave uninterpretable voltammograms due to surface
adsorption of the catalyst onto the working electrode. However,
ITO gave reasonably clean CVs under neutral aqueous condi-
tions despite the surface adsorption problem.
Figure 4. Pressure vs time curves for dioxygen evolution from solutions
of 1 and cerium(IV). CAN (1720 µmol) was contained in these 10 mL
aqueous reactions along with the specified amount of catalyst, bringing the
theoretical maximum O2 yield to 430 µmol for each experiment. At higher
catalyst concentrations, the rate of reaction is observed to fall from its
original rate, possibly due to competitive side reactions or an undesirable
change in the catalyst’s structure.
Rate Study. To better understand the mechanism of oxygen
evolution, a series of parallel experiments were performed in
which the reactant concentrations were varied. The headspace
pressures above these reactions were monitored dynamically,
and the GC-confirmed results are plotted in Figure 4. In each
of the four experiments shown, 1.7 mmol of CAN from a stock
solution were allowed to react with water in the presence of a
varied amount of compound 1. These reactions took place in a
completely aqueous medium with a total solution volume of
10 mL. The reactions were allowed to proceed to completion,
after which all evolved product volumes were within 4% of
the theoretical maximum yield of 430 µmol of O2.
One interesting phenomenon that appears in these traces is
the unusual kinetic progression that occurs after a half hour of
reaction. In particular, the shape of these kinetic traces can be
organized into two regimes. The first section involves a constant,
linear evolution of oxygen from the system. A comparison
between these initial slopes, which reveals a non-first order
dependence on catalyst concentration, can be found in the
Supporting Information. The initial linear kinetic reaction
proceeds for roughly 30 min before advancing to a parabolic
curve. This parabola is then maintained until the cerium
concentration dwindles, at which point the reaction tapers off
and finishes. Whereas the shape of this second regime can be
rationalized by assuming a general catalytic mechanism, the
reaction’s initiation and change of mechanism is more diffi-
cult to rationalize. With constant pH and roughly invariable
[Ce(IV)] during the initial stages of the O2 evolution process,
Other complications included differing solubilities of 1-5
in water and solvent oxidation at reduced onset potentials, which
tended to obscure the interesting Ir(III/IV) oxidation wave and
invalidate comparisons of current between CVs. Shown in
Figure 5 is the neutral, aqueous voltammogram of 1 as well as
the solvent baseline. Also shown for comparison is the CV of
1 in acetonitrile on a separate axis. Markedly, the Ir(III/IV) wave
is concealed for aqueous CVs by oxidation of the solvent, to
the extent that deconvolution of the overlapping waves becomes
problematic.
9
J. AM. CHEM. SOC. VOL. 130, NO. 1, 2008 213

A R T I C L E S
McDaniel et al.
Figure 6. Depicted on the left is dynamic oxygen evolution from 10 mL aqueous solutions containing 3 µmol of the substituted catalysts and 1720 µmol
of ceric ammonium nitrate. Though unbuffered except by the ceric ion, the pH of 0.7 ( 0.05 did not change appreciably during these reactions, presumably
due to the sequential release of OH- upon reduction of Ce(IV) to Ce(III) and liberation of H+ upon oxidation of water. On the right, the initial O2-evolution
rate from each of these curves (measured via linear least-squares regression of the oxygen generated during the first 45 min) is plotted against the oxidation
potentials of these catalysts in CH3CN, as determined by cyclic voltammetry.
nylpyridine substitution would be equivalent with either water
or acetonitrile at the ancillary positions.
Negligible rate-related isotope effect was observed upon
solvent substitution of D2O for H2O under constant pH 0.7
conditions (pH corrected for D2O). This does not preclude the
possibility that further processes involving proton-coupled
charge transfers also occur during catalyzed water oxidation,
but currently there is no evidence to suggest that this is the
case for the rate-limiting step of these reactions.
Characterizations in acetonitrile were thus used to observe
the relevant Ir(III/IV) electrochemical processes independently
Figure 7. Week-long oxygen evolution is shown from 10 mL aqueous
solutions containing 0.5 µmol of 1, 2, and 4 (the three most efficient
catalysts) and an excessive 15 000 µmol of ceric ammonium nitrate, along
with a catalyst-free control. The axis on the left measures micromoles of
oxygen produced, whereas the axis on the right gives the conversion to
turnover numbers. Analysis of the headspace gas revealed that 1240, 1140,
and 1380 µmol of oxygen were generated by 1, 2, and 4, respectively. These
correspond respectively to maximum catalytic turnover numbers of 2490,
2270, and 2760.
of the more complex steps that are necessarily involved when
aqueous conditions are employed. Oxidation waves were
observed for each of the five compounds, at potentials ranging
from 1.20 to 1.74 V vs NHE. Two such waves are shown in
the bottom half of Figure 5, performed on the parent complex
1 at sweep rates of 0.1 and 10 V/s. Notably, the shapes of these
scans are quite different; the faster scan is more reversible than
the slower scan, indicative of an EC mechanism. These results,
as well as comparison to density functional theory predictions
(discussed below), are summarized in Table 2.
Although this prevents a well-defined measurement of aque-
ous Ir(III/IV) oxidation potentials, there is nevertheless a rela-
tionship between aqueous oxidation onsets and the Ir(III/IV)
waves measured in acetonitrile. The absolute potentials are
shifted dramatically in going from one solvent to the other, due
both to the altered polarity of the medium as well as to dynamic
exchange between ancillary ligands and the solvent. However,
the observed effect from adding electron donating/withdrawing
substituents to the cyclometalated phenylpyridines is reasonably
predictable upon considering the molecular orbitals of these
complexes. The bottom of Figure 8 shows the calculated highest
occupied molecular orbitals (HOMOs) for molecules 3 and 4.
No electron density resides on the ancillary ligands in these
orbitals but rather is entirely located on the metal center and
cyclometalated phenylpyridine ligands of the complexes. This
agrees well with an earlier report by Goldsmith et al.36 in which
the HOMO of a similar complex (structure shown in Figure 3)
is also shown to be roughly independent of the ancillary groups.
It would not be unexpected, then, that the effect from phe-
These compounds were then tested for catalytic activity
toward water oxidation by injecting 1.66 mmol of CAN into
300 µM solutions of the catalysts. In the case of the least soluble
catalyst, 3, the catalyst did not dissolve fully until addition of
the oxidant. Oxygen was observed to evolve from all five of
these reactions at varying rates. Their respective catalytic
activities were quantified by calculating the maximum slope
from the reactions’ dynamic pressure traces (confirmed to be
oxygen evolution). Both the original pressure traces as well as
the initial catalytic rates, plotted against the catalysts’ measured
oxidation potentials, can be found in Figure 6. These initial rates
were determined via linear least-squares regression of O2
evolution from the first 45 min of the reaction. Compound 5 is
observed to catalyze water oxidation at a comparatively reduced
rate. This is believed to be a result of insufficient overpotential
involved in the charge transfer between the Ir(III) center and
the Ce(IV) ion, likely to be the initial step of water oxidation.
Stability of the Catalyst. To determine how many catalytic
cycles the catalysts can endure, 0.5 µmol of 1, 2, and 4 were
added to 15 000 µmol of aqueous CAN, after which the reaction
mixture was put under vacuum and allowed to react for several
(34) Reiss, H.; Heller, A. J. Phys. Chem. 1985, 89, 4207-4213.
(35) Cossi, M.; Iozzi, M. F.; Marrani, A. G.; Lavecchia, T.; Galloni, P.; Zanoni,
R.; Decker, F. J. Phys. Chem. B 2006, 110, 22961-22965.
(36) Goldsmith, J. I.; Hudson, W. R.; Lowry, M. S.; Anderson, T. H.; Bernhard,
S. J. Am. Chem. Soc. 2005, 127, 7502-7510.
9
214 J. AM. CHEM. SOC. VOL. 130, NO. 1, 2008

Cyclometalated Iridium(III) Aquo Complexes
A R T I C L E S
Figure 8. Frontier molecular orbital diagrams for compounds 3 and 4, generated from DFT calculations, as well as rendered 95% surfaces of selected
orbitals. Due to the mixed metal/π-system character of these orbitals, the addition of electron-donating or electron-withdrawing substituents (phenyl on the
left, fluorine on the right) can be used to adjust the orbital energies of these compounds. Of particular interest is the ability to fine-tune the energy of the
HOMO, which is strongly involved in the catalysis of water oxidation.
Table 2. Calculated and Measured Ionization Potentials for Removal of an Electron from Compounds 1-5
a All calculations were corrected by addition of a constant to bring the calculated potentials of 1 into agreement with its observed ionization potential.
days. Compounds 3 and 5 were omitted from this experiment
due to their relatively slow reaction kinetics. The pressure traces
of these reactions are shown in Figure 7. After a week, the
reactions’ headspaces were analyzed for oxygen by GC injec-
tion, and it was determined that 1240, 1140, and 1380 µmol of
oxygen were generated by 1, 2, and 4, respectively. This evolved
product gas represents roughly one-third of the expected 3750
µmol for complete conversion. In addition, trace CO2 was
detected by MS, indicating that the catalyst’s ligand sphere was
at least partially oxidized by the end of the reaction. These
values correspond to catalytic turnover numbers of 2490 for 1,
2270 for 2, and 2760 for 4.
Density Functional Theory Calculations. The electrochemi-
cal tunability of this catalyst architecture is a major advantage
for its practical incorporation into photosynthetic schemes. This
electronic flexibility would not be possible if the highest
occupied molecular orbitals (HOMOs) of these complexes did
not demonstrate mixed ligand/metal character, that is, the
cyclometalating nature of the ppy ligands allows them to interact
strongly with the metal center, forming frontier molecular
9
J. AM. CHEM. SOC. VOL. 130, NO. 1, 2008 215

A R T I C L E S
McDaniel et al.
Figure 9. Static DFT calculations were used as a predictive tool for determining an appropriate set of cyclometalating ligands for an electronic tunability
demonstration. Complexes generated from the chosen ligands exhibit properties that correlate well with their predicted values. Two methods were used to
make these predictions: in method 1 (left), the geometry-optimized HOMO energies of the various complexes were calculated and used to estimate their
ionization potentials, disregarding any electronic or geometric rearrangements that might occur in the oxidized species. In method 2 (right), both the ground
state singlet and oxidized doublet species were geometrically/electronically optimized, and the difference between their predicted absolute energies was used
as an estimation of their ionization potentials.
orbitals with the iridium that incorporate both the metal’s d-orbi-
tals and the ppy’s π-system. In general, by adding substituents
to the phenyl ring of the ppy, the ligands become either more
electron-withdrawing or electron-donating, thereby respectively
stabilizing or destabilizing the HOMO. Shown in Figure 8 are
the molecular orbital diagrams for compounds 3 (left) and 4
(right), whose HOMOs are known by cyclic voltammetry to
have a ∼0.5 V difference in potential. These MO diagrams, as
well as the 95% surfaces depicted for selected MOs, were
generated through static density functional theory calculations
for the geometry-optimized singlet ground states of each species.
As shown in the figure, their predicted ligand-field splitting
potentials are very similar, but the mixed orbitals of 4 are
expected to be lower in energy than those of 3, due to electronic
stabilization from the withdrawing fluorine substituent.
Two computational methods were used to extract meaningful
predictions from these calculations. In particular, “method 1”
and “method 2,” described below, were used to arrive at
expected relative ionization potentials for the complexes of
interest. Both were found to give accurate predictions for this
observable. In each of the offered methods, referencing to
standardized electrodes was achieved by shifting all predicted
values by a common constant, as is the general protocol for
these comparisons.34,35 The results of these calculations are
summarized both in Table 2 and Figure 9.
Method 1: As a rough but efficient predictor of ionization
potentials, the ground state, singlet geometries of the desired
complexes were optimized, and the energies of their HOMOs
were compared. To reference these values to the normal
hydrogen electrode, a common 6.96 V constant was subtracted
from the free ions’ calculated HOMO energies. The results are
plotted in Figure 9 (left side) against the voltammetrically
observed ionization potentials. Notably, a linear least-squares
regression through the data gives a respectable Pearson factor
of 0.90, with a slope of roughly unity.
Method 2: In a more extensive approach, the ionization
potentials were estimated by taking the difference in total
calculated energy between the geometry-optimized ground state
and first oxidized state of the compounds. Specifically, geometry
optimization was performed on both the singly charged, lowest
energy singlet configuration, as well as the doubly charged,
oxidized doublet configuration, prior to energetic comparison.
To reference the predictions to NHE, a common 7.92 V constant
was subtracted from the raw energy differences of the free ions.
The results of this calculation are likewise plotted against the
measured potentials in Figure 9 (right side). The Pearson factor
for these data is superior to that generated using Method 1, but
the slope of the regression line is farther from unity. In general,
one would expect Method 2 to better predict the ionization
potentials of a given structure, though it is also a lengthier
calculation.
Conclusions
This report details the homogeneous, chemically driven
oxidation of water by a class of easily synthesized cyclometa-
lated iridium(III) aquo complexes. These compounds have four
notable selling points as water-oxidation catalysts: First, their
design is quite simple, which makes their synthesis compara-
tively economical. Second, their ligand framework is bound by
a strong carbon-iridium bond, making them exceptionally robust
under typical reaction conditions. Third, they are water soluble
at various pH, eliminating the need for mixed solvent systems.
Fourth, their HOMO energies are herein shown to be highly
tunable through substitution on the cyclometalated ligands,
which allows us to selectively tune their oxidative potentials.
As previously shown, the HOMO and LUMO energy levels of
similar heteroleptic polypyridyl iridium(III) luminophores are
likewise highly tunable through ligand design.32,36,37 In particu-
lar, it is now known that the cyclometalating ligands in these
systems play a large role in determining the HOMO energy of
the complexes. Thus, the ability to tune the electronics of both
the catalyst and the photosensitizer may facilitate ongoing efforts
to couple these two species in photodriven half reactions, or
even full water-splitting schemes.
Acknowledgment. This work was supported by the National
Science Foundation (CAREER Award No. CHE-0449755). We
also thank Doctors Andrew Bocarsly and Robert A. Pascal, Jr.
for useful discussions.
Supporting Information Available: The detailed protocol for
dynamic monitoring and quantification of oxygen evolution, as
well as the structural characterizations of the compounds of
(37) Lowry, M. S.; Goldsmith, J. I.; Slinker, J. D.; Rohl, R.; Pascal, R. A.;
Malliaras, G. G.; Bernhard, S. Chem. Mater. 2005, 17, 5712-5719.
9
216 J. AM. CHEM. SOC. VOL. 130, NO. 1, 2008

Cyclometalated Iridium(III) Aquo Complexes
A R T I C L E S
interest (inlcuding 1H NMR, 13C NMR, electrospray ionization
MS, and elemental analyses) are provided. Also included is
a detailed comparison between the initial rates of the reac-
tions shown in Figure 4 and the complete citation for ref 33.
This material is available free of charge via the Internet at
http://pubs.acs.org.
JA074478F
9
J. AM. CHEM. SOC. VOL. 130, NO. 1, 2008 217