Homogeneous catalytic system for photoinduced hydrogen production utilizing iridium and rhodium complexes

Article abstract of DOI:10.1021/ic800988b

An efficient homogeneous catalytic system for the visible-light-induced production of hydrogen from water utilizing cyclometalated iridium(III) and tris-2,2′-bipyridyl rhodium(III) complexes is described. Synthetic modification of the photosensitizer lr(C∧N)₂(N∧N)⁺ and water reduction catalyst Rh(N∧N)₃³⁺ creates a family of catalysts with diverse photophysical and electrochemical properties. Parallel screening of the various catalyst combinations and photoreaction conditions allows the rapid development of an optimized photocatalytic system that achieves over 5000 turnovers with quantum yields (1/2 H₂ per photon absorbed) greater than 34%. Photophysical and electrochemical characterization of the optimized system reveals that the reductive quenching pathway provides the necessary driving force for the formation of [Rh(N∧N)₂]⁰, the active catalytic species for the reduction of water to produce hydrogen. Tests for system poisoning with mercury or CS₂ provide strong evidence that the system is a true homogeneous system for photocatalytic hydrogen production.

Full text of DOI:10.1021/ic800988b

Inorg. Chem. 2008, 47, 10378-10388
Homogeneous Catalytic System for Photoinduced Hydrogen Production
Utilizing Iridium and Rhodium Complexes
Eric D. Cline, Samantha E. Adamson, and Stefan Bernhard*
Frick Laboratory, Department of Chemistry, Princeton UniVersity, Princeton, New Jersey 08544
Received May 30, 2008
An efficient homogeneous catalytic system for the visible-light-induced production of hydrogen from water utilizing
cyclometalated iridium(III) and tris-2,2′-bipyridyl rhodium(III) complexes is described. Synthetic modification of the
∧
∧
+
∧
3+
3
photosensitizer Ir(C N) (N N) and water reduction catalyst Rh(N N)
2
creates a family of catalysts with diverse
photophysical and electrochemical properties. Parallel screening of the various catalyst combinations and photoreaction
conditions allows the rapid development of an optimized photocatalytic system that achieves over 5000 turnovers
with quantum yields (1/
2
2
H per photon absorbed) greater than 34%. Photophysical and electrochemical characterization
of the optimized system reveals that the reductive quenching pathway provides the necessary driving force for the
formation of [Rh(N N)
2
]0, the active catalytic species for the reduction of water to produce hydrogen. Tests for
∧
system poisoning with mercury or CS
photocatalytic hydrogen production.
2
provide strong evidence that the system is a true homogeneous system for
1
1-15
Introduction
neous
systems. Catalytic systems based on transition-
metal complexes involve multiple components, with the
distinct advantage that each component can be optimized
for its specific task through synthetic modification. The
classic example of a heterogeneous system for water reduc-
As researchers look for efficient methods to harness
renewable clean energy from sustainable sources, hydrogen
is a promising candidate as a medium for storage of this
energy. Direct conversion of solar energy to chemical energy
in the form of hydrogen is an attractive goal that can be
accomplished by photosynthetic means from one of Earth’s
2+
tion uses tris-(2,2′-bipyridine)ruthenium(II), [Ru(bpy)
the photosensitizer (PS), methyl viologen as the electron relay
ER), colloidal platinum as the redox mediator, and EDTA
3
] , as
(
1
most plentiful resources, water. Designing a complete water-
4
as the SR. Heterogeneous systems depend on many param-
eters that are difficult to control and their potential application
in a complete water-splitting system is limited because most
heterogeneous catalysts, such as colloidal platinum, promote
splitting system is a daunting task, and researchers generally
develop systems for the oxidative or reductive half-reaction
exclusively by replacing the other half-reaction with an
appropriate sacrificial reductant (SR) or oxidant.
2 2
the recombination of the O and H products. Thus, the
Photosynthetic methods for hydrogen production from the
reduction of water have been examined in several embodi-
(
7) Brown, G. M.; Brunschwig, B. S.; Creutz, C.; Endicott, J. F.; Sutin,
N. J. Am. Chem. Soc. 1979, 101, 1298–1300.
2,3
ments, including semiconductor-based devices and transi-
4
-10
(8) Amouyal, E.; Koffi, P. J. Photochem. 1985, 29, 227–242.
tion metal complexes in heterogeneous
or homoge-
(9) Du, P.; Schneider, J.; Jarosz, P.; Zhang, J.; Brennessel, W.; Eisenberg,
R. J. Phys. Chem. B 2007, 111, 6887–6894.
(10) Tinker, L.; McDaniel, N.; Curtin, P.; Smith, C.; Ireland, M.; Bernhard,
S. Chem.sEur. J. 2007, 13, 8726–8732.
(11) Krishnan, C. V.; Brunschwig, B. S.; Creutz, C.; Sutin, N. J. Am. Chem.
Soc. 1985, 107, 05–2015.
(12) Brewer, K. J.; Murphy, W. R.; Moore, K. J.; Eberle, E. C.; Petersen,
J. D. Inorg. Chem. 1986, 25, 2470–2472.
(13) Goldsmith, J. I.; Hudson, W. R.; Lowry, M. S.; Anderson, T. H.;
Bernhard, S. J. Am. Chem. Soc. 2005, 127, 7502–7510.
(14) Rau, S.; Sch a¨ fer, B.; Gleich, D.; Anders, E.; Rudolph, M.; Friedrich,
M.; G o¨ rls, H.; Henry, W.; Vos, J. G. Angew. Chem., Int. Ed. 2006,
45, 6215–6218.
(15) Fisher, J. R.; Cole-Hamilton, D. J. J. Chem. Soc., Dalton Trans. 1984,
809–813.
*
To whom correspondence should be addressed. E-mail: bern@
princeton.edu.
(
1) Lewis, N. S.; Nocera, D. G. Proc. Natl. Acad. Sci. U.S.A. 2006, 103,
5729–15735.
2) Gr a¨ tzel, M. Nature 2001, 414, 338–344.
3) Khaselev, O.; Bansal, A.; Turner, J. A Int. J. Hydrogen Energy 2001,
1
(
(
2
6, 127–132.
(
(
(
4) Moradpour, A.; Amouyal, E.; Keller, P.; Kagan, H NouV. J. Chim.
978, 2, 547.
5) Kirch, M.; Lehn, J.; Sauvage, J. HelV. Chim. Acta 1979, 62, 1345–
384.
6) Kalyanasundaram, K.; Kiwi, J.; Gratzel, M. HelV. Chim. Acta 1978,
1, 2720–2730.
1
1
6
10378 Inorganic Chemistry, Vol. 47, No. 22, 2008
10.1021/ic800988b CCC: $40.75  2008 American Chemical Society
Published on Web 10/22/2008

Photoinduced Hydrogen Production
+
development of efficient homogeneous photocatalytic sys-
tems for water reduction is of considerable importance to
advance the science of artificial photosynthesis.
bipyridine)iridium(III) [Ir(ppy)
2
(bpy)] , and derivatives thereof
that exhibit a wide range of photophysical and electrochemi-
cal properties. The efficiency of these complexes as PSs
2
4
Various examples of homogeneous systems have been
in a homogeneous photocatalytic water reduction system with
1
1-13
14,15
16,17
2+
developed utilizing Co(II),
Pd(II),
and Fe
S
2 2
[Co(bpy)
3
]
was studied, yet no direct correlation existed
complexes as the water reduction catalyst (WRC), which
stores reducing equivalents and carries out the reduction of
between photophysics or electrochemistry and catalytic
13,18
activity.
This emphasizes the difficulties associated with
protons to H
2
, without the need for an additional ER.
improving such complex catalytic systems based on physical
properties because differences in catalyst activity could be
from any combination of changes in light absorption, electron
transfer rates, or catalyst stability.
Recently, it was demonstrated that heteroleptic Ir(III)
complexes were superior to [Ru(bpy)
utilizing tris-(2,2′-bipyridine)cobalt(II), [Co(bpy)
WRC, and triethanolamine (TEOA) as the SR.
Ir-Co system, the Co(II) WRC was the weak component
with catalyst instability leading to rapid system deterioration
and limiting the overall efficiency of the system.
3
Tris-(2,2′-bipyridine)Rh(III), [Rh(bpy) ] , is an ideal
candidate for the WRC component in homogeneous systems
because it accumulates two electrons at a suitable potential
2+
3
]
as the PS in systems
2+
3
] , as the
1
3,18
In the
Herein, we describe a new homogeneous system for
visible-light-induced hydrogen production from water that
utilizes cyclometalated Ir(III) as the PS and tris-2,2′-bipyridyl
Rh(III) complexes as the WRC catalyst. Synthetic modifica-
tion allows the development of a group of catalysts with
diverse ground-state and excited-state redox properties.
Parallel screening of the structure-activity relationships for
the various PS-WRC catalyst combinations, along with
optimization of the reaction conditions, leads to the most
productive homogeneous photocatalytic system for water
reduction to date. Experimental results coupled with photo-
physical and electrochemical characterization of the opti-
mized system provides insight into the catalytic mechanisms
responsible for the photocatalytic reduction of water in this
new system.
3
+
1
9
for water reduction and is known to form hydrides.
Additionally, the one-electron reduction product,
2+
[
Rh(bpy)
3
] , is kinetically unstable and rapidly self-dispro-
portionates to form the doubly reduced species, eliminating
the need for a concerted two-electron reduction by the PS.
20
Tris-(2,2′-bipyridine)Rh(III) complexes have been imple-
mented in a two-component homogeneous system for the
photoreduction of water with UV irradiation in the presence
5
,21
of TEOA.
of [Ru(bpy)
mogeneous systems was unsuccessful in thorough investiga-
Water reduction using visible-light irradiation
Experimental Procedure
]2 with [Rh(bpy)
+
3+
in three-component ho-
3
3
]
General. H NMR and 13C NMR were recorded on a Varian
1
2
0,22
Inova or Bruker BioSpin Avance II 500 MHz spectrometer at room
tions by Sutin and co-workers;
workers reported one instance of H
excess of the Rh-WRC, which only achieved 12 turnovers
however, Lehn and co-
production using a large
19
temperature. F-NMR spectra were recorded on a Varian Mercury-
VX 300 MHz spectrometer at room temperature. Mass spectral data
were collected using a Hewlett-Packard 5898B mass spectrometer
or a Kratos MS50TC RF-High Resolution mass spectrometer.
Elemental analyses were conducted by the Microanalytical Labora-
tory at the University of Illinois, Urbana-Champaign.
2
5
(
2
TON, 0.5 H per PS or WRC). During the completion of
our own investigations with Rh-WRCs, Brewer and co-
workers reported the use of Rh-diimine complexes as the
WRC component in a supramolecular system achieving 30
Materials. 2-Phenyl-pyridine (ppy), 2,2′-bipyridine (bpy), 5,5′-
dimethyl-2,2′-bipyridine (dmbpy), 4,4′-di-tert-butyl-2,2′-bipyridine
(dtbbpy), and 1,2-bis-(diphenylphosphino)ethane (dppe) were pur-
2
3
TON, confirming the discoveries that are described in the
present account.
chased from Aldrich. RhCl
3
·2H
2
O and IrCl
3
2
·4H O were purchased
The natural photosynthetic machinery harvests visible light
to drive water-splitting using a series of finely tuned
chromophoric and redox active sites. This complex system
is difficult to replicate through artificial systems due to
technology, time, and resource constraints. In an attempt to
rapidly develop diverse photocatalysts, our group recently
used combinatorial techniques to synthesize and study a
series of ionic luminophores, bis-(2-phenylpyridine)-(2,2′-
from Pressure Chemical Company. TEOA, triethylamine (TEA),
and N,N′-dimethylaniline (DMA) were purchased from Alfa Aesar.
All synthesis and purification solvents were purchased from EM
Science. Acetonitrile (ACN), tetrahydrofuran (THF), and N,N′-
dimethylformamide (DMF) for photoreactions were purchased from
Acros. Commercial materials were used as received.
5
-Methyl-2-(4-methoxyphenyl)pyridine (MeO-mppy), 5-methyl-
-(4-fluorophenyl)pyridine (f-mppy), and 5-(trifluoromethyl)-2-(2,4-
difluorophenyl)pyridine (df-CF ppy) cyclometalating ligands were
synthesized by a Kr o¨ hnke pyridine synthesis according to
2
3
(
(
(
16) Na, Y.; Wang, M.; Pan, J.; Zhang, P.; Akermark, B.; Sun, L Inorg.
Chem. 2008, 47, 2805–2810.
17) Li, X.; Wang, M.; Zhang, S.; Pan, J.; Na, Y.; Liu, J.; Akerman, B.;
Sun, L. J. Phys. Chem. B 2008, 112, 8198–8202.
1
8,24
literature procedure.
(
The ligand 5,5′-dimethoxy-2,2′-bipyridine
dMeObpy) was prepared by a nickel-catalyzed homocoupling as
18) Lowry, M. S.; Goldsmith, J. I.; Slinker, J. D.; Rohl, R.; Robert, A.;
Pascal, J.; Malliaras, G. G.; Bernhard, S Chem. Mater. 2005, 17, 5712–
described in the literature, and the novel compound 5,5′-difluoro-
2,2′-bipyridine (dfbpy) was synthesized in a similar manner as
5
719.
19) Sutin, N.; Creutz, C.; Fujita, E. Comments Inorg. Chem. 1997, 19,
7–92.
2
5
(
(
described in the Supporting Information.
6
∧
Tetrakis-(C N)-µ-(dichloro)diiridium(III) complexes were pre-
pared with the appropriate cyclometalating ligand (C N ) ppy,
20) Chan, S.; Chou, M.; Creutz, C.; Matsubara, T.; Sutin, N. J. Am. Chem.
Soc. 1981, 103, 369–379.
∧
(
(
21) Kalyanasundaram, K NouV. J. Chim 1979, 3, 511–515.
22) Brown, G. M.; Chan, S. F.; Creutz, C.; Schwarz, H. A.; Sutin, N.
J. Am. Chem. Soc. 1979, 101, 7638–7640.
23) Elvington, M.; Brown, J.; Arachchige, S.; Brewer, K. J. Am. Chem.
Soc. 2007, 129, 10644–10645.
(24) Lowry, M. S.; Hudson, W. R.; Pascal, R. A.; Bernhard, S. J. Am.
Chem. Soc. 2004, 126, 14129–14135.
(25) Fukuda, Y.; Seto, S.; Furuta, H.; Ebisu, H.; Oomori, Y.; Terashima,
S. J. Med. Chem. 2001, 44, 1396–1406.
(
Inorganic Chemistry, Vol. 47, No. 22, 2008 10379

Cline et al.
3
f-mppy, MeO-mppy, and df-CF ppy) followed by cleavage with
∧
the appropriate neutral ligand (N N ) bpy, dmbpy, dppe, dtbbpy)
∧
∧
+
as described previously to form the series of [Ir(C N) (N N)]
2
18,24,26
products.
The complex [Co(bpy)
]Cl
3 2
was prepared according
]3+ where (N N ) bpy, dfbpy,
13
∧
∧
to literature procedure. [Rh(N N)
3
dMeObpy, dtbbpy) was synthesized by a modification of literature
Figure 1. General structure of the Ir(III) PS and Rh(III) WRC used for
27
the photochemical H -production studies (Figure 2 for C∧N and N∧N
2
procedure. The synthesis, purification, and characterization of the
6 3
title compounds [Ir(f-mppy) (dtbbpy)](PF ) and [Rh(dtbbpy) ](PF )
ligands).
2
6
3
are described below. All other synthetic notes and structural
information are reported in the Supporting Information.
Synthesis of [(4,4′-Ditert-butyl-2,2′-bipyridine)-bis-(2-(4-fluo-
rophenyl)-5-methyl-pyridine)-iridium(III)] Hexafluorophosphate.
To [(f-mppy)
161.0 mg, 600 µmol) was added ethylene glycol (5 mL), and the
reaction was heated at 120 °C under N for 24 h. The cooled
reaction mixture was diluted with H O (50 mL) and washed with
Et PF (1.0 g
O (3 × 25 mL). After counterion metathesis with NH
in H O), the solids were isolated by vac filtration, washed with
water, and dried under high vac overnight. The crude product was
purifiedbyrecrystallizationfromACNtoproduce[Ir(fmppy) (dtbbpy)]
2 2
Ir-µ-Cl] (383.8 mg, 300 µmol) and dtbbpy
(
2
2
2
4
6
2
2
(
PF
6
) as a yellow cubic crystal in 81% yield.
Figure 2. A) C N ligands used for [Ir(C N)2(bpy)]+ and
∧
∧
1
∧
+
H NMR (500 MHz, acetone-d
d, 2 H, J ) 8.5), 8.04 (d, 2 H, J ) 6.0), 7.94 (dd, 2 H, J ) 8.5,
.0), 7.82 (dd, 2 H, J ) 8.5, 1.5), 7.74 (dd, 2 H, J ) 6.0, 2.0), 7.48
d, 2 H, J ) 1.5), 6.80 (td, 2 H, J ) 9.5, 2.5), 5.93 (dd, 2 H, J )
.5, 2.5), 2.11 (s, 6 H), 1.43 (s, 18 H).
Synthesis of [Tris-(4,4′-ditert-butyl-2,2′-bipyridine)-rhodium(I-
6
): δ 8.88 (d, 2 H, J ) 2.0), 8.12
[Ir(C N) (dtbbpy)]
complexes; B) neutral ligands used for
2
∧
+
∧
+
∧
(
6
(
9
[Ir(ppy)2(N N)] and [Ir(ppy)2(P P)] complexes; C) N N ligands used
for [Rh(N N)3] complexes.
∧
3+
The photocatalytic reactions were evaluated in parallel fashion
using a home-built 16-sample photoreactor with real-time analysis
1
0
of gas evolution as described previously. Samples were deoxy-
genated with six consecutive cycles of a 30 s gas evacuation
followed by backfilling with an atmosphere of argon and then
subjected to continuous photolysis with bottom illumination
II)] Tris-hexafluorophosphate. To RhCl
3
2
·2H O (250 mg, 1.0 mmol)
and dtbbpy (808 mg, 3.0 mmol) was added EtOH (10 mL) and
O (10 mL), and the reaction was heated at 80 °C for 15 min.
We added 1.0 mL of a 1 M solution of hydrazine in EtOH/H
and then continued heating at 80 °C for an additional 1 h. Following
counterion metathesis with NH PF (1.0 g in H O), the reaction
H
2
2
O
(
460 nm, 500 ( 50 mW) until gas evolution ceased. Analysis of
the reaction headspace by gas chromatography or a residual gas
analyzer quantified hydrogen production. Detailed descriptions of
the experiments and equipment are included in the Supporting
Information.
4
6
2
mixture was concentrated to 10 mL volume by rotary evaporation.
The solids were isolated vac filtration, washed with water, and dried
under high vac overnight. The crude product was purified by vapor
diffusion
[
Electrochemistry. Electrochemical experiments were conducted
using a CH-Instruments Electrochemical Analyzer Model 600C
recrystalization
](PF as a white crystalline solid in 74% yield.
H NMR (500 MHz, acetone-d ): δ 9.07 (d, 6 H, J ) 2.0), 7.90
d, 6 H, J ) 6.0), 7.82 (dd, 6 H, J ) 6.0, 2.0), 1.41 (s, 54 H).
): δ 169.0, 156.0, 151.1, 128.5, 125.5,
(acetone/pentane)
to
give
Rh(dtbbpy)
1
3
6 3
)
2
potentiostat with a 1 mm platinum disk working electrode, a coiled
6
platinum wire supporting electrode, and a silver wire pseudoref-
erence electrode. Ferrocene (Sigma) was employed as an internal
reference, and measured potentials were later referenced to the
standard calomel electrode (SCE). Solutions for voltammetric
analysis were prepared in a 0.1 M solution tetra-n-butylammonium
hexafluorophospate (TBAH) in acetonitrile (ACN). The experiments
were conducted at approximately 1 mM concentration of the
complex of interest at various scan rates ranging from 10 mV/s to
1
3
(
C
NMR (125 MHz, acetone-d
7.0, 30.2. MS (m/e, ESI): 302 (67.5, M ), 453 (8.7, M ), 525
6
3
+
2+
3
(
4
3
+
-
24.1, M - PF
6
). Elem. Anal. Calcd for C54
8.29; H, 5.40; N, 6.26. Found: C, 48.11; H, 5.20; N, 6.29.
General Procedures for Photogeneration of Hydrogen. To
72 6 6 3
H N Rh(PF ) : C,
evaluate the activity of different catalysts in relationship to one
another, all reactions were composed in a uniform fashion through
the use of stock solutions prepared using volumetric techniques to
reduce the error associated with distributing such small quantities
of catalysts. Samples were prepared in 40 mL EPA vials with the
appropriate PS, WRC, SR, and solvent as defined for each set of
experiments. Reactions were prepared by distributing a stock ACN
solution of catalyst to ensure uniformity between reactions. These
samples were immediately concentrated to dryness by rotary
evaporation before adding 10 mL of the appropriate reaction media.
Alternatively, stock solutions of [Ir(f-mppy)
[
tributed. A typical sample for the PS-WRC combination study
contained 10 µmol of the PS and WRC catalysts in 10 mL of a
0
1
0 V/s.
Results and Discussion
Synthesis. [Rh(bpy) Cl
and the synthesis of Rh(bpy)
conditions for long time periods or the addition of a reducing
+
2
2
] is known to be kinetically inert,
3+
3
requires either harsh forcing
2
7,28
∧
3+
agent.
Figures 1 and 2) was accomplished by the reaction of the
appropriate bpy ligand with RhCl •2H O in refluxing EtOH/
O for 15 min, followed by the addition of a reducing
Facile synthesis of the [Rh(N N)
3
]
complexes
(
2 6
(dtbbpy)](PF ) and
3
2
Rh(dtbbpy) ](PF in the appropriate reaction media were dis-
3
6 3
)
H
2
equivalent of hydrazine and subsequent refluxing for 1 h
under aerobic conditions. This method has previously been
.6 M TEOA solution in 90% ACN-water.
3
+
3
described for the synthesis of [Rh(bpy) ] ; however, the
(26) Sprouse, S.; King, K. A.; Spellane, P. J.; Watts, R. J. J. Am. Chem.
Soc. 1984, 106, 6647–6653.
(27) Hillis, J. E.; DeArmond, M. K. J. Lumin. 1971, 4, 273–290.
(28) McKenzie, L. D.; Plowman, R. A. J. Inorg. Nucl. Chem. 1970, 32,
199–212.
10380 Inorganic Chemistry, Vol. 47, No. 22, 2008

Photoinduced Hydrogen Production
27,29
specific attributes of the reaction have not been detailed.
Titration of the hydrazine equivalency (Supporting Informa-
tion) indicates that under aerobic conditions the reaction is
not catalytic, requiring a full equivalent of hydrazine to
+
reduce [Rh(bpy)
2
2
Cl ]
and promote ligand labilization.
+
Subsequent reoxidation of [Rh(bpy)
2
] by dissolved oxygen
would promote association of the third bpy ligand to form
3
+
[
Rh(bpy)
3
] . This procedure allowed the efficient synthesis
∧
3+
∧
of three previously unreported [Rh(N N)
3
]
species (N N
is dfbpy, dMeObpy, or dtbbpy) and has proven to be
generally applicable and robust.
The novel ligand 5,5′-difluoro-2,2′-bipyridine was prepared
in 80% yield by a nickel-catalyzed homocoupling and
Figure 3. Evaluation of [Ir(ppy)2(bpy)](PF6) or [Ru(bpy)3](PF6)2 PS with
[
Co(bpy)3]Cl2 or [Rh(dtbbpy)3](PF6)3 WRC in photosynthetic-H2 reactions
(
10 µmol of PS and WRC, 10 mL of 0.6 M TEOA in 50% ACN-H2O, 460
∧
∧
+
recrystalization from EtOH. [Ir(C N)
2
(N N)]
and
nm, 500 mW, 18 h).
∧
∧
+
[
Ir(C N)
2
(P P)] complexes (Figures 1 and 2) were synthe-
∧
]2+ or [Ir(ppy) (bpy)]+ PS in a photosynthetic
3 2
sized by preparation of the tetrakis-(C N)-µ-(dichloro)-
[Ru(bpy)
reaction with 0.6 M TEOA in 50% ACN-H
kinetic trace in Figure 3 shows that [Ir(ppy)
diiridium(III) complexes with the appropriate cyclometalating
2
O. The reaction
∧
∧
+
ligand (C N) followed by dimer cleavage with the neutral
2
(bpy)] outper-
∧
24
2+
ligands (N N or P P). The Ir(III) and Rh(III) complexes
forms [Ru(bpy)
3
]
when evaluated with either WRC. When
3
+
(
Figures 1 and 2) were purified by column chromatography
evaluated with the Ir-PS, [Rh(bpy)
active with a maximum catalytic rate of 23 turnovers per
hour (TOF, 0.5 H per catalyst per hour) and overall
efficiency of 29 TON compared to 9.5 TOF (h ) and 5 TON
3
]
was much more
-
or recrystalization, as appropriate, and isolated as the PF
salt.
6
2
-1
Catalyst Evaluation Criteria. The photoreactions are
evaluated using a number of criteria that provide insight into
2+
for [Co(bpy)
3
]
2
(Supporting Information). Both H -evolution
kinetics plots can be fitted with a unimolecular exponential
decay function, as described in ref 10, which indicates that
the activity of the Co(II) system is decaying at nearly twice
the rate of the more robust Rh(III) system (Supporting
Information).
the activity of the system. End-point results such as H
2
produced and TON are a composite of both catalytic rate
and catalyst stability, so they provide little utility when
attempting to assess a system’s efficiency and robustness.
By monitoring gas evolution at 5 s intervals over the reaction
time-course and normalizing the result with a headspace
composition analysis at the end point, we get a very detailed
picture of the reaction in a kinetic trace. The first derivative
of the kinetic trace provides information about the catalytic
rate and the catalyst decay for each reaction. Because the
systems involve an induction period as the WRC is converted
to the active catalytic state, the maximum catalytic rate is
used as the relevent rate in kinetic analysis with the
assumption that the catalysts are still at the initial concentra-
tions.
Structure-Activity Relationships. The promising pre-
liminary results for the Ir-Rh system warranted further
investigation but the complicated nature of multicomponent
photocatalytic systems makes the improvement of such
systems through the study of their physical properties a nearly
impossible task. To make rapid improvements, synthetic
modification was used to develop a group of catalysts with
molecular diversity. The heteroleptic Ir(III) PSs are perfectly
suited for synthetic tuning because ligand substitutions on
the cyclometalating and neutral ligands allow the energy of
the highest occupied and lowest unoccupied molecular
orbitals to be modified independently. Seven diverse PSs with
a range of photophysical and electrochemical properties were
The maximum catalytic rate can be converted to quantum
yield (QY) (0.5 H per photon absorbed) using the calculated
2
photon flow and the molar absorptivity of the PS (Supporting
Information). The QY calculations disregard any light
1
3,18,24,30
prepared (Figures 1 and 2).
The Rh WRCs had not
been studied previously, so four diverse bpys were chosen
to determine the effect of electron withdrawing and donating
substituents on catalyst activity (Figures 1 and 2). The seven
Ir PSs and four Rh WRCs were evaluated in a 0.6 M TEOA
∧
3+
absorption from the WRC because [Rh(N N)
3
]
does not
∧
absorb light in the visible region (N N ) bpy or dtbbpy,
-
1
-1
ꢀ460 nm
< 1 M cm in ACN). The actual structure of the
WRC catalyst in solution is unknown and cannot be
accounted for in QY calculations, therefore making them a
minimum estimate of the actual value. However, the dilute
nature of the reactions and high photon flow of the LED
ensure sufficient irradiation to overcome spectral interference
from weakly absorbing molecules.
solution with 90% ACN-H
was the optimal ACN-H O mixture for the Ir-Rh system
Supporting Information). The end-point results, presented
2
O, for which studies indicated
2
(
∧
in Figure 4, indicate that dtbbpy is the optimal N N ligand
for both PS and WRC, whereas the f-mppy and MeO-mppy
cyclometallating ligands are equally effective.
Catalyst System Comparison. Initially, the utility of
To confirm the previous results and determine which
2
+
3+
[
Co(bpy)
3
]
and [Rh(bpy)
3
]
as the WRC was compared
∧
cyclometalating ligand outperforms the rest, all four C N
by evaluating them with an equivalent amount of
∧
ligands were evaluated with the optimal N N ligand, dtbbpy.
(
29) Gillard, R. D.; Heaton, B. T. Coor. Chem. ReV. 1972, 8, 149–157.
(30) Lowry, M. S.; Bernhard, S. Chem.sEur. J. 2006, 12, 7970–7977.
Inorganic Chemistry, Vol. 47, No. 22, 2008 10381

Cline et al.
Figure 6. Evaluation of [Ir(f-mppy)2(dtbbpy)](PF6) and [Rh(dtbbpy)3](PF6)3
in photosynthetic-H2 reactions (1 µmol of PS and WRC in 10 mL of 0.6 M
sol of SR in 90% solvent-H2O, 460 nm, 500 mW, 20-24 h) with the SR
and solvent indicated in the graph.
conditions to achieve the highest possible catalytic activity
while exploring the properties and limitations of the Ir-Rh
system.
A. Sacrificial Reductant. The SR can play a large role
in the activity of the catalytic system because it affects the
properties of the reaction medium (pH, dielectric constant,
solvent viscosity), which can influence the observed rates
of the various electron transfers and metal-hydride formation.
In the Ir-Rh system, the SR has an even greater influence
than in traditional ruthenium-based systems because the
Ir-Rh system functions through a reductive quenching
mechanism (below) that depends on efficient quenching of
the excited-state PS by the SR.
∧
∧
∧
Figure 4. Evaluation of [Ir(C N)2(N N)](PF6) and [Rh(N N)3](PF6)3 in
photosynthetic-H2 reactions (1 µmol of PS and WRC in 10 mL of 0.6 M
sol of SR in 90% Solvent-H2O, 460 nm, 500 mW, 20-24 h) with the SR
and solvent indicated in the graph.
DMA and TEA are two SRs that have been previously
2
3,31,32
utilized in photocatalytic water-reduction systems.
Using laser flash photolysis to measure dynamic quenching
(
Supporting Information), the quenching constants for DMA
9
-1 -1
8
-1 -1
(k
q
) 5.6 × 10 M s ) and TEA (k
q
+
) 1.2 × 10 M s )
were determined for [Ir(ppy)
literature value for TEOA (k
2
(bpy)] and compared to the
6
-1 -1
q
) 5.4 × 10 M s ). We
evaluated the use of these two amines as the SR with the
optimal PS-WRC combination, and the kinetic traces in
Figure 6 show that the aliphatic amine TEA gives a dramatic
increase in catalytic rate and a 6-fold increase in total
hydrogen production compared to TEOA. Counter-intu-
itively, the system using the DMA, the best quencher of the
three SRs studied, exhibited little catalytic activity compared
to the aliphatic amines TEOA or TEA. This result may
indicate that for DMA quenching involves energy-transfer
rather than electron-transfer quenching, that charge recom-
bination is occurring, or that DMA creates an unfavorable
environment for the WRC-hydride formation. All subse-
quent studies were conducted using TEA as the SR.
∧
Figure 5. Evaluation of [Ir(C N)2(dtbbpy)](PF6) and [Rh(dtbbpy)3](PF6)3
in photosynthetic-H2 reactions (10 µmol of PS and WRC in 10 mL of 0.6
M sol of TEOA in 90% ACN-H2O, 460 nm, 500 mW, 18 h).
The kinetic traces in Figure 5 indicate that the combination
+
]3+ is the most
of [Ir(f-mppy)
2
(dtbbpy)] and [Rh(dtbbpy)
3
effective catalyst system under the conditions employed with
-
1
more than 150 TON and a maximum rate of 39 TOF (h )
for a QY of 11.5%. The use of synthetic modification and
catalyst screening allowed for a rapid improvement in
catalytic activity and efficiency, further validating our
screening approach. All of the subsequent studies were
conducted using the optimal PS-WRC combination of
(dtbbpy)] and [Rh(dtbbpy)
Photoreaction Condition Optimization. Reaction condi-
tions such as sacrificial reductant, solvent environment, and
catalyst concentration have a dramatic effect on the perfor-
mance of artificial photosynthetic systems. Following the
completion of the SAR study, we optimized the reaction
B. Solvent Environment. In homogeneous catalysis, the
solvent environment often plays a critical role in the activity
and stability of the catalysts. The electrostatic properties of
the medium can affect the photophysical properties of the
PS as well as the Coulombic interactions and the rates of
+
]3+ described above.
[
Ir(f-mppy)
2
3
33,34
ET between the PS and WRC components.
Additionally,
the ligating ability of the solvent may influence the reaction
3
2
by stabilizing or destabilizing key redox intermediates.
(
31) DeLaive, P. J.; Sullivan, B. P.; Meyer, T. J.; Whitten, D. G. J. Am.
Chem. Soc. 1979, 101, 4007–4008.
10382 Inorganic Chemistry, Vol. 47, No. 22, 2008

Photoinduced Hydrogen Production
Table 1. Results for [Ir(f-mppy)2(dtbbpy)](PF6) and
is not the result of the direct dehydrogenation of TEA, isotope
labeling experiments were conducted. The optimal PS-WRC
[
Rh(dtbbpy)3](PF6)3 in Photosynthetic-H2 Reactions the Optimal Solvent
Ratio (1 µmol of PS and WRC in 10 mL of 0.6 M TEA in THF-H2O,
4
60 nm, 500 mW, 18 h)
combination using 0.6 M TEA in 8:2 THF-D
a gas composition of 2:12:86 H :HD:D for 92% deuterium
incorporation. A similar experiment using an 8:1:1 mixture
of THF-H O-D O showed little difference in the rate of
gas evolution but the gas composition was 70:27:3 H :HD:
for only 17% deuterium incorporation. Thus, although
2
O resulted in
THF-H2O (%)
H2 (µmol)
TON
max rate (µmol/h)
2
2
5
7
7
8
8
9
9
9
0
0
5
0
5
0
5
9.9
423
732
918
1063
934
840
625
360
846
1463
1835
2126
1868
1680
1251
721
127
173
180
182
173
203
153
86
2
2
2
D
2
there is an obvious kinetic isotope effect involving the
formation of the rhodium hydride species, this is not the rate-
limiting step in the photosynthetic reactions and the overall
rate of H
explains the 8% hydrogen that is incorporated in the gas
mixture for the 8:2 THF-D O reaction because the small
amount of protons generated in the SR oxidation and
decomposition (below) will be selectively reduced over the
available deuterium for a disproportionate representation in
the final product. The isotope labeling clearly demonstrates
that our system selectively reduces water to produce hydro-
gen.
2
production is unaffected. This isotope effect
Previous studies of the Ir-PS using mass spectrometry
showed that ACN displacement of the bipyridine ligand
1
0
accompanied the observed decay in catalyst activity. In
an effort to improve the catalyst activity and stability in the
Ir-Rh system, additional solvents with less coordinating
strength than ACN were tested with the optimal PS-WRC
combination in 0.6 M TEA with a 9:1 solvent-water
mixture. Improved results were achieved with both DMF and
THF, whose reaction kinetic traces are presented in Figure
2
D. Catalyst Concentration. As substantial increases in
catalyst activity were achieved, catalyst loading was reduced
so that the reactions would operate within the capacity of
the photoreactor setup. From preliminary results, a marked
increase in efficiency and catalytic activity was noticed as
the systems approached lower catalyst concentrations. This
result is counter-intuitive because outer-sphere ET processes
should be more efficient at higher concentration. Addition-
ally, the PS concentrations in the photoreactions are suf-
ficiently low to ensure that the reactions are not light-limited.
This indicates that competitive deactivation pathways are
prevalent at higher concentrations, such as the self-annihila-
6
. Both solvents are much weaker ligands than ACN, whereas
calculations of the static dielectric constant using the
Kirkwood theory for multicomponent mixtures reveals that
the ACN and DMF mixtures have very similar values,
3
5
whereas the THF mixture will be much lower. The results
show that the ACN and DMF reactions have identical
catalytic rates during the intial five hours, yet in DMF the
catalysts show greater stability allowing it to outperform the
ACN reaction over the extended reaction period. The THF
reaction shows the same improvement of catalyst stability
with the expected increase in catalytic rate due to its lower
dielectric constant.
The early studies for catalyst system comparison were
conducted in a 0.6 M TEOA solution using a 1:1 mixture of
ACN and water. A study evaluating solvent ratio (Supporting
Information) showed that catalyst performance improves at
3
+
tion between rhodium hydride and [Rh(bpy)
previously in studies of the radiolytic reduction of water.
3
]
noted
3
7
To examine the concentration dependence more thor-
oughly, an experiment was conducted varying PS and WRC
concentrations independently in the range of 0.05 to 0.15 M.
The results of this 25-reaction array (Figure 7) demonstrate
that catalyst performance improves when one component is
evaluated with an excess of the other component, and more
than 5000 TON are achieved for either the PS or the WRC.
The use of an equivalent amount of PS and WRC balances
the performance of both components. Looking at the PS and
WRC TON for the series of 1:1 PS-WRC reactions (parts A
and B of Figure 7), the catalysts have a maximum efficiency
around 0.1 mM concentration. The combination of QY,
catalytic rates and TON are optimized at 0.075 mM PS and
WRC concentration. This results in 3400 TON and a
2
higher ACN concentrations, with maximum H -production
in 9:1 ACN-water. This phenomenon could be attributed
to a balance between the benefit of the lower dielectric
constant at high ACN concentration with the need for water
36
as a proton source for Rh-hydride formation. To determine
the optimal solvent ratio for the TEA-THF-water system,
eight different compositions ranging from 1:1 to 1000:1 THF-
water were evaluated with the optimal PS-WRC combina-
tion. The results in Table 1 show that, although the highest
catalytic rates were achieved in 90% THF-water, the 80%
THF system was more productive over the course of the
reaction and the optimized ratio of THF-water was used
for all subsequent studies.
-
1
maximum catalytic rate of 520 TOF (h ), which equates to
a QY of 32% (part C of Figure 7). This marks more than
two orders of magnitude improvement over the original
Ir-Rh system (Figure 3).
C. Isotope Labeling. The results in Table 1 indicate that
with less than 0.1% water available the photocatalytic system
continues to function. To ensure that the hydrogen production
Reaction Mechanism. The photocatalytic water reduction
process reported above involves three critical components:
the PS, the SR, and WRC. In the catalytic light cycle, the
PS is responsible for the absorption of visible light to provide
(
32) Ziessel, R.; Hawecker, J.; Lehn, J. HelV. Chim. Acta 1986, 69, 1065–
084.
33) Brunschwig, B. S.; Ehrenson, S.; Sutin, N. J. Phys. Chem. 1987, 91,
714–4723.
1
(
4
(34) Borchardt, D.; Wherland, S. Inorg. Chem. 1984, 23, 2537–2542.
(35) Wang, P.; Anderko, A. Fluid Phase Equilib. 2001, 186, 103–122.
(36) Bakac, A. J. Chem. Soc., Dalton Trans. 2006, 1589–1596.
(37) Mulazzani, Q. G.; Venturi, M.; Hoffman, M. Z. J. Phys. Chem. 1982,
86, 242–247.
Inorganic Chemistry, Vol. 47, No. 22, 2008 10383

Cline et al.
Figure 8. Evaluation of [Ir(f-mppy)2(dtbbpy)](PF6) and [Rh(dtbbpy)3](PF6)3
in photosynthetic-H2 reactions (1 µmol of PS and WRC in 10 mL of 80%
THF-H2O, 460 nm, 500 mW, 20-24 h) with varying TEA concentrations.
3
8-40
ously.
The oxidation peak potential for TEA at pH 12
4
1
in water is reported to be 0.69 V (vs SCE). Following the
initial one-electron oxidation, the TEA radical cation is
+
rapidly deprotonated and undergoes a radical shift to the
R-carbon. This neutral carbon radical species is expected to
be highly reducing, and a second oxidation forms the
iminium cation, which is hydrolyzed to form DEA and
acetaldehyde. Thus, each TEA is capable of donating two
electrons and two protons. Rapid conversion of the oxidized
TEA species, as outlined in eqs 2 and 4, avoid any possible
back reactions between the oxidized TEA species and the
reduced PS or WRC species.
•
+
-
TEA f Et NCH CH + e
(1)
(2)
(3)
(4)
2
2
3
•
+
•
+
-
Et NCH CH f Et NCHCH + H
2
2
3
2
3
•
+
Et NCHCH f Et N d CHCH + e
2
3
2
3
+
+
Et N d CHCH f DEA + CH CHO + H
2
3
3
Figure 7. A) Ir TON, B) Rh TON, and C) kinetic and rate trace (at 0.075
mM PS and WRC) from a study evaluating the effect of catalyst
concentration on performance with [Ir(f-mppy)2(dtbbpy)](PF6) and
During the course of the photoreaction, the pH and solvent
composition may gradually change as a result of the
degradation of TEA or the reduction of water, potentially
leading to a substantial change in reaction conditions when
the system achieves high TON. As proven by the isotope
experiment, the protons from water are reduced in the dark
cycle to produce hydrogen, yet these protons are replenished
during the dielectronic reduction and degradation of TEA.
Thus, the net change in reaction condition is the partial
[
Rh(dtbbpy)3](PF6)3 in photosynthetic-H2 reactions (0.5-1.5 µmol of PS
and WRC in 10 mL of 0.6 M TEA in 80% THF-H2O, 460 nm, 500 mW,
2 h).
2
the energy necessary for the endothermic water reduction
process to occur. In the catalytic dark cycle, the WRC
collects the high-energy electrons from the light cycle and
protons from water to produce hydrogen. The SR provides
the electrons necessary for water reduction by replenishing
the lowest singly occupied molecular orbital of the PS* or
a
conversion of TEA to DEA, which have similar pK values,
and the consumption of 1-2% of the water. To ensure that
sufficient TEA is still available at all time points in the
reaction, the dependency on TEA concentration was studied.
The kinetic traces in Figure 8 indicate that the catalytic rate
and overall reaction are not affected at concentrations as low
as 0.3 M TEA. In the current investigations, we can be certain
that any decrease in activity at later time points during the
+
PS , depending on the mechanism of quenching. The
interactions of all three components and their corresponding
redox intermediates determine the mechanism and energetics
of the process. Photophysical and electrochemical charac-
terization of the components provides insight into the nature
of these interactions in the optimized system.
A. Sacrificial Reductant. Aliphatic tertiary amines have
traditionally been employed as the SR in photocatalytic
systems for water reduction, and their oxidation and subse-
quent degradation (eqs 1-4) have been studied previ-
(38) Ross, S. D. Tetrahedron Lett. 1973, 14, 1237–1240.
(
39) Cohen, S. G.; Parola, A.; Parsons, G. H. Chem. ReV. 1973, 73, 141–
1
61.
(40) DeLaive, P. J.; Foreman, T. K.; Giannotti, C.; Whitten, D. G. J. Am.
Chem. Soc. 1980, 102, 5627–5631.
(
41) Chow, Y. L.; Danen, W. C.; Nelsen, S. F.; Rosenblatt, D. H. Chem.
ReV. 1978, 78, 243–274.
10384 Inorganic Chemistry, Vol. 47, No. 22, 2008

Photoinduced Hydrogen Production
the classic Ru-PS, is E°′ ) -0.78 V (vs SCE). Reductive
quenching of the excited Ir-PS by the SR yields much higher
energy reducing equivalents than oxidative quenching by the
WRC.
C. Dark Cycle. Published electrochemical studies of
[
Rh(bpy)
3 6 3
](PF ) and related complexes in ACN have
demonstrated complex electrochemical behavior attributed
Figure 9. General reaction scheme for photocatalytic hydrogen production
by TEA-Ir-Rh system.
4
2,43
to an irreversible ECEC mechanism.
mmogram of [Rh(dtbbpy) ](PF (part B of Figure 10) shows
very similar electrochemistry to that of the unsubstituted form
part A of Figure 10). The increased electron donation of
The cyclic volta-
3
6 3
)
photoreactions are due to catalyst degradation instead of
changes in the reaction environment.
(
B. Light Cycle. The central component of the catalytic
system drives photosynthesis through charge transfer of a
high-energy electron to the dark-cycle catalytic components.
Following visible-light excitation (460 nm, ꢀ ) 553 (
the tert-butyl substitutions on the bpy ligands results in a
negative shift of all of the reduction peaks as expected.
Cathodic waves I and II correspond to irreversible one-
electron reductions (Epa ) -0.96 V and Epa ) -1.15 V vs
-
1
-1
5
M cm in 80% THF/H
achieves a long-lived charge-separated state (eq 5) with a
.15 µs decay lifetime similar to that seen for the classical
PS [Ru(bpy) ](PF . The excited-state is then quenched
2 2 6
O), [Ir(f-mppy) (dtbbpy)](PF )
SCE at 100 mV/s) that are accompanied by ligand labilization
+
to form [Rh(dtbbpy)
2
] . Parts C-F of Figure 10 show that
1
reversing the scans immediately following waves I or II
imparts some reversibility at high scan rates as expected with
3
6 2
)
through two different pathways: reductive quenching by the
SR (eq 6) or oxidative quenching by the components of the
dark cycle (eq 7). Laser flash photolysis and Stern-Volmer
analyses (Supporting Information) yielded the quenching
constants for the optimal system. The rate constants for
an EC mechanism. The reversible cathodic wave III repre-
0
sents the formation of [Rh(dtbbpy)
SCE, ∆E ) 66 mV).
2
] (E°′ ) -1.45 V vs
-
∧
3+
∧
2+
PS + Rh(N N)
f PS + Rh(N N)
(8)
(9)
3
3
+*
reductive quenching of [Ir(f-mppy)
2
(dtbbpy)] by TEA and
∧
2+
∧
2+
∧
3+
Rh(N N)
f Rh(N N)
2
+ N N
oxidative quenching by [Rh(dtbbpy)
3
]
are roughly equiva-
3
7
-1 -1
7
-1 -1
lent (kox ) 6.7 × 10 M s and kred ) 6.4 × 10 M s ).
PS- + Rh(N N)
∧
2+
∧
+
f PS + Rh(N N)
(10)
(11)
2
2
Because TEA is present at more than 1000 times the
3
+
-
∧
+
∧
0
concentration of [Rh(dtbbpy)
3
] , the reaction is operating
PS + Rh(N N) f PS + Rh(N N)
2
2
predominately through a reductive quenching mechanism,
0
The active catalytic state of the WRC component is
unknown, although the Ir-Rh system exhibits an induction
period up to 1 h before it reaches maximum activity, as
illustrated in the part C of Figure 7, indicating the conversion
of the WRC to a different species as the active catalyst. The
2
where the highly reactive [Ir(f-mppy) (dtbbpy)] species
reduces the WRC components as outlined in Figure 9.
hV
PS
9
8 PS*
(5)
(6)
(7)
∧
3+
electrochemical data indicates that [Rh(N N)
form [Rh(N N)
2
3
]
will initially
] following two consecutive reductions and
ligand labilization (eqs 8-10). This was confirmed by a
Kred
∧
+
-
+
PS * + TEA
8 PS + TEA
∧
3+
∧
+
comparison of [Rh(N N)
3
]
2 2
and [Rh(N N) Cl ] (Support-
Kox
∧
3+
+
∧
2+
ing Information) that showed no difference in catalytic
activity between the tris- and bis-bipyridyl Rh(III) species.
PS * + Rh(N N)
8 PS + Rh(N N)
3
3
∧
+
Although earlier publications postulated that [Rh(N N)
2
]
The reaction pathway has a major effect on the reaction
energetics because the reducing equivalents available to the
WRC come from the reduced or excited PS for the reductive
or oxidative quenching pathways, respectively. Cyclomet-
allated Ir(III) complexes are distinct from traditional Ru(II)
5
is the active catalyst species, it has been noted that the Rh(I)
species is stable in both acidic and alkaline solution for
months. Mulazanni and co-workers hypothesized that the
44
Rh(III)-hydride could form and then be further reduced to
37
2
form a Rh(II)-hydride as the direct precursor to H , similar
+
*
PSs in that the excited-state species, [Ir(ppy)
2
(bpy)] , can
45
to the mechanism proposed for Rh-porphyrins. The high-
energy reducing equivalents from the reductive quenching
mechanism provides the necessary energy for the
be reductively quenched by tertiary amines. The electro-
chemical and emission data for many of the compounds
1
3,18
evaluated in this study were previously reported.
With
∧
+
∧
0
[
2 2
Rh(N N) ] species to be reduced to form [Rh(N N) ] per
the ground-state oxidation potential and the triplet-singlet
relaxation energy, a minimal estimate of the excited-state
oxidation potential can be made using Hess’ Law. [Ir(f-
eq 11. This species would then form a Rh(II) hydride, which
(42) Kew, G.; DeArmond, K.; Hanck, K. J. Phys. Chem. 1974, 78, 727–
7
34.
mppy)
2
(dtbbpy)](PF
6
) has a reversible one-electron reduction
) 67 mV, vs SCE) and an
(
43) Creutz, C.; Keller, A. D.; Sutin, N.; Zipp, A. P. J. Am. Chem. Soc.
982, 104, 3618–3627.
potential of E°′ ) -1.50 V (∆E
p
1
excited-state oxidation potential of E°′ ) -0.94 V (vs SCE).
(44) Mulazzani, Q. G.; Emmi, S.; Hoffman, M. Z.; Venturi, M. J. Am.
Chem. Soc. 1981, 103, 3362–3370.
The corresponding excited-state oxidation potential for
(
45) Grass, V.; Lexa, D.; Saveant, J. J. Am. Chem. Soc. 1997, 119, 7526–
[
3 6
Ru(bpy) ](PF )
2
, the only reducing equivalent available for
7532.
Inorganic Chemistry, Vol. 47, No. 22, 2008 10385

Cline et al.
are often viewed as strong evidence of slow formation of a
highly active colloidal species followed by deactivation as
48
the nanoparticles agglomerate to form inactive bulk metal.
However, the kinetics could also be attributed to the
electrochemical transformation of the precatalyst to form the
active catalytic species, as outlined in eqs 8-11, followed
10
by the degradation of the PS as observed previously. There
is no single definitive experiment to make the distinction
between homogeneous and heterogeneous catalysis and a
variety of methods must be used to reach a convincing
4
9
conclusion regarding the actual catalytic mechanism.
A. Rhodium Colloids. Colloidal rhodium has been used
in the photochemical production of hydrogen in at least one
instance, although it was less active than the standard
platinum colloid and did not seem to be a promising
8
∧
3+
candidate. If [Rh(N N)
3
]
was a precursor for colloid
formation, the activity of all species would be expected to
be similar, so the SAR for the different ligand substitutions
(
Figure 4) seems to indicate a true homogeneous catalyst. It
∧
cannot be ruled out, however, that the different N N ligands
would not have an effect on colloid formation if the reaction
were occurring through a heterogeneous mechanism.
To evaluate the activity of rhodium colloids in the present
system, several attempts were made to prepare colloid
samples for comparison (Supporting Information). Tertiary
amines are known to stabilize rhodium colloids, so it is
feasible that colloid formation might be favored under the
reaction conditions employed in the present studies. An
5
0
experiment substituting RhCl
precursor shows little catalytic activity when compared to
Rh(dtbbpy) ](PF . Attempts to presynthesize colloids by
3
2
· 2H O as a potential colloid
[
3
6 3
)
the conditions employed in ref 50 resulted in the formation
of large particulate rhodium that showed no activity in the
Ir-Rh system. Additionally, a polymer-protected rhodium
5
1
colloid was synthesized according to literature procedure,
for use as a positive control in the poisoning tests described
below, yet it too showed negligible catalytic activity in our
system.
B. Mercury Poisoning. The ability of mercury to poison
nanoparticulate metal catalysts is a widely used test to
establish reaction mechanism, and the discovery of platinum
colloid formation in a supposed homogeneous system using
Figure 10. Cyclic voltammograms of [Rh(bpy)3](PF6)3 (Inset A) and
Rh(dtbbpy)3](PF6)3 (Insets B-F), 0.5 mM concentration in ACN with 0.1
M TBAH, 1 mm platinum disk working electrode, a coiled platinum wire
supporting electrode, silver wire pseudoreference electrode using ferrocene
as an internal standard referenced at 0.37 V vs SCE.
[
2
47
this method was recently reported. The rhodium amalgam
is known to form, albeit with more difficulty than the
could produce hydrogen in a homolytic bimolecular
pathway or form a rhodium dihydride that produces
hydrogen by reductive elimination in a homolytic unimo-
lecular pathway.
52
platinum amalgam, and mercury poisoning has been
48,53
reported for rhodium colloids.
To test the Ir-Rh system,
the pristine reaction was compared to both a reaction
pretreated with mercury then filtered prior to illumination
and a reaction treated with mercury during illumination. If
catalytic activity is the result of rhodium colloids forming
from degradation of the precatalyst during the photoreaction,
then pretreatment with mercury should not affect catalytic
Homogeneous or Heterogeneous Catalysis. In any
homogeneous system where reducing conditions exist, care
must be taken to ensure that the catalytic activity originates
from an actual molecular species and not a colloidal metal
formed through catalyst decomposition. It has recently been
proven in two separate instances that Pt(II) or Pd(II)
molecular catalysts were actually decomposing to form
(46) Lei, P.; Hedlund, M.; Lomoth, R.; Rensmo, H.; Johansson, O.;
Hammarstrom, L. J. Am. Chem. Soc. 2008, 130, 26–27.
(47) Du, P.; Schneider, J.; Li, F.; Zhao, W.; Patel, U.; Castellano, F. N.;
Eisenberg, R. J. Am. Chem. Soc. 2008, 130, 5056–5058.
(48) Weddle, K.; Aiken, J.; Finke, R. J. Am. Chem. Soc. 1998, 120, 5653–
5666.
4
6,47
colloids, which were the active catalyst species.
The
reaction kinetics for the Ir-Rh system reveal an induction
period and sigmoidal kinetics (part C of Figure 7), which
10386 Inorganic Chemistry, Vol. 47, No. 22, 2008

Photoinduced Hydrogen Production
Figure 11. [Ir(f-mppy)2(dtbbpy)](PF6) with [Rh(dtbbpy)3](PF6)3 in pho-
tosynthetic-H2 reactions (0.5 µmol of PS and WRC in 10 mL of 0.6 M sol
of TEA in 80% THF-H2O, 460 nm, 500 mW, 24 h) that are untreated,
pretreated with Hg (1 g) for 1 h then filtered, or treated with Hg (1 g)
during illumination; shaking 150 rpm or stirring 1200 rpm.
Figure 12. End-point results for quantitative poisoning of [Ir(f-
mppy)2(dtbbpy)](PF6) with [Rh(dtbbpy)3](PF6)3 in photosynthetic-H2 reac-
tions (0.5 µmol of PS and WRC in 10 mL of 0.6 M sol of TEA in 80%
THF-H2O, 460 nm, 500 mW, 24 h) with CS2 (0, 0.5, 1, 1.5, 2, 2.5, or 5
µmol).
activity and the addition of mercury during illumination
should poison the system such that little or no activity is
observed.
5
4
than 0.018 equiv of CS
system, the addition of CS
multiple equivalents of CS to poison catalyst activity (Figure
2), as would be expected for a homogeneous system. The
2
per rhodium. For the Ir-Rh
2
to the photoreactions required
The results of the mercury test, presented in Figure 11,
showed a similar effect for both the pretreated and treated
systems with increased initial rates and shorter system
lifetimes. The majority of the catalytic activity is retained
in the mercury test, indicating that the catalytic activity is
not due to rhodium colloid formation and that any effect is
due to an unknown interaction between mercury and the
solvent, SR, PS, or WRC, irrespective of the photoreaction
itself. Finke and co-workers highlighted the need for a large
excess of mercury (at least 300 equiv) and good stirring to
ensure that a lack of catalyst poisoning is in fact due to the
2
1
2
effect of CS poisoning shows increased poisoning at 2 and
3 equiv with only 10% of activity remaining at 5 equiv of
CS . The activity slowly approaches zero at higher concen-
trations, and it appears that competitive binding between
dtbbpy and CS occurs. These results reinforce the mercury
poison test and provide strong evidence that the Ir-Rh
system does operate through a true homogeneous mechanism.
2
2
Conclusion
4
9
absence of a colloid. The present study utilized more than
0 000 equiv of mercury, and vigorous stirring of the mercury
We have reported an efficient homogeneous catalytic
system for the visible-light-induced production of hydrogen
from water utilizing tris-2,2′-bipyridyl Rh(III) complexes.
The synthetic modification and catalyst screening identified
an optimal catalyst combination of [Ir(f-mppy)
PF ) and [Rh(dtbbpy) ](PF , and the systematic optimiza-
tion of the reaction conditions led to more than 2 orders of
magnitude improvement over the original system. We
currently achieve more than 5000 TON for the PS and WRC
with QYs greater than 34%. Photophysical and electrochemi-
cal characterization reveals that the reductive quenching
pathway provides the necessary driving force to efficiently
reduce water using the Rh-WRC. Mercury and quantitative
1
test photoreactions at 1200 rpm showed not significant
difference to orbital shaking at 150 rpm ensuring that the
test is not mass transfer limited. Without a catalytically active
rhodium colloid sample for comparison as a positive control,
however, we cannot be absolutely certain that the lack of
poisoning indicates a true homogeneous system.
2
(dtbbpy)]
(
6
3
6 3
)
2
C. Quantitative CS Poisoning. Quantitative poisoning
studies can discern whether a system operates by heteroge-
neous or homogeneous catalysis using ligands that bind
strongly to metal centers and have the ability to poison both
4
9
heterogeneous and molecular species. A heterogeneous
species will have only a fraction of the metal atoms on the
surface and will be poisoned completely with ,1 equiv of
the poison ligand, while a molecular catalyst will require
2
CS poison tests both provide strong evidence of a true
molecular catalyst, and we are confident that the catalyst
activity is the result of a homogeneous catalytic species. To
the best of our knowledge, the results are superior to any
other homogeneous photocatalytic system for water reduc-
tion.
Ongoing investigations of the complex mechanistic details
of the Ir-Rh systems will aid the design of more robust
catalysts. Additionally, we are working to incorporate the
Ir-PS and Rh-WRC components into supramolecular water
reduction systems for improved efficiency using a series of
2
G1 equiv for complete poisoning. For example, CS has been
shown to poison most of the catalytic activity of rhodium
nanoclusters used for cyclohexene hydrogenation with less
(
(
(
49) Widegren, J. A.; Finke, R. G. J. Mol. Catal. A: Chem. 2003, 198,
17–341.
50) Yonezawa, T.; Tominaga, T.; Richard, D. J. Chem. Soc., Dalton Trans.
996, 783–789.
51) Wang, Q.; Liu, H.; Han, M.; Li, X.; Jiang, D. J. Mol. Catal. A: Chem.
3
1
1
997, 118, 145–151.
(
(
(
52) Guminski, C. J. Mater. Sci. 1989, 24, 3285–3288.
53) Jaska, C. A.; Manners, I. J. Am. Chem. Soc. 2004, 126, 9776–9785.
54) Hornstein, B.; Aiken, J.; Finke, R. Inorg. Chem. 2002, 41, 1625–
5
5
tailored bridging ligands. Future work will attempt to
develop complete water splitting systems using the system
described herein and an iridium-based water oxidation
1
638.
(
(
55) Bernhard, S.; Belser, P. Synthesis 1996, 2, 192–194.
56) McDaniel, N.; Coughlin, F.; Tinker, L.; Bernhard, S. J. Am. Chem.
Soc. 2008, 130, 210–217.
5
6
catalyst.
Inorganic Chemistry, Vol. 47, No. 22, 2008 10387

Cline et al.
Acknowledgment. We thank Emily Barton, Neal McDaniel,
and Lenny Tinker for their useful discussions. We thank
Radhika Vemula for her assistance in the work. S.B.
acknowledges support through a NSF CAREER award
ing cyclometalated iridium(III) and tris-2,2′-bipyridyl rhodium(III)
complexes is described. Synthetic modification of the photosensi-
∧
∧
+
∧
3+
tizer Ir(C N) (N N) and water reduction catalyst Rh(N N)
2
3
creates a family of catalysts with diverse properties. Parallel
screening of the various catalyst combinations and photoreaction
conditions allows the rapid development of an optimized photo-
catalytic system that achieves over 5000 turnovers with quantum
yields greater than 34%. This material is available free of charge
via the Internet at http://pubs.acs.org.
(
CHE-0449755).
Supporting Information Available: Experimental notes and
characterization for all ligands and complexes. Detailed experi-
mental notes and additional results from photoreactions and dynamic
quenching studies. An efficient homogeneous catalytic system for
the visible-light-induced production of hydrogen from water utiliz-
IC800988B
10388 Inorganic Chemistry, Vol. 47, No. 22, 2008