Toward exceeding the Shockley-Queisser limit: Photoinduced interfacial charge transfer processes that store energy in excess of the equilibrated excited state

Article abstract of DOI:10.1021/ja060470e

Nanocrystalline (anatase), mesoporous TiO₂ thin films were functionalized with [Ru(bpy)₂(deebq)]-(PF₆)₂, [Ru(bq)₂(deeb)](PF₆)₂, [Ru(deebq) ₂(bpy)](PF₆)₂, [Ru(bpy)(deebq)(NCS) ₂], or [Os(bpy)₂(deebq)](PF₆)₂, where bpy is 2,2′-bipyridine, bq is 2,2′-biquinoline, and deeb and deebq are 4,4′-diethylester derivatives. These compounds bind to the nanocrystalline TiO₂ films in their carboxylate forms with limiting surface coverages of 8 (± 2) × 10^-8 mol/cm². Electrochemical measurements show that the first reduction of these compounds (-0.70 V vs SCE) occurs prior to TiO₂ reduction. Steady state illumination in the presence of the sacrificial electron donor triethylamine leads to the appearance of the reduced sensitizer. The thermally equilibrated metal-to-ligand charge-transfer excited state and the reduced form of these compounds do not inject electrons into TiO₂. Nanosecond transient absorption measurements demonstrate the formation of an extremely long-lived charge separated state based on equal concentrations of the reduced and oxidized compounds. The results are consistent with a mechanism of ultrafast excited-state injection into TiO₂ followed by interfacial electron transfer to a ground-state compound. The quantum yield for this process was found to increase with excitation energy, a behavior attributed to stronger overlap between the excited sensitizer and the semiconductor acceptor states. For example, the quantum yields for [Os(bpy)₂(dcbq)]/TiO₂ were φ(417 nm) = 0.18 ± 0.02, φ(532.5 nm) = 0.08 ± 0.02, and φ(683 nm) = 0.05 ± 0.01. Electron transfer to yield ground-state products occurs by lateral intermolecular charge transfer. The driving force for charge recombination was in excess of that stored in the photoluminescent excited state. Chronoabsorption measurements indicate that ligand-based intermolecular electron transfer was an order of magnitude faster than metal-centered intermolecular hole transfer. Charge recombination was quantified with the Kohlrausch-Williams-Watts model.

Full text of DOI:10.1021/ja060470e

Published on Web 06/06/2006
Toward Exceeding the Shockley-Queisser Limit:
Photoinduced Interfacial Charge Transfer Processes that
Store Energy in Excess of the Equilibrated Excited State
Paul G. Hoertz,† Aaron Staniszewski,† Andras Marton,† Gerard T. Higgins,†
Christopher D. Incarvito,‡ Arnold L. Rheingold,‡ and Gerald J. Meyer*,†
Contribution from the Departments of Chemistry and Materials Science and Engineering,
Johns Hopkins UniVersity, 3400 North Charles Street, Baltimore, Maryland 21218, and
Department of Chemistry and Biochemistry, UniVersity of California, San Diego,
9500 Gilman DriVe, La Jolla, California 92093-0358
Received January 31, 2006; E-mail: meyer@jhu.edu
Abstract: Nanocrystalline (anatase), mesoporous TiO2 thin films were functionalized with [Ru(bpy)2(deebq)]-
(PF6)2, [Ru(bq)2(deeb)](PF6)2, [Ru(deebq)2(bpy)](PF6)2, [Ru(bpy)(deebq)(NCS)2], or [Os(bpy)2(deebq)](PF6)2,
where bpy is 2,2′-bipyridine, bq is 2,2′-biquinoline, and deeb and deebq are 4,4′-diethylester derivatives.
These compounds bind to the nanocrystalline TiO2 films in their carboxylate forms with limiting surface
coverages of 8 (( 2) × 10-8 mol/cm2. Electrochemical measurements show that the first reduction of these
compounds (-0.70 V vs SCE) occurs prior to TiO2 reduction. Steady state illumination in the presence of
the sacrificial electron donor triethylamine leads to the appearance of the reduced sensitizer. The thermally
equilibrated metal-to-ligand charge-transfer excited state and the reduced form of these compounds do
not inject electrons into TiO2. Nanosecond transient absorption measurements demonstrate the formation
of an extremely long-lived charge separated state based on equal concentrations of the reduced and oxidized
compounds. The results are consistent with a mechanism of ultrafast excited-state injection into TiO2 followed
by interfacial electron transfer to a ground-state compound. The quantum yield for this process was found
to increase with excitation energy, a behavior attributed to stronger overlap between the excited sensitizer
and the semiconductor acceptor states. For example, the quantum yields for [Os(bpy)2(dcbq)]/TiO2 were
φ(417 nm) ) 0.18 ( 0.02, φ(532.5 nm) ) 0.08 ( 0.02, and φ(683 nm) ) 0.05 ( 0.01. Electron transfer to
yield ground-state products occurs by lateral intermolecular charge transfer. The driving force for charge
recombination was in excess of that stored in the photoluminescent excited state. Chronoabsorption
measurements indicate that ligand-based intermolecular electron transfer was an order of magnitude faster
than metal-centered intermolecular hole transfer. Charge recombination was quantified with the Kohlrausch-
Williams-Watts model.
Introduction
tion that multiple electron-hole pairs can be generated in a
semiconductor nanoparticle after absorption of a single photon
The maximum solar-to-electrical energy conversion efficiency
from a single junction photovoltaic cell is about 31% (1 sun,
air mass 1.5 spectral distribution).1 This celebrated limit was
first calculated by Shockley and Queisser under the assumption
that solar photons with energy larger than the semiconductor
band gap, Eg, lose all energy in excess of Eg.1 The theoretical
efficiency exceeds 60% when electron-hole pairs are converted
to electrical power prior to band edge thermalization.2 Attempts
to fabricate such “hot carrier” solar cells have thus far been
unsuccessful, but may be enabled by recent demonstrations that
a single photon can create multiple electron hole pairs in
semiconductor nanoparticles.3-5 However, the recent demonstra-
represents a significant advance.3-5
The Shockley-Queisser limit also applies to molecular solar
cells. Here too the realization of >31% efficiencies requires
that energy stored in molecular excited states be collected prior
to vibrational relaxation to the lowest electronic state. Relatively
few strategies for accomplishing this exist.6,7 Vibrational
relaxation in inorganic and organic excited states is known to
occur on ultrafast time scales. Therefore, any successful strategy
would require subpicosecond charge-transfer processes.
The recent demonstrations of ultrafast interfacial electron
transfer and rapid charge trapping at dye-sensitized TiO2
interfaces may provide unanticipated opportunities for exceeding
the Shockley-Queisser limit.8-16 Indeed, there is experimental
† Johns Hopkins University.
‡ University of California, San Diego.
(1) Shockley, W.; Queisser, H. J. J. Appl. Phys. 1961, 32, 510.
(2) Ross, R. T.; Nozik, A. J. J. Appl. Phys. 1982, 53, 3813.
(3) Schaller, R.; Klimov, V. Phys. ReV. Lett. 2004, 92, 186601.
(4) Ellingson, R. J.; Beard, M. C.; Johnson, J. C.; Yu, P.; Micic, O. I.; Nozik,
A. J.; Shabaev, A.; Efros, A. L. Nanolett. 2005, 5, 865.
(5) Nozik, A. J. Inorg. Chem. 2005, 44, 6893.
(6) (a) Becker, R. S.; Donlan, E.; Balke, D. E. J. Chem. Phys. 1969, 50, 239.
(b) Becker, R. S.; Pellicciolo, A. P.; Romani, A.; Favaro, G. J. Am. Chem.
Soc. 1999, 121, 2104.
(7) (a) Koval, C. A.; Segar, P. R. J. Am. Chem. Soc. 1989, 111, 2001. (b)
Koval, C. A.; Segar, P. R. J. Phys. Chem. 1990, 94, 203.
9
8234
J. AM. CHEM. SOC. 2006, 128, 8234-8245
10.1021/ja060470e CCC: $33.50 © 2006 American Chemical Society

Toward Exceeding the Shockley−Queisser Limit
A R T I C L E S
Scheme 1
evidence for excited state electron injection from the initially
formed Franck-Condon state, i.e., interfacial electron transfer
without vibrational energy loss in the photoexcited dye.9
However, whether such ultrafast electron injection is beneficial,
or even necessary, for efficient energy conversion in regenerative
dye-sensitized solar cells remains unknown.17 Quantitative
injection would be expected from the long-lived metal-to-ligand
charge transfer (MLCT) excited states of the commonly utilized
Ru(II) polypyridyl sensitizers even if the rate constant was
slowed by ∼3 orders of magnitude.18
state. Our approach exploits ultrafast dye-sensitized electron
injection into TiO2 nanocrystallites followed by electron transfer
to a molecular acceptor that would not be reduced by the
thermally equilibrated excited state. The interfacial energetics
shown on the right-hand side of Scheme 1, coupled with the
well-known ultrafast injection,8-16 are indeed expected to give
rise to such a semiconductor-mediated charge transfer pathway.
Since the acceptors are reduced only when electrons are injected
from upper vibrational excited states, their reduction serves as
a direct probe of “hot electron” involvement in interfacial
phenomena.19,20 The studies reported here represent proof-of-
principle examples of this behavior.
Complete excited state quenching, electron transfer rate
constants, and interfacial energetics reported in the literature
strongly suggest that the conduction band edge, Ecb, lies below
the reduction potential of the thermally equilibrated excited
(thexi) state, S+/*, in regenerative dye-sensitized solar cells,
Scheme 1, left hand side.18 Therefore, it matters little whether
the Franck-Condon state (path A) or the thermally equilibrated
excited state (path B) injects the electron. Loss of the excess
energy occurs on one side of the interface or the other: phonon
release in the solid or nonradiative decay of the excited state.
Both pathways ultimately yield the same thermalized products,
a conduction band electron and an oxidized dye.
Experimental Section
Materials. Tetrabutylammonium perchlorate (Fluka), acetonitrile
(Burdick and Jackson), ethanol (Pharmco), RuCl3‚xH2O (Alfa Aesar),
OsCl3‚xH2O (Alfa Aesar), lithium iodide (Alfa Aesar), and 4,4′-
dicarboxylic acid-2,2′-biquinoline (dcbqH2, Fluka) were used as
received. Acetone, sulfuric acid, nitric acid, potassium dichromate, 2,2′-
biquinoline (bq), 2,2′-bipyridine (bpy), LiCl, trifluoromethanesulfonic
acid, dichlorobenzene, dimethyl formamide, methylene chloride, chlo-
roform, tetrabutylammonium thiocyanate, tetrabutylammonium iodide,
ammonium hexafluorophosphate, and [Ru(bpy)3](Cl)2 were purchased
from Aldrich and were reagent grade or better.
Herein we describe novel molecular charge separation
processes that ultimately store more free energy than the thexi
Synthesis. 4,4′-Dicarboxylic acid-2,2′-bipyridine (dcbH2). 4,4′-
Dimethyl-2,2′-bipyridine (5.0 g) in heated H2SO4 (125 mL, 70-80 °C)
was oxidized by adding solid potassium dichromate (24 g) slowly. The
temperature was held consistently between 70 and 80 °C during the
transfer. The deep green mixture was then poured over 800 mL of ice/
H2O, affording a light yellow precipitate that was isolated by vacuum
filtration and was washed with H2O. The solid was then refluxed in
50% HNO3 (150 mL) for 4 h. The cooled solution was then poured
over ice and diluted with 800 mL of H2O. A white powder was isolated
by vacuum filtration and washed with H2O (5.99 g, 90%).21
(8) Tachibana, Y.; Moser, J. E.; Gra¨tzel, M.; Klug, D. R.; Durrant, J. J. Phys.
Chem. B 1996, 100, 20056.
(9) Hannappel, T.; Burfeindt, B.; Storck, W.; Willig, F. J. Phys. Chem. B 1997,
101, 6799.
(10) (a) Heimer, T. A.; Heilweil, E. J. J. Phys. Chem. B 1997, 101, 10990. (b)
Heimer, T. A.; Heilweil, E. J.; Bignozzi, C. A.; Meyer, G. J. J. Phys. Chem.
A 2000, 104, 4256.
(11) Ellingson, R. J.; Asbury, J. B.; Ferrere, S.; Ghosh, H. N.; Sprague, J. R.;
Lian, T.; Nozik, J. Phys. Chem. B 1997, 101, 6455.
(12) Piotrowiak, P.; Galoppini, E.; Wei, Q.; Meyer, G. J.; Wiewior, P. J. Am.
Chem. Soc. 2003, 125, 5278.
(13) (a) Benko, G.; Kallioinen, J.; Korppi-Tommola, J. E. I.; Yartsev, A. P.;
Sundstrom, V. J. Am. Chem. Soc. 2002, 124, 489. (b) Benko, G.;
Myllyperkio, P.; Pan, J.; Yartsev, A. P.; Sundstrom, V. J. Am. Chem. Soc.
2003, 125, 1118. (c) Kallioinen, J.; Benko, G.; Sundstrom, V.; Korppi-
Tommola, J. E. I.; Yartsev, A. P. J. Phys. Chem. B 2002, 106, 4396.
(14) Kuciauskas, D.; Monat, J. E.; Villahermosa, R.; Gray, H. B.; Lewis, N. S.;
McCusker, J. K. J. Phys. Chem. B 2002, 106, 9347.
4,4′-Diethylester-2,2′-bipyridine (deeb). This ligand was synthe-
sized from dcbH2 by a literature preparation.22 1H NMR δ (CD2Cl2):
8.93 (2H, dd), 8.85 (2H, dd), 7.89 (2H, dd), 4.44 (4H, q, CH2), 1.44
(6H, t, CH3).
(15) (a) Asbury, J. B.; Hao, E.; Wang, Y.; Lian, T. J. Phys. Chem. B 2000,
104, 11957. (b) Asbury, J. B.; Hao, E.; Wang, Y. Q.; Ghosh, H. N.; Lian,
T. Q. J. Phys. Chem. B 2001, 105, 4545.
(19) (a) Ferrere, S.; Gregg, B. A. J. Am. Chem. Soc. 1998, 120, 843. (b) Islam,
A.; Hara, K.; Singh, L. P.; Katoh, R.; Yanagida, M.; Murata, S.; Takahashi,
Y.; Sugihara, H.; Arakawa, H. Chem. Lett. 2000, 490. (c) Moser, J. E.;
Gra¨tzel, M. Chimia 1998, 52, 160.
(16) Hoertz, P. G.; Thompson, D. W.; Friedman, L. A.; Meyer, G. J. J. Am.
Chem. Soc. 2002, 124, 9690.
(17) Haque, S. A.; Palomeras, E.; Cho, B. M.; Green, A. N. M.; Hirata, N.;
Klug, D. R.; Durrant, J. R. J. Am. Chem. Soc. 2005, 127, 3456.
(18) (a) Gra¨tzel, M. Nature 2001, 414, 338. (b) Watson, D. F.; Meyer, G. J.
Annu. ReV. Phys. Chem. 2005, 56, 119.
(20) Huber, R.; Sporlein, S.; Moser, J. E.; Gra¨tzel, M.; Wachtveitl, J. J. Phys.
Chem. B 2000, 104, 8995.
(21) Oki, A. R.; Morgen, R. J. Synth. Commun. 1995, 25, 4093.
(22) Maerker, G.; Case, F. H. J. Am. Chem. Soc. 1958, 80, 2747.
9
J. AM. CHEM. SOC. VOL. 128, NO. 25, 2006 8235

A R T I C L E S
Hoertz et al.
4,4′-Dicarboxylato-2,2′-bipyridine Disodium Salt (dcb(Na)2). A
solid sample of dcbH2 (1.04 g) was added to 20 mL of deionized H2O
and treated with NaOH (aq) (25% w/v) until the solid was completely
dissolved. During this process, the pH increased from 3 to 9. Acetone
(700 mL) was then added to precipitate out dcb(Na)2 (1.21 g, 99%).
4,4′-Diethylester-2,2′-biquinoline (deebq). A 10 g sample of dcbqH2
was added to 360 mL of stirred EtOH and was then chilled in an ice/
H2O bath. Sulfuric acid (92 mL) was slowly added. The suspension
was refluxed for 27 h; 1 h into the reflux, the ligand completely
dissolved. Neutralization with NaOH (aq, 25% w/v) led to precipitation
of a yellow solid, which was filtered and dried in an oven. The solid
was dissolved in methylene chloride, filtered, and placed on a
rotoevaporator to remove the methylene chloride (9.7 g, 84%). 1H NMR
δ (CDCl3): 9.35 (2H, s), 8.80 (2H, d), 8.35 (2H, d), 7.85 (2H, td),
7.70 (2H, td), 4.60 (4H, q, CH2), 1.55 (6H, t, CH3).
(bpy)2(OTf)2] (250 mg, 0.31 mmol) and deebq (1.15 g, 2.87 mmol)
were dissolved in dry acetone (10 mL) and refluxed for 2 days under
argon. Upon cooling to room temperature, the acetone was removed
by rotoevaporation. The solid was added to 50 mL of methylene
chloride and filtered, and the methylene chloride was removed under
a vacuum. The resultant solid was dissolved in acetone (50 mL) and
filtered. The slow addition of diethyl ether (200 mL) resulted in
precipitation of the desired product. Recrystallization from acetone was
repeated 4 times with 10 mL of acetone and 40 mL of ether. The filtrate
in each case was brown-green in color; after the last recrystallization,
the filtrate was dark green with no hints of brown. The black solid
was dissolved in 15 mL of acetone. After the solution was filtered, the
filtrate was treated with 15 mL of deionized H2O. Dropwise addition
of 2 M NH4PF6 (aq, 5 mL) led to precipitation of a black solid. After
adding 20 mL of deionized H2O, the solid was isolated via filtration
and washed with H2O followed by ether. After drying in an oven at 75
°C, the solid was dissolved in 10 mL of acetone and placed in an ether
chamber for slow vapor diffusion recrystallization. This afforded good
4,4′-Dicarboxylato-2,2′-biquinoline Disodium Salt (dcbq(Na)2).
About 5 g of dcbqH2 was added to 50 mL of deionized H2O and treated
with NaOH (aq) (25% w/v) until the solid was completely dissolved.
During this process, the pH increased from 5 to 8. Acetone (750 mL)
was then added to precipitate dcbq(Na)2 (95%).
1
crystals (50 mg, 14%). H NMR δ (CD3CN): 8.90 (2H, s), 8.69 (2H,
dd), 8.42 (4H, dd), 7.90 (4H, m), 7.63 (4H, m), 7.49 (2H, d), 7.40
(2H, td), 7.23 (4H, m), 6.87 (2H, d), 4.44 (4 H, q, CH2), 1.51 (6H, t,
CH3). Elem. Anal. Calcd: C, 44.30; H 3.04; N, 7.05. Found: C, 45.22;
H 3.32; N, 6.90. MALDI-MS: Calcd for (MH+), 904; found, 904.
[Ru(bpy)2(deebq)](PF6)2. (a) Ru(bpy)2Cl2 was synthesized according
to a literature preparation.23 UV-vis (CH2Cl2), nm: 380, 557. (b) Ru-
(bpy)2(OTf)2 (where OTf is trifluoromethanesulfonate) was prepared
by adding Ru(bpy)2Cl2 (1.67 g) to 150 mL of argon-purged o-
dichlorobenzene. Trifluoromethanesulfonic acid (HOTf, 1 mL) was
added to the suspension, and the reaction mixture immediately turned
from purple to red. After 1 h of gentle stirring, the solution was filtered
and a brick-red precipitate was washed with copious amounts of ether
to isolate Ru(bpy)2(OTf)2, in 97% yield (2.28 g).24 1H NMR δ (CH3-
CN): 9.33 (2H, dd), 8.51 (2H, d), 8.39 (2H, d), 8.25 (2H, td), 7.94
(2H, td), 7.87 (2H, m), 7.60 (2H, d), 7.30 (2H, m). (c) Ru(bpy)2(OTf)2
(0.5 g, 0.7 mmol) was dissolved in dry acetone (150 mL) with deebq
(360 mg, 0.9 mmol) and was refluxed under argon for 1 day. The
acetone was removed by rotoevaporation. The red solid was treated
with deionized H2O (80 mL), and the solution was filtered. The filtrate
was treated with 2 M NH4PF6 (aq, 5 mL) to afford a red precipitate,
which was washed with H2O followed by ether. After drying in an
oven at 75 °C, the solid was dissolved in 10 mL of acetone and placed
in an ether chamber for slow vapor diffusion recrystallization. This
afforded good crystals (50 mg, 14%). 1H NMR δ (CD3CN): 8.97 (2H,
s), 8.69 (2H, dd), 8.42 (4H, dd), 8.06 (4H, m), 7.77 (4H, m), 7.69 (2H,
m), 7.36 (6H, m), 7.2 (2H, d), 4.59 (4 H, q, CH2), 1.51 (6H, t, CH3).
Elem. Anal. Calcd: C, 46.71; H 3.40; N, 7.43. Found: C, 48.44; H,
3.43; N, 7.74. FAB-MS: Calcd for (M - PF6) 958.8, found 958.8;
Calcd for (M - 2PF6) 813.8, found 814.0.
[Ru(bq)2(deeb)](PF6)2. (a) Ru(bq)2Cl2 was synthesized using the
literature procedure for Ru(bpy)2Cl2 (67%).23 UV-vis (CHCl3), nm:
431, 650 (sh), 737. (b) Ru(bq)2Cl2 (0.5 g, 0.7 mmol) and deeb (0.32 g,
1.1 mmol) were added to 4:1 EtOH/H2O (10 mL) and refluxed under
argon for 24 h protected from room light using aluminum foil. After
cooling to room temperature, the reaction solution was treated with 20
mL of deionized H2O and then filtered to remove any excess free ligand.
The reaction solution was treated with 2 M NH4PF6 (aq, 5 mL) giving
a precipitate that was filtered and washed with H2O followed by ether.
The solid was dissolved in a minimal amount of dry acetone and
recrystallized by slow ether addition with stirring. This recrystallization
was performed repetitively until the acetone/ether filtrates were clear,
1
approx 3-4 times (70 mg, 9%). H NMR δ (CD3CN): 9.00 (4H, q),
8.82 (2H, d), 8.49 (2H, d), 8.13 (6H, m), 7.85 (4H, m), 7.52 (2H, m),
7.37 (4H, m), 7.06 (4H, m), 6.94 (2H, m), 4.32 (4 H, q, CH2), 1.31
(6H, t, CH3). Elem. Anal. Calcd: C, 51.88; H, 3.35; N, 6.95. Found:
C, 50.94; H, 4.27; N, 6.85. FAB-MS: Calcd for (M - 2PF6) 914.0,
found 914.4; Calcd for (M - 2PF6 - bq) 657.7, found 658.1.
[Ru(bpy)(deebq)2](PF6)2. (a) Ru(deebq)2Cl2 was synthesized using
the following procedure. RuCl3‚xH2O (0.2 g), deebq (0.7 g, 1.8 mmol),
and LiCl (0.42 g, 9.9 mmol) were added to argon-purged 1:1 EtOH/
CHCl3 (20 mL) and refluxed under argon for 16 h. After cooling to
room temperature, ascorbic acid (0.22 g, 1.2 mmol) was added, and
the brown reaction solution was refluxed again under argon for 3 h,
during which time the reaction solution turned dark green. Upon cooling
to room temperature, the reaction solution was rotoevaporated to remove
most of the CHCl3. The solution was transferred to a filter using
deionized H2O, and the resulting black solid was washed with copious
amounts of H2O followed by ether (0.65 g, 80%). UV-vis (CH2Cl2),
nm: 464, 676, 750 (sh). (b) [Ru(bpy)(deebq)2](PF6)2 was prepared using
the following procedure. Ru(deebq)2Cl2 (1.6 g, 1.7 mmol) was added
to argon purged EtOH (100 mL). Silver(I) hexafluorophosphate (1.25
g, 5 mmol) dissolved in 5 mL of argon purged EtOH was then added,
producing a deep blue reaction solution. The reaction solution was
refluxed under argon for 1 h and cooled to room temperature. After
100 mL of argon purged deionized H2O and bpy (2.57 g, 16.5 mmol)
were added, the reaction solution was refluxed for 24 h under argon
and protected from room light using aluminum foil. Upon cooling to
room temperature, the purple reaction solution was filtered; the filtrate
was treated with 2 M NH4PF6 (aq, 10 mL), giving a precipitate that
was filtered and washed with H2O until the washings were clear. The
purple solid was then dissolved in acetonitrile and was filtered. After
rotoevaporation of the acetonitrile, the purple solid was dissolved in
[Os(bpy)2(deebq)](PF6)2. (a) Os(bpy)2Cl2 was prepared according
to a modified literature preparation.25 LiCl (0.95 g) and bpy (0.48 g)
were added to argon-purged DMF (25 mL) followed by OsCl3‚xH2O
(0.52 g). The solution was refluxed under argon for 2.5 h and then
allowed to cool to room temperature. The reaction solution was refluxed
for an additional hour under argon in the presence of 1 mL of
triethylamine, cooled to room temperature, poured over 100 mL of dry
acetone, and cooled to -5 °C overnight. Filtration yielded a purple-
1
brown solid (450 mg, 52%). H NMR δ (d6-DMSO): 9.67 (2H, d),
8.59 (2H, dd), 8.38 (2H, dd), 7.59 (4H, m), 7.30 (4H, t), 7.56 (2H, td).
(b) Os(bpy)2(OTf)2 was prepared by addition of Os(bpy)2Cl2 (440 mg)
to 125 mL of argon-purged o-dichlorobenzene. Trifluoromethane-
sulfonic acid (HOTf, 0.75 mL) was added to the suspension, and the
reaction mixture immediately turned from purple to red. After stirring
for 1 h, the reaction solution was filtered to isolate a dark green
precipitate that was washed with copious amounts of ether (320 mg,
52%).24 1H NMR δ (CH3CN): 9.26 (2H, d), 8.51 (2H, d), 8.37 (2H,
d), 7.95 (2H, td), 7.88 (4H, m), 7.56 (2H, d), 7.18 (2H, m). (c) [Os-
(23) Sullivan, B. P.; Salmon, D. J.; Meyer, T. J. Inorg. Chem. 1978, 17, 3334.
(24) Lay, P. A.; Sargerson, A. M.; Taube, H. Inorg. Synth. 1986, 24, 291.
(25) Geno, M. J. K.; Dawson, J. H. Inorg. Chim. Acta 1985, 97, 241.
9
8236 J. AM. CHEM. SOC. VOL. 128, NO. 25, 2006

Toward Exceeding the Shockley−Queisser Limit
A R T I C L E S
dry acetone, the solution was filtered, and the purple solution was placed
in an ether chamber for slow recrystallization (0.46 g, 21%). 1H NMR
δ (CD3CN): 9.38 (2H, s), 9.22 (2H, s), 8.74 (2H, d), 8.51 (2H, d),
7.85 (2H, d), 7.75 (2H, t), 7.60 (4H, m), 7.45 (4H, m), 7.30 (2H, d),
7.15 (4H, m), 7.00 (2H, m), 4.70 (4 H, q, CH2), 4.60 (4 H, q, CH2),
1.60 (6H, t, CH3), 1.50 (6H, t, CH3). Elem. Anal. Calcd: C, 51.68; H,
3.59; N, 6.23. Found: C, 50.47; H, 3.66; N, 6.07.
taking 128-256 scans at 4 cm-1 resolution. For TiO2 samples, the
background was taken using either a blank TiO2/glass or TiO2/FTO
slide. Elemental analysis was performed by Atlantic Microlabs, Inc.
X-ray crystallography was performed by Arnold Rheingold and co-
workers at the University of Delaware. Matrix Assisted Laser Desorp-
tion Ionization Mass Spectrometry (MALDI-MS) measurements were
performed on a Kratos MALDI-TOF model SEQ mass spectrometer
using R-cyano-4-hydroxycinnamic acid as the matrix. Fast Atom
Bombardment Mass Spectrometry (FAB-MS) measurements were
obtained on a VG70S mass spectrometer; samples were suspended in
a p-nitrobenzyl alcohol matrix.
MO2 Preparations. Transparent films of colloidal TiO2 or ZrO2
nanoparticles were prepared by a previously described sol-gel
procedure.27,28 Sensitizer binding was carried out by placing a MO2
film in an approximately millimolar sensitizer acetonitrile solution for
∼24 h.
[Ru(bpy)(deebq)(NCS)2]. (a) [Ru(bpy)(deebq)2](PF6)2 (0.17 g, 0.13
mmol) was dissolved in 80 mL of acetonitrile and irradiated (1000 W
Xe lamp, λ > 570 nm) in a 100 mL borosilicate round-bottomed flask
under argon for 4 days.26 Complete conversion to [Ru(bpy)(deebq)-
(CH3CN)2](PF6)2 was apparent from UV-vis spectra. The acetonitrile
was removed, and the resultant red solid was dissolved in 15 mL of
dry acetone that was treated with 15 mL of H2O. The deebq ligand
was removed by filtration. The filtrate was then treated with 2 M NH4-
PF6 (aq, 5 mL), giving a red precipitate, which was filtered and washed
with H2O followed by ether (0.11 g, 80%). 1H NMR δ (CD3CN): 9.75
(1H, dd), 9.05 (3H, m), 8.80 (1H, s), 8.70 (1H, d), 8.20 (2H, m), 8.00
(4H, m), 7.85 (1H, m), 7.77 (1H, d), 7.70 (1H, m), 7.50 (1H, m), 7.25
(1H, m), 7.00 (1H, d), 4.75 (4H, q, CH2), 4.55 (4H, q, CH2), 2.60 (3H,
s, CH3CN-Ru), 2.40 (3H, s, CH3CN-Ru), 1.60 (6H, t, CH3), 1.50
(6H, t, CH3). UV-vis (CH3CN), nm: 490 (sh), 525. (b) [Ru(bpy)-
(deebq)(CH3CN)2](PF6)2 (40 mg) was dissolved in argon purged EtOH
(5 mL) containing tetrabutylammonium thiocyanate (25 mg) and
refluxed for 4 h, during which time, a blue-green precipitate formed.
The solid was isolated by filtration and rinsed with EtOH and ether
(13 mg, 43%). 1H NMR δ (CD3CN): 9.49 (1H, d), 9.33 (1H, d), 8.98
(1H, dd), 8.89 (1H, s), 8.74 (1H, d), 8.72 (1H, s), 8.14 (1H, d), 8.00
(3H, m), 7.90 (1H, m), 7.75 (1H, m), 7.66 (1H, td), 7.55 (1H, m), 7.36
(1H, d), 7.25 (2H, m), 7.00 (1H, m), 4.66 (2H, q, CH2), 4.54 (2H, q,
CH2), 1.59 (3H, t, CH3), 1.50 (3H, t, CH3). ATR-IR: 2095 cm-1 and
2071 cm-1 (NdCdS).
Photoluminescence. Corrected photoluminescence (PL) spectra were
obtained with a Spex Fluorolog that had been calibrated with a standard
tungsten-halogen lamp using procedures provided by the manufacturer.
Sensitized films were placed diagonally in a 1 cm square quartz cuvette,
immersed in acetonitrile, and purged with nitrogen for at least 15 min.
The excitation beam was directed 45° to the film surface, and the
emitted light was monitored from the front face of the surface-bound
sample and from the right angle in the case of fluid solutions.
Photoluminescence quantum yield measurements were performed using
the optically dilute technique29 with Ru(bpy)3(PF6)2 in acetonitrile as
the actinometer and calculated by eq 1:
φPL ) (Ar/As)(Is/Ir)(ns/nr)2φr
(1)
where Ar and As are the absorbances of the actinometer and sample,
respectively, Ir and Is are the integrated photoluminescence of the
actinometer and sample, respectively, nr and ns are the refractive indexes
of the actinometer and sample solvents, respectively, and φr is the
quantum yield for Ru(bpy)3(PF6)2 in acetonitrile (φr ) 0.062). For
lithium titration experiments, 0.1 M LiClO4 in CH3CN was added via
syringe to the cuvette that was continuously purged with nitrogen.
Time-resolved photoluminescence decays were acquired on a
nitrogen-pumped dye laser (460 nm) apparatus that has been previously
described.30 For solution studies, the samples were optically dilute (A
≈ 0.1 at λmax), and the kinetic traces were fit to a first-order kinetic
model. For TiO2 and ZrO2 studies, the time-resolved photoluminescence
decays were fit to a parallel first- and second-order kinetic model, eq
2,
[Ru(bpy)2(dcbq)]. Ru(bpy)2Cl2‚2H2O (54 mg, 0.10 mmol) and dcbq-
(Na)2 (70 mg, 0.18 mmol) were added to N2-saturated deionized H2O
(5 mL). Refluxing under N2 for 8 h resulted in the formation of a brick-
red precipitate that was filtered and washed with deionized H2O
followed by acetone (69 mg, 88%). 1H NMR δ (d6-DMSO): 8.16 (6H,
m), 8.03 (2H, s), 7.54 (4H, t), 7.32 (4H, t), 6.92 (6H, m), 6.53 (2H,
td), 6.41 (2H, d).
[Os(bpy)2(dcbq)]. Os(bpy)2Cl2‚2H2O (33 mg, 0.052 mmol) and
dcbq(Na)2 (70 mg, 0.20 mmol) were added to N2-saturated deionized
H2O (5 mL). After refluxing under N2 for 8 h, the reaction solution
was filtered (removing a black solid), and acetone (250 mL) was added
to the filtrate. The solution was cooled in a freezer for 4 days, over
which time the desired product precipitates out as a black solid. The
product was isolated by vacuum filtration and washed with deionized
k1 exp(- k1t)
1
PLI(t) ) B
(2)
H2O and acetone (10 mg, 20%). H NMR δ (d6-DMSO): 8.80 (2H,
(
)
k1 + p - p exp(- k1t)
dd), 8.72 (4H, t), 8.59 (2H, s), 7.92 (4H, m), 7.70 (4H, dd), 7.43 (6H,
m), 7.07 (2H, m), 6.75 (2H, d).
where k1 is a first-order rate constant analogous to the solution and B
is a constant.31 The parameter p is the product of the observed second-
order rate constant, k2, and the initial concentration of ruthenium excited
[Ru(bq)2(dcbH2)](PF6)2. [Ru(bq)2(Cl)2] (150 g, 0.21 mmol) was
added to argon-purged DMF (12 mL) and refluxed under argon for 2
h. After cooling to room temperature, the reaction solution was treated
with dcb2- (200 mg, 0.62 mmol) dissolved in 12 mL of argon-purged
H2O, and the solution was refluxed again for 18 h. The solution was
filtered to remove a green precipitate, and the dark red filtrate was
treated with 0.4 M HPF6 (aq, 1 mL) giving a red-brown precipitate
states, [Ru2+*
]t)0. For studies involving the sensitizers on TiO2 or ZrO2,
the excitation beam was directed 45° to the film surface, and the emitted
light was collected at 90°.
Electrochemistry. Cyclic voltammetry was performed in 0.1 M
tetrabutylammonium hexafluorophosphate (TBAPF6/CH3CN) electrolyte
1
that was filtered and washed with H2O followed by ether. H NMR δ
(CD3CN): 8.98 (4H, q), 8.81 (2H, d), 8.46 (2H, d), 8.25 (2H, s), 8.12
(2H, d), 8.06 (2H, d), 7.81 (4H, m), 7.50 (2H, m), 7.31 (4H, m), 7.04
(4H, m), 6.90 (2H, m).
Characterization. 1H NMR spectra were obtained on a Bruker 300
Hz AMX FT-NMR spectrometer. Attenuated Total Reflectance (ATR)
measurements were obtained using a Golden Gate Single Reflection
Diamond ATR apparatus on a Nexus 670 Thermo-Nicolet FT-IR and
(27) (a) O’Regan, B.; Moser, J.; Anderson, M.; Gra¨tzel, M. J. Phys. Chem. 1990,
94, 8720. (b) Barbe, C. J.; Arendse, F.; Comte, P.; Jirousek, M.; Lenzmann,
F.; Shklover, V.; Gra¨tzel, M. J. Am. Ceram. Soc. 1997, 80, 3157. (c) Heimer,
T. A.; D’Arcangelis, S. T.; Farzad, F.; Stipkala, J. M.; Meyer, G. J. Inorg.
Chem. 1996, 35, 5319.
(28) Qu, P.; Meyer, G. J. Langmuir 2001, 17, 6720 and references therein.
(29) Demas, J. N.; Crosby, G. A. J. Phys. Chem. 1971, 75, 991.
(30) Castellano, F. N.; Heimer, T. A.; Tandhasetti, T.; Meyer, G. J. Chem. Mater.
1994, 6, 1041.
(31) (a) Kelly, C. A.; Farzad, F.; Thompson, D. W.; Meyer, G. J. Langmuir
1999, 15, 731. (b) Higgens, G. T.; Bergeron, B. V.; Hasselmann, G. M.;
Farzad, F.; Meyer, G. J. J. Phys. Chem. B 2006, 110, 2598.
(26) von Zelewsky, A.; Gremaud, G. HelVetica Chimica Acta 1988, 71, 1108.
9
J. AM. CHEM. SOC. VOL. 128, NO. 25, 2006 8237

A R T I C L E S
Hoertz et al.
Figure 1. X-ray crystal structures of (a) [Os(bpy)2(deebq)](PF6)2 and (b) [Ru(bpy)2(deebq)](PF6)2.
with a sensitizer concentration of ∼1 mM. A BAS model CV-50W
potentiostat was used in a standard three-electrode arrangement with a
glassy carbon working electrode, a Pt gauze counter electrode, and Ag/
AgCl as the reference electrode. Cyclic voltammetry of the sensitizers
bound to TiO2 was performed in a similar manner with the sensitizer/
TiO2 films deposited on FTO glass as the working electrodes submerged
in 0.1 M tetrabutylammonium perchlorate (TBAClO4) acetonitrile. The
CV experiments were carried out at room temperature under argon.
shots. Samples were nitrogen purged for at least 15 min prior to transient
absorption studies.
Comparative actinometry was carried out with [Ru(bpy)3](PF6)2
doped into a PMMA thin film deposited on a glass microscope slide
as the actinometer.32 The extinction coefficient difference between the
excited-state and ground-state has been previously reported to be (-1.00
( 0.09) × 104 M-1 cm-1 at 450 nm, with an excited-state quantum
yield of unity.32 Extinction coefficients for the reduced and oxidized
compounds were obtained from spectroelectrochemical measurements.
Resonance Raman. Spectra were collected using a Spex 14018
double monochromator equipped with a Spex Compudrive controller
and a Hamamatsu R943-02 photomultiplier (PMT) cooled by a Products
for Research thermoelectric liquid heat exchanged housing. The PMT
signal was processed with a Hamamatsu C3866 discriminator/amplifier
and counted with a computer interfaced Philips PM 6680 counter.
LabView was used for data acquisition and monochromator control.
Excitation at 406.7, 413.1, 415.4, 468.0, 476.2, 482.5, 520.8, 530.9,
and 568.2 nm was provided by a Coherent Innova Sabre Kr+ laser.
Plasma lines were eliminated with a grating. The laser beam was
focused on the sample (in a melting point tube) and maintained at 50
( 5 mW. Scattered light was collected at 45° from the excitation using
a conventional two-lens collection system. Spectra were acquired at 1
cm-1 steps and with a 10 s integration time.
Spectroelectrochemistry. Spectroelectrochemistry of derivatized
TiO2 electrodes was performed in a three-electrode custom designed
long cuvette cell with a sensitizer/TiO2/FTO film as the working
electrode, a Pt gauze as the counter electrode, and a Ag/AgCl reference
electrode, all submerged in 0.1 M TBAClO4/CH3CN under a nitrogen
atmosphere. Oxidative spectroelectrochemistry was performed on
derivatized TiO2 electrodes by stepping the potential from 1.2 to 1.65
V in the presence of ∼10-5 M redox mediator. The redox mediator
utilized depended on the sensitizer being studied. For [Os(bpy)2(dcbq)]2+
/
TiO2, the redox mediator was [Ru(bpy)3](PF6)2. For the remainder of
the sensitizers, the redox mediator was the solvated version of the
sensitizer. Comparison of E1/2(MIII/II) of the sensitizers in solution and
on TiO2 shows that the adsorbed complex has a slightly more negative
E
1/2(RuIII/II), allowing for the intended diffusional redox mediation. For
[Ru(bpy)(deebq)(NCS)2], the oxidation was irreversible, precluding
spectral determination of A(λ) for RuIII.20 For all oxidative spectro-
electrochemical experiments, the absorbance spectrum of the solvated
redox mediator was subtracted from the resulting spectrum of the
oxidized sensitizer. Reductive chronoabsorptometry measurements were
performed by stepping the potential from 0 V to a negative potential
and monitoring UV-vis spectral changes. Desorption during the course
of the experiment was found to be negligible.
Results
Crystals of [Ru(bpy)2(deebq)](PF6)2 and [Os(bpy)2(deebq)]-
(PF6)2 suitable for X-ray diffraction studies were isolated, Figure
1. Both crystals were monoclinic and of the same space group,
P2(1)/c.
The unit cell lengths for the compounds were quite different,
with the largest disparity occurring along the axis a, Table 1.
In addition, there was a significant difference in the â angles,
96.5° for [Ru(bpy)2(deebq)](PF6)2 and 112.9° for [Os(bpy)2-
(deebq)](PF6)2, as well as the deebq dihedral angles, 16.5° and
25.2°, respectively.
Absorbance Measurements. UV-vis absorbance measurements
were made on a Hewlett-Packard 8453 diode array spectrophotometer.
Nanosecond transient absorption measurements were acquired with 532
nm laser excitation (ca. 8 ns fwhm from a Nd:YAG Continuum Surelite
II laser), 417 nm (H2 Raman shifter with 355 nm laser light), or 683
nm (H2 Raman shifter with 532 nm laser light). The sample was
protected from a pulsed 150 W Xe probe beam using a fast shutter
and appropriate UV and heat absorbing glass and solution filter
combinations. Each kinetic trace was acquired averaging 60-100 laser
Absorption spectra for [Ru(bpy)2(deebq)](PF6)2 and [Os-
(bpy)2(deebq)](PF6)2 in acetonitrile are provided in Figure 2.
(32) Bergeron, B.; Kelly, C. A.; Meyer, G. J. Langmuir 2003, 19, 8389.
9
8238 J. AM. CHEM. SOC. VOL. 128, NO. 25, 2006

Toward Exceeding the Shockley−Queisser Limit
A R T I C L E S
Figure 2. Normalized absorbance spectra of (a) [Ru(bpy)2(deebq)](PF6)2 and (b) [Os(bpy)2(deebq)](PF6)2 in neat acetonitrile (dashed line) and anchored to
TiO2 (solid line).
Table 1. Crystal Data and Structure Refinement for
[Ru(bpy)2(deebq)](PF6)2 and [Os(bpy)2(deebq)](PF6)2
All five compounds were found to bind to nanocrystalline
TiO2 films with a limiting surface coverage, Γo, of ∼8.0 × 10-8
mol/cm2. Small but reproducible spectral shifts were observed
upon binding to the nanocrystalline thin films. The absorption
spectra of the surface bound compounds most closely resembled
the carboxylate forms of the sensitizers measured in methanol,
Figure 2. We abbreviate the carboxylate forms of the ligands
as dcb or dcbq.
The [Ru(bpy)2(deebq)](PF6)2, [Ru(bq)2(deeb)](PF6)2, and [Ru-
(deebq)2(bpy)](PF6)2 compounds displayed room temperature
photoluminescence (PL) in acetonitrile solution, while no PL
was observed from [Os(bpy)2(deebq)](PF6)2 and [Ru(bpy)-
(deebq)(NCS)2] from 500 to 800 nm under the same conditions.
Excited-state decay followed a first-order kinetic model, and
the extracted lifetimes, τ, are listed in Table 2. Excited-state
decay of the sensitizers anchored to ZrO2 or TiO2 were
nonexponential and were well described by a parallel first- and
second-order kinetic model.31
The corrected PL spectra were used to estimate the Gibbs
free energy stored in the thermally equilibrated MLCT excited
state. An attempt was made to accomplish this with a single
mode Franck-Condon line shape analysis as has been previ-
ously described.36 However, the PL was observed at such low
energy that a meaningful fit could not be obtained even at 77
K. We therefore utilized the more traditional approach of a
tangent line drawn on the high-energy side of the corrected room
temperature spectra.38 The reduction potentials of the thermally
equilibrated luminescent excited state were then calculated from
eq 3 as previously described.39
[Ru(bpy)2(deebq)](PF6)2
[Os(bpy)2(deebq)](PF6)2
formula
fw
C47H42F12N6O5P2Ru
1161.88
C47H42F12N6O5P2Os
1251.01
crystal system
space group
a, Å
b, Å
c, Å
monoclinic
P2(1)/c
17.5217(14)
14.1894(11)
18.7241(15)
96.4570(10)
16.5
monoclinic
P2(1)/c
19.0259(11)
13.9453(8)
19.2470(12)
112.8890(10)
25.2
â, deg
biq dihedral angle, deg
V, Å3
4625.7(6)
4
4704.6(5)
4
Z
cryst color
crystal dimensions, mm3
D(calcd), g cm-3
µ(Mo KR), cm-1
temp, K
diffractometer
radiation
deep red
black
0.2 × 0.2 × 0.2
0.35 × 0.25 × 0.2
1.668
1.766
5.12
173(2)
Siemens P4/CCD
Mo KR
2.88
173(2)
Siemens P4/CCD
Mo KR
(λ ) 0.710 73 Å)
(λ ) 0.710 73 Å)
R(F), %
7.80
3.74
R(wF2),a %
20.17
11.69
2
2
a Quantity minimized ) R(wF2) ) ∑ [w (Fo2 - Fc )2]/∑ [(wFo )2]1/2; R
) ∑ ∆/∑ (Fo), ∆ ) |(Fo - Fc)|.
Two broad metal-to-ligand charge transfer (MLCT) absorption
bands were observed in the visible region, the higher energy
band is assigned as Ru f bpy charge transfer and the lower
energy band as Ru f deebq. For [Ru(bq)2(deeb)](PF6)2 and [Ru-
(deebq)2(bpy)](PF6)2, the higher energy M f deeb/bpy band
exists as a lower intensity shoulder to the M f bq/deebq charge-
transfer absorption band. For [Os(bpy)2(deebq)](PF6)2, an ad-
ditional band in the near-IR was assigned to a direct singlet-
to-triplet transition.33 The absorbance spectra of the carboxylic
acid and carboxylate forms of the compounds were measured
in methanol. The carboxylic acid derivatives displayed absorp-
tion features that were very similar to that of the ethyl ester
compounds in acetonitrile. Deprotonation of the carboxylic acid
groups lead to a significant blue shift of the M f bq′ charge-
transfer band and a smaller red shift of the M f bpy band. The
singlet-to-triplet absorption band of [Os(bpy)2(dcbq)] also blue
shifted significantly upon deprotonation of the carboxylic acid
groups. These spectral properties compare well with previously
reported results for similar heteroleptic compounds.34,35
E1/2(RuIII/II*) ) E1/2(RuIII/II) - ∆Ges
(3)
Attenuated total reflectance FTIR spectroscopy was used to
characterize the compounds in the solid state and when anchored
to TiO2. An intense band at 1713 cm-1 assigned to the
(34) Nazeeruddin, Md. K.; Zakeeruddin, S. M.; Humphry-Baker, M. J.; Liska,
P.; Vlachopoulos, V. S.; Fischer, C.-H.; Gra¨tzel, M. Inorg. Chem. 1999,
38, 6298.
(35) Qu, P.; Meyer, G. J. Langmuir 2001, 17, 6720.
(36) Casper, J. V.; Meyer, T. J. J. Am. Chem. Soc. 1983, 105, 5583.
(37) Kober, E. M.; Caspar, J. V.; Lumpkin, R. S.; Meyer, T. J. J. Phys. Chem.
1986, 90, 3722.
(38) Arnold, D. R.; Baird, N. C.; Bolton, J. R.; Brand, J. C. D.; Jacobs, P. W.
M.; DeMayo, P.; Ware, W. R. Photochemistry. An Introduction; Academic
Press: New York and London, 1974; p 13.
(33) Kober, E.; Meyer, T. J. Inorg. Chem. 1982, 21, 1324.
(39) Rehm, D.; Weller, A. Isr. J. Chem. 1970, 8, 259.
9
J. AM. CHEM. SOC. VOL. 128, NO. 25, 2006 8239

A R T I C L E S
Hoertz et al.
Table 2. Photophysical and Electrochemical Properties of Sensitizers and TiO2 Bound Sensitizers in Acetonitrilea
1
b
c
/
/
e
λ
Ru
abs, nm (
ꢀ
, M-1 cm-
)
λPL
,
τ
ns
,
E1 2(MIII II),d
E1 2(M2+/+),d
E1 2(MIII II*),d
∆
eV
Ges
,
φPL
/
/
/
3
sensitizer
f
bpy′ | Ru
f
bq
′
nm
V
V
V
×
10-
[Ru(bpy)2(deebq)](PF6)2
[Ru(bpy)2(dcbq)]/TiO2
[Os(bpy)2(deebq)](PF6)2
[Os(bpy)2(dcbq)]/TiO2
[Ru(bq)2(deeb)](PF6)2
[Ru(bq)2(dcb)]/TiO2
[Ru(bpy)(deebq)2](PF6)2
[Ru(bpy)(dcbq)2]/TiO2
[Ru(bpy)(deebq)(NCS)2]
[Ru(bpy)(dcbq)(NCS)2]/TiO2
427 (6.4 × 103), 555 (7.4 × 103)
835
800
f
90
430
11
1.45
1.39
1.05
0.98
1.52
1.44
1.55
1.50
0.86
0.72
-0.60
-0.70
-0.59
-0.70
-0.79
-0.90
-0.56
-0.65
-0.84
-0.95
-0.20
-0.38
f
1.64
1.77
f
1.9
444, 539
442 (7.7 × 103), 579 (9.5 × 103)
451, 565
f
20
488 (7.1 × 103), 530 (9.8 × 103)
728
775
780
800
f
210
400
590
780
10
-0.33
-0.33
-0.15
-0.20
f
1.85
1.77
1.70
1.70
f
5.3
491, 542
490 (6.0 × 103), 573 (1.3 × 104)
500, 573
16
450 (1.1 × 104), 626 (1.2 × 104)
f
432, 593
15
a All measurements were performed at room temperature. b Absorption maxima of the visible MLCT bands. c Corrected photoluminescence maxima, (5
nm. d Half-wave potentials measured by cyclic voltammetry in 0.1 M TBAPF6 acetonitrile electrolyte versus the standard calomel electrode, SCE. The
abbreviation MIII/II represents the metal-based reductions, M2+/+ represents the first ligand-based reductions, and MIII/II* is the excited-state reduction potential.
e The Gibbs free energy stored in the thermally equilibrated excited state. f A room-temperature emission maximum could not be observed with λ < 850 nm
that precludes analysis of excited-state reduction potentials, photoluminescent quantum yields, and the free energy stored in the excited state.
nm excitation, most of the enhancement was from bpy with
only a small contribution from biquinoline.
All the compounds except [Ru(bpy)(deebq)(NCS)2] displayed
quasi-reversible MIII/II waves in cyclic voltammetery measure-
ments in acetonitrile electrolyte with scan rates from 10 to 500
mV/s. The redox processes were classified as quasi-reversible
because the anodic and cathodic currents were approximately
equal but the peak-to-peak separation was greater than 80 mV.44
The MIII/II reduction potential for [Os(bpy)2(deebq)]2+ was 400
mV more negative than that of [Ru(bpy)2(deebq)]2+, Table 1.
As the number of bq ligands increased in the heteroleptic Ru-
(II) compounds, E1/2(RuIII/II) became increasingly more positive.
Quasi-reversible waves at ∼ -0.7 V vs SCE were assigned to
the first ligand-based reductions, E1/2(M2+/+). The reductions
were about 200 mV more positive when ethyl ester groups
replaced hydrogen atoms in the 4 and 4′ positions of the diimine
ligand. Comparative studies show that deebq is the most easily
reduced followed by bq, deeb, and bpy. The electrochemical
data are consistent with previous reports for related com-
pounds.26
Figure 3. Resonance Raman spectra of [Ru(bpy)2(deebq)](PF6)2 in aceto-
nitrile-d6 at the indicated excitation wavelengths. Letters A and B denote
deebq and bpy vibrational modes, respectively.
asymmetric CdO stretch was observed for the ester derivatives,
Cyclic voltammetry performed on sensitized TiO2 films also
showed quasi-reversible waves in acetonitrile electrolyte. The
while the carboxylate forms showed two bands at 1597 and 1387
cm-1 40,41
Interestingly, when either the ester or the carboxylate
.
E
1/2(MIII/II) potentials were slightly more negative (∼80 mV)
forms were anchored to TiO2, their FTIR spectra were virtually
superimposable showing intense bands at approximately the
same energies as the carboxylate compound. The only exception
to this was [Ru(bpy)(deebq)2](PF6)2, which showed some
absorbance at ∼1713 cm-1 characteristic of unhydrolyzed ethyl
esters after surface attachment. Consistent with previous results,
the FTIR data indicated that the carboxylate form of the
compounds were present on the ZrO2 or TiO2 surfaces.28
Shown in Figure 3 is resonance Raman spectra of [Ru(bpy)2-
(deebq)](PF6)2 in acetonitrile collected at different excitation
wavelengths. The bipyrdine and biquinoline vibrational modes
enhanced were readily assigned based on previous publica-
tions.42,43 The biquinoline modes were selectively enhanced with
long wavelength excitation (530.9 or 568.2 nm) of [Ru-
(bpy)2(deebq)]2+. As the excitation energy increased, contribu-
tions from the bpy′ modes became more apparent. With 415.4
when compared to data obtained in the electrolyte solution. The
E
1/2(M2+/+) ligand-based reductions were consistently 100 mV
more negative on the surface than that measured in the
electrolyte solution. These data are summarized in Table 2.
The RuIII/II wave for [Ru(bpy)(deebq)(NCS)2] was quasi-
reversible at faster scan rates but became irreversible at slower
scan rates. The oxidative process was 82% reversible at 50 mV/s
and completely reversible at 500 mV/s, consistent with an ∼1
s lifetime of the oxidized compound. The oxidized N3 sensitizer,
[RuIII(dcb)2(NCS)2]+, also has a lifetime of 0.1-1 s in CH3-
CN.45 The estimated E1/2(RuIII/II) for [Ru(bpy)(deebq)(NCS)2]
is similar to that reported for N3, Table 1.
Spectroelectrochemistry was used to obtain the absorption
spectra of the reduced and oxidized forms of the compounds
anchored to the TiO2 surface. Application of a potential more
negative than that of the first ligand-based reduction for samples
(40) Finnie, K. S.; Bartlett, J. R.; Woolfrey, J. L. Langmuir 1988, 14, 2744.
(41) Dobson, K.; McQuillan, A. Spectrochim. Acta, Part A 2000, 5, 557.
(42) Tait, C. D.; MacQeen, D. B.; Donahue, R. J.; DeArmond, M. K.; Hanck,
K. W.; Wertz, D. W. J. Phys. Chem. 1986, 90, 1766.
(43) Chowdhury, J.; Ghosh, M.; Misra, T. N. Spectrochim. Acta, Part A 2000,
56, 2107.
(44) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and
Applications, 2nd ed.; John Wiley & Sons: New York, 2001.
(45) (a) Moser, J. E.; Noukakis, U. B.; Tachibana, Y.; Klug, D. R.; Durrant, J.
R.; Humphrey-Baker, R.; Gra¨tzel, M. J. Phys. Chem. B 1998, 102, 3649.
(b) Das, S.; Kamat, P. V. J. Phys. Chem. B 1998, 102, 8954.
9
8240 J. AM. CHEM. SOC. VOL. 128, NO. 25, 2006

Toward Exceeding the Shockley−Queisser Limit
A R T I C L E S
the MLCT excited state. The difference spectra showed positive
∆A features below ∼425 nm and beyond ∼600 nm, as well as
a bleach of the ground-state absorption from ∼500-600 nm.
Excited-state decay was exponential in fluid solution and
required a parallel first- and second-order kinetic model for
excited states anchored to ZrO2.31 In all cases, the kinetic rate
constants abstracted agreed well with those obtained by time-
resolved photoluminescence measurements. UV-visible absorp-
tion spectra recorded before and after transient absorption studies
of the mono-biquinoline compounds showed no evidence for
photochemistry. For [Ru(bq)2(deeb)](PF6)2 and [Ru(bpy)(deebq)2]-
(PF6)2 photochemical products were observed after photolysis
even at low irradiances. In agreement with the results of von
Zelewsky and Gremaud, the biquinoline ligand was found to
be photolabile.26 For example, laser excitation of [Ru(deebq)2-
(bpy)]2+ in acetonitrile yielded cis-[Ru(bpy)(deebq)(CH3CN)2]2+
1
that was isolated and characterized by H NMR. This photo-
Figure 4. Absorption changes observed after electrochemical reduction
and oxidation of [Ru(bpy)2(dcbq)]/TiO2 in acetonitrile electrolyte. The
positive absorption changes correspond to [Ru(bpy)2(dcbq)]/TiO2 reduction
monitored at 470 nm (upper black traces) and induced by cycling the
potential between 0.0 and -0.75 V vs SCE. The oxidation of [Ru(bpy)2-
(dcbq)]/TiO2 (lower red trace) was observed at 550 nm after stepping the
potential from 1.10 V to +1.60 V vs SCE.
product was successfully used as a precursor in the synthesis
of [Ru(bpy)(deebq)(NCS)2].
Nanosecond transient absorption spectra of the sensitizers
anchored to TiO2 showed the presence of the MLCT excited
state and an additional product that had a lifetime several orders
of magnitude greater. No corresponding long-lived state was
observed by time-resolved PL measurements. The absorption
difference spectrum of the long-lived component was simulated
by standard addition of the visible absorption spectrum of the
reduced and oxidized forms of the compound minus two ground-
state spectra, ∆A ) A(MIII) + A(M+) - 2A(MII). The simula-
tions were in excellent agreement with the observed difference
spectra for the long-lived component, Figure 5b inset right-hand
side. This charge separated state was easily distinguished from
the more common interfacial charge separated state with an
oxidized sensitizer and a conduction band electron, Figure 5b
inset left-hand side. The positive absorptions observed at ∼500
and 700 nm are characteristic of the reduced sensitizers.
The long-lived component returned cleanly to ground-state
products on a millisecond time scale. Transients recorded at
probe wavelengths that provided the best signal-to-noise were
well described by the Kohlrausch-Williams-Watts (KWW)
model, Figure 6 and eq 4,
results in visible absorbance changes. The spectral changes were
consistent with formation of the reduced compound on TiO2.
The same spectra could be observed upon steady-state photolysis
(514.5 nm) of the sensitized TiO2 films in 0.5 M triethylamine/
CH3CN. The application of a potential more positive than the
MIII/II potential resulted in the appearance of the oxidized
compound on TiO2. The time scale for the oxidation was much
longer than that for reduction, and the solvated version of the
sensitizer was often used to mediate oxidation.
Chronoabsorption measurements were made to obtain the
apparent diffusion coefficients, Dapp, for intermolecular hole and
electron hopping at saturation surface coverages across the
nanocrystalline TiO2 surfaces.46-48 For all the sensitizers, it took
at least 10 times as long to oxidize the surface bound sensitizers
as it did to reduce them. Figure 4 shows typical data that
demonstrates that the sensitizers could be reversibly reduced
and reoxidized in about a minute, while oxidation was incom-
plete after 20 min. These redox processes occurred with retention
of isosbestic points. Occasionally, some sensitizer desorption
was observed particularly during the sluggish oxidation.
About 60% of Anson plots, ∆A versus t1/2, were found to be
linear, and the apparent diffusion constant, Dapp, was abstracted
from these data as previously described.48 For reduction of Ru-
(bpy)2(dcbq)/TiO2 in acetonitrile electrolyte, a Dapp(Ru2+/+) )
(3.3 ( 0.3) × 10-12 m2/s was obtained from absorption changes
observed after stepping the applied potential from 0.00 V to
-0.75 V vs SCE, while Dapp(RuIII/II) was obtained similarly by
stepping the applied potential from 1.10 V to +1.60 V,
I(t) ) R exp - (kt)â 0 < â < 1
(4)
where k is the maximum rate constant in a distribution of rate
constants R is the initial amplitude, and â is inversely related
to the width of this underlying distribution.27c Kinetic fits to a
large number of samples gave k ) (8 ( 5) × 105 s-1 and â )
0.25 ( 0.05.
Comparative actinometry was used to determine quantum
yields for formation of the charge-separated states. The yields
were found to be independent of the excitation irradiance and
only weakly dependent on the sensitizer surface coverage.49
notable wavelength dependence was observed, Table 3. Sig-
nificant charge separation was also observed for Ru(bpy)(dcbq)-
(NCS)2/TiO2; however the extinction coefficients were un-
known, due to the reactivity of the oxidized sensitizer, so that
absolute yields could not be quantified. Also given in Table 2
are the reduction potentials of the initially formed Franck-
Condon excited state, eq 5,
D
app(RuIII/II) ) (2 ( 1) × 10-13 m2/s.
A
Nanosecond transient absorption spectra of the sensitizers
were obtained in acetonitrile and when anchored to TiO2 and
ZrO2 thin films, Figure 5. The absorption difference spectra of
the sensitizers measured in acetonitrile and anchored to ZrO2
were within the same experimental error and were assigned to
(46) Bonhote, P.; Gogniat, E.; Tingry, S.; Barbe, C.; Vlachopoulos, N.;
Lenzmann, F.; Comte, P.; Gra¨tzel, M. J. Phys. Chem. B 1998, 102, 1498.
(47) Trammel, S. A.; Meyer, T. J. J. Phys. Chem. B 1999, 103, 104.
(48) Anson, F. C.; Blauch, D. N.; Saveant, J.-M.; Shu, C.-F. J. Am. Chem. Soc.
1991, 113, 1922.
E°(MIII/II**) ) E1/2(MIII/II) - Eop
(5)
9
J. AM. CHEM. SOC. VOL. 128, NO. 25, 2006 8241

A R T I C L E S
Hoertz et al.
Figure 5. Transient absorbance spectra of (a) [Os(bpy)2(dcbq)]/ZrO2 and (b) [Os(bpy)2(dcbq)]/TiO2 following 417 nm light excitation (3.5 mJ/cm2, 8 ns
fwhm). The delay times for the ZrO2 sample were as follows: 5 ns (black line, 9), 10 ns (red line, (), 15 ns (green line, 2), 30 ns (cyan line, 1), and 100
ns (blue line, b). The delay times for the TiO2 sample were as follows: 10 ns (black line, 9), 25 ns (red line, (), 100 ns (green line, 2), 500 ns (cyan line,
1), and 2 µs (blue line, b). The inset shows a simulation over the same spectral range of the simulated spectra (dashed) overlaid with the experimental data
obtained 5 µs after pulsed excitation of [Os(bpy)2(dcbq)]/TiO2 in 0.1 M LiClO4 acetonitrile (left-hand side) and in neat acetonitrile (right-hand side).
were observed, Figure 5b inset left-hand side.49 For [Ru(bpy)2-
(dcbq)]/TiO2 quenching of the steady-state photoluminescence
was observed at >20 mM Li+ concentrations.
Discussion
Ultrafast electron injection at sensitized TiO2 interfaces, while
of debatable importance in regenerative solar cells, has been
exploited here to drive photoredox reactions that are thermo-
dynamically uphill in fluid solution. Indeed, control experiments
in fluid acetonitrile gave no evidence for excited-state electron
transfer. The ability to drive such “uphill” redox reactions and
utilize energy that would otherwise be irreversibly lost represents
an important step toward the realization of solar conversion
efficiencies that exceed the Shockley-Queisser limit.1,2
The proposed mechanism for photoinduced TiO2 mediated
charge separation is shown schematically (Scheme 2). Ultrafast
Figure 6. Absorption change monitored at 570 nm following 417 nm laser
electron injection into TiO2 is followed by rapid reduction (<10
light excitation (∼3.5 mJ/cm2, 8 ns fwhm) of [Os(bpy)2(dcbq)]/TiO2 (Γ )
6.5 × 10-8 mol/cm2). Overlaid on the data is a best fit (white line) to the
ns) of a surface bound sensitizer. Lateral intermolecular charge
Kohlrausch-Williams-Watts model. Residuals for the fit are also shown.
Table 3. Excitation Wavelength Dependent Quantum Yieldsa
transfer across the semiconductor surface ultimately yields
ground state products. Thus absorption of a single photon
yielded well-defined redox equivalents that typically stored 15-
25% more free energy than did the photoluminescent excited
state of the sensitizer. For example, the photoinduced charge
separation observed after light excitation of [Ru(bpy)2(dcbq)]/
TiO2 can be designated as shown in eq 6.
417 nm
532 nm
683 nm
/
/
/
sensitizer
(E
°
(MIII II**))b
(E
°
(MIII II**))b
(E
°
(MIII II**))b
[Ru(bpy)2(dcbq)]/TiO2
0.17 ( 0.02
(-1.59)
0.05 ( 0.02
(-0.95)
0.08 ( 0.02
(-1.36)
e0.02
(-0.90)
e0.02
(-0.84)
c
[Os(bpy)2(dcbq)]/TiO2
[Ru(bq)2(dcb)]/TiO2
[Ru(bpy)(dcbq)2]/TiO2
0.18 ( 0.02
(-2.00)
0.05 ( 0.01
(-0.84)
c
Ru(bpy)2(dcbq)**/TiO2 + RuII(bpy)2(dcbq)/TiO2 f
RuIII(bpy)2(dcbq)+/TiO2 + RuII(bpy)2(dcbq-)-/TiO2 (6)
0.06 ( 0.02
(-1.54)
0.04 ( 0.02
(-1.48)
c
The redox equivalents formed store 2.09 eV of free energy,
while the luminescent excited state stores only about 1.77 eV.
In addition, charge separation lasted for periods of milliseconds,
while the excited states decayed on a nanosecond time scale.
Below we discuss mechanistic details of the charge separation
processes, electron injection and sensitizer reduction as well as
lateral intermolecular charge transfer processes that ultimately
lead to recombination.
Ultrafast Electron Injection. Subpicosecond electron injec-
tion is now well documented at sensitized TiO2 interfaces.8-15
The transient data are highly nonexponential and show an
excitation wavelength dependence. The origin of the complex
a All measurements performed in acetonitrile at room temperature. b The
reduction potential of the initially formed Franck-Condon excited state,
calculated by eq 5. c Sensitizer did not appreciably absorb light at this
wavelength.
where Eop is the excitation energy.40 The addition of less than
∼1 mM concentrations of LiClO4 to the external acetonitrile
was found to increase the charge separation yields by about
10-20% with 417 nm light excitation. When higher Li+
concentrations were employed, difference spectra consistent with
an oxidized sensitizer and an injected electron, MIII/TiO2(e-),
(49) Hoertz, P. G. Thesis Johns Hopkins University, 2003.
9
8242 J. AM. CHEM. SOC. VOL. 128, NO. 25, 2006

Toward Exceeding the Shockley−Queisser Limit
A R T I C L E S
Scheme 2
kinetics remains unresolved but may reflect interfacial hetero-
geneity. Sundstrum has proposed that the singlet state injects
electrons in subpicoseconds, while the triplet injects on a slower
picosecond time scale.13 Triplet injection was thought to be
particularly important when light excitation promoted an electron
to a ligand remote or weakly coupled to the semiconductor
surface. Moser and Gra¨tzel have recently shown that the slower
picosecond components were essentially absent from sensitized
TiO2 films equilibrated in ionic solutions, behavior attributed
to removal of the weakly bound sensitizers.50
Under the experimental conditions employed here, the pho-
toluminescent excited state did not inject electrons into TiO2.
Nanosecond time resolution precludes the direct determination
of the injection dynamics but were consistent with kinj > 108
s-1. The excitation wavelength dependent quantum yields were,
however, characteristic of injection from vibrationally hot
excited states and are not easily explained by other mecha-
nisms.19,51 To a first approximation, these data can be rational-
ized based on the reduction potential of the initially formed
Franck-Condon excited state, Table 3. The stronger the
photoreductant, the higher the injection yield. Injection is thus
competitive with vibrational relaxation and/or intersystem
crossing of the photoexcited sensitizer. For Ru and Os poly-
pyridyl compounds, this is known to be a subpicosecond
process.55
reflect a larger overlap integral in eq 7. Electrochemical
measurements indicate that the density of TiO2 acceptor states
increases exponentially in energy.53 The number of vibrational
and electronic states accessible to the photoexcited sensitizer
also increases with excitation energy.55 Therefore, a greater
overlap integral is expected for sensitizers excited with blue
light relative to green or red light. This conclusion is consistent
with recent ultrafast injection measurements8-15 and with early
pioneering photoluminescence studies54 both of which have
provided compelling evidence for faster injection rates as the
energy separation between the excited donor and the conduction
band edge increases.
A close inspection of the quantum yield data in Table 3 shows
some subtle effects that cannot easily be rationalized solely on
reduction potentials. Fortunately, variable light excitation studies
into the energetically well-resolved charge transfer absorption
bands of these heteroleptic sensitizers provided new insights.
While some excited state mixing undoubtedly occurs, the
resonance Raman, spectroscopic, and electrochemical data
indicate that the lower energy absorption band (λmax ≈ 550 nm)
is predominately M f bq′ charge transfer, while the higher
energy band (λmax ≈ 450 nm) is M f bpy in nature. Therefore,
blue light excitation of [M(bpy)2(dcbq)]/TiO2 promotes an
electron to a remote and unbound bpy ligand, [Ru(bpy-)(bpy)-
(dcbq)]*/TiO2 that injects electrons more efficiently than does
Mfbq′ excitation, [M(bpy)2(dcbq-)]*/TiO2. This must reflect
the stronger reducing power of the bpy localized excited state
and a larger overlap integral in eq 7. This conclusion challenges
the notion that strong electronic coupling through the carboxylate
containing ligand and the semiconductor is required for ultrafast
electron injection. The Os sensitizer yields were lower than one
would expect based on comparisons with the Ru excited-state
reduction potentials. As pointed out by McKusker, the heavier
Os metal center is expected to increase intersystem crossing
dynamics, and this lowers the ultrafast injection yield.14 We
note that light excitation into the singlet-to-triplet absorption
band results in a significant injection. Therefore, to the degree
that spin effects can be realized with these heavy transition metal
Gerischer has stated that the electron transfer rate constants
are directly proportional to the overlap of the donor levels of
the excited sensitizer Wdon(E) with the density of unoccupied
acceptor levels of the semiconductor D(E),
kinj
≈
κ(E) D(E) Wdon(E) dE
(7)
∫
where κ(E) is the transfer frequency.52 There is good reason to
believe that the increased injection yields with excitation energy
(50) Wenger, B.; Gratzel, M.; Moser, J.-E. J. Am. Chem. Soc. 2005, 127, 12150.
(51) Liu, F.; Meyer, G. J. J. Am. Chem. Soc. 2005, 127, 824.
(52) (a) Gerischer, H. Photochem. Photobiol. 1972, 16, 243. (b) Gerischer, H.;
Willig, F. Top. Curr. Chem. 1976, 61, 31. (c) Gerischer, H. Pure Appl.
Chem. 1980, 52, 2649.
(53) Kay, A.; Humphrey-Baker, R.; Gra¨tzel, M. J. Phys. Chem. 1994, 98, 952.
(54) (a) Kajiwara, T.; Hasimoto, K.; Kawai, T.; Sakata, T. J. Phys. Chem. 1982,
86, 4516. (b) Kajiwara, T.; Hasimoto, K.; Kawai, T.; Sakata, T. J. Phys.
Chem. 1988, 92, 4636. (c) Sakata, T.; Hasimoto, K.; Hiramoto, M. J. Phys.
Chem. 1990, 94, 1788.
(55) (a) Damrauer, N. H.; Cerullo, G.; Yeh, A.; Boussie, T. R.; Shank, C. V.;
McCusker, J. K. Science 1997, 275, 54. (b) Bhasikuttan, A. C.; Suzuki,
M.; Nakashima, S.; Okada, T. J. Am. Chem. Soc. 2002, 124, 8398.
9
J. AM. CHEM. SOC. VOL. 128, NO. 25, 2006 8243

A R T I C L E S
Hoertz et al.
compounds, this observation suggests that vibrationally “hot”
triplet states can also undergo ultrafast injection processes.33
reported or investigated. It is for this reason that similar behavior
has not been previously observed, Scheme 1.
Since electron transfer from TiO2 to form the reduced
sensitizers is energetically favored, eq 8, back electron transfer
to the oxidized sensitizer must also be. This could form the
excited state or the ground state, eqs 9 and 10.
The quantum yields were found to be insensitive to the
equilibration time of the sensitized film50 but could be tuned
by the addition of Li+ cations to the external solution. Lithium
cations are known to be “potential determining ions” that adsorb
to TiO2 and shift the conduction band edge positive on an
electrochemical scale.56-58 Lithium concentrations less than
about 1 mM were found to increase the injection yields by 10-
20%. At higher Li+ concentrations, only the oxidized sensitizer
and the TiO2(e-) were observed spectroscopically, presumably
because the conduction band edge was below the sensitizer
reduction potential (see below for further discussion).
e-/TiO2 + RuII(bpy)2(dcbq)/TiO2 f
RuII(bpy)2(dcbq-)-/TiO2 (8)
e-/TiO2 + RuIII(bpy)2(dcbq)+/TiO2 f
RuIII(bpy)2(dcbq-)*/TiO2 (9)
Sensitizer Reduction. Photoinitiated electron transfer from
TiO2 to molecular acceptors has been previously studied.18 This
process represents an unwanted loss mechanism in dye sensitized
solar cells, for example. It is known that the electron-transfer
rate constants are sensitive to temperature and to thermodynamic
driving force in a manner consistent with Marcus theory.59,60 It
is also known that transport of the injected electron in TiO2
can influence the observed dynamics.61,62
e-/TiO2 + RuIII(bpy)2(dcbq)+/TiO2 f
RuII(bpy)2(dcbq)/TiO2 (10)
The +3 formal oxidation state favors dcbq reduction by about
400 meV over the M(II) ground state, Table 2. For related Ru
and Os compounds, the difference in reduction potentials
between the excited and ground state is typically 300-500
meV.64 No evidence for back electron transfer to generate an
excited state was found in transient studies. This is attributed
to the number of ground-state sensitizers relative to the number
of oxidized sensitizers present at the interface. We estimate that
there are about 500 sensitizers anchored to each TiO2 nanopar-
ticle. On average, pulsed light excitation promotes a few of these
to excited states, only a fraction of which inject an electron
into TiO2. The injected electron has mobility and is statistically
more likely to encounter a ground-state sensitizer than the
oxidized sensitizer from which it originated.
In the present study, the molecular acceptors were ground
state sensitizers that were photoreduced in less than 10 ns.
Electrochemical studies demonstrate that the sensitizers were
more easily reduced than the TiO2 nanoparticles. Furthermore,
when the sensitizers were excited with visible light in the
presence of a sacrificial electron donor, the reduced forms of
the sensitizers were observed. Taken together, these results
demonstrate that the first reduction potential of the sensitizers
lies below the conduction band edge Ecb < E1/2(M2+/+).
Lateral Charge Transfer. Isoenergetic intermolecular charge
transfer provides a mechanism by which charge can “hop” across
the semiconductor surface. This allows encounters between the
reduced and oxidized sensitizers to yield ground-state products.
Previous researchers have shown that lateral charge transfer
reactions can occur efficiently and rapidly with an applied
bias.27c,46,47 Control experiments with ZrO2 substrates, for
example, demonstrate that the semiconducting properties of TiO2
are not responsible for the observed redox chemistry. Bonhote
and co-workers have also shown that a percolation threshold
exists for the oxidation of amines anchored to nanocrystalline
TiO2 thin films.46
In the present photo-initiated studies, we have modeled lateral
charge transfer with the Kohlrausch-Williams-Watts function,
which is a paradigm for transport in disordered media.65,66 An
advantage of this model is that the normalized data could be
quantified with only two variables; â, which is inversely related
to the width of an underlying Levy distribution of rate constants,
and k, which is the rate constant at the maximum amplitude of
the distribution. The average rate constants for a large number
of samples were found to be (8 ( 5) × 105 s-1. The â values
varied between 0.2 and 0.3 which corresponds to a highly
skewed distribution of rates with a significant amplitude over
6 orders of magnitude.66 A broad distribution would be expected
if the donors and acceptors were located at variable distances
These energetics enabled us to measure the ligand-based
reduction potentials of the sensitizers anchored to TiO2 for the
first time. The sensitizer/TiO2 reduction potentials were about
100 mV negative of that measured in fluid electrolyte. Infrared
studies demonstrated that the carboxylate forms of the sensitizers
were present on the TiO2 surface. Transformation of the
electron-withdrawing ester groups to carboxylates is expected
to shift the first reduction potentials of M(bpy)2(deebq)2+
,
Ru(deebq)2(bpy)2+, and Ru(deebq)(bpy)(NCS)2 negative. How-
ever, the magnitude of the shift (∼100 mV) was not as large as
one would anticipate based on solution studies, consistent with
TiO2 stabilization of the carboxylates.63 The ∼90 mV negative
shift observed after binding Ru(bq)2(deeb)2+ to TiO2 is harder
to rationalize as the first reduction is biquinoline based. To our
knowledge, sensitizers that have the first ground- and excited-
state reduction potentials below Ecb have never before been
(56) Wang, P.; Wenger, B.; Humphry-Baker, R.; Moser, J. E.; Teuscher, J.;
Kantlehner, W.; Mezger, J.; Stoyanov, E. V.; Zakeeruddin, S. M.; Gratzel,
M. J. Am. Chem. Soc. 2005, 127, 6850.
(57) (a) Redmond, G.; Fitzmaurice, D. J. Phys. Chem. 1993, 97, 1426. (b)
Enright, B.; Redmond, G.; Fitzmaurice, D. J. Phys. Chem. 1994, 98, 6195.
(c) Redmond, G.; Gra¨tzel, M.; Fitzmaurice, D. J. Phys. Chem. 1993, 97,
6951. (d) Boschloo, G.; Fitzmaurice, D. J. Phys. Chem. B 1999, 103, 2228.
(e) Boschloo, G.; Fitzmaurice, D. J. Phys. Chem. B 1999, 103, 7860.
(58) Watson, D. F.; Meyer, G. J. Coord. Chem. ReV. 2004, 248, 1391.
(59) Clifford, J. N.; Opalomeras, E.; Nazeerudding, Md. K.; Graztel, M.; Nelson,
J.; Long, N. J.; Durrant, J. R. J. Am. Chem. Soc. 2004, 126, 5225.
(60) Kuciauskas, D.; Freund, M. S.; Gray, H. B.; Winkler, J. R.; Lewis, N. S.
J. Phys. Chem. B 2001, 105, 392.
(61) Nelson, J. Phys. ReV. B 1999, 59, 15374.
(64) Kalyanasundaram, K. Photochemistry of Polypyridine and Porphyrin
Complexes; Academic Press: London, 1992.
(62) Hasselmann, G. M.; Meyer, G. J. J. Phys. Chem. B 1999, 103, 7671.
(63) (a) Wolfbauer, G.; Bond, A. M.; Deacon, G. B.; MacFarlane, D. R.; Spiccia,
L. J. Am. Chem. Soc. 2000, 122, 130. (b) Nazeeruddin, Md. K.;
Zakeeruddin, S. M.; Humphry-Baker, M. J.; Liska, P.; Vlachopoulos, V.
S.; Fischer, C.-H.; Gra¨tzel, M. Inorg. Chem. 1999, 38, 6298.
(65) (a) Kohlrausch, R. Ann. 1847, 5, 430. (b) Williams, G.; Watts, D. C. Trans.
Faraday Soc. 1971, 66, 80.
(66) (a) Linsey, C. P.; Patterson, G. D. J. Chem. Phys. 1980, 73, 3348. (b)
Majumdar, C. K. Solid State Commun. 1971, 9, 1087.
9
8244 J. AM. CHEM. SOC. VOL. 128, NO. 25, 2006

Toward Exceeding the Shockley−Queisser Limit
A R T I C L E S
from each other. The long approximately millisecond lifetimes
could correspond to redox equivalents generated on different
TiO2 nanocrystals for example.
above the first reduction potential. However, when LiClO4 was
added to the external acetonitrile, the ligand based reduction
potential and Ecb became more energetically proximate. At high
[Li+], quenching of the MLCT excited state was apparent and
photoinduced sensitizer reduction was not observed. This
indicates that the sensitizer π* orbitals and Ecb are energetically
proximate and can be tuned relative to one another. Thus a TiO2
mediated intermolecular electron transfer mechanism, either
through a Boltzmann population of conduction band states or a
superexchange type mechanism, is a reasonable explanation for
the rapid reduction of the sensitizers anchored to TiO2.75 The
MIII/II potentials, on the other hand, are almost 2 eV positive of
the conduction band, and a purely localized “hopping” mech-
anism must be operative for intermolecular “hole” transfer.
A disadvantage of the KWW model is that it does not provide
molecular insights into the recombination mechanism(s). An
issue that naturally arises concerns which charges move faster,
the electrons or the “holes”. Hole transfer involves the metal
t2g orbitals while the biquinoline π* orbitals mediate lateral
2+
electron transfer. For Ru(bpy)3 in fluid acetonitrile solution,
the RuIII/II self-exchange rate constants are 2 × 109 M-1 s-1
and the Ru2+/+ constants are 8 × 108 M-1 s-1 67,68
.
The high
self-exchange rate constants have previously been discussed and
reflect low intrinsic barriers to electron transfer.69 However, the
rate constants may be very different at the interface where the
reorganization terms are expected to differ. To investigate this
issue, the reduction and oxidation of the surface bound sensitiz-
ers were quantified by chronoabsorption measurements.46-48 The
potential of the sensitized TiO2 films were stepped positive of
the RuIII/II potential (or negative of the Ru2+/+ potential), and
the oxidation (reduction) of the surface bound sensitizers was
monitored spectroscopically. With this approach, reduction was
consistently found to occur on a seconds time scale, while
oxidation required minutes. Diffusion constants abstracted from
these data were consistently an order of magnitude larger for
reduction. The energetic proximity of the first reduction potential
with the conduction band edge suggests that TiO2 may mediate
the intermolecular electron transfer.
Conclusions
Photoinduced electron transfer reactions that store energy in
excess of thermally equilibrated excited states have been realized
for the first time. The approach described is general. Indeed
one can envision exploiting this behavior in a variety of
arrangements that employ semiconductor nanoparticles as (1)
acceptors of electrons from vibrationally hot excited states and
(2) conduits for electron transport to remote acceptors. In
favorable cases, about 90% of the energy of an absorbed green
photon and 93% of the energy of an absorbed red photon were
converted to long-lived charge separated states. A disappointing
aspect of the current work was the low yields that could not be
optimized above 50%. This was clearly due to inefficient
electron injection. While the factors that control ultrafast charge
separation in these heteroleptic compounds remain unknown,
it is encouraging to note that researchers have found conditions
where ultrafast injection from related sensitizers are quantita-
tive.9,50 A significant challenge for future research is to couple
these ultrafast charge separation processes with catalysts that
can utilize the excess energy to drive fuel forming and
environmentally friendly reactions. Studies of this type are
underway in our laboratories.76
The possible role TiO2 conduction band states may play in
the electrochemical reduction of hemes and photochromic dyes
has previously been suggested.70-72 Sugihara and co-workers
have also reported evidence that diimine ligands with low-lying
π* orbitals can mediate back electron transfer from TiO2 to tri-
iodide in regenerative solar cells.73 Similar behavior has been
reported more recently by Bignozzi.74 Our ability to observe
an injected electron fully localized and “trapped” on a ligand
supports this general idea. At issue is the energetic position of
the conduction band relative to the sensitizer π* orbitals. In
neat acetonitrile, the conduction band states clearly reside >kT
Acknowledgment. The Division of Chemical Sciences, Office
of Basic Energy Sciences, Office of Energy Research, U.S.
Department of Energy are gratefully acknowledged for research
support. The authors thank Amy A. Narducci Sarjeant for help
with the crystallographic data.
(67) (a) Chou, M.; Creutz, C.; Sutin, N. J. Am. Chem. Soc. 1977, 99, 5615. (b)
Young, R.; Keene, R. J. Am. Chem. Soc. 1977, 99, 2468.
(68) (a) Motten, A. G.; Hanck, K.; DeArmond, M. K. Chem. Phys. Lett. 1981,
79, 541. (b) Heath, G. A.; Yellowless, L. J.; Braterman, P. S. Chem. Phys.
Lett. 1982, 92, 646.
(69) Sutin, N.; Creutz, C. AdV. Chem. Ser. 1978, 168, 1.
(70) Wang, Q.; Zakeeruddin, X. M.; Cremer, J.; Bauerle, P.; Humphry-Baker,
R.; Gra¨tzel, M. J. Am. Chem. Soc. 2005, 127, 5706.
(71) Biancardo, M.; Argazzi, R.; Bignozzi, C. A. Inorg. Chem. 2005, 44, 9619.
(72) Obare, S. O.; Ito, T.; Meyer, G. J. EnViron. Sci. Technol. 2005, 39, 6266.
(73) Yanagida, M.; Yamaguchi, T.; Kurashige, M.; Hara, K.; Katoh, R.; Sugihara,
H.; Arakawa, H. Inorg. Chem. 2003, 42, 7921.
(74) Altobello, S.; Argazzi, R.; Caramori, S.; Contado, C.; Da Fre, S.; Rubino,
P.; Chone, C.; Larramona, G.; Bignozzi, C. A. J. Am. Chem. Soc. 2005,
127, 15342.
Supporting Information Available: Crystallographic data as
CIF files. This material is available free of charge via the Internet
at http://pubs.acs.org.
JA060470E
(75) Davis, W. B.; Svec, W. A.; Ratner, M. A.; Wasielewski, M. R. Nature
1998, 396, 60.
(76) Obare, S. O.; Ito, T.; Meyer, G. J. J. Am. Chem. Soc. 2006, 128, 712-713.
9
J. AM. CHEM. SOC. VOL. 128, NO. 25, 2006 8245