Cobaloxime-based photocatalytic devices for hydrogen production

Article abstract of DOI:10.1002/anie.200702953

(Figure Presented) It is the dawning of the age of supramolecular photocatalysts for H₂ production using first-row transition-metal catalytic centers. These catalysts consist of a ruthenium tris(diimine) light-harvesting unit covalently linked to various catalytic cobaloxime centers (see picture). The catalyst stability, Co^II/Co^I redox potential, and nucleophilicity of the cobaloxime moiety all affect the photocatalytic properties.

Full text of DOI:10.1002/anie.200702953

Communications
DOI: 10.1002/anie.200702953
Photocatalysis
Cobaloxime-Based Photocatalytic Devices for Hydrogen Production**
Aziz Fihri, Vincent Artero,* Mathieu Razavet, Carole Baffert, Winfried Leibl, and
Marc Fontecave
Dedicated to Professor P. Gouzerh on the occasion of his 65th birthday
Homogeneous light-driven catalytic systems for hydrogen
production and, more generally, efficient photoactivated
[
1]
synthetic multielectron catalysts remain relatively scarce.
[
2–4]
Such systems
generally consist of 1) a photosensitizer,
[5]
often based on the ruthenium tris(diimine) moiety, 2) a
metal-based catalytic center, and in some cases 3) an addi-
tional redox mediator. However, their efficiency remains to
be improved in terms of both turnover numbers (stability)
and turnover frequencies, and these systems should prefera-
bly rely on inexpensive first-row transition-metal catalysts
rather than unsustainable noble metals. We and others
recently reported that cobaloximes are very efficient and
[
6–9]
cheap electrocatalysts for hydrogen evolution.
We thus
decided to couple cobaloximes with ruthenium tris(diimine)
moieties in order to make a supramolecular variant of the
system previously studied by Lehn et al. for photochemical
[
3]
production of hydrogen. In such a molecular device, the
intramolecular electron transfer from the photoactivated
center to the catalytic center can potentially be controlled,
and the charge-recombination processes limited, to an extent
larger than in intermolecular systems, by fine-tuning both the
distance between metal centers and the nature of the
[
2,10]
bridge.
gen-evolving green algae, where the photosystem I is tightly
Such an organized assembly is found in hydro-
Scheme 1. Structures of 1–4.
[11]
coupled to hydrogenase enzymes. In this paper we describe
the synthesis and activity of a series of novel heterodinuclear
ruthenium–cobaloxime photocatalysts able to achieve the
photochemical production of hydrogen with the highest
turnover numbers so far reported for such devices.
Compounds 1–3 (Scheme 1) were synthesized in good
[12]
[*] Dr. A. Fihri, Dr. V. Artero, Dr. M. Razavet, Dr. C. Baffert,
yields by replacing one axial ligand of cobaloxime moieties
Prof. M. Fontecave
Laboratoire de Chimie et Biologie des MØtaux
UniversitØ JosephFourier
CNRS, UMR 5249
CEA, DSV/iRTSV/LCBM
with the pyridine residue of the previously reported
2+
[
[
(bpy) Ru(l-pyr)]
4,5-f]phenanthroline). NMR measurements and ESI-MS
complex (l-pyr= (4-pyridine)oxazolo-
2
[13]
analysis are consistent with the l-pyr ligand connecting the
ruthenium and cobalt centers. This was further supported by
CEA-Grenoble, Bat K’
[
12]
cyclic voltammetry:
in addition to ruthenium-centered
17 rue des Martyrs, 38054 Grenoble Cedex 9 (France)
Fax: (+33)4-3878-9124
E-mail: vincent.artero@cea.fr
processes, which are not significantly modified upon complex-
ation to the cobalt center, cyclic voltammograms of 1–3 show
II
I
Dr. W. Leibl
Laboratoire de Photocatalyse et Biohydrog›ne
CEA, DSV/iBiTecS
Gif-sur-Yvette (France)
CNRS, URA 2096 (France)
Co /Co reversible processes shifted by ꢀ 80 mV to more
positive potentials relative to the starting cobaloximes,
probably because of the overall 2 + charge of the compounds.
We checked by cyclic voltammetry that the cobaloxime
moieties retain their electrocatalytic properties for hydrogen
production in all three heterobinuclear complexes: an elec-
trocatalytic wave corresponding to proton reduction develops
at À0.45 V vs. Ag/AgCl upon addition of increasing amounts
of p-cyanoanilinium tetrafluoroborate to a solution of 1 in
[
**] The authors acknowledge SØbastien Camensuli for experimental
work during his Master’s training and the Life Science Division of
the Commissariat à l’Energie Atomique for financial support within
the BioHydrogen program.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
[
12]
CH CN (electrocatalytic waves are observed at À0.9 V vs.
3
5
64
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 564 –567

Angewandte
Chemie
AgCl for 2 in the presence of Et NHCl and À0.65 V vs AgCl
of 1 was observed during the first hour of irradiation but
3
for 3 in the presence of p-cyanoanilinium tetrafluoroborate).
hydrogen production was sustained for the following time
À1
H was produced when a CdI-doped Hg light source was
2
with a turnover frequency (TOF) of 7–8 h . Up to 85
used to irradiate acetone solutions of 1–3 in the presence of
turnovers could be obtained in an eight-hour experiment
(run 4, Table 1). Lower amounts of H2 were obtained in
1
00 equiv of Et N as the sacrificial electron donor and
3
1
00 equiv of Et NHBF as the proton source (Table 1). No
CH CN (10 turnovers within 4 h, Figure 1), MeOH (TON =
3
4
3
[14]
9
), DMF (TON = 3), and 1,2-dichloroethane (TON = 0).
When the reaction was carried out using water as the proton
Table 1: Photocatalytic hydrogen production in acetone using Et N as
the sacrificial electron donor. TONs are calculated based on the cobalt
moiety.
3
+
source instead of Et NH , much lower turnover numbers
3
were obtained (run 7, Table 1), probably because the pH
value was too high. Compounds 1 and 2 were active when a
UV cut-off filter was used (runs 5, 6, and 9, Table 1) as well.
Under these conditions more than 100 turnovers (run 6,
Table 1) could be achieved in 15 h. Comparison of runs 5 and
1 indicates that both UV and visible light contribute to the
[
a]
[b]
Run Photocatalyst
Irrad.
TON
time [h]
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
3
–
1
4
4
8
1
15
4
4
4
4
4
4
4
1
4
32
56
[
c]
c]
d]
[
60
[
photochemical production of H using 1 as catalyst. Finally 4,
2
85
[
a close analogue of 1 in which the bridging ligand was
modified by replacing the central oxazole unit with an
imidazole unit (Scheme 1), showed a two-fold increased
photocatalytic activity with 104 turnovers achieved within
16
d]
103
[
e]
22
17
[
d]
14
12
4
h in the same conditions as used in run 2.
0
1
2
3
4
5
The mechanism for cobaloxime-catalyzed hydrogen evo-
<1
<1
<1
lution is now well documented. It implies the reduction of the
[(bpy) Ru(l-pyr)](PF )
6 2
[Co(dmgBF ) (OH ) ]
[Ru(bpy) ]Cl + 1 equiv [Co(dmgBF ) (OH ) ]
[Ru(bpy) ]Cl + 2.4 equiv [Co(dmgH) (OH ) ]
2
I
cobaloxime to the Co state followed by protonation to yield a
2
2
2 2
[
f]
[6,7,15]
20
2
cobalt(III) hydride intermediate.
The latter is then
3
2
2
2
2 2
[
f]
III
3
2
2
2
2
protonated to generate dihydrogen. The resulting Co
species is finally reduced either at the electrode or through
a retrodismutation reaction with bulk Co . Since the observed
[
g]
+
15.4 equiv dmgH2
À1
I
[
a] [Ru]=0.43 mmolL . [b] Irradiation was performed in Pyrex glassware
using a 150-W CdI-doped Hg lamp; unless otherwise stated, 100 equiv
electrocatalytic potentials are more positive than the standard
+
Et N and 100 equiv Et NH were added. [c] 300 equiv Et N and 300 equiv
3+
2+
0
3
3
3
potentials of the [Ru(bpy) ] /[Ru(bpy) ] * (E = À0.87 V vs.
3
3
+
Et NH were added. [d] A UV cut-off filter was placed between the lamp
3
2+
standard hydrogen electrode (SHE)) or the [Ru(bpy) ] /
3
and the Schlenk tube. [e] 100 equiv Et N and 100 equiv H O were added.
3
2
+
0
[16]
[
Ru(bpy)3] (E = À1.28 V vs. SHE) couples, photochem-
[
[
f] Only traces of H were detected when a UV cut-off filter was used.
2
g] dmg =dimethylglyoximato dianion.
2
À
ical production of hydrogen mediated by compounds 1–3 is
likely to proceed from the thermodynamically favorable
electron transfer to the catalytic cobalt center either directly
from the photoexcited ruthenium moiety (oxidative quench-
ing) or from the reduced sensitizer resulting from reductive
quenching by an electron donor. This could be confirmed by
H could be detected when the catalyst was omitted (run 11,
2
Table 1) or replaced by the mononuclear ruthenium complex
(
(
(
run 12, Table 1) or the mononuclear cobaloxime moiety
run 13, Table 1). With up to 56 turnovers achieved over 4 h
Figure 1), 1 proved to be more active as a photocatalyst than
irradiation experiments carried out in the presence of Et N as
3
the electron donor but without any added proton source:
flashing a solution of 1 in DMF with 440-nm laser light
resulted in the appearance of a broad absorption band located
at 600 nm, which could be assigned to reduced cobaloxime,
2
and 3 (compare runs 2, 8, and 10, Table 1). Increasing the
amounts of Et N and Et NHBF had only a slight effect
3
3
4
(
compare runs 2 and 3, Table 1). Prolonged irradiation led to
I
II
increased H production (compare runs 2 and 1, 4 and 3, and 6
either described as a Co species or as a Co ion in interaction
2
[9,15]
and 5, Table 1). The maximum turnover frequency in the case
with a radical located on the ligand.
Photoaccumulation of
this species reached a saturation level after about 100 flashes.
Addition of a great excess of p-cyanoanilinium tetrafluoro-
borate leads to recovery of the original absorption spectrum
of the sample.
Measurements of luminescence lifetimes at 650 nm upon
laser excitation at 532 nm were performed in deaerated
acetone in order to gain more insight into the electron-
transfer mechanism. All complexes displayed the classical
exponential decay of the metal–ligand charge-transfer
(
MLCT) state (lowest triplet excited state resulting from
[
12]
intersystem crossing) observed for other ruthenium tris-
diimine) compounds. First-order decay lifetimes of 1.63, 1.17,
(
Figure 1. Photochemical production of hydrogen from acetone (^)
À1
and 1.72 ms were obtained for 1, 2, and 3, respectively. These
data are consistent with an oxidative quenching mechanism
and acetonitrile (~) solutions (10 mL) of Et N (0.043 molL ) and
3
À1
À1
Et NHBF (0.043 molL ) catalyzed by 1 (0. 43 mmolL ).
3
4
Angew. Chem. Int. Ed. 2008, 47, 564 –567ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
565

Communications
(
light-induced electron transfer from the photoexcitated
achieved in 15 h; run 6, Table 1) provides the largest reported
turnover number: the Ru–Pt photocatalyst synthesized by
Sakai et al. was shown to achieve 4.8 turnovers over 10 h of
ruthenium moiety to the cobaloxime center). However, this
process is only slightly competitive with the intrinsic decay of
the MLCT (1.72 ms) measured on [(bpy) Ru(l-pyr)](PF ) .
[
20]
irradiation in water, the Ru–Pd photocatalyst designed by
Rau et al. stops after 56 turnovers over 29 h irradiation in
2
6 2
[17]
On the basis of the comparison of these lifetimes yields of
%, 32%, and < 1% and intrinsic time constants for electron
transfer of 30 ms, 4 ms, and > 300 ms , respectively, could be
[
21]
5
acetonitrile, and up to 60 TON was reported for the Ru–Rh
[22,23]
system from Brewer et al.
competitive with regard to [Rh (dfpma) (PPh )(CO)]
Compound 1 proves also
[
18]
estimated for 1, 2, and 3.
2
3
3
In Table 1, we also compare the photocatalytic properties
of our novel supramolecular systems with those of multi-
(dfpma = bis(difluorophosphino)methylamine) which ach-
ieves 80 turnovers for H production in 0.1m HCl in THF
2
2
+
[24]
component {[Ru(bpy) ] /cobaloxime} systems. In the dmgH
with lexc > 338 nm.
3
series, the multicomponent system afforded only 2 turnovers
The supramolecular compounds presented here pave the
way towards efficient photocatalytic devices for hydrogen
production. First of all, substituting cobalt for rare and
expensive platinum, palladium, or rhodium metals in photo-
catalysts is a first step toward economically viable hydrogen
production. Cobaloximes appear to be good candidates for
(
(
run 15, Table 1) after 4 h, whereas 17 were obtained with 2
run 8, Table 1). In the dmgBF series, the supramolecular
2
system was also superior but to a lesser extent (compare
runs 1 and 14, Table 1). The difference became much greater
when the samples were irradiated using a UV cut-off filter
since under these conditions, the multicomponent systems
H -evolving catalysts, and they may provide a good basis for
2
[
3]
were very slow, in agreement with previous reports, while 2
maintains its activity and 1 retains half of it.
the design of photocatalysts that function in pure water as
both the solvent and the sustainable proton source. Secondly,
a molecular connection between the sensitizer and the H2-
evolving catalyst seems to provide advantages regarding the
photocatalytic activity. Structural modifications of this con-
nection should allow a better tuning of the electron transfer
between the light-harvesting unit and the catalytic center and
thus an increase of the efficiency of the system. Further
developments may eventually lead to the replacement of the
rare and expensive ruthenium center by other photosensi-
tizers such as inorganic nanoparticles (quantum dots) or
nanocrystalline materials meeting all the specifications
required for technological applications.
The high catalytic activity of the supramolecular systems
described here relies on the following considerations. First,
the H -evolving catalytic center in 1 is quite stable towards
2
both hydrolysis and hydrogenation reactions. The BF2-
bridged catalytic center is known to be more resistant towards
[
19]
acidic hydrolysis,
and the lowered nucleophilicity of its
[
7]
hydride derivative limits undesired hydrogenation reac-
tions. By contrast, addition of 6 to 15 equiv of free dmgH was
2
found to be necessary in the multicomponent system
described by Lehn et al. to prevent dissociation of [Co-
(
dmgH) (OH ) ] and replace hydrogenated ligand formed by
2
2
[
2
3]
side reactions.
II
Second, the Co state is more easily reducible in BF2-
bridged than in H-bridged cobaloximes, which facilitates Experimental Section
electron transfer from the ruthenium to the cobalt center.
This is demonstrated by comparison of the activity of 1 and 2
under the same conditions as well as in the multicomponent
See the Supporting Information for experimental details including
synthetic and photocatalytical assay procedures, cyclic voltammo-
grams of 1, [Co(dmgBF ) (dmf) ], [(bpy) Ru(l-pyr)](PF )
2
2
2
2
6 2
(Figure S1), and 3 (Figure S2), electrochemical parameters for 1–3
0
II
I
system: substituting [Co(dmgBF ) (OH ) ] (E (Co /Co ) =
2
2
2 2
and corresponding cobaloximes (Table S1), electrocatalytic behavior
of 1 (Figure S3), and emission kinetics and steady-state emission
spectra of 1–3 (Figures S4 and S5).
0
II
À0.55 V vs. Ag/AgCl) for [Co(dmgH) (OH ) ] (E (Co /
2
2 2
I
Co ) = À0.98 V vs. Ag/AgCl) results in a tenfold increase in
hydrogen production (runs 14 and 15, Table 1).
Third, the presence of a conjugated bridging ligand
facilitates the transfer of photogenerated electrons either
through bonds or by an outer-sphere mechanism favored by
the spatial proximity of the ruthenium center and the catalytic
cobaloxime moiety. Under similar conditions, the supra-
molecular systems are indeed from 1.5 to 8.5 times more
efficient than the multicomponent system, in which gener-
Received: July 3, 2007
Keywords: cobaloxime · hydrogen · photolysis · ruthenium ·
.
supramolecular chemistry
I
[1] N. Armaroli, V. Balzani, Angew. Chem. 2007, 119, 52 – 67;
Angew. Chem. Int. Ed. 2007, 46, 52 – 66; S. Licht, Chem.
Commun. 2005, 4635 – 4646; M. Grätzel, Nature 2001, 414,
ation of the catalytically active Co center occurs through
intermolecular electron transfer from the light-harvesting
unit. When 1 equiv of pyridine was added to the systems
described in runs 14 and 15 (Table 1) in order to compare
catalysts with the same coordination sphere at cobalt, no
effect on the photocatalytic activity was observed.
338 – 344; A. J. Esswein, D. G. Nocera, Chem. Rev. 2007, 107,
4022 – 4047.
[2] S. Rau, D. Walther, J. G. Vos, Dalton Trans. 2007, 915 – 919.
[
3] J. Hawecker, J. M. Lehn, R. Ziessel, Nouv. J. Chim. 1983, 7, 271 –
77.
4] M. Kirch, J. M. Lehn, J. P. Sauvage, Helv. Chim. Acta 1979, 62,
345 – 1384; U. Kölle, M. Grätzel, Angew. Chem. 1987, 99, 572 –
2
The ruthenium–cobaloxime compounds reported here are
the first supramolecular photocatalysts for hydrogen produc-
[
1
tion using first-row transition-metal H -evolving catalytic
2
574; Angew. Chem. Int. Ed. Engl. 1987, 26, 567 – 570; C. V.
Krishnan, B. S. Brunschwig, C. Creutz, N. Sutin, J. Am. Chem.
Soc. 1985, 107, 2005 – 2015; J. Rosenthal, J. Bachman, J. L.
[2]
centers. Furthermore, compared to the three previously
reported supramolecular systems, 1 (with up to 103 turnovers
5
66
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 564 –567

Angewandte
Chemie
Dempsey, A. J. Esswein, T. G. Gray, J. M. Hodgkiss, D. R.
Manke, T. D. Luckett, B. J. Pistorio, A. S. Veige, D. G. Nocera,
Coord. Chem. Rev. 2005, 249, 1316 – 1326.
[17] Essentials of Molecular Photochemistry (Eds.: A. Gilbert, J.
Baggott), Blackwell Scientific Publications, Oxford, 1991.
[18] It should be noted that these experiments achieved in the
absence of a sacrificial electron donor give only an indication
about the rate of the first electron transfer. The more efficient
quenching observed for 2 may be explained by the fact that it
[
[
5] J. G. Vos, J. M. Kelly, Dalton Trans. 2006, 4869 – 4883.
6] M. Razavet, V. Artero, M. Fontecave, Inorg. Chem. 2005, 44,
4
786 – 4795.
7] C. Baffert, V. Artero, M. Fontecave, Inorg. Chem. 2007, 46,
817 – 1824.
III
II
[
[
[
involves electron transfer to a Co center instead of a Co
1
center as in the case of 1 and 3.
8] X. L. Hu, B. M. Cossairt, B. S. Brunschwig, N. S. Lewis, J. C.
Peters, Chem. Commun. 2005, 4723 – 4725.
9] O. Pantani, E. Anxolabehere-Mallart, A. Aukauloo, P. Millet,
Electrochem. Commun. 2007, 9, 54 – 58.
[19] A. Bakac, J. H. Espenson, J. Am. Chem. Soc. 1984, 106, 5197 –
5202.
[20] H. Ozawa, M. A. Haga, K. Sakai, J. Am. Chem. Soc. 2006, 128,
4926 – 4927.
[
[
[
[
10] C. Konigstein, J. Photochem. Photobiol. A 1995, 90, 141 – 152.
11] A. Melis, T. Happe, Plant Physiol. 2001, 127, 740 – 748.
12] See the Supporting Information.
[21] S. Rau, B. Schafer, D. Gleich, E. Anders, M. Rudolph, M.
Friedrich, H. Gorls, W. Henry, J. G. Vos, Angew. Chem. 2006,
118, 6361 – 6364; Angew. Chem. Int. Ed. 2006, 45, 6215 – 6218.
[22] K. Brewer, M. Elvington, US 20060120954A1, 2006; M. Elving-
ton, J. Brown, S. M. Arachchige, K. J. Brewer, J. Am. Chem. Soc.
2007, 129, 10644 – 10645.
[23] It should be noted that the last two systems were assayed using
LED irradiation at 470 nm.
[24] A. F. Heyduk, D. G. Nocera, Science 2001, 293, 1639 – 1641.
13] C. R. Mayer, E. Dumas, F. Secheresse, Chem. Commun. 2005,
345 – 347.
[
14] Decomposition occurs in DMF and 1,2-dichloroethane with the
formation of a black precipitate.
15] P. Connolly, J. H. Espenson, Inorg. Chem. 1986, 25, 2684 – 2688.
16] U. Kölle, New J. Chem. 1992, 16, 157 – 169.
[
[
Angew. Chem. Int. Ed. 2008, 47, 564 –567ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
567