Impact of ligand exchange in hydrogen production from cobaloxime-containing photocatalytic systems

Article abstract of DOI:10.1021/ic2010166

Ligand exchange on the Co(dmgH)₂(py)Cl water reduction catalyst was explored under photocatalytic conditions. The photosensitizer fluorescein was connected to the catalyst through the axially coordinated pyridine. While this two-component compl

Full text of DOI:10.1021/ic2010166

ARTICLE
pubs.acs.org/IC
Impact of Ligand Exchange in Hydrogen Production from
Cobaloxime-Containing Photocatalytic Systems
Theresa M. McCormick, Zhiji Han, David J. Weinberg, William W. Brennessel, Patrick L. Holland,* and
Richard Eisenberg*
Department of Chemistry, University of Rochester, Rochester, New York 14627, United States
S Supporting Information
b
ABSTRACT: Ligand exchange on the Co(dmgH)2(py)Cl water reduction
catalyst was explored under photocatalytic conditions. The photosensitizer
fluorescein was connected to the catalyst through the axially coordinated
pyridine. While this two-component complex produces H2 from water
under visible light irradiation in the presence of triethanolamine (TEOA),
it is less active than a system containing separate fluorescein and
[CoIII(dmgH)2(py)Cl] components. NMR and photolysis experiments
show that the Co catalyst undergoes pyridine exchange. Interestingly,
glyoximate ligand exchange was also observed photocatalytically and by
NMR spectroscopy, thereby showing that integrated systems in which the
photosensitizer is linked directly to the Co(dmgH)2(py)Cl catalyst may not remain intact during H2 photogeneration. These
studies have also given rise to insights into the catalyst decomposition mechanism.
’ INTRODUCTION
excited state from which electron transfer occurred. Efforts to
restart hydrogen generation in this system required addition of
both PS and catalyst to become active. In 2010, we reported an
even more active system in which a Se-containing rhodamine
derivative (Rhod-Se) was employed as the photosensitizer along
with Co(dmgH)2(py)Cl as the catalyst, but it too suffered from
photoinstability.12 In both sets of studies, it was noted that at
higher concentrations of the Co catalyst or with additional free
dmgH2 present the systems exhibited increased longevity for H2
generation, suggesting a possible connection between catalyst
and PS decomposition.11,19,20
Artificial photosynthesis to split water into its constituent
elements is the most promising approach in the conversion of
solar energy into the stored energy of chemical bonds.1À4 The
reductive side of this reaction involves the light-driven generation
of hydrogen from aqueous protons. Aspects of this reaction have
been studied for more than three decades using molecularly
based systems containing a chromophore or photosensitizer for
light absorption, a catalyst for H2 generation, an electron source
for proton reduction, and a means of electron transfer to the
catalyst.5À10 Advances in the photosensitizer (PS)10À16 and
catalyst13,15,17,18 components have led to systems having both
high turnover frequencies (TOF) and overall turnover numbers
(TON), but these systems in general possess limited photo-
stabilities and have rates that may be controlled by diffusion of
the components.
In 2009 we reported the first homogeneous system for photo-
reduction of water that did not contain platinum group metals.7
The system employed the xanthene-based dye Eosin Y as the PS,
the Co(III) complex Co(dmgH)2(py)Cl (dmgH = dimethyl-
glyoxime, py = pyridine) as the catalyst, and triethanolamine (TEOA)
as the sacrificial electron donor and gave >900 turnovers of H2
with a quantum yield of ∼4% for 520À540 nm light. While this
system exhibited good rates of H2 production (∼70 TON/h,
based on photosensitizer), it suffered from photodecomposition,
with hydrogen generation ceasing after 12 h. Under the same
conditions, the nonhalogenated dye fluorescein (Fl) did not
produce H2 but also did not bleach. The dominant electron
transfer behavior of the Eosin Y dye was thought to be linked
to the heavy atom effect of the bromide substituents, thereby
The present paper describes two different but related studies
that were done to address the activity and durability of these H2
photogenerating systems. The first was based on the reported
observation that reduction of the Co(dmgH)2(py)Cl catalyst led
only to ClÀ dissociation and not dissociation of pyridine.21,22
This observation stimulated construction of the fluoresein-
derivatized pyridine 1 and the resultant integrated system shown
as 2 in which the fluorescein dye is linked directly to the catalyst
through the pyridyl ligand to facilitate intramolecular electron
transfer upon photoexcitation relative to bimolecular reaction
Received: May 13, 2011
Published: October 07, 2011
3
facilitating intersystem crossing (ISC) to the longer-lived ππ*
r
2011 American Chemical Society
10660
dx.doi.org/10.1021/ic2010166 Inorg. Chem. 2011, 50, 10660–10666
|

Inorganic Chemistry
ARTICLE
between the indivisual PS and the catalyst molecules.23À26 The
second study arose out of the first in order to rationalize the
observation that the newly synthesized linked component system
2 offered no improvement over the corresponding system having
separate components. Specifically, as a result of that conclusion,
ligand exchange was examined in the Co(dmgH)2(py)Cl com-
plex under catalytic conditions, and the results give insight into
the detailed mechanism of operation of the dye + Co(dmgH)2-
(py)Cl catalyst system.
acetonitrile was prepared. Equal volumes of the dye and cobaloxime
solutions were added to total 5 mL in 40 mL scintillation vials. Varying
amounts of dmgH2 and dpgH2 were added as solids. The samples were
placed into a temperature-controlled block at 15 °C and sealed with
airtight caps fitted with a pressure transducer and septa. The samples
were degassed by bubbling with a 20% mixture of CH4 in N2, in which
CH4 was later used as the internal standard for GC analysis of evolved
H2. The cells were irradiated from below with high-power Philips
LumiLEDs Luxeon Star Hex, green (520 nm) 700 mA, LEDs. The light
power of each LED was set to 0.15 W, measured with a L30 A Thermal
sensor and Nova II Power meter (Ophir-Spiricon LLC). The samples
were mixed by placing the apparatus on an orbital shaker. The pressure
changes in the vials were recorded using a Labview program read from a
Freesale semiconductor pressure sensor (MPX4250A series). At the end
of the irradiation, the headspace of the vials was sampled by gas
chromatography to ensure that the measured pressure changes were
caused by H2 generation and to double check that the amount of
generated hydrogen calculated by the change in pressure corresponded
to the amount determined by the GC. The amounts of hydrogen evolved
were determined using a Shimadzu GC-17A gas chromatograph with a
5 Å molecular sieve column (30 m, 0.53 mm) and a TCD detector by
injecting 100 μL of headspace into the GC and were quantified by a
calibration plot to the internal CH4 standard. The LED photolysis setup
was built with guidance and help from Profs. Stefan Bernhard and Karen
Brewer.15,30
’ EXPERIMENTAL SECTION
UVÀvis Absorption Spectra. A 1 Â10À5 M solution of 1, 2, and
fluorescein in 0.1 M NaOHaq was analyzed on a Hitachi U2000
spectrometer in a 1 cm path length quartz cuvette.
Materials. The complexes [CoIII(dmgH)2(py)Cl],27 [CoIII(dpgH)2-
(py)Cl],28 and [CoIII(dmgH)2(py)2]PF629 were synthesized by previously
reported methods and recrystallized from methylene chloride. All solvents were
used without further purification unless otherwise stated. Dimethylglyoxime
(dmgH2) and diphenylglyoxime (dpgH2) were purchased from Aldrich and
used without further purification. 1H NMR spectra were recorded on a Bruker
Avance 500 MHz spectrometer; data are reported in ppm.
Syntheses. Pyridine-4-amido-aminofluorescein (1). Aminofluor-
escein (1.5 g, 4.3 mmol) and isonicotinoyl chloride hydrochloride (0.92 g,
6.5 mmol) were stirred in 100 mL of anhydrous pyridine for 24 h under a
nitrogen atmosphere. The yellow precipitate was collected by vacuum
filtration. The product was purified on a silica column with an ethyl
acetate:methanol:hexanes 3:2:1 eluent. The collected product was
dissolved in a 0.1 M NaOH in MeOH solution and precipitated by
adding acetic acid dropwise. The orange precipitate was collected by
vacuum filtration and washed three times each with water, ethanol, and
Fluorescence Quantum Yield. The absorbance of dilute solu-
tions of 1, 2, and fluorescein in 0.1 M NaOHaq solutions at 491 nm were
set to 0.1 recorded on a Hitachi U2000 spectrometer. The area under the
emission spectra was recorded on a Spex Fluoromax-P fluorimeter with a
photomultiplier tube detector. The quantum yields were calculated
using the ratio of the numerical integration of the emission spectra of
1 and 2 to the fluorescein standard.
Fluorescence Quenching. A solution of fluorescein in a 1:1 mixture
of CH3CN:H2O was prepared in a quartz cuvette fitted with a septum cap.
Aliquots of 30 μL of TEOA were added, and the intensity of the fluorescence
was monitored by steady state fluorescence exciting at 500 nm on a Spex
Fluoromax-P fluorimeter with a photomultiplier tube detector.
Kinetic NMR Experiments for Pyridine Exchange. A 0.012 M
solution of [CoIII(dmgH)2(py)2]PF6 in CD3CN was prepared and
transferred to a J. Young resealable NMR tube. Immediately before the
first NMR measurement, 4 mol equiv of 4-tert-butylpyridine (tBupy)
was added. NMR spectrawere subsequently recordedperiodicallyatroom
temperature until spectral changes were no longer observed.
Kinetic NMR Experiments for Dimethyl- or Diphenylglyox-
ime Exchange. A 0.012 M solution of [CoIII(dmgH)2(py)Cl] in 9:1
CD3CN/D2O was preparedand transferred to a J. Young resealable NMR
tube. Immediately before the first NMR measurement, 4 mol equiv of
dpgH2 was added. NMR spectra were subsequently recorded periodically
at room temperature until spectral changes were no longer observed.
X-ray Crystal Structure Determination of 2. A crystal of 2
(0.22 Â 0.16 Â 0.06 mm3) was placed onto the tip of a 0.1 mm diameter
glass capillary tube or fiber and mounted on a Bruker SMART APEX II CCD
Platform diffractometer for a data collection at 100.0(1) K.31 A prelim-
inary set of cell constants and an orientation matrix were calculated from
reflections harvested from three orthogonal wedges of reciprocal space.
The full data collection was carried out using Mo Kα radiation (graphite
monochromator) with a frame time of 60 s and a detector distance of
3.97 cm. A randomly oriented region of reciprocal space was surveyed:
four major sections of frames were collected with 0.50° steps in ω at four
different j settings and a detector position of À38° in 2θ. The intensity
data were corrected for absorption.32 Final cell constants were calculated
diethyl ether (1.05 g, 53.7%). Anal. Calcd for C26H16N2O6 CH3OH: C,
3
67.08; H, 3.96; N, 5.79. Found: C. 67.27; H. 3.62; N. 5.93. 1H NMR (400
MHz, DMSO-d6): δ 10.91 (s, 1H), 10.15 (br, 2H), 8.81 (d, J = 6.0 Hz,
2H), 8.45 (d, J = 1.2 Hz, 1H), 8.06 (dd, J = 8.4, 1.6 Hz, 1H), 7.88 (d, J =
5.6 Hz, 2H), 7.27 (d, J = 8.4 Hz, 1H), 6.65 (d, J = 2.0 Hz, 2H), 6.60 (d, J =
8.4 Hz, 2H), 6.54 (d, J = 2.0 Hz, 2H) ppm.
Co(dmgH)2(py-aminofluorescein)Cl (2). Co(dmgH)2Cl2 (158 mg,
0.442 mmol) was mixed in 20 mL of methanol, resulting in a green
suspension. Sodium hydroxide (17.7 mg, 0.442 mmol) was added, and
after 3 min, 1 (200 mg, 0.442 mmol) was added with the mixture left to
stir for 24 h. The orange-brown mixture was filtered, and 67 mg of 2 was
collected. By slow evaporation of the solvent from the filtrate, red-brown
crystals were obtained (130 mg, 40%). HRMS-EI C34H30O10N6Co1:
calcd 741.1350, found, 741.13245, Δ = À3.4 ppm. 1H NMR (400 MHz,
DMSO-d6): δ 10.97 (s, 1H) 10.11 (s, 2H), 8.34 (d, J = 1.6 Hz, 1 H), 8.22
(d, J = 6.8 Hz, 2H), 7. 92 (dd, 1H), 7.86 (d, J = 6.8 Hz, 2H), 7.24
(d, J = 8.4 Hz, 1 H), 6.64 (d, J = 2.4 Hz, 2 H), 6.57À6.52 (m, 4H), 2.26
(s, 12H) ppm.
Hydrogen Evolution Studies. A1Â 10À4 M solution of Eosin Y in
9:1 H2O/TEOA (adjusted to pH 7 using concentrated HCl) was prepared
and protected from light before use. A 5 Â 10À4 M solution of cobaloxime
complex (either [CoIII(dmgH)2(py)Cl] or [CoIII(dpgH)2(py)Cl]) in
10661
dx.doi.org/10.1021/ic2010166 |Inorg. Chem. 2011, 50, 10660–10666

Inorganic Chemistry
ARTICLE
from the xyz centroids of 4000 strong reflections from the actual data
collection after integration.33
around the Co center is a distorted octahedron, with the NÀCoÀN
chelate angles being 81.1(1)° and 81.9(1)° for the two dmgH ligands.
These values are similar to those in [Co(dmgH)2(4-COOMe-py)Cl]
(81.90(7)° and 81.30(7)°) and [Co(dmgH)2(4-Me2N-py)Cl]
(81.68(4)° and 81.36(4)°).6 The CoÀN bond lengths of 2 are
within error the same as those reported for the related complexes
Co(dmgH)2(4-COOMe-py)Cl (1.959(2) Å for Co to pyridyl
and ∼1.89 Å for Co to dmgH) and Co(dmgH)2(4-Me2N-py)Cl
(1.9464(9)Å for Co to pyridyl and avg. 1.894(5) Å for Co
to dmgH).6
The crystal structure contains one cocrystallized methanol and
one and a half cocrystallized water solvent molecules per cobalt
complex. There are two major intermolecular hydrogen-bonding
motifs present: one between the Co complex and the water and
methanol molecules and the other due between the lactone
carbonyl oxygen O10 and the neighboring fluorescein O6. Intra-
molecular hydrogen bonding is also found within the complex
between the dimethylglyoximate ligands.
The structure was solved using SIR9734 and refined using SHELXL-
97.35 The space group P21/n was determined based on systematic
absences and intensity statistics. A direct-methods solution was calcu-
lated which provided most non-hydrogen atoms from the E-map. Full-
matrix least-squares/difference Fourier cycles were performed which
located the remaining non-hydrogen atoms. All non-hydrogen atoms
were refined with anisotropic displacement parameters. When possible
hydrogen atoms involved in hydrogen bonding were found from the
difference Fourier map, and their positions were refined. Others were
placed in reasonable positions to maximize hydrogen bonding. All other
hydrogen atoms were placed in idealized positions and refined as riding
atoms with relative isotropic displacement parameters. The final full-
matrix least-squares refinement converged to R1 = 0.0584 (F2, I > 2σ(I))
and wR2 = 0.1555 (F2, all data). A table with crystal data and structure
refinement for 2 can be found in the Supporting Information (Table S1).
Photogeneration of Hydrogen. Using 2 as both PS and
catalyst, photogeneration of H2 occurs under irradiation with
visible light (5 Â 10À4 M 2, 5% v/v TEOA in 1:1 CH3CN:H2O,
pH 7). During 24 h of irradiation with 520 nm light a total 1.0 mL
of H2 is produced at a constant rate (TON = 20). Addition of
10 equiv of dmgH2 prolongs this constant rate of H2 generation
for 72 h. The Co(dmgH)2(py)Cl catalyst is minimally soluble in
water, but complex 2 is soluble in 5% TEOA/water (v:v with pH
adjusted to ∼7 with HCl). When this solution is used as the
solvent without any additional organic solvent, H2 generation is
observed. However, when using water as the solvent under other-
wise identical conditions, the rate of H2 generation is ca. one-
third of that seen in the water/MeCN system (11 vs 32 TON
over 28 h). Hydrogen generation from systems with molecular
catalysts in water only has been reported a few times.37
To assess the benefit of the attached PSÀcatalyst complex, a
H2-generating system composed of the derivatized fluorescein 1
(5Â 10À4 M) asthePSand[CoIII(dmgH)2(py)Cl] (5Â 10À4 M)
as the catalyst in 1:1 CH3CN:H2O with TEOA was prepared.
Photogeneration of H2 from this system resulted in 31 TON over
24 h (1.3 mL/day), which was slightly greater in rate and TON
than when 2 was used as both PS and H2-generating catalyst under
the same conditions (pH 7, λ = 520 nm). A system containing
fluorescein as the PS (rather than 1) and [CoIII(dmgH)2(py)Cl]
as the catalyst at the same concentrations as those used for the
system containing 2 (5 Â 10À4 M each) produced H2 at even
higher rates and TON’s (63 TON in 24 h, 3.6 mL/day) than for
either 2 or 1 + [CoIII(dmgH)2(py)Cl], as shown in Figure 2.
These observations indicate that attachment of fluorescein to the
catalyst complex via pyridine in 2 offers no benefit in H2
generation and led to the suggestion that despite the earlier
assertion that pyridine does not dissociate upon reduction of
’ RESULTS AND DISCUSSION
Synthesis and Structure. A version of fluorescein with an
appended pyridine was synthesized from the reaction of 4-
amino-fluorescein and isonicotinoyl chloride hydrochloride. This
reaction yields compound 1, which in turn displaces ClÀ from
H[Co(dmgH)2Cl2]27 to give [Co(dmgH)2(1)Cl] (2), as con-
firmed by NMR spectroscopy and high-resolution mass spectro-
metry. X-ray crystal structure analysis of 2 verifies the CoIII
coordination sphere and reveals that the fluorescein of 1 exists as
its ring-closed lactone isomer (2b, Figure 1). This isomer is
known to undergo tautomerization to the ring-opened, anionic
form at neutral or basic pH.36 The molecular structure of 2
reveals little influence on the Co(dmgH) portion of the structure
from the attached fluorescein moiety. The bond length to the
pyridyl nitrogen is longer (1.952(2) Å, Table 1) than the CoÀN
bonds to the dmgH ligands (avg. 1.894(5) Å). The geometry
Figure 1. Molecular structure of 2b with thermal ellipsoids at the 50%
probability level. Atom-numbering schemes are shown; the hydrogen
atoms are omitted for clarity.
Table 1. Selected Bond lengths [Å] and angles [deg] for 2
Co(1)ÀN(4)
Co(1)ÀN(3)
Co(1)ÀN(2)
Co(1)ÀN(1)
Co(1)ÀN(5)
Co(1)ÀCl(1)
N(1)ÀC(1)
N(1)ÀO(1)
C(18)ÀC(22)
1.891(2)
1.892(2)
1.895(2)
1.899(2)
1.952(2)
2.2390(7)
1.289(4)
1.348(3)
1.505(4)
O(7)ÀC(22)
1.511(3)
1.493(4)
1.507(4)
1.346(3)
1.368(3)
1.372(3)
1.367(3)
81.9(1)
N(3)ÀCo(1)ÀN(2)
N(2)ÀCo(1)ÀN(1)
N(2)ÀCo(1)ÀN(5)
N(1)ÀCo(1)ÀN(5)
N(2)ÀCo(1)ÀCl(1)
N(1)ÀCo(1)ÀCl(1)
C(23)ÀC(22)ÀC(34)
C(23)ÀC(22)ÀO(7)
C(34)ÀC(22)ÀO(7)
98.2(1)
81.1(1)
C(22)ÀC(23)
C(22)ÀC(34)
C(26)ÀO(9)
89.18(9)
88.83(9)
92.41(7)
91.11(7)
111.6(2)
108.4(2)
107.4(2)
C(28)ÀO(8)
O(8)ÀC(29)
C(31)ÀO(10)
N(4)ÀCo(1)ÀN(3)
N(4)ÀCo(1)ÀN(2)
179.56(10)
10662
dx.doi.org/10.1021/ic2010166 |Inorg. Chem. 2011, 50, 10660–10666

Inorganic Chemistry
ARTICLE
Figure 2. Hydrogen generation from 5.0 Â 10À4 M solutions of (A)
fluorescein and [CoIII(dmgH)2(py)Cl], (B) derivatized fluorescein 1
and [CoIII(dmgH)2(py)Cl], and (C) the fluorescein-containing CoIII-
(dmgH)2 complex 2. All samples are in 5% TEOA in 1:1 CH3CN:H2O,
pH 7; irradiated with 520 nm light.
Figure 3. 1H NMR spectra for ligand exchange reactions: (a) [Co(dmgH)2-
(py)2]+ with 4 equiv of tBupy and (b) [Co(dmgH)2(py)Cl] with 4 equiv
of dpgH2. Labels for resonances are as follows: (A) free py, (B) free tBupy,
(C) py coordinated to Co, (D) tBupy coordinated to Co, (E) dmgH methyl
on [Co(dmgH)2(py)(tBupy)]+, (F) dmgH methyl on [Co(dmgH)2-
(tBupy)2]+, (G) dmgH methyl on [Co(dmgH)2(py)2]+, (H) [Co(dmgH)-
(dpgH)pyCl], (J) [Co(dmgH)2(py)Cl], and (K) free dmgH2 (intensity
of each segment is scaled for clarity).
[CoIII(dmgH)2(py)Cl], the py ligand is indeed labile during
hydrogen-generating conditions, thereby precluding any
value from the PSÀcatalyst linkage.
Previous reports have stated that Fl itself is not an effective PS
for photogeneration of hydrogen,7 but the present observations
indicate otherwise. However, for H2 photogeneration to occur
with Fl as the photosensitizer, the PS concentration must be
>1 Â 10À4 M. At this concentration, the fluorescein emission is
red shifted to λmax = 545 nm and its emission intensity is
diminished relative to dilute solutions (λmax = 515 nm). The
observed shift is attributed to dye aggregation and a longer lived
dimer emission, τ ≈ 12 ns, relative to Fl monomer emission from
dilute solution with τ ≈ 5 ns.38 When [PS] is decreased from
5 Â 10À4 to 1.2 Â 10À4 M while [CoIII(dmgH)2(py)Cl] is main-
tained at 5 Â 10À4 M, the system produces 175 TON per mol of
PS over 50 h. Decreased PS concentration decreases hydrogen
production rates (see Figure S1, Supporting Information). How-
ever, a change in catalyst concentration does not affect the rate but
rather alters system longevity for H2 generation (see Figure S2,
Supporting Information). These observations are consistent with
the turnover-limiting step in the light-driven production of H2
involving the photosensitizer only and not the catalyst.
Addition of extra py to the system containing the tethered
PSÀcatalyst complex 2 and TEOA results in a higher rate of H2
production. Over 24 h, a 4.8 Â 10À4 M solution of 2 to which
2 mol equiv of py is added produces 45 TON, as compared to 32
TON when no extra py is added (see Supporting Information,
Figure S3). This observation is also consistent with the higher
rates found using Co(dmgH)2(py)Cl as the catalyst with 1 as the
PS since py is already coordinated to the catalyst. In contrast,
addition of sterically hindered 2,6-dimethylpyridine that is inhibited
from coordinating to Co exhibits no effect on H2 generation. The
results thus suggest that the effect of py on H2 generation rates is
associated solely with coordination.
1
after 86 h (Figure 3a), as judged by H NMR spectroscopy.
After 1 week the system is near equilibrium, with 82% of the
initial [CoIII(dmgH)2(py)2]PF6 complex showing exchange in a
mixture of 18% [CoIII(dmgH)2(py)2]+, 59% [Co(dmgH)2(py)-
(tBupy)]+, and 23% [Co(dmgH)2(tBupy)2]+. This indicates
that slow exchange of the py ligands is occurring at a rate
consistent with the inertness of CoIII. The py exchange reactions
were found to proceed more rapidly under other conditions
consistent with those produced during hydrogen generation. For
example, when 2 equiv of tBupy was added to [CoIII(dmgH)2-
(py)Cl] in CD3CN under irradiation by light of λ > 420 nm
for 2 h, a mixture of 40% [CoIII(dmgH)2(py)Cl] and 60%
[CoIII(dmgH)2(tBupy)Cl] was observed by 1H NMR spectros-
copy. The exchange is accelerated by light, consistent with
photosubstitution. It was also anticipated that reduction of
Co(III) to Co(II) would result in much faster py exchange
because of the lability of CoII. In support of this hypothesis, when
a mixture of [CoIII(dmgH)2(py)2]+ and 4 equiv of tBupy in
CD3CN is stirred over Zn amalgam for 10 min, a mixture of 82%
[Co(dmgH)2(py)(tBupy)]+ and 18% [CoIII(dmgH)2(py)2]+ is
obtained.
Exchange of Glyoximate Ligands. The observation of pyri-
dine exchange from the Co glyoximate complexes raised the
question of whether glyoximate exchange was also occurring
during active catalysis of H2 generation. Such exchange was not
expected because of the bidentate nature of these ligands and the
H bonding between the glyoximate ligands that holds the two
monoanionic chelates in the equatorial positions of a tetragonally
distorted octahedral complex. However, reports of increased
system stability through addition of free dmgH2 ligand suggested
the possibility of exchange.7,12,19,37
In order to probe the possibility of py exchange from the
catalyst, complexes [CoIII(dmgH)2(py)Cl] and [CoIII(dmgH)2-
The notion of glyoximate exchange was examined in two
different ways. The first was by monitoring NMR spectra of
samples containing both complex and additional glyoximate
ligand, and the second was by examination of H2 photogenera-
tion using systems with either one or two different glyoximate
ligands present. The initial NMR experiments showed that when
1
(py)2]PF6 were examined by H NMR spectroscopy. Addition
of 4 equiv of 4-tert-butylpyridine (tBupy) to a solution of
[CoIII(dmgH)2(py)2]PF6 in CD3CN at room temperature
results in a mixture of 32% [CoIII(dmgH)2(py)2]+, 53%
[Co(dmgH)2(py)(tBupy)]+, and 15% Co(dmgH)2(tBupy)2
+
10663
dx.doi.org/10.1021/ic2010166 |Inorg. Chem. 2011, 50, 10660–10666

Inorganic Chemistry
ARTICLE
with the starting Co complex in each case. Instead, the rate
depends on the sum and types of bound and unbound diglyoxime
ligands present in the system. It can thus be concluded that
during photolysis with water present the chelating ligands are
labile. These results help to explain the finding that with
[CoIII(dmgH)2(py)Cl] as the catalyst extra dmgH2 in solution
increases system longevity through exchange.7,20,37
We previously found that to restart hydrogen generation after
the system has stopped producing H2 upon irradiation both
photosensitizer and catalyst must be added to the system. In the
present case, hydrogen generation can also be revived with
addition of Eosin and dimethylglyoxime, with the subsequent
rate of hydrogen production being slightly less than that
observed before photodecomposition. The facts that dimethyl-
glyoxime ligands are being exchanged during H2 generation and
that addition of dmgH2 + Eosin reconstitutes activity indicates
that the dimethylglyoxime ligand, whether coordinated or free, is
being converted into an inactive form for hydrogen generation.
One possibility is that the dimethylglyoxime ligand is hydro-
genated.39 While the hydrogenated or otherwise decomposed
dimethylglyoximate could act as a ligand for Co, it may exist in a
form that does not allow reduction to CoI at a potential accessible
by either the excited state of Eosin directly or the reduced Eosin
produced by reductive quenching with the sacrificial donor. We
suggested previously that the path that leads to photodecompo-
sition of the dye is also the path that leads to H2 generation.12
With regard to the initial photochemical steps in the present
systems, electron transfer quenching leading to H2 generation
can proceed by PS* oxidation of TEOA to form PSÀ that
transfers an electron to the Co catalyst or by PS* reduction
of the Co catalyst yielding PS+ that is reduced by TEOA. To
distinguish between these possibilities, emission quenching studies
of fluorescein were carried out separately using CoIII(dmgH)2-
(py)Cl as the electron acceptor or TEOA as the electron donor.
In a test of the former, addition of 50 equiv of CoIII(dmgH)2-
(py)Cl leads to diminished but still strong Fl emission. Since the
Co complex possesses some absorption at an excitation energy of
500 nm, the moderate emission decrease can be attributed, at
least in part, to competitive absorption by CoIII(dmgH)2(py)Cl.
In contrast, when 5% by volume of TEOA is added to a 5 Â 10À4 M
solution of Fl, the emission is quenched by 90% (see Figure S4,
Supporting Information). These results indicate that for the
Fl/[CoIII(dmgH)2(py)Cl]/TEOA system electron transfer lead-
ing to H2 generation proceeds by reductive quenching of PS*
by TEOA.
On the basis of these observations and similar ones obtained
using Se-derivatized rhodamine dyes with the CoIII(dmgH)2-
(py)Cl catalyst,12 we conclude that H2 production and system
decomposition proceed by reductive quenching of PS* to give
PSÀ. In essence, there exists a branchpoint in the reaction scheme
following reaction of PS* with TEOA.19 If the resultant PSÀ is able
to transfer an eÀ to the Co catalyst, H2 production will follow. On
the other hand, if the Co species is no longer able to accept an
electron from PSÀ, the latter will decompose by a slow unim-
olecular path. The exchange of coordinated ligand with free
dmgH2 thus accounts for the observed increase in system long-
evity, since the transformed ligand can exchange with dmgH2 in
solution. Further work is underway to better understand the ligand
transformation that renders the Co species incapable of hydrogen
generation in order to help develop a stable photoreduction catalyst.
Are Linked Systems Really Better? In other studies that have
been reported, connection of the PS to the catalyst through covalent
Figure 4. Photolyis of Eosin Y 5 Â 10À5 M in 5% TEOA in 1:1
CH3CN:H2O, pH 7; irradiated with 520 nm light, 2.5 Â 10À4
[CoIII(dmgH)2(py)Cl] or [CoIII(dpgH)2(py)Cl] with 5 Â 10À4
dmgH2 or dpgH2.
M
M
4 equiv of free diphenylglyoxime (dpgH2) was added to a solu-
tion of [CoIII(dmgH)2(py)Cl] in CD3CN, no exchange oc-
curred over 2 days. However, by changing the solvent to a mixed
CD3CNÀD2O solvent system (9:1 v/v), exchange did occur to
form [CoIII(dpgH)(dmgH)(py)Cl] with 35% conversion in 68 h
(Figure 3b). To verify if reduction to Co(II) facilitates dmgH
ligand exchange, a mixture of [CoIII(dmgH)2(py)Cl] and 4 equiv
of free dpgH2 in a CD3CNÀD2O solvent mixture (9:1 v/v) was
stirred over Zn amalgam for 1 min, resulting in a fast exchange
of glyoximate ligands with 38% conversion to [CoIII(dpgH)-
(dmgH)(py)Cl]. The rate of dmgH ligand exchange can also be
catalyzed by addition of 3 mol % CoII(BF4)2 6H2O under N2.
3
After mixing for 10 min, followed by subsequent exposure to
air to oxidize all Co to the +3 state, more than 50% of the
[CoIII(dmgH)2(py)Cl] complex was converted to [CoIII(dpgH)-
(dmgH)(py)Cl]. While the CoIII complex [CoIII(dmgH)2(py)Cl]
is substitutionally inert, the exchange process is accelerated
significantly by Co(II), possibly via inner-sphere electron trans-
fer through the chloride ligand of the Co(III) dmgH complex.
Ligand Exchange During H2 Generation. Both the rate of
hydrogen generation and the total amount of hydrogen gener-
ated are substantially less when using the diphenylgloximate
complex CoIII(dpgH)2(py)Cl as catalyst as compared to those
obtained with the dimethylglyoximate catalyst CoIII(dmgH)2-
(py)Cl. On the basis of this observation, it was possible to probe
diglyoxime ligand exchange during H2 photogeneration experi-
ments. For these experiments, Eosin Y was used as the PS
because of its greater TOF relative to the fluoroscein system.
In Figure 4, the amounts of hydrogen evolved as a function of
time are plotted for these runs. Curve A shows the results for a
system composed of CoIII(dmgH)2(py)Cl + 2 equiv of dmgH2,
while curve D exhibits the corresponding results for CoIII-
(dpgH)2(py)Cl + 2 equiv of dpgH2. The difference in activity
between these two catalysts in otherwise identical systems is
striking. When the two mixed systems, [CoIII(dmgH)2(py)Cl] + 2
equiv of dpgH2 and [CoIII(dpgH)2(py)Cl] + 2 equiv of dmgH2,
are similarly employed, the rates and overall H2 production amounts
are identical (see curves B and C in Figure 4) and approximately
intermediate between those of [CoIII(dmgH)2(py)Cl] + 2 equiv
of dmgH2 and [CoIII(dpgH)2(py)Cl] + 2 equiv of dpgH2
(Figure 4). The results are clear evidence of glyoxime exchange.
If the cobaloxime catalyst remained intact during photolysis,
the two experiments would have yielded different rates, consistent
10664
dx.doi.org/10.1021/ic2010166 |Inorg. Chem. 2011, 50, 10660–10666

Inorganic Chemistry
ARTICLE
bonds has yielded at best only modest improvement in overall H2
production relative to systems containing separate PS, catalyst,
and a sacrificial electron donor.10,40 For example, linkage of a
polypyridyl ruthenium- or iridium-based PS to a cobaloxime
catalyst(Co(dmgH)2Cl or Co(dmgBF2)2Cl, where dmg = dimethyl-
glyoxime) has led to relatively small increases in TON relative to
the systems of separate components, from 7 to 38 for a Ru photo-
sensitizer23,25 and from 165 to 210 for an Ir PS.24
A linked PSÀcatalyst complex 2 for H2 photogeneration offers
no advantages and poorer rates of H2 production than a system
consisting of separated PS and catalyst components. We identi-
fied three factors which mitigate any potential benefits to attach-
ment. First, the derivatized fluorescein sensitizer increases non-
radiative excited state decay pathways for the dye. Second, H2
photogeneration using the system based on fluorescein and
[CoIII(dmgH)2(py)Cl] proceeds through reductive quenching,
in which the turnover-limiting step is not facilitated by connec-
tion of the PS to the catalyst. Third, NMR and H2 photogenera-
tion experiments indicate that there is exchange of both the py
and diglyoxime ligands on the catalyst. These studies highlight
the importance of analyzing in detail the behavior of system
components in the context of system operation so that efforts to
design more efficient and active systems for H2 photogeneration
are soundly based.
Several factors may contribute to the slower generation of H2
when Fl is linked to py as in 1, whether free in solution or
incorporated into 2, relative to the system in which the compo-
nents are not connected. First is a difference in the photophysical
properties of 1, 2, and Fl. While the molar absorptivity of the
Fl absorption in all three species is very similar (72 000À
75 000 MÀ1 cmÀ1), the emissions from 1 and 2 are severely
quenched relative to Fl (ϕFl 6.6% for 1, 4.8% for 2, and 93%
for Fl).36 Previously reported amide derivatives of fluorescein
have only slightly quenched emission compared to the parent
compound (ϕFl ≈ 80%).41 The fact that emission from 1 is
quenched to the same extent as emission for 2 indicates that the
observed quenching is not by electron transfer from Fl to the Co
catalyst but instead occurs via nonradiative decay, a factor that
would lower activity for both 1 and 2 for H2 photogeneration.
Since the Fl emission is quenched by TEOA and not by the Co
catalyst in the turnover-limiting step of H2 generation, facilitating
electron transfer from Fl to Co is not expected to increase the
efficiency of H2 production. The reductive quenching mechan-
ism is thus a second factor contributing to the absence of any
benefit from attaching the PS to the catalyst in 2. Additionally, at
lower dye concentrations, quenching of Fl emission is signifi-
cantly diminished by TEOA (50% quenched by 5% v/v TEOA at
[Fl] of 1 Â 10À5 M), which helps to explain the absence of
H2 generation at lower Fl concentrations.
A third factor contributing to the similarity between the
tethered and the untethered catalysts is that the PS-containing
ligand does not remain attached to the catalyst during the
reaction. Under H2-generating conditions, the resting state of
the catalyst has been found to be CoII.7 The well-known lability
of CoII suggests that ligand exchange occurs readily on the time
scale of the catalytic reaction.42À44 This hypothesis has been
supported by both the hydrogen generation and the NMR
experiments presented here, which indicate that both pyridine
and dmgH ligands exchange readily on the cobalt catalyst during
the photogeneration experiments. Recently, using X-ray scatter-
ing techniques, a report has indicated that similar cobaloximeÀ
photsensitzer assemblies exist in equilibrium between bound and
unbound photosensitizer in acetonitrile solution.25 The fact that
the dmgH ligands exchange under these conditions casts doubt
on the feasibility of productive attachment of catalyst and chromo-
phore through any of the supporting ligands on the cobalt center.
’ ASSOCIATED CONTENT
S
Supporting Information. Detailed experimental descrip-
b
tion, crystal structure data for 2 in CIF format, plots of photolysis
experiments, SternÀVolmer plots, and full NMR spectra. This
material is available free of charge via the Internet at http://pubs.
acs.org.
’ AUTHOR INFORMATION
Corresponding Author
*E-mail: holland@chem.rochester.edu (P.L.H.); eisenberg@
chem.rochester.edu (R.E.).
’ ACKNOWLEDGMENT
We thank William W. Brennessel and the X-ray Crystallo-
graphic Facility of the Department of Chemistry at the University
of Rochester. We acknowledge the Division of Chemical
Sciences, Geosciences, and Biosciences, Office of Basic Energy
Sciences of the U.S. Department of Energy Grant DE-FG02-
09ER16121 for funding. T.M. thanks NSERC of Canada for
funding.
’ REFERENCES
(1) Esswein, A. J.; Nocera, D. G. Chem. Rev. 2007, 107, 4022–4047.
(2) Nocera, D. G. Inorg. Chem. 2009, 48, 10001–10017.
(3) Eisenberg, R. Science 2009, 324, 44–45.
(4) Gray, H. B. Nat. Chem. 2009, 1, 7.
(5) Du, P.; Knowles, K.; Eisenberg, R. J. Am. Chem. Soc. 2008,
130, 12576–12577.
(6) Du, P.; Schneider, J.; Luo, G.; Brennessel, W.; Eisenberg, R.
Inorg. Chem. 2009, 48, 4952–4962.
(7) Lazarides, T.; McCormick, T.; Du, P.; Luo, G.; Lindley, B.;
Eisenberg, R. J. Am. Chem. Soc. 2009, 131, 9192–9194.
(8) Dempsey, J.; Winkler, J.; Gray, H. J. Am. Chem. Soc. 2010, 132,
16774–16776.
(9) Dempsey, J.; Winkler, J.; Gray, H. J. Am. Chem. Soc. 2010, 132,
1060–1065.
(10) Tinker, L.; McDaniel, N.; Bernhard, S. J. Mater. Chem. 2009,
19, 3328–3337.
(11) Probst, B.; Kolano, C.; Hamm, P.; Alberto, R. Inorg. Chem.
2009, 48, 1836–1843.
’ CONCLUSION
In summary, ligand exchange on the [CoIII(dmgH)2(py)Cl]
hydrogen production catalyst during the photoreduction of
protons accounts for the lack of increase in hydrogen production
upon connection of the photosensitizer to the catalyst. It also
indicates why system longevity for H2 generation is increased
upon addition of free dmgH2 and explains how transformation of
the dmgH2 ligand provides the path for system decomposition
and cessation of hydrogen generation.
(12) McCormick, T.; Calitree, B.; Orchard, A.; Kraut, N.; Bright, F.;
Detty, M.; Eisenberg, R. J. Am. Chem. Soc. 2010, 132, 15480–15483.
(13) Arachchige, S.; Brown, J.; Brewer, K. J. Photochem. Photobiol., A
2008, 197, 13–17.
10665
dx.doi.org/10.1021/ic2010166 |Inorg. Chem. 2011, 50, 10660–10666

Inorganic Chemistry
ARTICLE
(14) Arachchige, S.; Brown, J.; Chang, E.; Jain, A.; Zigler, D.;
Rangan, K.; Brewer, K. Inorg. Chem. 2009, 48, 1989–2000.
(15) Elvington, M.; Brown, J.; Arachchige, S. M.; Brewer, K. J. J. Am.
Chem. Soc. 2007, 129, 10644–10645.
(16) Wang, X.; Goeb, S.; Ji, Z.; Pogulaichenko, N.; Castellano, F.
Inorg. Chem. 2011, 50, 705–707.
(17) Goldsmith, J.; Hudson, W.; Lowry, M.; Anderson, T.; Bernhard,
S. J. Am. Chem. Soc. 2005, 127, 7502–7510.
(18) Cline, E.; Adamson, S.; Bernhard, S. Inorg. Chem. 2008, 47,
10378–10388.
(19) Probst, B.; Rodenberg, A.; Guttentag, M.; Hamm, P.; Alberto,
R. Inorg. Chem. 2010, 49, 6453–6460.
(20) Zhang, P.; Wang, M.; Dong, J.; Li, X.; Wang, F.; Wu, L.; Sun, L.
J. Phys. Chem. C 2010, 114, 15868–15874.
(21) Connolly, P.; Espenson, J. Inorg. Chem. 1986, 25, 2684–2688.
(22) Baffert, C.; Artero, V.; Fontecave, M. Inorg. Chem. 2007, 46,
1817–1824.
(23) Li, C.; Wang, M.; Pan, J.; Zhang, P.; Zhang, R.; Sun, L. J. Organomet.
Chem. 2009, 694, 2814–2819.
(24) Fihri, A.; Artero, V.; Pereira, A.; Fontecave, M. Dalton Trans.
2008, 5567–5569.
(25) Mulfort, K.; Tiede, D. J. Phys. Chem. B 2010, 114, 14572–14581.
(26) Kobayashi, M.; Masaoka, S.; Sakai, K. Molecules 2010, 15, 4908–
4923.
(27) Schrauzer, G. Inorg. Synth. 1968, 11, 61.
(28) Lꢀopez, C.; Alvarez, S.; Aguilꢀo, M.; Solans, X.; Font-Altaba, M.
Inorg. Chem. Acta. 1987, 127, 153–159.
(29) Gerli, A.; Marzilli, L. Inorg. Chem. Acta. 1992, 31, 1152–1160.
(30) Tinker, L.; McDaniel, N.; Curtin, P.; Smith, C.; Ireland, M.;
Bernhard, S. Chem.—Eur. J. 2007, 13, 8726–8732.
(31) APEX2, version 2009.9-0; Bruker AXS: Madison, WI, 2009.
(32) Sheldrick, G. M. SADABS, version 2008/1; University of
G€ottingen: G€ottingen, Germany, 2008.
(33) SAINT, version 7.68A; Bruker AXS: Madison, WI, 2009.
(34) Altomare, A.; Burla, M.; Camalli, M.; Cascarano, G.; Giacovazzo,
C.; Guagliardi, A.; Moliterni, A.; Polidori, G.; Spagna, R. SIR97; CNR:
Bari, Italy, 1999.
(35) Sheldrick, G. Acta Crystallogr. 2008, A64, 112–122.
(36) Sj€oback, R.; Nygren, J.; Kubista, M. Spectrochim. Acta, Part A 1995,
51, L7–L21.
(37) Probst, B.; Guttentag, M.; Rodenberg, A.; Hamm, P.; Alberto,
R. Inorg. Chem. 2011, 50, 3404–3412.
(38) Math, N.; Naik, L.; Suresh, H.; Inamdar, S. J. Lumin. 2006, 121,
475–487.
(39) Simꢀandi, L. I.; Szeverꢀenyi, Z.; Budꢀo-Zꢀahony, E. Inorg. Nucl.
Chem. Lett. 1975, 11, 773.
(40) Wang, M.; Na, Y.; Gorlov, M.; Sun, L. Dalton Trans. 2009,
6458–6467.
(41) Munkholm, C.; Parkinson, D.; Walt, D. J. Am. Chem. Soc. 1990,
112, 2608–2612.
(42) Ma, H.; Wan, C.; Zewail, A. Proc. Natl. Acad. Sci. U.S.A. 2008,
105, 12754–12757.
(43) Parker, W.; Bresciani-Pahor, N.; Zangrando, E.; Randaccio, L.;
Marzilli, L. Inorg. Chem. 1986, 25, 1303–1309.
(44) Parker, W.; Bresciani-Pahor, N.; Zangrando, E.; Randaccio, L.;
Marzilli, L. Inorg. Chem. 1985, 24, 3908–3913.
10666
dx.doi.org/10.1021/ic2010166 |Inorg. Chem. 2011, 50, 10660–10666