Photochemical Formic Acid Dehydrogenation by Iridium Complexes: Understanding Mechanism and Overcoming Deactivation

Article abstract of DOI:10.1021/acscatal.5b01995

The mechanism of photochemical formic acid dehydrogenation catalyzed by [CpIr(bpy)(Cl)]⁺ (1, bpy = 2,2′-bipyridine) and [Cp Ir(bpy-OMe)(Cl)]⁺ (1-OMe, bpy-OMe = 4,4′-dimethoxy-2,2′-bipyridine) is examined. The catalysts operate with good turnover frequency (TOF) across an unusually wide pH range. Above pH 7, the evolved gas is >95% pure H₂ (along with traces of CO₂ but no detectable CO). Light-triggered H₂ release from a metal hydride intermediate is found to be the turnover-limiting step, based on the observed first-order dependence on catalyst concentration, saturation behavior in formate concentration, and direct in situ observation of a metal hydride resting state during turnover. Deactivation pathways are identified, including ligand loss and aggregate formation, precipitation of insoluble forms of the catalyst, and deprotonation of the iridium hydride intermediate. Guided by mechanistic insights, improved catalytic activity (initial TOF exceeding 50 h^-1), stability (>500 turnovers at nearly 5 atm), and selectivity (>95% H₂ gas) are achieved.

Full text of DOI:10.1021/acscatal.5b01995

Research Article
pubs.acs.org/acscatalysis
Photochemical Formic Acid Dehydrogenation by Iridium Complexes:
Understanding Mechanism and Overcoming Deactivation
Seth M. Barrett,‡ Samuel A. Slattery,‡ and Alexander J. M. Miller*
Department of Chemistry, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-3290, United States
S
* Supporting Information
ABSTRACT: The mechanism of photochemical formic acid dehydrogenation catalyzed by [Cp*Ir(bpy)(Cl)]+ (1, bpy = 2,2′-
bipyridine) and [Cp*Ir(bpy-OMe)(Cl)]+ (1-OMe, bpy-OMe = 4,4′-dimethoxy-2,2′-bipyridine) is examined. The catalysts
operate with good turnover frequency (TOF) across an unusually wide pH range. Above pH 7, the evolved gas is >95% pure H2
(along with traces of CO2 but no detectable CO). Light-triggered H2 release from a metal hydride intermediate is found to be the
turnover-limiting step, based on the observed first-order dependence on catalyst concentration, saturation behavior in formate
concentration, and direct in situ observation of a metal hydride resting state during turnover. Deactivation pathways are
identified, including ligand loss and aggregate formation, precipitation of insoluble forms of the catalyst, and deprotonation of the
iridium hydride intermediate. Guided by mechanistic insights, improved catalytic activity (initial TOF exceeding 50 h−1), stability
(>500 turnovers at nearly 5 atm), and selectivity (>95% H2 gas) are achieved.
KEYWORDS: hydrogen evolution, iridium, hydride, photocatalysis, hydrogen storage
would be accomplished by thermal hydrogenation of CO2 in
the dark, followed by light-triggered hydrogen release. In
pursuit of this eventual goal, and motivated by our interest in
photochemical H2 evolution,18,19 we considered the very rare
examples of photocatalytic formic acid dehydrogenation.
Building on two prior studies,20,21 Beller and co-workers
recently reported iron carbonyl catalysts that produced up to
126 turnovers of H2:CO2 (1:1 mixture) from formic acid in
DMF solvent at 60 °C under photolysis.22,23 In one of the only
other examples of this reaction, Ziessel and co-workers
demonstrated that Cp*Ir-based complexes produce H2:CO2
(1:1 mixture) from aqueous formate at pH 5, with a turnover
number (TON) up to 53 after 2 h of irradiation under
vacuum.24 Under these acidic conditions (and in organic
solvents), H2 and CO2 gas are coevolved in a 1:1 ratio.
INTRODUCTION
■
Carbon dioxide is a promising hydrogen storage medium,
capable of storing 4.3 wt % H2 in formic acid (HCO2H) and up
to 12 wt % H2 in subsequent hydrogenation products CH3OH
and H2O.1−5 The CO2/HCO2H couple (eq 1) is particularly
promising because the process is readily reversible, as evidenced
by a substantial body of work on both the CO2 hydro-
genation6−9 and the formic acid dehydrogenation steps.2,5,10−12
H2 + CO2 ⇆ HCO2H
(1)
A few catalysts can perform both CO2 hydrogenation and
HCO2H dehydrogenation.13−17 Reversible storage and release
is most commonly accomplished by applying high H2/CO2
pressures to promote hydrogenation, then releasing the
pressure to trigger dehydrogenation.13−16 Solution pH can
also be used to control the reaction, with hydrogenation
occurring at basic pH (where production of formate is
thermodynamically favorable) and dehydrogenation occurring
at acidic pH (where the thermochemistry shifts to favor H2/
CO2 release).17
Little is known about the mechanism of photochemical
formic acid dehydrogenation. For example, metal carbonyl-
catalyzed reactions may be initiated by photochemical CO
dissociation, or photochemical H2 release may occur, or light
An alternative strategy for reversible hydrogen storage would
utilize visible light to regulate the chemistry of eq 1. Storage
Received: September 8, 2015
© XXXX American Chemical Society
6320
DOI: 10.1021/acscatal.5b01995
ACS Catal. 2015, 5, 6320−6327

ACS Catalysis
Research Article
could play multiple roles simultaneously.25 Mechanistic
information on the Cp*Ir-based photocatalysis is limited to
an Arrhenius analysis,24 which gave activation energies higher
than the Cp*Ir-catalyzed photochemical water−gas shift
reaction.26 Photochemical formic acid dehydrogenation cata-
lysts are currently slower and less stable than corresponding
thermal catalysts.3,27,28 A deeper mechanistic understanding of
photochemical formic acid dehydrogenation would provide
guidance for improving these systems, which offer a promising,
mild photochemical alternative to thermal reactions controlled
by pressure or pH.
not been studied previously as a photocatalyst for formic acid
dehydrogenation, shows significantly improved photocatalytic
activity relative to 1 (Figure 1). As shown in Figure 1,
The mechanistic studies reported here reveal several new
aspects of Cp*Ir-photocatalyzed formic acid dehydrogenation
that helped guide significant improvements in stability, activity,
and selectivity. Kinetic studies and in situ reaction monitoring
indicate turnover-limiting photochemical H2 release under most
conditions, directing subsequent improvements by tuning the
light source wavelength and intensity. Several deactivation
pathways are also identified, guiding changes to the catalyst
structure and reaction conditions that yielded more active and
longer-lived catalysts. The catalysts operate efficiently in basic
conditions, where the evolved gas is comprised of >95% H2
because of efficient sequestration of CO2 by hydroxide. Our
mechanistic findings lead to substantial improvements in
performance, indicating that photochemical H2 release is a
promising strategy complementary to thermal reactions.
□
●
Figure 1. Gas evolution from formate solutions of 1 ( ), 1-OMe ( ),
○
and no catalyst ( ) during intermittent illumination (white back-
ground: lamp on, gray background: lamp off). Conditions: 0.32 mM Ir,
pH 10 1 M NaCO2H(aq), 18 mM phosphate, 296 K.
intermittent illumination (1 h cycles) confirmed the critical role
of visible light in promoting the reaction. In the absence of the
catalyst, only traces of gas evolved over 6 h, with no response to
light.
Both catalysts exhibit gradually diminishing activity during
prolonged illumination. While the initial yellow color of the
solutions was maintained for the first 24 h, prolonged
photolysis (beyond 48 h) in the eudiometer setup led to
formation of a dark, catalytically inactive precipitate. The
chloride complexes 1 and 1-OMe are both air-stable; however,
a ∼30% drop in activity was observed when the reactions were
carried out under air. In order to avoid this decomposition and
better understand the system, a series of mechanistic studies
were carried out.
RESULTS AND DISCUSSION
■
Initial Catalytic Studies. The air-stable and water-soluble
precatalysts 1 and 1-OMe were prepared according to
previously reported procedures.29,30 Initial catalytic runs were
carried out in 20 mL scintillation vials containing 0.36 mM
precatalyst in 10 mL of 1 M aqueous sodium formate,
illuminated by a 460 nm LED lamp at room temperature
(Scheme 1). The evolved gas was collected in an inverted buret
Scheme 1. Typical Conditions for Photochemical Formic
Acid Dehydrogenation by 1 and 1-OMe
pH Dependence. The pH dependence of the photo-
chemical reaction was examined first. The volume of evolved
gas dropped noticeably at higher pH, as shown in Figures S9
and S10. Rather than indicating a loss of activity, GC headspace
analysis revealed that the decline in evolved gas correlated with
a decrease in the amount of CO2 released during the reaction.
In fact, H2 comprised 96(3)% of the evolved gas above pH 7, as
shown in Figure 2a (the fuel cell poison CO was not detected).
After prolonged illumination, reaction mixtures with initial pH
between 6 and 10 were observed to approach a final pH of 8,
while more acidic solutions showed no significant pH change
during the reaction. This behavior is consistent with CO2
sequestration and engagement of carbonate equilibria above pH
6, as in eq 2 (an authentic sample of ∼30 mM HCO3− in 1 M
formate gave a pH 8 solution).
eudiometer and analyzed by headspace gas chromatography
(GC) in many cases (see Experimental Section and Supporting
Information for full experimental details).
HCO2 + H2O ⇆ H2 + HCO3−
When protected from light, solutions of 1 do not evolve
appreciable amounts of gas at room temperature. Under 460
nm illumination, however, a pH 5 formate solution containing
1 produced 1.2 mL of gas in 2 h, corresponding to a turnover
frequency (TOF) of ∼4 h−1 (slightly higher than Ziessel’s prior
report, TOF = 0.5 h−1).24 Although the prior report examined
only pH 5 conditions, we found that the catalyst is also active in
basic solutions: upon photolysis of a pH 10 formate solution of
1, gas evolution commenced immediately and continued
steadily. The methoxy-substituted catalyst 1-OMe, which had
−
(2)
When the volume of evolved gas is corrected based on the
H2:CO2 ratio of the evolved gas, it becomes clear that the rate
of H2 evolution is faster under more basic conditions (Figure
2b). Little H2 release activity is observed below pH 3, but the
rate increases sharply between pH 3 and 5 for both 1 and 1-
OMe. Both catalysts exhibit a roughly constant TOF between
pH 5 and 10, with 1-OMe operating roughly 3 times faster than
1 over this wide pH range. Catalyst 1 maintains good
6321
DOI: 10.1021/acscatal.5b01995
ACS Catal. 2015, 5, 6320−6327

ACS Catalysis
Research Article
443 nm LED lamp was applied instead, a higher initial TOF (20
h−1) was obtained and the pressure in the vessel approached 4
atm. The initial rates in the pressure vessel were similar to
initial rates observed by eudiometry (which maintains 1 atm H2
during the reaction), and pressures over 4 atm were required
before H2 pressure appeared to inhibit the reaction. This
pressure range is similar to a recently reported charge−
discharge process for hydrogen storage.13
Kinetic Analysis and in Situ Monitoring. Initial rates
methods (linear fits of first 2 h of reaction) were used to
determine the molecularity of the reaction. The rate of gas
evolution exhibits a first-order dependence on the concen-
tration of both catalysts 1 (Figure S11) and 1-OMe (Figure
3a). The reaction catalyzed by 1 exhibits saturation kinetics as a
●
Figure 2. (a) H2 content of evolved gas as a function of pH for 1 ( )
□
and 1-OMe ( ). (b) Initial TOF (H2 over 2 h) as a function of initial
pH for 1 ( ) and 1-OMe ( ). Conditions: 0.36 mM catalyst, 1 M
□
●
NaCO2H(aq), 296 K.
photocatalytic activity up to pH 10, and 1-OMe performs well
even at pH 11.
The broad region of pH independence is quite uncommon
for aqueous formic acid dehydrogenation catalysts. Thermal
aqueous formic acid dehydrogenation catalysts typically exhibit
a sharp maximum between pH 3 and 4, close to the pKa of
formic acid (3.75).17,31−33 The ability of the photocatalysts 1
and 1-OMe to maintain activity under alkaline conditions
enables the exciting observation of pure H2 release. Only a few
catalysts are capable of releasing a pure stream of H2 from
formic acid,1,3,13,14,34 and this is the first example of
photochemically driven production of pure H2.
Figure 3. (a) Gas volume measured from the eudiometer after 2 h of
irradiation with varying concentrations of precatalyst 1-OMe in pH 9 1
M formate. (b) Gas evolved after 2 h of 460 nm irradiation of 0.36
mM 1 ( ) and 0.36 mM 1-OMe ( ) with varying concentrations of
formate at pH 9.
The unusual ability to release H2 from alkaline solutions is
ascribed to the intermediate [Cp*Ir(bpy)(H)]+, a species that
becomes a stronger net hydride donor upon illumination.18,19
Given that most catalysts produce 1:1 H2:CO2 mixtures in the
“hydrogen release” step, photochemical methods offer a
complementary selectivity for high-purity hydrogen gas. Basic
conditions also may facilitate integration with hydrogen storage
□
●
function of formate concentration, with the initial rate
increasing at higher formate concentration before becoming
formate-independent above 4 M (Figure 3b).
−
reactions, because the best CO2 and HCO3 hydrogenation
catalysts thrive with added base,17,35 similar to the conditions in
which our catalysts are most active.
In contrast, an unexpected deactivation process was
encountered in 1-OMe-catalyzed reactions carried out at high
formate concentrations (pH 9). As shown in Figure 3b, the rate
increases initially with increasing formate concentration, but
slows and eventually loses activity at formate concentrations
around 6 M. The appearance of a yellow precipitate above 3 M
formate correlates with the decrease in activity of 1-OMe;
further investigations of this deactivation process are described
below.
Pressure Dependence. The dehydrogenation was carried
out in a closed system to determine if H2 inhibited the reaction.
A thick-walled Pyrex glass pressure vessel was charged with a
standard catalyst solution of 1-OMe (20 mL in a ∼ 40 mL
vessel), equipped with either a 4.1 or 10.9 atm pressure gauge,
and sealed under N2. Hydrogen evolved with a TOF of 10 h−1
under 460 nm illumination at pH 9. When a higher-powered
6322
DOI: 10.1021/acscatal.5b01995
ACS Catal. 2015, 5, 6320−6327

ACS Catalysis
Research Article
The pale yellow charge-transfer band of the chloride complex
1-OMe (λmax = 330 nm) provided a convenient handle for in
situ ultraviolet−visible (UV−vis) spectroscopy studies. Initial
studies were carried out protected from light to prevent
turnover. The reaction was monitored over time immediately
following injection of a stock solution of catalyst 1-OMe to a
cuvette containing a stirred solution of 1 M formate (pH 9).
The characteristic charge-transfer band of the hydride complex
[Cp*Ir(bpy-OMe)(H)]+ (2-OMe, λmax = 390 nm)18 grew in
over ∼2 min (Figure 4). The hydride complex is a proposed
dehydrogenation by [Cp*Ir(4,4-dihydroxy-bipyridine)
(OH2)]2+, which was also proposed to involve rate-determining
H2 evolution.37−39
Table 1 shows KIE values obtained when other mixtures of
isotopologues were examined. The observation that HCO2H/
D2O mixtures produced a normal, primary KIE of similar
magnitude to the DCO2D/D2O is inconsistent with rate-
limiting hydride formation: conversion of Ir−O2CH to Ir−H
should be sensitive only to isotopic substitution on formate,
and a primary KIE would be expected only when comparing
rates of formate with formate-d. On the other hand, the KIE
experiments are all consistent with a turnover-limiting process
involving H−H bond formation.
Lamp Intensity, Wavelength Dependence, and Quan-
tum Yields. A rate-limiting photochemical process could be
limited by the kinetics of the photochemical reaction or by the
light-absorbing characteristics. To further probe this important
step, the influence of light intensity was examined. A 443 nm
LED lamp with programmable output intensity was used to
perform formic acid dehydrogenation between 2 and 62 mW·
cm−2 power density. The initial TOF increased from 3 h−1 at
low intensity to 36 h−1 at high intensity (Figure S20), again
consistent with photochemical H2 evolution limiting the rate of
catalysis. A slight deviation from linearity was observed at high
intensity, which might indicate that sufficient photon flux was
present that other steps were starting to contribute to the rate.
To probe the role of overlap between the light source and
the absorbing molecules, photocatalysis was performed under
406, 443, and 503 nm illumination at a constant, relatively low
power density (∼13 mW·cm−2) in a closed system (Figure
S19). Illumination with either the 406 or 443 nm lamp gave
similar activity (10 h−1 for 1-OMe). Under 503 nm light, the
rate dropped to 5 h−1 for 1-OMe. The chloride complex 1-
OMe does not absorb as much low-energy light, nor with as
much intensity as 2-OMe, so the ability to operate with >400
nm light suggests that 2-OMe is the active light absorber.18 The
performance trends correlate well with the predicted number of
photons absorbed by hydride 2-OMe; calculations suggest that
2-OMe would absorb a similar number of 406 and 443 nm
photons, but about 4 times fewer 503 nm photons (see
derivation in Supporting Information).
The conventional TON and TOF metrics of catalysis provide
a convenient point of comparison with thermal catalysts, but
photochemical reactions can also be assessed based on the
photon-to-chemical efficiency. The quantum yield, defined as
the moles of H2 produced divided by the moles of photons
absorbed by the catalyst, was determined under several
conditions. Under typical conditions of 443 nm illumination
at 13 mW·cm−2 intensity, Φ = 0.05 is obtained for formic acid
dehydrogenation by 1-OMe. As the 443 nm lamp intensity was
increased, the quantum yield decreases slightly, with Φ = 0.03
at 62 mW·cm−2. As the irradiation wavelength was changed,
only slight variations in quantum yield are observed, with
Φ406 nm = 0.05 and Φ503 nm = 0.10 (Table S2). With relatively
high photon efficiency maintained over the course of 2 h,24,26
and under a variety of illumination conditions, catalyst 1-OMe
has encouraging photochemical behavior.
Figure 4. Absorbance profile after injection of 0.36 mM 1-OMe into
pH 9 1 M NaCO2H(aq) at 296 K. The growth of 2-OMe was
observed in 20 s intervals for 1 min in the dark, followed by
illumination. The inset follows the hydride absorbance at 390 nm; after
1 min in the dark, illumination with 460 nm light continued for 60
min.
intermediate in thermal formic acid dehydrogenation by Cp*Ir-
based catalysts.31,36,37 Formation of hydride 2-OMe occurs
much faster than the observed rate of catalysis, suggesting that
these dark processes are not rate-limiting.
Photocatalytic dehydrogenation was then spectroscopically
monitored under continuous illumination to elucidate the
resting state during catalysis. The cuvette was attached to a
eudiometer to confirm that H2 was indeed released during
monitoring. The charge-transfer band at 390 nm remained
prominent during active catalysis (Figure 4), identifying
hydride 2-OMe as the resting state and indicating turnover-
limiting photochemical H2 release.
A significant kinetic isotope effect (KIE) is anticipated if H2
evolution is indeed the turnover-limiting process. The initial
rate of gas evolution from HCO2H/H2O and DCO2D/D2O
was determined in two separate experiments, providing a
normal, primary KIE of 2.6(3), as shown in Table 1. This KIE is
similar to the value of 2.6(1) reported for thermal formic acid
Table 1. Kinetic Isotope Effects of Photochemical
a
Dehydrogenation
formic acid isotopologue
solvent
kinetic isotope effect
HCO2H
DCO2D
HCO2H
DCO2D
H2O
D2O
D2O
H2O
−
2.6(3)
Mechanistic Proposal. A plausible catalytic mechanism
that is consistent with the preceding mechanistic studies is
depicted in Scheme 2. Entry to the catalytic cycle is afforded by
the chloride precatalysts 1 and 1-OMe. Previous studies have
shown that [Cp*Ir(bpy)(Cl)][Cl] undergoes facile ligand
exchange in water18 and is readily converted to the hydride
2.4
1.8
a
Conditions: 0.36 mM 1-OMe, 1 M formate, pH 8, 460 nm
illumination for 2 h at 296 K. The formic acid OH proton is assumed
to exchange with water protons.
6323
DOI: 10.1021/acscatal.5b01995
ACS Catal. 2015, 5, 6320−6327

ACS Catalysis
Research Article
improved reaction rates are possible by supplying a higher
photon flux (see above), which can be accomplished by
changing the photochemical reactor design or by employing
more powerful lamps.43−45
Scheme 2. Proposed Mechanism of Photochemical Formic
Acid Dehydrogenation
The precise mechanism of photochemical H2 release
warrants further discussion.46,47 The primary event is likely
excitation to the previously observed triplet excited state of
2.26,48,49 Once formed, the triplet excited state has a potent
thermodynamic driving force for hydride transfer: the excited
state is at least 20 kcal·mol−1 more hydridic than the ground
state in acetonitrile.19 While the kinetic details of excited-state
H2 release from 2 remain the subject of ongoing studies in our
laboratory, the photohydride reactivity of 2 clearly enables
some of the unique features of Cp*Ir-based photochemical
formic acid dehydrogenation. Light absorption renders the M−
H fragment more hydridic, and therefore more reactive toward
weak acids (H2O in this case), facilitating H2 release under
strongly basic conditions. Photoelectrochemical H2 evolution
proceeds efficiently until pH 10,18 in contrast to the acidic
conditions (pH ∼ 4) required by most aqueous formic acid
dehydrogenation catalysts. Photochemically enabling H2 release
above pH 7 in turn facilitates the release of pure H2 gas with
concomitant trapping of CO2 by hydroxide. The rarity of pure
H2 release likely stems from the fact that basic conditions are at
odds with the fundamental organometallic chemistry under-
lying H2 release: the metal hydride intermediates typically
require acidic conditions to rapidly release H2.
Typical thermal H2 release pathways slow rapidly as the
solution pH increases, as indicated by the narrow, acidic pH
range at which most aqueous catalysts operate.17,31,32 In
contrast, catalysts 1 and 1-OMe operate with similar TOF
over a wide pH range. Although pH does not greatly impact the
reaction rate, it does impose several restrictions based on the
pKa values of various species. The pKa of formic acid (3.75)
dictates the availability of formate anion and therefore the
efficiency of the substitution process; under more acidic
conditions formic acid is present and undergoes substitution
more slowly, as observed in related systems that exhibit
maximal performance near pH 4.31,32 The pKa of the metal
hydride limits the operational pH range in the basic regime. A
pKa of ∼10 is estimated for hydride 2, based on changes in
electrochemical behavior,18 at which point deprotonation forms
an off-cycle Ir(II)-bpy•− complex (3, Scheme 2).49 When
hydride 1 was stirred at pH 13, a purple precipitate formed;
extraction into benzene confirmed that the precipitate was the
deprotonation product 3.49
Deactivation Pathways. Understanding deactivation and/
or decomposition processes of homogeneous catalysts can help
identify molecular weaknesses and guide catalyst redesign.50,51
After 72 h illumination of 1 in pH 8 or pH 10 formate solutions
while collecting gases by eudiometry, formation of a dark blue
solution was observed. A UV−vis spectrum after the reaction
revealed the growth of a broad new feature with λmax = 750 nm
that is consistent with nanoparticle formation.40,52 Nano-
particles could result from photochemical or thermal bpy
dissociation, forming the photocatalytically inactive tris-
hydroxide dimer 4 (Scheme 2),41 which could go on to form
insoluble aggregates or nanoparticles.
[Cp*Ir(bpy)(H)]+ in aqueous formate.40 Ligand substitution
processes are not turnover-limiting, as evidenced by the similar
rates of dehydrogenation by chloride 1 and aquo [Cp*Ir(bpy)-
(H2O)]2+ (TOF ∼ 4 h−1 for each catalyst). Under basic
conditions hydroxide substitution would be expected, based
upon the reported pKa value for aquo [Cp*Ir(bpy)-
(H2O)]2+.29,41,42 Indeed, stirring either chloride 1 or aquo
[Cp*Ir(bpy)(H2O)]2+ in a pH 10 solution leads to the
formation of the same product by NMR spectroscopy, assigned
as the hydroxide complex [Cp*Ir(bpy)(OH)]+. Our in situ
observations of rapid conversion of chloride 1-OMe to hydride
2-OMe in formate solutions in the dark suggest that formate
binding and decarboxylation occur rapidly and without
photochemical assistance.
Photochemical H2 release (shown in blue in Scheme 2) has
been identified as the turnover-limiting process at high
concentrations of formate. The first-order dependence on Ir
and saturation behavior with respect to formate concentration
suggest that at lower formate concentrations a pre-equilibrium
step involving formate (such as reversible formation of the
formate complex) is also involved in the rate law. Observation
of the hydride 2-OMe as the resting state during photocatalysis
strengthens this conclusion. Finding the reaction to be photon-
limited holds promise for future catalyst development:
To prevent ligand loss, the reaction was run in the presence
of free 2,2′-bipyridine. In parallel eudiometry experiments
monitoring the ability of 1 to dehydrogenate 1 M formate
solutions at pH 9, the catalyst produced more H2 in the
presence of 3 equiv free ligand (Figure S22). Interestingly,
6324
DOI: 10.1021/acscatal.5b01995
ACS Catal. 2015, 5, 6320−6327

ACS Catalysis
Research Article
analogous experiments in the sealed medium-pressure vessel
showed little evidence of aggregation, suggesting that slow
leaking of air into the eudiometer setup may also be partially
responsible for nanoparticle formation. The more strongly
donating methoxy-substituted bipyridine ligand in 1-OMe may
help avoid ligand loss during catalysis, as 1-OMe maintains its
initial rate for longer.
EXPERIMENTAL SECTION
■
General Considerations. Manipulations were performed
under the inert nitrogen atmosphere of a Schlenk line or a
glovebox. Water was thoroughly degassed by sparging with
nitrogen before use. D2O, DCO2D, NaOD(aq), and DCl(aq)
were purchased from Cambridge Isotopes Laboratories, Inc.
[Cp*Ir(Cl)2]2,53 [Cp*Ir(bpy)(Cl)][Cl] (1),29 and [Cp*Ir-
(bpy-OMe)(Cl)][Cl] (1-OMe)30 were synthesized following
established procedures. All other materials were readily
commercially available and were used as received. UV−vis
spectra were obtained with an Ocean Optics USB2000+
spectrometer with a DT-MINI-2GS deuterium/tungsten
A deactivation process specific to 1-OMe is encountered at
high formate concentrations, as discussed above in the context
of formate dependence from Figure 3b. To probe this process
further, 1-OMe was stirred in 8 M formate solutions protected
from light. A yellow precipitate formed in the dark, ruling out
photochemical degradation. When the solids were assessed by
attenuated total reflection infrared (ATR-IR) spectroscopy, a
prominent Ir−H stretch was observed at 2046 cm−1, closely
matching the previously reported stretch for the hexafluor-
ophosphate salt.42 The solids dissolved readily in CD3CN, and
NMR and electrospray ionization mass spectrometry (ESI-MS)
analyses confirm the presence of hydride 2-OMe as well as
some [Cp*Ir(bpy-OMe)(O2CH)]+. Taken together, the data
suggest that the formate salt of 2-OMe, [Cp*Ir(bpy-OMe)-
(H)][O2CH], is less soluble than the chloride salt, leading to
precipitation. The methoxy substituents apparently provide
enough lipophilicity that the formate salt of 2-OMe is
substantially less water-soluble than the formate salt of 2. It is
noteworthy that activity can be restored simply by diluting the
solution below 6 M formate.
1
halogen light source controlled by OceanView software. H
NMR spectra were recorded on 400 or 600 MHz spectrometers
at 25 °C. Chemical shifts are reported with respect to residual
proteo solvent.54 ESI-MS data was acquired on a Thermo
Scientific LTQ-FT-ICR Mass Spectrometer. Infrared spectros-
copy was carried out with a Bruker Alpha FT-IR equipped with
an ATR module.
Visible light photolysis was carried out using either an
EagleLight 460 nm ( 12 nm at half-max intensity) LED lamp
(500 lumens at 15 W; ∼ 2 mW·cm−2 in typical experimental
conditions) or a ThorLabs multiwavelength LED source
containing 406
10 nm, 443
9 nm, and 503
13 nm
LED lamps controlled by a ThorLabs LED driver for tunable
power density. A Coherent FieldMaxII Laser Power/Energy
Meter (photodiode) was used for electronic photon flux
measurements (to determine the number of incident photons
on a sample). The quantum yield of the reaction was
determined based on the moles of H2 produced divided by
the moles of absorbed photons after 2 h of irradiation.
Measurements of pH were obtained with either a Fisher
AccupHast pH probe, a Hanna pH probe, or a Hach ISFET pH
probe.
Improved Catalytic Conditions. Guided by our mecha-
nistic findings, catalytic activity was reassessed in a pressure
vessel under conditions designed to minimize decomposition
while maintaining optimal activity. When a 3 M aqueous
formate solution at pH 8 was treated with 0.39 mM catalyst 1 at
296 K and illuminated with a 443 nm lamp, the initial TOF
exceeds 12 h−1 and TON > 360 is achieved. Under analogous
conditions with 0.37 mM 1-OMe, the initial TOF exceeds 50
Photocatalytic Formic Acid Dehydrogenation Proce-
dures. General Procedure A: Eudiometry. Reactions were
carried out in 20 mL scintillation vials containing a stir bar, 10
mL of pH-adjusted aqueous formate, and the desired
precatalyst [Cp*Ir(bpy)(Cl)][Cl] (1) or [Cp*Ir(bpy-OMe)-
(Cl)][Cl] (1-OMe). Formate solutions were made by
dissolving the appropriate amount of NaCO2H in HPLC
grade water; pH adjustments were made by addition of dilute
HCl(aq) and/or KOH(aq). In a typical reaction, an aqueous
precatalyst solution containing 3.6 μmol 1 or 1-OMe was
delivered to a scintillation vial containing 10 mL of pH-adjusted
formate via micropipette. Specialized vial caps were constructed
using septum caps, PTFE tubing, needles, parafilm, and grease
(Figure S1). Caps were sealed on each vial with grease and
parafilm for an airtight seal. Once capped and sealed, the
reaction vial was wrapped in aluminum foil to shield the
reaction from light. A large recrystallization dish was then filled
halfway with water, and a graduated centrifuge tube, filled with
water, was inverted in the water of the recrystallization dish.
The PTFE tubing attached to the reaction vial was placed into
the inverted centrifuge tube, completing the eudiometer setup
(Figure S1). All reaction vials were then degassed with N2 from
a Schlenk line by inserting a needle through the septum cap to
the bottom of the solution. After degassing for at least 15 min,
the eudiometer tube was once again filled with water and any
trace N2 gas left in the eudiometer (typically <0.1 mL) was
recorded and subtracted from subsequent data. After the
aluminum foil was removed, the samples were illuminated with
h
−1 and TON > 500 is achieved over 30 h illumination. After 30
h, nearly 5 atm had built up in the sealed vessel. In addition to
outperforming other photocatalysts that we are aware of,
photocatalyst 1-OMe also produces H2 in 96% purity according
to GC analysis.
CONCLUSIONS
■
Mechanistic studies of Cp*Ir-based photochemical formic acid
dehydrogenation catalysts have led to a detailed understanding
of the active catalyst cycle as well as decomposition and
deactivation pathways. Direct spectroscopic evidence for every
proposed intermediate was accumulated. Kinetic evidence
identified the turnover-limiting process as photochemical H2
release, via the photohydride [Cp*Ir(bpy)(H)]+, which is
capable of producing H2 beyond pH 10 and consequently
enables the release of pure H2 from formic acid in basic water.
CO2 is trapped in the alkaline solution, and no CO is produced,
leading to headspace gas compositions of 96(3)% H2. The H2
levels are remarkably impervious to the makeup of the catalytic
system, as H2 levels above 90% were obtained regardless of
catalyst loading and formate concentration (Figure S7). Guided
by these mechanistic findings, improved visible light photo-
catalysis was realized at room temperature, even at elevated
pressures of H2. These findings offer a promising outlook for
photochemical H2 release as a strategy for hydrogen storage
involving the CO2/HCO2H couple.
6325
DOI: 10.1021/acscatal.5b01995
ACS Catal. 2015, 5, 6320−6327

ACS Catalysis
Research Article
a 460
12 nm LED lamp with constant stirring from a
more sensitive to CO2, with even small amounts leading to
substantial response. The resulting calibration curves for H2
(Figure S3) and CO2 (Figure S4) were validated by comparison
with an authentic sample of 50/50 H2/CO2, purchased from
Airgas (certificate of analysis: 51 2% H2). The commercial
50/50 H2/CO2 sample was 48 2% H2 and 52 4% CO2
according to the calibration curves.
magnetic stir bar. The gas evolved from these reactions was
collected, via the PTFE tubing, in the inverted-buret style
eudiometer suspended in water, and the volume of evolved gas
was recorded for the duration of the experiment.
A representative photocatalytic dehydrogenation procedure,
following General Procedure A, is included for completeness. In
a beaker, 1.701 g (0.025 mol) of NaCO2H was dissolved in
HPLC grade water; a volumetric flask was used to dilute the
total solution volume to 25 mL, resulting in a 1.0 M formate
solution. This solution was transferred to a beaker equipped
with a stir bar. The pH was adjusted by addition of small
volumes of KOH(aq). A 20 mL scintillation vial was then filled
with 10 mL of the pH-adjusted solution using a micropipette. A
second scintillation vial was prepared the same way, and stir
bars were added to both vials. A 76.9 μL aliquot of a 0.042 M
stock solution of 1-OMe (3.25 μmol) was added to the first vial
with a micropipette. The second vial was used as a control and
contained no precatalyst. The specialized caps were then put on
the two vials, and the vials were sealed with grease and parafilm
around the cap−vial junction. Aluminum foil was wrapped
around each vial to shield from light. A long needle connected
to a Schlenk line was then inserted into the solution through
the septum cap. The PTFE tubing was immersed into the
inverted centrifuge vial. N2 was then bubbled through the
solution from the Schlenk line for at least 15 min. The inverted
centrifuge vial was then refilled with water, while the end of the
PTFE tubing was kept submerged. The tubing was then
reinserted in the centrifuge vial. Under constant stirring, the
aluminum foil was removed from the vials and the 460 nm LED
lamp was turned on. Volume measurements were made
periodically by inspection of the water level in the inverted
centrifuge vial and recording the volume of collected gas.
General Procedure B: Pressure Vessel. Reactions were
performed in ∼40 mL Pyrex glass pressure vessels equipped
with a stir bar and capped with a pressure gauge (60 or 160
psig). Each vessel was charged with 20 mL of 1 M formate at
pH 8−9, prepared according to the general procedure above. In
a typical experiment, 100 mL of 1 M formate was prepared,
transferred to a small bomb, and degassed for 20−30 min prior
to being pumped into an N2-filled glovebox. Next, 4.5−5.2 mg
(7.32−8.46 μmol) 1-OMe was weighed into a vial, and a
micropipette was used to deliver 20 mL of the 1 M formate
solution to the vial. The dissolved catalyst was then added to
the pressure vessel along with a stir bar. The vessel was then
capped with the pressure gauge to create an air-free, closed
system. The vessel was then removed from the glovebox and
placed in a photolysis hood. Under constant stirring, the vessel
was monitored in the dark for ∼10 min prior to irradiation.
After 10 min in the dark (allowing for equilibration in any
temperature change between the glovebox and photolysis
hood), the vessel was irradiated with LED light. At periodic
intervals, the light was turned off, the pressure gauge was
checked, and the pressure buildup was recorded.
Quantification of Catalytic Activity. Turnover number
(TON) is defined as the moles of H2 produced divided by the
moles of catalyst present in the reaction. Based on the H2
composition of the evolved gas determined by GC, the moles
of H2 were calculated from the total volume of gas evolved.
Turnover frequency (TOF) is calculated by dividing the TON
by the time of reaction (the irradiation time). TOF values are
typically calculated as initial rates, using the first 2 h of the
reaction, over which time the TOF was constant. Many
experiments were run three times to ensure consistency and
minimize experimental error. Error bars represent the standard
deviation when two or more experiments were performed.
When only one experiment was performed, uncertainty of 3%
was assigned based on typical run-to-run variation and the GC
instrument response variation.
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acscatal.5b01995.
Additional experimental details, photographs of reaction
setups, and NMR spectra (PDF)
AUTHOR INFORMATION
Corresponding Author
*E-mail: ajmm@email.unc.edu.
■
Author Contributions
‡S.M.B. and S.A.S.: These authors contributed equally.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We gratefully acknowledge funding from the University of
North Carolina at Chapel Hill for the University Research
Council’s James Moeser Award and a J. Thurman Frieze
Research Fellowship (S.A.S.) and from the NSF (Graduate
Research Fellowship to S.M.B.; DGE-1144081). This research
made use of an IR spectrometer funded by the UNC EFRC:
Center for Solar Fuels, an Energy Frontier Research Center
supported by the U.S. Department of Energy, Office of Science,
Office of Basic Energy Sciences, under Award DE-SC0001011.
REFERENCES
■
(1) Zaidman, B.; Wiener, H.; Sasson, Y. Int. J. Hydrogen Energy 1986,
11, 341−347.
(2) Fukuzumi, S.; Suenobu, T. Dalton Trans. 2013, 42, 18−28.
(3) Grasemann, M.; Laurenczy, G. Energy Environ. Sci. 2012, 5,
8171−8181.
GC Analysis. H2 quantification was performed using a
Varian 450-GC with a pulsed discharge helium ionizer detector.
A 1.0 mL Vici Pressure-Lok Precision Analytical gastight
syringe was used to sample the headspace of the reaction vials
by puncturing the septum cap. Calibration curves were
constructed independently for H2 and CO2. Mixed gas samples
for calibration were prepared by performing gas dilutions in
sealed round-bottom flasks. Data for each gas mixture was
collected at least three times. The instrument is considerably
(4) Enthaler, S.; von Langermann, J.; Schmidt, T. Energy Environ. Sci.
2010, 3, 1207−1217.
(5) Johnson, T. C.; Morris, D. J.; Wills, M. Chem. Soc. Rev. 2009, 39,
81−88.
(6) Appel, A. M.; Bercaw, J. E.; Bocarsly, A. B.; Dobbek, H.; DuBois,
D. L.; Dupuis, M.; Ferry, J. G.; Fujita, E.; Hille, R.; Kenis, P. J. A.;
Kerfeld, C. A.; Morris, R. H.; Peden, C. H. F.; Portis, A. R.; Ragsdale,
6326
DOI: 10.1021/acscatal.5b01995
ACS Catal. 2015, 5, 6320−6327

ACS Catalysis
Research Article
S. W.; Rauchfuss, T. B.; Reek, J. N. H.; Seefeldt, L. C.; Thauer, R. K.;
Waldrop, G. L. Chem. Rev. 2013, 113, 6621−6658.
(7) Wang, W.; Wang, S.; Ma, X.; Gong, J. Chem. Soc. Rev. 2011, 40,
3703−3727.
(8) Jessop, P. G.; Joo,
2425−2442.
(9) Jessop, P. G.; Ikariya, T.; Noyori, R. Chem. Rev. 1995, 95, 259−
272.
(39) Wang, W.-H.; Ertem, M. Z.; Xu, S.; Onishi, N.; Manaka, Y.;
Suna, Y.; Kambayashi, H.; Muckerman, J. T.; Fujita, E.; Himeda, Y.
ACS Catal. 2015, 5, 5496−5504.
(40) Miller, A. J. M.; Heinekey, D. M.; Mayer, J. M.; Goldberg, K. I.
Angew. Chem., Int. Ed. 2013, 52, 3981−3984.
(41) Makihara, N.; Ogo, S.; Watanabe, Y. Organometallics 2001, 20,
497−500.
́
F.; Tai, C.-C. Coord. Chem. Rev. 2004, 248,
(42) Ogo, S.; Kabe, R.; Hayashi, H.; Harada, R.; Fukuzumi, S. Dalton
Trans. 2006, 4657−4663.
(10) Fukuzumi, S. Eur. J. Inorg. Chem. 2008, 2008, 1351−1362.
(11) Fukuzumi, S.; Yamada, Y.; Suenobu, T.; Ohkubo, K.; Kotani, H.
Energy Environ. Sci. 2011, 4, 2754−2766.
(43) Garlets, Z. J.; Nguyen, J. D.; Stephenson, C. R. J. Isr. J. Chem.
2014, 54, 351−360.
(44) Andrews, R. S.; Becker, J. J.; Gagne,
2012, 51, 4140−4143.
́
M. R. Angew. Chem., Int. Ed.
(12) Fujita, E.; Muckerman, J. T.; Himeda, Y. Biochim. Biophys. Acta,
Bioenerg. 2013, 1827, 1031−1038.
(45) Elliott, L. D.; Knowles, J. P.; Koovits, P. J.; Maskill, K. G.; Ralph,
M. J.; Lejeune, G.; Edwards, L. J.; Robinson, R. I.; Clemens, I. R.; Cox,
B.; Pascoe, D. D.; Koch, G.; Eberle, M.; Berry, M. B.; Booker-Milburn,
K. I. Chem. - Eur. J. 2014, 20, 15226−15232.
́
(13) Papp, G.; Csorba, J.; Laurenczy, G.; Joo, F. Angew. Chem., Int.
Ed. 2011, 50, 10433−10435.
(14) Boddien, A.; Gartner, F.; Federsel, C.; Sponholz, P.; Mellmann,
̈
D.; Jackstell, R.; Junge, H.; Beller, M. Angew. Chem., Int. Ed. 2011, 50,
6411−6414.
(46) Esswein, A. J.; Nocera, D. G. Chem. Rev. 2007, 107, 4022−4047.
(47) Teets, T. S.; Nocera, D. G. Chem. Commun. 2011, 47, 9268−
9274.
(15) Boddien, A.; Federsel, C.; Sponholz, P.; Mellmann, D.; Jackstell,
R.; Junge, H.; Laurenczy, G.; Beller, M. Energy Environ. Sci. 2012, 5,
8907−8911.
(48) Suenobu, T.; Guldi, D. M.; Ogo, S.; Fukuzumi, S. Angew. Chem.,
Int. Ed. 2003, 42, 5492−5495.
(16) Hsu, S.-F.; Rommel, S.; Eversfield, P.; Muller, K.; Klemm, E.;
Thiel, W. R.; Plietker, B. Angew. Chem., Int. Ed. 2014, 53, 7074−7078.
(17) Hull, J. F.; Himeda, Y.; Wang, W.-H.; Hashiguchi, B.; Periana,
R.; Szalda, D. J.; Muckerman, J. T.; Fujita, E. Nat. Chem. 2012, 4, 383−
388.
(49) Ladwig, M.; Kaim, W. J. Organomet. Chem. 1992, 439, 79−90.
(50) Widegren, J. A.; Finke, R. G. J. Mol. Catal. A: Chem. 2003, 198,
317−341.
(51) Crabtree, R. H. Chem. Rev. 2012, 112, 1536−1554.
(52) Hoertz, P. G.; Kim, Y.-I.; Youngblood, W. J.; Mallouk, T. E. J.
Phys. Chem. B 2007, 111, 6845−6856.
(18) Pitman, C. L.; Miller, A. J. M. ACS Catal. 2014, 4, 2727−2733.
(19) Barrett, S. M.; Pitman, C. L.; Walden, A. G.; Miller, A. J. M. J.
Am. Chem. Soc. 2014, 136, 14718−14721.
(53) White, C.; Yates, A.; Maitlis, P. M.; Heinekey, D. M. (η5-
Pentamethylcyclopentadienyl)Rhodium and -Iridium Compounds. In
Inorganic Syntheses; Grimes, R. N., Ed.; John Wiley & Sons, Inc.:
Hoboken, NJ, 1992; Vol. 29, pp 228−234.
(20) Onishi, M. J. Mol. Catal. 1993, 80, 145−149.
(21) Linn, D. E.; King, R. B.; King, A. D. J. Mol. Catal. 1993, 80,
151−163.
(54) Fulmer, G. R.; Miller, A. J. M.; Sherden, N. H.; Gottlieb, H. E.;
Nudelman, A.; Stoltz, B. M.; Bercaw, J. E.; Goldberg, K. I.
Organometallics 2010, 29, 2176−2179.
(22) Boddien, A.; Loges, B.; Gartner, F.; Torborg, C.; Fumino, K.;
̈
Junge, H.; Ludwig, R.; Beller, M. J. Am. Chem. Soc. 2010, 132, 8924−
8934.
(23) Loges, B.; Boddien, A.; Junge, H.; Noyes, J. R.; Baumann, W.;
Beller, M. Chem. Commun. 2009, 28, 4185−4187.
(24) Watson, K. J.; Ziessel, R. Inorg. Chim. Acta 1992, 197, 125−127.
(25) Loges, B.; Boddien, A.; Junge, H.; Beller, M. Angew. Chem., Int.
Ed. 2008, 47, 3962−3965.
(26) Ziessel, R. J. Am. Chem. Soc. 1993, 115, 118−127.
(27) Boddien, A.; Mellmann, D.; Gartner, F.; Jackstell, R.; Junge, H.;
Dyson, P. J.; Laurenczy, G.; Ludwig, R.; Beller, M. Science 2011, 333,
1733−1736.
(28) Bielinski, E. A.; Lagaditis, P. O.; Zhang, Y.; Mercado, B. Q.;
Wurtele, C.; Bernskoetter, W. H.; Hazari, N.; Schneider, S. J. Am.
̈
Chem. Soc. 2014, 136, 10234−10237.
(29) Dadci, L.; Elias, H.; Frey, U.; Hoernig, A.; Koelle, U.; Merbach,
A. E.; Paulus, H.; Schneider, J. S. Inorg. Chem. 1995, 34, 306−315.
(30) Himeda, Y.; Onozawa-Komatsuzaki, N.; Sugihara, H.; Kasuga,
K. Organometallics 2007, 26, 702−712.
(31) Himeda, Y. Green Chem. 2009, 11, 2018−2022.
(32) Fukuzumi, S.; Kobayashi, T.; Suenobu, T. J. Am. Chem. Soc.
2010, 132, 1496−1497.
(33) Fukuzumi, S.; Kobayashi, T.; Suenobu, T. ChemSusChem 2008,
1, 827−834.
(34) Wiener, H.; Sasson, Y.; Blum, J. J. Mol. Catal. 1986, 35, 277−
284.
(35) Tanaka, R.; Yamashita, M.; Nozaki, K. J. Am. Chem. Soc. 2009,
131, 14168−14169.
(36) Himeda, Y.; Onozawa-Komatsuzaki, N.; Miyazawa, S.; Sugihara,
H.; Hirose, T.; Kasuga, K. Chem. - Eur. J. 2008, 14, 11076−11081.
(37) Wang, W.-H.; Xu, S.; Manaka, Y.; Suna, Y.; Kambayashi, H.;
Muckerman, J. T.; Fujita, E.; Himeda, Y. ChemSusChem 2014, 7,
1976−1983.
(38) Wang, W.-H.; Hull, J. F.; Muckerman, J. T.; Fujita, E.; Hirose,
T.; Himeda, Y. Chem. - Eur. J. 2012, 18, 9397−9404.
6327
DOI: 10.1021/acscatal.5b01995
ACS Catal. 2015, 5, 6320−6327