Iridium, rhodium, and ruthenium catalysts for the "aldehyde-water shift" reaction

Article abstract of DOI:10.1021/cs500843a

A series of half-sandwich complexes of iridium, rhodium, and ruthenium are shown to be active catalysts for the conversion of aldehydes and water to carboxylic acids. Depending on the catalyst, H2 is either released (the "aldehyde-water shift") or transferred to a second equivalent of aldehyde (aldehyde disproportionation). Mechanistic studies suggest hydride transfer to be the selectivity-determining step along the reaction pathway. Using [(p-cymene)Ru(bpy)OH₂][OTf]₂as precatalyst, we demonstrate a novel example of a highly selective aldehyde-water shift in the absence of a hydrogen acceptor or base. (Chemical Equation Presented).

Full text of DOI:10.1021/cs500843a

Letter
pubs.acs.org/acscatalysis
Iridium, Rhodium, and Ruthenium Catalysts for the “Aldehyde−
Water Shift” Reaction
Timothy P. Brewster,§ William C. Ou,‡ Jeremy C. Tran,§ Karen I. Goldberg,§ Susan K. Hanson,†
Thomas R. Cundari,*,‡ and D. Michael Heinekey*,§
§Department of Chemistry, University of Washington, Seattle, Washington 98195-1700, United States
†Division of Chemistry, Los Alamos National Laboratory, MS J572, Los Alamos, New Mexico 87545, United States
‡Department of Chemistry, Center for Advanced Scientific Computing and Modeling (CASCaM), University of North Texas, P.O.
Box 305070, Denton, Texas 76203-5070, United States
S
* Supporting Information
ABSTRACT: A series of half-sandwich complexes of iridium,
rhodium, and ruthenium are shown to be active catalysts for the
conversion of aldehydes and water to carboxylic acids. Depending on
the catalyst, H2 is either released (the “aldehyde−water shift”) or
transferred to a second equivalent of aldehyde (aldehyde
disproportionation). Mechanistic studies suggest hydride transfer
to be the selectivity-determining step along the reaction pathway.
Using [(p-cymene)Ru(bpy)OH2][OTf]2 as precatalyst, we demon-
strate a novel example of a highly selective aldehyde−water shift in the absence of a hydrogen acceptor or base.
KEYWORDS: aldehyde oxidation, homogeneous catalysis, dehydrogenation, disproportionation, water
arboxylic acids are important organic compounds, utilized
in a wide variety of consumer products, and serve as
serves as the terminal oxidant. Utilizing this strategy, aldehydes
could be oxidized in an atom-economical fashion under mild
conditions while also liberating hydrogen gas as a valuable
byproduct.
The AWS reaction was first demonstrated in 1987 by
Murahashi using a ruthenium catalyst (Scheme 3) in the
C
synthetic precursors to esters, amides, and polymers.
Production of carboxylic acids from aldehydes can be achieved
with a wide variety of oxidants, such as permanganate,
hydrogen peroxide, and oxygen.1 Aldehyde disproportionation,
in which the aldehyde itself serves as the oxidant, is also well
established. In the Cannizzaro reaction, an aldehyde bearing no
α-hydrogens undergoes disproportionation under basic con-
ditions to afford a carboxylate salt and a primary alcohol
(Scheme 1).1 Catalysis of aldehyde disproportionation under
Scheme 3. AWS Reported by Murahashi5
Scheme 1. Cannizzaro Reaction
presence of a hydrogen acceptor as part of a larger study of the
conversion of alcohols and aldehydes to esters and lactones.5
Much later, Stanley observed carboxylic acids as side products
of hydroformylation reactions catalyzed by a dimeric rhodium
catalyst and suggested the AWS as a likely reaction pathway.4
To the best of our knowledge, these are the only direct
demonstrations of the AWS reaction.
neutral conditions has also been reported by Maitlis using a
series of half-sandwich iridium, rhodium, and ruthenium
precursors.2,3 An intriguing alternative method for generating
commercially valuable carboxylic acids from aldehydes is the
“aldehyde−water shift” (AWS) reaction,4 analogous to the well-
studied water−gas shift reaction (Scheme 2), in which water
The AWS reaction may also play a role in the dehydrogen-
ative oxidation of primary alcohols. Beller6−8 and Grutzmacher9
̈
have independently published reports demonstrating oxidation
of aqueous methanol to carbon dioxide with release of three
equivalents of H . Grutzmacher10 and Milstein11 have shown a
̈
2
more general oxidation reaction in which a wide range of
primary alcohols can be oxidized to the corresponding
Scheme 2. Aldehyde−Water Shift and Water−Gas Shift
Received: June 17, 2014
Revised: July 24, 2014
Published: July 29, 2014
© 2014 American Chemical Society
3034
dx.doi.org/10.1021/cs500843a | ACS Catal. 2014, 4, 3034−3038

ACS Catalysis
Letter
carboxylate salt in the presence of base. The reactions are
thought to follow a reaction pathway similar to that suggested
by Milstein in his oxidative esterification and amidation
reactions of terminal alcohols and the “oxidant-free” trans-
formation of cyclic amines to lactams.12−14 In the rate-limiting
step, the alcohol is dehydrogenated to generate an aldehyde,
which is in equilibrium with its gem-diol.11 The gem-diol is then
dehydrogenated to generate the carboxylic acid, which is rapidly
deprotonated by the added base (Scheme 4).15
and heated at 105 °C for 20 h in a Teflon-sealed glass reaction
vessel. Products and product yields were determined by H
1
NMR spectroscopy utilizing phenol as an internal standard (see
Supporting Information). In addition to the desired carboxylic
acid product and unreacted starting material (aldehyde and the
corresponding gem-diol) the corresponding alcohol reduction
product resulting from aldehyde disproportionation was
observed as the remainder of the converted starting material,
as summarized in Table 1.
When using Ir(bpy) or Ru(bpy) as precatalyst, the 1H NMR
spectrum of the product mixture shows resonances correspond-
ing to a coordinated bipyridine ligand. Signals attributable to
alkyl groups of coordinated Cp* and p-cymene can also be
observed when acetaldehyde is utilized as substrate further
suggesting that the precatalyst remains intact for the duration of
the reaction.19 However, due to the low catalyst concentration
relative to substrate (and the corresponding low signal-to-noise
ratio observed for catalyst resonances), it could not be
conclusively shown that catalyst decomposition was not
occurring. Therefore, homogeneity of the catalysts was further
investigated by performing test reactions in the presence of a
drop of metallic mercury. In all cases, the addition of mercury
did not significantly alter the composition of the final reaction
mixture, suggesting that the active catalyst for the reaction is a
homogeneous metal complex.
Two metrics were utilized to evaluate the viability of each
precatalyst. First, percent conversion of starting material was
used to gauge the overall activity of the respective complexes.
Second, the reaction rate for the aldehyde−water shift relative
to the competing aldehyde disproportionation was inferred
from the amount of carboxylic acid present in the converted
products.20
The iridium bipyridyl complexes were the most active
catalyst precursors, displaying near quantitative conversion of
propionaldehyde. The rhodium precatalysts followed a reversed
activity trend to that observed from the iridium analogues;
Rh(ppy) displayed a high conversion rate when compared to
the bpy and bpy−OMe analogues. Surprisingly, precatalysts
featuring the bpy−OH ligand, which might be expected to
serve as a proton shuttle,17,21,22 did not function differently
from the electronically similar bpy−OMe complexes which do
not possess this potential for bifunctionality.
Scheme 4. Acceptorless Alcohol Dehydrogenation Reported
by Grutzmacher and Milstein.10,11
̈
Here we report a novel example of the AWS reaction in the
absence of a hydrogen acceptor or base. The reaction proceeds
under neutral aqueous conditions using iridium, rhodium, and
ruthenium catalysts. For all catalysts investigated, a competition
between the AWS reaction and aldehyde disproportionation
was observed. Experimental and computational studies suggest
a possible pathway for this reaction.
The AWS reaction is the microscopic reverse of the aqueous
hydrogenation of a carboxylic acid. Consequently, we
hypothesized that a series of half-sandwich iridium and
rhodium complexes recently found to catalyze the hydro-
genation of carboxylic acids16 may also be able to catalyze the
AWS reaction. In addition, related complexes are known to be
capable of generating H2 from alcohols under aqueous
conditions.17 Initially, several complexes were evaluated as
precatalysts (Chart 1). Dicationic complexes of the form [(p-
Chart 1. Catalysts Screened for the AWS Reaction
Ruthenium complexes consistently exhibited slow rates of
reaction. Although Maitlis and co-workers did not observe
disproportionation utilizing [(C6Me6)Ru(OH)3]Cl·4H2O in
the presence of bpy at 50 °C,3 we observed turnover at 105 °C
using the closely related Ru(bpy) catalyst. Moreover, the low
conversion measured after 18 h is not a result of catalyst
deactivation, as longer reaction times utilizing the Ru(bpy)
catalyst lead to further conversion to products. For the Ru
catalysts tested, doubling the catalyst loading doubled the
percent conversion of starting material, suggesting a first-order
rate dependence on the metal (Table 1).
For many of the catalysts tested, propionic acid is present as
approximately 50% of the converted product, indicating that
disproportionation is the reaction preferentially catalyzed, and
there is little AWS occurring. However, the Ir(ppy) and
Ru(bpy) precatalysts stand out (Table 1), both demonstrating
greater than 70% selectivity for acid production. At 75%
selectivity for acid production, two-thirds of the generated acid
is derived from the AWS, while one-third results from
disproportionation.
cymene)Ru(L-L)OH2][OTf]2 and [Cp*M(L-L)OH2][OTf]2
(Cp* = 1,2,3,4,5-pentamethylcyclopentadienide, M = Ir, Rh,
L-L = 2,2′-bipyridine (bpy), 4,4′-dimethoxy-2,2′-bipyridine
(bpy-OMe), 6,6′-dihydroxy-2,2′-bipyridine (bpy−OH), OTf =
trifluoromethanesulfonate) were prepared.18 Monocationic
[Cp*M(ppy)OH2][OTf] (M = Ir, Rh, ppy = κ-C1-phenylene-
2,2′-κ-N-pyridine) complexes were synthesized by reaction of
phenylpyridine with [Cp*M(OH2)3][OTf]2.18 For conven-
ience, complexes will hereafter be referred to by the metal and
pyridyl ligand (i.e., [Cp*Ir(bpy)OH2][OTf]2 = Ir(bpy)).
The Ru, Rh, and Ir complexes were tested for catalytic
activity in the AWS reaction utilizing propionaldehyde as a
model substrate. In a typical reaction, the catalyst (0.4 mol %)
was dissolved in 10 mL of 0.500 M aqueous propionaldehyde
3035
dx.doi.org/10.1021/cs500843a | ACS Catal. 2014, 4, 3034−3038

ACS Catalysis
Letter
a
Table 1. Precatalyst Screen Results
0.4 mol % Catalyst
% Acid
0.8 mol % Catalyst
% Acid
Precatalyst
Rh(bpy)
Conversion
% AWS
Conversion
% AWS
16(4)
95(1)
17(3)
39(3)
99(1)
3(1)
55(7)
51(1)
60(3)
54(1)
52(2)
73(4)
53(3)
50(1)
76(5)
71(2)
62(2)
18(20)
4(4)
28(4)
95(6)
26(2)
82(1)
99(1)
4(2)
53(1)
54(2)
56(1)
51(1)
54(5)
73(6)
52(2)
51(1)
75(5)
57(4)
59(3)
11(4)
15(8)
21(3)
4(4)
Rh(ppy)
Rh(bpy−OMe)
Rh(bpy−OH)
Ir(bpy)
33(7)
15(4)
8(8)
15(15)
63(13)
8(8)
Ir(ppy)
63(8)
11(11)
0(4)
Ir(bpy−OMe)
Ir(bpy−OH)
Ru(bpy)
99(1)
99(1)
5(1)
99(1)
99(1)
9(1)
4(4)
68(9)
59(4)
39(6)
67(10)
25(14)
31(10)
Ru(bpy−OMe)
Ru(bpy−OH)
5(1)
9(2)
6(1)
12(6)
a
% Reaction run in 10 mL of aqueous propionaldehyde. % Acid based on converted propionaldehyde. % AWS: percent of acid product obtained
from AWS; calculated as (% Acid − (1 − % Acid))/% Acid.20 In parentheses: Standard deviation for “Conversion” and “% Acid.” Error in “% AWS”
calculated based on 1 standard deviation of % Acid.
To confirm that the carboxylic acid product is produced via
the AWS, production of hydrogen was investigated by exposing
the headspace gas from a propionaldehyde reaction to a
solution of Ir(Cl)(CO)(PPh3)2 in benzene.23 1H and 31P{1H}
NMR analysis demonstrated the expected formation of
Ir(Cl)(CO)(PPh3)2(H)2, confirming the production of hydro-
gen gas (see Supporting Information for full details).24
The substrate scope of the reaction was then investigated
utilizing active catalyst Ir(bpy) and selective catalyst Ru(bpy)
(Table 2). Both catalysts demonstrated essentially identical
reactivity with acetaldehyde and propionaldehyde (entries 1
and 2 for Ir(bpy) and 4 and 5 for Ru(bpy)). The more sterically
hindered pivaldehyde showed similar selectivity but decreased
activity when compared to propionaldehyde. This may be due
to the lower aqueous solubility of pivaldehyde when compared
to propionaldehyde (reactions of pivaldehyde were biphasic).
Though the reactivity of benzaldehyde with the Ir(bpy)
catalyst is quite similar to that of the other substrates (low
conversion is again due to poor substrate solubility), the
reactivity of benzaldehyde and water in the presence of
Ru(bpy) is markedly different. In this case, the reaction is 95%
selective for the production of carboxylic acid over alcohol.
Thus, 95% of the generated carboxylic acid can therefore be
attributed to AWS reactivity.
Investigation of a plausible AWS mechanism was carried out
using DFT. Our working hypotheses for steps in the AWS
mechanism are based on proposals for related reactions by
Stanley and co-workers,4 Maitlis and co-workers,3 Yamaguchi
a
Table 2. Substrate Screen for AWS Reactivity
and co-workers,17 Grutzmacher and co-workers,10 and Milstein
̈
and co-workers,11 as well as the DFT studies of aqueous alcohol
oxidation reported by Li and Hall.15 Our mechanism starts with
acetaldehyde and model catalyst [Cp*Ir(bpy)]2+(A), a 16-
electron species with a vacant coordination site. Initial
calculations with Ir(I) and Ir(V) entities indicated these to be
much higher in free energy, so DFT modeling focused on
Ir(III) complexes. The cycle begins with coordination of
MeCHO to Ir, forming B. Computations suggest that κ1-O is
the preferred ligation mode (versus κ2-C,O) of the aldehyde
substrate in B. Next, as hypothesized by Stanley et al.,4 a weak
donor−acceptor interaction between H2O and the carbonyl
carbon of the ligated acetaldehyde was sought and found,
yielding a minimum, C, termed the hydrate tautomer. Ligation
of the two substrates is computed to be mildly endergonic in
aqueous solvent model, ΔGA→B = +2 kcal/mol and ΔGB→C
=
+8 kcal/mol. Hydrate tautomer C may then tautomerize to a
“diol” form, D, the latter being lower in energy by 3 kcal/mol
and requiring the surmounting of a modest free energy barrier
of 14 kcal/mol. Next, in a concerted dehydrogenation, H+ and
H− are simultaneously transferred to water and the metal,
respectively, producing hydronium and hydride complex E. In
species E, acetic acid is no longer coordinated to the metal, but
is instead stabilized by a hydrogen bond between the acetic acid
hydroxyl group and the hydride ligand. Acetic acid may then
dissociate from E, forming species F. This step is entropically
favored, ΔGE→F = −8 kcal/mol. Finally, H2 is released from F
to close the cycle; this step is computed to be exergonic by 11
kcal/mol. In the competing disproportionation reaction, F
a
b
0.4 mol % catalyst in 10 mL H2O, 5 mmol substrate, 105 °C, 20 h. %
Acid based on converted aldehyde. % AWS: percent of acid product
obtained from AWS; calculated as (% Acid − (1 − % Acid))/% Acid.20
In parentheses: Standard deviation for “Conversion” and “% Acid.”
Error in “% AWS” calculated based on 1 standard deviation of % Acid.
3036
dx.doi.org/10.1021/cs500843a | ACS Catal. 2014, 4, 3034−3038

ACS Catalysis
Letter
transfers the hydride to an additional equivalent of aldehyde as
proposed in an analogous hydrogenation mechanism.16 Thus,
though other pathways could be envisioned, all of the steps in
Scheme 5 yield an energetically feasible route for the AWS
reaction. Full investigation into the mechanism is underway.
aldehydes allows this reaction to proceed under exceptionally
mild conditions. This represents an exciting addition to the
suite of fundamental organic transformations and could have
great impact within synthetic organic chemistry. Studies into
the development of catalysts that are both highly efficient and
highly selective for AWS reactivity are currently underway.
Scheme 5. Modeled Mechanism for AWS with Cp*Ir
a
ASSOCIATED CONTENT
* Supporting Information
Full experimental details, NMR spectra, and computational
details. This material is available free of charge via the Internet
at http://pubs.acs.org.
■
Catalyst
S
AUTHOR INFORMATION
Corresponding Authors
*E-mail: t@unt.edu (T.R.C.).
■
*E-mail: heinekey@chem.washington.edu (D.M.H.).
Author Contributions
All authors have given approval to the final version of the
manuscript.
Notes
The authors declare no competing financial interest.
a
Computed free energies (kcal/mol) in continuum aqueous solvent
ACKNOWLEDGMENTS
are noted.
■
This work was supported by the Camille and Henry Dreyfus
Postdoctoral Program in Environmental Chemistry (T.P.B.,
K.I.G.), by the Washington NASA Space Grant (J.C.T.), and by
NSF under the CCI Center for Enabling New Technologies
through Catalysis (CENTC), CHE-1205189 (T.P.B., J.C.T.,
K.I.G., S.K.H., T.R.C., D.M.H.). W.C.O. is a student in the
Texas Academy of Math and Science (TAMS) at the University
of North Texas (UNT) and thanks the TAMS Summer
Research Fellowship for its support of this research. The
authors acknowledge Dr. David L. Thorn for helpful
discussions.
Observed reactivity trends are consistent with this mecha-
nistic proposal. Dicationic iridium complexes, which form
stronger MH bonds than their rhodium analogues,25 would
be expected to react more quickly as hydride formation is
calculated to be rate limiting. However, the resultant iridium
hydrides are very weak bases,26,27 and therefore the reaction
favors disproportionation. In an effort to promote the
formation of H2 via protonation of the generated metal hydride
(F in Scheme 5), substrates were screened utilizing the Ir(bpy)
and Ru(bpy) catalysts in aqueous HBF4 (1M) (See Supporting
Information).28 In all cases, the percent carboxylic acid
produced was not observed to increase in the presence of
HBF4. In the case of the slower Ru(bpy) catalyst, substrates
bearing α-hydrogens were observed to undergo aldol
condensation yielding an immiscible organic layer over the
course of the reaction.29
Reaction selectivity observed in the substrate screen is also
consistent with our mechanistic hypothesis. Acetaldehyde,
propionaldehyde, and pivaldehyde are electronically quite
similar and likely will have similar propensity for hydride
transfer from the metal center. Benzaldehyde, which is fully
conjugated, is less electrophilic, thereby allowing for high
selectivity with the Ru(bpy) catalyst. Ir(bpy), whose corre-
sponding hydride is not sufficiently basic to competitively
deprotonate water,26 still displays disproportionation behavior.
Future studies will focus on the development and utilization of
catalysts, which are known to be basic enough to achieve this
disproportionation, thereby allowing for high AWS selectivity.27
In conclusion, we have demonstrated the viability of the
aldehyde−water shift reaction, an attractive method for the
oxidation of aldehydes using water as terminal oxidant. AWS
reactivity was found to be in competition with aldehyde
disproportionation. High selectivity (95(4)%) for the AWS was
achieved when using [(p-cymene)Ru(bpy)OH2][OTf]2 as
precatalyst and benzaldehyde as substrate, a novel example of
a selective AWS in the absence of a hydrogen acceptor.
Utilization of water as the terminal oxidant in the oxidation of
REFERENCES
■
(1) Smith, M. B.; March, J. March’s Advanced Organic Chemistry:
Reactions, Mechanisms, and Structure, 6th ed.; John Wiley and Sons,
Inc.: Hoboken, NJ, 2007.
(2) Cook, J.; Hamlin, J. E.; Nutton, A.; Maitlis, P. M. J. Chem. Soc.,
Chem. Commun. 1980, 144−145.
(3) Cook, J.; Hamlin, J. E.; Nutton, A.; Maitlis, P. M. J. Chem. Soc.,
Dalton Trans. 1981, 2342−2352.
(4) Stanley, G. G.; Aubry, D. A.; Bridges, N.; Barker, B.; Courtney, B.
Prepr. Pap. - Am. Chem. Soc., Div. Fuel Chem. 2004, 49, 712−713.
(5) Murahashi, S.-I.; Naota, T.; Ito, K.; Maeda, Y.; Taki, H. J. Org.
Chem. 1987, 52, 4319−4327.
(6) Nielsen, M.; Alberico, E.; Baumann, W.; Drexler, H.-J.; Junge, H.;
Gladiali, S.; Beller, M. Nature 2013, 495, 85−89.
(7) Monney, A.; Barsch, E.; Sponholz, P.; Junge, H.; Ludwig, R.;
Beller, M. Chem. Commun. 2014, 50, 707−709.
(8) Alberico, E.; Sponholz, P.; Cordes, C.; Nielsen, M.; Drexler, H.-J.;
Baumann, W.; Junge, H.; Beller, M. Angew. Chem., Int. Ed. 2013, 52,
14162−14166.
(9) Rodriguez-Lugo, R. E.; Trincado, M.; Vogt, M.; Tewes, F.;
Santiso-Quinones, G.; Grutzmacher, H. Nat. Chem. 2013, 5, 342−347.
̈
(10) Zweifel, T.; Naubron, J.-V.; Grutzmacher, H. Angew. Chem., Int.
Ed. 2009, 48, 559−563.
̈
(11) Balaraman, E.; Khaskin, E.; Leitus, G.; Milstein, D. Nat. Chem.
2013, 5, 122−125.
(12) Zhang, J.; Leitus, G.; Ben-David, Y.; Milstein, D. J. Am. Chem.
Soc. 2005, 127, 10840−10841.
(13) Gunanathan, C.; Ben-David, Y.; Milstein, D. Science 2007, 317,
790−792.
3037
dx.doi.org/10.1021/cs500843a | ACS Catal. 2014, 4, 3034−3038

ACS Catalysis
Letter
(14) Khusnutdinova, J. R.; Ben-David, Y.; Milstein, D. J. Am. Chem.
Soc. 2014, 136, 2998−3001.
(15) Li, H.; Hall, M. B. J. Am. Chem. Soc. 2014, 136, 383−395.
(16) Brewster, T. P.; Miller, A. J. M.; Heinekey, D. M.; Goldberg, K.
I. J. Am. Chem. Soc. 2013, 135, 16022−16025.
(17) Kawahara, R.; Fujita, K.-I.; Yamaguchi, R. J. Am. Chem. Soc.
2012, 134, 3643−3646.
(18) For synthetic details, see Supporting Information.
(19) Under analogous conditions, [(p-cymene)RuCl2]2 rapidly
decomposes to heterogeneous material.
(20) For example, if 4 mmol of aldehyde is converted to products,
and 3 mmol (75%) of the observed product is carboxylic acid, 1 mmol
of the carboxylic acid product results from disproportionation, and 2
mmol result from AWS.
(21) Wang, W.-H.; Hull, J. F.; Muckerman, J. T.; Fujita, E.; Himeda,
Y. Energy Environ. Sci. 2012, 5, 7923−7926.
(22) Nieto, I.; Livings, M. S.; Sacci, J. B., III; Reuther, L. E.; Zeller,
M.; Papish, E. T. Organometallics 2011, 30, 6339−6342.
(23) For a related example, see Finney, E. E.; Ogawa, K. A.;
Boydston, A. J. J. Am. Chem. Soc. 2012, 134, 12374−12377.
(24) Burk, M. J.; McGrath, M. P.; Wheeler, R.; Crabtree, R. H. J. Am.
Chem. Soc. 1988, 110, 5034−5039.
(25) Simoes, J. A. M.; Beauchamp, J. L. Chem. Rev. 1990, 90, 629−
688.
(26) Vogt, M.; Pons, V.; Heinekey, D. M. Organometallics 2005, 24,
1832−1836.
(27) Morris, R. H. J. Am. Chem. Soc. 2014, 136, 1948−1959.
(28) Rh(bpy) was also tested utilizing propionaldehyde as substrate.
This led to aldol products.
(29) 31% aldol product was also observed using Ir(bpy) with
propionaldehyde as substrate. This was not observed using
acetaldehyde in the presence of Ir(bpy).
3038
dx.doi.org/10.1021/cs500843a | ACS Catal. 2014, 4, 3034−3038