Base-Free Methanol Dehydrogenation Using a Pincer-Supported Iron Compound and Lewis Acid Co-catalyst

Article abstract of DOI:10.1021/acscatal.5b00137

Hydrogen is an attractive alternative energy vector to fossil fuels if effective methods for its storage and release can be developed. In particular, methanol, with a gravimetric hydrogen content of 12.6%, is a promising target for chemical hydrogen storage. To date, there are relatively few homogeneous transition metal compounds that catalyze the aqueous phase dehydrogenation of methanol to release hydrogen and carbon dioxide. In general, these catalysts utilize expensive precious metals and require a strong base. This paper shows that a pincer-supported Fe compound and a co-catalytic amount of a Lewis acid are capable of catalyzing base-free aqueous phase methanol dehydrogenation with turnover numbers up to 51 000. This is the highest turnover number reported for either a first-row transition metal or a base-free system. Additionally, this paper describes preliminary mechanistic experiments to understand the reaction pathway and propose a stepwise process, which requires metal-ligand cooperativity. This pathway is supported by DFT calculations and explains the role of the Lewis acid co-catalyst.(Chemical Equation Presented).

Full text of DOI:10.1021/acscatal.5b00137

Research Article
pubs.acs.org/acscatalysis
Base-Free Methanol Dehydrogenation Using a Pincer-Supported Iron
Compound and Lewis Acid Co-catalyst
Elizabeth A. Bielinski,† Moritz Forster,‡ Yuanyuan Zhang,§ Wesley H. Bernskoetter,*,§ Nilay Hazari,*,†
̈
and Max C. Holthausen*,‡
†Department of Chemistry, Yale University, P.O. Box 208107, New Haven, Connecticut 06520, United States
‡Institut fur Anorganische und Analytische Chemie, Goethe-Universitat, Max-von-Laue-Strasse 7, 60438 Frankfurt am Main,
̈
̈
Germany
§Department of Chemistry, Brown University, Providence, Rhode Island 02912, United States
S
* Supporting Information
ABSTRACT: Hydrogen is an attractive alternative energy
vector to fossil fuels if effective methods for its storage and
release can be developed. In particular, methanol, with a
gravimetric hydrogen content of 12.6%, is a promising target
for chemical hydrogen storage. To date, there are relatively few
homogeneous transition metal compounds that catalyze the
aqueous phase dehydrogenation of methanol to release hydrogen and carbon dioxide. In general, these catalysts utilize expensive
precious metals and require a strong base. This paper shows that a pincer-supported Fe compound and a co-catalytic amount of a
Lewis acid are capable of catalyzing base-free aqueous phase methanol dehydrogenation with turnover numbers up to 51 000.
This is the highest turnover number reported for either a first-row transition metal or a base-free system. Additionally, this paper
describes preliminary mechanistic experiments to understand the reaction pathway and propose a stepwise process, which
requires metal−ligand cooperativity. This pathway is supported by DFT calculations and explains the role of the Lewis acid co-
catalyst.
KEYWORDS: iron, catalysis, methanol dehydrogenation, metal−ligand cooperativity, pincer ligands, DFT calculations
catalysts for MeOH dehydrogenation since the 1980s,7 only
recently have a number of well-defined systems that can
INTRODUCTION
■
As the worldwide demand for energy increases, the develop-
ment of large-scale alternatives to fossil fuels will become more
catalyze the full dehydrogenation of MeOH and water to H2
and CO2 been reported (Table 1).8 These systems generally
important from both environmental and economic stand-
points.1 H2 is a potential clean energy source as its combustion
operate at significantly lower temperatures than heterogeneous
catalysts and produce less CO. However, to date the most
results only in the generation of water as a byproduct.2 In
active homogeneous catalysts are based on expensive precious
particular, chemical H2 storage (CHS) based on the reversible
(de)hydrogenation of organic molecules represents a method
by which a liquid organic carrier (LOC) may serve as a safe and
easily transportable fuel.3 The selective release of H2 from a
LOC, followed by either direct combustion or use as a
feedstock in a proton-exchange membrane fuel cell,4 would
allow for the generation of energy from H2 while avoiding the
dangers and difficulties associated with its transport.3 Methanol
(MeOH) is a promising target for CHS,3a,5 as it has a high
gravimetric H2 content (12.6%) and can be dehydrogenated in
the presence of water to release 3 equiv of H2 (eq 1).
Furthermore, it can be generated from renewable sources.5b
metals such as Ru,8b and with the exception of Grutzmacher’s
̈
seminal system,8a require either the use of a strong base or a
precious metal co-catalyst.
The only reported first-row transition metal catalyst for
MeOH dehydrogenation was described by Beller and co-
workers.8c This complex, (iPrPNHP)Fe(CO)H(HBH3) (iPrPNP
= N(CH2CH2PiPr2)2, C),9 features a bifunctional PNP ligand
and is able to achieve ∼10 000 TONs in the presence of 8 M
KOH. We, along with several other groups, have been studying
the related amido compounds (RPNP)FeH(CO) (RPNP =
10
i
N{CH2CH2(PR2)}2; R = Pr (1a) or Cy (1b)), which in the
case of 1a can be formed by the addition of base to C. It has
been demonstrated that 1a can dehydrogenate primary alcohols
such as 1-butanol to the corresponding esters without a base or
H2 acceptor (Scheme 1),10e whereas 1a and 1b can be used as
CH3OH + H2O → 3H2 + CO2
(1)
Currently in re-formed MeOH fuel cells, heterogeneous
catalysts are used to release H2 from MeOH for the generation
of electricity.3a,6 These catalysts operate at high temperatures
and pressures and produce a significant amount of CO, which
ultimately poisons the fuel cell.3a,6a Although there has been
ongoing research into the development of homogeneous
Received: January 22, 2015
Revised: March 3, 2015
© XXXX American Chemical Society
2404
DOI: 10.1021/acscatal.5b00137
ACS Catal. 2015, 5, 2404−2415

ACS Catalysis
Research Article
Table 1. Homogeneous Transition Metal Catalysts for Aqueous Phase MeOH Dehydrogenation to CO2 and H2
catalyst
ratio MeOH/H2O
solvent
additive
KOH
TON
yield (%)
reference
A
B
C
D
E
1:1 MeOH/H2O
4:1 MeOH/H2O
4:1 MeOH/H2O
1:1 MeOH/H2O
4:1 MeOH/H2O
toluene
neat
28,000
350,000
9800
77
27
6
Milstein8d
Beller8b
Beller8c
8 M KOH
8 M KOH
neat
THF
540
90
26
Grutzmacher8a
̈
triglyme
4200
Beller8e
Scheme 1. Summary of Selected Previous and Current Reactions Catalyzed by 1a, 1b, 2a, and 2b
highly efficient catalysts for formic acid dehydrogenation in
combination with a Lewis acid (LA).10d Similarly, the related
formate complexes 2a and 2b, which are proposed to generate
1a and 1b in situ, are also active catalysts for formic acid
dehydrogenation in the presence of a LA.10d Herein we
demonstrate that 1a, 1b, 2a, and 2b in the presence of a LA co-
catalyst can be used as catalysts for the dehydrogenation of
MeOH without added base. We report a maximum TON of
∼51 000, the highest for either a first-row transition metal
based catalyst or a base free system. In addition, we describe
preliminary mechanistic studies and suggest a connection
between Beller’s catalyst C and 1a.
dehydrogenation of MeOH produces formaldehyde and
releases 1 equiv of H2. Subsequently, the reaction of water
with formaldehyde generates methanediol, which undergoes a
second dehydrogenation to produce formic acid and a second
equivalent of H2. Finally, formic acid dehydrogenation results in
the release of the third equivalent of H2, along with CO2. Given
that 1a and related Fe complexes catalyze both the
dehydrogenation of primary alcohols such as 1-butanol to
esters (analogous to steps i−iii in Scheme 2) and formic acid
dehydrogenation (step iv in Scheme 2) without a base, we
postulated that they may be able to perform base-free aqueous
phase MeOH dehydrogenation if compatible conditions for
both reactions could be developed. To achieve this tandem
reaction we pursued a strategy in which we first studied the
dehydrogenation of MeOH in the absence of water (step i) and
then explored full aqueous phase MeOH dehydrogenation.
MeOH Dehydrogenation in the Absence of Water.
Although catalysts 1a and 1b were used to dehydrogenate
primary alcohols,10e MeOH was not used as a substrate.
Initially, we screened conditions for the dehydrogenation of
MeOH in the absence of water using 1b as the catalyst (Tables
2 and 3). The primary products of this reaction were methyl
formate and H2; the latter was identified using gas
chromatography (GC) (see Supporting Information Figure
RESULTS AND DISCUSSION
■
Previously it has been proposed that the complete aqueous
phase dehydrogenation of MeOH to CO2 and H2 occurs
following the stepwise pathway shown in Scheme 2.8a,b Initial
Scheme 2. Proposed Stepwise Pathway for Aqueous Phase
Dehydrogenation of MeOH to CO2 and H2
2405
DOI: 10.1021/acscatal.5b00137
ACS Catal. 2015, 5, 2404−2415

ACS Catalysis
Research Article
Under the optimized conditions shown in Table 3, 1a and 1b
show nearly identical activities for MeOH dehydrogenation to
methyl formate, giving 73% (1460 turnovers) and 71% yield
(1421 turnovers), respectively. To the best of our knowledge
these are the highest turnovers reported to date for this
reaction.7,11 Interestingly, the borohydride complex C is
significantly less active, achieving only 19% yield (384
turnovers). This is consistent with recent observations by
Guan and co-workers suggesting that the activation of C
through loss of BH3 results in increased catalytic activity for the
hydrogenation of esters to alcohols by rapidly generating the
active catalyst 1a.12 In our current system there is no additive to
facilitate the formation of the active species. Presumably, one of
the roles of KOH in Beller’s aqueous phase dehydrogenation of
MeOH using C8c is to facilitate catalyst activation through the
removal of BH3. The formate complexes 2a and 2b are also
poor catalysts for MeOH dehydrogenation. This is most likely
due to the inability of these species to readily undergo
decarboxylation and 1,2-elimination of H2 to access catalytically
active 1a or 1b, in the absence of base or LA.10d
Table 2. Solvent Screen for MeOH Dehydrogenation
Catalyzed by 1b
a
b
c
solvent
time (min)
TON
yield (%)
dioxane
20
20
10
11
12
10
15
10
10
53
71
26
35
40
48
53
56
58
85
88
propylene carbonate
chlorobenzene
dimethyl sulfoxide
tetrahydrofuran
cyclopentylmethyl ether
toluene
80
96
107
112
116
170
176
acetonitrile
ethyl acetate
a
Reaction conditions: MeOH (36 μL, 0.91 mmol), 1b (9.1 μmol, 1
b
mol %) 5 mL solvent, reflux. Time at which no further increase in
TON was observed. TON was measured using a gas buret. Each
c
equivalent of H2 generated is counted as a TON. All numbers are the
average of two runs.
Recently we demonstrated that LA co-catalysts assist in the
dehydrogenation of formic acid using 1a and 1b.10d The
addition of 10 mol % LiBF4 did not influence the yield or
kinetic profile for the dehydrogenation of MeOH to methyl
formate using 1b (see Table S3), suggesting that the rate-
determining steps in MeOH dehydrogenation and formic acid
dehydrogenation are not equivalent. Furthermore, the kinetic
isotope effect (KIE) for catalytic MeOH dehydrogenation
(determined from rate constants for parallel reactions using
MeOH and d4-MeOH and 1b, see the Supporting Information)
is kH/kD = 2.5(2). In contrast, for catalytic formic acid
dehydrogenation using 1b the KIE is kH/kD = 4.2(3) (see the
Supporting Information). This indicates that the rate-
determining steps in the two processes are not an identical
elementary reaction, such as H2 elimination. Although the
addition of LA co-catalysts did not enhance catalysis using 1a
and 1b, a remarkable increase was observed in the cases of 2a
and 2b (Table 3). In the presence of 10 mol % LiBF4 complete
conversion of MeOH to methyl formate was observed using 0.1
mol % 2a and the catalyst loading could be decreased to 0.01
mol % without any loss in yield.13 We believe that this dramatic
increase occurs because LiBF4 facilitates the decarboxylation of
the formate complexes to access the catalytically active
species.10d The combination of 2 and a LA may provide an
alternative strategy for dehydrogenating challenging organic
substrates10e such as 1-cyclohexylmethanol using low catalyst
loading, as it appears to generate a more stable catalytic system
than using 1 without any additives.
Table 3. Catalyst Optimization for MeOH
Dehydrogenation
a
b
c
catalyst
mol % [Fe] time (min)
TON
yield (%)
1a
1b
C
0.1
50
45
1460
1421
384
73
71
19
12
>99
6
0.1
0.1
40
2a
0.1
25
255
2a + 10 mol % LiBF4
2b
0.1
40
>1999
126
0.1
30
2b + 10 mol % LiBF4
2a + 10 mol % LiBF4
2a + 10 mol % LiBF4
0.1
45
1920
>19999
12400
96
>99
6
0.01
0.001
265
340
a
Reaction conditions: MeOH (36 μL, 0.91 mmol), [Fe] (0.001−0.1
b
mol %), 10 mL of ethyl acetate, reflux. Time at which no further
increase in TON was observed. TON was measured using a gas buret.
c
Each equivalent of H2 generated is counted as a TON. All numbers are
the average of two runs.
S10). High yields of methyl formate were obtained only when
the moderately polar solvents ethyl acetate and acetonitrile
were utilized (Table 2). In contrast, excellent yields were
previously obtained for the dehydrogenation of 1-butanol in
nonpolar toluene.10e The reasons for the excellent performance
in ethyl acetate and acetonitrile, and the relatively low yield for
MeOH dehydrogenation in nonpolar solvents such as toluene
are unclear. In subsequent optimization reactions, ethyl acetate
was used as the solvent because there is a significant decrease in
catalytic activity when formic acid dehydrogenation catalyzed
by 1 is performed in acetonitrile, which therefore is not a
suitable solvent for full aqueous phase MeOH dehydrogenation
(see Table S1). The dehydrogenation of MeOH was dependent
not only on the identity of solvent but also on the
concentration (see Table S2). Dilution studies indicate that
the reaction fails at high concentrations. This is consistent with
a bimolecular catalyst decomposition pathway, and single
crystals of the Fe(0) complex (CyPNHP)Fe(CO)2, which is
proposed to form in a bimolecular fashion, were obtained from
a completed catalytic reaction mixture (see Figure S16).
To further probe the mechanism of MeOH dehydrogenation,
stoichiometric reactions were performed (Scheme 3). When 1
equiv of MeOH was added to a d8-toluene solution of 1b at low
temperature (−80 °C), a new PNP-supported Fe species, 3b,
was observed by 1H and 31P NMR spectroscopy along with 1b
(see Figures S3 and S4). Compound 3b has a triplet resonance
in the 1H NMR spectrum at −23.95 ppm and a singlet
resonance in the 31P{1H} NMR spectrum at 84.8 ppm. It is
assigned as (CyPNHP)Fe(CO)(H)(OCH3), which arises from
1,2-addition of MeOH to 1b. Further evidence for this
assignment was obtained through an experiment between
CD3OD and 1b at −75 °C (see Figure S5). In this reaction two
resonances in a 3:1 ratio were observed at 3.57 and 2.04 ppm in
2
the H NMR spectrum, along with the previously observed
resonance at 84.8 ppm, in the 31P{1H} NMR spectra. The
2406
DOI: 10.1021/acscatal.5b00137
ACS Catal. 2015, 5, 2404−2415

ACS Catalysis
Research Article
Scheme 3. Stoichiometric Reactions of 1b with MeOH and/or Water
resonances in the 2H NMR spectrum are proposed to
correspond to the OCD3 ligand (3.57 ppm) bound to Fe and
the N−D (2.04 ppm) moiety. Free CD3OD was also observed
in the 2H NMR spectrum, which is consistent with the presence
of unreacted 1b. In both experiments using CH3OH and
CD3OD, the amount of 3b decreased relative to the amount of
1b as the temperature was increased. In fact, at 0 °C only trace
amounts of 3b were observed by 1H NMR or 2H NMR
spectroscopy and the predominant species is 1b. Cooling the
solutions to −80 °C resulted in the conversion of 1b and
CH3OH/CD3OH back to 3b. These experiments suggest that
1b and MeOH are in equilibrium with 3b and that the 1,2-
addition of MeOH is temperature dependent. It was not
possible to isolate 3b, as the removal of solvent resulted in the
regeneration of 1b, along with substantial amounts of free
ligand and unidentified solid precipitate.
the observation that 4b is the resting state during catalysis as
determined using 31P NMR spectroscopy.
In the reaction pathway shown in Scheme 2, water is
necessary to fully dehydrogenate MeOH and generate 3 equiv
of H2. The stoichiometric reaction of 1b with 1 equiv of both
MeOH and water led to the formation of the previously
characterized formate complex 2b along with H2.10d
A
proposed pathway for this reaction is summarized in Scheme
5. Initially, MeOH is dehydrogenated by 1b to generate
Scheme 5. Proposed Pathway for Stoichiometric Reaction of
1 equiv of MeOH and Water with 1b
When the reaction between 1 equiv of MeOH and 1b in d8-
toluene was performed at 50 °C, there was no evidence for the
Scheme 4. Proposed Stepwise Pathway for MeOH
Dehydrogenation in the Absence of Water
formation of 3b. Instead, the major Fe-containing products
were the dihydride (CyPNHP)Fe(CO)(H)2 (4b), which we
previously characterized,10d and (CyPNHP)Fe(CO)2. Also
present were a significant amount of free ligand, H2, methyl
formate, and MeOH (see Figures S6 and S7). The analogous
reaction between 1b and 2 equiv of MeOH (Scheme 3b)
resulted in the formation of the same products, but no MeOH
was observed. We believe that the pathway for this reaction
involves initial dehydrogenation of MeOH to produce
formaldehyde and H2 followed by esterification of form-
aldehyde with MeOH to form methoxymethanol, which is
subsequently dehydrogenated to generate methyl formate and
the second equivalent of H2 (Scheme 4). This is the same
sequence of reactions previously proposed for butanol
dehydrogenation10e and is consistent with our catalytic results.
The observation of the dihydride complex 4b suggests that
release of H2 to regenerate 1b is slow and is in agreement with
formaldehyde and H2, with the latter presumably formed via
the dihydride intermediate 4b. Subsequently, formaldehyde is
trapped by water to form methanediol, which is dehydro-
genated to form formic acid and 4b. Formic acid then
protonates 4b to form H2 and the formate product 2b. There is
precedent for all of the steps in this reaction sequence.10d,e In a
control experiment, 1b was treated with 1 equiv of water
(Scheme 3d). Even at low temperature a large number of
different products were observed by 31P NMR spectroscopy,
none of which were identifiable. Furthermore, removal of the
solvent led to almost complete decomposition of the
complexes, with only a small amount of 1b recovered,
indicating that the addition of water is largely irreversible.
This strongly suggests that when both water and MeOH are
present, 1b initially reacts with MeOH.
Aqueous Phase MeOH Dehydrogenation. Given the
similarity of 1 to Beller’s catalyst, C, we began our investigation
2407
DOI: 10.1021/acscatal.5b00137
ACS Catal. 2015, 5, 2404−2415

ACS Catalysis
Research Article
of aqueous phase MeOH dehydrogenation using Beller’s
optimized conditions of 4:1 (molar ratio) MeOH/H2O.8c In
the absence of a base or other additive, 1b catalyzes the
generation of H2 from an aqueous solution of MeOH in 58%
yield, based on water as the limiting reagent (Table 4).
However, methyl formate is also generated as a significant
product, suggesting that complete MeOH dehydrogenation to
H2 and CO2 is not occurring. When this reaction was
monitored using 31P NMR spectroscopy, the major Fe-
containing species at the end of the reaction was the formate
complex 2b. Presumably, if 2b, which we believe is formed
through 1,2 addition of formic acid to 1b (vide supra), cannot
undergo facile decarboxylation, it represents a highly stable
intermediate in catalysis. To prevent the accumulation of 2b,
the catalytic reaction was performed in the presence of a variety
of different LAs (Table 4). Several different LAs facilitate the
complete aqueous phase dehydrogenation of MeOH, without
formation of methyl formate. In general, the smaller, more
oxophilic cations such as Li+ and Na+ are the most active.
Additionally, non-coordinating or weakly coordinating anions
reaction using LiBF4 was analyzed by GC and found to contain
a 3:1 ratio of H2/CO2 and <0.1% CO (see Figures S12−S15).
This percentage of CO is significantly lower than that observed
with the best current heterogeneous catalysts3a,6a and
comparable with state-of-the art precious metal homogeneous
systems.8a,b Using LiBF4 as the LA, we explored the effect of
changing the quantity of LA on TON (Table 5). When the
catalyst loading of 1b is 0.5 mol %, the optimal LA loading is
between 5 and 10 mol %; however, at a lower loading of 1b
(0.1 mol %), a 10 mol % LA loading gives more efficient
catalytic activity. The decrease in performance at both higher
and lower LA loading is comparable to the LA effect that was
observed in formic acid dehydrogenation using 1 and 2.10d
The effect of changing the ratio of MeOH/H2O was explored
using a catalyst system including 1b and LiBF4 (Table 6). A
Table 5. Effect of Amount of LiBF4 on MeOH
a
Dehydrogenation in the Presence of Water
are necessary, with PF6 , BF4 , and OTf− giving the best
results. It is possible to use LAs as simple as NaCl, but the
poisoning effect of the chloride anion appears to be similar
regardless of the cation, as there is little difference in activity
between LiCl, NaCl, and CsCl. Even in the presence of LAs,
31P NMR spectroscopy indicates that the formate complex 2b is
the resting state during catalysis.
−
−
b
c
mol % LA
mol % 1b
time (min)
TON
yield (%)
1
2
0.5
25
25
20
25
60
350
375
58
62
5
>599
>599
>599
>99
>99
>99
10
20
The six LAs that gave quantitative conversion of MeOH and
water to H2 and CO2 at 0.5 mol % loading of 1b were tested at
lower catalyst loading to further explore the differences in their
activities (see Table S4). Despite their impressive performances
at high catalyst loading, both LiOTf and NaOTf performed
poorly under these conditions, whereas LiBF4 was the most
active, giving >99% yield in 12.5 h. The gas produced from the
5
10
20
0.1
320
600
750
975
>2999
>2999
32
>99
>99
a
Reaction conditions: water (18 μL, 1.0 mmol), MeOH (161 μL, 4.0
mmol), 1b (x mol % with respect to water), LA (mol % with respect to
water),10 mL of ethyl acetate, reflux. Time at which no further
increase in TON was observed. TON measured using a gas buret.
b
c
Each equivalent of H2 generated is counted as a TON. All numbers are
the average of two runs.
Table 4. LA Screening for Aqueous Phase MeOH
Dehydrogenation Using 1b
a
large excess of either MeOH or water afforded poor yields.
More moderate ratios of 2:1 or 4:1 MeOH/H2O gave
significantly higher TON and yields, with a ratio of 4:1 giving
a TON of 9500 (95% yield) in 41 h. The significant decrease in
catalytic activity at high water and/or MeOH concentrations
may be related to the instability of 1b in either neat MeOH or
water. This is in contrast to the reaction of 1b with 1 equiv of a
mixture of MeOH and water, which gave near quantitative
conversion to 2b, with very little evidence of decomposition.
Our optimal conditions are similar to those employed by Beller
and co-workers to achieve a TON of approximately 10 000 in
43 h using C and 8 M KOH.8c
Using our optimized conditions we tested the catalytic
activity of 1a, 1b, 2a, 2b, and C (Table 7). In combination with
LiBF4, both the amido complexes 1a and 1b and the formate
complexes 2a and 2b give yields >80% (8000 turnovers) at 0.03
mol % catalyst loading. In an analogous fashion to the
dehydrogenation of MeOH to methyl formate, C is not highly
active under these base-free conditions, giving only 25% yield.
This is presumably because it is not efficiently activated.
Further optimization using 2a at 0.01 mol % loading gave a
TON of 30 000 and yield of >99%. Lowering the catalyst
loading to 0.006 mol % gave a TON of 51 000, but the yield
was reduced to 50%. Overall, 2b in combination with 10 mol %
LiBF4 represents the first example of base-free MeOH
b
d
b
d
TON (time, yield
TON (time, yield
c
c
LA
min)
(%)
LA
NaCl
min)
(%)
e
no
additive
350 (20)
58
497 (15)
82
LiCl
499 (15)
496 (25)
427 (15)
297 (10)
>599 (20)
410 (15)
542 (20)
83
82
LiPF6
>599 (30)
425 (20)
>599 (35)
>599 (25)
542 (20)
530 (30)
>599 (15)
>599 (35)
393 (20)
>99
70
KCl
NaPF6
KPF6
CsCl
CaCl2
NaBArF
71
>99
>99
90
49
f
LiBF4
NaBF4
KBF4
>99
68
4
NaBPh4
NaBF4
88
90
LiOTf
NaOTf
KOTf
>99
>99
65
a
Reaction conditions: water (18 μL, 1.0 mmol), MeOH (161 μL, 4.0
mmol), 1b (0.5 mol % with respect to water), LA (0.1 mmol, 10 mol
% with respect to water), 10 mL of ethyl acetate, reflux. TON
measured using a gas buret. Each equivalent of H2 generated is
counted as a TON. All numbers are the average of two runs. Time at
which no further increase in TON was observed. Based on water as
b
c
d
e
the limiting reagent. Methyl formate was observed as a major product.
f
−
BArF = B{3,5-(CF3)2C6H3}4 .
4
2408
DOI: 10.1021/acscatal.5b00137
ACS Catal. 2015, 5, 2404−2415

ACS Catalysis
Research Article
is formed, it does not decompose so rapidly. Support for this
hypothesis was obtained from the following experiment: the
addition of an excess of a 4:1 molar mixture of MeOH/H2O to
an ethyl acetate solution containing 1b and LiBF4 at room
temperature led to the formation of some free ligand, indicative
of catalyst decomposition (see the Supporting Information). In
contrast, no reaction occurred using 2b under the same
conditions.
Table 6. Effect of MeOH/H2O Ratio on MeOH
Dehydrogenation in the Presence of Water Using 1b
a
b
c
MeOH/H2O (M)
time (h)
TON
yield (%)
8:1
6:1
4:1
2:1
1:1
1:2
4
30
500
2780
9500
6000
5000
604
5
28
95
60
50
6
Mechanistic Studies. DFT calculations were performed to
provide further insight into the mechanism of aqueous phase
MeOH dehydrogenation using 1. We employed smaller model
Fe complexes in the calculations by replacing the isopropyl or
cyclohexyl groups of the phosphine ligands with methyl
substituents. It has previously been demonstrated that this
change has only a minor effect on the energetics of PNP
supported Fe complexes.10e The relative free energies reported
below were obtained at the B3LYP/def2-SVP level of DFT and
relate to standard conditions (gaseous species at 298 K and 1
bar). Our chosen level of DFT was validated by comparison to
coupled-cluster single-point results obtained at the extrapolated
basis-set limit, CCSD(T)/CBS(T,Q), for a set of minima and
transition structures representative of the system under study
(see Figure S17). The benchmarking study indicates sufficient
agreement for a qualitative assessment of reaction pathways,
which is our aim here, but with a maximum deviation of about 6
kcal mol−1 for relative electronic energies, our expectations as
to a quantitative description are limited. In our computational
study we explore pathways for full aqueous phase dehydrogen-
ation of MeOH along the four-step reaction sequence depicted
in Scheme 2. Accordingly, we discriminate between four
individual reaction sequences in the following discussion of the
computational results: (A) MeOH dehydrogenation, (B)
hemiacetal formation, (C) methanediol dehydrogenation, and
(D) formate dehydrogenation. Stationary points identified
along the individual routes are denoted correspondingly by
preceding capital letters; please note that the numbering
scheme deviates from that used in the Experimental Section.
The first step in the dehydrogenation of MeOH is the
formation of formaldehyde and 1 equiv of H2. The lowest
energy pathway for this process is shown in Scheme 6. Initially,
an encounter complex A2 is formed via an N···H···O hydrogen
bond between MeOH and the five-coordinate amido species
A1. Subsequently, concerted transfer of the hydrogen atoms
associated with both the C−H and O−H bonds occurs to
generate A4 (corresponding to 4a or 4b discussed above) and
formaldehyde, via an intermediate encounter complex A3. The
barrier for this process via TS-A2 is relatively low (15 kcal
mol−1 relative to MeOH and A1), and this reaction sequence is
analogous to the calculated first step in the conversion of
alcohols to esters and 2 equiv of H2, which was previously
reported by our group using 1a.10e This pathway is also closely
41
38
5.5
2.5
a
Reaction conditions: water (18 μL, 1.0 mmol), appropriate molar
quantity MeOH, 1b (0.03 mol % with respect to water), LiBF4 (0.10
b
mmol, 10 mol %), 10 mL of ethyl acetate, reflux. Time at which no
c
further increase in TON was observed. TON was measured using a
gas buret. Each equivalent of H2 generated is counted as a TON. All
numbers are the average of two runs.
Table 7. Screening of Catalysts for MeOH Dehydrogenation
a
in the Presence of Water
b
c
catalyst
mol % [Fe]
time (h)
TON
yield (%)
1a
1b
C
0.03
0.03
0.03
0.03
0.03
0.01
0.006
42
41
21
39
41
52
94
8200
9500
82
95
2500
25
2a
2b
2a
2a
>9999
>9999
30,000
51,000
>99
>99
>99
50
a
Reaction conditions: water (18 μL, 1.0 mmol), MeOH (160 μL, 4.0
mmol), [Fe], LiBF4 (0.10 mmol, 10 mol %),10 mL of ethyl acetate,
b
reflux. Time at which no further increase in TON was observed.
c
TON measured using a gas buret. Each equivalent of H2 generated is
counted as a TON. All numbers are the average of two runs.
dehydrogenation by a first-row metal, giving 5 times greater
turnover than previous Fe catalysts and 12 times better
turnover than other base-free systems.
It is surprising that the formate complexes 2a and 2b give
slightly superior yields and reach completion more rapidly for
both MeOH dehydrogenation in the absence of water and
aqueous phase MeOH dehydrogenation, even though it is
proposed that they need to access 1a or 1b, respectively, for
productive catalysis to occur. Our explanation for this strange
observation is that the five coordinate species 1 undergo
nonproductive side reactions with MeOH and water at room
temperature. For example, we have already demonstrated the
facile 1,2-addition of MeOH to 1b at low temperature, which
we do not believe leads to dehydrogenation (vide infra) and
results in some decomposition. Thus, in catalysis using the five
coordinate species, these unproductive side reactions deactivate
some of the catalyst before MeOH dehydrogenation can
initiate. In contrast, the formate complexes 2 are less prone to
these side reactions because the Fe center must first undergo
decarboxylation, which typically only occurs at temperatures at
which MeOH dehydrogenation is also operative.10d Further-
more, at the elevated temperatures at which decarboxylation
occurs, 1,2-addition of MeOH is no longer preferred, so once 1
related to that proposed by Grutzmacher and co-workers for
̈
the dehydrogenation of ethanol to acetaldehyde using a Rh
system, although in this case it is proposed that the O−H bond
is cleaved before the C−H bond.14 Yang15 in turn, calculated a
stepwise ionic pathway for the dehydrogenation of ethanol to
acetaldehyde using 1a.16 An alternative pathway for the reaction
of MeOH and A1 to generate formaldehyde and A4, involving
1,2-addition of MeOH across the Fe−N bond to generate an
alkoxide followed by β-hydride elimination, was calculated to be
significantly higher in energy (see Figure S20). However, initial
1,2-addition of MeOH was slightly energetically favored,
consistent with our experimental observation of the alkoxide
2409
DOI: 10.1021/acscatal.5b00137
ACS Catal. 2015, 5, 2404−2415

ACS Catalysis
Research Article
Scheme 6. Calculated Lowest Free Energy Pathway for MeOH Dehydrogenation (A)
Scheme 7. Calculated Pathways for Conversion of Formaldehyde and Water into Methanediol (B)
complex 3b at low temperature (vide supra). The loss of
dihydrogen from A4 to regenerate A1 is mediated by MeOH
and, with a barrier of 23 kcal mol−1, represents the rate-
determining step in the conversion of MeOH to formaldehyde
and H2. We have previously described this process for H2
elimination in detail.10e The barrier for H2 elimination is higher
when mediated by water compared to MeOH (see Figure S21).
In line with expectation, we find that the thermodynamic
favorability of H2 loss from A4 to form A1 varies as a function
of the H2 pressure (see Figure S18), consistent with
experimental results on the stability of 4a/4b.10d
Recently, both our group10e and Azofra et al.17 reported that
there was a large barrier to uncatalyzed hemiacetal formation
from MeOH and formaldehyde. Similarly, the barrier to
uncatalyzed methanediol formation from MeOH and water is
also high (40 kcal mol−1, Scheme 7a). A pathway in which a
second molecule of water acts as a shuttle is significantly lower
in energy, with a barrier of 19 kcal mol−1 for the six-membered
2410
DOI: 10.1021/acscatal.5b00137
ACS Catal. 2015, 5, 2404−2415

ACS Catalysis
Research Article
Scheme 8. Calculated Lowest Free Energy Pathway for Methanediol Dehydrogenation (C)
transition state (Scheme 7b). However, the lowest energy
pathway for methanediol formation is mediated by B1 (Scheme
7c). It involves initial 1,2-addition of water across the Fe−N
bond to generate the hydroxide complex (B3, similar to 3b
discussed under Experimental Section). This species forms an
encounter complex with formaldehyde (B4), which facilitates
the formation of the new C−O and O−H bonds, via a low-
energy transition state (TS-B4, 4 kcal mol−1). The facile
calculated pathway for 1,2-addition of water to B1 is consistent
with our experimental observation that 1 reacts rapidly with
water (Scheme 3), although we were unable to identify any
well-defined products in the experiments. In analogy to our
stabilized by a formate ion. The whole reaction cascade C1 →
C5 occurs without significant activation barriers, and also the
subsequent displacement of the coordinated H2 ligand by
formate to form C6 has a small barrier only (12 kcal mol−1 via
TS-C5). Overall, methanediol dehydrogenation represents a
highly exergonic process (−37 kcal mol−1 relative to C1 or −26
kcal mol−1 relative to A1).
The final step in aqueous phase MeOH dehydrogenation is
the decarboxylation of the formate complex D1, followed by
release of H2 to regenerate D7 (Scheme 9). We have previously
demonstrated that in the absence of LA the barrier for the
decarboxylation of D1 is high,10d consistent with the
experimental observation that a LA is required for this process.
We decided to model the effect of the LA on decarboxylation
using Na+.18 As we have no relevant information on the nature
of the Na+ coordination environment in the rather complex
reaction mixture, we chose Na(H2O)6+ as a model LA to study
its influence on the formate decarboxylation at least in a
qualitative fashion. Our calculations predict that expulsion of
one water ligand from the coordination sphere of Na+ and
results, Grutzmacher and co-workers reported a closely related,
̈
low-barrier pathway for the corresponding reaction mediated
by a Rh system.14
The third step in the overall process is the dehydrogenation
of methanediol (Scheme 8). The initial approach of
methanediol to C1 is similar to that described for MeOH,
with an encounter complex C2, formed via an N···H···O
hydrogen bond. However, the subsequent dehydrogenation
occurs through a stepwise rather than a concerted pathway.
Initially, the amido ligand is protonated by methanediol to form
intermediate C3. Subsequent transfer of the hydrogen atom
associated with one of the C−H bonds of methanediol
generates C4, an encounter complex between formic acid and
the dihydride complex. Formic acid, which is the strongest acid
generated in the reaction cascade, then protonates an Fe−H
bond in C4 to form C5, a cationic molecular H2 complex
+
binding of Na(H2O)5 to the formate ligand in D1 is
thermodynamically favorable by −12 kcal mol−1 (see Figure
S19). This value is almost certainly an overestimation, as
experimentally we do not observe significant changes in the
NMR properties of the formate complexes when LAs are added
and have never isolated a species with a LA coordinated. We
hence assume that in reality, a LA adduct of the formate
complex is rather approximately isoenergetic with the separated
2411
DOI: 10.1021/acscatal.5b00137
ACS Catal. 2015, 5, 2404−2415

ACS Catalysis
Research Article
Scheme 9. Lowest Free Energy Pathways in the Absence (Black) and the Presence of LA (Red) for Formic Acid
Dehydrogenation (D)
LA and the formate in solution. Nevertheless, as we previously
hypothesized,10d Na+ coordination to the formate ligand
stabilizes the negative charge that develops in the decarbox-
ylation step, thereby significantly lowering its activation barrier.
Correspondingly, we find that both transition states, TS-D1
and TS-D2, and the H-bound formate intermediate D2 are
substantially lowered by LA coordination (Scheme 9).19
Subsequent release of H2 from D3 to regenerate D7 is then
again facilitated by MeOH as described before. In the presence
of the LA, the latter step is overall rate limiting for the sequence
D1 → D7. In this scenario, there is a pre-equilibrium involving
reversible CO2 insertion/decarboxylation and the formate
complex D1 represents the resting state during catalysis,
which is consistent with our experimental observations.
steps involved in aqueous phase MeOH dehydrogenation
(Scheme 10 shows the overall catalytic cycle resulting from our
computational study). We have identified the role of the LA in
facilitating decarboxylation, and in principle this effect could be
transferable to other systems that are proposed to decarbox-
ylate via an outer sphere mechanism. Additionally, LAs may
also assist in CO2 insertion reactions into metal hydrides, which
are the microscopic reverse of this outer sphere decarbox-
ylation; however at this stage there are no direct data to support
this proposal. Our calculations demonstrate the difficulty
associated in 1,2-elimination of H2 from 4a/4b, and to design
improved catalysts based on our Fe systems, it is crucial to
reduce the barrier for H2 elimination. This may result in
systems that operate at lower temperature and are more stable,
which is necessary to achieve TONs that are comparable to the
best precious metal systems.
In summary, notwithstanding a number of assumptions made
in our theoretical assessment, which certainly limit the
expectable accuracy in the description of the realistic process,
our calculations provide a qualitative picture of the elementary
2412
DOI: 10.1021/acscatal.5b00137
ACS Catal. 2015, 5, 2404−2415

ACS Catalysis
Research Article
Scheme 10. Overall Catalytic Cycle Computed for Lewis Acid Assisted Methanol Dehydrogenation (Effective Activation
Barriers for Key Elementary Steps in kcal mol−1; B3LYP/def2-SVP)
CONCLUSIONS
EXPERIMENTAL SECTION
■
■
General Methods. Experiments were performed under a
dinitrogen or argon atmosphere in an inert atmosphere glovebox or
using standard Schlenk techniques, unless otherwise noted. Under
standard inert atmosphere glovebox conditions, purging was not
performed between uses of pentane, diethyl ether, benzene, toluene,
and THF; thus, when any of these solvents were used, traces of all
these solvents were in the atmosphere and could be found intermixed
in the solvent bottles. Moisture- and air-sensitive liquids were
transferred by stainless steel cannula on a Schlenk line or in an inert
atmosphere glovebox. Solvents were dried by passage through a
column of activated alumina followed by storage under dinitrogen or
argon. Ethyl acetate, propylene carbonate, and dioxane were dried over
CaH2 and distilled before use. All commercial chemicals were used as
received, except where noted. Lithium hexafluorophosphate, sodium
hexafluorophosphate, potassium hexafluorophosphate, lithium triflate,
sodium triflate, potassium triflate, and sodium tetraphenylborate were
purchased from Fisher Scientific Co. Sodium chloride, lithium
chloride, potassium chloride, cesium chloride, calcium chloride,
lithium tetraphenylborate, sodium tetraphenylborate, and potassium
tetraphenylborate were purchased from Acros. Deuterated solvents
were obtained from Cambridge Isotope Laboratories. d8-Toluene was
dried over sodium metal and vacuum-transferred prior to use.
Literature procedures were used to prepare sodium tetrakis[(3,5-
We have demonstrated that a family of PNP-supported Fe
complexes generates highly active catalysts for the dehydrogen-
ation of MeOH. In the absence of water, these catalysts rapidly
convert MeOH to methyl formate and H2. Although a LA is not
required for this reaction, the catalytic system that gives the
highest TON (approximately 20 000) requires a LA for
activation. In the presence of water, our Fe complexes fully
dehydrogenate MeOH to H2 and CO2, provided a LA co-
catalyst is present. For this reaction our best Fe/LiBF4 system
gives a TON of approximately 51 000, which is the highest
reported to date for a homogeneous first-row transition metal
catalyst or a system that does not require a Brønsted base.
There are two major advantages to using a LA instead of a
Brønsted base: (i) the reaction conditions are milder, which
could both extend catalyst lifetime and increase the potential
range of improved catalysts that could be developed in the
future; and (ii) the loading of LA required is significantly lower
than the loading of base typically used. The mechanism of
MeOH dehydrogenation is proposed to involve four steps: (i)
initial dehydrogenation of MeOH to formaldehyde; (ii)
reaction of water with formaldehyde to form methanediol;
(iii) dehydrogenation of methanediol to form an Fe formate
species and H2; and (iv) decarboxylation of the Fe formate
species to release CO2 and the final equivalent of H2. The LA is
required to facilitate the decarboxylation of the Fe formate
species, whereas the ability of the PNP ligand to undergo
bifunctional reactivity is crucial to many of the elementary steps
in the reaction pathway. In future work, we will continue to
explore the potential of Fe complexes supported by PNP
ligands to release H2 from small molecules with the potential to
be used for chemical hydrogen storage.
trifluoromethyl)phenyl]borate (NaBArF ),20 1a,10d 1b,10d 2a,10d 2b,10d
4
and C.9 NMR spectra were recorded on Bruker AMX-400, AMX-500,
and AMX-600 spectrometers at ambient probe temperatures, unless
otherwise noted.
Computational Details. DFT calculations were performed using
the Gaussian 09 program package.21 The B3LYP hybrid functional,22
as implemented in Gaussian 09, was used in combination with the
def2-SVP basis set.23 Unscaled zero-point vibrational energies as well
as thermal and entropic corrections were obtained from computed
Hessians using the standard procedures implemented in Gaussian 09
and were used to obtain Gibbs free energies at 298.15 (1 bar
atmospheric pressure).
2413
DOI: 10.1021/acscatal.5b00137
ACS Catal. 2015, 5, 2404−2415

ACS Catalysis
Research Article
Gas Chromatography. Gas chromatography experiments were
performed on a Buck Scientific 910 gas chromatograph with FID/
TCD and methanizer. The system uses N2 as a carrier gas and allows
for the determination of the following gases and detection limits: H2 ≥
100 ppm, CO ≥ 1 ppm, and CO2 ≥ 1 ppm.
(3) (a) Navarro, R. M.; Pena, M. A.; Fierro, J. L. Chem. Rev. 2007,
107, 3952−3991. (b) Boddien, A.; Loges, B.; Junge, H.; Beller, M.
ChemSusChem 2008, 1, 751−758. (c) Eberle, U.; Felderhoff, M.;
Schuth, F. Angew. Chem., Int. Ed. 2009, 48, 6608−6630. (d) Staubitz,
A.; Robertson, A. P.; Manners, I. Chem. Rev. 2010, 110, 4079−4124.
(e) Grasemann, M.; Laurenczy, G. Energy Environ. Sci. 2012, 5, 8171−
8181. (f) Dalebrook, A. F.; Gan, W.; Grasemann, M.; Moret, S.;
Laurenczy, G. Chem. Commun. 2013, 49, 8735−8751.
Representative Procedure for Catalytic MeOH Dehydrogen-
ation in the Presence and Absence of Water. In an inert
atmosphere glovebox, a 50 mL Schlenk flask was loaded with the
appropriate catalyst, MeOH, additive (LA), water (for aqueous phase
reactions), and the desired solvent. The Schlenk flask was sealed with a
glass stopper and removed from the inert atmosphere glovebox and
attached to a gas buret setup and reflux condenser (see Figure S1).
The gas buret and tubing was subjected to three vacuum/N2 purge
cycles and allowed to equilibrate. For aqueous phase reactions the U-
tube trap was cooled with liquid nitrogen. The solution flask was then
lowered into an oil bath preheated to the desired temperature upon
which gas evolution began immediately. The change in water level in
the gas buret (Vobs) was used to determine turnover using previously
reported methods (see Figure S1).8c,10d Each equivalent of H2
produced was taken to be one turnover. For aqueous phase reactions,
upon completion of the reaction, the U-tube was removed from the
liquid nitrogen bath and the CO2 gas evolved was measured by the
buret to be a third of total turnover. In all cases, a blank reaction was
run in which no catalyst was added to the solution. The volume of gas
obtained from this reaction (trace solvent and MeOH) was recorded
(4) Carmo, M.; Fritz, D. L.; Mergel, J.; Stolten, D. Int. J. Hydrogen
Energy 2013, 38, 4901−4934.
(5) (a) Olah, G. A. Angew. Chem., Int. Ed. 2005, 44, 2636−2639.
(b) Olah, G. A. Angew. Chem., Int. Ed. 2013, 52, 104−107.
(6) (a) Palo, D. R.; Dagle, R. A.; Holladay, J. D. Chem. Rev. 2007,
107, 3992−4021. (b) Gartner, F.; Losse, S.; Boddien, A.; Pohl, M.-M.;
̈
Denurra, S.; Junge, H.; Beller, M. ChemSusChem 2012, 5, 530−533.
(c) Okamoto, Y.; Ida, S.; Hyodo, J.; Hagiwara, H.; Ishihara, T. J. Am.
Chem. Soc. 2011, 133, 18034−18037.
(7) (a) Itagaki, H.; Saito, Y.; Shinoda, S. J. Mol. Catal. 1987, 41, 209−
220. (b) Morton, D.; Cole-Hamilton, D. J. J. Chem. Soc., Chem.
Commun. 1987, 248−249.
(8) (a) Rodríguez-Lugo, R. E.; Trincado, M.; Vogt, M.; Tewes, F.;
Santiso-Quinones, G.; Grutzmacher, H. Nat. Chem. 2013, 5, 342−347.
̈
(b) Nielsen, M.; Alberico, E.; Baumann, W.; Drexler, H.-J.; Junge, H.;
Gladiali, S.; Beller, M. Nature 2013, 495, 85−89. (c) 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.
(d) Hu, P.; Diskin-Posner, Y.; Ben-David, Y.; Milstein, D. ACS Catal.
2014, 2649−2652. (e) Monney, A.; Barsch, E.; Sponholz, P.; Junge,
H.; Ludwig, R.; Beller, M. Chem. Commun. 2014, 50, 707−709.
(f) Campos, J.; Sharninghausen, L. S.; Manas, M. G.; Crabtree, R. H.
Inorg. Chem. 2015, in press, DOI: 10.1021/ic502521c.
as Vblank
.
ASSOCIATED CONTENT
* Supporting Information
■
S
The following files are available free of charge on the ACS
Publications website at DOI: 10.1021/acscatal.5b00137.
(9) Koehne, I.; Schmeier, T. J.; Bielinski, E. A.; Pan, C. J.; Lagaditis,
Further experimental details, X-ray information for
(CyPNHP)Fe(CO)2, and details of DFT calculations
(PDF)
P. O.; Bernskoetter, W. H.; Takase, M. K.; Wurtele, C.; Hazari, N.;
̈
Schneider, S. Inorg. Chem. 2014, 53, 2133−2143.
(10) (a) Chakraborty, S.; Dai, H.; Bhattacharya, P.; Fairweather, N.
T.; Gibson, M. S.; Krause, J. A.; Guan, H. J. Am. Chem. Soc. 2014, 136,
7869−7872. (b) Chakraborty, S.; Brennessel, W. W.; Jones, W. D. J.
Am. Chem. Soc. 2014, 136, 8564−8567. (c) Bornschein, C.;
Werkmeister, S.; Wendt, B.; Jiao, H.; Alberico, E.; Baumann, W.;
Junge, H.; Junge, K.; Beller, M. Nat. Commun. 2014, 5, 4111.
(d) Bielinski, E. A.; Lagaditis, P. O.; Zhang, Y.; Mercado, B. Q.;
CIF format data (CIF)
AUTHOR INFORMATION
Corresponding Authors
*(W.H.B.) E-mail: wesley_bernskoetter@brown.edu.
*(N.H.) E-mail: nilay.hazari@yale.edu.
■
Wurtele, C.; Bernskoetter, W. H.; Hazari, N.; Schneider, S. J. Am.
̈
Chem. Soc. 2014, 136, 10234−10237. (e) Chakraborty, S.; Lagaditis, P.
*(M.C.H.) E-mail: max.holthausen@chemie.uni-frankfurt.de.
Notes
The authors declare no competing financial interest.
O.; Forster, M.; Bielinski, E. A.; Hazari, N.; Holthausen, M. C.; Jones,
̈
W. D.; Schneider, S. ACS Catal. 2014, 4, 3994−4003.
(11) (a) Smith, T. A.; Aplin, R. P.; Maitlis, P. M. J. Organomet. Chem.
1985, 291, c13−c14. (b) Yang, L.-C.; Ishida, T.; Yamakawa, T.;
Shinoda, S. J. Mol. Catal. A: Chem. 1996, 108, 87−93. (c) Yamakawa,
T.; Hiroi, M.; Shinoda, S. J. Chem. Soc., Dalton Trans. 1994, 2265−
2269. (d) Shinoda, S.; Yamakawa, T. J. Chem. Soc., Chem. Commun.
1990, 1511−1512.
ACKNOWLEDGMENTS
■
W.H.B. and N.H. thank the National Science Foundation for
support through Grant CHE-1240020, a Center for Chemical
Innovation. M.C.H. acknowledges support through the COST
Action 1205 (CARISMA). Calculations were performed at the
Center for Scientific Computing (CSC) Frankfurt on the
FUCHS and LOEWE-CSC high-performance compute clus-
ters. N.H. and W.H.B. are fellows of the Alfred P. Sloan
Foundation, and N.H. is a Camille and Henry Dreyfus
Foundation Teacher Scholar. We are grateful to Professor
Sven Schneider for valuable discussions, Steven Ahn for help
with GC, and Dr. Brandon Mercado for assistance with
crystallography.
(12) Qu, S.; Dai, H.; Dang, Y.; Song, C.; Wang, Z.-X.; Guan, H. ACS
Catal. 2014, 4377−4388.
(13) In the experiment using a low catalyst loading of 2a (0.001 mol
%) it was possible to reduce the amount of LiBF4 from 10 to 1 mol %
with no decrease in catalytic activity.
(14) Zweifel, T.; Naubron, J.-V.; Buttner, T.; Ott, T.; Grutzmacher,
̈
̈
H. Angew. Chem., Int. Ed. 2008, 47, 3245−3249.
(15) Yang, X. ACS Catal. 2013, 3, 2684−2688.
(16) For a more detailed discussion of the differences between Yang’s
proposed pathway and our pathway see ref 10e.
́
(17) Azofra, L. M.; Alkorta, I.; Elguero, J.; Toro-Labbe, A. J. Phys.
Chem. A 2012, 116, 8250−8259.
(18) Even though the best catalytic activity was observed using Li+ as
the LA, we computationally modeled the LA as Na+. This is due to the
difficulties in modelling the aggregation of Li+, which can lead to the
presence of clusters in solution.
REFERENCES
■
(1) (a) Dincer, I. Energy Policy 1999, 27, 845−854. (b) Lewis, N. S.;
Nocera, D. G. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 15729−15735.
(2) (a) Bockris, J. O. M. Science 1972, 176, 1323. (b) Bockris, J. O.
M. Int. J. Hydrogen Energy 2002, 27, 731−740. (c) Midilli, A.; Ay, M.;
Dincer, I.; Rosen, M. A. Renewable Sust. Energy Rev. 2005, 9, 255−271.
(19) The decarboxylation of D1 to form D3 is thermodynamically
disfavored. This indicates that CO2 insertion into D3 is thermody-
2414
DOI: 10.1021/acscatal.5b00137
ACS Catal. 2015, 5, 2404−2415

ACS Catalysis
Research Article
namically favored and that these complexes may act as catalysts for
CO2 hydrogenation to formate. This is a topic of ongoing investigation
in our laboratories.
(20) Yakelis, N. A.; Bergman, R. G. Organometallics 2005, 24, 3579−
3581.
(21) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci,
B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H.
P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.;
Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima,
T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.;
Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin,
K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.;
Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega,
N.; Millam, N. J.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.;
Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.;
Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.;
Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.;
̈
Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.;
Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09,
revision D.02; Gaussian, Inc.: Wallingford, CT, USA, 2009.
(22) (a) Lee, C.; Parr, R. G.; Yang, W. Phys. Rev. B 1988, 37, 785−
789. (b) Becke, A. D. Phys. Rev. A: At., Mol., Opt. Phys. 1988, 38,
3098−3100. (c) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652.
(23) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7,
3297−3305.
2415
DOI: 10.1021/acscatal.5b00137
ACS Catal. 2015, 5, 2404−2415