Hydrogenation of carboxylic acids catalyzed by half-sandwich complexes of iridium and rhodium

Article abstract of DOI:10.1021/ja408149n

A series of half-sandwich Ir and Rh compounds are demonstrated to be competent catalysts for the hydrogenation of carboxylic acids under relatively mild conditions. Of the structurally diverse group of catalysts tested for activity, a Cp*Ir complex supported by an electron-releasing 2,2′-bipyridine ligand was the most active. Higher activity was achieved with employment of Bronsted or Lewis acid promoters. Mechanistic studies suggest a possible reaction pathway involving activated carboxylic acid substrates. The hydrogenation reaction was shown to be general to a variety of aliphatic acids.

Full text of DOI:10.1021/ja408149n

Communication
pubs.acs.org/JACS
Hydrogenation of Carboxylic Acids Catalyzed by Half-Sandwich
Complexes of Iridium and Rhodium
†
,‡
†
,†
Timothy P. Brewster, Alexander J. M. Miller,* D. Michael Heinekey, and Karen I. Goldberg*
†
Department of Chemistry, University of Washington, Seattle, Washington 98195-1700
‡
Department of Chemistry, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-3290
*
S Supporting Information
Scheme 1. Reductive Coupling Reaction Sequence
ABSTRACT: A series of half-sandwich Ir and Rh
compounds are demonstrated to be competent catalysts
for the hydrogenation of carboxylic acids under relatively
mild conditions. Of the structurally diverse group of
catalysts tested for activity, a Cp*Ir complex supported by
an electron-releasing 2,2′-bipyridine ligand was the most
active. Higher activity was achieved with employment of
Brønsted or Lewis acid promoters. Mechanistic studies
suggest a possible reaction pathway involving activated
carboxylic acid substrates. The hydrogenation reaction was
shown to be general to a variety of aliphatic acids.
We began our investigation of direct hydrogenation of
carboxylic acids by 1 utilizing acetic acid as substrate. Glacial
acetic acid solutions containing 2 mM 1 were pressurized with
12
3
0 atm H and heated to 120 °C. After 65 h, ethyl acetate was
2
observed to be the major product by gas chromatography
(approximately 42:1 ethyl acetate/ethanol vs dioxane internal
standard), as outlined in Scheme 1. Reactions in glacial acetic
acid consistently produced reductively coupled ester products
with high selectivity (>95%). No additional products are
observed under these conditions.
eduction of carboxylic acids to alcohols is a fundamental
R
transformation in organic chemistry. Traditionally, this
reaction has been carried out using stoichiometric quantities of
strong reducing agents such as lithium aluminum hydride or
1
lithium triethylborohydride. The use of stoichiometric
amounts of strong reductants is undesirable due to issues of
reagent compatibility and the poor atom economy of the
overall transformation.
Mechanistic studies were undertaken to understand and
optimize the reaction. Relative reaction rates were assessed
13
from the turnover number (TON) after 65 h (up to 30%
conversion). Measurements at relatively low conversion are
required to avoid rate effects due to inhibition by water (vide
An attractive alternative would be the direct hydrogenation
of the carboxylic acid by dihydrogen. Unfortunately, carboxylic
acids are among the most difficult carbonyl substrates to
hydrogenate due to the low electrophilicity of the carbonyl
14
infra). Direct temperature dependence was observed with the
15
TON increasing from 73 at 60 °C to 425 at 120 °C after 65 h.
2
carbon. Recently, a biocatalytic route was reported to catalyze
Unfortunately, at temperatures above 120 °C, deposition of
iridium black on the reactor liners and irreproducible rates
demonstrated catalyst decomposition. The heterogeneous
decomposition products were tested and found to be inactive
3
acid hydrogenation under mild conditions. However, only two
homogeneous, transition-metal based systems for acid hydro-
genation have been reported, both of which operate under
4
−9
15
relatively forcing conditions (≥160 °C, ≥100 atm H ).
for catalyzing the hydrogenation reaction.
2
We recently reported that formic acid can be reduced to
methanol via an apparent transfer hydrogenation mechanism in
which formic acid acts as both the hydrogen donor and
The rate of hydrogenative coupling of acetic acid was found
to be first order in both substrate and catalyst at 120 °C, and
saturation behavior was observed with respect to the pressure
1
0
of H (Figure S6, Figure 1). Addition of water to the reaction
acceptor. Mechanistic studies suggested that catalyst [Cp*Ir-
2
dramatically slowed the rate. In 8.7 M acetic acid only 26
turnovers occurred after 65 h. Conversely, in the presence of
(
bpy)(OH )][OTf] (bpy = 2,2′-bipyridine, OTf = trifluor-
2
2
omethanesulfonate) (1) was converted to [Cp*Ir(bpy)(H)]-
16
added HBF4 (8.7 M acetic acid in a 1:1 (w:w) H O/HBF
[
OTf] (2) during turnover; hydride transfer from 2 to
2
4
solution), the rate increased markedly to 1338 turnovers after
5 h, as shown in Figure 1. Under higher concentrations of
protonated formic acid appears to be an important step in
the reaction. Hydride 2 has also been implicated in hydro-
6
water and protons the distribution of products began to shift;
diethyl ether was now observed as a minor product, and more
genation reactions using H directly, notably in the hydro-
2
11
genation of CO to formic acid. It seemed likely, therefore,
2
ethanol was present (8:13:1 ratio of EtOH:EtOAc:Et O). This
that replacement of formic acid by H2 might facilitate
hydrogenation of a wide range of carboxylic acid substrates
while avoiding competitive decomposition pathways. We report
that 1 and related homogeneous catalysts hydrogenate aliphatic
carboxylic acids under the mildest conditions yet reported.
2
is unsurprising, as esterification to ethyl acetate should be less
Received: August 6, 2013
©
XXXX American Chemical Society
A
dx.doi.org/10.1021/ja408149n | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
Figure 1. Dependence of the hydrogenation of acetic acid by 1 on H pressure and pH. (A) 2 mM 1, 2 mL glacial AcOH, 120 °C, 65 h. (B) 2 mM 1,
2
2
mL 8.7 M AcOH (aq) at given pH16 (adjusted with HBF ), 120 °C, 65 h.
4
Scheme 2
12
Figure 2. Precatalysts screened for AcOH hydrogenation.
acid solvent to generate the hydrogenative coupling product
ethyl acetate (major product) or acid-catalyzed homocoupling
to generate diethyl ether (minor product).
favored in the presence of excess water. The diethyl ether is
presumably formed through an acid-catalyzed dehydrative
homocoupling of ethanol.
A variety of homogeneous metal catalysts for the hydro-
genolysis of esters under mild conditions have been
A reaction mechanism consistent with the above data is
shown in Scheme 2. An initial equilibrium between aqua
complex 1, acetic acid complex [Cp*Ir(bpy)(HOAc)][OTf]2,
and dihydrogen complex [Cp*Ir(bpy)(H )][OTf] is estab-
2
,21−23
reported.
In separate experiments, methoxy-substituted
catalyst 5 (Figure 2) was also demonstrated to be competent
for hydrogenolysis of ethyl acetate at a rate comparable to that
2
2
1
5
17
15
lished. Such dihydrogen complexes are highly acidic, so
of acetic acid hydrogenation under similar conditions. Thus,
rapid deprotonation to form 2 would be expected. This pre-
at the low conversions of acetic acid used in our kinetic studies,
hydrogenolysis of the ester product is not expected to be a
significant contributor.
equilibrium is consistent with the observed H saturation
2
kinetics. At sufficiently high pressure, formation of the
dihydrogen complex is favored, leading to rapid formation of
To probe the intermediacy of iridium monohydride complex
1
2
2
and rate-limiting hydride transfer that does not depend on H2
2, the previously reported hydride was prepared and
employed as the precatalyst. Interestingly, the reaction rate
under standard conditions was much slower when 2 was used
instead of 1. We hypothesized that complex 1 effectively acts as
pressure. After protonation of acetic acid, 2 transfers a hydride
to the activated acetic acid molecule, producing acetaldehyde,
and then rejoins the equilibria between dicationic Ir species.
The proposal that the reaction involves protonated acetic acid
is consistent with our observations involving aqueous dilution;
the concentration of protonated acetic acid drops significantly
2+
+
a strong acid, as [Cp*Ir(bpy)(H )] quickly releases H into
2
solution to generate hydride 2 (Scheme 2). Accordingly,
reactions catalyzed by a combination of 2 and HOTf (1 μL, 3
+
15
in the presence of water which is more basic (pK (H O (aq))
equiv) gave similar rates to reactions catalyzed by 1. These
a
3
1
8
+
19
=
−1.7, pK (CH CO H (aq)) = −6.1). Further, addition
results confirm the viability of complex 2 as an intermediate in
a
3
2
2
of strong acid accelerates the reaction.
the hydrogenation reaction.
The known iridium chloride complex [Cp*Ir(bpy)Cl][Cl]
12
The acetaldehyde produced after one hydrogenation cycle
would then be converted to ethanol following a similar
mechanism. No aldehydes (or aldehyde hydrates) were
observed, consistent with previous reports of rapid (transfer)
(3) notably shows no catalytic activity for acetic acid
hydrogenation. This result is in contrast to the case of formic
acid disproportionation where complex 3 was shown to be a
20
10
hydrogenation of aldehydes under even milder conditions.
competent catalyst2.4 The relatively nonpolar glacial acetic acid
Depending on conditions, the initial hydrogenation product,
solvent (ε = 6.1 ) may inhibit chloride dissociation and
ethanol, can undergo either Fischer esterification with the acetic
prevent formation of the key hydride intermediate. The
B
dx.doi.org/10.1021/ja408149n | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
9,30
2
Table 1. Comparison of Acetic Acid Hydrogenation
should render the dihydrogen complex less acidic.
For Ir,
a,b
Catalysts
the saturation kinetics of Figure 1A suggest that above ∼30 atm
H the limiting step involves hydride transfer, which would be
accelerated by more strongly donating ligands that render
hydride 2 a stronger H donor. For Rh catalyst 13, saturation
kinetics were not observed up to a H pressure of 50 atm
Figure S7 in the SI). In this case, formation of the Rh−H
complex is likely involved in the rate law. The aforementioned
slowing of catalytic rates with more donating ligands could then
be attributed to a lowering of the concentration of the Rh
dihydrogen species or inhibiting the proton loss from the Rh
dihydrogen complex.
The rate of formation of the metal hydride species will
depend on the pre-equilibrium between the solvento and
dihydrogen complexes. Efforts to observe ligand effects on the
position of the pre-equilibrium, and thus the effect on catalytic
rate, are currently underway. Consistent with our mechanistic
proposal, preliminary data using complex 1 show two
observable species in solution under H2 pressures below
saturation: the solvento complex and a second species formed
2
Cat.
TON ± dev.
Cat.
TON ± dev.
e
1
2
425 ± 8
41 ± 12
8
9
40
−
e
9
2
c
e
15
2
503 ± 34
777 ± 15
1637 ± 25
615 ± 50
389 ± 56
12
13
14
15
21
(
5
96 ± 5
56 ± 1
125 ± 23
trace
d
5
6
7
none
a
For results of all catalysts shown in Figure 2, see Table S1 in the SI.
Average of three trials with calculated standard deviation; 2 mM
b
c
catalyst in glacial acetic acid, 27 atm H , 120 °C, 65 h. 5 mM HOTf.
2
d
e
1
6 mM Sc(OTf) added. Average of two trials.
3
12
chloride-bridged dimer [Cp*IrCl2]2 (4) was also tested, but
only traces of product and large amounts of Ir black were
observed.
The above mechanistic studies suggest that, under sufficient
reversibly upon addition of H (Figure S10 in the SI).
As mentioned previously, our mechanistic proposal suggests
that, at pressures where the reaction rate is independent of
H pressure, hydride transfer is involved in the rate-limiting
2
2
step. Thus, increasing the hydricity of this intermediate should
increase the rate of reaction. A series of structurally related
catalysts with varying electronic properties were therefore
investigated (Figure 2). All catalysts were tested for activity in
[
H ], hydride transfer to a protonated substrate is rate limiting.
2
We have already seen that increasing the proton concentration
increases the reaction rate. By analogy, Lewis acid activation of
the carbonyl could be expected to accelerate the reaction.
Indeed, addition of NaOTf, Zn(OTf) , or Sc(OTf) resulted in
glacial acetic acid at 120 °C under 27 atm H (above the H2
2
saturation pressure for 1), mild conditions compared to those
4
−9
of related previous literature reports.
The results are
2
3
a rate enhancement over reactions lacking these promoters (for
a full list of screened Lewis acids see Table S2 in SI). A specific
interaction of the Lewis acid with the carbonyl is implied by the
complete suppression of the acid effect when 15-crown-5 was
added to a reaction containing NaOTf. Under these conditions,
turnover returned to the level observed in the absence of Lewis
acids.
summarized in Table 1. Methoxy-substituted catalyst 5, and
12
hydroxy-substituted catalyst 6 (Figure 2) showed markedly
increased reaction rates when compared to bpy itself, consistent
with our hypothesis that hydride donor ability is an important
factor. Phenanthroline complex 7 showed activity nearly
identical to that of 1. In contrast, catalyst 8 featured electron-
withdrawing methyl ester groups and showed no activity.
Replacing Cp* with the unsubstituted Cp ligand (9) also led to
a decrease in rate. No decomposition to Ir black was observed
in any of these reactions.
We also investigated hydrogenation of other carboxylic acids.
An experiment involving catalyst 1 in a mixture of 3 M formic
acid (aq) and 3 M acetic acid (aq) yielded only methanol and
methyl ester products, suggesting that carboxylic acids with
shorter aliphatic carbon chains react more rapidly (consistent
Noticing that electron-rich complexes were better catalysts,
we sought other viable catalysts supported by strong donor
4
12
with previous observations). Indeed, using our most active
ligands. The reported phenylpyridine complex 10 and bis(N-
heterocyclic carbene) complex 11 seemed promising, but these
complexes showed little activity at 120 °C, and a large amount
of iridium black was observed to form in these reactions.
Half-sandwich complexes of alternative metals were also
catalyst 5, propionic acid and butyric acid were each
hydrogenated in 3:1 H O/HBF at diminished rates compared
2
4
to acetic acid (Table 2). Reaction selectivity also changes from
12
Table 2. Comparing Reactivity of Aliphatic Carboxylic Acids
explored as catalysts for this reaction. First, an analogous half-
sandwich Ru complex, 12, was investigated. While similar
complexes are highly active toward asymmetric hydrogenation
a
b
Substrate
Aqueous TON ± dev.
AcOH selectivity (%)
2
5−28
acetic acid
68 ± 3
37 ± 1
21 ± 3
100
87
of ketones,
hydrogenation. In contrast, the Rh complex 13, [Cp*Rh(bpy)-
OH )][OTf] , was found to be a competent precatalyst for
12 showed no activity for acetic acid
propionic acid
butyric acid
91
(
2
2
a
3
.5 mmol substrate and 4 μmol 5 in 2 mL 3:1 (v:v) H O/HBF , 30
this reaction, albeit at a rate slower than that of its iridium
2
4
b
atm H , 120 °C, 18 h (three runs per substrate). 2 mM 5, 20 mM
2
analogue. [Cp*Rh(bpy-OMe)(OH )][OTf] , 14, and [Cp*Rh-
2
2
NaOTf in 2 mL 1:1 (mol:mol) AcOH/substrate, 30 atm H , 120 °C,
2
(
bpy-COOMe)(OH )][OTf] , 15, were also found to be active
2 2
1
8 h (three runs per mixture). Selectivity calculated from the ratio of
precatalysts. Surprisingly, Rh and Ir exhibit opposite trends in
terms of how the ligand-donating ability affects the reactivity of
the catalyst. Rh catalyst 14, with electron-donating methoxy
groups, showed the lowest activity while catalyst 15, with
electron-withdrawing ester groups, exhibited the highest activity
hydrogenated acetic acid to hydrogenated substrate.
Scheme 3
(
Table S1 in the SI).
This difference between Rh and Ir may be explained by a
change in the resting state in the mechanism proposed in
Scheme 2. For both metals, more strongly donating ligands
C
dx.doi.org/10.1021/ja408149n | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
.5:1 ester/alcohol for acetic acid to 1:2 ester/alcohol for
Communication
2
REFERENCES
■
propionic and butyric acids. The observed decrease in reactivity
is consistent with our proposed mechanism; as the size of the
carboxylic acid increases from C to C , the carboxylate carbon
(
1) Seyden-Penne, J. Reductions by the Alumino- and Borohydrides in
Organic Synthesis, 2nd ed.; Wiley: New York: 1997.
(2) Dub, P. A.; Ikariya, T. ACS Catal. 2012, 2, 1718.
(3) Ni, Y.; Hagedoorn, P.-L.; Xu, J.-H.; Arends, I. W. C. E.;
Hollmann, F. Chem. Commun. 2012, 48, 12056.
1
4
would be expected to become more electron-rich and thus less
susceptible to nucleophilic attack from the iridium hydride. The
magnitude of the electronic effect would likely diminish with
increasing aliphatic chain size, but this could not be directly
(
4) Bianchi, M.; Menchi, G.; Francalanci, F.; Piacenti, F.; Matteoli,
U.; Frediani, P.; Botteghi, C. J. Organomet. Chem. 1980, 188, 109.
(5) Salvini, A.; Frediani, P.; Giannelli, C.; Rosi, L. J. Organomet.
Chem. 2005, 690, 371.
31
verified due to substrate solubility. The relative rates of
aliphatic carboxylic acid hydrogenation were further probed
through competition experiments. Equimolar mixtures of acetic
(
1
6) Rosi, L.; Frediani, M.; Frediani, P. J. Organomet. Chem. 2010, 695,
314.
acid and C or C acids were subjected to 30 atm H in the
3
4
2
(7) Geilen, F. M. A.; Engendahl, B.; Harwardt, A.; Marquardt, W.;
presence of 2 mM 5 and 20 mM NaOTf for 18 h at 120 °C.
The trend in relative rates of hydrogenation observed in this
experiment reflected the activity previously observed as
displayed in the ratio of observed hydrogenation products
Klankermayer, J.; Leitner, W. Angew. Chem., Int. Ed. 2010, 49, 5510.
(8) Geilen, F. M. A.; Engendahl, B.; Holscher, M.; Klankermayer, J.
R.; Leitner, W. J. Am. Chem. Soc. 2011, 133, 14349.
9) Wesselbaum, S.; vom Stein, T.; Klankermayer, J.; Leitner, W.
Angew. Chem., Int. Ed. 2012, 51, 7499.
10) Miller, A. J. M.; Heinekey, D. M.; Mayer, J. M.; Goldberg, K. I.
Angew. Chem., Int. Ed. 2013, 52, 3981.
11) Ogo, S.; Kabe, R.; Hayashi, H.; Harada, R.; Fukuzumi, S. Dalton
Trans. 2006, 4657.
12) Full synthetic details for new and previously reported catalysts
̈
(
(
6
Table 2). Under these conditions, acetic acid was reduced with
-fold selectivity over propionic acid and 10-fold selectivity
(
relative to butyric acid.
(
Finally, hydrogenation of levulinic acid was examined in
order to make direct comparisons to the system recently
(
7
,8
reported by Leitner. Using an in situ generated Ru tris-
are available in the SI.
phosphine catalyst (0.1 mol % catalyst in 1,4-dioxane, 100 atm
(13) Turnover number is based on combined yield of all
hydrogenation products. Ethanol and ethyl acetate yield one turnover.
Diethyl ether yields two turnovers.
H , 160 °C, 18 h), they reported 95% yield of 1,4-pentanediol
2
(
PDO), 3% γ-valerolactone (GVL, Scheme 3), and no 2-
7
(14) Turnover frequencies after 18 h were similar to turnover
methyltetrahydrofuran (MeTHF). With the addition of 1 mol
p-toluenesulfonic acid, the products shifted to 1% PDO, 58%
frequencies after 65 h (catalyst 5: Table 1 and Table S2, SI).
(15) See SI for further details.
%
32
GVL, and 39% MeTHF. At a lower temperature and pressure
(
16) The ionic strength at this acid concentration is too high for
accurate pH measurement. These are meant as qualitative pH values.
17) Vogt, M.; Pons, V.; Heinekey, D. M. Organometallics 2005, 24,
832.
18) Starkey, R.; Norman, J.; Hintze, M. J. Chem. Educ. 1986, 63, 473.
(19) Goldfarb, A. R.; Mele, A.; Gutstein, N. J. Am. Chem. Soc. 1955,
with catalyst 5 (0.08 mol % catalyst, 8 mol % HOTf, 30 atm H2,
1
20 °C, 18 h in 1,2-dimethoxyethane), we observed complete
(
1
(
consumption of levulinic acid, a 90% yield of partially reduced
GVL, and a small amount (10% yield) of fully reduced MeTHF.
Notably, no dehydration products were observed in our system.
In conclusion, we have demonstrated a novel system for
catalytic hydrogenation of a variety of carboxylic acids. The
proposed reaction mechanism (based on experimental
evidence) guided optimization of the reaction, including the
use of Lewis acid additives. The optimized catalyst system
exhibits activity similar to those of previously published systems
under significantly milder conditions. Extension of this
reactivity is currently under investigation.
77, 6194.
(20) Abura, T.; Ogo, S.; Watanabe, Y.; Fukuzumi, S. J. Am. Chem. Soc.
2003, 125, 4149.
(21) Grey, R. A.; Pez, G. P.; Wallo, A.; Corsi, J. J. Chem. Soc., Chem.
Commun. 1980, 783.
(
(
(
22) Teunissen, H. T.; Elsevier, C. J. Chem. Commun. 1997, 667.
23) Teunissen, H. T. Chem. Commun. 1998, 1367.
24) Gordon, A. J.; Ford, R. A. The Chemist’s Companion: A
Handbook of Practical Data, Techniques, and References; John Wiley and
Sons: New York, 1972.
ASSOCIATED CONTENT
(25) Hashiguchi, S.; Fujii, A.; Takehara, J.; Ikariya, T.; Noyori, R. J.
■
Am. Chem. Soc. 1995, 117, 7562.
*
S
Supporting Information
(
(
26) Noyori, R.; Hashiguchi, S. Acc. Chem. Res. 1997, 30, 97.
27) Samec, J. S. M.; Backvall, J.-E.; Andersson, P. G.; Brandt, P.
Full experimental details. This material is available free of
charge via the Internet at http://pubs.acs.org.
̈
Chem. Soc. Rev. 2006, 35, 237.
(
28) Ikariya, T.; Blacker, A. J. Acc. Chem. Res. 2007, 40, 1300.
AUTHOR INFORMATION
■
(29) Ng, S. M.; Fang, Y. Q.; Lau, C. P.; Wong, W. T.; Jia, G.
Corresponding Authors
ajmm@email.unc.edu
Organometallics 1998, 17, 2052.
(
(
30) Jia, G.; Morris, R. H. J. Am. Chem. Soc. 1991, 113, 875.
31) Preliminary experiments investigating fluorinated alcoholic or
goldberg@chem.washington.edu
ethereal solvents met with difficulties due to solvent decomposition
and/or esterification with the carboxylic acid starting material. For
example, in the case of 1,2-dimethoxyethane, large amounts of methyl
ester were observed.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
(32) When a sulfonic acid containing ionic liquid rather than p-
toluenesulfonic acid was added to the reaction mixture, 87% yield of
MeTHF was obtained in addition to 1% PDO and 5% GVL.
■
The authors thank Loren Kruse for assistance with mass
spectrometry and Prof. James M. Mayer for many fruitful
discussions. This work was supported by the Camille and
Henry Dreyfus Postdoctoral Program in Environmental
Chemistry (T.P.B., A.J.M.M., K.I.G.) and by NSF under the
CCI Center for Enabling New Technologies through Catalysis
(
CENTC), CHE-1205189.
D
dx.doi.org/10.1021/ja408149n | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX