Pd-catalyzed transfer hydrogenolysis of primary, secondary, and tertiary benzylic alcohols by formic acid: A mechanistic study

Article abstract of DOI:10.1021/cs300785r

A palladium-catalyzed transfer hydrogenolysis of primary, secondary, and tertiary benzylic alcohols by formic acid has been developed and studied. The product hydrocarbons were obtained in excellent yields from both secondary and tertiary benzylic alcohols and in good yields for primary benzylic alcohols. The rate of disappearance of 1-phenylethanol (1) follows zero-order dependence in 1 and first-order dependence in formic acid and palladium. Catalytic amounts of base inhibit a competing disproportionation reaction of alcohol to alkane and ketone, and an optimum was obtained when 5 equiv of base to palladium was used. Deuterium kinetic isotope studies for the transfer hydrogenolysis reveal individual isotope effects for the hydridic position (k_CHOH/k _CDOH = 2.26 ± 0.24) and the protic position (k _CHOH/k_CHOD = 0.62 ± 0.06) of the formic acid. Simultaneous deuteration in both positions of formic acid gave a combined isotope effect of (k_CHOH/k_CDOD = 1.41 ± 0.11). We propose a mechanism involving the following steps: a competitive inhibition of the open palladium site by adsorption of the formate anion to generate formato-palladium species, followed by a reversible protonation and a rate-limiting hydride transfer to obtain the active palladium with chemisorbed hydrogen that performs the hydrogenolysis of the alcohol in a fast reaction step.

Full text of DOI:10.1021/cs300785r

Research Article
pubs.acs.org/acscatalysis
Pd-Catalyzed Transfer Hydrogenolysis of Primary, Secondary, and
Tertiary Benzylic Alcohols by Formic Acid: A Mechanistic Study
Supaporn Sawadjoon, Anna Lundstedt, and Joseph S. M. Samec*
Department of Chemistry, BMC, Uppsala University, Box 576, 751 23 Uppsala, Sweden
S
* Supporting Information
ABSTRACT: A palladium-catalyzed transfer hydrogenolysis of
primary, secondary, and tertiary benzylic alcohols by formic acid
has been developed and studied. The product hydrocarbons were
obtained in excellent yields from both secondary and tertiary
benzylic alcohols and in good yields for primary benzylic alcohols.
The rate of disappearance of 1-phenylethanol (1) follows zero-order
dependence in 1 and first-order dependence in formic acid and
palladium. Catalytic amounts of base inhibit a competing
disproportionation reaction of alcohol to alkane and ketone, and an optimum was obtained when 5 equiv of base to palladium
was used. Deuterium kinetic isotope studies for the transfer hydrogenolysis reveal individual isotope effects for the hydridic
position (kCHOH/kCDOH = 2.26 0.24) and the protic position (kCHOH/kCHOD = 0.62 0.06) of the formic acid. Simultaneous
deuteration in both positions of formic acid gave a combined isotope effect of (kCHOH/kCDOD = 1.41 0.11). We propose a
mechanism involving the following steps: a competitive inhibition of the open palladium site by adsorption of the formate anion
to generate formato-palladium species, followed by a reversible protonation and a rate-limiting hydride transfer to obtain the
active palladium with chemisorbed hydrogen that performs the hydrogenolysis of the alcohol in a fast reaction step.
KEYWORDS: heterogeneous catalysis, transfer hydrogenolysis, alcohols, kinetic isotope effect
hydrogen gas, the overall process is atom efficient20,21 and only
INTRODUCTION
■
water is formed as a byproduct. Another advantage of formic
The reduction of C−O bonds to the corresponding C−H bond
acid over hydrogen gas is that the latter can lead to over-
is a fundamental transformation in organic synthesis.1−3 While
reduction of aromatic substrates, thus resulting in lower
the reduction of benzylic ethers is a common reaction, the
chemoselectivity.22
corresponding reaction of benzylic alcohols is studied to a
lesser extent. There is an ambition to utilize lignocellulose as a
carbon feed-stock for organic synthesis and biofuels.4,5 Taking
into account that benzylic alcohols are abundant functional
group in lignin,6 efficient methodologies to reduce the C−O
bond in benzylic alcohols are highly desired. Traditionally,
either Clemmensen7 or Wolff−Kishner8 reductions are
performed on ketones, or Barton−McCombie deoxygenation9
is performed on manipulated alcohols. A drawback of the
traditional methodologies is the use of a stoichiometric amount
of hazardous reagents.
An alternative methodology for transforming benzylic
alcohols is the catalytic hydrogenolysis of nonmanipulated
hydroxyl groups, in which the C−O bond is cleaved and
substituted for a hydride.10,11 Most studies have used catalysts
based on palladium,12−15 but ruthenium16 and rhodium17 have
also been explored. Traditionally, hydrogen gas has been used
in catalytic hydrogenolysis to generate the alkane and water as
side product.14,15 More recently, formic acid has been
employed as the source of hydrogen, and the ensuing reaction
has been termed “transfer hydrogenolysis” (eq 1).12,13 Formic
acid as a hydrogen source has many advantages with regards to
handling, transport, and storage18 and can easily be generated
from hydrogen gas and carbon dioxide.19 Because carbon
dioxide can be recycled into formic acid with the addition of
[M]
R−OH + HCOOH ⎯⎯→ R−H + CO2 + H2O
(1)
Different reaction mechanisms have been proposed for
catalytic hydrogenolysis. Van Bekkum et al. have performed
detailed mechanistic studies in which they demonstrated by
isotopic labeling that the acidic carbon support facilitates an
initial elimination to generate the corresponding styrene, which
is then attacked by a hydride to produce the alkane product.16
Kwak et al. even referred to this effect as part of a bifunctional
catalysis.23 The corresponding study of transfer hydrogenolysis
has been lagging behind. When formic acid was used as the
hydrogen donor in the transfer hydrogenolysis of alcohols, a
competing disproportionation of the alcohol was observed
(Scheme 1, pathway a).12 Formic acid may result in
esterification of the alcohol (Scheme 1, pathway b).24 It has
been proposed that the esterification may precede the
hydrogenolysis by converting the hydroxyl group into a better
leaving group.25 There are also reports where the formate ester
intermediate decomposes into the alkane26 or undergoes
Received: December 4, 2012
Revised: February 20, 2013
© XXXX American Chemical Society
635
dx.doi.org/10.1021/cs300785r | ACS Catal. 2013, 3, 635−642

ACS Catalysis
Research Article
disproportionation, a slow transfer hydrogenation of 3 to 1 and
subsequent formation of 2 was observed. After 1 h, 60%
conversion to 2 was observed. Palladium on carbon (5 mol %)
was used as catalyst, since it has previously been reported to be
most active in both catalytic hydrogenolysis and catalytic
transfer hydrogenolysis.12−15 The disproportionation was also
observed in the absence of formic acid albeit at a lower reaction
rate. Attempts to activate the palladium by vacuum and heat
prior to the reaction did not affect the outcome. Attempts to
activate the palladium by exposure to hydrogen gas prior to
catalysis or running the transfer hydrogenolysis under an
atmosphere of hydrogen gas gave marginal improvement; a
70% conversion to ethylbenzene was observed after 30 min.
Even though the mechanism of the disproportionation of
alcohols has not been elucidated, a pathway proceeding through
a palladium with chemisorbed hydrogen, in analogy to a
hydrogenolysis,14,15 would be expected. Both previous studies
describing the use of formic acid as a hydrogen donor for the
catalytic transfer hydrogenolysis of benzylic alcohols reported
that the addition of base gave a lower yield of the product.12,13
This is in contrast to the transfer hydrogenolysis of esters25 and
halides,29 in which formic acid was found to be inactive in the
reaction. Instead, formate salts of either Group 1A or
ammonium were used for a successful catalytic transfer
hydrogenolysis. Trials using pure formate salts in the transfer
hydrogenolysis of 1 did not lead to any reactivity and only the
original material was observed in the reaction mixture.
Ammonium formate is known to generate Pd−H− species,
and transfer hydrogenolysis of both benzylic esters and halides
are promoted using formate salts.25,29 The greater leaving
group ability for these substrates would not require proton
transfer as in the case of benzylic alcohols.
We found that the addition of a catalytic amount of base
generated a powerful reduction medium for the catalytic
transfer hydrogenolysis of 1. Within 30 min, full conversion to
2 was observed (Scheme 2). Triethylamine, pyridine,
ammonium formate, and sodium carbonate were tested as
bases, and all gave similar performance in the catalytic transfer
hydrogenolysis of 1. The order of addition was important.
Successful transfer hydrogenolysis was achieved when the
palladium on carbon was activated by the base in a solvent−
water mixture for at least 2 min before the addition of formic
acid and substrate. The amount of base affected the amount of
3 formed and also how efficiently 2 was generated. With a 5:1
molar ratio of base to palladium, 70% conversion of 1 was
observed after 5 min, and the concentration of disproportio-
nation product 3 was below 10%. Using a 10:1 ratio of base to
palladium gave a 45% conversion of 1 and only traces of 3 were
observed in the reaction medium. At higher concentration of
base, the transfer hydrogenolysis reaction was inhibited.
Effect of Solvent Mixture. Water is necessary for efficient
catalytic transfer hydrogenolysis, as is the presence of a
cosolvent, without which only poor results were obtained.
However, a variety of cosolvents, ranging in polarity from
benzene to methanol, worked well in the Pd-catalyzed transfer
hydrogenolysis of 1. In the absence of water, a disproportio-
nation reaction was observed. The water is expected to
promote the solubility of the formate ion and thereby the
generation of the formato-palladium species that inhibit the
disproportionation reaction.12,13
Scheme 1. Intermediates Observed (a−d) and Proposed (e)
for the Transfer Hydrogenolysis of Alcohols
elimination to form the alkene.24 The alcohol may also undergo
etherification to generate the symmetrical ether as the first
step,27 followed by hydrogenolysis (Scheme 1, pathway c).28
Formic acid may promote initial elimination to generate the
alkene, followed by hydrogenation in line with van Bekkum’s
study on the catalytic hydrogenolysis of benzylic alcohols
(Scheme 1, pathway d).23 Alternatively, the acidity may
promote activation of the hydroxyl group by generating a
carbenium ion followed by a hydride addition (Scheme 1,
pathway e).12−15
In the present work, we have studied the transfer
hydrogenolysis of different benzylic alcohols, using palladium
on carbon as the catalyst. A remarkable effect of added base was
observed whereby the competing disproportionation reaction
(Scheme 1, pathway a) was inhibited, allowing an efficient and
general transfer hydrogenolysis to be demonstrated. A detailed
mechanistic study of the transfer hydrogenolysis has been
performed. On this basis, we propose that the significant
element to the reaction is the generation of the formato-
palladium intermediates that inhibit the competing dispropor-
tionation. This is followed by a reversible protonation and a
rate-limiting hydride transfer to generate palladium with
chemisorbed hydrogen, which is responsible for the hydro-
genolysis of the alcohol.
RESULTS AND DISCUSSION
■
(A). Transfer Hydrogenolysis of Alcohols. Effect of
Hydrogen Donor and Base. When attempts were made to
perform the Pd-catalyzed transfer hydrogenolysis of phenyl-
ethanol (1) in ethanol−water mixture (4:1) using formic acid
as hydrogen donor at 80 °C, a rapid disproportionation was
observed to yield a 1:1 mixture of ethylbenzene (2) and
acetophenone (3) after 2 min (Scheme 2). After the initial
Scheme 2. Effect of Base in the Transfer Hydrogenolysis of 1
Transfer Hydrogenolysis of Benzylic Alcohols. The
optimized reaction conditions, using a 5:1 molar ratio of base
to palladium, and formic acid (3 molar equiv. to alcohol) in an
636
dx.doi.org/10.1021/cs300785r | ACS Catal. 2013, 3, 635−642

ACS Catalysis
Research Article
ethanol/water mixture, were employed for the transfer
hydrogenolysis of different benzylic alcohols. The reactions
were run at 80 °C and performed in a 0.42 mmol scale. The
yield was determined by isolation or quantitative NMR
spectroscopy using mesitylene as an internal standard. Alcohol
1 was reduced to 2 in 98% conversion within 40 min using 5
mol % palladium (Table 1, entry 1). We were curious whether a
tertiary alcohol would work, since the dehydrogenation
pathway would not be possible. 2-Phenyl-propan-2-ol (4) was
reduced to the isopropylbenzene (5) in 87% conversion using 5
mol % Pd. We were also curious as to whether a primary
alcohol would work in spite of its presumed reluctance to
generate a carbenium ion intermediate. Gratifyingly, benzyl
alcohol (6) was reduced to toluene (7) in a 61% conversion
within 40 min. These substrates demonstrate a broad scope in
which primary, secondary, and tertiary benzylic alcohols were
reduced to their corresponding hydrocarbons (Table 1, entries
1−3).
1,1-Diphenylethanol (8) was reduced to generate 1,1-
diphenylethane (9) in a 95% yield within 40 min.
Diphenylmethanol (10) was reduced to diphenylmethane
(11) in a 94% yield within 40 min. 1-Phenylethane-1,2-diol
(12) was selectively reduced in the benzylic position to
generate 1-phenyl-2-propanol (13) (Table 1, entry 6).
Substitution of the phenyl group in the para-position by
electron donating groups gave the corresponding hydrocarbons
in excellent yields (Table 1, entries 7 and 8). Substitution of the
phenyl group in the para-position by an electron withdrawing
group gave the corresponding hydrocarbons in lower yields
(Table 1, entry 9). The allylic alcohol 20 was reduced to
saturated 21 in a good yield, in which both the hydroxyl group
and the alkene were reduced (Table 1, entry 10). Naphthyl
ethanol (22) was reduced to generate naphtyl ethane (23) in a
good yield (Table 1, entry 11). Triphenyl methanol (24) was
reduced to triphenylmethane (25) in a quantitative yield (Table
1, entry 12).
Table 1. Transfer Hydrogenolysis of Different Benzylic
a
Alcohols to the Corresponding Hydrocarbon
(B). Mechanistic Studies. While the mechanism of
palladium-catalyzed alkyne transfer hydrogenation by formic
acid has been well studied, the corresponding study on the
transfer hydrogenolysis of alcohol has not.30 The importance of
developing atom efficient reductions of alcohols in general and
for biomass related compounds in particular,4,5 in addition to
our interest in the mechanism of C−O activation,31−33
motivated us to perform a mechanistic study.
Nature of the Catalyst. The catalyst was characterized by
Transmission Electron Microscope (TEM) and Brunauer−
Emmett−Teller (BET) techniques. The mean particle size was
2−3 nm and the BET surface area 813 m2/g. At stirring rates
below 300 rounds per minute, diffusion control was operating,
whereas at higher stirring rates, kinetic control was operating
(Supporting Information). High reproducibility in the kinetic
experiments (relative standard deviation of 3.9%, n = 6)
supported the observation that kinetic control, rather than
diffusion control, was operating under our reaction conditions
(Supporting Information). The reactions showed good
reproducibility between different batches of Pd/C and were
also unaffected by using inert and purified reagents and solvent.
Inhibition experiments were performed to determine the
heterogeneous nature of the catalyst where the addition of
triphenylphosphine inhibited the reaction and polymer bound
triphenylphosphine did not affect the rate of the transfer
hydrogenolysis (Supporting Information).34
Initial Rates for the Transformation of Postulated
Intermediates to Ethylbenzene. Different intermediates for
the catalytic hydrogenolysis and transfer hydrogenolysis of
alcohols have previously been proposed.12,13 During the
optimization of the reaction conditions, we observed a few of
1
these intermediates (3, 27, 28, 29) by H NMR spectroscopy.
To be able to exclude the observed intermediates as
intermediates in the mechanism of 1 to 2, we compared the
initial rates of conversion of these possible intermediates.
The observed initial rate for the disappearance of 1 was 3.3 ×
10−3 mol L−1 s−1 (Table 2, entry 1). Styrene (26) underwent a
fast transfer hydrogenation with a higher initial rate than 1
a
Conditions: Alcohol (0.4 mmol), 5% Pd/C (42 mg), HCOONH4
(0.095 mmol), HCOOH (1.3 mmol), MeOH (2.4 mL), water (0.6
b
mL), temperature 80 °C. Yields were calculated using 1H NMR with
c
d
mesitylene as an internal standard. Isolated yield. 10 mol % of Pd
was used.
637
dx.doi.org/10.1021/cs300785r | ACS Catal. 2013, 3, 635−642

ACS Catalysis
Research Article
Table 2. Transfer Hydrogenolysis of Different Intermediates
a
Found in Catalysis
Figure 1. Dependence of initial rate on concentration of substrate 1 in
the Pd/C catalyzed transfer hydrogenolysis of 1 by formic acid.
Reaction conditions: 1 (0.065−0.26 M), Pd/C (4.9 mol %),
HCOONH4 (0.095 mmol), HCOOH (1.3 mmol), MeOH (2.4
mL), water (0.6 mL), temperature 25 °C, stirring speed 350 rpm.
agreement with the lower relative rates observed for the transfer
hydrogenolysis of 27−29 compared to 1 and 3 (Table 2).
The plot of initial rate of reaction versus the concentration of
hydrogen donor is shown in Figure 2. The reaction rate
a
Conditions: The optimized reaction conditions were used [substrates
(0.4 mmol), Pd/C (4.9 mol %), HCOONH4 (0.095 mmol), HCOOH
(1.3 mmol), MeOH (2.4 mL), water (0.6 mL), temperature 25 °C,
stirring speed 350 rpm]. Initial rate was determined as an average of
two runs (Supporting Information).
(Table 2, entry 2). The disappearance of acetophenone 3 also
gave a high initial rate of starting material (Table 2, entry 3).
Thereby these two intermediates, especially 26 would be
difficult to observe as such during the transfer hydrogenolysis of
1. 1-Phenylethyl formate (27) was transformed into 2 at a
lower rate relative to 1 (Table 2, entry 4). Intermediate 27 was
observed under acidic conditions, but not in the presence of the
formate/formic acid mixture used in the successful hydro-
genolysis. Also, the diphenylethyl ether (28) and 1-methoxy-1-
phenylethane (29) were reduced at a lower rate than 1 (Table
2, entries 5−6).
Reaction Order Determination. Initial rates were deter-
mined for the Pd/C catalyzed transfer hydrogenolyis of 1 by
formic acid at 25 °C. Concentrations of 1 were varied between
0.06 and 2.6 M, and initial rates were determined at a
conversion of 1 below 10%. The initial rates for the transfer
hydrogenolysis of 1 was performed in a thermostatted oil bath
where aliquots were withdrawn from the reaction mixture at the
different time intervals and the conversions were determined by
1H NMR spectroscopy. As seen in Figure 1, the initial rate for
transfer hydrogenolysis of 1 using Pd/C and formic acid and
formate was independent of the substrate concentration in the
examined range.
Figure 2. Dependence of initial rate on concentration of formic acid in
the Pd/C catalyzed transfer hydrogenolysis of 1 by formic acid.
Reaction conditions: 1 (0.4 mmol), Pd/C (4.9 mol %), HCOONH4
(0.095 mmol), HCOOH (0.087−0.427 M), MeOH (2.4 mL), water
(0.6 mL), temperature 25 °C, stirring speed 350 rpm.
followed a linear relationship with the concentration of formic
acid over the range 0.1−0.45 M. The first-order dependence of
the hydrogen donor supports a reaction mechanism in which
the proton and/or hydride transfer is involved in, or precede,
the rate-limiting step.
The plot of initial rate versus concentration of the catalyst is
displayed in Figure 3. The first-order dependence of the catalyst
is clear from the plot. The high reproducibility of the kinetic
runs with the catalyst implies that kinetic-, rather than diffusion
control, is operating (Supporting Information).34 Noteworthy,
no sigmoidal kinetic curve was observed during these
experiments. This supports that palladium on carbon and not
leaked colloidal metal is reactive in the transfer hydrogenolysis
of alcohols.34
This zero-order dependence of 1 is in accordance with a
mechanism where the hydrogen transfer to the substrate or C−
O bond cleavage is not the rate-determining step. It should be
noted that intermediates 27−29 were not observed during the
course of the reaction, and would have been if they were
intermediates in the mechanism from 1 to 2. This is in
The rate order of the base was studied (Figure 4). Without
base, a disproportionation reaction occurs, in which the
substrate also acts as a hydrogen donor, generating a 1:1
ratio of 2 and 3. The rate of the disproportionation reaction is
proportionally inhibited by the base (HCOONH4 concen-
638
dx.doi.org/10.1021/cs300785r | ACS Catal. 2013, 3, 635−642

ACS Catalysis
Research Article
Deuterium Kinetic Isotope Effect. The kinetic deuterium
isotope effect of the transfer hydrogenolysis of 1 was studied at
room temperature using the optimized reaction conditions
monitored by removing aliquots at regular intervals and
1
measuring the disappearance of 1 by H NMR spectroscopy.
The water (H2O or D2O) and methanol (CH3OH or CD3OD)
present in the reaction medium determined whether deuterium
or hydrogen was present at the protic position of formic acid. In
the transfer hydrogenolysis of 1 using HCOOH and H2O, we
found kobs = 3.27 ( 0.12) × 10−3 mol L−1 s−1. When running
the reaction in deuterated formic acid and deuterated solvents,
kobs = 2.33 ( 0.25) × 10−3 mol L−1 s−1was determined. The
observed deuterium kinetic isotope effect (kH/kD) was
therefore 1.41 ( 0.11) (Table 3). When running the reaction
Figure 3. Dependence of initial rate on catalyst loading in the Pd/C
catalyzed transfer hydrogenolysis of 1 by formic acid. Reaction
conditions: 1 (0.4 mmol), Pd/C (2.47−7.40 mol %), HCOONH4
(0.095 mmol), HCOOH (1.3 mmol), MeOH (2.4 mL), water (0.6
mL), temperature 25 °C, stirring speed 350 rpm.
Table 3. Kinetic Deuterium Isotope Effect for Transfer
Hydrogenolysis of 1
kCHOH/kCDOH
2.26 0.24
0.62 0.06
1.41 0.11
kCHOH/kCHOD
CHOH/kCDOD
k
with selective deuterium incorporation at the OD position, we
measured kobs = 5.33 ( 0.34) × 10−3 mol L−1 s−1, giving a
deuterium kinetic isotope effect of 0.62 ( 0.06) (Table 3).
When running the reaction with formic acid selectively
deuterated at the hydridic position, we determined a value of
k
obs = 1.47 ( 0.21) 10−3 mol L−1 s−1, giving a deuterium kinetic
isotope effect of 2.26 ( 0.24) (Table 3).35,36 These kinetic
isotope effect values are consistent with a reversible proton
transfer and a rate-determining hydride transfer from formic
acid to palladium.
Figure 4. Dependence of initial rate on concentration of ammonium
formate in the Pd/C catalyzed transfer hydrogenolysis of 1 by formic
acid. Reaction conditions: 1 (0.4 mmol), Pd/C (4.9 mol %),
HCOONH4 (0.0079−0.3806 M), HCOOH (1.3 mmol), MeOH
(2.4 mL), water (0.6 mL), temperature 25 °C, stirring speed 350 rpm.
Attempts to perform a labeling study using selectively
deuterated formic acid (DCOOH or HCOOD) to distinguish
whether the proton or hydride of formic acid can be
discriminated in the product were unsuccessful. The proton
source determined the label of both the α- and the β-position of
the product, even at a 10% conversion. In a competitive
experiment, we observed that the deuterium incorporation in
toluene by HCOOD was faster than the transfer hydrogenolysis
of benzylic alcohol by HCOOD. The inverse kinetic isotope
effect for the proton transfer is consistent with a scrambling of
the proton and may explain why the labeling studies were
unsuccessful.
trations 0−15.9 mM). Between 15.9 and 63 mM of the
ammonium formate, the rate is independent of the base. At
these concentrations, negligible concentration of 3 was
observed, and a transfer hydrogenolysis is operating, in which
the formic acid acts as a hydrogen donor. At base
concentrations above 63 mM, the transfer hydrogenolysis is
negatively affected by the concentration of base, and with only
formate, no reaction occurs.
This gives a rate equation for the transfer hydrogenolysis in
which the reaction rate is dependent on Pd and formic acid,
depicted in eq 2. At lower concentration of the base, a fast
disproportionation reaction occurs. At higher concentrations of
the base, the transfer hydrogenolysis is inhibited and another
mechanism is operating, involving a formato-palladium or
palladium monohydride species not active in the transfer
hydrogenolysis of alcohols.
MECHANISTIC DISCUSSION
■
Disproportionation and Transfer Hydrogenation
Pathway. Palladium on carbon catalyzes the disproportiona-
tion of 1 in the presence or absence of formic acid to generate a
1:1 mixture of 2 and 3 (Scheme 2).13 The reaction is fast and
within 2 min, a 90% conversion of 1 is observed. A possibility is
that a fast disproportionation of the alcohol is followed by a
slower transfer hydrogenation of the ketone to regenerate the
alcohol. In this case, the formic acid would operate as a
hydrogen donor to regenerate 1 from 3 (Scheme 3). Because
the conversion of 3 to 1 is faster than the conversion of 1, this
may be a possible reaction mechanism. However, the transfer
hydrogenolysis was successful even with tertiary alcohols
(which cannot undergo β-hydride elimination); this is
inconsistent with a disproportionation pathway, as is the rate
d[1]
dt
−
= k[Pd][HCOOH]
(2)
The rate equation implies that the hydrogen transfer between
the formic acid and palladium is involved in the rate-
determining step, and that C−O bond cleavage is not.
Therefore, we decided to study the hydrogen transfer in detail.
639
dx.doi.org/10.1021/cs300785r | ACS Catal. 2013, 3, 635−642

ACS Catalysis
Research Article
but this was not observed. Furthermore, primary alcohols that
were successfully transformed to the corresponding hydro-
carbons are not susceptible toward an elimination pathway.
Also, secondary alcohols without β-hydrogens, such as 10, were
successfully transformed, which would not be expected if the
transformation proceeded through an elimination pathway.
Protonation and Carbocation Formation. The formic
acid may facilitate protonation to convert the hydroxyl group of
1 into a better leaving group, which is subsequently eliminated
to leave behind a carbenium ion (Scheme 6). That para-
Scheme 3. Disproportionation and Transfer Hydrogenation
Pathway
Scheme 6. Reaction Mechanism for the Transfer
Hydrogenolysis Involving a Carbenium Intermediate
expression in which the substrate is absent (zero-order
dependence in 1). The fact that the initial rate was independent
of 1 is inconsistent with a disproportionation pathway, where a
rate-determining transfer dehydrogenation is expected to be
operating during the initial turnovers (>10% conversion,
measured in Figure 1).
Ester and Ether Pathways. Both the ether and ester have
been observed by other research groups as well as by our group
during the optimization studies with low catalyst loadings under
acidic conditions (Scheme 4). However, under our optimized
Scheme 4. Transfer Hydrogenolysis via Ether or Ester
Intermediates
substitution by electron-donating substituents facilitates trans-
fer hydrogenolysis, while electron-withdrawing substituents
impede the reaction and supports the elimination hypothesis
(Table 1, entries 7−9). In addition, the primary benzylic
alcohols showed lower reactivity than the secondary and
tertiary benzylic alcohols (Table 1, entries 1−3). If the
carbenium ion was generated, the excess methanol in the
reaction medium would be expected to trap the carbenium ion
and form 1-methoxy-1-phenylethane (29), via an SN1 reaction
mechanism (MeOH/Pd is 2650:1). If the ether was an
intermediate in the reaction, it would have been observed
because the rate of reduction of the ether is lower compared to
the transfer hydrogenolysis of 1, and this is not the case.
Furthermore, the reaction rate for a tertiary benzylic alcohol
would be expected to be faster than a secondary alcohol if the
reaction proceeded through a carbenium ion intermediate. The
fact that the transfer hydrogenolysis works on unactivated
primary benzylic alcohols that are highly unlikely to generate a
carbenium ion counts heavily against this proposed pathway.
Proposed Mechanism. In the proposed reaction mecha-
nism, a formate anion adsorbs with palladium to generate
formato-palladium intermediate A (Scheme 7). This adsorption
inhibits the disproportionation pathway by blocking the open
coordination site of palladium for 1 (K2 > k1). This is consistent
with the negative rate dependence for the base between 0 and
16 mM (Figure 4), where the mechanism changes from a fast
disproportionation to a slower transfer hydrogenolysis.
Complex A or a generated monohydride palladium species is
not active in the transfer hydrogenolysis of benzylic alcohols,
because of the poor leaving group ability of the hydroxyl group
(Scheme 2). In the presence of formic acid a proton
equilibrates with complex A and complex B. This is consistent
with the inverse kinetic isotope effect for the proton transfer
(kCHOH/CHOD = 0.62 0.06) observed when using formic acid
deuterated in the protic position. The proton transfer is
followed by a rate-limiting hydride transfer to generate
palladium with chemisorbed hydrogen (C). This is consistent
with the primary kinetic isotope effect (kCHOH/CDOH = 2.26
0.24) observed for the hydride transfer.30,37,38 The transfer of
reaction conditions, these intermediates were not observed.
Since the transfer hydrogenolysis rates for both the ester and
the two ethers were much slower (20% and 35% conversion
after 20 min, respectively) than the rate of transfer hydro-
genolysis of 1 (a 90% conversion after 20 min), one would
1
expect them to be detectable by H NMR spectroscopy if they
were true intermediates. However, we were unable to observe
27−29 under the reaction conditions.
Elimination and Transfer Hydrogenation Pathway. An
elimination to form an alkene intermediate either via an acid-
catalyzed E1-mechanism or by an insertion of palladium into
the C−O bond, followed by a β-hydride elimination has been
suggested as the pathway for the hydrogenolysis of alcohols by
hydrogen gas (Scheme 5).14 The acid-catalyzed elimination
reaction would generate styrene in the absence of palladium,
but this was not observed in subsequent studies. Moreover, if
this were the case, tertiary alcohols would be expected to
undergo faster transfer hydrogenolysis than secondary alcohols,
Scheme 5. Mechanism Proceeding through Elimination and
a Transfer Hydrogenation
640
dx.doi.org/10.1021/cs300785r | ACS Catal. 2013, 3, 635−642

ACS Catalysis
Research Article
is, the hydrogenolysis of the benzylic alcohol, occurs in a fast
reaction step.
Scheme 7. Proposed Reaction Mechanism of the Transfer
Hydrogenolysis of 1 by Formic Acid
EXPERIMENTAL SECTION
■
NMR spectra were recorded on a Varian UNITY (1H at 400
MHz, 13C at 100 MHz) or a Varian INOVA (1H at 500 MHz,
13C at 125 MHz) spectrometer. The chemical shifts are
reported using the residual solvent signal as an indirect
reference to TMS 1H (CHCl3): 7.26 ppm, 13C (CDCl3):
77.16 ppm. The Transmission Electron Microscope (TEM)
studies were done in a JEM 2100F (Jeol, Japan) operating at
200 kV and equipped with an Energy Dispersive X-ray
Spectrometer (EDS). For the BET measurements, the uptake
of N2 was measured at the temperature of liquid nitrogen
(−196 °C or 77 K) using a Micromeritics ASAP2020
volumetric adsorption analyzer. Column chromatography was
performed with Merck silica gel (0.04−0.06 mm). Thin layer
chromatography (TLC) was performed using Fluka precoated
silica gel (0.2 mm) 60-F254 plates. All reactions were
performed in heavy-walled glass Emrys process vials (2−5
mL) sealed with aluminum crimp caps fitted with a septum. All
yields were determined by isolation or by using qNMR where a
1H NMR spectrum was recorded on a Varian UNITY at 400
MHz with d1 = 30 and using mesitylene as an internal standard.
All chemicals including Pd/C were purchased from Sigma-
Aldrich. No further purification was needed to obtain
reproducible kinetic data (Supporting Information). Controls
using purified reagents and solvents were performed, without
change in performance or reproducibility.
the proton and the hydride from the formic acid to palladium
may occur in a concerted process. The product of the two
individual isotope effects (0.62 × 2.26 = 1.40), falls within
experimental error of the combined D isotope effect measured
for (kCHOH/CDOD = 1.41
0.11), in accordance to Casey’s
methodology to determine concerted hydrogen transfers in
transition metal catalysis.39 Complex C reduces the benzylic
alcohol in a fast reaction step. The fact that palladium on
carbon alone catalyzes the disproportionation, which is known
to proceed through a palladium with chemisorbed hydrogen
intermediate, counts in support of this mechanistic proposal.40
The observation that the transfer hydrogenolysis showed zero-
order dependence in benzylic alcohol is consistent with a
reaction mechanism in which both the adsorption of the
alcohol to C and the reduction step, and thus also C−O bond
cleavage, is not rate-limiting. The formic acid may promote the
C−O bond cleavage by protonation, however not in a rate-
determining step.
Reduction of 1-Phenylethanol (1) with Pd/C. 5% Pd/C
(42 mg, 0.021 mmol) and HCOONH4 (6 mg, 0.095 mmol)
were added to a predried vial. The vial was capped and flushed
with argon, and EtOH (2.4 mL) and H2O (0.6 mL) were added
via syringe. The mixture was placed into a preheated oil-bath at
80 °C. Formic acid (50 μL, 1.05 mmol) was added after 3 min
and after an additional 2 min, 1 (50 μL, 0.42 mmol) was added.
The mixture was stirred at 80 °C for 30 min. NMR samples
were prepared by removing 0.05 mL aliquots from the reaction
mixture and filtration through Celite using CDCl3 (1.3 mL)
containing mesitylene (1 μL) as eluent. The filtrate was dried
CONCLUSION
■
Pd/C catalyzes the disproportionation of benzylic alcohols to
generate a 1:1 ratio of ketone and hydrocarbon. In the presence
of formic acid and a catalytic amount of base, this ratio is
changed in favor of the hydrocarbon. The base can be an
organic or an inorganic base. The transfer hydrogenolysis works
equally well for secondary and tertiary alcohols to generate the
corresponding hydrocarbons in excellent yields. Primary
benzylic alcohols are transformed to the corresponding
hydrocarbons in moderate yields. We have found an inverse
1
over Na2SO4 and added to an NMR tube. The H NMR
spectrum of 1 shows a characteristic multiplet at δ 7.20 and a
doublet at δ 1.46 ppm, 2 shows a characteristic multiplet at δ
7.36 and a triplet at δ 1.26, and 3 a characteristic multiplet at δ
7.50 and at singlet at δ 2.62 and compared to mesitylene signal
at δ 6.76.
kinetic isotope effect for the protic position (kCHOH/kCHOD
=
General Procedure for Kinetics of Reduction of 1-
Phenylethanol (1) with Pd/C. 5% Pd/C (42 mg, 0.021
mmol) and HCOONH4 (6 mg, 0.095 mmol) were added to a
predried vial. The vial was capped and flushed with argon, and
MeOH (2.4 mL) and H2O (0.6 mL) were added via syringe.
Formic acid (50 μL, 1.05 mmol) was added after 3 min and
after 2 min 1 (50 μL, 0.42 mmol) was added. The mixture was
stirred at 25 °C. NMR samples were prepared by removing 0.05
mL aliquots from the reaction mixture, filtration through Celite
using CDCl3 (1.3 mL) as eluent, followed by the addition of
mesitylene (1 μL) as an internal standard. The filtrate was dried
over Na2SO4 and added to an NMR tube. The disappearance of
substrate 1 was monitored by comparison of the signal for a
proton in 1 resonating at δ 4.85 with the mesitylene signal at δ
6.76.
0.62 0.06) and a primary kinetic isotope effect for the
hydridic position of formic acid (kCHOH/kCHOD = 2.26 0.24).
The product of the two individual isotope effects (0.62 × 2.26
= 1.40) is within experimental error of the combined deuterium
kinetic isotope effect measured for the transfer hydrogenolysis
(kCHOH/CDOD = 1.41 0.11).
We propose a mechanism in which the formate anion
adsorbs to the palladium. The role of the base is to inhibit the
disproportionation pathway by blocking the vacant coordina-
tion sites on Pd, preventing the undesired disproportionation
reaction of 1. A proton from formic acid equilibrates with the
formato-palladium complex A. The hydride transfer from
formic acid to Pd occurs in a rate-determining step. The proton
and hydride transfer from formic acid to palladium may occur
in a concerted step. The transfer of the hydrogens from Pd, that
641
dx.doi.org/10.1021/cs300785r | ACS Catal. 2013, 3, 635−642

ACS Catalysis
Research Article
(29) Wiener, H.; Blum, J.; Sasson, Y. J. Org. Chem. 1991, 56, 6145−
6148.
(30) Ahlquist, M.; Fabrizi, G.; Cacchi, S.; Norrby, P.-O. J. Am. Chem.
ASSOCIATED CONTENT
■
S
* Supporting Information
The heterogeneous nature of the catalyst have been studied by
kinetics and also TEM and BET measurements. This material is
available free of charge via the Internet at http://pubs.acs.org.
Soc. 2006, 128, 12785−12793.
(31) Mirzaei, A.; Biswas, S.; Samec, J. S. M. Synthesis 2012, 44, 1213−
1218.
(32) Sawadjoon, S.; Samec, J. S. M. Org. Biomol. Chem. 2011, 9,
2548−2554.
AUTHOR INFORMATION
■
(33) Howard, F.; Sawadjoon, S.; Samec, J. S. M. Tetrahedron Lett.
2010, 51, 4208−4011.
Corresponding Author
(34) Widegren, J. A.; Finke, R. G. J. Mol. Catal. A: Chem. 2003, 198,
317−341.
*E-mail: joseph.samec@kemi.uu.se.
Notes
(35) Lange, S.; Leitner, W. J. Chem. Soc., Dalton Trans. 2002, 752−
758.
The authors declare no competing financial interest.
(36) Leitner, W.; Brown, J. M.; Brunner, H. J. Am. Chem. Soc. 1993,
115, 152−159.
ACKNOWLEDGMENTS
■
(37) Darenbourg, D. J.; Wiegreffe, P.; Riordan, C. G. J. Am. Chem.
Soc. 1990, 112, 5759−5762.
We thank the Swedish Energy Agency and Magnus Bergvalls
stiftelse for financial support. We thank Prof. C. P. Casey and
Dr. L. Pilarski for fruitful discussions.
(38) Yu, J.; Spencer, J. B. Chem. Commun 1998, 1935−1936.
(39) Casey, C. P.; Singer, S. W.; Powell, D. R.; Hayashi, R. K.;
Kavana, M. J. Am. Chem. Soc. 2001, 123, 1090−1100.
(40) Muzart, J. Tetrahedron 2003, 59, 5789−5816.
REFERENCES
■
(1) Yasuda, M.; Onishi, Y.; Ueba, M.; Miyai, T.; Baba, A. J. Org.
Chem. 2001, 66, 7741−7744.
(2) Radinov, R.; Hutchings, S. D. Tetrahedron Lett. 1999, 40, 8955−
8960.
(3) Muzart, J. Tetrahedron 2005, 61, 9423−9463.
(4) Huber, G. W.; Iborra, S.; Corma, A. Chem. Rev. 2006, 106, 4044−
4098.
(5) Corma, A.; Iborra, S.; Velty, A. Chem. Rev. 2007, 107, 2411−
2502.
(6) Zakzeski, J.; Bruijnincx, P. C. A.; Jogerius, A. L.; Weckhuysen, B.
M. Chem. Rev. 2010, 110, 3552−3599.
(7) Vedejs, E. Org. React. 1975, 22, 401−422.
(8) Todd, D. Org. React. 1948, 4, 378−422.
(9) Barton, D. H. R.; McCombie, S. W. J. Chem. Soc., Perkin Trans. 1
1975, 1574−1585.
(10) Schlaf, M. J. Chem. Soc., Dalton Trans. 2006, 4645−4653.
(11) Alonso, D. M.; Bond, J. Q.; Dumesic, J. Green Chem. 2010, 12,
1493−1513.
(12) Feng, J.; Yang, C.; Zhang, D.; Wang, J.; Fu, H.; Chen, H.; Li, X.
Appl. Catal., A 2009, 354, 38−43.
(13) Liu, X.; Lu, G.; Guo, Y.; Guo, Y.; Wang, Y.; Wang, X. J. Mol.
Catal A: Chem. 2006, 252, 176−180.
(14) Thakar, N.; Polder, N. F.; Djanashvili, K.; van Bekkum, H.;
Kapteijn, F.; Moulijn, J. A. J. Catal. 2007, 246, 344−350.
(15) Kieboom, A. P. G.; de Kreuk, J. F.; van Bekkum, H. J. Catal.
1971, 20, 58−66.
(16) Taher, D.; Thibault, M. E.; Di Mondo, D.; Jennings, M.; Schlaf,
M. Chem.Eur. J. 2009, 15, 10132−10143.
(17) Ranade, V. S.; Prins, R. Chem.Eur. J. 2000, 6, 313−320.
(18) Jessop, P. G.; Joo, F.; Tai, C.-C. Coord. Chem. Rev. 2004, 248,
2425−2442.
(19) Federsel, C.; Jackstell, R.; Beller, M. Angew. Chem., Int. Ed. 2010,
49, 6254−6257.
(20) Sheldon, R. A. Pure Appl. Chem. 2000, 72, 1233−1246.
(21) Trost, B. M. Science 1991, 254, 1471−1477.
(22) Rylander, P. N. Catalytic Hydrogenation over Platinum Metals;
Academic Press: New York, 1967.
(23) Kwak, B.-S.; Kim, T.-J.; Lee, S.-I. Appl. Catal., A 2003, 238,
141−148.
(24) Arceo, E.; Marsden, P.; Bergman, R. G.; Ellman, J. A. Chem.
Commun. 2009, 3357−3359.
(25) Rajagopal, S.; Spatola, A. F. Appl. Catal., A 1997, 152, 69−81.
(26) Watkins, S. T.; Bowden, S. T. J. Chem. Soc. 1940, 1333−1334.
(27) Miller, K. J.; Abu-Omar, M. M. Eur. J. Org. Chem. 2003, 1294−
1299.
(28) Saulnier, D. G.; Dodier, M.; Frennesson, D. B.; Langley, D. R.;
Vyas, D. M. Org. Lett. 2009, 11, 5154−5157.
642
dx.doi.org/10.1021/cs300785r | ACS Catal. 2013, 3, 635−642