Directing the Cleavage of Ester C-O Bonds by Controlling the Hydrogen Availability on the Surface of Coprecipitated Pd/Fe3O4

Article abstract of DOI:10.1002/cctc.201501414

The ready availability of esters from natural sources makes them promising substrates for the sustainable production of key intermediates of the chemical industry. In this context, methods to selectively cleave the alkoxy or ester C-O bond to obtain either alkane+acid or alcohol+aldehyde were investigated. Using Pd/Fe₃O₄^(r) [(r)=reduced] as the catalyst of choice and either 2-propanol - as indirect H source - or molecular hydrogen - as ample H source - the hydrogen coverage on the surface of the catalyst was controlled. This enabled selective cleavage of either the alkoxy C-O bond or the ester C-O bond in aromatic esters as shown for the hydrogenolysis of 2-phenylethylacetate as model substrate. Noteworthy, hydrogenation of the aromatic ring was not observed. Thus, an exceptionally high chemoselectivity of Pd/Fe₃O₄^(r) to hydrogenolysis of the ester group is combined with the option to readily direct the regioselectivity.

Full text of DOI:10.1002/cctc.201501414

DOI: 10.1002/cctc.201501414
Full Papers
Directing the Cleavage of Ester CÀO Bonds by Controlling
the Hydrogen Availability on the Surface of Coprecipitated
Pd/Fe3O4
Daniela Cozzula,[a] Alessandro Vinci,[b] Francesco Mauriello,*[b] Rosario Pietropaolo,[b] and
Thomas E. Müller*[a]
The ready availability of esters from natural sources makes
them promising substrates for the sustainable production of
key intermediates of the chemical industry. In this context,
methods to selectively cleave the alkoxy or ester CÀO bond to
obtain either alkane+acid or alcohol+aldehyde were investi-
surface of the catalyst was controlled. This enabled selective
cleavage of either the alkoxy CÀO bond or the ester CÀO bond
in aromatic esters as shown for the hydrogenolysis of 2-phe-
nylethylacetate as model substrate. Noteworthy, hydrogena-
tion of the aromatic ring was not observed. Thus, an excep-
(r)
(r)
gated. Using Pd/Fe3O4 [(r)=reduced] as the catalyst of choice
tionally high chemoselectivity of Pd/Fe3O4 to hydrogenolysis
and either 2-propanol—as indirect H source—or molecular hy-
drogen—as ample H source—the hydrogen coverage on the
of the ester group is combined with the option to readily
direct the regioselectivity.
Introduction
The directed cleavage of one of the two different CÀO single
bonds in esters derived from natural sources represents one of
the major challenges in green chemistry.[1] Recent reports con-
cern the synthesis of long- and short-chain alcohols[2] from nat-
ural fats and oils,[3] acetates and formiates as well as diesters.[4]
To accelerate the hydrogenolysis reaction various transition-
metal catalysts have been used.[5,6] Nickel has found wide-
spread application in the hydrotreatment of vegetable oils to
produce diesel-range alkanes.[7] Copper-based catalysts are em-
ployed commonly for cleaving CÀO bonds in esters[2] to form
alcohol products but require harsh reaction conditions. Similar-
ly, homogeneous[8] and heterogeneous[9] catalysts have been
used for cleaving the CÀO bond in aryl ethers to arenes and al-
cohols including the depolymerization of soft- and hardwood
lignin.[8,9]
transfer hydrogenolysis of polyalcohols employing 2-propanol
as H source.[14] As 2-propanol can be obtained from biological
sources, it is a particularly “green” reductant that can be dehy-
drogenated readily to acetone in the presence of an appropri-
ate catalyst. The use of such alternative hydrogen donors has
recently gained increased attention to improve the sustainabili-
ty and economics of hydrogenation reactions.[15] In the hydro-
genolysis of esters, the cleavage of the alkoxy CÀO bond pro-
vides alkane and acid, while cleavage of the ester CÀO bond
provides alcohol and aldehyde (Scheme 1). If aromatic esters
are used as substrates, a high chemoselectivity to hydrogenol-
ysis of the ester group versus hydrogenation of the aromatic
ring is also essential.
(r)
Recently, coprecipitated Pd/Fe3O4 [(r)=reduced] has been
claimed to give superior performance in a number of reactions,
including CÀO reduction,[10] aqueous-phase reforming of ethyl-
ene glycol,[11] and hydrogenolysis of glycerol[12] and ethylene
(r)
glycol.[13] Furthermore Pd/Fe3O4 has been used in the catalytic
[a] Dr. D. Cozzula, Dr. T. E. Müller
CAT Catalytic Center
RWTH Aachen University
Worringerweg 2, 52074 Aachen (Germany)
E-mail: thomas.mueller@catalyticcenter.RWTH-Aachen.de
Scheme 1. Options for the cleavage of the two different CO bonds in 2-
phenylethylacetate (PEA); under the conditions of ester hydrolysis, the acid
and aldehyde are converted to the corresponding iPr-ester and alcohol.
[b] Dr. A. Vinci, Dr. F. Mauriello, Prof. R. Pietropaolo
Department of Civil Engineering, Energy, Environment and Materials
(DICEAM)
(r)
Universita’ Mediterranea di Reggio Calabria
loc. Feo di Vito—89122 Reggio Calabria (Italy)
E-mail: francesco.mauriello@unirc.it
In the present study, coprecipitated Pd/Fe3O4 was explored
as a catalyst for the hydrogenolysis of 2-phenylethylacetate
(PEA). The principal goal was to develop a methodology for
the directed cleavage of one of the ester CÀO single bonds, at
the same time avoiding the concurrent hydrogenation of the
Supporting Information and the ORCID identification number(s) for the
author(s) of this article can be found under http://dx.doi.org/10.1002/
cctc.201501414.
ChemCatChem 2016, 8, 1515 – 1522
1515
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
aromatic ring. 2-Propanol was chosen as indirect H source and
molecular hydrogen as readily available H source. Depending
on the choice of the H source, a surprising switch in selectivity
with respect to the cleavage of either the alkoxy or the ester
CÀO bond was observed. To study the significance of the dis-
tance between phenyl ring and ester group and the influence
of the phenyl ring, substrates with different length of the
linker as well as alkylacetates were tested under similar
reaction conditions.
0.4 eV higher than that reported for metallic palladium.[14] Anal-
ysis of the extended X-ray absorption fine structure revealed
that some iron was alloyed into the palladium nanoparticles
forming Pd–Fe bimetallic ensembles.[13] Both findings imply
that also strong interactions between the palladium nanoparti-
cles and the iron oxide support were present (Supporting
Information).
(r)
Hydrogenolysis of 2-phenylethylacetate over Pd/Fe3O4
The hydrogenolysis of the ester CÀO bond in PEA was studied
in detail as model reaction. In principle, the hydrogenolysis of
PEA may give rise to two possible reaction sequences
(Scheme 2). Cleavage of the alkoxy CÀO bond produces ethyl-
benzene (EBE) and acetic acid (route A). Cleavage of the ester
CÀO bond yields phenylethanol (PET) and ethanol (route B). In
a successive reaction, phenylethanol can be reduced to EBE.
Moreover, the aromatic ring of PEA as well as of any of the re-
action intermediates that occur in the two reaction pathways
may be hydrogenated.
Results and Discussion
Catalyst preparation and characterization
(r)
A sample of Pd/Fe3O4 with a Pd loading of 8.7 wt% was pre-
pared by coprecipitation of palladium chloride and iron nitrate
followed by calcination and reduction with molecular hydro-
gen.[10–14] The obtained catalyst had a high BET surface area of
170 m2 gÀ1. Palladium nanoparticles were well dispersed on the
surface as shown by X-ray diffraction and high-resolution trans-
mission electron microscopy. Analysis of the particle size distri-
bution revealed a bimodal distribution with a majority of parti-
cles of 1.2 nm diameter and a second population of slightly
larger palladium particles with 2.3 nm average diameter
(Figure 1). X-ray photoelectron spectroscopy (see the Support-
ing Information) revealed that the Pd3d5/2 binding energy was
An initial exploratory study on the best reaction conditions
for heterogeneous palladium catalysts showed that harsher re-
action conditions were required (1508C) for the cleavage of
CÀO bonds in phenyl alkyl esters than for benzylesters (benzyl-
acetate, BAC: 508C, 10 bar H2, see below).[16]
(r)
1) Using Pd/Fe3O4 as catalyst, the reduction of 2-phenyl-
ethylacetate with 2-propanol as H source proceeded rapidly
at 1508C (94% conversion in 24 h). Ethylbenzene was ob-
tained as the main product with 93% selectivity (Table 1,
entry 1). Acetic acid, formed during the reaction as by-
product, reacted with 2-propanol to isopropyl acetate.
Traces of 2-phenylethanol were detected. Products, where
the aromatic ring had been hydrogenated, were not
observed.
2) For comparison, the reduction of 2-phenylethylacetate was
(r)
also studied over Pd/Fe3O4 in the presence of molecular
hydrogen at 1508C (Table 1, entry 2). The reaction proceed-
ed more slowly and, though unexpected, 2-phenylethanol
was obtained with a high selectivity of 98%.
Figure 1. Size distribution of the palladium nanoparticles and high-resolu-
tion transmission electron micrograph of the Pd/Fe3O4 catalyst (inset).
Thus, the product selectivity was reversed if a different
reductant was used as H source. Whereas the use of 2-propa-
(r)
(r)
Scheme 2. Alternative pathways for the CÀO bond cleavage in 2-phenylethylacetate in the presence of Pd/Fe3O4 as catalyst. With 2-propanol as H source,
route A prevails (left) whereas the use of molecular hydrogen as H source directs the reaction pathway to route B (right).
ChemCatChem 2016, 8, 1515 – 1522
www.chemcatchem.org
1516
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 1. Hydrogenolysis of phenylethylacetate in the presence of different catalysts by using 2-propanol or molecular hydrogen as H source.
Entry
Catalyst
H source
Conversion
[%]
Selectivity [%]
CÀO bonds Aromatic ring
Selectivity [%]
Chemoselectivity [%]
route A
route B
EBE
PET
HEX
HET
HEA
(r)
(r)
1
2
3
4
5
6
Pd/Fe3O4
Pd/Fe3O4
Pd/C
Pd/C
CuCr
iPrOH[b]
94
66
4
88
0
100
100
100
54
0
0
0
46
–
93
0
50
0
–
(0)
7
100
50
100
–
93
2
50
34
–
7
98
50
5
0
0
0
15
–
0
0
0
0.1
–
0
0
0
46
–
0
[a]
H2
iPrOH[b]
[a]
H2
iPrOH[b]
[a]
–
–
CuCr
H2
3
100
0
(100)
10
90
0
0
Conditions: catalyst (0.5 g), 1508C, 24 h; [a] pressure 10 bar; [b] under N2; HEX, ethylcyclohexane; EBE, ethylbenzene; PET, phenylethanol; HET, 2-cyclohexy-
lethanol; HEA, 2-cyclo-hexylethylacetate.
nol led to cleavage of the alkoxy bond (Scheme 2, route A),
the use of molecular hydrogen involved nearly exclusive cleav-
age of the ester CÀO bond (Scheme 2, route B).
Reversing the selectivity in the hydrogenolysis of 2-phenyl-
ethylacetate
To corroborate its hydrogen donor ability under our reaction
conditions, 2-propanol was heated in the presence of Pd/
In the next step, the effect of temperature and pressure on the
(r)
product selectivity was investigated for the Pd/Fe3O4 catalyst.
(r)
Fe3O4 for 24 h to 1508C. Acetone was obtained as main prod-
uct, confirming the ability of 2-propanol to release hydro-
gen.[14] Thus, the different product selectivity must be related
to differences in the elementary steps that take place on the
catalyst surface (see below).
1) In the presence of 2-propanol, an appreciable conversion
(51%) and excellent selectivity towards EBE (97%) was ach-
ieved also at temperatures as low as 1008C (Supporting
Information).
(r)
To verify the outstanding selectivity of the Pd/Fe3O4 cata-
2) In the presence of molecular hydrogen, the conversion in
the hydrogenolysis of PEA increased, as expected, with in-
creasing temperature in the range from 100 to 2008C
(Figure 2). PET was the main product at low temperatures,
lyst for cleaving the ester CÀO bond, EBE and PET were sub-
jected to the same reaction conditions as employed for PEA
(Supporting Information). In both cases, very low conversions
were obtained; products derived from aromatic ring hydroge-
nation were not observed. This confirms that, under the condi-
(r)
tions employed in this study, Pd/Fe3O4 is highly selective to
hydrogenolysis of the CÀO bond versus hydrogenation of the
aromatic ring.
Comparison of Pd/Fe3O4(r) with other catalysts
(r)
The performance of Pd/Fe3O4 was compared with Pd/C and
CuCr as benchmark catalysts. In the hydrogenolysis of PEA
over Pd/C in the presence of 2-propanol, an unexpectedly low
conversion was obtained (4%, Table 1, entry 3). The ester CÀO
and the alkoxy CÀO bond were equally cleaved. Products de-
rived from aromatic ring hydrogenation were not detected. In
contrast, in the presence of H2 at 10 bar (entry 4), the reaction
proceeded to 88% conversion, and a mixture including ring
hydrogenated products was obtained (2-cyclohexylethylace-
tate, HEA, 46%; EBE, 34%; HEX, 15%; PET, 5%). In a reference
experiment, PET was subjected to the same reaction conditions
(Supporting Information). Within 24 h, PET was fully converted
to EBE (79%) and HEX (13%). Formation of HET was not ob-
served. This suggests that hydrogenolysis of the alcohol CÀO
bond precedes the hydrogenation of the aromatic ring
(Scheme 2). Hydrogenolysis of PEA over CuCr in the presence
of 2-propanol or molecular hydrogen provided very low con-
version of PEA to a mixture of PET and EBE (Table 1, entries 5,
6). This is in agreement with reports on the low activity of the
CuCr catalyst.[2]
Figure 2. Product selectivity in the hydrogenolysis of 2-phenylethylacetate
over Pd/Fe3O4 as a function of temperature (reaction at 10 bar of H2).
(r)
whereas an increasing fraction of EBE was formed at higher
(r)
temperatures. This suggests that for Pd/Fe3O4 the activa-
tion energy for cleavage of the ester CÀO bond in PEA is
much lower than that for cleavage of the alkoxy CÀO bond.
Upon increasing the initial hydrogen pressure from 10 to
40 bar while keeping the temperature at 1508C, there was
a slight decrease in conversion of PEA though the selectivity to
PET remained above 95%. Such decrease in conversion with in-
creasing hydrogen pressure was also reported for other hydro-
genolysis reactions over supported palladium catalysts and is
attributed to progressive displacement of the substrate ad-
sorbed on the surface of the catalyst by hydrogen.[17]
ChemCatChem 2016, 8, 1515 – 1522
www.chemcatchem.org
1517
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Upon exploring the effect of the availability of hydrogen,
a surprising change in chemoselectivity was observed, if the
reaction was performed at lower hydrogen pressures. Whereas
PET was the main product at pressures as low as 5 bar, a mix-
ture of PET and EBE was obtained at 2 bar. Noteworthy, EBE
was also obtained as the main product if 2-propanol was used
as H source. Hydrogen chemically bound in 2-propanol is less
available and, thus, resembles the use of molecular hydrogen
at low pressures. This implies that it is the availability of hydro-
gen that determines the selectivity in the cleavage of CÀO
bonds as outlined in Figure 3.
performed at high temperatures (1508C) showed full conver-
sion. Toluene was obtained as the only product (route A). In
contrast, at 508C using molecular H2 (10 bar) as H source, full
conversion and a selectivity of >99% for the cleavage of the
CÀO alkoxy bond was achieved.[16] Toluene and acetic acid (de-
tected as isopropyl acetate) were the only products (route A).
This exemplifies that iPrOH is a potent hydrogen donor only at
high reaction temperatures, whereas molecular hydrogen func-
tions as a readily available H source at low reaction tempera-
tures.
Reaction of 3-phenylpropylacetate (PPA) with iPrOH in the
(r)
presence of Pd/Fe3O4 at 1508C showed a conversion of 14%,
with a selectivity of 92% to phenylpropanol (PPO) and of 8%
to propylbenzene (prB) (route B). The reaction performed in
the presence of H2 at 10 bar provided a conversion of 19%
and a selectivity to PPO and prB of 98 and 2%, respectively
(route B). Thus, the experimental data demonstrate that for
(r)
BAC and PPA as well as for PEA, Pd/Fe3O4 has a high selectivi-
ty towards CÀO cleavage, while the phenyl ring is not
converted.
For comparison, BAC and PPA were also tested in the pres-
ence of Pd/C. The hydrogenolysis of BAC with iPrOH over Pd/C
at 508C provided no conversion of BAC similar to the reaction
(r)
over Pd/Fe3O4 while at 1508C a conversion of 10% was ob-
served. Toluene was obtained as the only product (route A). By
contrast, the hydrogenolysis of BAC at 508C with molecular H2
provided a conversion of >99% with complete selectivity to
toluene >99% (route A).
Figure 3. Effect of the availability of hydrogen on the product selectivity in
the hydrogenolysis of 2-phenylethylacetate over Pd/Fe3O4 upon varying be-
tween the use of 2-propanol and different pressures of molecular hydrogen.
The hydrogenolysis of PPA in the presence of iPrOH over
Pd/C at 1508C provided a conversion of 23% and a selectivity
of 10 and 89% to PPO and prB, respectively (route A). The re-
action in the presence of molecular H2 (10 bar) over Pd/C at
1508C provided a very high conversion (99%) with a selectivity
of 84% to propylcyclohexane (PCH) resulting from consecutive
hydrogenolysis of the CÀOH bond of PPO, and some minor
products (15%).
Role of the distance between phenyl ring and ester group
To explore the substrate scope, the length of the alkyl linker
between the phenyl ring and the ester group was varied
(Table 2). Benzylacetate (BAC) was not converted at all over Pd/
(r)
Fe3O4 using iPrOH as H source at 508C. The same reaction
The influence of the pressure of molecular hydrogen was
(r)
also investigated for the hydrogenolysis of PPA over Pd/Fe3O4
Table 2. Selectivity to cleavage of the alkoxy CÀO bond (route A) or the
ester CÀO bond (route B) in phenylesters dependent on the choice of
catalyst and H source.
(1508C) in comparison to iPrOH as H source. The conversion of
PPA increased from 14%, when iPrOH was used as H source, to
21 and 19% in the presence of molecular H2 (5 and 10 bar, re-
spectively). In all cases, the selectivity to PPO was very high
(84–98%). In the presence of Pd/C, the conversion of PPA in-
creased markedly from 23% upon using iPrOH to 79 and
>99% using molecular hydrogen at 5 and 10 bar, respectively.
Noteworthy was that the selectivity was reversed and prB was
obtained as main product upon using iPrOH, while 37 and
84% of PCH were obtained at a H2 pressure of 5 and 10 bar, re-
spectively. Thus, in the hydrogenolysis of PPA the influence of
(r)
Substrate
Pd/Fe3O4
iPrOH[a]
Pd/C
iPrOH[a]
[b]
[b]
H2
H2
BAC
PEA
PPA
A
A
B
A[c]
A
A[c]
the H source is much more pronounced for Pd/C than for Pd/
(r)
Fe3O4
.
B
A/B
A
B
Upon raising the temperature, the conversion of PPA in the
presence of iPrOH (Pd/Fe3O4) increased from 7, 14 to 86%
(100, 150 to 2008C, respectively), while the selectivity to the
major product PPO increased from 82% (1008C) to 92%
(1508C) and then decreased to 52% (2008C). With the increase
in temperature, the selectivity to prB also varied (18, 8, 46%,
B
B
Conditions: catalyst (0.5 g), 1508C, 24 h; [a] under N2; [b] 10 bar pressure;
[c] 508C.
ChemCatChem 2016, 8, 1515 – 1522
www.chemcatchem.org
1518
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
respectively) indicative of consecutive hydrogenolysis of the C-
OH group in PPA. The reaction performed in the presence of
H2 (10 bar) gave very similar conversions (5, 19 and 96%) and
selectivities to PPO (97, 98 and 61% at 100, 150 and 2008C,
respectively).
lecular hydrogen as H source. This low reactivity of alkylace-
tates is consistent with a previous study in which very harsh
conditions were needed for hydrogenolysis of alkylacetates to
alkanes, CO2, and alcohols.[18]
The hydrogenolysis of PPA with iPrOH in the presence of Pd/
C showed a rise in conversion from 7, to 23% and full conver-
sion upon increasing the temperature from 100 to 1508C and
then to 2008C. The selectivity for the major product prB de-
creased from 92, 89 to 7% (100, 150 to 2008C). In the presence
of molecular hydrogen, complete conversion was obtained in-
dependent of the temperature applied, whereas the selectivity
to propylcyclohexane (PCH) decreased from 91, 84 to 72%
(100, 150 to 2008C) in favor of the formation of prB (20% at
2008C, route A).
Mechanism of CÀO bond cleavage
An important difference between the two H sources is the dif-
ferent availability of hydrogen and, consequently, the very dif-
(r)
ferent hydrogen coverage on the surface of the Pd/Fe3O4 cat-
alyst. Even though hydrogen can be transferred from 2-propa-
nol to the metal surface, the rate of 2-propanol dissociation
limits the surface coverage with hydrogen atoms.[19] In con-
trast, in the presence of an increasing pressure of molecular
hydrogen the palladium surface becomes increasingly saturat-
ed with hydrogen atoms (Ead =À2.96 eV for PdÀH).[19a] Further-
more, interstitial octahedral sites of the palladium lattice can
be occupied.[19b] Consequently, different reaction pathways
appear more likely in dependence of the choice of the
reductant.
Hydrogenolysis of alkyl acetates
To better understand the importance of the aromatic group in
the cleavage of the CÀO bond, the cleavage of alkyl acetates
varying in the length of the alkyl chain was investigated
(Table 3). Butylacetate (BTA) and octylacetate (OCA) were con-
For 2-propanol, a large number of free Pd sites enables
ready coordination of the substrate PEA (Ead =À0.21 eV),[20]
whereby it is likely that both oxygen atoms of the ester group
interact with Pd. Owing to steric constraints, the adsorption of
PEA involves an agostic interaction of the a-methylene group
with a neighboring Pd atom (Scheme 3) thereby enhancing
the positive partial charge at the carbon atom. In contrast, the
aromatic ring is adsorbed preferentially onto exposed Fe
atoms in vicinity (Ead =À1.10 eV for benzene).[21]
Table 3. Selectivity to cleavage of the alkoxy CÀO bond (route A) or the
ester CÀO bond (route B) in alkylesters dependent on the choice of cata-
lyst and H source.
We propose that the reaction proceeds through a transfer
hydrogenation mechanism[22] without involvement of surface-
bound H-atoms. Transfer of the a-hydrogen atom of 2-propa-
nol to the a-methylene moiety of adsorbed PEA is promoted
by the activation of the neighboring OH-group of 2-propanol
through coordination to Pd. The agostic interaction in ad-
sorbed PEA leads to weakening of the alkoxy CÀO bond, ena-
bling simultaneous transfer of the a-hydrogen atom and cleav-
age of the alkoxy CÀO bond. Subsequently, the 2-propanol OH
proton is transferred to the acetate moiety. Overall, the sur-
face-assisted transfer hydrogenation enables cleavage of the
PEA alkoxy CÀO bond to provide EBE and acetic acid.
(r)
Substrate
Pd/Fe3O4
iPrOH[a]
Pd/C
iPrOH[a]
[b]
[b]
H2
B
H2
MBT
BTA
B
B
B
B[c]
B[c]
B[c]
B[c]
B[c]
B[c]
B
B
OCA
n=0, 3, 6; conditions: catalyst (0.5 g), 1508C, 24 h; [a] under N2; [b] pres-
sure 10 bar; [c] conversion <1%.
In the presence of H2, a large number of Pd sites are occu-
pied by hydrogen atoms, and the substrate PEA is adsorbed
preferentially through the aromatic ring onto exposed Fe
(r)
verted at 1508C over Pd/Fe3O4 using 2-propanol as H source.
(r)
The conversion was 8.4 and 15%, respectively, with a selectivity
of 100% to the corresponding alcohols 1-butanol and 1-octa-
nol (route B). The same reactions performed in the presence of
10 bar of H2 provided similar conversion (7.6% for butylacetate
and 14% for octylacetate) as well as 100% of selectivity to the
corresponding alcohols. In contrast, the use of methyl butyrate
(MBT) as substrate led to even lower conversion at similar high
selectivities (>80%) to 1-butanol. Thus, route B dominated for
alkylacetates. The surprisingly low reactivity of aliphatic sub-
strates demonstrates that the adsorption of the aromatic ring
atoms of the Fe3O4 support (Scheme 4). Transfer of two hy-
drogen atoms from Pd to the ester functionality leads to cleav-
age of the ester CÀO bond to provide PET and acetaldehyde.
The latter is then hydrogenated rapidly to ethanol.
The fact that hydrogenation of the aromatic ring is not ob-
served is most likely related to the strong preference for ad-
sorption of the aromatic ring onto exposed Fe atoms of the
(r)
Fe3O4 support. This preference results from the electronic
structure of Fe3O4 that remains unaffected by the interactions
with the palladium nanoparticles.[21] Thereby, the orbitals of
the Fe atoms are electronically oriented in such a way that
they readily overlap with the aromatic system. In contrast, the
(r)
on the Pd/Fe3O4 catalyst plays a key role. In comparison, the
conversion for all three alkylacetates was very low (1%) in the
presence of Pd/C both in the presence of isopropanol and mo-
(r)
interaction with the Fe3O4 support leads to strong deviation
ChemCatChem 2016, 8, 1515 – 1522
www.chemcatchem.org
1519
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
(r)
Scheme 3. Hydrogen transfer mechanism proposed for the hydrogenolysis of the alkoxy CÀO bond of 2-phenylethylacetate over Pd/Fe3O4 in the presence
of 2-propanol.
(r)
Scheme 4. Reaction mechanism proposed for the hydrogenolysis of the ester CÀO bond of 2-phenylethylacetate over Pd/Fe3O4 in the presence of molecular
hydrogen.
of the Pd d-band of the palladium nanoparticles from the
Fermi level.[21] In a reference experiment, PEA was allowed to
react over the parent Fe3O4 in the presence of molecular hy-
drogen (10 bar). No conversion of the substrate was observed
confirming that the presence of Pd nanoparticles is essential
for the hydrogenolysis reaction to occur.
at higher fields (64.0 ppm). Similarly, the chemical shift of the
alkoxy carbon was 63.8 ppm for PPA. Thus, the alkoxy CÀO
bond becomes less polarized in the series BAC–PEA–PPA con-
sistent with the increasing propensity for route B. A low elec-
tron density on the carbon atom, as in the case of BAC, ren-
ders the CÀO bond more exposed to the approach of a nega-
tively polarized hydrogen atom. In contrast, increased electron
density on the carbon atom, as in the case of PPA, results in
a less polarized and stronger alkoxy CÀO bond allowing the
Pd-H species to react more readily with the C=O carbon atom.
For the aliphatic acetates, the carbonyl stretch vibration
nst(C=O) was blueshifted to 1736–1742 cmÀ1 relative to that of
the aromatic acetates (Figure 5) indicative of a stronger C=O
bond. The CÀO stretch vibration was broadened or involved
two vibration modes, whereby the absorbance maximum was
As a measure for the relative bond strengths, the IR vibra-
tions of the substrates as well as the NMR data were analyzed
in detail. For the aromatic acetates (Figure 4) the carbonyl
stretch vibration nst(C=O) was observed at a similar position
(1734-1735 cmÀ1). The CÀO stretch vibration of BAC was found
at 1221 cmÀ1, whereas the CÀO stretch vibration was observed
at 1231 cmÀ1 for PEA and PPA. Lower wave numbers corre-
spond to a weakened bond[23] and it appears reasonable that
the alkoxy CÀO bond of benzylacetate is relatively easy to
break compared to those of PEA and PPA. The 13C NMR signal
of the CH2 group of BAC was observed at 66.2 ppm. In com-
parison, the corresponding alkoxy carbon of PEA was observed
observed at varying frequencies (1174, 1227, and 1169 cmÀ1
,
MBT, BTA, and OCA, respectively). The 13C NMR signal of the
methyl group neighboring the carbonyl group in MTB (CH3-
ChemCatChem 2016, 8, 1515 – 1522
www.chemcatchem.org
1520
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
phatic acetate esters (route B) and the low reactivity must
therefore be related to the absence of an aromatic group facili-
tating the coordination of the substrate on the catalyst
surface.
As compared to PEA (route A or B dependent on the reac-
tion conditions), the selectivity of the CÀO bond cleavage is
specific for BAC (route A), PPA (route B), and alkyl esters
(route B). This suggests that the electronic state of the carbon
in the (RÀCH2ÀOÀ) group plays a synergistic role. The decisive
factors controlling the selectivity include (i) coordination of the
(r)
aromatic ring to the Fe atoms of the Fe3O4 support, (ii) the
availability of hydrogen on the surface of the palladium nano-
particles as well as (iii) the electronic factors in the substrate.
For PEA, the control of the availability of hydrogen enables the
specific hydrogenolysis of either the alkoxy or the ester CÀO
bond.
Conclusions
The principles allowing for the selective cleavage of either the
(r)
alkoxy CÀO bond or the ester CÀO bond over Pd/Fe3O4 as
catalyst were studied for 2-phenylethylacetate (PEA) as base
case. The alkoxy CÀO bond was cleaved preferentially in the
presence of 2-propanol, whereas cleavage of the ester CÀO
bond was favored in the presence of molecular hydrogen. Sim-
ilarly, by using very low H2 pressures, the selectivity for cleav-
age of the ester CÀO bond was reversed, allowing for the pref-
erential cleavage of the alkoxy CÀO bond. This suggests that it
is the low availability of hydrogen on the surface of the Pd
nanoparticles that makes a surface-assisted hydrogen transfer
mechanism likely in the case that 2-propanol is used as H
source. This is a variant of the classic hydrogenolysis mecha-
nism that occurs if molecular hydrogen is used.
Figure 4. FTIR analysis of the aromatic acetate esters employed in this study
and selected NMR data.
Noteworthy, a very high selectivity to hydrogenolysis was
observed, whereas hydrogenation of the aromatic ring was not
observed independent of the choice of the H source or the re-
action conditions adopted. This exceptional preference to-
wards cleavage of the alkoxy or ester CÀO bond without paral-
lel hydrogenation of the aromatic ring is explained by the
strong interaction between the aromatic system and exposed
(r)
Fe atoms derived from the support. The Pd/Fe3O4 catalyst,
thus, revealed superior performance for the selective cleavage
of CÀO bonds in the presence of aromatic rings compared to
other heterogeneous catalysts. Furthermore, the activity for
(r)
conversion of alkylacetates demonstrated Pd/Fe3O4 to be an
exceptionally active and selective catalyst for CÀO bond
cleavage.
Figure 5. FTIR analysis of the aliphatic acetate esters employed in this study
and selected NMR data.
Experimental Section
(c)
Pd/Fe2O3 [(c)=coprecipitated] with a nominal palladium loading
OC(O)nPr) was observed at 52.3 ppm and compares to a posi-
tion of the signal assigned to the methylene group in BTA and
OCA at 64.0 and 64.9 ppm, respectively. The low fields of the
13C NMR signals of the CH3 group of MTB and the CH2 group
of BTA and OCA are comparable to that of the aromatic ace-
tates. The preference for cleavage of the ester CÀO bond in ali-
of 5 wt% was prepared by coprecipitation. Palladium chloride (an-
hydrous, Fluka, 60 wt% Pd) was dissolved in aqueous HCl (33%).
The solution was poured into an aqueous solution of iron(III)nitrate
nonahydrate (Fluka). The mixture was then added dropwise to an
aqueous solution of Na2CO3 (1m). The precipitate was filtered off,
washed with water and dried at 808C under a partial vacuum for
ChemCatChem 2016, 8, 1515 – 1522
www.chemcatchem.org
1521
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
(r)
J. H. Bitter, D. S. van Es, ChemSusChem 2013, 6, 1576–1594; c) A. M.
Ruppert, K. Weinberg, R. Palkovits, Angew. Chem. Int. Ed. 2012, 51,
2564–2601; Angew. Chem. 2012, 124, 2614–2654.
1 day. Thereafter, the sample was reduced to Pd/Fe3O4 by heating
for 2 h to 2008C under flowing hydrogen. Characterization of the
Pd/Fe3O4 catalyst revealed a Pd loading of 8.7 wt% (ICP), a BET
(r)
[7] a) A. F. H. Studentschnig, S. Schober, M. Mittelbach, Energy Fuels 2013,
27, 7480–7484; b) T. Morgan, E. Santillan-Jimenez, A. E. Harman-Ware, Y.
Ji, D. Grubb, M. Crocker, Chem. Eng. J. 2012, 189–190, 346–355; c) G. W.
Huber, A. Corma, Angew. Chem. Int. Ed. 2007, 46, 7184–7201; Angew.
Chem. 2007, 119, 7320–7338.
surface of 170 m2 gÀ1 (H2-chemisorption) and an average particle
size for the Pd nanoparticles of 1.2 nm (TEM). The benchmark cata-
lysts Pd/C (5 wt% Pd) and CuCr were purchased from commercial
sources and used as received.
[8] a) Z. Li, R. S. Assary, A. C. Atesin, L. A. Curtiss, T. J. Marks, J. Am. Chem.
Soc. 2014, 136, 104–107; b) M. Chatterjee, T. Ishizaka, A. Suzuki, H. Ka-
wanami, Chem. Commun. 2013, 49, 4567–4569; c) A. G. Sergeev, J. D.
Webb, J. F. Hartwig, J. Am. Chem. Soc. 2012, 134, 20226–20229; d) B. S.
Samant, G. W. Kabalka, Chem. Commun. 2012, 48, 8658–8660; e) A. G.
Sergeev, J. F. Hartwig, Science 2011, 332, 439–443.
[9] a) V. Molinari, C. Giordano, M. Antonietti, D. Esposito, J. Am. Chem. Soc.
2014, 136, 1758–1761; b) M. Zaheer, J. Hermannsdorfer, W. P. Kretsch-
mer, G. Motz, R. Kempe, ChemCatChem 2014, 6, 91–95; c) Y. Li, H. Liu, L.
Maab, D. He, RSC Adv. 2014, 4, 5503–5512; d) C. Zhao, J. A. Lercher,
Angew. Chem. Int. Ed. 2012, 51, 5935–5940; Angew. Chem. 2012, 124,
6037–6042; e) T. K. Mukhopadhyay, M. Flores, T. L. Groy, R. J. Trovitch, J.
Am. Chem. Soc. 2014, 136, 882–885; f) S. P. Y. Cutulic, N. J. Findlay, S.
Zhou, E. J. T. Chrystal, J. A. Murphy, J. Org. Chem. 2009, 74, 8713–8718;
g) H. Tatamidani, K. Yokota, F. Kakiuchi, N. Chatani, J. Org. Chem. 2004,
69, 5615–5621; h) J. Zakzeski, P. C. A. Bruijnincx, A. L. Jongerius, B. M.
Weckhuysen, Chem. Rev. 2010, 110, 3552–359.
Catalytic reactions were performed in a 160 mL stainless-steel reac-
tor (Parr Instruments) equipped with a gas-entrainment stirrer. The
reactor was loaded with a suspension of catalyst (0.5 g) in a solu-
tion of the desired substrate in 2-propanol (75 mL, 4 wt%). The re-
actor was purged with N2 (99.99%), the system was then pressur-
ized with the desired gas (H2 or N2), stirred at 500 rpm, and, finally,
heated up to the reaction temperature for 24 h. Then, the reactor
was cooled to RT, the pressure released and the composition of
the organic phase was analyzed by gas chromatography.
Gas chromatographic analyses were performed on a HP 5890 gas
chromatograph equipped with a wide-bore capillary column (CP-
WAX 52CB, 60 m, i.d. 0.53 mm) and flame ionization detector. Con-
version and selectivity to the product i were calculated on the
basis of Equations (1) and (2), respectively.
[10] M. G. Musolino, C. Busacca, F. Mauriello, R. Pietropaolo, Appl. Catal. A
2010, 379, 77–86.
[11] J. Liu, B. Sun, J. Hu, Y. Pei, H. Li, M. Qiao, J. Catal. 2010, 274, 287–295.
[12] M. G. Musolino, L. A. Scarpino, F. Mauriello, R. Pietropaolo, ChemSu-
sChem 2011, 4, 1143–1150.
[13] a) C.-T. Wu, K. M. K. Yu, F. Liao, N. Young, P. Nellist, A. Dent, A. Kroner,
S. C. E. Tsang, Nat. Commun. 2012, 3, 1050; b) F. Liao, T. W. B. Lo, D.
Sexton, J. Qu, C. T. Wu, S. C. E. Tsang, Catal. Sci. Technol. 2015, 5, 887–
896.
niniðiÞÀntðiÞ
ð1Þ
ð2Þ
Conversion ¼
Selectivity ¼
 100 ½%Š
 100 ½%Š
niniðiÞ
ntðjÞ
niniðiÞÀntðiÞ
nini =molar amount of substrate i at time 0; nt =molar amount of
substrate i at time t; nt(j)=molar amount of product j at time t.
[14] a) M. G. Musolino, L. A. Scarpino, F. Mauriello, R. Pietropaolo, Green
Chem. 2009, 11, 1511–1513; b) F. Mauriello, H. Ariga, M. G. Musolino, R.
Pietropaolo, S. Takakusagi, K. Asakura, Appl. Catal. B 2015, 166–167,
121–131; c) D. Scholz, C. Aellig, I. Hermans, ChemSusChem 2014, 7,
268–275; d) F. Mauriello, A. Vinci, C. Espro, B. Gumina, M. G. Musolino,
R. Pietropaolo, Catal. Sci. Technol. 2015, 5, 4466–4473.
[15] a) A. Martin, U. Armbruster, I. Gandarias, P. L. Arias, Eur. J. Lipid Sci. Tech-
nol. 2013, 115, 9–27; b) A. Shrotri, H. Kobayashi, A. Tanksale, A. Fukuo-
ka, J. Beltramini, ChemCatChem 2014, 6, 1349–1356; c) H. Kobayashi, H.
Matsuhashi, T. Komanoya, K. Haraa, A. Fukuoka, Chem. Commun. 2011,
47, 2366–2368.
Acknowledgements
The financial support of POR Calabria – FSE 2007/2013 (Project
Number 2012.M2.114) and Mediterranea University of Reggio
Calabria is gratefully acknowledged. F.M. acknowledges the
German Academic Exchange Service (DAAD) for financial support
(Grant Number A/13/04366).
[16] a) G. V. Smith, F. Notheisz, Heterogeneous catalysis in organic chemistry,
1999, 144–151, Chap. 4; b) S. Rajagopal, A. F. Spatola, Appl. Catal. A
1997, 152, 69–81; c) H. L. Smith, E. S. Brown, J. D. Smith, J. Andrako, J.
Pharm. Sci. 1965, 54, 1269–1273.
Keywords: chemoselectivity
hydrogen · hydrogenolysis · palladium
·
heterogeneous catalysis
·
[17] H. Bernas, A. Taskinen, J. Wärnåärnaen, D. Y. Murzin, J. Mol. Catal. A
2009, 306, 33–39.
[18] J. Han, H. Sun, Y. Ding, H. Lou, X. Zheng, Green Chem. 2010, 12, 463–
467.
[1] L. A. Saudan in Sustainable Catalysis: Challenges and Practices for the
Pharmaceutical and Fine Chemical Industries, Wiley-VCH, Weinheim,
2013, pp. 37–61.
[2] a) T. Turek, D. L. Trimm, N. W. Cant, Catal. Rev. Sci. Eng. 1994, 36, 645–
683; b) T. Turek, D. L. Trimm, D. StC. Black, N. W. Cant, Appl. Catal. A
1994, 116, 137–150; c) J. W. Evans, N. W. Cant, D. L. Trimm, M. S. Wain-
wright, Appl. Catal. 1983, 6, 355–362.
[3] Global Markets for Oleochemical Fatty Acids, BCC Research LLC, Welles-
ley, MA, USA, 2013, pp. 1–326.
[4] a) E. Tílvez, G. I. Cardenas-Jiron, M. I. Menendez, R. Lopez, Inorg. Chem.
2015, 54, 1223–1231; b) Solvents Market (Alcohols, Hydrocarbons, Ke-
tones, Esters, Chlorinated And Others)—Global Industry Analysis, Size,
Share, Growth, Trends And Forecast 2012–2018, Research and Markets,
Dublin, 2013, pp. 1–94; c) R. J. Trovitch, E. Lobkovsky, M. W. Bouwkamp,
P. J. Chirik, Organometallics 2008, 27, 6264–6278.
[19] a) M. Tamura, K. Kon, A. Satsuma, K. Shimizu, ACS Catal. 2012, 2, 1904–
1909; b) W. Reeve, C. M. Erikson, P. F. Aluotto, Can. J. Chem. 1979, 57,
2747–2754.
[20] a) L. Xu, Y. Xu, Surf. Sci. 2010, 604, 887–892; b) L. Xu, Y. Xu, Catal. Today
2011, 165, 96–105.
[21] a) A. J. R. Hensley, R. Zhang, Y. Wang, J.-S. McEwen, J. Phys. Chem. C
2013, 117, 24317–24328; b) A. J. R. Hensley, Y. Hong, R. Zhang, H.
Zhang, J. Sun, Y. Wang, J.-S. McEwen, ACS Catal. 2014, 4, 3381–3392.
[22] a) X. Liu, G. Lu, Y. Guo, Y. Guo, Y. Wang, X. Wang, J. Mol. Catal. A 2006,
252, 176–180; b) F. Alonso, P. Riente, M. Yus, Tetrahedron 2009, 65,
10637–10643; c) J. Feng, C. Yang, D. Zhang, J. Wang, H. Fu, H. Chen, X.
Li, Appl. Catal. A 2009, 354, 38–43.
[5] a) E. McCusker-Orth, J. Ellen, Pat. Appl., US 6919489 B1, 2005; b) G.
Irick; P. N. Mercer, Pat. Appl. US 4929777A; c) D. S. Brands, E. K. Poels,
T. A. Krieger, O. V. Makarova, C. Weber, S. Veer, A. Bliek, Catal. Lett. 1996,
36, 175–182.
[23] R. Badger, J. Chem. Phys. 1934, 2, 128.
Received: December 27, 2015
[6] a) T. V. Choudhary, C. B. Phillips, Appl. Catal. A 2011, 397, 1–12; b) R. W.
Gosselink, S. A. W. Hollak, S.-W. Chang, J. Van Haveren, K. P. de Jong,
Published online on April 13, 2016
ChemCatChem 2016, 8, 1515 – 1522
www.chemcatchem.org
1522
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim