Structure, activity, and selectivity of bimetallic Pd-Fe/SiO2 and Pd-Fe/Γ-Al2O3 catalysts for the conversion of furfural

Article abstract of DOI:10.1016/j.jcat.2017.03.016

The conversion of furfural has been investigated in vapor and liquid phases over a series of supported monometallic Pd and bimetallic Pd-Fe catalysts. Over the monometallic Pd/SiO₂ catalyst, the decarbonylation reaction dominates, yielding furan as the main product. By contrast, over the bimetallic Pd-Fe/SiO₂ catalyst a high yield of 2-methylfuran is obtained with much lower yield to furan. Interestingly, changing the catalyst support affects the product distribution. For instance, using γ-Al₂O₃ instead of SiO₂ as support of the bimetallic catalyst changed the dominant product from 2-methylfuran to furan. That is, Pd-Fe/γ-Al₂O₃ behaves more like monometallic Pd/SiO₂ than bimetallic Pd-Fe/SiO₂. A detailed characterization of the catalysts via XPS, XRD, and TEM indicated that a Pd-Fe alloy is formed on the SiO₂ support but not on the γ-Al₂O₃ support. Theoretical density functional theory calculations suggest that on the Pd-Fe alloy binding of the furan ring to the surface is weakened compared to on pure Pd. This weakening disfavors the ring hydrogenation and decarbonylation paths, while the oxophilic nature of Fe atoms enhances the interaction of the C[dbnd]O and the OH groups with the metal surface, which favors the C[dbnd]O hydrogenation and C–O bond cleavage paths. The presence of the solvent has a less pronounced effect, but clearly has a stronger inhibition on C–C bond cleavage (decarbonylation to furan) than on C–O bond cleavage (hydrogenolysis to methylfuran).

Full text of DOI:10.1016/j.jcat.2017.03.016

Journal of Catalysis 350 (2017) 30–40
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Structure, activity, and selectivity of bimetallic Pd-Fe/SiO2 and Pd-Fe/
Al2O3 catalysts for the conversion of furfural
c-
Natalia Pino a, Surapas Sitthisa c, Qiaohua Tan c, Talita Souza b, Diana López a, Daniel E. Resasco c,
⇑
a Química de Recursos Energéticos y Medio Ambiente, Instituto de Química, Facultad de Ciencias Exactas y Naturales, Universidad de Antioquia UdeA, Calle 70 No. 52-21,
Medellín, Colombia
b University of Minas Gerais, Chemistry Department, Belo Horizonte, MG, Brazil
c School of Chemical, Biological and Materials Engineering and Center for Biomass Refining, University of Oklahoma, Norman, OK 73019, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
The conversion of furfural has been investigated in vapor and liquid phases over a series of supported
monometallic Pd and bimetallic Pd-Fe catalysts. Over the monometallic Pd/SiO2 catalyst, the
decarbonylation reaction dominates, yielding furan as the main product. By contrast, over the bimetallic
Received 2 February 2017
Revised 16 March 2017
Accepted 17 March 2017
Pd-Fe/SiO2 catalyst
Interestingly, changing the catalyst support affects the product distribution. For instance, using
-Al2O3 instead of SiO2 as support of the bimetallic catalyst changed the dominant product from
2-methylfuran to furan. That is, Pd-Fe/ -Al2O3 behaves more like monometallic Pd/SiO2 than bimetallic
Pd-Fe/SiO2. A detailed characterization of the catalysts via XPS, XRD, and TEM indicated that a Pd-Fe alloy
is formed on the SiO2 support but not on the -Al2O3 support. Theoretical density functional theory
a high yield of 2-methylfuran is obtained with much lower yield to furan.
c
Keywords:
Furfural
c
Furfuryl alcohol
Methylfuran
Pd-Fe alloys
Hydrogenolysis
Hydrogenation
Decarbonylation
Bio-oil upgrading
c
calculations suggest that on the Pd-Fe alloy binding of the furan ring to the surface is weakened com-
pared to on pure Pd. This weakening disfavors the ring hydrogenation and decarbonylation paths, while
the oxophilic nature of Fe atoms enhances the interaction of the C@O and the OH groups with the metal
surface, which favors the C@O hydrogenation and CAO bond cleavage paths. The presence of the solvent
has a less pronounced effect, but clearly has a stronger inhibition on CAC bond cleavage (decarbonylation
to furan) than on CAO bond cleavage (hydrogenolysis to methylfuran).
Ó 2017 Elsevier Inc. All rights reserved.
1. Introduction
quently reduces the carbon chain length. Bimetallic catalysts
[13–17] have attracted attention because they display catalytic
Furfural (FAL) is a renewable platform molecule derived from
lignocellulosic biomass with a good potential to be directly used
in the production of fuel components and chemicals [1–3]. Due
to its high reactivity, furfural requires selective deoxygenation to
obtain compounds stable enough to undergo further upgrading
[4]. The selective deoxygenation to 2-methylfuran (MF) is of partic-
ular interest since this compound has been proposed as a potential
additive to gasoline [5] with good energy density, boiling point,
and octane number, as well as a chemical intermediate [6–8].
Therefore, it is important to examine catalysts that effectively
break C@O bonds while preserving CAC bonds.
properties that differ from those of their parent metals, particularly
in the conversion of biomass-derived compounds. As a result, a ser-
ies of novel catalysts with enhanced selectivity, activity, and stabil-
ity have been investigated. For example, studies on bimetallic Pd-
Cu and Ni-Fe catalysts [7,18] have shown that bimetallic alloys can
greatly alter the furfural reaction paths. The incorporation of Cu
onto Pd/SiO2 catalyst, reduced the decarbonylation rate, while
increasing the selectivity to hydrogenation products such as fur-
furyl alcohol. Furthermore, the addition of Fe to silica-supported
Ni, changed the product selectivity from furan to 2-methylfuran.
This behavior was ascribed to the oxophilic nature of Fe, which
Previous studies [9–11] have shown that Cu-based catalysts are
highly selective for furfural hydrogenation to furfuryl alcohol,
preserving the CAC bond. Other metals such as Ni and Pd [6,12]
catalyze side reactions involving CAC bond scission that conse-
makes the di-bonded g
2(C,O)-furfural more stable than on the
pure Ni surface and hinders the formation of acyl species, involved
in the decarbonylation reaction. In a combined study of density
functional theory (DFT) and surface science measurements on Fe/
Ni(111) model surfaces, Yu et al. [19] gave further evidence for
the preferential adsorption of furfural via the carbonyl group, with
the furan ring tilted away from the surface. In good agreement
⇑
Corresponding author.
E-mail address: resasco@ou.edu (D.E. Resasco).
http://dx.doi.org/10.1016/j.jcat.2017.03.016
0021-9517/Ó 2017 Elsevier Inc. All rights reserved.

N. Pino et al. / Journal of Catalysis 350 (2017) 30–40
31
with the conclusions derived from the supported catalyst study
[18], these authors concluded that such configuration enhances
the production of 2-methylfuran, with furfuryl alcohol acting as
an intermediate.
Expanding the investigations of bimetallic catalysts for the
conversion of furfural, in this contribution, we have inspected the
reaction pathways of furfural over Pd-Fe catalysts. Specifically,
we have explored whether different supports may result in varying
degrees of metal-metal interaction and consequent modifications
of catalytic behavior. To that end, we prepared and characterized
bimetallic Pd-Fe catalysts supported on silica (SiO2) and
situ dehydrated samples from a glove box to the analysis chamber
without exposure to air. A 93.9 eV and 58.7 eV pass energy was
typically used for survey and specific element analysis, respec-
tively. The electron take-off angle was 45° with respect to the sam-
ple surface. The binding energies were adjusted to the C signal at
284.6 eV as an internal reference.
Morphology and particle size of the Pd/SiO2, Pd-Fe/SiO2 and
Pd-Fe/c-Al2O3 catalysts were characterized by transmission
electron microscopy (TEM – Tecnai F20 Super Twin TMP), with a
resolution of 0.1 nm, at 200 kV accelerating voltage. Prior to the
analysis, the samples were reduced ex situ in flowing H2 at
250 °C for 3 h. The reduced samples were suspended in propanol
by sonication, deposited on a TEM (Cu) grid, and dried overnight.
The particle size and crystallographic planes were measured from
HRTEM images using DigitalMicrograph and ImageJ software.
c
-alumina (c-Al2O3), as two examples of supports that represent
weak and strong interaction with the metal components, respec-
tively. To complement and help analyze the experimental results,
theoretical density functional theory (DFT) calculations were con-
ducted for Pd and Pd-Fe surfaces. As discussed below, Pd and Fe
have been known to produce a range of FCC (face-centered cubic)
solid solutions, which depending on composition and temperature
can adopt regular FCC, distorted FCC, or even FCT (face-centered
tetragonal) structures [20,21]. Our aim in this work was to under-
stand how the formation of the Pd-Fe bimetallic alloy affects the
HDO pathway of furfural to produce 2-methylfuran and to deter-
mine whether the extent of alloy formation is affected by the sup-
port used.
2.3. Catalytic activity
The furfural conversion over Pd and Pd-Fe catalysts was evalu-
ated in liquid and vapor phase, with the purpose of investigating
how the reaction medium affects the mechanism and product
distribution.
2.3.1. Vapor phase
The vapor-phase conversion of furfural over 1%Pd/SiO2, 1%Pd-
0.5%Fe/SiO2 and 1%Pd-0.5%Fe/
1/400 tubular quartz reactor. The pelletized catalyst (size range:
250–425 m) was placed at the center of the reactor between
c-Al2O3 catalysts was studied in a
2. Experimental and theoretical methods
l
2.1. Catalytic materials
two layers of glass beads and quartz wool. Calculations were done
to ensure that external and internal mass transfer limitations were
eliminated following the criteria proposed by Madon and Boudart
[22]. The catalyst was pre-reduced in flow of H2 (60 ml/min, Airgas,
99.99%) for 1 h at 250 °C. After reduction, the catalyst was cooled
down to the selected temperature (210–250 °C) under the same
H2 flow rate. Prior to the reaction, the as-received furfural
(Sigma-Aldrich, 99.5%, brown color) was purified by vacuum distil-
lation to remove residues and any oligomers formed during stor-
age. The purified liquid was kept under He atmosphere until its
use in the reaction test. This is an important step to minimize deac-
tivation by deposition of oligomers over the surface.
A 0.5 ml/h flow of liquid furfural was fed continuously from a
syringe pump (Cole Palmer) and vaporized into a gas stream of
60 ml/min H2. To keep all the compounds in the vapor phase all
lines before and after the reactor were kept heated at 220 °C using
heating tapes. To vary the space time (W/F = catalyst mass/mass
flow rate of reactant), the amount of catalyst was varied in the
range 0.02–0.15 g. Catalysts were mixed with glass bead when
loading and the catalyst bed L/D ratio is about 5. The reaction prod-
ucts were analyzed online on a gas chromatograph (Agilent model
Monometallic Pd and bimetallic Pd-Fe catalysts were prepared
by incipient wetness impregnation and co-impregnation, respec-
tively, using aqueous solutions of the corresponding metal precur-
sors: Pd(NO3)2ꢀ6H2O (98%, Alfa Aesar) and Fe(NO3)3ꢀ9H2O (98%
Sigma-Aldrich). The two supports investigated were SiO2 (PPG Sil-
ica Hi-Sil-915) and c-Al2O3 (99.9%, Alfa Aesar). Prior to impregna-
tion, the supports were dried overnight at 120 °C. The Pd loading
was kept constant at 1.0 wt% in all preparations, while the Fe load-
ing on the bimetallic Pd-Fe catalysts was 0.5 wt%. After impregna-
tion, the catalysts were first dried overnight at room temperature
and then placed in an oven at 120 °C for 12 h. The oven-dried cat-
alysts were finally calcined for 4 h at 500 °C, with a linear heating
ramp of 10 °C/min, under 100 ml/min flow of pure air.
The supported Pd and Pd-Fe catalyst samples were reduced in a
flow of hydrogen (30 ml/min) at 250 °C and then used in character-
ization and catalytic activity measurements. For the catalytic activ-
ity in gas phase, the catalyst powders were pressed (1500 psi),
crushed, and sieved through 40–60 mesh before testing.
2.2. Characterization of the catalysts
6890) using an HP-5 capillary column (30 m/0.25 mm/0.25
and a FID detector.
lm)
Several techniques were employed to characterize the proper-
ties and structure of the catalysts. They include N2 physisorption,
X-ray diffraction (XRD), X-ray photoelectron microscopy (XPS)
and high-resolution transmission electron microscopy (HRTEM).
The BET surface area (Sg) was measured by conventional N2
physisorption on a Micromeritics ASAP 2010 unit, after evacuation
at 350 °C for 3 h. X-ray diffraction studies were conducted in a Pan-
The product yield and selectivity for each product were calcu-
lated as follows:
moles of product
moles of furfural fed
Yield ð%Þ ¼
ꢁ 100ð1Þ
ð1Þ
ð2Þ
moles of product
ꢁ 100
moles of furfural consumed
Selecti
v
ity ð%Þ ¼
alytical X’PERT PRO MPD diffractometer with Cu K
a radiation
(k = 1.5406 Å), operated at 45 kV and 40 mA. The data were
recorded over 2h ranges of 30–50° with a step size of 0.026°.
X-ray photoelectron spectroscopy (XPS) analysis was conducted
on a Physical Electronics PHI 5800 ESCA system with standard non-
monochromatic Al X-rays (1486.6 eV) operated at 250 W and
15 kV in a chamber pumped down to a pressure of approximately
1.0 ꢁ 10ꢂ8 Torr. A sealed transfer cell was used to transport the ex
2.3.2. Liquid phase
The liquid phase conversion of furfural was investigated in a
50 ml Parr Reactor using decahydronaphthalene (decalin, cis
+ trans mixture, 98% purity, Sigma-Aldrich) as a solvent. For the
catalytic reaction runs, 20 ml of decalin, 0.5 ml of furfural and

32
N. Pino et al. / Journal of Catalysis 350 (2017) 30–40
20 mg of catalysts (1%Pd/SiO2, 1%Pd-0.5%Fe/SiO2 and 1%Pd-0.5%Fe/
-Al2O3) were added into the reactor vessel. The liquid phase
3. Results and discussion
c
experiments were performed under H2 at 4 MPa and 250 °C. To
minimize the effect of catalyst deactivation, the reaction time
was kept at 1 h for all the runs with a stirring speed of 400 rpm.
At the end of each run, the products were separated from the
catalyst by centrifugation and filtration. The liquid products were
identified and quantified by gas chromatography. A Shimadzu
3.1. Catalyst characterization
3.1.1. BET surface area
The BET surface areas were measured by N2 physisorption. In
line with the surface areas reported by the manufacturers, the
silica support exhibited a higher surface area (130 m2/g) than
c
-alumina (90 m2/g). No significant drops in surface area were
QP2010 GC–MS equipped with
a
mid-polarity (Phenomenex
m nominal,
ZB-1701) capillary column, 60.0 m ꢁ 0.25 mm ꢁ 0.25
l
observed after impregnation of the metals and thermal treatment.
was used for product identification, while a GC-FID Agilent 7890B
was used for quantitative analysis. In all the GC-FID analyses,
phenol was used as internal standard to help close the mass
balances. The product yield and selectivity for each product were
calculated using Eq. (1) and Eq. (2), respectively.
3.1.2. X-ray diffraction
XRD patterns of the mono and bimetallic samples were
obtained on pre-reduced samples. Fig. 2 compares the diffrac-
tograms of Pd-Fe on the two supports to those of Pd/SiO2. The char-
acteristic diffraction peaks at 2h = 40.1° and 46.8° corresponding to
Pd(111) and (200) crystal faces are clearly observed in the XRD
pattern of Pd/SiO2. Remarkably, these peaks do not appear in the
XRD pattern of the Pd-Fe/SiO2 catalyst, but rather two new peaks
are observed at around 41.2 and 47.2°, which can be ascribed to
the formation of a Pd-Fe alloy when the SiO2 support is used. Since
Fe atoms are smaller than Pd atoms, when incorporated in the Pd
FCC structure, they cause a shift of the 2h diffraction angles to lar-
ger values, as observed here for the Pd-Fe/SiO2 catalyst and in pre-
vious studies on analogous Pd-Fe bimetallic catalysts [33,34]. This
is also in agreement with our DFT calculations that the lattice con-
stants for the Pd-Fe alloys Pd3Fe1 (3.88 Å) and Pd2Fe2 (3.83 Å) are
smaller than the pure Pd (3.94 Å). There is a linear decrease in
the unit cell edge with the incorporation of iron. By contrast, no
evidence of crystalline Pd or Pd-Fe alloys was seen in the XRD pat-
2.4. Density functional theory (DFT) calculations
All DFT calculations were performed with the Vienna ab initio
simulation package (VASP) [23]. The GGA-PBE functional [24],
all-electron plane-wave basis sets with an energy cutoff of
400 eV, and a projector augmented wave (PAW) [25] method were
adopted. A Brillouin-zone of p(4 ꢁ 4) lateral supercell was sampled
by 3 ꢁ 3 ꢁ 1 k-points using the Monkhorst–Pack scheme. The con-
vergence threshold was set to 10ꢂ6 eV in total energy and
10ꢂ2 eV/Å in force on each atom.
It was reported that both Pd and Fe in the PdFe/SiO2 catalyst can
be reduced to their metallic state (zero valence) at the temperature
as low as 400 K, and the main Pd-Fe alloy phases are PdFe and Pd3-
Fe [26–29]. Thus the two Pd3-Fe1 and Pd2-Fe2 crystal structures,
shown in Fig. 1 along with those of pure Pd, were used to model
the Pd-Fe alloy herein. Their structures were optimized and the lat-
tice constants were calculated with a = b = c = 3.94 Å for the Pd,
tern of the Pd-Fe/
c -Al2O3 catalyst. Only the peaks associated with
c-Al2O3 [35] were detectable in this sample. This could be due to a
higher dispersion of Pd on this type of support.
In the Pd-Fe/SiO2 catalyst, the diffraction peaks became much
weaker compared to those in the pattern of the monometallic Pd
catalyst, which indicate the better dispersion of Pd particles with
the introduction of Fe by the formation of Pd-Fe alloys and show
the advantages of a bimetallic catalyst over a monometallic cata-
lyst in the dispersion of the metal sites.
a = b = c = 3.88 Å for the Pd3-Fe1 alloy and a = b = 3.83 Å
&
c = 3.78 Å for the Pd2-Fe2 alloy, which were close to the lattice con-
stants measured by experiments [30]. The closed packed 4 ꢁ 4 Pd
(111), Pd3-Fe1(111) and Pd2-Fe2(111) surfaces were modeled by
a four-layer slab with the bottom two layers fixed in their bulk
positions while the top layers were allowed to relax. The two suc-
cessive slabs were separated by a 15 Å vacuum region to ensure
that the adsorbate (e.g., furfural, furfuryl alcohol) and the subse-
quent slab would not interact.
3.1.3. X-ray photoelectron spectroscopy (XPS)
XPS was used to verify the formation of a binary alloy in the
bimetallic catalyst. For this purpose, binding energies for the Pd-
Fe catalyst were compared to those of pure Pd. For instance, the
XPS spectra for the ex situ reduced (transferred avoiding exposure
to air) Pd and Pd-Fe catalysts are shown in Fig. 3.
The adsorption energy (Eads) in this work is defined as Eads
=
Eadsorbate/surf ꢂ Esurf ꢂ Eadsorbate, where Eadsorbate/surf, Esurf, and Eadsorbate
are the total energy of the adsorbate on the surface, the clean
surface, and the gas-phase adsorbate in vacuum.
The core-level shifts of Pd3d in the bulk Pd3Fe1 and Pd2Fe2 crys-
tals compared to Pd were calculated in reference to the Fermi
energy using the final state approximation [31,32].
The binding energies observed on the Pd/SiO2 catalyst of 335.4
and 340.8 eV can be definitely assigned to Pd 3d5/2 and 3d3/2 of
metallic palladium [36–38]. In comparison, a clear shift of 0.4 eV
Fig. 1. Bulk monometallic Pd and bimetallic Pd-Fe alloy structures. Blue and purple spheres represent Pd and Fe, respectively.

N. Pino et al. / Journal of Catalysis 350 (2017) 30–40
33
to higher binding energies is reproducibly observed for the Pd-Fe/
SiO2 catalyst. This shift has been previously detected when Pd-Fe
alloys are formed. For example, Tang et al. [34] observed a positive
binding energy shift in Pd-Fe/C samples compared to a Pd/C sam-
ple. Similarly, Felicissimo et al. [39] observed shifts to higher bind-
ing energies when Fe was deposited on top of Pd. As the amount of
Fe increased, the binding energy for Pd 3d continuously shifted up
to 0.6 eV above the value of pure Pd, while the intensity decreased,
as expected. The Fe 2p spectra showed an inverse trend, with a
continuous shift to lower binding energies. The same trend has
been reported by other authors [40,41], who have consistently
observed a positive shift with respect to the pure Pd when the
Pd-Fe alloy is formed. Therefore, based on previous observations,
it is reasonable to conclude that the observed binding energy shift
for Pd-Fe/SiO2 (Fig. 3) can be taken as an evidence of alloy
formation.
While we can take the binding energy shift as an indication of
Pd-Fe alloy formation, the fundamental cause for this shift may
be a matter of debate. Some authors have concluded that the con-
comitant shifts in binding energies, that is Pd up and Fe down,
reflect an electron transfer from Pd to Fe, which would be the
opposite to that predicted by simple analysis of electronegativities
of the two metals (Pd:2.2 and Fe:1.83).
Felicissimo et al. [39] have cautioned that binding energy shifts
may not only contain contributions of initial state effects, such as
electron transfer, but also orbital rehybridization and final state
effects. This is further confirmed by our DFT calculations of the
Pd3d core-level shift in the monometallic Pd and bimetallic Pd-Fe
alloy bulk structures (Table 1), which clearly show that alloying
Pd with Fe leads to the Pd3d core-level shift of 0.35 eV in the Pd3-
Fe1 alloy and 0.52 eV in the Pd2Fe2 eV.
Contrary to the XPS spectrum of the Pd-Fe/SiO2 catalyst that
clearly shows the presence of a Pd-Fe alloy, the Pd-Fe/c-Al2O3 spec-
Fig. 2. XRD patterns of the samples after reduction. The (111) and (200) peaks of
the fcc crystal structure (Pd or Pd-Fe alloy) are indicated.
trum shows no shift in binding energy compared to that of Pd/SiO2,
which would indicate that Pd on alumina is not affected by the
presence of Fe. That is, while formation of Pd-Fe alloys is favorable
on the SiO2 support it is much less so on c-Al2O3. In a study of
bimetallic Rh-Au catalysts supported on silica or alumina, Nunez
and Rouco [42] observed similar differences in the extent of
metal-metal interaction between the two supports, as those
reported here. Specifically, they observed that the hydrogenolysis
activity and the reducibility (TPR) of the bimetallic catalyst sup-
ported on c-Al2O3 could be considered as a simple sum of the inde-
pendent monometallic catalysts. By contrast, both the activity and
reducibility (TPR) of the bimetallic catalysts supported on SiO2
clearly gave evidence of the formation of a bimetallic cluster,
showing a significant drop in activity and a single reduction peak
in TPR. That is, a weak interacting support such as SiO2 allows high
enough mobility to the metal precursors to intermingle efficiently
and form a metallic alloy upon reduction. Indeed, this conclusion is
in good agreement with the differences in the product distribution
described in the next section.
TEM analysis was conducted on pre-reduced samples to obtain
information on the size and microstructure of the metal particles.
Fig. 4 shows the micrographs and corresponding particle size dis-
tribution histograms of Pd/SiO2, Pd-Fe/SiO2 and Pd-Fe/
samples. It is interesting to note the effect of the support on the
metal dispersion. With the Pd-Fe/ -Al2O3 catalyst, the distribution
c-Al2O3
c
of metal particles is more homogeneous, with an average size
smaller than on the other samples. It appears that the metal inter-
action with alumina is higher than on silica, which is consistent
with XRD results and the discussion above. Specifically, the aver-
age particle sizes, estimated by random measurement of 50
nanoparticles from the corresponding TEM images, were
6.9 1.7 nm, 4.1 0.8 nm, 2.5 0.8 nm for Pd/SiO2, Pd-Fe/SiO2,
Fig. 3. X-ray photoelectron spectroscopy (XPS) of Pd3d for Pd and Pd-Fe catalysts
with different supports.
Pd-Fe/cAl2O3, respectively. Comparing Pd/SiO2 and Pd-Fe/SiO2

34
N. Pino et al. / Journal of Catalysis 350 (2017) 30–40
Table 1
Comparison of XPS measured Pd3d binding energies in Pd/SiO2 and Pd-Fe/SiO2 and the DFT calculated core-level binding energy of Pd3d in Pd (fcc) and Pd-Fe (fct) bulk structures.
XPS
DFT
Pd/SiO2
335.4
Pd-Fe/SiO2
335.8 eV
Shift
+0.4 eV
Pd (fcc)
–
Pd3Fe1
–
Pd2Fe2
–
Pd3Fe1
+0.35 eV
Pd2Fe2
+0.52 eV
Fig. 4. TEM images of (a) Pd/SiO2, (b) Pd-Fe/SiO2, and (c) Pd-Fe/c-Al2O3 catalysts with the corresponding particle size distribution histograms.
catalysts, it can be observed that the Pd-Fe alloy nanoparticles on
the silica support are noticeably smaller and more uniform in size
than those on the monometallic Pd/SiO2 sample. The introduction
of Fe reduced the aggregation of Pd particles significantly during
the preparation procedure, which resulted in a better dispersion
and smaller sizes of the metal particles. It seems that the interac-
tion between the two metals has a stabilizing effect that results
in inhibition of metal sintering.
ples. They give further evidence that on the silica support, Pd-Fe
alloy particles are obtained, but on the alumina-supported
bimetallic catalyst, the two metals are segregated.
3.2. Catalytic activity
3.2.1. Vapor phase conversion of furfural over Pd/SiO2 and Pd-Fe/SiO2
The product distribution obtained over the Pd/SiO2 catalyst at
250 °C is shown in Fig. 5. In agreement with our previous study
of furfural conversion over pure Pd catalysts [43], the main product
Additional HRTEM and STEM-EDX data are included in the Sup-
plemental Information for the Pd-Fe/SiO2 and Pd-Fe/cAl2O3 sam-

N. Pino et al. / Journal of Catalysis 350 (2017) 30–40
35
Fig. 5. Product distribution from the reaction of furfural over a monometallic 1%Pd/
SiO2 catalyst, at 250 °C, H2/Feed molar ratio = 25, pressure = 1 atm. FOL = furfuryl
alcohol, THF = tetrahydrofuran, HFOL = tetrahydrofurfuryl alcohol.
Fig. 6. Product distribution from the reaction of furfural over a bimetallic 1%Pd-
0.5%Fe/SiO2 catalyst, at 250 °C, H2/Feed molar ratio = 25, pressure = 1 atm.
FOL = furfuryl alcohol, MF = 2-methylfruan, HFOL = tetrahydrofurfuryl alcohol.
from the reaction of furfural on 1%Pd/SiO2 at 250 °C is furan, from
direct decarbonylation [44]. At the same time, hydrogenation of
furfural to furfuryl alcohol was also observed, but to a much lesser
extent than decarbonylation (ꢃ10% vs. 60% yield at W/F = 0.1 h). In
fact, furfuryl alcohol is the dominant product over Cu, on which the
metal-carbon bond strength is very weak [39–46]. The secondary
ring hydrogenation of the primary products (furan and furfuryl
alcohol) to tetrahydrofuran and tetrahydrofurfuryl alcohol (HFOL),
respectively, was only observed at higher W/F.
It must be noted that, regardless of the conversion level, no
2-methylfuran was formed. That is, pure Pd does not exhibit a high
catalytic activity for CAO bond cleavage, but the high selectivity
toward furan, accompanied by an equimolar production of carbon
monoxide, indicates that Pd is highly selective for decarbonylation,
as previously shown [39,40].
As shown in Fig. 6, interesting differences in product distribu-
tion were observed when Fe was incorporated into the Pd/SiO2
catalyst. A remarkable change in product selectivity is observed
over 1%Pd-0.5%Fe/SiO2 catalyst compared to the pure Pd catalyst.
It is clear that the main products from the reaction of furfural over
1%Pd-0.5%Fe/SiO2 catalyst are furfuryl alcohol and, notably,
2-methylfuran, which is not observed over 1%Pd/SiO2.
From the evolution of product distribution as a function of W/F,
it can be seen that 2-methylfuran is in fact a secondary product
derived from the hydrogenolysis of the CAO bond in furfuryl
alcohol, rather than directly from furfural. That is, the yield of fur-
furyl alcohol is dominant at low W/F and increases with increasing
W/F, reaching a maximum and then decreases. Simultaneously, the
production of 2-methyl furan starts with a very low slope at
W/F = 0, but gradually increases with W/F. The secondary product
nature of 2-methyl furan is further demonstrated in Table 2, which
summarizes the conversion and product distribution obtained
from the reaction of furfural (FAL) and furfuryl alcohol (FOL) over
the Pd/SiO2 and Pd-Fe/SiO2 catalysts. A much higher yield of
2-methylfuran was observed when furfuryl alcohol was used as a
feed over Pd-Fe/SiO2 instead of furfural (compare entries 2 vs. 6).
1%Pd/SiO2 and 1%Pd-0.5%Fe/SiO2 catalysts, respectively. The yield
of furan derived from decarbonylation is dramatically decreased
on Pd-Fe/SiO2 (13.5%) compared to 54% on Pd/SiO2. Furthermore,
contrary to monometallic Pd catalyst that does not form any
2-methylfuran,
the
1%Pd-0.5%Fe/SiO2
catalyst
produced
2-methylfuran in significant amounts, particularly at high W/F. To
show that the enhanced rate of 2-methylfuran production on the
Pd-Fe bimetallic catalyst prepared by co-impregnation technique
is due to a direct interaction between Pd and Fe, a physical mixture
of monometallic 1%Pd/SiO2 and 1%Fe/SiO2 (entry 3) was tested and
compared in the table. The results show that the product distribu-
tions and furfural conversions from 1%Pd/SiO2 and a physical mix-
ture of 1%Pd/SiO2 and 1%Fe/SiO2 are essentially the same. That is,
furan from decarbonylation of furfural is the main product while
2-methylfuran is not formed. It should be noted that under the
present conditions, the monometallic Fe/SiO2 exhibits no activity
for the reaction of furfural and furfuryl alcohol (entries 4 and 8).
It is clear that the presence of Fe on bimetallic catalyst has both
a promoting effect and a suppressing effect on a product yield and
selectivity. The yield of furfuryl alcohol from hydrogenation of fur-
fural and 2-methylfuran from hydrogenolysis of furfuryl alcohol is
promoted while the yield of furan from decarbonylation is sup-
pressed. This will be further illustrated by DFT calculations below.
The similar behavior was also observed when having furfuryl
alcohol as feed over Pd/SiO2 and bimetallic Pd-Fe/SiO2 catalysts.
The most important result on Pd-Fe bimetallic catalyst is the for-
mation of 2-methylfuran which can be derived from the CAO
hydrogenolysis of furfuryl alcohol. As shown in Table 2 (entry 5),
when feeding furfuryl alcohol there is no significant amount of
2-methyfuran formed on Pd/SiO2 (yield < 1%). It is also interesting
to note that the major product from furfuryl alcohol on 1%Pd/SiO2
is a saturation of furanyl ring to tetrahydro furfuryl alcohol. In con-
trast, more than 83% of furfuryl alcohol is converted to 2-
methylfuran with 1%Pd-0.5%Fe/SiO2 catalyst. The results from this
work have demonstrated that bimetallic Pd-Fe catalyst is very
selective toward CAO bond cleavage reactions instead of CAC
bonds which mainly occur on monometallic Pd catalyst.
Also,
a comparison of product distributions at the same
furfural conversion and temperature is made in entries 1 and 2 for

36
N. Pino et al. / Journal of Catalysis 350 (2017) 30–40
Table 2
Conversion and yield of products from the reaction of furfural and furfuryl alcohol over 1%Pd/SiO2 and 1%Pd-0.5%Fe/SiO2 catalysts in vapor phase.
Entry
Catalysts
Feed
Conv. (%)
Yield (%)
Furan
Selectivity (%)
DeCOb
THF
FOL
MF
HFOL
HyCOc
1
2
3
4
5
6
7
8
Pd/SiO2
FAL
FAL
FAL
FAL
FOL
FOL
FOL
FOL
82.1
81.4
80.3
0.0
19.2
92.7
19.6
0.0
54
13.5
53
0.0
2.0
2.8
1.6
0
13.4
0.0
11.9
0.0
2.8
0.4
2.0
0
11
19.0
10.6
0.0
–
–
–
–
0
3.8
10.2
4.8
0.0
13.7
6.2
15.1
0
82.1
16.6
80.8
–
–
–
18.0
83.3
19.3
–
–
–
Pd-Fe/SiO2
aPd/SiO2 + Fe/SiO2
Fe/SiO2
38.6
0.1
0.0
0.7
83.3
1.0
0
Pd/SiO2
Pd-Fe/SiO2
aPd/SiO2 + Fe/SiO2
Fe/SiO2
–
–
–
–
Reaction conditions: Temp. = 250 °C, H2/Feed molar ratio = 25, pressure = 1 atm, W/F = 0.075 h.
a
Physical mixture of two catalysts.
DeCO = decarbonylation (see Scheme 1) DeCO products = Furan + THF.
HyCO = hydrogenation (see Scheme 1) HyCO products = FOL + MF + HFOL.
b
c
3.2.2. Effect of support on the 2-methylfuran selectivity
To explore the effect of the support on the gas phase furfural
conversion, a bimetallic Pd-Fe supported over
ated. The results are shown in Fig. 7.
c-Al2O3 was evalu-
A significant difference with the product distribution from Pd-
Fe/SiO2 is observed. First, the yield of 2-methylfuran was much
lower than over the SiO2-supported bimetallic. In this case, the
CAC scission is the preferred pathway displaying a high yield of
furan, similar to that of pure Pd. As a result, the selectivity to MF
is dramatically lower than on SiO2 and similar to that of Pd/SiO2,
as illustrated in Fig. 8.
As it was expected, with the Pd-Fe/SiO2 catalyst the selectivity
toward 2-methylfuran increases with the furfural conversion,
while with Pd/SiO2 catalyst there was not selectivity since there
was no formation of 2-methylfuran. The product selectivity was
drastically shifted to furan when Pd-Fe nanoparticles were sup-
ported on
the metallic nanoparticles with the
the two-metal segregation as shown in the XPS, XRD and TEM
analysis. In this way, Pd-Fe/ -Al2O3 catalyst behaves a lot more like
c
-Al2O3. This could be due to the higher interaction of
c-Al2O3 support, which caused
c
monometallic Pd/SiO2. Since the physical mixture of Pd/SiO2 and
Fig. 8. 2-Methylfuran selectivity as a function of furfural conversion over supported
Pd-Fe catalysts at 250 °C, H2/Feed molar ratio = 25, pressure = 1 atm.
Fe/SiO2 showed a low yield of 2-methylfuran (Table 2), one could
expect the formation of Pd-Fe bimetallic alloy, which is required
for the CAO hydrogenolysis may not be totally formed on
c-alumina support; therefore, nanoparticles of Pd and Fe behave
individually.
Fig. 8 shows the yield and selectivity to 2-methylfuran (MF)
over the Pd-Fe/SiO2 and Pd-Fe/c-Al2O3 catalysts. It is clear that,
at the same furfural conversion (ꢃ80%) the selectivity to
2-methylfuran over the Pd-Fe/SiO2 catalyst was much higher
(46.3%) than that over Pd-Fe/c-Al2O3 (only 8.5%). The high selectiv-
ity toward 2-methylfuran on the silica supported catalyst is a clear
indication of the Pd-Fe alloy formation, which is not present when
alumina was used as support.
3.2.3. Reaction in liquid phase
The product distribution obtained from furfural conversion in
liquid phase (decalin solvent) was investigated on the same
catalyst series used in the vapor phase. It was found that the main
Fig. 7. Product distribution from the reaction of furfural over a bimetallic 1%Pd-
0.5%Fe/g-Al2O3 catalyst, at 250 °C, H2/Feed molar ratio = 25, pressure = 1 atm.
FOL = furfuryl alcohol, MF = 2-methylfuran, HFOL = tetrahydrofurfuryl alcohol.

N. Pino et al. / Journal of Catalysis 350 (2017) 30–40
37
Scheme 1. Reaction pathways for furfural conversion.
product from the liquid phase reaction of furfural on 1%Pd/SiO2 at
250 °C was furan. Hydrogenation products, such as tetrahydrofu-
ran, obtained from secondary furan hydrogenation, and furfuryl
alcohol, from primary furfural hydrogenation (see Scheme 1) were
also observed, but at lower yields than that of furan. This means
that decarbonylation is the preferred reaction pathway for the con-
version of furfural over pure Pd. Even at high furfural conversion,
no formation of 2-methylfuran was observed over Pd/SiO2. Similar
results have been found by Bhogeswararao et al. [6], who obtained
high furan yield (82%) and selectivity using supported Pd catalysts
in liquid phase (isopropanol solvent). Likewise, Wang et al. [47]
achieved high selectivity to furan (98%) in a reaction of furfural
over a core-shell structural Pd@S-1 catalyst, using butanol as a
solvent.
By contrast, over the 1%Pd-0.5%Fe/SiO2 catalyst, 2-methylfuran
and furfuryl alcohol were the main reaction products (see Table 3
and Scheme 1). In the batch reactor, at increasing reaction
times the yield of furfuryl alcohol decreased as the yield of
2-methylfuran increased. Similar behavior was observed by Scholz
et al. [48], who investigated the liquid phase conversion of furfural
over Pd catalysts. They found that the formation of 2-methylfuran
greatly increased when Pd was supported on Fe2O3 compared to
other supports. The same trend has been reported for other
bimetallic catalysts in the liquid phase, including Cu-Co and
Cu-Fe at varying molar ratios [13,15,49].
Piancatelli rearrangement of furfuryl alcohol, favored by the
presence of water [50–52], which is produced in the CAO
hydrogenolysis of furfuryl alcohol to 2-methylfuran.
The major differences between the results in gas and liquid
phase can be summarized as follows. Over Pd/SiO2 the product dis-
tribution obtained at the same overall conversion was very similar,
regardless of whether the reaction was conducted in vapor or liq-
uid phase. Over Pd-Fe/SiO2, the major difference was the almost
complete disappearance of furan in the products when the reaction
was conducted in the liquid phase. That is, the effect of the alloy,
which causes a great decrease in decarbonylation rate in the vapor
phase is further enhanced in the presence of the solvent, since
decarbonylation requires an ensemble of Pd sites with strong bond
to the adsorbate and the addition of Fe suppresses this activity.
Similar conclusions have been recently derived by Yang et al.
[53] in their study of furfural and methyl-isobutylketone conver-
sion on Pd-FeOx/SiO2 catalysts. Working under solvent-free
hydrodeoxygenation conditions, they found that relative to
monometallic Pd, the Pd-Fe catalyst gave less decarbonylation
products. They explained this behavior in terms of Pd ensemble
dilution on the basis of experimentally demonstrated (EXAFS,
Mossbauer) decrease in Pd-Pd coordination number by addition
of Fe, which inhibits CAC bond breaking by decarbonylation. Our
results further suggest that the solvent may further enhance the
site blockage, thus reducing the production of furan. Interestingly,
the production of MF via CAO bond hydrogenolysis, enhanced by
alloying with Fe, is not greatly affected by the presence of the
solvent.
Table 3 shows that hydrogenation of furan to tetrahydrofuran
also occurred over 1%Pd-0.5%Fe/SiO2, but with a significantly lower
yield than over Pd/SiO2. Interestingly, when c-Al2O3 was used as a
support in the liquid phase, the unbalance carbon greatly
increased, leading to lower yields of 2-methylfuran and furfuryl
alcohol. This loss in carbon yield may be associated with the
3.3. DFT calculations on Pd(111), Pd3Fe1(111) and Pd2Fe2(111)
surfaces
acidity of c-Al2O3, which promotes oligomerization, coke forma-
tion, and consequently, catalyst deactivation. Hydrogenation of
2-methylfuran to 2-methyltetrahydrofuran was also observed on
this catalyst.
Another interesting by-product, observed over Pd-Fe/SiO2 and
Pd-Fe/c-Al2O3 catalysts in liquid phase but not in vapor phase,
DFT calculations were conducted to interpret the changes
observed in product selectivity on Pd-Fe bimetallic catalysts com-
pared to the monometallic Pd catalyst. Two aspects should be
explored; first, the addition of Fe causes a decrease in the CAC
bond cleavage reaction [7], but also it causes an increase in the
HDO reaction, in good agreement with the results reported by
was cyclopentanone. This CAC bond forming product has been
previously observed in furfural reactions and is ascribed to the

38
N. Pino et al. / Journal of Catalysis 350 (2017) 30–40
Table 3
Conversion and yield of products from the reaction of furfural over Pd and Pd-Fe catalysts in vapor and liquid phase.a
Product yield (%)
Pd/SiO2
Vapor
Pd-Fe/SiO2
Vapor
Pd-Fe/
c-Al2O3
Liquid phase
Liquid phase
Vapor
Liquid phase
Conversion (%)
Furan
FOL
THF
MF
HFOL
MTHF
CPON
68
47
10
11
0
0.3
0
0
68
36
7
13
0
0
0
0
82
13
20
0
40
9
84
0.13
21
7
36
0
80
40
21
0
10
9
89
6
7
0
20
0
0
0
0
2
0
0
14
9
Unbalance carbon (%)
0
12
0
18
0
33
a
FOL = furfuryl alcohol, THF = tetrahydrofuran, MF = 2-methylfuran, THFOL = tetrahydrofurfuryl alcohol, MTHF = 2-methyltetrahydrofuran, CPON = cyclopentanone.
Fig. 9. DFT optimized adsorption structures of furfural, furfuryl alcohol and furan on Pd(111), Pd3Fe1(111) and Pd2Fe2(111) surfaces and the corresponding adsorption
energies.
Wang et al. on the conversion of phenolic compounds such as
cresol [54] and guaiacol [55] over bimetallic Pd-Fe catalysts.
To investigate the interaction of furanic compounds with the
monometallic and bimetallic surfaces, we conducted DFT
calculations to obtain adsorption energies of furfural, furfuryl alco-
hol, furan and methylfuran on optimized structures of Pd(111),
Pd3Fe1(111) and Pd2Fe2(111).
The possible adsorption modes of these furanics on different
metal surfaces were shown in the supporting information Fig. S6,
and the most favorable adsorption structures and the correspond-
ing adsorption energies are shown in Fig. 9. On Pd(111), all these
four furanics favorably adsorb via the furanic ring with the adsorp-
tion energy around ꢂ79 to ꢂ89 kJ/mol. The addition of Fe weakens
the adsorption of furanic ring. Therefore, the adsorption energies of
all the four furanics decrease along with the content of Fe. That is,
on Pd3Fe1(111), the adsorption energies of furan, methylfuran and
furfuryl alcohol decrease to ꢂ67 kJ/mol, ꢂ58 kJ/mol and ꢂ64 kJ/-
mol respectively. On Pd2Fe2(111), they further decrease to
ꢂ33 kJ/mol, ꢂ18 kJ/mol and ꢂ42 kJ/mol, respectively. The furanic
ring adsorption on Pd2Fe2(111) is so weak that it is more favorable
for furfuryl alcohol to adsorb via the AOH side group rather than
the furanic ring, as shown in Fig. 9. The weaker binding of the fura-
nic ring on the Pd-Fe alloy surface is probably due to both the elec-
tronic effects and the ensemble effects. The d band center of
Pd3Fe1(111) and Pd2Fe2(111) was calculated to be ꢂ1.66 eV and
ꢂ1.70 eV respectively, which are lower than the monometallic Pd
(111) surface (ꢂ1.46 eV). As the d-band shifts up in energy, the
number of antibonding states above eF (Fermi level) increases,

N. Pino et al. / Journal of Catalysis 350 (2017) 30–40
39
which leads to a stronger bond, as clearly occurring with Pd(111)
surface. Furthermore, according to the literature, lower d band
center leads to weaker binding of species [56].
The weak binding of the furan ring on Pd-Fe alloy surface com-
pared to the monometallic Pd(111) surface leads to the increased
difficulty in ring hydrogenation, which is consistent with the
experimental observation that much less ring hydrogenation
products (THF, MTHF) were observed over Pd-Fe/SiO2 than over
Pd/SiO2.
Unlike furan, methylfuran or furfuryl alcohol, the adsorption of
furfural on Pd3Fe1(111) and Pd2Fe2(111) does not change much
relative to adsorption on Pd(111). This is clearly due to the stron-
ger bond of the carbonyl group to the oxophilic Fe atoms, as
illustrated in Fig. 9. This is further confirmed by our calculations
of the adsorption of formaldehyde (H2C@O) on Pd(111),
Pd3Fe1(111) and Pd2Fe2(111) via only the carbonyl group. Indeed,
they showed stronger adsorption on Pd3Fe1(111) (ꢂ92 kJ/mol) and
Pd2Fe2(111) (ꢂ85 kJ/mol) rather than on Pd(111) (ꢂ55 kJ/mol).
Similarly, the stronger adsorption of the side group AOH of furfuryl
alcohol on Pd-Fe alloy surface than the monometallic Pd(111)
surface was shown by the adsorption energies of CH3OH on
Pd3Fe1(111) (ꢂ46 kJ/mol), Pd2Fe2(111) (ꢂ42 kJ/mol) and Pd(111)
(ꢂ26 kJ/mol), as shown in Fig. S7.
uid phase the maximum methylfuran/furan product ratio is
obtained. However, carbon losses and catalyst deactivation are
much higher in the liquid phase compared to gas phase operation.
DFT calculations show that furfural, furfuryl alcohol, furan and
methylfuran favorably adsorb via the furanic ring on Pd(111),
while the addition of Fe clearly weakens these adsorption energies.
The stronger binding of the carbonyl group (strongly bound to the
oxophilic Fe) and AOH group on Pd-Fe alloy surface promotes the
furfural hydrogenation and furfuryl alcohol hydrogenolysis toward
2-methylfuran.
Acknowledgments
This work was supported by the Department of Energy, with
funding of the experimental part under grant DE-SC0004600
(DoE-EPSCOR) and theoretical part under grant DE-EE0006287
(BETO/CHASE program). NP gratefully acknowledges the Universi-
dad de Antioquia (Colombia) for a scholarship and COLCIENCIAS
(Colombia) for financial support (1115-569-33557).
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2017.03.016.
These results clearly explain the great differences in selectivity
between Pd and Pd-Fe alloys, with the former preferring CAC bond
cleavage via decarbonylation and the latter CAO bond cleavage via
hydrogenolysis. The weaker binding of the furanic ring but stron-
ger binding of the carbonyl group and AOH group on Pd-Fe alloy
surface than Pd(111) leads to enhanced HDO activity. At the same
time, the stronger binding of the carbonyl group on Pd-Fe alloy sur-
face facilitates hydrogenation of furfural to furfuryl alcohol and
further deoxygenation to methylfuran. By contrast, the decarbony-
lation of furfural requires the adsorption of the furanic ring after
the CAC bond breaking, which is expected to be difficult due to
the unfavorable binding of furanic ring on the Pd-Fe alloy surface,
as shown above. This is in good agreement with the experiments
over Pd-Fe alloys that show lower selectivity to furan with
enhanced selectivity to furfuryl alcohol and methylfuran. The
opposite is true on pure Pd or bimetallic catalysts in which the
alloying of Pd and Fe is prevented by the interaction with the sup-
port (e.g. alumina).
References
[1] R. Mariscal, P. Maireles-Torres, M. Ojeda, I. Sádaba, M. López Granados, Energy
Environ. Sci. 9 (2016) 1144–1189.
[2] D.E. Resasco, S. Sitthisa, J. Faria, T. Prasomsri, M.P. Ruiz, in: D. Kubicka, I.
Kubicková (Eds.), ‘‘Heterogeneous Catalysis in Biomass to Chemicals and Fuels”
Research Signpost, 2012, pp. 155–188.
[3] A. Mandalika, T. Runge, Green Chem. 14 (2012).
[4] T.N. Pham, D. Shi, D.E. Resasco, Appl. Catal. B Environ. 145 (2014) 10–23.
[5] Y. Roman-Leshkov, C.J. Barrett, Z.Y. Liu, J.A. Dumesic, Nat. Lett. 447 (2007) 982.
[6] S. Bhogeswararao, D. Srinivas, J. Catal. 327 (2015) 65–77.
[7] S. Sitthisa, W. An, D.E. Resasco, J. Catal. 284 (2011) 90–101.
[8] H.-Y. Zheng, Y.-L. Zhu, B.-T. Teng, Z.-Q. Bai, C.-H. Zhang, H.-W. Xiang, Y.-W. Li, J.
Mol. Catal. A Chem. 246 (2006) 18–23.
[9] B.-M. Nagaraja, A.-H. Padmasri, B. Raju, K.-S. Rama, J. Mol. Catal. A: Chem. 265
(2007) 90–97.
[10] M.-M. Villaverde, N.-M. Bertero, T.-F. Garetto, A.-J. Marchi, Catal. Today 213
(2013) 87–92.
[11] C.-P. Jiménez, J.-A. Cecilia, I. Márquez, R. Moreno, J. Santamaría, J. Mérida, P.
Maireles, Catal. Today 279 (2017) 327–338.
[12] L. Liu, H. Lou, M. Chen, Int. J. Hydrogen Energy (2016).
[13] H. Sheng, R.-F. Lobo, ChemCatChem. 8 (2016) 1–8.
[14] K. Yan, A. Chen, Fuel 115 (2014) 101–108.
[15] X. Chang, A.-F. Liu, B. Cai, J.-Y. Luo, H. Pan, Y.-B. Huang, ChemSusChem. 9
(2016) 1–9.
4. Conclusion
[16] K. Xiong, W. Wan, J. Chen. Surf. Sci. 652 (2016) 91–97.
[17] W.T. Yu, M.D. Porosoff, J.G. Chen, Chem. Rev. 112 (2012) 5780.
[18] S. Sitthisa, T. Pham, T. Prasomsri, T. Sooknoi, R. Mallinson, D.E. Resasco, J. Catal.
280 (2011) 17–27.
[19] W. Yu, K. Xiong, N. Ji, M.D. Porosoff, J.-G. Chen, J. Catal. 317 (2014) 253–262.
[20] M. Hansen, Constitution of Binary Alloys, McGraw-Hill, New York, 1958, p.
696.
[21] G. Willimas, J.W. Loram, J. Phys. Chem. Solids 30 (1969) 1827–1833.
[22] R. Madon, M. Boudart, Ind. Eng. Chem. Fundament. 21 (1982) 438–447.
[23] G. Kresse, J. Hafner, Phys. Rev. B 47 (1993) 558–561.
[24] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865–3868.
[25] G. Kresse, D. Joubert, Phys. Rev. B 59 (1999) 1758–1775.
[26] G. Ghosh, C. Kantner, G.B. Olson, J. Phase Equilib. 20 (1998) 295–310.
[27] G.L.W. Hart, S. Curtarolo, T.B. Massalski, O. Levy, Phys. Rev. X 3 (2013), 041035-
1–041035-33.
[28] R.V. Chepulskii, S.V. Barabesh, A. Zunger, Phys. Rev. B 85 (2012), 144201-1–
144201-23.
[29] Y. Chen, T. Atago, T. Mohri, J. Phys.: Condens. Matter 14 (2002) 1903–1913.
[30] K.H.J. Buschow, P.G. van Engen, R. Jongebreur, J. Magn. Magn. Mater. 38 (1983)
1–22.
[31] W. Olovsson, T. Marten, E. Holmstrom, B. Johansson, I. Abrikosov, J. Electron.
Spectrosc. Relat. Phenom. 178 (2010) 88–99.
[32] Q.T. Trinh, K.F. Tan, A. Borgna, M. Saeys, J. Phys. Chem. C 117 (2013) 1684–
1691.
The gas phase conversion of furfural over bimetallic Pd-Fe/SiO2
catalyst gives predominantly 2-methylfuran, while over
monometallic Pd/SiO2 catalyst, the decarbonylation reaction
dominates, yielding furan as the main product. This means that
an alloying metal like Fe with a strong affinity for oxygen enhances
the selectivity toward CAO hydrogenolysis, forming 2-methylfuran
as a final product.
The change of the support affects the formation of Pd-Fe alloy
and consequently the selectivity toward 2-methylfuran. When
using
the selectivity greatly changes from 2-methylfuran to furan. That
is, Pd-Fe/ -Al2O3 behaves much more like monometallic
c-Al2O3 instead of SiO2 as a support of the bimetallic catalyst,
c
a
Pd/SiO2 than a bimetallic Pd-Fe/SiO2. These results suggest that
the intrinsic properties of support have significant influence on
the catalytic performance of Pd-Fe catalysts. The observed
contrasting behavior of the two supports is explained in terms of
a segregation of the metals induced by a higher interaction with
the support, in the case of alumina that does not occur on silica,
as evidenced by XPS, XRD, and TEM characterization.
[33] F. Pinna, M. Selva, M. Signoretto, G. Strukul, F. Boccuzzi, A. Benedetti, P. Canton,
G. Fagherazzi, J. Catal. 150 (1994) 356–367.
[34] Y. Tang, S. Cao, Y. Chen, T. Lu, Y. Zhou, L. Lu, Bao, J. Appl. Surf. Sci. 256 (2010)
4196–4200.
The presence of the solvent enhances the inhibition of the
decarbonylation path via acyl intermediate. As a result, in the liq-

40
N. Pino et al. / Journal of Catalysis 350 (2017) 30–40
[35] S.N. Babu, N. Lingaiah, J.V. Kumar, P.S.S. Prasad, Appl. Catal. A Gen. 367 (2009)
70–76.
[47] C. Wang, L. Wang, J. Zhang, H. Wang, J.-P. Lewis, F.–Sh. Xao, J. Am. Chem. Soc.
138 (2016) 7880–7883.
[36] B.L. Gustafson, P.S. Wehner, Appl. Surf. Sci. 52 (1991) 261–270.
[37] K. McEleney, C.M. Crudden, J.H. Horton, J. Phys. Chem. C 113 (5) (2009) 1901–
1907.
[38] J. Tarabay, W. Al-Maksoud, F. Jaber, C. Pinel, S. Prakash, L. Djakovitch, Appl.
Catal. A Gen. 388 (2010) 124–133.
[48] D. Scholz, Ch. Aellig, I. Hermans, ChemPubSoc. 7 (2014) 268–275.
[49] S. Srivastava, G.-C. Jadeja, J. Parikh, RSC Adv. 6 (2016) 1649–1658.
[50] G. Piancatelli, A. Scettri, S. Barbadoro, Tetrahedron Lett. 17 (1976) 3555–3558.
[51] K. Fulajtarova, T. Sotak, M. Hronec, I. Vavra, E. Dobrocka, M. Omastova, Appl.
Catal. A 502 (2015) 78–85.
[39] M.P. Felicissimo, O.N. Martyanov, T. Risse, H.J. Freund, Surf. Sci. 601 (2007)
2105–2116.
[52] T.V. Bui, S. Crossley, D.E. Resasco, in: Cavani et al. (Eds.), Chemicals and Fuels
from Bio-Based Building Blocks, 2016 (Chapter 17).
[40] J.C. Bertolini, J.L. Rousset, P. Miegge, J. Massardier, B. Tardy, Surf. Sci. 287 (288)
(1993) 346.
[53] J.F. Yang, S.S. Li, L.L. Zhang, X.Y. Liu, J.H. Wang, X.L. Pan, N. Li, A.Q. Wang, Y.
Cong, X.D. Wang, T. Zhang, Appl. Catal.
266–277.
B – Environ. 201 (2017)
[41] S.L. Zhang, J.R. Zhang, Phys. Stat. Sol. B 182 (1994) 421.
[42] G.M. Nunez, A.J. Rouco, J. Catal. 111 (1988) 41–49.
[43] S. Sitthisa, D.E. Resasco, Catal. Lett. 141 (2011) 784–791.
[44] R.D. Srivastava, A.K. Guha, J. Catal. 91 (1985) 254–262.
[45] S. Sitthisa, T. Sooknoi, Y.G. Ma, P.B. Balbuena, D.E. Resasco, J. Catal. 277 (2011)
1–13.
[54] Y. Hong, H. Zhang, J. Sun, A.M. Karim, A. Hensley, M. Gu, M.H. Engelhard, J.S.
McEwen, Y. Wang, ACS Catal. 4 (2014) 3335–3345.
[55] J. Sun, A.M. Karim, H. Zhang, L. Kovarik, X.S. Li, A. Hensley, J.S. McEwen, Y.
Wang, J. Catal. 306 (2013) 47–57.
[56] J.-K. Norskov, F.-A. Pedersen, F. Studt, T. Bligaard, PNAS 108 (2011) 937–943.
[46] J.A. Cecilia, D. Duran-Martin, R. Moreno-Tost, Ramon, J. Santamaria-Gonzalez,
J. Catal. 336 (2016) 107–115.