Geometric and electronic effects of bimetallic Ni–Re catalysts for selective deoxygenation of m-cresol to toluene

Article abstract of DOI:10.1016/j.jcat.2017.01.001

Ni–Re/SiO₂ bimetallic catalysts were prepared using a co-impregnation method and tested in vapor phase hydrodeoxygenation of m-cresol at 300?°C and 1?atm H₂. In contrast to the use of unselective monometallic Ni/SiO₂ for catalyzing deoxygenation, hydrogenation, and C–C hydrogenolysis reactions, bimetallic 5%Ni–2.5%Re/SiO₂ improved the intrinsic reaction rate of the hydrodeoxygenation reaction by a factor of 6, with the turnover frequency for selective deoxygenation to toluene increased by four times, while that for C–C hydrogenolysis to methane was reduced by one-half. Characterization results from X-ray diffraction, Raman, transmission electron microscopy, X-ray photoelectron spectroscopy, infrared spectroscopy of CO adsorption, H₂ temperature-programmed reduction, and CO chemisorption indicate that adding Re increased Ni dispersion and resulted in Ni–Re surface alloy formation after reduction. The presence of Re in the surface alloy breaks the continuous Ni surface into smaller ensembles (geometric effect) and reduces the d-band electron density of Ni (electronic effect). Results from density functional theory calculations indicate that the Ni–Re neighboring site is the active site for breaking the C–O bond by adsorbing the O atom on Re and the phenyl ring on the neighboring Ni atoms, which facilitates deoxygenation to toluene. The reduced Ni ensemble size inhibits the hydrogenolysis of the C–C bond by destabilizing the transition state, whereas the reduction of the electronic density in d states of Ni weakens the adsorption of the phenyl ring, and both contribute to the greatly reduced methane production from successive C–C hydrogenolysis.

Full text of DOI:10.1016/j.jcat.2017.01.001

Journal of Catalysis 349 (2017) 84–97
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Geometric and electronic effects of bimetallic Ni–Re catalysts for
selective deoxygenation of m-cresol to toluene
a,
Feifei Yang a, Dan Liu a, Hua Wang a, Xiao Liu a, Jinyu Han a, Qingfeng Ge a,b, , Xinli Zhu
⇑
⇑
a Collaborative Innovation Center of Chemical Science and Engineering, School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China
b Department of Chemistry and Biochemistry, Southern Illinois University, Carbondale, IL 62901, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Ni–Re/SiO2 bimetallic catalysts were prepared using a co-impregnation method and tested in vapor phase
hydrodeoxygenation of m-cresol at 300 °C and 1 atm H2. In contrast to the use of unselective monometal-
lic Ni/SiO2 for catalyzing deoxygenation, hydrogenation, and CAC hydrogenolysis reactions, bimetallic 5%
Ni–2.5%Re/SiO2 improved the intrinsic reaction rate of the hydrodeoxygenation reaction by a factor of 6,
with the turnover frequency for selective deoxygenation to toluene increased by four times, while that for
CAC hydrogenolysis to methane was reduced by one-half. Characterization results from X-ray diffraction,
Raman, transmission electron microscopy, X-ray photoelectron spectroscopy, infrared spectroscopy of CO
adsorption, H2 temperature-programmed reduction, and CO chemisorption indicate that adding Re
increased Ni dispersion and resulted in Ni–Re surface alloy formation after reduction. The presence of
Re in the surface alloy breaks the continuous Ni surface into smaller ensembles (geometric effect) and
reduces the d-band electron density of Ni (electronic effect). Results from density functional theory cal-
culations indicate that the Ni–Re neighboring site is the active site for breaking the CAO bond by adsorb-
ing the O atom on Re and the phenyl ring on the neighboring Ni atoms, which facilitates deoxygenation to
toluene. The reduced Ni ensemble size inhibits the hydrogenolysis of the CAC bond by destabilizing the
transition state, whereas the reduction of the electronic density in d states of Ni weakens the adsorption
of the phenyl ring, and both contribute to the greatly reduced methane production from successive CAC
hydrogenolysis.
Received 17 October 2016
Revised 21 December 2016
Accepted 4 January 2017
Keywords:
Ni–Re/SiO2
Bimetallic catalyst
m-Cresol
Hydrodeoxygenation
Deoxygenation
Hydrogenation
CAC hydrogenolysis
Ó 2017 Elsevier Inc. All rights reserved.
1. Introduction
genation of the phenyl ring, deoxygenation, CAC hydrogenolysis,
and transalkylation, may take place at the same time, depending
The lignin fraction of lignocellulosic biomass has relatively low
oxygen content in comparison to cellulose and hemicellulose [1,2].
Depolymerization (for example, fast pyrolysis) of lignin results in
phenolic compounds, such as phenol, cresol, anisole, guaiacol,
and syringol [3]. These are important building blocks for fuels
and chemicals after complete or partial removal of the oxygen con-
tent. Catalytic hydrodeoxygenation (HDO) is a key step in oxygen
removal [2].
Many studies on HDO of phenolics have been performed using
various catalysts, including hydrodesulfurization (HDS) catalysts
(CoMoS and NiMoS) [4,5], hydrogenating catalysts (Ni and Pt)
[6,7], zeolites [8], metal oxides [9], and metal phosphides [10],
and operated under very different conditions to achieve a high
degree of oxygen removal and to explore the reaction mechanism.
During HDO of phenolics, a number of reactions, such as hydro-
on the catalysts and operating conditions. These reactions, occur-
ring in an uncontrolled manner, lead to a mixture of products with
high hydrogen consumption and low liquid products yield.
Recently, more attention has been paid to selective deoxygenation
of phenolics to aromatics with reduced hydrogen consumption
under mild conditions (low pressure and intermediate tempera-
ture) [11]. These mild conditions are suitable for processing pheno-
lics produced from fast pyrolysis before their condensation to
unstable liquids at low temperatures. Reducible metal oxides, such
as FeOx and MoOx, appear to be highly selective for deoxygenation
of phenolics to aromatics at relatively high temperatures [9,12]. On
the other hand, typical hydrogenating catalysts, such as Pt, Pd, Ni,
and Ru, are more active and could be operated at lower tempera-
tures. However, they tend to hydrogenate the phenyl ring rather
than to deoxygenate it [6,7,13–15]. In addition, Ni and Ru have
been reported to be highly active for catalyzing CAC hydrogenoly-
sis to form low-value short-chain hydrocarbons [6,13,14]. Combin-
ing a hydrogenating metal and an oxophilic metal (reducible metal
⇑
Corresponding authors.
E-mail addresses: qge@chem.siu.edu (Q. Ge), xinlizhu@tju.edu.cn (X. Zhu).
http://dx.doi.org/10.1016/j.jcat.2017.01.001
0021-9517/Ó 2017 Elsevier Inc. All rights reserved.

F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
85
oxide) forms a bimetallic catalyst, such as Ni–Co [16], Pt–Mo [17],
Pt–Co [18], Pt–Zn [19], Pd–Fe [20], or Ni–Fe [21]. These bimetallic
catalysts are expected to balance the activity and selectivity of
deoxygenation and hydrogenation, and have been explored for
HDO of phenolics. For example, Nie and Resasco reported that
the Ni–Fe catalyst is selective for deoxygenation of m-cresol to
toluene, and proposed that Fe helped the hydrogenation of the tau-
tomerization intermediate from m-cresol to 3-methyl-3,5-
cyclohexadienol, which is then quickly dehydrated to toluene
[21]. It is noted that the Ni–Fe catalyst is less active than the Ni cat-
alyst. Sun et al. reported deoxygenation of guaiacol and m-cresol to
aromatics over a Pd–Fe catalyst and suggested that Fe is the active
site for direct deoxygenation, whereas Pd is the active site for acti-
vating hydrogen [20,22]. The addition of 5% Pd improves the activ-
ity of Fe2O3 by a factor of 2 [22]. Clearly, more active and selective
catalysts for conversion of phenolics to aromatics are desirable.
Although catalysts based on bimetallics show promise in improv-
ing the selectivity, the nature of the active site and the mechanism
for selective deoxygenation are not yet clear.
ment, a catalyst sample of 40 mg was loaded in a U-tube quartz
reactor and pretreated with flowing N2 at 300 °C for 1 h. After
the sample was cooled to room temperature, the gas was shifted
to 5% H2/Ar (25 mL/min). After the signal was stabilized, the sam-
ple was heated linearly to 800 °C at a rate of 10 °C/min. The
amount of hydrogen uptake was monitored on line by a thermal
conductive detector.
The transmission electron microscopy (TEM) images were
obtained on a JEM 2010F field emission system operated at
200 kV. The fine catalyst powder was dispersed ultrasonically in
ethanol for 30 min and a drop of the suspension was deposited
on a carbon-coated copper grid for TEM measurement.
X-ray photoelectron spectroscopy (XPS) was performed on a
PHI 1600 ESCA spectrometer equipped with an AlK
a X-ray radia-
tion source (h = 1486.6 eV). Catalyst samples were prereduced at
v
450 °C for 1 h ex situ before each measurement. The binding
energy (BE) values were referenced to the C1s peak at 284.5 eV.
Fourier transform infrared spectra (FTIR) of CO adsorption were
recorded on a Nicolet 6700 (Thermo Scientific), equipped with a
In this work, we report a highly active bimetallic Ni–Re/SiO2
catalyst for selective deoxygenation of m-cresol to toluene that
greatly inhibits CAC hydrogenolysis under mild conditions. m-
Cresol was used as a model compound because it is an important
intermediate product from HDO of complex phenolic compounds,
such as guaiacol [23,24]. The inert silica was used to avoid compli-
cations due to the support, and therefore, to allow us to focus on
better understanding the role of the bimetallic catalysts. Using a
combination of detailed characterizations and density functional
theoretical calculation, the active site for deoxygenation is deter-
mined to be the Ni–Re neighboring sites.
liquid-N2-cooled mercury–cadmium–tellurium detector and a
transmittance cell. The catalyst wafer (20 mg) was reduced
in situ at 450 °C for 1 h with flowing H2 at 30 mL/min. After that,
the sample was cooled to room temperature in flowing He at
30 mL/min and then the background was recorded. The sample
was exposed to 5% CO/He (30 mL/min) for 30 min, followed by
He purging for another 30 min. The spectra were recorded at a res-
olution of 4 cmꢁ1 and 128 scans.
The dispersion of Ni was estimated based on dynamic CO
chemisorption, measured in a microreactor system equipped with
a Cirrus 200 mass spectrometer (MKS). The system consisted of an
oven with a temperature controller, gas delivery system, micro
quartz reactor (6 mm o.d.), and six-port valve. Catalyst samples
(50 mg, 40–60 mesh) were reduced at 450 °C for 1 h in flowing
H2, followed by He purging (30 mL/min) for another 30 min. Then
the temperature was decreased to room temperature. Pulses of 5%
CO/He (100 mL) were sent to the catalyst through the six-port valve
until a constant CO peak area was reached.
2. Experimental
2.1. Catalyst preparation
The bimetallic Ni–Re catalysts were prepared using incipient
wetness co-impregnation of SiO2 (Sigma, SBET = 200 m2/g) with
aqueous solutions of calculated amounts of Ni(NO3)2ꢀ6H2O (Strem
Chemicals) and NH4ReO4 (Alfa). After impregnation for 12 h, the
samples were dried overnight at 120 °C, followed by calcination
at 400 °C for 4 h with a heating rate of 2 °C/min. The amount of
Ni loading was 5 wt.%, while Re loading was varied from 0 to
10 wt.%. The catalyst was denoted as 5NiXRe, where 5 and X repre-
sent the weight percentages of Ni and Re, respectively, in the sam-
ple. The catalyst powder was pressed, crushed, and sieved to 40–60
mesh for reaction.
NH3 and isopropylamine temperature–programmed desorption
(NH3 TPD and IPA TPD) were measured on the same system as CO
chemisorption. A catalyst sample of 100 mg was reduced at 450 °C
for 1 h and then purged with flowing He (30 mL/min) for another
30 min. The temperature was reduced to 100 °C. For NH3 TPD,
the sample was exposed to a stream of 2% NH3/He (30 mL/min)
for 30 min. For IPA TPD, liquid IPA (5 lL/pulse, 10 pulse, 3 min/
pulse) was injected using a microsyringe manually and vaporized
before entering the reactor. After the sample was flushed with
flowing He (30 mL/min) for 30 min, the temperature was increased
to 650 °C at a rate of 10 °C/min. The quantification was done by
injecting pulses of 2% NH3/He (10 mL/pulse).
2.2. Catalyst characterization
Diffuse reflectance infrared Fourier transform spectroscopy
(DRIFTS) of pyridine adsorption were measured using a Frontier
spectrometer (PerkinElmer), equipped with a diffuse reflectance
accessory and a reaction chamber (Harrick). A fine powder catalyst
was loaded into the sample cup of the chamber. After the sample
was reduced at 450 °C for 1 h and purged with He for 30 min, it
was cooled to 100 °C, at which a background spectrum was
recorded. Pyridine vapor was then introduced for 30 min, followed
by purging using He for another 30 min. The DRIFT spectra were
recorded at a resolution of 4 cmꢁ1 with 64 scans.
Nitrogen adsorption was recorded at liquid nitrogen tempera-
ture in an automatic Micrometrics Digisorb 2600 analyzer. The cat-
alysts were outgassed at 350 °C prior to measurements. The
specific surface area of the catalysts was calculated using the Bru
nauer–Emmett–Teller (BET) method in the relative pressure range
0.005 < p/p0 < 0.27.
X-ray diffraction (XRD) patterns of the catalyst samples were
recorded on a Rigaku D/max 2500 diffractometer with a Cu K
a
radiation source (k = 1.54056 A) in the 2h range 20°–80°.
The Raman spectra were obtained on a Renishaw Raman spec-
trometer, with an Ar+ ion laser (532 nm) being the exciting light
2.3. Density functional theory calculations of phenol adsorption
source. The focusing spot size was about 1
collected at room temperature, with an acquisition time of 10 s/
scan and a resolution of 2 cmꢁ1
Hydrogen temperature-programmed reduction (H2 TPR) was
performed on a Chemisorb 2750 (Micrometrics). Prior to measure-
lm. All spectra were
Density functional theory (DFT) periodic slab calculations were
carried out using the Vienna Ab Initio Simulation Package (VASP)
[25–27]. The periodic DFT code uses the projector-augmented
wave (PAW) method to describe the effective core potentials and
.

86
F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
plane waves for the valence electrons [27,28]. The exchange-
correlation energy of interacting electrons was evaluated using
the Perdew–Burke–Ernzerhof (PBE) functional [29].
The XRD patterns of Ni–Re catalysts before and after reduction
are shown in Fig. 1, with the particle sizes estimated using the
Scherrer equation reported in Table 1. For the fresh catalysts
(Fig. 1A), sharp diffraction peaks attributed to NiO were observed
for the monometallic 5Ni sample. Increasing the amount of Re
resulted in diffraction peaks of NiO broadening and weakening
gradually. Therefore, the estimated particle size of NiO decreased
as the amount of Re increased (Table 1). However, no diffraction
peaks attributable to ReOx were observed in either the monometal-
lic 2.5Re sample or bimetallic Ni–Re samples, regardless of the Re
loadings. These results indicate that ReOx species were highly dis-
persed in all bimetallic Ni–Re samples and were probably in prox-
imity to NiO, which, in turn, increases the dispersion of NiO on
silica as compared with that for the monometallic Ni samples.
When the catalysts were reduced at 450 °C for 1 h (Fig. 1B),
metallic Ni of face-centered cubic (fcc) structure and metallic Re
of hexagonal closed packed (hcp) structure were clearly observable
for the monometallic 5Ni and 2.5Re catalysts, respectively. The
presence of Re in bimetallic catalysts led to the broadening of the
Ni diffraction peaks, indicating that the size of Ni or bimetallic
Ni–Re particles was reduced, in good agreement with the Ni–Re/
ZrO2 catalyst [32]. Interestingly, the diffraction peak of Ni(1 1 1)
at 2h of 44.41° for monometallic 5Ni shifted gradually to a lower
value of 43.95° for bimetallic 5Ni2.5Re. This shift resulted from
the lattice expansion of the Ni unit cell (Table 1), likely due to Re
(which has a larger atomic diameter than Ni) penetrating into
the Ni lattice during the reduction. This phenomenon has been
considered as an indication of alloy formation in the bimetallic
Ni–Re catalyst [33] and other bimetallic catalysts, such as Ni–Fe
[34] and Ni–Co [16]. When the Re loading was increased to
>2.5%, the diffraction peaks shifted further to lower values and
the shape of the patterns changed from fcc Ni to hcp Re, which
might indicate that the structure of bimetallic particles changed
from fcc to hcp as a result of increased Re content. It is worth not-
ing that the estimated particle size for the Ni–Re samples after
reduction (Table 1) reflects contributions from both Ni and Re as
Re penetrated into the Ni lattice during the reduction, particularly
at high Re loadings. The particle size of 5Ni10Re is slightly greater
than that of 5Ni5Re as a result of the doubled Re loading.
Bare Ni and Re-doped Ni (111) (referred to as (Re)Ni(111) in
the following discussion) surfaces were modeled with a four-
layer slab in a p (4 ꢂ 4) surface unit cell, separated by at least
14 Å of vacuum in the z-direction. The computed bulk lattice con-
stant of 3.519 Å for Ni was used to construct the slabs for surface
calculations. The atoms in the bottom two layers were fixed, while
those in the top two layers were allowed to relax. A cutoff energy
of 400 eV and a (2 ꢂ 2 ꢂ 1) k-point grid were found to provide con-
verged adsorption energies [30]. The atomic structures were
relaxed using either the quasi-Newton scheme or the conjugate
gradient algorithm as implemented in the VASP code.
The adsorption energy was defined as
Eads ¼ Eðadsorbate=surfaceÞ ꢁ EðadsorbateÞ ꢁ E
ð1Þ
ðbare surfaceÞ
where E(adsorbate/surface) is the total energy of an adsorbate bound to
Ni or the Ni–Re slab, E(adsorbate) is the total energy of a molecule in
gas phase, and E(bare surface) is the total energy of the bare slab. Phe-
nol was used as a model compound. The transition state for deoxy-
genation was confirmed to have only one imaginary mode in
normal mode frequency analysis.
2.4. Catalytic activity
The vapor-phase conversion of m-cresol over the Ni, Ni–Re, and
Re catalysts was evaluated in a downflow fixed-bed quartz tube
reactor (6 mm o.d.) under atmospheric pressure, as has been
reported [13,31]. Briefly, the catalyst sample (40–60 mesh) was
placed between two layers of quartz wool in the middle of the
reactor. The catalyst was reduced at 450 °C for 1 h in flowing H2
(30 ml/min) and then cooled down to reaction temperature. m-
Cresol was fed using a syringe pump (KDS100, kd scientific) and
vaporized before entering the reactor. H2/m-cresol molar ratio
was kept at 60 in all runs. The products were quantified on line
in a gas chromatograph (GC7890B, Agilent). The effluent was
trapped by methanol in an ice water bath, and its components
were identified using GC-MS (Shimadzu QP2010SE). Fresh samples
ꢁ1
were used for each space time (W/F, g
g
h). The conversion
cat reactant
and yield are reported in molcarbon%.
3.1.2. Raman spectroscopy
Raman spectroscopy was used to probe the surface structure of
ReOx species in Ni–Re catalysts. As shown in Fig. 2, 5Ni and bare
SiO2 showed similar feature with bands at 603 and 489 cmꢁ1 and
a broad band at ꢃ400 cmꢁ1, which are characteristic bands of silica
supports [35]. The NiAO stretching band at 550–500 cm-1 is very
weak [36] and may be masked by the bands of silica. In contrast,
the 5Re sample showed sharp and strong bands at 965, 914, 891,
3. Results and discussion
3.1. Catalyst characterizations
3.1.1. BET and XRD
As shown in Table 1, the specific surface area of the catalysts
decreased moderately as the amount of Re increased.
and 345 cmꢁ1. These bands can be assigned to
vs(Re@O), vas(-
Table 1
Physical properties of monometallic 5Ni, 2.5Re, and bimetallic Ni–Re catalysts.
Catalyst
Ni/Re
SBET (m2/g)
Particle size (nm)
Lattice
Crystal structure
Lattice
CO/Ni
Acid densityc
(lmol/g)
distance (nm)
constant (Å)
Mass
Mole
XRDa
XRDb
TEMb
25.5
XRD
TEM
a or b
c
5Ni
–
5
2
1
0.5
–
–
181
184
182
167
145
197
22.9
13.6
11.5
3.4
3.0
–
21.3
8.0
7.5
2.3
3.0
4.9
0.204
0.205
0.206
0.209
0.210
0.212
0.204
fcc
fcc
fcc
hcp
hcp
hcp
3.530
3.551
3.565
2.744
2.759
2.764
3.530
3.551
3.565
4.441
4.451
4.482
0.016
0.04
0.045
0.085
0.103
n.d.d
15.3
15.4
25.1
51.2
60.5
20.7
5Ni1Re
5Ni2.5Re
5Ni5Re
5Ni10Re
2.5Re
15.9
6.35
3.17
1.59
–
9.9
5.5
0.207
–
a
b
c
Particle size of calcined catalysts.
Particle size of reduced catalysts.
Derived from NH3 TPD.
d
Not detected.

F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
87
Re(002)
Re(100)
(A)
(B)
Re(110)
NiO(101)
NiO(012)
NiO(113)
Re(101)
NiO(110)
f
f
e
d
e
d
c
c
b
b
NiO(202)
Ni(200)
Ni(111)
a
Ni(220)
a
20
30
40
50
( )
60
70
80
20
30
40
50
60
70
80
2
2 ( )
Fig. 1. XRD patterns of monometallic 5Ni, 2.5Re, and bimetallic Ni–Re catalysts: (A) fresh catalysts; (B) reduced at 450 °C for 1 h. (a) 5Ni; (b) 5Ni1Re; (c) 5Ni2.5Re; (d) 5Ni5Re;
(e) 5Ni10Re; (f) 2.5Re.
489
603
872
900
5Ni5Re*
5Re*
990
1010
977
950
5Ni10Re
345
1002
5Ni7.5Re
5Ni5Re
965
5Ni5Re
5Re
914
891
×1/3
5Ni2.5Re
5Ni1Re
5Ni
2.5Re
5Ni
320
347
SiO2
100
200
300
Temperature (°C)
400
500
1200
1000
800
600
400
200
0
Raman Shift (cm-1)
Fig. 3. H2 TPR profiles of monometallic 5Ni, 2.5Re, and bimetallic Ni–Re catalysts.
Fig. 2. Raman spectra of different catalysts. An * indicates that the catalysts were
reduced ex situ at 450 °C for 1 h; the others were fresh catalysts.
broad peak at a slightly lower temperature of 320 °C, indicating
that isolated ReOꢁ4 species with strong interaction with silica are
more difficult to reduce than the larger NiO particles with weak
interaction with silica. Interestingly, all bimetallic Ni–Re samples
showed lower reduction temperatures than monometallic 2.5Re.
This is expected, as Ni facilitates the reduction of adjacent ReOx
species through hydrogen activation on Ni and spill over to ReOx.
The lowered reduction temperature indicates that the ReOx species
are in close contact with the NiO particle, consistent with the
Raman result. At Re loadings ꢄ5%, bimetallic NiRe only showed
one reduction peak. When the Re loading was further increased,
the reduction peak appears to contain at least two components,
which is probably due to the formation of Ni–Re alloys with differ-
ent Ni/Re ratios.
Quantitative results show that the molar ratio of H2 consump-
tion to Ni (H2/Ni) for monometallic 5Ni is 0.95, indicating almost
complete reduction of NiO. In contrast, the H2/Re ratio is 2.4 for
monometallic 2.5Re. Similar H2/Re ratio values of 2.1–2.5 are
obtained for bimetallic Ni–Re catalysts, assuming complete reduc-
tion of NiO. As indicated in the Raman results, the main ReOx spe-
cies is ReOꢁ4 , in which Re has a formal valence state of +7, requiring
a H2/Re ratio of 3.5 for complete reduction. The difference in mea-
sured and theoretical H2/Re ratios indicates that the Re species
were not completely reduced in TPR. Based on the H2 consumption,
we estimated the average valence state of the Re species after the
reduction is about +2. Since particles of metallic Re were detected
in XRD, we can conclude that the resulting catalyst contains both
Re@O),
vs(ReAOASi), and d(OAReAO), respectively, of isolated
tetrahedral ReOꢁ4 species with a structure of SiAOARe(@O)3 on
the silica surface [35,37,38]. No characteristic bands of ReAOARe
at 800, 450, and 200–150 cmꢁ1 were observed, suggesting that
the ReOx are in the state of isolated ReOꢁ4 highly dispersed on the
surface instead of polymeric Re2O7 [38]. Consequently, XRD pat-
terns of the fresh catalyst did not show any signal of ReOx. In com-
parison with monometallic 5Re, the bands of isolated tetrahedral
ReOꢁ4 species in bimetallic 5Ni5Re shifted to higher wavenumbers
of 1002, 965, 950, and 900 cmꢁ1, with reduced peak intensity and
broadened peak width. By analogy to Ir–Re/SiO2 catalysts [39],
these changes might be a result of the decrement in the bond
length of Re@O due to the formation of NiAOARe(@O)3 species
on the surfaces of NiO particles. This close proximity between
NiO and ReOx species would prevent the growth of NiO particle
during calcination and facilitate the penetration of Re atom into
the Ni unit cell during reduction. When the samples were reduced
at 450 °C, the band intensities of ReOx species decreased signifi-
cantly, indicating that most of the ReOx species were reduced by
the hydrogen treatment.
3.1.3. H2 TPR
H2 TPR was conducted to investigate the interaction between
NiO and ReOx (Fig. 3). Monometallic 2.5Re displayed a hydrogen
consumption peak at 347 °C, while monometallic 5Ni showed a

88
F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
completely and partially reduced Re species after reduction treat-
ment. Because Ni helps to reduce ReOx at lower temperatures,
we anticipate that part of the incompletely reduced Re species stay
away from the Ni particles due to their strong interaction with the
silica support, consistent with the TEM results shown below.
3.1.4. TEM
Representative TEM images of the reduced 5Ni, 5Ni2.5Re, and
2.5Re samples are displayed in Fig. 4. Large nickel particles of
ꢃ20 nm were observed for monometallic 5Ni (Fig. 4A). In contrast,
small and uniform particles of Re with sizes <5 nm were observed
for monometallic 2.5Re (Fig. 4C). These results clearly indicate that
Ni tends to aggregate into large particles due to weak interaction
with silica, while Re tends to be well dispersed on silica due to
the strong oxophilicity of Re, which leads to strong interaction
with silica. The presence of Re in the 5Ni2.5Re sample significantly
reduced the particle size (Fig. 4B). Although large particles of
ꢃ20 nm were still present for 5Ni2.5Re, most particles are smaller
than 10 nm. In good agreement with the XRD results, the estimated
average particle sizes decreased from 25.5 nm for monometallic
5Ni to 9.9 nm for 5Ni2.5Re (Table 1). High-resolution TEM (HR-
TEM) measurements were performed to examine the fine struc-
tures (see insets in Fig. 4). For 5Ni, the measured lattice fringe is
0.204 nm and agrees well with Ni(111). For 5Ni2.5Re, the lattice
fringe is 0.207 nm, which is larger than that of Ni(111) but smaller
than that of Re(101) (0.210 nm). In good agreement with the XRD
results (Table 1), these results indicate that the penetration of Re
atoms into the Ni lattice during reduction resulted in the expan-
sion of lattice, which may be indicative of Ni–Re alloy formation
in the bimetallic sample.
The composition of the particles of bimetallic 5Ni2.5Re was fur-
ther analyzed by high-angle annular dark field scanning transmis-
sion electron microscopy (HAADF-STEM), accompanied by energy-
dispersive spectroscopy (EDS) analysis. The HAADF-STEM image
showed a bimodal distribution of particles with bright larger par-
ticles and dark smaller particles (Fig. 5). Single-point EDS analysis
of several bright particles showed an Ni/Re ratio ꢃ10 (see insets),
which is higher than the theoretical ratio of 6.35 for 5Ni2.5Re. This
result indicates that the bright larger particle is rich in Ni. Line-
scanning EDS analysis (see insets) of a larger particle (along the
red line in Fig. 5B) showed that the Ni/Re ratio is higher at the cen-
ter of the particle, suggesting that Re content decreases from the
external surface to the core of the Ni particle due to a longer diffuse
path during Re penetration into the Ni particle. EDS mapping con-
formed that the highly dispersed dark smaller particles (indicated
by an arrow in Fig. 5B) on the silica support are rich in Re. As dis-
played in Fig. 6, the bright larger particles contain Ni and Re uni-
formly distributed in the particle and are rich in Ni. On the other
hand, the small dark particles rich in Re are highly dispersed on
the silica support. The STEM results indicate that the Re species
in close contact with the Ni species stabilizes the dispersion of Ni
on silica.
3.1.5. XPS
The electronic interactions and surface compositions of
bimetallic Ni–Re catalysts were analyzed using XPS. Due to the
limitations of the instrument, the samples can only be reduced
ex situ for XPS measurement. Consequently, the surfaces may have
been partially oxidized while being transferred for XPS measure-
ments. To analyze Ni–Re interaction, spectral analysis was per-
formed by fixing the BE of oxidized Ni2+2p3/2 at 856.1 eV and the
corresponding shakeup satellite peak at 861.1 eV [40], and oxi-
dized Re4+ and Re3+ of 4f7/2 at 42.4 and 40.7 eV, respectively [41].
The results are shown in Fig. 7. In comparison with monometallic
5Ni at 852.7 eV, the BE of Ni02p3/2 for bimetallic 10Ni5Re shifts
to a higher BE by 0.5–853.2 eV (Fig. 7A). On the other hand, the
Fig. 4. TEM images of (A) 5Ni; (B) 5Ni2.5Re; (C) 2.5Re. The inserts show the HR-
TEM image. The samples were prereduced at 450 °C for 1 h.

F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
89
Fig. 5. HAADF-STEM image of 5Ni2.5Re. (A) A larger area with insert shows the Ni/Re molar ratio of several particles; (B) a smaller area with inset shows line scanning results
of EDS. The catalyst was prereduced at 450 °C for 1 h.
Fig. 6. (A) HAADF-STEM of 5Ni2.5Re revealing the EDS mapping region; (B) EDS mapping of Ni element; (C) EDS mapping of Re element. The catalyst was prereduced at
450 °C for 1 h.
Re3+
Ni2+
856.1
(B) Re 4f
(A) Ni 2p
Re4+
40.7
Re0
42.4
39.9
861.1
Ni0
853.2
10Ni5Re
40.0
10Ni5Re
852.7
850
10Ni
865
5Re
48
46
44
42
40
38
860
855
Binding Energy (eV)
Binding Energy (eV)
Fig. 7. XPS spectra of (A) Ni2p and (B) Re4f. Dots, experimental data; lines, curve fittings. The samples were prereduced ex situ at 450 °C for 1 h.
BE of Re04f7/2 only changed slightly, from 40.0 eV for 5Re to 39.9 eV
for 10Ni5Re (Fig. 7B). Consistent with previous work on unsup-
ported Ni–Re [40], the results indicate that the relatively
electron-rich d band of Ni loses its electron density through
hybridization either to its 4s and 4p states or to the relatively

90
F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
electron-poor d band of Re [42], resulting in the observed shift of
the Ni2p state to higher BE.
2047
To further understand the electronic interactions, we analyzed
the charge distribution of a Ni–Re dimer. As shown in the isosur-
faces of positive and negative charges (Fig. S1 in the Supplemen-
tary Material), the Ni end of the Ni–Re dimer concentrated both
positive and negative charges, indicating a strong redistribution
of electronic charges around the Ni atom. Although the overall
charge on Ni is negative, the net reduction of the d electronic den-
sity is expected to dominate the observed peak shifts in XPS. In
fact, a similar observation has been reported on the Pd–Cu bimetal-
lics, where the overall electron transfer occurred from Cu to Pd but
the population of the Pd4d orbitals is lower in the alloy than in the
pure metal [43].
0.004
2002
1906
1910
2090
5Ni2.5Re
2036
5Ni
2.5Re
As shown in Table 2, the bimetallic 10Ni5Re is more easily oxi-
dized than the monometallics (either 10Ni or 5Ni) after exposure
to air. The surface Ni/Si ratio increases significantly from 10Ni to
10Ni5Re, whereas the Re/Si ratio decreases from 5Re to 10Ni5Re,
indicating that the Re–Ni interactions in the bimetallic Ni–Re cat-
alyst improve the dispersion of Ni on the surface of SiO2. The sur-
face Ni/Re ratio was found to be 3.3, lower than the theoretical
value of 6.3. The results indicate a higher dispersion of Re than of
Ni, as well as the enrichment of Re on the surface of the bimetallic
Ni–Re particles, in good agreement with the EDS line scanning
result (Fig. 5B).
2200
2100
2000
1900
1800
Wavenumber (cm-1)
Fig. 8. FTIR spectra of CO adsorption on 5Ni, 2.5Re, and 5Ni2.5Re.
ported by the fact that the band intensity increases with increasing
amounts of Re (data not shown). Interestingly, the linear CO band
is blue-shifted from 2036 cmꢁ1 on 5Ni to 2047 cmꢁ1 on bimetallic
5Ni2.5Re. Consistent with the XPS results, this shift could result
from electronic interaction between Ni and Re that reduces the
d-band electron densities of Ni [40,42,50]. Consequently, the elec-
tron back-donation from d orbitals of Ni to the
p⁄ orbital of CO is
3.1.6. FTIR of CO adsorption
reduced, and therefore the NiAC bond is weakened while the
CAO bond is strengthened, resulting in a blue shift in the CAO
stretching frequency. The most significant change from
monometallic 5Ni to bimetallic 5Ni2.5Re is the significantly
increased L/B ratio, reaching 1.96 from 0.59. The high L/B ratio
indicates that the addition of Re increases the population of termi-
nal sites but reduces the bridge sites for CO adsorption. This is
believed to be achieved by breaking the continuity of the Ni surface
and forming smaller Ni atom ensembles due to the presence of Re.
The reduction of the surface Ni ensembles provided more sites for
linear adsorption of CO than for bridge adsorption. We expect that
the changes of surface properties also influence the adsorption of
phenolics on the surface, and thus the mechanism and selectivity
of the HDO reactions.
Fig. 8 shows IR spectra of CO adsorption on 2.5Re, 5Ni, and
5Ni2.5Re. Two absorption bands at 2036 and 1910 cmꢁ1 are pre-
sent in monometallic 5Ni, which can be ascribed to linear- and
bridge-adsorbed CO, respectively [44,45]. The ratio of integrated
peak area of linear to bridge adsorption (L/B) is 0.59. This low L/B
ratio for 5Ni sample is consistent with the TEM and XRD results,
which showed large size of Ni particles. The large Ni particle con-
tains more close-packed Ni(111) facets, which favor bridge over
linear adsorption of CO. Interestingly, no CO adsorption band was
observed for monometallic 2.5Re. It has been reported that CO only
adsorbs on metallic Re but not on ReOx [46]. Moreover, CO adsorp-
tion on metallic Re was reported to be extremely weak at room
temperature [47]. Considering the TPR result that Re was not com-
pletely reduced during the pretreatment, we can infer that CO
barely adsorbed on the Re species in Re-containing samples under
this condition.
Compared with CO adsorption on monometallic samples,
changes in the important adsorption feature are evident for
bimetallic 5Ni2.5Re. In addition to the bands of linear- and
bridge-bound CO at 2047 and 1906 cmꢁ1, respectively, a small
band at 2090 cmꢁ1 and a shoulder band at 2002 cmꢁ1 emerged.
The former has been assigned to CO linearly adsorbed on defect
Ni0 sites [45]. The latter could not be assigned to CO adsorption
on Re0 due to the absence of characteristic adsorption bands at
2059, 2039, 2016, and 1963 cmꢁ1 [48]. Based on the literature
[49], we tentatively assign this band to CO linearly adsorbed on
Ni0 (e.g., the Ni atom of the Ni–Re alloy) that is perturbed by vicinal
Re, since Re is a strongly oxophilic element. This assignment is sup-
3.1.7. CO chemisorption
The number of surface active sites was quantified by dynamic
pulse chemisorption of CO. The amount of CO adsorbed onto
monometallic 2.5Re was under the detection limit of MS, which
suggests that Re species do not adsorb CO under this condition,
in good agreement with IR results. Since a similar degree of reduc-
tion of Re species was achieved over Re-containing samples, the CO
adsorbed on bimetallic Ni–Re catalysts is only related to the sur-
face Ni sites. As shown in Table 1, CO/Ni increased significantly
with the increasing amount of Re. This result is consistent with
the XRD and TEM characterization, suggesting that addition of Re
stabilizes the smaller Ni particles through NiAOARe interactions
Table 2
Distribution of different valence states of Ni and Re as well as atomic ratios derived from XPS.
Catalyst
Distribution of valence states (%)
Ni
Atomic ratio
Ni/Si
Re
Re/Si
Ni/Re
Ni0
Ni2+
Re0
Re3+
Re4+
10Ni
5Re
10Ni5Re
57.4
42.6
–
67.1
41.3
–
21.0
33.1
–
11.9
25.7
0.027
0.154
–
–
0.072
0.047
45.3
54.7
3.3

F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
91
during calcination and provides more surface Ni sites after
reduction.
r and p bonds of NiAC, while the O in the AOH group was pushed
away from the surface, with a distance of 2.82 Å between O and the
nearest Ni atom. The strong repulsion between Ni and O also lifted
the connecting carbon atom (C1) away from the aromatic plane by
ꢃ0.1 Å with respect to other carbon atoms (Table 3). This adsorp-
tion configuration is similar to the results of theoretical calcula-
tions [56–58] and high-resolution electron energy loss
spectroscopy experiments [19,59] on phenol adsorption on Group
VIII metals. Compared with phenol adsorption on Ni(111),
although flat adsorption was still the most favorable configuration
on the (Re)Ni(111) surface, some important features were evident,
as shown in Fig. 10C. First, the aromatic ring rotates counterclock-
wise by 6° on the bridge site (configuration Bri22.1) to place the O
atom of phenol on top of the Re atom. Second, the O moves toward
Re in the surface due to the strong oxophilicity of Re. The distance
between O and Re is 2.41 Å, suggesting the formation of a ReAO
bond, which is supported by density of states (DOS) analysis. As
shown in Fig. S3, there are obvious overlaps between the d orbitals
of Re and the p orbitals of O. Meanwhile, the angle of the O–
C1―plane decreases from 30.3° on bare Ni(111) to 22.3° on (Re)
Ni(111). Third, in contrast to the H atom of the hydroxyl of phenol
pointing to the surface on Ni(111), the H atom is tilted away from
the surface on (Re)Ni(111). These structural changes lead to the
CAO bond of phenol being lengthened from 1.37 Å in the gas phase
and on bare Ni(111) to 1.42 Å on (Re)Ni(111). In addition, the
bond length of C1–Ni was reduced by 0.1 Å. In summary, the pres-
ence of an oxophilic Re atom in the Ni surface makes the O more
easily bound to the surface and the C1 closer to the Ni atom in
the surface. Therefore, the CAO bond may cleave more readily on
the Ni–Re neighboring sites, resulting in higher HDO reactivity.
We also note that the adsorption of phenol is slightly weaker on
(Re)Ni(111) than on bare Ni(111) by 6.8 kJ/mol (Table 3).
Recently, McEwen and co-workers showed that flat adsorption of
phenol is stronger on Pd(111) than on oxophilic Fe(110), even
though there is FeAO bond but no PdAO bond formation [60].
Although the aromatic ring is closer to the surface on Fe than on
Pd, the charge transfer between Pd and the aromatic ring is more
pronounced than that between Fe and the aromatic ring [60], lead-
ing to a higher adsorption energy of phenol on Pd(111) than on Fe
(110). The XPS results and FTIR results of CO adsorption on
bimetallic Ni–Re have shown that the d-band electron density of
Ni becomes deficient in the presence of adjacent Re. Thus, one
would expect that the deficient Ni d band in Ni–Re reduces the
charge transfer between surface Ni and the aromatic ring, and con-
sequently weakens the adsorption strength of phenol on (Re)Ni
(111). This weakened adsorption of the aromatic ring facilitates
the desorption of aromatic products and prevent the subsequent
hydrogenolysis of the CAC bonds.
3.1.8. NH3 TPD, IPA TPD, and DRIFTS of pyridine adsorption
Since the presence of strong acid sites could affect the reaction
pathway and therefore the product distributions [31,51], the acidic
property of the bimetallic catalysts has been investigated using
NH3 TPD and IPA TPD. The IPA TPD (data not shown) did not show
any Brønsted acid-catalyzed decomposition products of propylene
or NH3 [52,53] for any samples, indicating the absence of strong
Brønsted acid sites (BAS) in these samples. This result is consistent
with Kawai et al., who performed pyridine TPD on Re/Al2O3 and
showed that a higher reduction temperature changed BAS to Lewis
acid sites (LAS) [54]. Fig. 9 shows the NH3 TPD profiles, with quan-
tified results summarized in Table 1. Again, silica support exhibited
no acidity, and both monometallic 5Ni and 2.5Re contain very few
acid sites. However, the number of acid sites increases with the
increasing Re loading in the bimetallic Ni Re samples. These results
suggest that new acid sites are created in the presence of Re upon
reduction, in line with previous work, which showed that acid sites
could be created in bimetallic catalysts containing a noble hydro-
genating metal and an oxophilic metal [55]. It should be noted that
the center of the desorption peak is lower than 200 °C in all sam-
ples. Taking the IPA TPD results into account, we believe that the
acid sites in the bimetallic Ni–Re catalysts are weak LAS. These
sites could be the completely and/or partially reduced Re sites
adjacent to Ni, which serve as active sites to adsorb the oxygen
atom of cresol. Additional DRIFTS measurements of pyridine
adsorption were performed (Fig. S2). The adsorption band at
1446 cmꢁ1 in both monometallic and bimetallic samples can be
assigned to pyridine adsorption on LAS [54]. In contrast, the char-
acteristic adsorption band for pyridine adsorption on BAS at
ꢃ1540 cmꢁ1 [54] was absent in all samples, indicating no BAS in
presence under this pretreatment condition.
3.2. Phenol adsorption on Ni(111) and (Re)Ni(111)
Based on the characterizations, the Ni(111) surface and the sur-
face with one surface Ni atom replaced by one Re atom were used
to represent the monometallic 5Ni and bimetallic Ni–Re catalysts,
respectively. The most stable configuration of phenol adsorption
on Ni(111) was found to be flat adsorption on a bridge site with
an angle between the axis of phenol and the close-packed direction
of the (111) surface of 28.2° (configuration Bri28.2), as shown in
Fig. 10B. Similarly to phenol adsorption on Pd and Pt surfaces
[30], the aromatic ring was adsorbed on the surface through mixed
3.3. Catalytic performance
Fig. 11 compares the major products’ evolution as a function of
W/F during HDO of m-cresol over monometallic 5Ni and bimetallic
5Ni2.5Re catalysts. Apparently, bimetallic 5Ni2.5Re is much more
active than monometallic 5Ni. To reach 20% conversion, the
required W/F on 5Ni is five times longer than that on 5Ni2.5Re
(Fig. 11A). At low conversions, both catalysts showed major prod-
ucts toluene (Tol), 3-methylcyclohexanone (MCHone), 3-
methylcylcohexanol (MCHol), phenol (Ph), and methane (CH4), as
a result of three parallel reaction pathways: apparent direct deoxy-
genation, hydrogenation, and CAC hydrogenolysis (Scheme 1).
However, the product distributions are significantly different on
the two catalysts. On 5Ni, the hydrogenolysis products (Ph and
CH4) had yields comparable to those of Tol and MCHone. This
result is consistent with previous works that showed that nickel
catalysts favor multiple CAC hydrogenolysis (ring opening) and
5Ni10Re
5Ni5Re
5Ni2.5Re
5Ni1Re
2.5Re
5Ni
SiO2
100
200
300
400
500
Temperature (°C)
Fig. 9. NH3 TPD profiles of different catalysts.

92
F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
Fig. 10. Structure of (A) gas phase phenol; (B) phenol adsorption onto Ni(111) surface with a structure of Bri28.2; and (C) phenol adsorption onto (Re)Ni(111) surface with a
structure of Bri22.1. The unit of the numbers in the figure is Å.
successive hydrogenolysis at the terminal CAC bond [6,13], result-
ing in the formation of low-value CH4 with very high hydrogen
consumption. On 5Ni2.5Re, Tol is the dominant product, with
Table 3
The distances of CANi, CAO, OANi, and OARe, the angle of the OAC1 plane, and the
CH4 and Ph being significantly reduced and becoming minor prod-
ucts, while the yields of MCHone and MCHol are slightly reduced.
Evidently, selective deoxygenation of m-cresol to Tol is achieved
on the bimetallic NiRe catalysts while CAC hydrogenolysis is
greatly suppressed.
As shown in Table 4, the minor products include the products
from acid-catalyzed primary reactions of isomerization and
transalkylation (p- and o-cresol, xylenols, and phenol) and sec-
ondary hydrogenation, deoxygenation, and CAC hydrogenolysis
products (cyclohexanone, cyclohexanol, benzene, cyclohexane,
methylcyclohexane, C2–C6 hydrocarbons). Since the acid strength
of bimetallic NiRe is weak (see Fig. 9), acid-catalyzed reactions
occur to a limited extent. Furthermore, in contrast to the strong
adsorption energy (Ead) of phenol on Ni(111) and (Re)Ni(111) surfaces.
Surface
Ni(111)
(Re)Ni(111)
C1ANi (Å)
C2ANi (Å)
C3ANi (Å)
C4ANi (Å)
C5ANi (Å)
C6ANi (Å)
CAO (Å)
OANi (Å)
OARe (Å)
\OAC1―plane (°)
Ead (kJ/mol)
2.21
2.07
2.06
2.07
2.05
2.03
1.37
2.82
–
2.12
2.03
2.06
2.12
2.03
2.04
1.42
–
2.41
22.33
ꢁ109.9
30.09
ꢁ116.7
14
25
20
15
10
5
14
(A)
(C)
(B)
12
10
8
12
Tol
Tol
MCHone
MCHol
CH4
MCHone
MCHol
CH4
5Ni2.5Re
10
8
Ph
Ph
5Ni
6
6
4
4
2
2
0
0
0
0
4
8
12 16 20 24
0.0 0.2 0.4 0.6 0.8 1.0
W/F (h)
0
4
8
12 16 20 24
m-Cresol conversion (%)
m-Cresol Conversion (%)
Fig. 11. Effect of W/F on m-cresol conversion over 5Ni and 5Ni2.5Re (A) and major products distribution as a function of m-cresol conversion on 5Ni (B) and 5Ni2.5Re (C).
Reaction conditions: T = 300 °C, P = 1 atm, H2/m-cresol = 60, TOS = 30 min.

F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
93
OH
C-C hydrogenolysis
C-C hydrogenolysis
CH4 +
CH4
Ph
C-C hydrogenolysis
Hydrogenation
OH
Deoxygenation
Tol
MCHane
m-Cr
Hydrogenation
Hydrogenation
O
OH
C-O hydrogenolysis
Hydrogenation
MCHone
MCHol
Scheme 1. Major reaction pathway of m-cresol conversion on Ni and Ni–Re catalysts.
Table 4
Effect of catalyst on m-cresol conversion and products distribution.
a
Catalyst
5Ni
2.5Re
3.0
2.5ReOx
0.8
5Ni2.5Re
47.6
5Ni+2.5Reb
14.9
Conversion (%)
10.3
Product yield (%)
Tol
MCHone
MCHol
CH4
Ph
Ben
MCHane
C2-6
1.9
2.2
0.5
1.9
2.4
0.1
0.0
0.2
0.7
0.4
2.9
0
0
0
0
0
0
0
0
0.5
0
0
0
0
0
0
0
0
23.7
3.0
0.7
5.6
5.9
5.8
2.4
0.4
1.5
2.3
0.3
0.03
0.09
0.6
1.5
2.6
0.04
0.14
0.9
Xylenols
Othersc
0.1
0.3
5.0
Note: Reaction conditions: T = 300 °C, P = 1 atm, H2/m-cresol = 60, W/F = 1 h, TOS = 30 min.
a
2.5ReOx refers to the monometallic 2.5Re catalyst without pretreatment at 450 °C in H2 for 1 h.
5Ni+2.5Re refers to the physical mixture of 5Ni and 2.5Re.
Others include p-/o-cresol, cyclohexanone, cyclohexanol, and heavy products with carbon number higher than 8.
b
c
acid sites in zeolite [31], the weak acid sites in Ni–Re do not change
the reaction pathway. Tests by feeding MCHol directly (see
Table S1) produced very small amounts of dehydration products
(methylcyclohexene, MCHene) as well as their hydrogenation
products (methylcyclohexane, MCHane), but large amounts of
dehydrogenation products (m-cresol and MCHone). Additional
tests by feeding MCHone directly produced very little MCHene
and MCHane (Table S1). These results indicate that complete
hydrogenation of the aromatic ring of m-cresol to MCHol followed
by dehydration to MCHene and further hydrogenation to MCHane
or CAO hydrogenolysis to MCHane (HYD path) is not important
over either catalyst.
Table 4 compares m-cresol conversion and product distribu-
tions over different type of catalysts at the same W/F of 0.5 h.
The monometallic 2.5Re is much less active than 5Ni, while it is
highly selective to Tol, and no hydrogenation or CAC hydrogenol-
ysis products were detected. Interestingly, the 2.5Re sample with-
out prereduction shows even lower activity. These results are
consistent with Feng et al., who showed that Re has very low activ-
ity for HDO of 4-propylphenol [32], but in contrast to Leiva et al.,
who showed that Re is very active for HDO of guaiacol [61]. The
differences may be results of very different reaction conditions
(temperature, pressure, support, solvent). The catalytic perfor-
mance of a physical mixture of 5Ni + 2.5Re appears to be the
sum of individual 5Ni and 2.5Re, indicating that Ni and Re work
independently. However, compared with monometallic 5Ni, the
conversion and the Tol yield improved by more than 4 and 10
times, respectively, on bimetallic 5Ni2.5Re. This improvement
highlights the crucial role of Ni–Re neighboring sites working in
synergy for high activity and selectivity toward toluene during
HDO of m-cresol. Moreover, the promotional effect of Re on Ni is
much greater than that of Fe on either Ni or Pd in the correspond-
ing bimetallic catalysts [20,21].
The promotional role of Re was further studied by varying the
amount of Re loading. The mass-based intrinsic reaction rate mea-
sured under differential reaction conditions is plotted in Fig. 12.
The intrinsic rꢁe1action rate increases almost lineaꢁrl1y from
0.536 mmol gꢁca1t s
on 5Ni to 3.367 and 6.764 mmol gcat sꢁ1 on
5Ni2.5Re and 5Ni5Re, respectively. When the Re loading is further
increased to 5Niꢁ110Re, the intrinsic reaction rate only increases to
8.851 mmol gꢁca1t s . The improvement factors are 6 and 12 for
5Ni2.5Re and 5Ni5Re, respectively. When the intrinsic reaction
rate was normalized to moles of Re loading, the reaction rate
was found to decrease when the Re loading was >5%.
The effect of the Re loading on the primary product distribution
were investigated at m-cresol conversion of ꢃ11.5 1%, which was
achieved by adjusting W/F (Fig. 13). Clearly, increasing the Re load-
ing improved the direct deoxygenation to Tol, while it inhibited the

94
F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
0.12
0.10
0.08
0.06
0.04
0.02
Conversion
Tol
MCHone+MCHol
CH4
0.00
0
2
4
6
8
10
Re loading (wt%)
Fig. 14. Turnover frequencies of m-cresol conversion and production of toluene,
methylcyclohexanone + methylcyclohexanol, and CH4. The reaction conditions are
the same as in Fig. 12.
Fig. 12. Effect of Re loading on intrinsic reaction rate of m-cresol conversion based
on mass of catalyst or moles of Re loading, measured under differential conditions
with conversion <15%. Reaction conditions: T = 300 °C, P = 1 atm, H2/m-Cresol = 60,
TOS = 30 min. W/F was varied to achieve m-cresol conversion <15%.
to 0.027 sꢁ1 on 5Ni10Re. The results indicate that the increased
number of active sites in bimetallic Ni–Re catalysts works syner-
gistically with the structure of active sites to enhance the HDO
activity and to improve the selectivity of deoxygenation to toluene.
60
50
40
30
20
10
0
Tol
MCHone
MCHol
CH4
3.4. Discussion
As shown in Scheme 2, the Re species in the fresh catalyst
appears to be in the state of isolated ReOꢁ4 species attaching to
NiO particles or SiO2 support. This Ni–Re interaction during calci-
nation resulted in better dispersion of NiO on silica. Upon reduc-
tion, the Re atoms penetrate into the unit cell of Ni, forming
intimate interactions, likely due to the similar d-spacing values
of Ni(111) and Re(101). The Ni and Re form bulk Ni–Re alloys at
Ni/Re ratios of 4/1 and 1/3, respectively [62], and their formation
requires high temperatures. Consequently, reduction at 450 °C
may not lead to formation of bulk stoichiometric Ni–Re alloys,
but formation of a nonstoichiometric surface Ni–Re alloy is likely,
as shown by the enrichment of Re on the surface of Ni–Re particles
(EDS and XPS results). The penetration of Re atoms into the surface
layers of Ni particles will disrupt the continuity of Ni atoms and
break the surface into smaller Ni ensembles, consequently chang-
ing the catalyst surface geometrically, as reflected in the increased
L/B ratio of CO adsorption. In addition, the close proximity of Re
and Ni also modifies the neighboring Ni atoms electronically, as
indicated by the shifted core level of Ni to a higher BE in XPS and
the blue shift of linearly adsorbed CO in FTIR. This is consistent
with the previous work, which showed that Ni in bimetallic Ni–
Re/CeO2 and Ca-promoted Ni–Re catalysts become electron-
deficient (cationic) due to the presence of Re in the vicinity, as evi-
denced by X-ray adsorption near-edge structure [50] and XPS [40].
In contrast to the Ni–Re catalysts, the electronic effect was absent
in the bimetallic Ir–Re and Rh–Re catalysts [63,64]. This difference
may be a result of a small Re ensemble only attached to the sur-
faces of Ir or Rh particles without penetration into their unit cells.
Under those conditions, one would expect little electronic effect
due to the limited interaction between Re and either Ir or Rh.
Several mechanisms have been proposed for selective direct
deoxygenation of phenolics, including direct deoxygenation
(DDO) [4], partial hydrogenation [65], and tautomerization [7,21].
In each mechanism, the CAO bond cleavage is the key step. On
Group VIII metals, the cleavage of the CAO bond requires the
adsorption of both C and O atoms on the two adjacent metal atoms
[66–68]. On a monometallic Ni catalyst, phenol adsorbs on a bridge
site through its phenyl ring with the O atom pointing away from
Ph
0
2
4
6
8
10
Re loading (wt%)
Fig. 13. Major products distribution as a function of Re loading at similar m-cresol
conversion of ꢃ11.5 1%. The reaction conditions are the same as in Fig. 12.
CAC hydrogenolysis to CH4 and Ph, with a slight reduction of the
hydrogenation products (MCHone and MCHol). The effect of Re
on the product distributions is more pronounced at Re loadings
less than 3%.
The mass-based intrinsic reaction rate may be influenced by
both the number and the structure of the active sites. To distin-
guish these effects, the turnover frequencies (TOFs) of m-cresol
conversion, deoxygenation to Tol, hydrogenation to MCHone and
MCHol, and hydrogenolysis to CH4 were calculated based on the
number of active sites measured with CO chemisorption and plot-
ted as a function of Re loading (Fig. 14). It should be noted that the
stoichiometry of CO adsorption changes as a function of Re loading
even though the CO/Ni ratio was assumed to be 1. However, this
will not change the trend in Fig. 14. The TOF of m-cresol conversion
doubled from 0.043 sꢁ1 on 5Ni to 0.093 sꢁ1 on 5Ni2.5Re, and then
slowly increased to 0.108 sꢁ1 on 5Ni10Re. A similar trend was
observed for TOF of toluene formation, which increased more than
four times from 0.008 sꢁ1 on 5Ni to 0.037 sꢁ1 on 5Ni2.5Re, and
again slowly increased to 0.058 sꢁ1 on 5Ni10Re. On the other hand,
the TOF for MCHone + MCHol doubled from 0.011 sꢁ1 on 5Ni to
0.025 sꢁ1 on 5Ni2.5Re but only increased slightly to 0.026 sꢁ1 on
5Ni10Re. In contrast, the TOF for CH4 was almost halved from
0.052 sꢁ1 on 5Ni to 0.033 sꢁ1 on 5Ni2.5Re and further decreased

F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
95
O
Re
O
O
O
O
O
O
O
O
O
(B)
(A)
O
Re
O
O
e
R
O
O
O
O
O
e
O
O
O
Re
O
Re
O
R
Re
O
O
O
O
O
Ni O
R
N
O
i
e
N
R
i
O
i
N
e
ReO
O
x
e
Ni-Re
O
NiO
R
SiO2
SiO2
Scheme 2. Schematic representation of the structure of bimetallic Ni–Re catalyst before (A) and after (B) reduction at 450 °C for 1 h.
Fig. 15. Structure of the transition sates of direct dehydroxylation of phenol on (A) Ni(111) surface and (B) (Re)Ni(111) surface. The unit of the numbers in the figure is Å.
the surface. To break the CAO bond, the adsorption of the O atom
onto the surface Ni atom requires overcoming the strong repulsion
between Ni and O, resulting in an increased activation barrier. Add-
ing H atoms to the phenyl ring appears to occur more readily, since
the ring is already on the surface. Therefore, hydrogenation of m-
cresol to MCHone and MCHol is more favorable than deoxygena-
tion of m-cresol to Tol at low conversions, particular when the
reaction temperature is low [13]. On the other hand, when oxophi-
lic Re is present in the surface of the bimetallic Ni–Re catalyst, both
the phenyl ring and the O atom can adsorb onto the surface, with
the phenyl ring on Ni with shortened C1–Ni bond and O on Re with
lengthened CAO bond. The cleavage of the CAO bond occurs more
readily, as the adsorption structure of phenolic on the Ni–Re sur-
face alloy resembles the transition state where both C and O bound
to the surface. Consequently, the TOF for deoxygenation of m-
cresol to Tol increased by four times on 5Ni2.5Re from that on
5Ni. These results highlight the crucial role of the Ni–Re neighbor-
ing sites for the adsorption of the phenolics and the subsequent
cleavage of the CAO bond. Similarly, one recent work suggested
that the selective deoxygenation of phenol to benzene at the inter-
face of Ru/TiO2, where the O atom of phenol adsorbed in the oxy-
gen vacancy or hydroxyl of TiO2 and the phenyl ring adsorbed onto
the Ru particle [69].
the reaction starts from the initial state, with the O atom of hydro-
xyl moving toward the nearest surface Ni or Re site, accompanied
by elongation of the C1AO bond along the reaction coordinate. At
the same time, C1 moves toward the hollow site and the phenyl
ring rotates slightly. At the transition state on Ni(111), the dis-
tance of C1AO increases to 2.09 Å with OH adsorption on Ni (OANi
distance of 2.22 Å) (Fig. 15A). In contrast, the transition state on the
(Re)Ni(111) surface has a significantly shorter C1AO distance of
1.90 Å, attributable to a stronger O–Re interaction (O–Re distance
of 2.17 Å) (Fig. 15B). In the final states (Fig. S4), the OH group
moved further away from the phenyl ring to an adjacent Ni–Ni
(C1AO distance 3.14 Å) or Ni–Re bridge site (C1AO distance
2.92 Å), with C1 in the hollow site. This elementary step is
endothermic (reaction energy of 54.0 kJ/mol) with a high activa-
tion energy of 175.6 kJ/mol on Ni(111). In contrast, the same step
becomes slightly exothermic (reaction energy of ꢁ5.8 kJ/mol) with
a significantly reduced activation barrier of 94.6 kJ/mol on (Re)Ni
(111). These results strongly indicate that the presence of the oxo-
philic Re next to Ni enhances the CAO cleavage by significantly
reducing both the reaction energy and the activation barrier for
direct deoxygenation, making this reaction pathway accessible,
and even favorable under mild conditions.
Recent experimental and computational studies [7,14,70] sug-
gest that HDO of m-cresol on Pt, particularly Pt supported on the
reducible TiO2 support, follows a tautomerization pathway to pro-
duce toluene. The current results do not allow us to exclude the
To demonstrate the importance of Ni–Re neighboring sites for
the HDO reaction, we examined the direct dehydroxylation of phe-
nol on both Ni(111) and (Re)Ni(111) surfaces. As shown in Fig. 10,

96
F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
contribution of partial hydrogenation and tautomerization mecha-
nisms to the apparent direct deoxygenation. In fact, the enhanced
direct CAO cleavage (dehydroxylation) on the Ni–Re neighboring
sites could also be an integral part of the CAO cleavage of the inter-
mediate by a partial hydrogenation or tautomerization pathway to
aromatics.
Even though Re plays an important role in adsorption of O
atoms of phenolics and facilitates CAO cleavage, the activities of
monometallic 2.5Re and 2.5ReOx were rather low (Table 4). This
may be due to the strong oxophilicity of Re, which leads to the
accumulation of surface hydroxyl species from dehydroxylation,
which eventually blocks the active sites for deoxygenation. When
Ni is present next to Re, the surface hydroxyl on Re could be hydro-
genated to H2O on the Ni–Re neighboring sites by H atoms from Ni,
resulting in the regeneration of active sites. Under working condi-
tions, Ni–ReOx with Re at a low degree of oxidation may also serve
as an active site for deoxygenation.
Nickel catalysts have been shown to be highly active for
hydrogenolysis of (cyclo)hydrocarbons directly to methane [71],
and the reaction rate of hydrogenolysis of benzene increases with
the Ni particle size [72], i.e., for large ensembles of close packed
facets. Theoretical calculations suggest that the hydrogenolysis of
CAC bonds requires complete dehydrogenation of two C atoms
and formation of three CANi bonds for each C atom on two hollow
sites prior to CAC cleavage [73,74]. That is, at least 4–6 neighboring
Ni atoms (ensemble size 4–6) are required in the transition state
for CAC cleavage. On monometallic 5Ni, large particles provide
needed Ni ensembles for breaking the CAC bonds of ring and the
successive hydrogenolysis of the terminal CAC bonds to form
CH4 [71]. On bimetallic Ni–Re, both TEM and FTIR of CO adsorption
indicate that the proximity of Ni and Re breaks the Ni surface into
small ensembles that do not favor the formation of the transition
state, leading to the CAC bond breaking. Furthermore, the elec-
tronic effect would reduce the adsorption strength of deoxygena-
tion products (Tol, Ben) on the surface, resulting in desorption of
the products and preventing the subsequent degradation of the
products through successive CAC hydrogenolysis. Consequently,
the hydrogenolysis was greatly inhibited on the bimetallic catalyst.
The TOFs for hydrogenation of m-cresol to MCHone and MCHol
were slightly increased (Fig. 14), while their selectivities were
slightly reduced (Fig. 13). This might be a consequence of the
strong promotional effect of Re for deoxygenation, making deoxy-
genation dominate hydrogenation, and resulting in slightly
reduced selectivity toward hydrogenation.
electronic effects: (1) Re breaks the Ni surface into small ensem-
bles and provides Ni–Re neighboring sites, and (2) the proximity
between Ni and Re reduce the Ni d-band occupancy. In contrast
to phenol adsorption on the bare Ni(111) surface through the phe-
nyl ring with O pointing away from the surface, the adsorption of
phenol on (Re)Ni(111) through the phenyl ring on Ni and O on
Re on the Ni–Re neighboring sites facilitates CAO bond cleavage.
The Ni ensembles with a reduced size inhibit the formation of
the structure that is favored by the transition state of CAC
hydrogenolysis. In addition, electron deficiency in the Ni d band
weakens the adsorption of phenyl rings on the surface and there-
fore prevents the CAC hydrogenolysis of aromatic products.
Acknowledgments
The authors thank the National Natural Science Foundation of
China (21676194 and 21373148) and the Ministry of Education
of China for the Program of New Century Excellent Talents in
University (NCET-12-0407) for their support.
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.2017.01.001.
References
[1] J. Zakzeski, P.C.A. Bruijnincx, A.L. Jongerius, B.M. Weckhuysen, The catalytic
valorization of lignin for the production of renewable chemicals, Chem. Rev.
110 (2010) 3552–3599.
[2] M. Saidi, F. Samimi, D. Karimipourfard, T. Nimmanwudipong, B.C. Gates, M.R.
Rahimpour,
Upgrading
of
lignin-derived
bio-oils
by
catalytic
hydrodeoxygenation, Energy Environ. Sci. 7 (2014) 103–129.
[3] X. Zhu, L.L. Lobban, R.G. Mallinson, D.E. Resasco, Bifunctional transalkylation
and hydrodeoxygenation of anisole over a Pt/HBeta catalyst, J. Catal. 281
(2011) 21–29.
[4] H. Weigold, Behaviour of Co-Mo-Al2O3 catalysts in the hydrodeoxygenation of
phenols, Fuel 61 (1982) 1021–1026.
[5] Y. Romero, F. Richard, S. Brunet, Hydrodeoxygenation of 2-ethylphenol as a
model compound of bio-crude over sulfided Mo-based catalysts: Promoting
effect and reaction mechanism, Appl. Catal. B 98 (2010) 213–223.
[6] E.J. Shin, M.A. Keane, Gas-phase hydrogenation/hydrogenolysis of phenol over
supported nickel catalysts, Ind. Eng. Chem. Res. 39 (2010) 883–892.
[7] L. Nie, D.E. Resasco, Kinetics and mechanism of m-cresol hydrodeoxygenation
on a Pt/SiO2 catalyst, J. Catal. 317 (2014) 22–29.
[8] X. Zhu, R.G. Mallinson, D.E. Resasco, Role of transalkylation reactions in the
conversion of anisole over HZSM-5, Appl. Catal. A 379 (2010) 172–181.
[9] M. Shetty, K. Murugappan, T. Prasomsri, W.H. Green, Y.R. Leshkov, Reactivity
and stability investigation of supported molybdenum oxide catalysts for the
hydrodeoxygenation (HDO) of m-cresol, J. Catal. 331 (2015) 86–97.
[10] J.S. Moon, E.G. Kim, Y.K. Lee, Active sites of Ni2P/SiO2 catalyst for
hydrodeoxygenation of guaiacol: A joint XAFS and DFT study, J. Catal. 311
(2014) 144–152.
We note that the promotional effect of Re is more pronounced
at Re loadings of <5%. Re would interact directly with Ni at these
low Re loadings. When Re loading is increased beyond 5%, addi-
tional Re may be isolated from Ni and disperse directly on silica.
In the latter case, Re will have little promotional effect on Ni.
[11] W.S. Lee, Z. Wang, R.J. Wu, A. Bhan, Selective vapor-phase hydrodeoxygenation
of anisole to benzene on molybdenum carbide catalysts, J. Catal. 319 (2014)
44–53.
[12] R. Olcese, M.M. Bettahar, B. Malaman, J. Ghanbaja, L. Tibavizco, D. Petitjean, A.
Dufour, Gas-phase hydrodeoxygenation of guaiacol over iron-based catalysts.
Effect of gases composition, iron load and supports (silica and activated
carbon), Appl. Catal. B 129 (2013) 528–538.
4. Conclusions
[13] C. Chen, G. Chen, F. Yang, H. Wang, J. Han, Q. Ge, X. Zhu, Vapor phase
hydrodeoxygenation and hydrogenation of m-cresol on silica supported Ni, Pd
and Pt catalysts, Chem. Eng. Sci. 135 (2015) 145–154.
[14] Q. Tan, G. Wang, L. Nie, A. Dinse, C. Buda, J. Shabaker, D.E. Resasco, Different
product distributions and mechanistic aspects of the hydrodeoxygenation of
m-Cresol over platinum and ruthenium catalysts, ACS Catal. 5 (2015) 6271–
6283.
[15] E.J. Shin, M.A. Keane, Catalytic hydrogen treatment of aromatic alcohols, J.
Catal. 173 (1996) 450–459.
[16] T.M. Huynh, U. Armbruster, M.M. Pohl, M. Schneider, J. Radnik, D.L. Hoang, B.
M.Q. Phan, D.A. Nguyen, A. Martin, Hydrodeoxygenation of phenol as a model
compound for bio-oil on non-noble bimetallic nickel-based catalysts,
ChemCatChem 6 (2014) 1940–1951.
Monometallic 5%Ni/SiO2 was unselective in catalyzing the pri-
mary reactions of deoxygenation to toluene, hydrogenation to 3-
methylcyclohexanone and 3-methylcyclohexanol, and CAC
hydrogenolysis to CH4. Bimetallic 5%Ni–2.5%Re/SiO2 improved
the intrinsic reaction rate for hydrodeoxygenation of m-cresol by
a factor of 6, and its TOF improved by a factor of 2. The bimetallic
Ni–Re catalyst was highly selective for deoxygenation to toluene,
with the TOF for deoxygenation increased by a factor of 4, while
that for CAC hydrogenolysis is reduced to half.
The addition of Re to the Ni/SiO2 catalyst stabilized the highly
dispersed NiO through the strong NiAOARe interactions. Upon
reduction, Re penetrated into the unit cell of Ni, forming surface
Ni–Re alloy. The bimetallic catalysts exhibit both geometric and
[17] A. Robinson, G.A. Ferguson, J.R. Gallagher, S. Cheah, G.T. Beckham, J.A. Schaidle,
J.E. Hensley, J.W. Medlin, Enhanced hydrodeoxygenation of m-Cresol over
bimetallic Pt–Mo catalysts through an oxophilic metal-induced
tautomerization pathway, ACS Catal. 6 (2016) 4356–4368.

F. Yang et al. / Journal of Catalysis 349 (2017) 84–97
97
[18] P.T.M. Do, A.J. Foster, J.G. Chen, R.F. Lobo, Bimetallic effects in the
hydrodeoxygenation of meta-cresol on -Al2O3 supported Pt–Ni and Pt–Co
[47] E.L. Kunkes, D.A. Simonetti, J.A. Dumesic, W.D. Pyrz, L.E. Murillo, J.G. Chen, D.J.
Buttrey, The role of rhenium in the conversion of glycerol to synthesis gas over
carbon supported platinum–rhenium catalysts, J. Catal. 260 (2008) 164–177.
[48] J.A. Anderson, F.K. Chong, C.H. Rochester, IR study of CO adsorption on Pt, Re
and Pt–Re/Al2O3 catalysts before and after coking, J. Mol. Catal. A 140 (1999)
65–80.
c
catalysts, Green Chem. 14 (2012) 1388–1397.
[19] D. Shi, L.A. Ramírez, J.M. Vohs, The use of bimetallics to control the selectivity
for the upgrading of lignin-derived oxygenates: Reaction of anisole on Pt and
PtZn catalysts, J. Catal. 340 (2016) 219–226.
[20] J. Sun, A.M. Karim, H. Zhang, L. Kovarik, X.S. Li, A.J. Hensley, J.S. McEwen, Y.
Wang, Carbon-supported bimetallic Pd–Fe catalysts for vapor-phase
hydrodeoxygenation of guaiacol, J. Catal. 306 (2013) 47–57.
[21] L. Nie, P.M. de Souza, F.B. Noronha, W. An, T. Sooknoi, D.E. Resasco, Selective
conversion of m-cresol to toluene over bimetallic Ni–Fe catalysts, J. Mol. Catal.
A 388–389 (2014) 47–55.
[22] Y. Hong, H. Zhang, J. Sun, K.M. Ayman, A.J.R. Hensley, M. Gu, M.H. Engelhard, J.
S. McEwen, Y. Wang, Synergistic catalysis between Pd and Fe in gas phase
hydrodeoxygenation of m-Cresol, ACS Catal. 4 (2014) 3335–3345.
[23] S. Ted Oyama, T. Onkawa, A. Takagaki, R. Kikuchi, S. Hosokai, Y. Suzuki, K.K.
Bando, Production of phenol and cresol from guaiacol on nickel phosphide
catalysts supported on acidic supports, Top. Catal. 58 (2015) 201–210.
[24] X. Zhu, L. Nie, L.L. Lobban, R.G. Mallinson, D.E. Resasco, Efficient conversion of
m-Cresol to aromatics on a bifunctional Pt/HBeta catalyst, Energy Fuels 28
(2014) 4104–4111.
[25] G. Kresse, J. Hafner, Efficiency of ab-initio total energy calculations for metals
and semiconductors using a plane-wave basis set, Phys. Rev. B Condens.
Matter 47 (1993) 558–561.
[26] G. Kresse, J. Furthmüller, Efficiency of ab-initio total energy calculations for
metals and semiconductors using a plane-wave basis, Comput. Mater. Sci. 6
(1996) 15–50.
[27] G. Kresse, Efficient iterative schemes for ab initio total-energy calculation
using a plane-wave basis set, Phys. Rev. B 54 (1996) 11169–11186.
[28] P.E. Blöchl, Projector augmented-wave method, Phys. Rev. B 50 (1994) 17953–
17979.
[29] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made
simple, Phys. Rev. Lett. 77 (1996) 3865–3868.
[49] Y. Sato, Y. Soma, T. Miyao, S. Naito, The water-gas-shift reaction over Ir/TiO2
and Ir–Re/TiO2 catalysts, Appl. Catal. A 304 (2006) 78–85.
[50] K. Chayakul, T. Srithanratana, S. Hengrasmee, Catalytic activities of Re–Ni/CeO2
bimetallic catalysts for water gas shift reaction, Catal. Today 175 (2011) 420–
429.
[51] C. Zhao, S. Kasakov, J. He, J.A. Lercher, Comparison of kinetics, activity and
stability of Ni/HZSM-5 and Ni/Al2O3-HZSM-5 for phenol hydrodeoxygenation,
J. Catal. 296 (2012) 12–23.
[52] X. Zhu, L.L. Lobban, R.G. Mallinson, D.E. Resasco, Tailoring the mesopore
structure of HZSM-5 to control product distribution in the conversion of
propanal, J. Catal. 271 (2010) 88–98.
[53] T.J.G. Kofke, R.J. Gorte, G.T. Kokotailo, W.E. Farneth, Stoichiometric adsorption
complexes in H-ZSM-5, H-ZSM-12, and H-mordenite zeolites, J. Catal. 115
(1989) 265–272.
[54] K.M. Jiang, T. Kawai, T. Ishikawa, FT-IR and TPD studies of adsorbed pyridine on
Re2O7-Al2O3 catalysts, J. Catal. 159 (1996) 288–295.
[55] D. Hibbitts, Q. Tan, M. Neurock, Acid strength and bifunctional catalytic
behavior of alloys comprised of noble metals and oxophilic metal promoters, J.
Catal. 315 (2014) 48–58.
[56] L.D. Site, A. Alavi, C.F. Abrams, Adsorption energies and geometries of phenol
on the (111) surface of nickel: An ab initio study, Phys. Rev. B 67 (2003).
[57] M.L. Honkela, J. Bjork, M. Persson, Computational study of the adsorption and
dissociation of phenol on Pt and Rh surfaces, Phys. Chem. Chem. Phys. 14
(2012) 5849–5854.
[58] Y. Yoon, R. Rousseau, R.S. Weber, D. Mei, J.A. Lercher, First-principles study of
phenol hydrogenation on Pt and Ni catalysts in aqueous phase, J. Am. Chem.
Soc. 136 (2014) 10287–10298.
[30] G. Li, J. Han, H. Wang, X. Zhu, Q. Ge, Role of dissociation of phenol in its
selective hydrogenation on Pt(111) and Pd(111), ACS Catal. 5 (2015) 2009–
2016.
[31] Q. Sun, G. Chen, H. Wang, X. Liu, J. Han, Q. Ge, X. Zhu, Insights into the major
reaction pathways of vapor-phase hydrodeoxygenation of m-Cresol on a Pt/
HBeta catalyst, ChemCatChem 8 (2016) 551–561.
[59] H. Ihm, J.M. White, Stepwise dissociation of thermally activated phenol on Pt
(111), J. Phys. Chem. B 104 (2000) 6202–6211.
[60] A.J.R. Hensley, Y. Wang, J.S. McEwen, Adsorption of phenol on Fe (110) and Pd
(111) from first principles, Surf. Sci. 630 (2014) 244–253.
[61] K. Leiva, N. Martinez, C. Sepulveda, R. García, C.A. Jiménez, D. Laurenti, M.
Vrinat, C. Geantet, J.L.G. Fierro, I.T. Ghampson, N. Escalona,
Hydrodeoxygenation of 2-methoxyphenol over different Re active phases
supported on SiO2 catalysts, Appl. Catal. A 490 (2015) 71–79.
[62] O. Levy, M. Jahnátek, R.V. Chepulskii, G.L.W. Hart, S. Curtarolo, Ordered
structure in rhenium binary alloys from first-principles calculations, J. Am.
Chem. Soc. 133 (2010) 158–162.
[63] Y. Amada, Y. Shinmi, S. Koso, T. Kubota, Y. Nakagawa, K. Tomishige, Reaction
mechanism of the glycerol hydrogenolysis to 1,3-propanediol over Ir–ReOx/
SiO2 catalyst, Appl. Catal. B 105 (2011) 117–127.
[32] B. Feng, H. Kobayashi, H. Ohta, A. Fukuoka, Aqueous-phase
hydrodeoxygenation of 4-propylphenol as
a
lignin model to n-
propylbenzene over Re-Ni/ZrO2 catalysts, J. Mol. Catal. A 388–389 (2014)
41–46.
[33] L. Wang, K. Murata, Y. Matsumura, M. Inaba, Lower-temperature catalytic
performance of bimetallic Ni-Re/Al2O3 catalyst for gasoline reforming to
produce hydrogen wiht the inhibition of methaen formation, Energy Fuels 20
(2006) 1377–1381.
[34] S. Sitthisa, W. An, D.E. Resasco, Selective conversion of furfural to methylfuran
over silica-supported Ni-Fe bimetallic catalysts, J. Catal. 284 (2011) 90–101.
[35] E.L. Lee, I.E. Wachs, In situ spectroscopic investigation of the molecular and
electronic structures of SiO2 supported surface metal oxides, J. Phys. Chem. C
111 (2007) 14410–14425.
[36] A. Aminzadeh, H.S. Fard, Raman spectroscopic study of Ni/Al2O3 catalyst,
Spectrochim. Acta A 55 (1999) 1421–1425.
[37] D.S. Kim, I.E. Wachs, Surface rhenium oxide-support interaction for supported
Re2O7 catalysts, J. Catal. 141 (1993) 419–429.
[64] M. Chia, Y.J.P. Torres, D. Hibbitts, Q. Tan, H.N. Pham, A.K. Datye, M. Neurock, R.J.
Davis, J.A. Dumesic, Selective hydrogenolysis of polyols and cyclic ethers over
bifunctional surface sites on rhodium-rhenium catalysts, J. Am. Chem. Soc. 133
(2011) 12675–12689.
[65] F.E. Massoth, P. Politzer, M.C. Concha, J.S. Murray, J. Jakowski, J. Simons,
Catalytic hydrodeoxygenation of methyl-substituted phenols: correlations of
kinetic parameters with molecular properties, J. Phys. Chem. B 110 (2006)
14283–14291.
[66] G.H. Gu, C.A. Mullen, A.A. Boateng, D.G. Vlachos, Mechanism of dehydration of
phenols on noble metals via first-principles microkinetic modeling, ACS Catal.
6 (2016) 3047–3055.
[67] C. Chiu, A. Genest, A. Borgna, N. Rösch, Hydrodeoxygenation of guaiacol over
Ru(0001): A DFT Study, ACS Catal. 4 (2014) 4178–4188.
[68] J. Lu, A. Heyden, Theoretical investigation of the reaction mechanism of the
hydrodeoxygenation of guaiacol over a Ru(0001) model surface, J. Catal. 321
(2015) 39–50.
[69] R.C. Nelson, B. Baek, P. Ruiz, B. Goundie, A. Brooks, M.C. Wheeler, B.G.
Frederick, L.C. Grabow, R.N. Austin, Experimental and theoretical insights into
the hydrogen-efficient direct hydrodeoxygenation mechanism of phenol over
Ru/TiO2, ACS Catal. 5 (2015) 6509–6523.
[70] M.B. Griffin, G.A. Ferguson, D.A. Ruddy, M.J. Biddy, G.T. Beckham, J.A. Schaidle,
Role of the support and reaction conditions on the vapor-phase deoxygenation
of m-cresol over Pt/C and Pt/TiO2 catalysts, ACS Catal. 6 (2016) 2715–2727.
[71] G. Leclercq, S. Pietrzyk, M. Peyrovi, M. Karroua, Hydrogenolysis of saturated
hydrocarbons: V. Influence of hydrocarbon structures on the activity and
selectivity of Ni on silica, J. Catal. 99 (1986) 1–11.
[38] M.A. Vuurman, D.J. Stufkens, A. Oskam, Structural determination of surface
rhenium oxide on various oxide supports (Al2O3, ZrO2, TiO2 and SiO2), J. Mol.
Catal. A 76 (1992) 263–285.
[39] Y. Amada, H. Watanabe, M. Tamura, Y. Nakagawa, K. Okumura, K. Tomishige,
Structure of ReOx clusters attached on the Ir metal surface in Ir–ReOx/SiO2 for
the hydrogenolysis reaction, J. Phys. Chem. C 116 (2012) 23503–23514.
[40] S. Engels, J. Eick, W. Morke, U. Maier, I. Boszormenyi, K. Matusek, Z. Schay, L.
Guczi, Characterization and catalytic activity of unsupported nickel-rhenium
catalysts modified by calcium oxide, J. Catal. 103 (1987) 105–114.
[41] G. Beamson, A.J. Papworth, C. Philipps, A.M. Smith, R. Whyman, Selective
hydrogenation of amides using bimetallic Ru/Re and Rh/Re catalysts, J. Catal.
278 (2011) 228–238.
[42] J.A. Rodriguez, Physical and chemical properties of bimetallic surfaces, Surf.
Sci. Rep. 24 (1996) 223–287.
[43] S. Sitthisa, T. Pham, T. Prasomsri, T. Sooknoi, R.G. Mallinson, D.E. Resasco,
Conversion of furfural and 2-methylpentanal on Pd/SiO2 and Pd–Cu/SiO2
catalysts, J. Catal. 280 (2011) 17–27.
[44] X. Zhu, Y. Zhang, C. Liu, CO adsorbed infrared spectroscopy study of Ni/Al2O3
catalyst for CO2 reforming of methane, Catal. Lett. 118 (2007) 306–312.
[45] D.G. Blackmond, E.I. Ko, Structural sensitivity of CO adsorption and H2 CO
coadsorption on NiSiO2 catalysts, J. Catal. 96 (1985) 210–221.
[46] M.S. Nacheff, L.S. Kraus, M. Ichikawa, B.M. Hoffman, J.B. Butt, W.M.H. Sachtler,
Characterization and catalytic function of Re0 and Re4+ in Re/Al2O3 and PtRe/
Al2O3 catalysts, J. Catal. 106 (1987) 263–272.
[72] P.H. Desai, J.T. Richardson, Crystallite size effects in Nickel catalysts:
cyclohexane dehydrogenation and hydrogenolysis, J. Catal. 98 (1986) 392–
400.
[73] D.D. Hibbitts, D.W. Flaherty, E. Iglesia, Role of branching on the rate and
mechanism of C-C cleavage in alkanes on metal surfaces, ACS Catal. 6 (2016)
469–482.
[74] Z. Zhao, L.V. Moskaleva, N. Rösch, Tuning the selectivity for ring-opening
reactions of methylcyclopentane over Pt catalysts: A mechanistic study from
first-principles calculations, J. Catal. 285 (2012) 124–133.