Product tunable behavior of carbon nanotubes-supported Ni–Fe catalysts for guaiacol hydrodeoxygenation

Article abstract of DOI:10.1016/j.apcata.2016.10.011

Bimetallic Ni–Fe nanoparticles supported on carbon nanotubes (CNTs) are prepared and evaluated for the catalytic hydrodeoxygenation (HDO) of a lignin-derived model compound guaiacol. Appropriate combination of Ni and Fe affords high activity and significantly enhances selectivity to cyclohexane or phenol, whereas monometallic Ni and Fe catalysts display poor activities or selectivities. The product tunable behavior of guaiacol HDO is found to be dependent on Ni/Fe atomic ratios. Cyclohexane and phenol are the major products over Ni₅–Fe₁/CNT with Ni/Fe atomic ratio at 5/1 and Ni₁–Fe₅/CNT with Ni/Fe atomic ratio at 1/5, respectively. Characterization results confirm that Ni–Fe alloys are formed and elicit synergistic effects on the HDO performance. The selectivity-switchable performance of Ni–Fe/CNT can be assigned to the synergism between Ni domains, where H₂ can be easily activated, and Fe domains, which exhibited strong oxophilicity. The bimetallic catalysts give an enhanced stability without significant sintering of metal nanoparticles, while the monometallic catalysts show obvious deactivation due to the agglomeration of metal nanoparticles. Further results reveal that the conversion of guaiacol depends on not only the chemical state but also the size of the metallic nanoparticles. The catalysts with appropriate Ni/Fe atomic ratio and smaller particle perform better hydrogenolysis of C–O bonds, resulting in high selectivity to cyclohexane or phenol.

Full text of DOI:10.1016/j.apcata.2016.10.011

Accepted Manuscript
Title: Product tunable behavior of carbon nanotubes-supported
Ni−Fe catalysts for guaiacol hydrodeoxygenation
Author: Huihuang Fang Jianwei Zheng Xiaolin Luo Junmou
Du Alberto Roldan Stefano Leoni Youzhu Yuan
PII:
S0926-860X(16)30503-8
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.apcata.2016.10.011
APCATA 16027
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
16-6-2016
6-10-2016
12-10-2016
Please cite this article as: Huihuang Fang, Jianwei Zheng, Xiaolin Luo, Junmou Du,
Alberto Roldan, Stefano Leoni, Youzhu Yuan, Product tunable behavior of carbon
nanotubes-supported Ni−Fe catalysts for guaiacol hydrodeoxygenation, Applied
Catalysis A, General http://dx.doi.org/10.1016/j.apcata.2016.10.011
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Product tunable behavior of carbon nanotubes-supported
NiFe catalysts for guaiacol hydrodeoxygenation
Huihuang Fang1, Jianwei Zheng1, Xiaolin Luo1 Junmou Du1, Alberto Roldan2, Stefano
Leoni2*, Youzhu Yuan1*
1 State Key Laboratory of Physical Chemistry of Solid Surfaces, National Engineering
Laboratory for Green Chemical Productions of Alcohols-Ethers-Esters and iChEM,
College of Chemistry and Chemical Engineering, Xiamen University, Xiamen, 361005,
China
2 School of Chemistry, Cardiff University, Main Building, Park Place, Cardiff, CF10 3AT,
United Kingdom
*
To whom correspondence should be addressed. E-mail: yzyuan@xmu.edu.cn,
LeoniS@cardiff.ac.uk
1

Graphical abstract
Highlights:
 Non-noble Ni-Fe bimetallic catalysts are preparation by varying Ni/Fe ratios.
 Ni-Fe/CNT shows higher performance for guaiacol HDO than mono-metal catalysts.
 High selectivity to cyclohexane or phenol is tunable by varying the Ni/Fe ratios.

The enhanced performance is ascribed to the synergism Ni and Fe in Ni-Fe alloy.
Abstract
Bimetallic NiFe nanoparticles supported on carbon nanotubes (CNTs) are prepared and
evaluated for the catalytic hydrodeoxygenation (HDO) of a lignin-derived model
compound guaiacol. Appropriate combination of Ni and Fe affords high activity and
significantly enhances selectivity to cyclohexane or phenol, whereas monometallic Ni and
Fe catalysts display poor activities or selectivities. The product tunable behavior of
guaiacol HDO is found to be dependent on Ni/Fe atomic ratios. Cyclohexane and phenol
2

are the major products over Ni5Fe1/CNT with Ni/Fe atomic ratio at 5/1 and Ni1Fe5/CNT
with Ni/Fe atomic ratio at 1/5, respectively. Characterization results confirm that NiFe
alloys are formed and elicit synergistic effects on the HDO performance. The
selectivity-switchable performance of NiFe/CNT can be assigned to the synergism
between Ni domains, where H2 can be easily activated, and Fe domains, which exhibited
strong oxophilicity. The bimetallic catalysts give an enhanced stability without significant
sintering of metal nanoparticles, while the monometallic catalysts show obvious
deactivation due to the agglomeration of metal nanoparticles. Further results reveal that
the conversion of guaiacol depends on not only the chemical state but also the size of the
metallic nanoparticles. The catalysts with appropriate Ni/Fe atomic ratio and smaller
particle perform better hydrogenolysis of CO bonds, resulting in high selectivity to
cyclohexane or phenol.
Keywords: NiFe catalyst; hydrodeoxygenation; guaiacol; cyclohexane; phenol
1. Introduction
Renewable feedstocks, such as lignocellulose, have gained considerable attention for
sustainable production of fuel and chemicals because of the limited availability and
passive effect of fossil fuels [15]. Lignin, an amorphous tri-dimensional biopolymer
containing abundant methoxylated propyl phenol units, is the second most important
compound in lignocellulosic biomass (15%20%) [6]. In contrast to cellulose and
hemicellulose, lignin is the most similar to petroleum in structure and composition, which
3

is industrially available from biomass-to-ethanol processes and other biorefineries [7].
Generally, lignocellulosic biomass is directly converted into aromatics and bio-oils
through a traditional method known as rapid pyrolysis [3]. However, the methoxylated
propyl phenol unit contains numerous CarylOCH3, CarylOH, and CarylOR bonds; as such,
the pyrolysis product is comprised high amounts of oxygen and oxo-functionalized groups
than commercial fuel and thus presents limited applications [8]. In addition, lignin-derived
phenolic compounds are highly reactive during fast pyrolysis, which leads to the
formation of oligomers and deactivation of catalysts [9,10]. Therefore, upgrading of
bio-oils through hydrodeoxygenation (HDO) with hydrogen consumption is valuable [11].
Guaiacol, a model compound of lignin-derived bio-oils, contains two types of
O-containing functional groups (CarylOCH3 and CarylOH), which are characteristic
components of lignin-derived compounds [12,13]. Hence, the catalytic hydrogenolysis and
HDO of guaiacol have gained increased research attention to elucidate the roles of
catalysts and identify their industrial applications.
Metal sulfides, such as MoS2, NiMoS, and CoMoS, as well as supported noble metals,
such as Ru, Pt, and Pd which are widely used in hydrodesulfurization, are also applied in
HDO because of their similarity to S and O [14,15]. Metal sulfides exhibit satisfactory
activities but deactivate rapidly, especially during the formation of water as the byproduct
[2,6,16]. The activated sulfide form of catalysts is maintained through continuous addition
of S to the reactant stream, causing serious negative effects on downstream processes
[17,18]. Using noble metals for HDO of guaiacol could avoid these problems
[11,13,19,20]. Gates et al. [21,22] used Pt catalysts for HDO of lignin-derived species and
4

reported that hydrogenolysis of CO bonds is the most kinetically available route that
does not remove oxygen. Varma et al. [23] systematically surveyed the HDO of guaiacol
on noble metal catalysts; the developed Pt/C catalysts showed superior activity at the
optimal operating temperature of 573 K. Supported noble metal catalysts demonstrate
satisfactory activity but have limited applications because of the high cost of noble metals.
Hence, novel metal catalysts such as non-noble metallic catalysts should be developed for
economic viability and engineering aspects.
Ni is cheaper than noble metals and shows high activity in hydrogenation reactions;
as such, this element is widely used as a metallic catalyst [2428]. Zhao et al. [27]
reported that Ni/HZSM exerts synergistic effects on HDO and CO bond cleavage of
substituted phenols. Ma et al. [24] investigated Ni-based catalysts supported by mixed
oxides and found that Ni/TiO2ZrO2 displayed the optimal performance, with a
cyclohexane yield of 86.4% in the solvent decalin at 573 K and 4.0 MPa H2. Recently,
Dongil et al. [29] obtained tunable selectivities of cyclohexane and cyclohexene over
CNTs supported Ni nanoparticles inside and outside. On the other hand, several reports
also indicated that Fe was active in selective cleavage of aryl CO bonds [25,26]. Olcese
et al. [26] found that the Fe/SiO2 catalyst was efficient for the guaiacol HDO. However,
these mono-metallic catalysts exhibit inherent low activity and poor stability. In this
regard, a second metal is added to form bimetallic nanoparticles, such as NiCu, PtSn
and PdFe, which can remarkably promote catalytic properties as a result of geometric
and electronic effects [28,30]. Ardiyanti et al. [28] found that bimetallic catalysts
promoted higher HDO activity than the mono-metallic Ni and Cu catalysts for
5

hydrotreatment of anisole over NiCu catalysts; Sun et al. [30] reported that PdFe/C
bimetallic catalyst benefit the HDO of guaiacol to benzene with 83.2% yield at 723 K.
In this study, bimetallic NiFe catalysts supported on carbon nanotubes (CNTs) were
prepared and firstly applied to investigate the relationship between the bimetallic structure
and the HDO performance of guaiacol, particularly selectivity. In all cases, CNTs are used
as carriers due to their special electrical properties, hydrogen spillover, and chemical
stability, which have triggered wide research interest [31,32]. Product distribution was
correlated with the amounts of Ni and Fe. The synthesized catalysts displayed tunable
selectivity to cyclohexane or phenol at Ni/Fe atomic ratio of 5/1 or 1/5, respectively.
Further
experiments
including
powder
X-ray
diffraction
(XRD),
H2-temperature-programmed reduction (H2-TPR), transmission electron microscopy
(TEM), X-ray photoelectron spectroscopy (XPS), high-sensitivity low-energy ion
scattering spectroscopy (HS-LEIS), Raman spectroscopy, and computational modelling
were performed to elucidate the interaction between Ni and Fe. The results revealed the
key role of different metals on the HDO pathway and confirmed the synergistic effect of
NiFe alloy nanoparticles, in which the Ni domains demonstrated high ability to activate
H2 and the Fe domain displayed strong oxophilicity. Meanwhile, deactivation of
monometallic catalysts was observed which may be ascribed to the agglomeration of
nanoparticles. Size effects of the catalysts were further investigated.
2. Experimental
2.1. Catalyst preparation
6

CNTs with 97% purity were purchased from Shenzhen Nanotech Port Co., Ltd. The
obtained CNTs have parallel multiwall structures with an outer diameter of 2040 nm.
Prior to use, the as-received CNTs were pretreated in concentrated HNO3 (68 wt. %) at
353 K for 16 h under reflux conditions to remove residual catalysts and amorphous
carbons. The treated CNTs were thoroughly washed with a large quantity of deionized
water until the pH became neutral and then dried overnight. All other carbon materials
before use were pretreated in the same process as above.
NiFe bimetallic catalysts were prepared through co-impregnation using aqueous
solutions of Ni(NO3)26H2O and Fe(NO3)39H2O. The total loading weight of metal was
maintained at 7 wt. %, and the Ni/Fe atomic ratio was accordingly adjusted. Typically, 1.0
g of CNTs was added to 30 mL of distilled water containing required amounts of
Ni(NO3)26H2O and Fe(NO3)39H2O with vigorous stirring for 6 h. Then, the aqueous
solution removed the excess solvent by evaporation to keep an incipient wetness
1
liquid/solid ratio of 1 mL g . After impregnation for 12 h at room temperature and solvent
removal by evaporation, the mixture was dried at 373 K overnight. All dry solids were
1
calcined for 4 h at 573 K with a heating ramp of 2 K min in the atmosphere. The
1
catalysts were reduced by 5% H2−95% N2 (25 mL min ) for 4 h prior to the reaction. The
bimetallic catalyst was denoted as NixFey/CNT, where x and y represent the atomic ratios
of Ni and Fe, respectively. Monometallic catalysts, namely, Ni/CNT and Fe/CNT, were
prepared using the same method.
2.2. Catalyst characterization
7

Ni and Fe contents of the catalysts were determined through inductively coupled
plasma optical emission spectrometry (ICP-OES) using a Thermo Elemental IRIS Intrepid
II XSP. The samples were pretreated with aqua regia at 353 K for 1 h, and the remaining
solution was heated to remove the solvent. The residues were diluted with 5% HNO3 and
filtered in a 50 mL volumetric flask before measurement.
Specific surface area, pore volume and average pore diameter were measured by
static N2 physisorption method at 77.3 K on a Micromeritics TriStar II 3020 surface area
and pore analyzer. The sample was degassed at 573 K for 3 h prior to the measurements.
Specific surface area was determined through Brunauer-Emmett-Teller (BET) method.
Pore size and average pore diameter distributions were calculated from the desorption
isotherm branch according to the Barret-Joyner-Halenda (BJH) method. The total pore
volume of the samples depended on the adsorbed N2 volume at a relative pressure of
approximately 0.99.
Powder XRD patterns were obtained on a Phillips PANalytical X’pert Pro
diffractometer equipped with a graphite monochromator and CuK radiation (40 kV and
30 mA) at scanning 2 from 10 to 90. The obtained diffraction data were identified and
compared with reference patterns in the JCPDS database.
H2-TPR spectra were determined using a Micromeritics AutoChem II 2920
1
chemisorption analyzer. The sample was pretreated in Ar (30 mL min ) at 573 K for 30
min and then cooled to room temperature. The reducing gas (5% H2−95% N2) was then
flowed through the sample at a rate of 30 mL/min, and the ambient temperature was raised
1
to 1173 K at a rate of 10 K min . A thermal conductivity detector was used to monitor H2
8

consumption.
TEM studies were performed using an electron microscope (Philips Analytical FEI
Tecnai 20) operated at an acceleration voltage of 300 kV. The samples for TEM analysis
were ultrasonically dispersed in ethanol. Drops of the suspension were deposited on
copper grids coated with C films.
XPS spectra were recorded on an Omicron Sphera II photoelectron spectrometer
equipped with an AlKα X-ray radiation source (h = 1486.6 eV). The machine is
connected to HS-LEIS, in which measurement was performed by IONTOF Qtac100 using
20Ne+ (5 keV) as Fe source. The scattering angle was 145. The sample was treated in an
in situ chamber with 5% H2−95% N2 at 673 K for 4 h before the measurement. All
measurements were performed in a system with combined elevated pressure reaction cell
and ultrahigh vacuum (UHV). Sample preparation and characterization were conducted in
9
the primary UHV chamber (base pressure, 7 × 10 Torr).
Raman spectra were obtained using a Renishaw inVia Raman microscope equipped
with an Ar ion laser at an excitation wavelength of 514 nm and a laser source intensity of
10%.
2.3. Catalytic testing
The catalytic reactions for HDO were performed on a conventional fixed-bed flow
reactor equipped with an auto-sampling system (Figure S1). Briefly, 100 mg of the
as-calcined catalyst was placed in the center of the reaction tube and sandwiched by quartz
powders. Prior to the experiment, the catalyst precursor was reduced under 5% H295%
9

1
N2 at 673 K for 240 min at a ramping rate of 3 K min . After cooling the catalyst bed to
the reaction temperature, pure H2 was fed into the reactor and held at 3.0 MPa. For the
HDO reaction, pure guaiacol was pumped into the reactor with varying weight liquid
hourly space velocity (WLHSVGUA) by using a Series III digital HPLC pump (Scientific
Systems, Inc.). Then, the introduced guaiacol reacted with H2 at the catalyst bed. The
product line was heated at 583 K before cold trap to avoid condensation of liquid products.
An Agilent 7890A gas chromatograph (GC), equipped with an auto-sampling valve,
DB-Wax (30 m  0.32 mm  0.25 m) capillary column, and flame ionization detector
(FID) was used to analyze the products. The mixture collected through condensation was
analyzed by an Agilent 7890GC-5975MS system. After the condensation of the products,
the dry gas was conveyed into a GC 2060 with a TDX column and thermal conductivity
detector for analysis of gas products (CO, CH4, CO2, and H2O). The response factors of
guaiacol and products were determined using pure compounds with known concentrations.
The carbon balance was about 95%  2%, unless otherwise noted. Guaiacol conversion
(XGUA), selectivity to C6 ring products (SC6i) and selectivity to C1 products (SC1i) were
calculated as follows:
(moles of GUA)in − (moles of GUA)out
XGUA
=
× 100%
Eq. 1
Eq. 2
Eq. 3
(moles of GUA)in
moles of ring product i
SC6i
=
× 100%
the sum moles of guaiacol consumed
moles of produced methane or methanol
the sum moles of guaiacol consumed
SC1i
=
× 100%
10

Unless otherwise mentioned, the reported conversion and selectivity were obtained after
the reaction proceeded for 2 h when the ready state was reached. SC6i was calculated based
on the number of C6 rings except for C1 products (CH4 and CH3OH); SC1i was calculated
based on the number of methoxy groups from the reacted substrate. In some cases, the
total selectivity to C1 products is less than 100% because of the formation of coke or trace
transalkylation reaction. The turnover frequency (TOF) was based on the metal dispersion,
indicating the moles of guaiacol converted by per atom at the catalyst surface per hour
1
1
1

(mol-guaiacol mol-metalsurf h , for short, h ). The guaiacol conversion for the TOF
calculation was lower than 40%.
2.4. Computational modelling
The catalyst surface was simulated with a periodic slab model of Ni exposing the
most stable surface (111). Each slab represents a p(3x3) cell with an area of 44.6478 Å2
and five atomic layers where the top two are fully relaxed while the bottom surface remain
fix at the optimized interatomic distance simulating the bulk. A vacuum of 15 Å avoids the
interactions between the slab and its images. As a doped model, we replaced three Ni from
the pristine surface with Fe. We studied two different configurations and found that an
alternating distribution of Fe atoms is 0.19 eV more stable than Fe agglomeration. To
obtain the properties of isolated molecules, we placed them in the center of a 19x20x21 Å3
simulation cell to avoid lateral interactions, with broken symmetry, and using the same
criteria of convergence as for the metallic surfaces.
All calculations were carried out with the density-functional theory-D2 (DFT-D2) as
11

implemented in the Vienna ab initio simulation package (VASP) [33,34]. The Kohn-Sham
valence states were expanded in a plane-wave basis set with a cut off at 450 eV for the
kinetic energy [35]. The electron exchange-correlation was denoted by the generalized
gradient approximation (GGA) with the Perdew-Burke-Ernzerhof revised for solids
(PBEsol) [36]. All the calculations include long-range dispersion correction Grimme
approach [37], which is an improvement on standard DFT [3840]. A Monkhorst-Pack
grid of 5x5x1 k-points was used to evaluate the wavefunctions which ensures the
electronic and ionic convergence [41]. The systems were considered converged when the
force on each ion dropped below 0.01 eV/Å and the energy threshold defining
5
self-consistency of the electron density was set to 10 eV.
We have calculated the energy variations (ΔE) along the reaction pathway according
to Eq. 4, ESystem being the energy of the slab and adsorbates, ESurf the energy of the naked
slab, EGua and n.EH2 the energy of the isolated guaiacol and H2 molecules.
EB  ESystem


ESurf  EGua  n EH

Eq. 4
2
3. Results and discussion
3.1. Catalyst characterization
A series of supported NiFe bimetallic catalysts were prepared with a fixed total
metal loading (7 wt. %) but varied Ni/Fe atomic ratios. The physicochemical properties of
the catalysts are summarized in Table 1. The ICP-OES results show that actual metal
loading is similar to the preset values in the preparation. The BET surface area of CNTs is
1
approximately 147 m2 g , and all surface areas of catalysts approximate those of CNTs
12

after loading with metals.
The XRD patterns of the samples after reduced under 5% H295% N2 at 673 K for 4
h are shown in Figures 1 and S2. Distinct diffraction peaks of the metal Ni species are
observed at 2θ = 44.50°, 51.60°, 76.08°, which correspond to the (111), (200), and (220)
reflections of the Ni phase (PDF 00-001-1258). No significant diffraction peaks of Fe0
species are observed, probably due to their highly dispersion on the catalyst surfaces. The
diffraction lines of the NiFe alloy are similar to those of monometallic Ni and Fe, which
exhibit similar face-centered cubic structures and lattice constants. The peak position
corresponding to (111) are shifted to lower angles (from 44.50° to 43.90°) with increasing
Fe content, indicating the formation of NiFe alloy. This finding is consistent with the
results previously reported [42,43]. In addition, diffraction peaks emerge at 2θ = 44.28°,
51.53°, 75.87° for bimetallic catalysts, which is ascribed to the diffractions of Fe1Ni3
alloy phase (PDF 00-038-0419). With increasing Fe content, the diffraction peaks
belonging to Fe3O4 (2θ = 30.06°, 35.45°; PDF 00-001-1111) appear because of incomplete
reduction of iron oxide at 673 K. The XRD patterns of samples after reduced at 773 K and
873 K are shown in Figures S3 and S4, respectively. The diffraction peak at 2θ = 44.64°
belonging to Fe0 species appears while the Fe3O4 phase disappears when all the samples
were reduced at 873 K, indicating that iron oxides can be completely reduced at this
temperature.
The average diameters of nanoparticles reduced at 673 K, 773 K, and 873 K based on
the Scherrer equation are summarized in Tables 1 and 2. Crystallite size decreases with
increase amount of Fe, suggesting that Fe promoted the dispersion of metallic Ni. Further,
13

the nanoparticle size increases with the increase of reduction temperatures. Figure S5-a–f
shows the TEM images and metallic particle size distribution of 673 K-reduced catalysts
containing varied Ni/Fe atomic ratios. The mean particle sizes of monometallic Ni/CNT
and Fe/CNT are 8.7 and 5.1 nm, respectively. The average particle sizes of bimetallic
particles are smaller than that of Ni/CNT and decreased (7.3, 6.9, 6.2, and 5.3 nm at a
Ni/Fe ratio of 5/1, 2/1, 1/1, and 1/5, respectively) with decreasing Ni/Fe atomic ratio. Both
the XRD and TEM results indicate that Fe species is beneficial for Ni dispersion.
The microstructure of Ni5Fe1/CNT was characterized using high-resolution TEM
(HRTEM) and scanning transmission electron microscope-energy dispersive X-ray
(STEM-EDX) elemental mapping and linear scanning pattern to observe distribution of Ni
and Fe in bimetallic catalysts (Figure 2). The representative HRTEM image (Figure 2-a)
shows a 0.205 nm distance between the corresponding lattice fringes, and this distance is
consistent with the lattice spacing of NiFe alloy [42]. The results of nanoscale elemental
STEM-EDX mapping and linear scanning analyses are shown in Figure 2-bc. Ni is well
distributed throughout the Fe domain, indicating that Ni and Fe are homogeneous
distributed along the bimetallic particles. Ni species display improved distribution than Fe
possibly because some Fe3O4 particles were not reduced completely. The linear scanning
analysis also confirms the mapping results. The distribution of Ni is similar to that of Fe in
an individual particle, and the formation of NiFe alloy in the Fe domain was verified.
The HRTEM image and STEM-EDX line scanning analysis of Ni1Fe5/CNT catalyst are
also verified the formation of NiFe alloys (Figure S6). These results are consistent with
the XRD patterns and the computational models.
14

In situ XPS and HS-LEIS analyses were performed to determine the structure and
electronic properties of Ni–Fe nanoparticles. The Ni 2p and Fe 2p spectra of Fe/CNT,
Ni/CNT, Ni2Fe1/CNT, and Ni5Fe1/CNT are displayed in Figure 3. Data on XPS curve
fitting are summarized in Table S1. All XPS peaks are identified according to published
data and an on-line database (www.lasurface.com). In contrast to those of bimetallic
NiFe/CNT samples, the Fe 2p spectra of Fe/CNT [Figure 3(A)-a] show a binding energy
of Fe 2p3/2 at 711.0 eV and that of Fe 2p1/2 at 724.0 eV, a typical characteristic of Fe3+. To
check the rationality of peak-fitting we executed the curve-fitting by fitting two peaks at
707.3 and 709.0 eV. The results became worse, indicating that only Fe3+ exists on the
surface of Fe/CNT sample. Hence, Fe species supported on CNTs are hardly reduced at
673 K within 4 h. Nevertheless, the results for bimetallic NiFe/CNT samples reveal that
the Fe 2p spectra evidently change [Figure 3(A)-bc]. Three peaks of the binding energy
of Fe 2p3/2 are identified at 707.3, 709.0, and 711.0 eV for Ni2Fe1/CNT and Ni5Fe1/CNT.
In addition, a binding energy of approximately 715.0 eV is ascribed to the satellite peak of
Fe2+ species, and is commonly observed in Fe2+ compounds. The original peaks at 707.3
and 709.0 eV are ascribed to the binding energy of Fe0 and Fe2+, respectively; this finding
indicates that some Fe3+ species are reduced into FeO and Fe0 in the presence of Ni
species. Thus, Ni addition promotes the reduction of iron oxides. On the other hand, the
binding energy of Ni 2p are observed at 853.1, 854.7, and 856.7 eV for monometallic
Ni/CNT [Figure 3(B)-a]; these peaks are attributed to Ni0, Ni2+, and Ni3+, respectively. The
results indicate that a certain proportion of nickel oxides still exist when the sample was
reduction at 673 K [44,45]. Figure 3(B)-bc shows the XPS results of Ni2Fe1/CNT, and
15

Ni5Fe1/CNT. In these cases, the significant intensity due to the contribution of Ni0
suggests that Ni oxides are more easily reduced than Fe ones under identical reduction
condition.
Ni and Fe concentrations were calculated by peak fitting (Table S1). Bimetallic
NiFe/CNT catalysts display higher concentrations of Fe0 and Ni0 than those in the
monometallic Fe/CNT or Ni/CNT catalysts. Values increase with increasing Fe content.
The higher metallic Ni concentration in bimetallic NiFe/CNT samples demonstrates the
improved reducibility than that of monometallic Ni/CNT. The introduction of Fe does not
inhibit but promote Ni reduction. As shown in Table 1, the Ni/Fe atomic ratios revealed by
the XPS results are lower than those in the ICP-OES results, demonstrating that Fe easily
segregates on the particle surface. Similarly, previous study [42] indicated that the
cohesive energy of Ni is slightly higher than that of Fe. In addition, the HS-LEIS results
show that Ni coexists with Fe on the outermost surface of NiFe nanoparticles even if Fe
will segregate on the surface (Figure S7). Therefore, it is presumable that the Fe0 species
interact with Ni0 by forming NiFe alloys and are located on the Ni domain, whereas the
iron oxides are doped on the NiFe nanoparticles. Table S2 shows the Ni/C and Fe/C
atomic surface ratios of NiFe/CNT catalysts from XPS peak fitting. The surface Ni/C
atomic ratio represents the relative Ni dispersion on the surface of CNTs. Despite the
lower loading of Ni, the surface Ni/C atomic ratio in Ni5Fe1/CNT is higher than that of
monometallic Ni/CNT catalyst (0.040 versus 0.021), which suggested that the introduction
of Fe increases the Ni dispersion on the surface of support. These findings are consistent
with the XRD and TEM results. Similar results were also observed in previous reports
16

[46,47].
Figure 4 shows the H2-TPR reduction profiles of the as-calcined catalyst precursors
containing different Ni/Fe ratios under a total metal loading of 7 wt. %. The monometallic
Ni/CNT exhibits a distinct reduction peak at approximately 579 K with a shoulder peak at
approximately 679 K. In previous reports [44,48,49], the reducibility of Ni species may be
affected by the nature of the support used and the synthesis conditions. Zhang [48] et al.
investigated the reduction of NiO deposited on CNTs and bulk NiO, suggesting that the
reduction temperature of highly dispersed NiO nanoparticles is lower than bulk NiO. The
first reduction peak of Ni/CNT at approximately 579 K is assigned to the reduction of
highly dispersed NiO nanoparticles, and the shoulder peak is assignable to the reduction of
larger NiO particles or some bulk-like NiO. The H2-TPR reduction profile of Fe/CNT is
different from those of the other samples. We observed that a tiny reduction peak appeared
at approximately 667 K over Fe/CNT catalyst, which is ascribed to the reduction of Fe2O3
to Fe3O4. However, the H2 consumption is smaller than other catalysts, which means only
trace Fe3+ oxides were reduced. In addition, the reduction peak above 835 K is assigned to
the reduction of iron oxides to Fe0 species and the interaction of FeO with C species [42].
This result reveals that Fe species are more difficult to be reduced than the Ni species. The
H2 consumption peaks of the as-calcined samples containing different Ni/Fe ratios shift to
low temperatures with increasing Ni content. Compared with Fe/CNT, Ni1Fe5/CNT
shows a larger H2 consumption peak, demonstrating that Ni promotes the reduction of Fe
species. Furthermore, with decreasing Ni/Fe atomic ratios, the shoulder peak of Ni
disappears and forms a larger reduction peak, suggesting that the introduction of Fe can
17

improve the Ni dispersion. It has been reported that close interaction between metal
species can promote or inhibit the reduction of metal oxides [47,50]. In our case, Ni and
Fe species possibly form NiFe alloys and Ni efficiently dissociates H2. In addition, H2
consumption peak at 769 K decreases with decreasing Ni content; this finding is consistent
with the H2 consumption of amorphous C species and reduction in the C functional group
in the presence of metals [46,51].
The defect densities of the synthesized NiFe/CNT catalysts were verified through
analysis of the Raman spectra (Figure S8). Two peaks with a shoulder peak are observed
1
at approximately 1340 and 1571 cm , which are ascribed to the disordered graphite
structure (D-band) and the ideal graphitic lattice vibration mode of Raman-active E2g
phonon of sp2C atoms (G-band), respectively. High ID/IG ratios, denoted as R, suggest the
high extent of defects in CNTs. The ID/IG ratios for original and pretreated CNTs are 0.77
and 0.89; hence, the pretreated CNTs contain more defects than the original CNTs, whose
surfaces can facilitate loading metal particles. With increasing Ni content, the R values of
the supported samples range from 0.84 to 0.90, which are lower than those of the
pretreated CNTs. These results indicate that the loaded metal particles occupied the defects
on the surface of CNTs decreasing amount of defects. However, increase Fe loading
slightly increased the number of defects.
Therefore, in situ XPS and HS-LEIS characterizations show that several species, such
as NiFe alloys and Fe oxides, present on the surface of catalysts, and Fe species
preferentially segregate on the surface of particles. The results of XRD, TPR, and STEM
analyses reveal the formation of NiFe alloys assignable to Fe1Ni3 structure. In addition,
18

it can be observed that the nanoparticle size decreases with decrease of Ni/Fe atomic ratio,
which suggests that Fe promotes the dispersion of Ni.
3.2. HDO of guaiacol
Initially, the catalytic performance of guaiacol HDO was evaluated over Ni catalysts
supported on different carbon materials at 573 K and 3.0 MPa (Table S3). Of all carbon
materials used, CNTs-supported catalyst is the optimal one for the guaiacol HDO reaction.
The as-received CNTs show a negligible activity of 0.5% conversion at 573 K, and the
purified CNTs display no activity. The superior catalytic results are obtained over the
NiFe/CNT catalysts in comparison with Ni/CNT and Fe/CNT, presumably due to the
integrating properties of bimetals and CNTs for enhancing the absorption and activation of
reactants. In all cases, the main C1 products are methane and methanol (Figure 5, the
upper columns) while catechol, the product of demethylation, was not detected by FID. In
this study, 2-methoxycyclohexanol is not detected, demonstrating that demethylation and
demethoxylation occurr prior to the hydrogenation of aromatic rings. Additionally,
products through aromatic ring condensation are detected by GC-MS in liquid phase.
Only trace amounts of transalkylation products like cresols are detected, indicating that
transalkylation reaction is negligible under present conditions. However, these products
and other unknown products are classified as “Others” in the tables.
3.2.1. Effect of Ni/Fe atomic ratio
Figure 5 shows the effect of Ni/Fe atomic ratio on the catalytic performance.
19

Monometallic Ni/CNT demonstrates a high conversion activity of 79.0% but poor
selectivity, meaning that the reaction pathways for the different products are energetically
similar. The major C6 ring products include cyclohexane and cyclohexanol, while phenol
and “Others” are also observed over Ni/CNT catalyst. Previous reports indicated that Ni
can efficiently dissociate H2, which is used as active metal in hydrogenation of aromatic
rings [43,52]. Hence, hydrogenation of aromatic rings during guaiacol conversion easily
occurs after demethylation and demethoxylation over the Ni/CNT catalysts. However, the
Fe/CNT catalyst shows lower catalytic activity than Ni/CNT, affording phenol as the
major product in C6 ring aromatics and many unknown products classified as “Others”, in
contrast to the Ni-based catalysts. Therefore, Fe species demonstrates poor performance in
guaiacol HDO. Similarly, Olcese et al. [26] reported that Fe/SiO2 demonstrated
deoxygenation activity without saturation of aromatic rings during conversion of guaiacol.
However, the conversion of guaiacol and the selectivity to cyclohexane are promoted
significantly upon introduction of a small amount of Fe into the catalysts studied. The
selectivity to cyclohexane presented a volcano curve with decreasing Ni/Fe atomic ratio.
The Ni5Fe1/CNT catalyst shows higher selectivity to cyclohexane and conversion of 83.4%
and 96.8%, respectively, compared with the other catalysts (Figure 5). This result indicates
that the synergistic effect of bimetallic Ni–Fe catalysts promotes the performance of HDO
facilitating a reaction mechanism. The synergy between Ni and Fe is commonly
observable, Sitthisa et al. [50] investigated the possible adsorption of furfural on the
surface of Ni(111) and NiFe(111) alloy through DFT calculations. The furan ring is
preferentially adsorbed on the surface of the Fe domain because of its oxophilic nature.
20

The strong interaction between NiFe and furfural resulted in satisfactory conversion and
selectivity. Sun et al. [30] studied the adsorption energies of phenol on the surface of
PdFe alloy through DFT calculations and revealed that Fe species favorably adsorbed
phenol and weakened the C–O bond. Furthermore, the conversion and selectivity to C6
ring products decrease with decreasing Ni/Fe atomic ratio. The conversion of guaiacol and
the selectivity to cyclohexane over the Ni2Fe1/CNT catalyst are 85.0% and 78.7%,
respectively. For the Ni1Fe1/CNT catalyst in our case, the reaction shows a conversion of
75.0% and a cyclohexane selectivity of 39.1%. Apparently, addition of suitable Fe content
enhanced the HDO of guaiacol. This phenomenon can be due to the fact that the HDO of
guaiacol requires dissociation of H2 on Ni species and adsorption of the O species of the
substrate on Fe species to induce the cleavage of the CO bond, corresponding to the TOF
values summarized in Table 2. The amount of Ni species on the surface decrease with
decreasing Ni/Fe atomic ratio (the total metal loading was maintained at 7 wt. %), thereby
reducing the number of active sites for hydrogen activation. Hence, the activated hydrogen
atoms are considered insufficient for hydrogenolysis of CO bond and hydrogenation of
aromatic rings. Other catalysts were also evaluated by maintaining the Ni loading constant
at 7 wt. % but varying Fe content (Table S4). When Fe was introduced into the Ni/CNT,
the conversion increased from 79.0% to 100.0% with incremental increase in Fe content.
The selectivity to cyclohexane increased at large increments from 53.0% to 89.2%,
supporting our explanation above.
To investigate whether the formation of NiFe alloy resulted in high conversion and
selectivity, we designed a control experiment by physical mixing of Ni/CNT and Fe/CNT
21

(Ni/Fe = 5/1) for forming a contrastive catalyst (Figure 5-h). The conversion is 69.0%,
which is lower than that of Ni5Fe1/CNT and Ni2Fe1/CNT; moreover, the product
distribution differs between the samples. The main C6 ring product is cyclohexanol, which
demonstrates a selectivity of 55.9%. This result indicates that improvement in catalytic
activity over NiFe/CNT is mainly ascribed to the formation of NiFe alloys, as proven
by XRD, TEM, and other characterization techniques.
The conversion of guaiacol and the selectivity to phenol over the Ni1Fe5/CNT
increased compared with that of Fe/CNT. The Ni1Fe5/CNT catalyst shows an incremental
conversion of 47.2% and a phenol selectivity of 82.3%. The main C6 ring product of the
reaction over Fe-based catalysts is phenol, in contrast with that of the reaction over
Ni-based catalysts with saturation of aromatic rings. Fe demonstrates strong affinity to O
but weak ability to adsorb and activate H2. Thus, the preferred reaction is the
hydrogenolysis of the CarylOCH3 bond over the Fe-based catalysts as the thermodynamic
profile in Figure 6 indicates. Addition of small quantity of Ni increased the conversion of
guaiacol and the selectivity to phenol as shown by the lower intermediate energy in
CarylOCH3 scission mechanism on the Fe3Ni(111) energy profile. Hence, activation of
hydrogen on the Ni surface is a necessary step to hydrogenolyze CarylOCH3 bonds. When
the amount of Ni is adequate, phenol as the intermediate will further react with activated
hydrogen, thereby increasing the amount of cyclohexanol and cyclohexane with increasing
guaiacol conversion. Suitable Ni contents obviously increase the yield of phenol.
Given that Ni5Fe1/CNT and Ni1Fe5/CNT exhibited high selectivity to cyclohexane
and phenol, respectively, we investigated the effects of WLHSV. Figure 8-a shows the
22

effect of WLHSV on the performance of the Ni5Fe1/CNT catalyst. The conversion and
selectivity to cyclohexane increased from 58.9% to 99.8% and from 49.0% to 99.9%,
respectively, as the WLHSV decreased from 12 to 3. The amount of cyclohexanol, as the
major by-product, decreased with decreasing WLHSV; hence, cyclohexane was formed
from cyclohexanol through the hydrogenolysis of the CalkylOH bond. This result indicates
that cyclohexanol, as an intermediate, reacts easily with H2 as increasing residence time on
the active sites, resulting in increased amount of cyclohexane. Ni1Fe5/CNT performs
lower activity than the Ni5Fe1/CNT catalyst (Figure 8-b). At high WLHSV (10 h−1), the
conversion is calculated as 22.1% with 65% selectivity of phenol. The “Others” are
mainly M-cresol and P-cresol, which are identified through GC-MS and FID detector.
Over the Ni1Fe5/CNT catalyst, with decreasing WLHSV, the conversion and selectivity
to phenol increase from 22.1% to 62.7% and from 64.9% to 83.7%, respectively, whereas
the selectivity to “Others” decrease obviously. These findings suggest that decreased
WLHSV, which corresponds to increase in residence time on the surface of the catalysts,
the guaiacol reacts with H2 adequately to form phenol with high selectivity.
It is known that not only catalytic properties but chemical stability of monometallic
active phase can be significantly promoted through adding second metal to form
bimetallic nanoparticles. Figure 9-ab shows the conversion and the selectivity over on
NiFe catalysts as a function of time. For monometallic Ni/CNT, the initial conversion is
approximately 80% and displays a rapid deactivation after 5 h. The decrease of selectivity
to cyclohexane and cyclohexanol reveal the disappearance of HDO ability. Whereas, both
bimetallic Ni5Fe1/CNT and Ni1Fe5/CNT show promoted stabilities without noticeable
23

deactivation; the selectivity to cyclohexane and phenol slightly fluctuate the initial
selectivities. XRD patterns and TEM images of fresh and spent catalysts (Figures S9 and
S10) reveal that bimetallic nanoparticles showed less sintering in contrast to that of
monometallic Ni/CNT and formed thermodynamically stable alloy structure which is
presumable that cohesive energy decrease with the incorporation of Fe in agreement with
previous reports [42,53].
Briefly, the results demonstrate the good performance of bimetallic NiFe/CNT
catalysts on the HDO and hydrogenolysis of guaiacol. Variation in Ni/Fe atomic ratios
remarkably changes the distribution of products, resulting in tunable selectivity to phenol
and cyclohexane. When the major domain is Ni species, the main products of the
hydrogenation of aromatic rings include cyclohexanol and cyclohexane. The yield of
cyclohexane peaks over the Ni5Fe1/CNT catalyst. The formation of NiFe alloys also
enhances the selectivity to cyclohexane. In addition, the reaction over the Fe-based
catalysts produces phenol as the major product, and the selectivity to phenol peaked at the
Ni/Fe atomic ratio of 0.2. Nonetheless, the HDO performance is inevitably related to the
bimetallic particle size and reduction extent of NiFe/CNT catalysts.
3.2.2. Effects of NiFe nanoparticle size and reduction extent
To investigate the effects of NiFe nanoparticle size and phase change on the HDO
of guaiacol, we prepared a series of catalysts with different reduction temperatures and
evaluated their HDO performances under identical conditions (Table 2). Entries 1–3 in
Table 2 show the performances for HDO of guaiacol over Ni/CNT, which was reduced at
24

673 K, 773 K, and 873 K, respectively. The conversion slightly fluctuates within
77%81%; however, the selectivity to cyclohexane decreases sharply from 53.0% to 4.6%,
and the selectivity to cyclohexanol increases from 31.7% to 69.7%. Moreover, the
selectivity to phenol over Entry 3 rapidly increases and reaches 10.4%. To try to make a
possible discrimination of the influences of metallic particle size and its chemical state on
the catalyst performance, XPS measurements of Ni/CNT catalyst reduced at different
temperatures were carried out and the results are showed in Figure S11 and Table S5. It
can be seen that the surface Ni0 concentration increase from 31.1% to 43.3% with the
increase of metallic particle sizes from 9.2 nm to 16.0 nm. However, the catalytic
performances decrease clearly even if the extent of reduction increase, suggesting that the
metallic particles size is the main factor that influences the catalytic performance in
comparison with the nature of nanoparticles on the catalyst surface. This phenomenon
reveals that reaction activity of the catalyst decreases with increasing metallic particle size,
resulting in reduced ability for hydrogenation of aromatic rings and hydrogenolysis of
CalkylOH bonds. These results are consistent with those of previous studies, which
showed that small metal particles are beneficial for hydrogenation reactions, whereas large
particles and bulks can weakly activate hydrogen [54,55].
The performance tendency of Ni5Fe1/CNT catalysts from Entries 46 is similar to
that over monometallic Ni/CNT catalysts. The selectivity to cyclohexane rapidly decreases
and the selectivity to cyclohexanol increases with increasing reduction temperature,
indicating a decline of catalytic performance. The XPS results (Figure S12 and Table S6)
show that the surface Ni0 concentration has a slight increase from 38.0% to 44.3% while
25

the Fe0 concentration increases greatly from 9.1% to 26% with the increase of reduction
temperature. In this case, the metallic particle sizes increase from 7.7 nm to 11.2 nm.
Therefore, the metallic particles size is considered as one of the crucial factors that
influence the catalytic performance, particularly in diversity of product distribution.
Furthermore, comparison of activity between Ni/CNT and Ni5Fe1/CNT indicates that the
bimetallic Ni5Fe1/CNT catalysts always display smaller metallic particle size and
improved performance than Ni/CNT catalysts at the same reduction temperature. This
finding reveals that addition of Fe species enable improved dispersion of nanoparticle and
promote hydrogenolysis of CalkylOH bonds. Compared with that of Ni-based catalysts,
the metallic particle size of Fe-based catalysts shows less sensitivity to reduction
temperature. Entries 79 for Ni1Fe5/CNT catalyst reduced at 673 K, 773 K, and 873 K
exhibit similar particle sizes and present similar HDO performance. The main product is
phenol, and the conversion and selectivity are approximately 40%50% and 80%,
respectively. The by-product cyclohexanol is also observed, indicating that a competitive
relationship existed between cleavage of CarylOCH3 bond and hydrogenation of phenol.
For the Fe/CNT catalysts, Entry 11 shows improved performance than Entry 10, despite
their similar metallic particle sizes; hence, Fe0 species is beneficial for hydrogenolysis of
guaiacol. The catalyst of Entry 12 with a particle size of 11.5 nm displays weak activity.
Thus, a small metallic particle is preferred for cleavage of CO bonds and HDO of
guaiacol.
To better understand the essence of NiFe nanoparticles on the catalytic behaviour,
the TOF values based on the metal dispersions (Table S7) by particle size estimated by
26

TEM and XRD were calculated and the results are summarized in Table 2. Ni/CNT
catalyst with different reduction temperatures (Entry 13) exhibits similar TOF values but
quite different product distributions, indicating that in addition to the simple relationship
between the amount of surface metal atoms and the catalytic activity, particle size effect
also exists on the formation of catalytic sites with hydrogenation of aromatic ring and
hydrogenolysis of CO bond. The different distributions of products suggest that the
conversion of guaiacol is a structure-sensitive reaction. Mortensen [56] et al. observed a
strong influence of nickel particle size on HDO of phenol and suggested that the large
particles are very active for hydrogenation of aromatic ring while small particles are
required to facilitate deoxygenation; it is consistent with our results. Compared with
Ni/CNT catalyst, Ni5Fe1/CNT catalyst (Entry 46) performs higher TOF value and better
selectivity to deoxygenation products, which is ascribed to modification of Fe to the
surface of nanoparticles. The decline of TOF values with the increase of reduction
temperature and the diversity of product distribution indicate that the metallic particles
size is the significant factor that influences the HDO of guaiacol.
To summarize, the effect of nanoparticle size and reduction extent caused by different
reduction temperatures truly influence the catalytic activity and selectivity. The particle
size effect shows a more significant influence in comparison with the phase change of
nanoparticles. Therefore, the performance of guaiacol HDO over the bimetallic NiFe
catalysts can be tuned through variations of metal nanoparticle size and reduction extent in
addition to the Ni/Fe atomic ratio.
27

3.2.3. HDO network of guaiacol over NiFe/CNT catalysts
Several reaction networks for HDO of guaiacol have been proposed to elucidate the
reaction pathways over different catalysts [2,6,13]. Demethylation and demethoxylation
are proposed to be the first step in reaction pathways. In many cases, methane is the
by-product of demethylation over various catalysts, such as Ru/MgO, Ru/C, and Fe/SiO2
[11,19,26]. Previous studies reported that demethylation occurred on activated metal sites,
and methyl group remained on the surface [6,57]. As an alternative pathway, methanol and
phenol are formed through the cleavage of CarylOCH3, which explains the presence of
CH3OH. The demethoxylation reaction is observed over several catalysts, such as sulfided
CoMo/Al2O3, CoMo/ZrO2, and supported Ni2P. Phenol can be deoxygenated to obtain
completely deoxygenated products. However, the conversion of phenol proceeds in
different pathways, depending on the activated species of catalysts. Studies show that the
hydrogenation of the aromatic ring occurs before dehydroxylation of phenol over many
catalysts [6,13,58]. Moreover, transalkylation occurs during HDO of guaiacol, especially
on Lewis acidic catalysts [21,22]. Consequently, different products (i.e., phenol, benzene,
cyclohexanol, cyclohexane, and catechol) can be obtained by controlling the reaction
pathways over different catalysts.
Scheme 1 illustrates the proposed pathways of CNT-supported Ni–Fe bimetallic
catalyst used in this study. Methane and methanol are observed as C1 by-products during
the HDO of guaiacol over NiFe bimetallic catalysts, indicating that both demethylation
and demethoxylation occurr in the reaction. However, no catechol is detected in all cases,
suggesting that this compound is rapidly deoxygenated into phenol. In addition, we
28

introduced pure methanol to the reactor as a blank test and observed the production of
methane. That is to say, methane can also be derived from the CH3OH reduction. Phenol is
the major intermediate formed from demethoxylation of guaiacol. Phenol can further react
to yield products with saturated rings in some catalysts, such as Ni/CNT and Ni5Fe1/CNT.
Consistent with data presented in Figure 5, trace benzene is observed, and products with
saturated rings (cyclohexanol and cyclohexane) are found to be the major products.
Furthermore, cyclohexene and the products of aromatic rings are observed by analyzing
the mixture collected through condensation by using GC-MS (Figure S13). Table S8 also
shows the conversion of phenol as the intermediate compound at which case that
cyclohexanon and cyclohexanol could be observed. These results suggest that phenol
hydrogenation occurr before the cleavage of the last CarylOH. Cyclohexanol can be
further deoxygenated when suitable Fe contents were introduced.
To elucidate the synergism of NiFe nanoparticles for HDO of guaiacol over NiFe
bimetallic catalysts containing various Ni/Fe atomic ratios, we propose appropriate
reaction mechanisms in Scheme 2. Our results show that saturation of aromatic rings over
monometallic Ni/CNT catalyst is favored to produce cyclohexanol and cyclohexane. As
depicted in Scheme 2-a, Ni demonstrates high hydrogen absorption and activation abilities
but weak substrate absorption. After demethoxylation of guaiacol, phenol can be adsorbed
plainly on the Ni surface and then react with activated hydrogen to form cyclohexanol.
However, cyclohexanol can hardly further react to yield cyclohexane because of weak
oxygen affinity, causing low selectivity to cyclohexane. When suitable Fe contents are
introduced into the surface of Ni, the HDO ability of catalyst is considerably enhanced, as
29