A novel surface modification approach for synthesizing supported nickel phosphide catalysts with high activity for hydrodeoxygenation of benzofuran

Article abstract of DOI:10.1016/j.apcata.2015.08.007

A novel and simple surface modification approach for preparing a highly active nickel phosphide hydrodeoxygenation (HDO) catalyst is proposed, in which the original nickel phosphide catalyst is modified with oxygen at 373 K under a flow of air for 1 h ins

Full text of DOI:10.1016/j.apcata.2015.08.007

Accepted Manuscript
Title: A novel surface modification approach for synthesizing
supported nickel phosphide catalysts with high activity for
hydrodeoxygenation of benzofuran
Author: Hua Song Jing Gong Hualin Song Feng Li
PII:
S0926-860X(15)30102-2
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.apcata.2015.08.007
APCATA 15501
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
28-5-2015
3-8-2015
4-8-2015
Please cite this article as: Hua Song, Jing Gong, Hualin Song, Feng Li, A novel
surface modification approach for synthesizing supported nickel phosphide catalysts
with high activity for hydrodeoxygenation of benzofuran, Applied Catalysis A, General
http://dx.doi.org/10.1016/j.apcata.2015.08.007
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.

A novel surface modification approach for synthesizing supported nickel
phosphide catalysts with high activity for hydrodeoxygenation of
benzofuran
a,b,
∗
a
c,∗∗
a,b
Hua Song , Jing Gong , Hualin Song , Feng Li
a
College of Chemistry & Chemical Engineering, Northeast Petroleum University,
Daqing 163318, China
b
Provincial Key Laboratory of Oil & Gas Chemical Technology, College of Chemistry
&
Chemical Engineering, Northeast Petroleum University, Daqing 163318,
Heilongjing, China
c
Key Laboratory of Cancer Prevention and Treatment of Heilongjiang Province,
Mudanjiang Medical University, Mudanjiang 157011, Heilongjing, China
*
To whom correspondence should be addressed. Email: songhua2004@sina.com.
Tel: +86 0459 6503167; Fax: +86 0459 6506498.
**
To whom correspondence should be addressed: Email: songhualin@126.com. Tel:
Highlights ►
+86 0453 6984321; Fax: +86 0453 6984321.
Highlights ►
＞
A novel modification approach for preparing a highly active Ni P is
2
proposed. ►
is needed. ►
＞
The catalyst was modified with oxygen, no pre-treatment of the catalyst
＞
The Ni P preparation step is simplified and therefore energy-efficient.
2
►
＞
2
The O-free products yield of Ni P-O/MCM41 improved by 60% compared to
Ni P-N/MCM41.
2
Abstract

A novel and simple surface modification approach for preparing a highly active
nickel phosphide hydrodeoxygenation (HDO) catalyst is proposed, in which the
original nickel phosphide catalyst is modified with oxygen at 373 K under a flow of air
for 1 h instead of being passivated with an O /N mixture. In this way, no pre-treatment
2
2
of the catalyst prior to reaction is needed. The effect of surface modification with
oxygen on catalytic performance in HDO of benzofuran was investigated. This surface
modification apparently promotes the formation of smaller Ni P particles in catalyst,
2
favoring more dissociation of molecular hydrogen. The Ni P preparation step is
2
simplified and therefore energy-efficient, and the obtained catalyst exhibits a much
higher O-free products yield (88%) than that with the unmodified catalyst (38%).
：
Key words Nickel phosphide; Surface modification; Oxygen; Hydrodeoxygenation;
Benzofuran
1
. Introduction
The increasing demand for energy, environmental issues, and depletion of oil
reserves have highlighted the need to develop alternatives to fossil fuels. As a kind of
renewable energy resource, biomass has been attracting much attention for the
production of liquid fuels [1]. One method of biomass conversion is pyrolysis, but
unfortunately the liquid products obtained by pyrolysis contain a high fraction of
oxygen (35–50 wt.%) [2]. The high oxygen content of pyrolysis oil leads to higher
viscosity, acidity, and lower heating values [3]. To upgrade the physical and chemical
stability of these liquid products, the oxygen must be removed as water or carbon
oxides by a hydrodeoxygenation (HDO) process [4]. Oxygen removal can be carried
out by conventional hydrodeoxygenation processes, which are generally operated
under a high pressure (3–10 MPa) of hydrogen at moderate temperatures (573–773 K)
[
5]. The majority of organooxygen compounds that are found in synthetic feed-stocks
have either a phenolic or furanic structure [6,7]. Benzofuran (BF) is a commonly used

probe molecule to evaluate catalyst performance in HDO reactions [6,8-14]. It has a
structure similar to those of the most commonly studied organosulfur and
organonitrogen compounds for hydrodesulfurization (HDS) and hydrodenitrogenation
(HDN) reactions, respectively, and thus provides an analogue that facilitates
comparisons of HDO with these processes.
HDO catalysts composed mainly of conventional sulfide catalysts, noble metal
catalysts, and transition metal phosphides [3,15-20]. However, sulfide catalysts suffer
from a progressive deactivation during the HDO reaction due to oxidation of the active
phase[21]. Noble metal catalysts exhibit higher HDO activity than sulfide catalysts, but
their high cost prevents their wide practical application. In recent years, transition metal
phosphides have attracted great attention as new hydroprocessing catalysts. Ni P has
2
recently been reported as a novel catalyst with unique physical and chemical properties
in the HDO of furan, benzofuran, phenol, guaiacol, and anisole [4, 20, 22-23].
Metal phosphides and metal catalysts synthesized by reduction of precursors in H2
are usually passivated prior to their removal from the synthesis apparatus. Passivation
of a freshly prepared catalyst most frequently involves passing a dilute oxygen flow
(
0.5 vol.% O /N ) over it in order to form a protective layer. In such cases, re-reduction
2 2
to the metal phosphide is necessary, and this operation consumes energy. Duan et al.
23] examined H S as a passivation agent for Ni P catalysts. They found that a small
[
2
2
degree of surface reconstruction took place during HDS reaction at the surface of the
H S-passivated catalyst, and that more extensive reconstruction occurred at the surface
2

of the in-situ reduced catalyst. Therefore, surface modification of the metal phosphide
is required.
In this work, a novel and simple surface modification approach for preparing a
highly active supported nickel phosphide hydrodeoxygenation (HDO) catalyst is
proposed. The supported nickel phosphide catalyst was synthesized according to a
previously described method [24] from ammonium hypophosphite and nickel chloride
at a lower reduction temperature of 673 K. Then as-obtained nickel phosphide catalyst
was modified with oxygen at 373 K under a flow of air for 1 h instead of the passivation
with an O /N mixture. In this way, pre-treatment of the catalyst prior to reaction was
2
2
not needed, thereby saving energy. Finally, a possible mechanism was proposed to
interpret the improved catalytic behavior of the modified catalyst.
2
. Experimental
2
.1. Preparation of supports and catalysts
Siliceous MCM-41 was synthesized using tetraethyl orthosilicate (TEOS) as the
silica source and cetyltrimethylammonium bromide (CTAB) as the template, following
the procedure as described in the literature [25]. The supports obtained were named
MCM-41.
The supported Ni P catalyst precursor was prepared by impregnating an
2
ammonium hypophosphite (NH H PO ) and nickel chloride (NiCl ·6H O) solution
4
2
2
2
2
with the mesoporous MCM-41. The precursor, was prepared with Ni loading of 8.8
wt.% and an initial Ni/P molar ratio of 1/2. In a typical experiment, 1.8 g (20.62 mmol)

NH H PO and 2.5 g NiCl ·6H O (10.31 mmol) were dissolved in 20 mL of deionized
4
2
2
2
2
water at room temperature to form a uniform solution. A quantity of 6.0 g MCM-41
was wet-impregnated with the above solution for 8 h. The volume of the solution of
precursors during the impregnation was 26 mL. The mixture was heated to evaporate
the water, the obtained solid was dried at 363 K overnight. The precursor was then
pelleted, crushed, and sieved to 16/20 mesh. For the reduction, precursor was placed in
a fixed-bed reactor by heating to 673 K, at a rate of 2 K/min in a flow of H (200
2
mL/min), held for 2 h, then naturally cooled to room temperature in a continuous H2
flow. The prepared catalyst was passivated in O /N mixture (0.5 vol.% of O ) at a rate
2
2
2
of 20 mL/min for 1 h. The catalyst obtained was named as Ni P-N/MCM41. A highly
2
active supported nickel phosphide catalyst (Ni P-O/MCM41) for HDO was
2
synthesized in flowing H at 673 K(200 mL/min), at a rate of 2 K/min, held for 2 h, then
2
cooled to 373 K under flowing air (20 mL/min) for 1 h instead of passivation with
O /N mixture.
2
2
2
.2 Characterization methods
The X-ray diffraction (XRD) analysis were carried out on a D/max-2200PC-X-ray
diffractometer using CuKα radiation under the setting conditions of 40 kV, 30 mA,
scan range from 10 to 80° at a rate of 4°/min.
The typical physico-chemical properties of supports and catalysts were analyzed
by BET method using Micromeritics adsorption equipment of NOVA2000e. All the

samples were outgassed at 473 K until the vacuum pressure was 6 mm Hg. The
adsorption isotherms for nitrogen were measured at 77 K.
Transmission electron microscope (TEM) examinations were performed using the
JEM-1010 instrument supplied by JEOL. The samples were dispersed in ethanol and
placed on a carbon grid before TEM examinations.
The X-ray photoelectron spectroscopy (XPS) spectra were acquired using
ESCALAB MKII spectrometer under vacuum. XPS measurements have been
performed for Mg radiation (E = 1253.6 eV) and equipped with a hemi-spherical
analyzer operating at fixed pass energy of 40 eV. The recorded photoelectron binding
energies were referenced against the C 1s contamination line at 284.8 eV.
Irreversible CO uptake measurements were used to titrate the surface nickel atoms
and to provide an estimate of the active sites on the catalysts [26,27]. The CO
adsorption capacity of Ni P-N/MCM41 catalyst was measured after passivation
2
without re-reduction, while Ni P-O/MCM41 catalyst was measured after surface
2
modification with oxygen without re-reduction using a Builder PCA-1200 analyzer.
CO uptake was obtained by pulsing calibrated volumes of CO into a He carrier. CO
uptake was calculated by measuring the decrease in the peak areas caused by adsorption
in comparison with the area of a calibrated volume.
2
.3 Catalytic activities
The HDO of BF over prepared catalysts were performed in a flowing high-pressure
fixed-bed stainless steel catalytic reactor (8 mm in diameter, and 400 mm in length),

using a feed consisting of a decalin solution of BF (2 wt.%). Prior to reaction, 0.7 g of
the Ni P-N/MCM41 catalyst was pretreated at 673 K for 2 h in a H flow (30
2
2
–
1
mL·min ). For the Ni P-O/MCM41 catalyst this step was saved. For comparison, we
2
also tried treatment Ni P-N/MCM41 catalyst without H reduction prior to HDO in the
2
2
same reactor. The catalyst obtained was named as Ni P-N/MCM41-nr. Catalytic
2
–
1
activities were measured at different temperatures (493-573 K) 3.0 MPa, WHSV=4 h ,
and hydrogen/oil ratio of 500 (V/V). Sampling of liquid products was started 6 h after
the steady reaction conditions had been achieved. Temperature was varied as follows:
5
73 K → 533 K → 493 K → 513 K → 553 K → 573 K, with the reaction held at each
temperature for 3 h. The feed and reaction product was analyzed by FID gas
chromatography with a GC-14C-60 column.
The total conversion was calculated from the ratio of converted benzofuran/initial
benzofuran. The selectivity to the different reaction products was calculated based on
the moles of each chemical as follows: ethylbenzene (EB), ethylcyclohexane (ECH),
methylbenzene (MB), benzene (B), methylcyclohexane (MCH),
2
,3-dihydrobenzofuran (2,3-DHBF), 2-ethylphenol (2-EtPh), and phenol (Ph).
Turnover frequency (TOF) values of the samples containing supported nickel
phosphide were calculated using Eq. (1) : [28]
F y
TOF =
(1)
W M

-1
Where F is the molar rate of BF fed into the reactor (µmol·s ), W is the weight of
catalyst (g), y is the conversion of BF (%), and M is the mole of sites loaded determined
by the CO uptake.
3
. Results and discussion
3
.1. XRD
XRD patterns were used to identify the crystalline phases formed after reduction (Fig. 1). In order to
study the effect of surface modification on the crystal structure of supported nickel phosphide catalysts,
the structures of MCM-41 support, Ni P-O/MCM41 and Ni P-N/MCM41 were determined. For
2
2
MCM-41, a broad diffraction line located at 2θ = 23° is due to the amorphous nature of mesoporous
structure [29]. For all catalysts, the peaks at 2θ = 40.6°, 44.5°, 47.1°, 54.1°, and 54.8° (PDF: 03–0953)
can be ascribed to Ni P. No other additional phase related to Ni and P was observed, indicating that the
2
active phase formed was mainly Ni P for both samples. After modification by the oxidation process,
2
Ni P-O/MCM41 retained the crystal structure of Ni P, suggesting that such surface modification of the
2
2
catalyst did not disrupt its crystal structure. The average size of the Ni P crystallites (column 5 of Table
2
1
) estimated by the Scherrer equation was 20 nm for Ni P-O/MCM41, smaller than that for
2
Ni P-N/MCM41 (24 nm). This indicates that surface modification with oxygen can promote the
2
formation of smaller, highly dispersed Ni P particles on the catalyst.
2
3
.2 N adsorption
2
Fig. 2 shows the N adsorption–desorption isotherms and BJH pore size
2
distributions for the MCM-41 support, Ni P-O/MCM41 and Ni P-N/MCM41 catalysts.
2
2
Table 1 summarizes the BET surface areas and pore volume results. The surface area

2
-1
3
-1
and pore volume of the MCM-41 support used were 1051 m ·g and 0.84 cm ·g ,
respectively. It can be seen that both catalysts led to decreases in surface area and pore
volume due to blocking of their pores by Ni P. The catalyst Ni P-N/MCM41 showed a
2
2
2
-1
surface area of 502 m ·g , whereas that of Ni P-O/MCM41 was slightly higher at 509
2
2
-1
m ·g . This can be attributed to the decrease in the enrichment of phosphorus on the
surface of Ni P-O/MCM41 compared with Ni P-N/MCM41 (see Table 2). This is
2
2
discussed in Section 3.5 below.
As can be seen from Fig. 2a, all the samples displayed a type IV isotherm and a
standard H4 type hysteresis loop according to the IUPAC classification, showing the
presence of some mesopores. The pore size distributions obtained by
Barrett−Joyner−Halenda (BJH) analysis of these curves show no marked differences
among the samples. The average pore diameters of these samples lie in the range of
2
3
.5−3.5 nm (see Fig. 2b).
.3. CO uptake
The CO uptakes at room temperature of the samples are listed in column 6 of Table
. It may be assumed that CO molecules are mainly adsorbed at Ni sites and so the
1
amount of CO molecules adsorbed on P sites may be very small [30]. As shown in
–
1
Table 1, the CO uptake of Ni P-O/MCM41 was 53 µmol·g , much higher than that of
2
–
1
Ni P-N/MCM41 (31 µmol·g ). This is possibly because the enrichment of phosphorus
2
on the surface of Ni P-O/MCM41 is suppressed by surface modification with
2
oxidization, which leads to the increase in the amount of exposed nickel atoms on the

surface (See Table 2, the superficial atomic ratio of Ni/P/O obtained from XPS
analysis). This will discussed further in section 3.5. Similar results were reported by Li
et. al [31]. The high CO uptake of Ni P-O/MCM41 is also attributed to the highly
2
dispersed smaller Ni P particles. (See the provided TEM spectra).
2
3
.4. TEM
The morphologies and microstructures of Ni P-O/MCM41 and Ni P-N/MCM41
2
2
catalysts were observed by TEM (Figure 3). It can be seen from Fig. 3(a), the Ni2P
particle sizes in the Ni P-N/MCM41 catalyst ranged from approximately 6 to 18 nm.
2
Unlike the typical stacked morphologies of Mo and W sulfides, Ni P are not layered
2
and form spherical particles that can be well dispersed on supports [24]. Fig. 3(b) shows
that the Ni P particle sizes in the Ni P-O/MCM41 catalyst ranged from approximately
2
2
2
to 6 nm. The TEM data, combined with the crystallite size estimates from XRD
indicate that the Ni P-O/MCM41 catalyst has smaller and more uniform Ni P particles
2
2
than Ni P-N/MCM41, and surface modification with oxygen is beneficial to the
2
formation of small Ni P particles.
2
3
.5. X-ray photoelectron spectroscopy
In order to gain further insight into the surface composition of surface oxidation
layer and the valence states of active components, the X-ray photoelectron
spectroscopy (XPS) spectrum of fresh and spent catalysts were performed. Table 2
shows the binding energy values and the Ni/P/O atomic ratios. Table 3 shows the
proportion of each species for the Ni2p and P2p spectra obtained by XPS analysis. As
shown in Fig. 4(a), all spectra were decomposed, taking into account the spin-orbital

splitting of the Ni 2p3/2 and Ni 2p1/2 lines (about 17 eV) and the presence of satellite
peaks at about 6 eV higher than the binding energy of the parent signal. This
observation is consistent with that of Korányi et al. [32]. For the fresh and spent
catalysts, Ni 2p3/2 core-level spectrum consists of three components, the first of which
δ+
centered at 851.6–853.5 eV can be assigned to Ni in the Ni P phase, and the second at
2
2
+
8
56.3–857.6 eV corresponds to the possible interaction of Ni ions with phosphate
ions, as a consequence of superficial passivation, along with the broad satellite at
2
+
approximately 6.0 eV higher than that of the Ni species, and this shake-up peak is
ascribed to divalent species [33]. Other broad peaks on the high binding energy side can
be assigned to the Ni 2p components [34]. The P 2p binding energy spectrum features
1/2
δ-
a peak at 129.5 eV due to P of the metal phosphides and a peak at 134.5eV due to
surface metal phosphate species arising from the superficial oxidation of the nickel
phosphide particles [35].
As shown in Fig. 4(a), the XPS pattern of the as-prepared Ni P-N/MCM41 exhibited Ni 2p peaks at
2
δ+
2+
8
52.3 and 856.3 eV, which can be attributed to the Ni bands of Ni P (40.6%, See Table 3) and Ni
2
species (59.4%), respectively. Compared with fresh Ni P-N/MCM41, the binding energy for Ni species
2
on the fresh Ni P-O/MCM41 catalyst was almost the same. However, the intensity of the peak at around
2
8
52.4 eV (5.7%) was clearly lower for the fresh Ni P-O/MCM41 catalyst, indicating that some of the
2
δ+
reduced Ni on the surface had been oxidized during the surface modification. Compared with the fresh
δ+
Ni P-O/MCM41 catalyst, the intensity of the Ni signal (24.7%) for spent Ni P-O/MCM41 was
2
2
stronger, indicating that more Ni P particles were exposed on its surface. This may result from the fact
2

2+
that Ni species in the oxidized layer formed during surface modification with oxygen were reduced
δ+
δ+
back to Ni species by the hydrogen flow during the HDO reaction. In contrast, the intensity of the Ni
signal (31.3%) from the spent Ni P-N/MCM41 catalyst decreased rather than increased after the HDO
2
δ+
reaction. Interestingly, compared with the fresh Ni P-O/MCM41 catalyst, the binding energy for Ni
2
shifts to a slightly lower value, while on the opposite, a shift to higher binding energies is observed for
2+
Ni on the spent Ni P-O/MCM41. This indicates that during HDO reaction, the modifier oxygen
2
changed the electron density of nickel species.
−
As shown in Fig. 4(b), for fresh Ni P-N/MCM41, peaks centered at 133.7 eV for H PO (41.8%,
2
2
3
3
−
See Table 3) and 134.6 eV for PO4 (28.1%) were observed, together with a low intensity peak at 129.3
δ–
eV that can be assigned to the P band (0.9%) of Ni P. It is noteworthy that a new band at 132.8 eV was
2
also observed. This binding energy is higher than the value reported for elemental phosphorus (130.2 eV)
−
−
and lower than the 133.5 eV reported for H PO . The band at 132.8 eV may possibly be due to H PO
2
2
3
2
(
29.2%) [35]. However, the fresh Ni P-O/MCM41 shows only the two bands at 133.8 (49.7%) and 134.9
2
eV (49.4%), with the band at 132.8 eV diminished. This implies that the phosphorus on the surface of the
catalyst was further oxidized during the modification with oxygen. Compared with fresh
Ni P-N/MCM41, the total intensity of the P 2p features of fresh Ni P-O/MCM41 decreased, with a
2
2
broadening towards the high binding energy side, indicating a strong interaction between oxygen and
phosphorus to generate an oxy-phosphide. This is in accordance with the results obtained by Liu et al.
[
36]. In addition, an increase in the content of surface oxygen was observed for fresh Ni P-O/MCM41,
2
which also confirms this point. This result discussed in the Ni/P/O atomic ratios below. For spent
δ−
Ni P-N/MCM41, the contribution from P (0.9%) species remained essentially unchanged, and the peak
2

−
due to the H PO species at 132.8 eV disappeared compared to the fresh sample. The disappearance of
2
2
−
-
−
3-
δ-
the H PO peak confirmed that the H PO was transformed into H PO (21.6%), PO (77.5%), and P
2
2
2
2
2
3
4
−
3-
species (0.9%), and mainly existed in the forms of H PO3 and PO4 during the HDO reaction.
2
As shown in Fig. 4(c), the fresh Ni P-N/MCM41and Ni P-O/MCM41 catalysts
2
2
-
showed a peak at around 532.5 eV, which could be attributed to OH [36]. Interestingly,
-
for the fresh Ni P-O/MCM41 catalyst, the binding energy of OH shifts to a slightly
2
higher value. This indicates that oxygen was strongly adsorbed on the surface of
Ni P-O/MCM41. Compared with fresh Ni P-O/MCM41, a new peak at 533.2 eV was
2
2
observed for spent Ni P-O/MCM41, while the band at 532.6 eV was diminished. The
2
band at 533.2 eV could be assigned to adsorbed water [37]. This is discussed in Section
3
.6 below.
XPS analyses were used to calculate the surface Ni/P/O atomic ratios. The
theoretical Ni/P ratio corresponding to the precursor materials was 1/2. However, all as
prepared samples exhibited lower
Ni/P values than 1/2, which may be due to the enrichment of phosphorous on the surface of the catalysts.
However, an increase in the Ni/P ratio was observed for all spent samples compared to their
corresponding as prepared samples. This may be due to a partial loss of phosphorous due to the formation
of volatile P species, such as PH , during the HDO process, leading to a decrease in the amount of P on
3
the surface. Both the fresh and spent Ni P-O/MCM41 samples have comparatively higher Ni/P atomic
2
ratio compared to the Ni P-N/MCM41 catalysts. This implies that the enrichment of phosphorus on the
2
surface of Ni P-O/MCM41 can be suppressed by surface modification with oxidization, which leads to
2

the increase in the amount of exposed nickel atoms on the surface of the sample. The enrichment of
nickel atoms on the surface of catalysts modified with oxidization is possibly due to the formation of an
oxide layer. This is in accordance to the CO uptake results. The oxygen contents of the fresh
Ni P-O/MCM41 catalyst was slightly higher than that of the fresh Ni P-N/MCM41 catalyst, which
2
2
indicates a much higher proportion of Ni and P atoms are at higher oxidation state on the surface of the
Ni P-O/MCM41 catalyst. Moreover, after HDO reaction, the oxygen contents increased slightly for all
2
spent samples compared to their corresponding fresh samples. We conclude that the high oxygen
contents in the superficial for all spent samples may be due to adsorbed oxygen species.
3
.6. Catalytic results and reaction pathways
The prepared catalysts have been tested using the HDO of BF for 48h. The
Ni P-O/MCM41 catalyst did not suffer deactivation for 48 h, displaying high stability.
2
Fig. 5 depicts the HDO conversion as a function of the reaction temperature. As
expected, BF conversion increased with increasing reaction temperature over all the
catalysts. It is clear that surface modified Ni P-O/MCM41 catalyst showed a higher BF
2
conversion than the Ni P-N/MCM41 catalyst prepared with O /N mixture
2
2
2
passivation–H reduction (673 K for 2 h), and Ni P-N/MCM41-nr catalyst prepared
2
2
with O /N mixture passivation without H reduction over the entire temperature range
2
2
2
tested. Compared with Ni P-N/MCM41 (31%), Ni P-O/MCM41 showed much higher
2
2
BF HDO conversion (57%) at relatively low temperature (493 K). On increasing the
reaction temperature, the conversion over the Ni P-O/MCM41 catalyst increased and
2
reached 95% at 573 K.

The BF conversion and CO uptake for samples were used to calculate the HDO
TOF and listed in Table 1, column 8. As can be seen from Table 1, the HDO TOF of the
-1
Ni P-N/MCM41 is 0.0016 s . The Ni P-O/MCM41 possesses a TOF value of 0.0018
2
2
-1
s , indicating the intrinsic activity of sites is increased slightly (12.5%) by the surface
modification with oxidization.
With regards to the selectivity and reaction pathways of BF HDO over the
as-prepared catalysts, the products and intermediates formed were analyzed. The main
oxygen-free compounds products detected were: ethylbenzene (EB), ethylcyclohexane
(ECH), methylbenzene (MB), benzene (B), and methylcyclohexane (MCH). The
oxygen-containing intermediates observed were 2,3-dihydrobenzofuran (2,3-DHBF),
2
-ethylphenol (2-EtPh), and phenol (Ph).
Fig. 6 shows the effect of reaction temperature on the selectivity to different
reaction products. It can be seen that at a lower temperature (Fig. 6a, 6b), the main
products over the Ni P-N/MCM41 and Ni P-O/MCM41 catalysts were O-containing
2
2
intermediates, mainly 2,3-DHBF and 2-EtPh. With increasing reaction temperature, the
yield of 2-EtPh initially increased and then decreased, and that of the 2,3-DHBF
decreased. With these two catalysts, the first step involved in the transformation of BF
was hydrogenation of the C=C bond leading to 2,3-DHBF as reported previously[11].
2
-EtPh was formed as a secondary product by C–O bond cleavage in 2,3-DHBF.
For the Ni P-N/MCM41 catalyst, 2,3-DHBF was the main product at 493–513 K,
2
whereas 2-EtPh was the main product at 533–573 K. 2-EtPh was probably transformed

by two pathways: a primary pathway leading to EB by direct deoxygenation (DDO)
pathway and a secondary way leading to phenol by dealkylation. At 493 K, the
oxygen-free products over the Ni P-N/MCM41 catalyst were ECH (3%), EB (4%) and
2
MCH (2%). At 573 K, the main oxygen-free products were ECH (17%), EB (9%), and
MCH (7%), B (4%) and MB (2%). EB was converted ECH by hydrogenation and to
MB and B by dealkylation. B was also formed by DDO of Ph. ECH converted to MCH
by dealkylation. According to the detected product distribution and reaction pathways
previously discussed in the literature [38], possible reaction routes that may occur
during HDO of BF over the Ni P-N/MCM41 catalyst are shown in Scheme 1. A similar
2
reaction mechanism was established by Bunch et al. [8], who suggested that C–O bond
cleavage occurs through Hofmann-type elimination reaction (E ) and nucleophilic
2
substitution reaction (SN2).
From Fig. 6b it can be seen that the selectivity for the deoxygenated product ECH
was much higher than those for all of the other products over the Ni P-O/MCM41
2
catalyst. In addition, with increasing reaction temperature, selectivity for ECH over the
Ni P-O/MCM41 catalyst increased significantly. ECH (selectivity of 68.0% at 573 K)
2
became the major product at temperatures higher than 553 K. However, the selectivity
in favor of EB and MCH increased much more slowly than that for ECH with
increasing temperature. The selectivity for MCH, which was formed by the
dealkylation of ECH, was about 11%, and the selectivity for EB was about 9% at 573
K. This suggests that the transformation of EB to ECH was fast, whereas that of ECH to

MCH was slow at higher temperature. Compared with the products formed over the
Ni P-N/MCM41 catalyst, phenol was hardly detected in the products over the
2
Ni P-O/MCM41 catalyst, indicating that 2-EtPh was probably transformed by one
2
pathway, which is leading to EB by DDO. B was not detected and the selectivity for
MB was less than 0.6%, indicating that EB was mainly converted to ECH. On the basis
of the above description, possible reaction routes that may occur during HDO of BF
under the experimental conditions over the Ni P-O/MCM41 catalyst are shown in
2
Scheme 2. Moreover, the selectivity of the non-reduced catalyst (Ni P-N/MCM41-nr)
2
is shown in Figure S1.
Fig. 7 shows the effect of reaction temperature on yields of O-free products over all
the catalysts. Over the entire temperature range tested, the Ni P-O/MCM41 catalyst
2
gave higher yields of O-free products than the Ni P-N/MCM41 and
2
Ni P-N/MCM41-nr catalysts, the difference becoming more marked with increasing
2
temperature. The Ni P-N/MCM41 showed higher yields of O-free products compare to
2
Ni P-N/MCM41-nr, indicating that the re-reduction at 673 K before HDO reaction by
2
hydrogen improved the HDO performance of the Ni P-N/MCM41 catalyst. However,
2
the yields of O-free products of Ni P-N/MCM41 was significantly lower that of the
2
Ni P-O/MCM41 catalyst. These results suggest that H re-reduction at 673 K before
2
2
HDO reaction is essential to restoring the desired catalytic properties of the
Ni P-N/MCM41 catalyst and this operation consumes energy. For Ni P-O/MCM41
2
2
catalyst, the re-reduction before HDO reaction is omitted, therefore, the proposed

surface modification approach is simplified and therefore energy-efficient.
Ni P-O/MCM41 catalyst was highly effective for promoting O-removal reactions in the
2
range 553–573 K. At a reaction temperature of 573 K, the yield of O-free products over
the Ni P-O/MCM41 catalyst reached 88%, improved by 50% compared with that over
2
the Ni P-N/MCM41 catalyst (38%).
2
It is worth noting that the selectivity for ECH was significantly increased at > 553
K over Ni P-O/MCM41. This suggests that surface modification with oxygen
2
promoted the conversion of 2-EtPh, and that the transformation of EB to ECH over
Ni P-O/MCM41 was very fast since the selectivity for ECH over the Ni P-O/MCM41
2
2
catalyst was much higher than that for EB. Edelman [12] proposed that after
ring-opening in 2, 3-DHBF, the subsequent HDO of the 2-EtPh intermediate will
govern the overall HDO of BF. HDO is generally conducted over traditional
hydrotreating catalysts, which consist of metal sulfides such as Ni-MoS/Al O or
2
3
CoMoS/Al O [8-9,38]. The active sites of these catalysts are coordinatively
2
3
unsaturated sites at the edges of MoS slabs. In addition, the surface configuration of
2
the catalyst is known to strongly affect the reaction activity [11]. Recently, Romero
[
11] proposed that over sulfided NiMoP/Al O catalysts the C–O bond cleavage of
2 3
2
-EtPh could occur by an elimination process involving a basic site (presumably a
sulfur anion) and a coordinatively unsaturated site. In a similar manner, the active site
configuration of the Ni P catalyst upon the production of H O during the HDO can be
2
2
understood. Ji-Sue Moon et al. [40] reported HDO of guaiacol over Ni P/SiO catalysts,
2
2

and suggested that the active sites of these catalysts involved H or OH groups on Ni or
P sites on the surface of the Ni P, and that the DDO pathway was favored by the surface
2
OH groups. The mechanism involved in the transformation of 2-EtPh into EB over
supported nickel phosphide catalysts is shown in Scheme 3. As mentioned above, there
are OH groups on the fresh Ni P-O/MCM41 catalyst and water on the spent
2
Ni P-O/MCM41 catalyst (Fig. 4(c)). The adsorbed OH groups on the surface of
2
Ni P-O/MCM41 were effectively activated at a certain temperature. They could then
2
bond with hydrogen from 2-EtPh to form water, which was subsequently removed from
the surface of the Ni P-O/MCM41. The OH absorption sites were thereby regenerated.
2
These regenerated absorption sites could further accommodate the OH groups of
2
-EtPh, allowing the reaction to proceed. These absorbed OH groups may behave
somewhat like a reaction initiator in the initial reaction. In this way, the DDO pathway
can be significantly promoted. As a result, the transformation of 2-EtPh to EB is
remarkably enhanced, and then EB is rapidly converted to ECH by hydrogenation. In
the case of Ni P-N/MCM41, the adsorbed OH groups were not as active as those on
2
Ni P-O/MCM41 catalyst since no water was detected on the spent Ni P-N/MCM41.
2
2
This indicated the transformation of 2-EtPh to EB over the Ni P-N/MCM41catalyst
2
proceeded by a different route.
Surface modification of the supported nickel phosphide catalysts with oxygen
greatly improved their HDO performance. It has been reported that the HDS
performance was improved when an Ni P catalyst was partially covered with sulfur
2

[
23,40]. Analogously, modification of Ni P with oxygen may exert a similar effect on
2
HDO. On the basis of the above discussion, a possible mechanism may be proposed to
explain the effect of surface modification of the supported nickel phosphide catalysts
with oxygen on their HDO performance (Fig. 8). In the HDO process, molecular
hydrogen is cleaved into active atomic hydrogen over the catalyst. The surface
modification with oxygen changes the electronic environment of the Ni P active phase,
2
which may induce more dissociation of molecular hydrogen (Ni P behaves somewhat
2
like hydrogenase) and improve the properties of the active atomic hydrogen. This is
understandable, since the surface modification of Ni P/MCM41 with oxygen, which
2
mainly leads to an interaction between oxygen and the phosphorus, generates a new
active phase for HDO. The shift of the P2p binding energy of Ni P-O/MCM41 catalyst
2
confirms interaction between oxygen and phosphorus (Fig. 4b). Finally, the surface
modification results in an increase in catalytic performance (mainly through the DDO
route shown in scheme 3). This is much like the modification by adsorbed sulfur
leading to an improvement in the HDS performance of Ni P catalysts [40, 41]. As with
2
sulfur, the modification with oxygen introduces highly electronegative atoms, which
attract electrons and thus decrease the electron density around the H atoms bonded to
Ni. Li et al. [31] have reported that the oxygen modifier improves the ionic property of
the catalyst surface, which in turn improves the generation of active atomic hydrogen.
On the other hand, the surface modification is beneficial for the formation of smaller
Ni P particles (see the TEM results in Fig. 3) and enhances the performance of HDO of
2

BF. This is in good agreement with the results of a previous study reported by Liu et. al
36]. They concluded that the cooperation between oxygen and metal centers with
[
moderate activity was responsible for the high performances of catalysts in the
water-gas-shift reaction.
4
. Conclusions
A novel surface modification approach for preparing a highly active
hydrodeoxygenation (HDO) catalyst (Ni P-O/MCM41) was proposed. A supported
2
nickel phosphide catalyst was synthesized from ammonium hypophosphite and nickel
chloride at a relatively low reduction temperature of 673 K. Then as-obtained catalyst
was then modified with oxygen at 373 K under a flow of air for 1 h instead of the usual
passivation with an O /N mixture. Prior to its use in catalytic reactions, no
2
2
pre-treatment of the modified Ni P-O/MCM41catalyst was needed. The effect of the
2
surface modification of the Ni P catalyst with oxygen on its performance of HDO of BF
2
was investigated. The results showed that this surface modification did not disrupt the
crystal structure of Ni P. The oxygen on the modified Ni P catalyst interacts mainly
2
2
with the phosphorus to generate a new active phase for HDO, which may induce more
dissociation of molecular hydrogen on the surface. The Ni P-O/MCM41 catalyst
2
showed very high catalytic activity, with BF conversion close to 95% at 573 K, and a
much higher O-free products yield (88%) than that with the Ni P-N/MCM41 catalyst
2
(38%), indicating that the surface modification of the support nickel phosphide catalyst
was highly effective for promoting O-removal reactions. In conclusion, the proposed

surface modification approach of the Ni P catalyst is simplified and energy-efficient
2
because of no pre-treatment of the catalyst prior to reaction is needed, and the prepared
catalyst exhibits much higher HDO activity and selectivity for O-free products. Finally,
a possible mechanism was proposed to interpret the improved catalytic behavior of the
highly active Ni P-O/MCM41 catalyst.
2
Acknowledgments
The authors acknowledge the financial supports from the National Natural Science
Foundation of China (21276048), the Natural Science Foundation of Heilongjiang
Province (ZD201201).
References
[
[
[
1] G. Yang, E.A. Pidko, E.J.M. Hensen, J. Catal. 295 (2012) 122–132.
2] A. Oasmaa, S. Czernik, Energy Fuels 13 (1999) 914–921.
3] J. Wildschut, F.H. Mahfud, R.H. Venderbosch, H.J. Heeres, Ind. Eng. Chem. Res.
4
8 (2009) 10324–10334.
4] J.A. Cecilia, A. Infantes-Molina, E. Rodríguez-Castellón, A. Jiménez-López, S.T.
Oyama, Appl. Catal. B 136–137 (2013) 140–149.
[
[
[
5] E. Furimsky, Appl. Catal. A 199 (2000) 147–190.
6] L.D. Rollman, J. Catal. 46 (1977) 243–252.
[
[
7] J.B. Bredenberg, M. Huuska, P. Toropainen, J. Catal. 120 (1989) 401–408.
8] A.Y. Bunch, U.S. Ozkan, J. Catal. 206 (2002) 177–187.

[
[
[
[
(
[
[
[
[
9] A.Y. Bunch, W. Wang, U.S. Ozkan, J. Mol. Catal. 270 (2007) 264–272.
10] C.N. Satterfield, S.H. Yang, J. Catal. 81 (1983) 335–346.
11] Y. Romero, F. Richard, Y. Rene` me, S. Brunet. Appl. Catal. A 353 (2009) 46–53.
12] M.C. Edelman, M.K. Maholland, R.M. Baldwin, S.W. Cowley, J. Catal. 111
1988) 243–253.
13] H. Abe, A.T. Bell, Catal. Lett. 18 (1993) 1–8.
14] S. Ramanathan, S.T. Oyama, J. Phys. Chem. 99 (1995) 16365–36382.
15] N. Li, G.W. Huber, J. Catal. 270 (2010) 48–59.
16] A. Gutierrez, R.K. Kaila, M.L. Honkela, R. Slioor, A.O.I. Krause, Catal. Today
1
47 (2009) 239–246.
[
[
[
17] V.M.L. Whiffen, K.J. Smith, Energy Fuels 24 (2010) 4728–4737.
18] K. Li, R. Wang, J. Chen, Energy Fuels 25 (2011) 854–863.
19] R.H. Bowker, M.C. Smith, M.L. Pease, K.M. Slemkamp, L. Kovarik, M.E.
Bussell, ACS Catal. 1 (2011) 917–922.
[
[
[
(
[
20] H.Y. Zhao, D. Li, P. Bui, S.T. Oyama, Appl. Catal. A 391 (2011) 305–310.
21] O.I. Senol, T.R. Viljava, A.O.I. Krause, Appl. Catal. A 326 (2007) 236–244.
22] W.Y. Wang, K. Zhang, H. Liu, Z.Q. Qiao, Y.Q. Yang, K. Ren, Catal. Commun. 41
2013) 41–46.
23] X.P Duan, Y. Teng, A.J. Wang, V.M. Kogan, X. Li, Y. Wang, J. Catal. 261 (2009)
2
32–240.

[
24] H. Song, M. Dai, H.L. Song, X. Wan, X.W. Xu, Appl. Catal. A 462–463 (2013)
47–255.
25] H.I. Meléndez-Ortiz, L.A. García-Cerda, Y. Olivares-Maldonado, G. Castruita,
J.A. Mercado-Silva, Y.A. Perera-Mercado, Ceram. Int. 38 (2012) 6353–6358.
2
[
[
[
[
26] Y. Shu, Y.K. Lee, S.T. Oyama, J. Catal. 236 (2005) 112–121.
27] S.T. Oyama, X. Wang, Y.K. Lee, W.J. Chun, J. Catal. 221 (2004) 263–273.
28] V. Teixeira da Silva, L.A. Sousa, R.M. Amorimb, L. Andrini, S.J.A. Figueroa,
F.G. Requejo, F.C. Vicentini, J. Catal. 279 (2011) 88–102.
29] J.A. Cecilia, A. Infantes-Molina, E. Rodríguez-Castellón, A. Jiménez-López, J.
Catal. 263 (2009) 4–15.
[
[
[
30] K.A. Layman, M.E. Bussell, J. Phys. Chem. B 108 (2004) 10930–10941.
31] R.G. Li, Q.X. Guan, R.C. Wei, S.Q. Yang, Z. Shu, Y. Dong, J. Chen, W. Li, J.
Phys. Chem. C 119 (2015) 2557–2565.
32] T. I. Korányi, Z. Vít, D. G. Poduval, R. Ryoo, H. S. Kim, E. J. M. Hensen, J. Catal.
53 (2008) 119–131.
[
2
[
[
[
33] J.N. Kuhn, N. Lakshminarayanan, U.S. Ozkan, J. Mol. Catal. A 282 (2008) 9–21.
34] L. Song, S. Zhang, Q. Wei, Catal. Commun. 12 (2011) 1157–1160.
35] H. Song, J. Wang, Z.D. Wang, H.L. Song , F. Li, Z.S. Jin, J. Catal. 311 (2014)
2
57–265.
[
[
36] P. Liu, J.A. Rodriguez, Y. Takahashi, K. Nakamura, J. Catal. 262 (2009) 294–303.
37] K. Sutthiumporn, S. Kawi, Int. J. Hydrogen Energ. 36 (2011) 14435–14446.

[
[
[
[
38] C.L. Lee, D.F. Ollis, J. Catal. 87 (1984) 325–331.
39] J.S. Moon, E.G. Kim, Y.K. Lee , J. Catal. 311 (2014) 144–152.
40] Y.K. Lee, Y. Shu, S.T. Oyama, Appl. Catal. A 322 (2007) 191–204.
41] S.T. Oyama, J. Catal. 216 (2003) 343–351.
List of Tables
Tab. 1 Properties of catalyst samples
Tab. 2 Spectral parameters obtained by XPS analysis
Tab.3 proportion of each species for the Ni2p and P2p spectra obtained by XPS analysis
List of Figures
Fig. 1 XRD patterns of MCM-41, Ni P-O/MCM41and Ni P-N/MCM41
2
2
Fig. 2 Nitrogen adsorption/desorption isotherms (a) and BJH pore size distributions (b)
for MCM-41, Ni P-O/MCM41 and Ni P-N/MCM41
2
2
Fig. 3 TEM spectra of Ni P-N/MCM41 (a), Ni P-O/MCM41 (b)
2
2
Fig. 4 XPS spectra in the Ni(2p), P(2p) and O(1s) regions for Ni P-N/MCM41 and
2
Ni P-O/MCM41 catalysts (a) Ni 2p core level spectra, (b) P 2p core level spectra, (c) O
2
1
s core level spectra
Fig. 5 Conversion of BF HDO over Ni P-O/MCM41, Ni P-N/MCM41 and
2
2
Ni P-N/MCM41-nr catalysts (Reaction temperature=493–573 K, pressure=3.0 MPa,
2
–
1
WHSV=4.0 h , and hydrogen/oil ratio=500 (V/V))

Fig. 6 Product selectivity of BF HDO over Ni P-N/MCM41 and Ni P-O/MCM41
2
2
catalysts (a) Ni P-N/MCM41, (b) Ni P-O/MCM41
2
2
Fig. 7 Yields of O-free compounds over Ni P-O/MCM41 and Ni P-N/MCM41
2
2
–
1
catalysts (Reaction temperature=493–573 K, pressure=3.0 MPa, WHSV=4.0 h , and
hydrogen/oil ratio=500 (V/V))
Fig. 8 Possible mechanism of HDO of BF over highly active Ni P-O/MCM41 catalyst
2
Fig. S1 Product selectivity of BF HDO over Ni P-N/MCM41-nr catalyst
2
List of Schemes
Sch.1 Reaction pathways for HDO of BF over Ni P-N/MCM41catalyst
2
Sch. 2 Reaction pathways for HDO of BF over Ni P-O/MCM41catalyst
2
Sch. 3 Proposed reaction schematic of 2-EtPh HDO over Ni P catalysts
2
Tab. 1 Properties of catalyst samples
Conversion
a
b
SBET
VBJH
3
D
dXRD
CO uptake
TOF
(s )
(%)
Sample
-1
2
-1
-1
–1
(
m ·g )
(cm ·g )
(nm)
(nm)
(µmol·g )
4
93 K
MCM-41
1051
502
0.84
0.37
0.37
3.2
2.9
2.9
—
24
20
—
31
53
—
31
57
—
Ni P-N/MCM41
0.0016
0.0018
2
Ni P-O/MCM41
509
2

a
D : pore diameter, D≈4VBJH/SBET
.
b
dXRD : Calculated from the Dc = Kλ/β cos(θ) (Scherrer equation) based on the Ni P {1
2
1
1}.
Tab. 2 Spectral parameters obtained by XPS analysis
Binding energy (eV)
Superficial
atomic ratio
Ni/P/O
Sample
Ni 2p3/2
P 2p
O 1s
2
+
3-
-
-
Ni
Ni2P
852.3
852.4
852.5
851.7
PO4
H PO3
Ni2P
129.2
129.2
129.3
129.3
OH
H O
2
2
Ni P-N/MCM41-fresh
2
856.3
856.3
856.7
857.4
134.6
134.8
134.8
134.5
133.7
133.8
133.7
—
532.4
532.7
532.7
—
—
—
1/3.16/3.40
1/2.54/4.99
1/3.11/4.48
1/2.15/5.27
Ni P-O/MCM41-fresh
2
Ni P-N/MCM41-spent
2
—
Ni P-O/MCM41-spent
2
533.2
Tab.3 proportion of each species for the Ni2p and P2p spectra obtained by XPS analysis
Ni (%)
P (%)
Sample
2
+
3-
-
-
Ni
Ni2P
40.6
PO4
H PO3
H PO2
Ni2P
0.9
2
2
Ni P-N/MCM41-fresh
59.4
28.1
41.8
29.2
2

Ni P-O/MCM41-fresh
94.3
68.7
75.3
5.7
49.4
77.5
61.7
49.7
21.6
31.6
0
0
0
0.9
0.9
6.7
2
Ni P-N/MCM41-spent
2
31.3
24.7
Ni P-O/MCM41-spent
2
Tab. 1 Properties of catalyst samples
Conversion
a
b
SBET
VBJH
3
D
dXRD
CO uptake
TOF
(s )
(%)
Sample
-1
2
-1
-1
–1
(
m ·g )
(cm ·g )
(nm)
(nm)
(µmol·g )
4
93 K
MCM-41
1051
502
0.84
0.37
0.37
3.2
2.9
2.9
—
24
20
—
31
53
—
31
57
—
Ni P-N/MCM41
0.0016
0.0018
2
Ni P-O/MCM41
509
2
a
D : pore diameter, D≈4VBJH/SBET
.
b
dXRD : Calculated from the Dc = Kλ/β cos(θ) (Scherrer equation) based on the Ni P {1
2
1
1}.
Tab. 2 Spectral parameters obtained by XPS analysis

Binding energy (eV)
Superficial
atomic ratio
Ni/P/O
Sample
Ni 2p3/2
P 2p
O 1s
2
+
3-
-
-
Ni
Ni2P
852.3
852.4
852.5
851.7
PO4
H PO3
Ni2P
129.2
129.2
129.3
129.3
OH
H O
2
2
Ni P-N/MCM41-fresh
2
856.3
856.3
856.7
857.4
134.6
134.8
134.8
134.5
133.7
133.8
133.7
—
532.4
532.7
532.7
—
—
—
1/3.16/3.40
1/2.54/4.99
1/3.11/4.48
1/2.15/5.27
Ni P-O/MCM41-fresh
2
Ni P-N/MCM41-spent
2
—
Ni P-O/MCM41-spent
2
533.2
Tab.3 proportion of each species for the Ni2p and P2p spectra obtained by XPS analysis
Ni (%)
P (%)
Sample
2+
3-
-
-
Ni
Ni2P
PO4
H PO3
H PO2
2
Ni2P
2
Ni P-N/MCM41-fresh
59.4
94.3
68.7
75.3
40.6
5.7
28.1
49.4
77.5
61.7
41.8
49.7
21.6
31.6
29.2
0.9
0.9
0.9
6.7
2
Ni P-O/MCM41-fresh
2
0
0
0
Ni P-N/MCM41-spent
2
31.3
24.7
Ni P-O/MCM41-spent
2