Comparison of phosphide catalysts prepared by temperature-programmed reduction and liquid-phase methods in the hydrodeoxygenation of 2-methylfuran

Article abstract of DOI:10.1016/j.apcata.2017.06.009

Two types of Ni₂P/SiO₂ were prepared by different methods, a conventional method involving temperature-programmed reduction (TPR) and a liquid-phase synthesis (LP). Both methods resulted in phase-pure materials with the Ni₂P hexagonal phase, but the sample prepared by TPR had lower particle size (7 nm) than the sample prepared by LP synthesis (25 nm). The samples were tested in the gas-phase hydrodeoxygenation of 2-methylfuran (2-MF) in H₂ at 0.5 MPa and 250–450 °C, and it was found that the Ni₂P/SiO₂-TPR gave higher conversions and turnover frequencies than the Ni₂P/SiO₂-LP. It was conjectured that the difference in activity was due to the higher phosphorus content of the sample prepared by TPR, and this was confirmed by measurements of the P/Ni ratio by x-ray photoelectron spectroscopy. The main products formed by the two catalysts were 2-pentanone and n-pentane and small amounts of 1-pentanol and 2-pentanol. Contact time measurements on the Ni₂P/SiO₂-TPR sample indicated that 2-pentanone was a primary product, the pentanols were intermediary compounds, and the n-pentane was a final product. Kinetic measurements on the Ni₂P/SiO₂-LP revealed orders of 0.6 for the 2-MF and 0.5 for H₂, consistent with adsorption of both reactants on the catalyst and a surface reaction between the adsorbed species. TPR was also employed to prepare a Cu₃P/SiO₂ sample. Although a phase-pure material was formed, the substance showed very low activity for the HDO of 2-MF.

Full text of DOI:10.1016/j.apcata.2017.06.009

Accepted Manuscript
Title: Comparison of Phosphide Catalysts Prepared by
Temperature-Programmed Reduction and Liquid-Phase
Methods in the Hydrodeoxygenation of 2-Methylfuran
Authors: Jianwen Zhang, Kazuki Matsubara, Gwang-Nam
Yun, Huidong Zheng, Atsushi Takagaki, Ryuji Kikuchi, S. Ted
Oyama
PII:
S0926-860X(17)30259-4
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.apcata.2017.06.009
APCATA 16270
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
31-3-2017
2-6-2017
3-6-2017
Please cite this article as: Jianwen Zhang, Kazuki Matsubara, Gwang-Nam Yun,
Huidong Zheng, Atsushi Takagaki, Ryuji Kikuchi, S.Ted Oyama, Comparison of
Phosphide Catalysts Prepared by Temperature-Programmed Reduction and Liquid-
Phase Methods in the Hydrodeoxygenation of 2-Methylfuran, Applied Catalysis A,
Generalhttp://dx.doi.org/10.1016/j.apcata.2017.06.009
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.

Comparison of Phosphide Catalysts Prepared by Temperature-Programmed
Reduction and Liquid-Phase Methods in the Hydrodeoxygenation of 2-Methylfuran
Jianwen Zhang1,2, Kazuki Matsubara1, Gwang-Nam Yun1, Huidong Zheng1,2, Atsushi
Takagaki1,2, Ryuji Kikuchi1, S. Ted Oyama1,2,3*
1The University of Tokyo
Department of Chemical Systems Engineering
7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
2Fuzhou University, College of Chemical Engineering, Fuzhou 350116, China
3Virginia Tech
Department of Chemical Engineering
Blacksburg, VA 24061
* Corresponding author.
E-mail address: oyama@vt.edu (S. Ted Oyama).

Graphical Abstract
Research Highlights
 Two Ni2P/SiO2 samples were prepared by temperature-programmed reduction (TPR) and
liquid-phase (LP) methods
 The samples were active in converting 2-methylfuran (2-MF) at 523-723 K and 0.5
MPa
 The TP sample had higher hydrodeoxygenation (HDO) activity than the LP sample
 X-ray photoelectron spectroscopy shows higher phosphorus content for the TP sample
 A Cu3P/SiO2 catalyst prepared by temperature-programmed reduction had low HDO
activity
Abstract
Two types of Ni2P/SiO2 were prepared by different methods, a conventional method
involving temperature-programmed reduction (TPR) and a liquid-phase synthesis (LP).
Both methods resulted in phase-pure materials with the Ni2P hexagonal phase, but the
sample prepared by TPR had lower particle size (7 nm) than the sample prepared by LP
synthesis (25 nm). The samples were tested in the gas-phase hydrodeoxygenation of 2-
methylfuran (2-MF) in H2 at 0.5 MPa and 250-450 ˚C, and it was found that the
Ni2P/SiO2-TPR gave higher conversions and turnover frequencies than the Ni2P/SiO2-LP.
It was conjectured that the difference in activity was due to the higher phosphorus content
of the sample prepared by TPR, and this was confirmed by measurements of the P/Ni
ratio by x-ray photoelectron spectroscopy. The main products formed by the two
catalysts were 2-pentanone and n-pentane and small amounts of 1-pentanol and 2-
pentanol. Contact time measurements on the Ni2P/SiO2-TPR sample indicated that 2-
pentanone was a primary product, the pentanols were intermediary compounds, and the
n-pentane was a final product. Kinetic measurements on the Ni2P/SiO2-LP revealed
2

orders of 0.6 for the 2-MF and 0.5 for H2, consistent with adsorption of both reactants on
the catalyst and a surface reaction between the adsorbed species. TPR was also
employed to prepare a Cu3P/SiO2 sample. Although a phase-pure material was formed,
the substance showed very low activity for the HDO of 2-MF.
Keywords:
Nickel phosphide; Liquid-phase synthesis; 2-Methylfuran
hydrodeoxygenation; Reaction network; Kinetics; Cu3P/SiO2
Introduction
Great advances have been made in the utilization of methane in natural gas for the
production of fuels and chemicals [1,2], but renewable resources are desirable for the
future. Lignocellulosic biomass, which is an abundantly available raw material, has
attracted much attention as a possible resource [3]. The use of biomass instead of fossil
fuels is of interest to help solve the problem of global warming because biomass produces
low net CO2 emissions, is highly abundant on the earth, and can be converted into liquid
fuels, which are more useful for transportation and storage than electrical energy.
However, the energy density of lignocellulosic biomass is low, and so to convert it to a
liquid fuel suitable for transportation, considerable processing is required. [4]. Fast
pyrolysis is one of the processing methods, in which lignocellulosic biomass is rapidly
heated in the absence of oxygen and converted into a liquid known as pyrolysis oil or bio-
oil [5,6]. Bio-oil is expected to be an eco-friendly and easily available liquid fuel, but
actually it is not directly usable because of its high content of oxygenates, typically 40-
60%, The oxygen is in the form of phenols, furans, carboxylic acids, ethers, and aromatic
alcohols, and these oxygen groups result in poor properties for a liquid fuel such as low
heating value and poor stability [7]. Therefore, hydrodeoxygenation (HDO), the catalytic
removal of oxygen, has been studied for the upgrading of bio-oil [8,9].
Many kinds of catalysts have been investigated for HDO [10,11].
Hydrodesulfurization (HDS) [12,13] and hydrodenitrogenation (HDN) [14] catalysts
have been applied in HDO processes because the reactions for the removal of sulfur and
3

nitrogen are similar to those for the removal of oxygen. Transition metal sulfide catalysts,
such as supported NiMo and CoMo, are well known as highly activity catalysts [15] for
HDS and HDN. Although they also have moderate activity for HDO, they require sulfur
in the feed to maintain them in their active sulfide form. A problem is that bio-oil does
not contain high sulfur levels [16].
Noble metal catalysts such as Ru, Rh, Pd [17,18], and Pt [19] have also been
investigated for HDO [20]. A study of supported noble metal catalysts (Ru/C, Ru/TiO2,
Ru/Al2O3, Pt/C, and Pd/C) indicated that Ru/C catalyst was better than the classical HDO
catalysts (sulfided NiMo/Al2O3 and CoMo/Al2O3). Another study on zirconia supported
noble metal catalysts among them (Rh/ZrO2, Pd/ZrO2 and Pt/ZrO2) showed that these
catalysts were active at 300 °C in the HDO of guaiacol (GUA), which was used as a model
compound of bio-oil. However, noble metals have problems of high cost and the
occurrence of carbon deposition [21].
Transition metal phosphide catalysts such as FeP, Fe2P, CoP, MoP [22], WP, Ni2P [23]
and NiP have been reported to be highly active for HDS [24,25] and HDN [26,27] and
have been applied in HDO in the last decade [28,29]. A study of metal phosphide
catalysts showed that these catalysts are active for dibenzothiophene HDS and quinoline
HDN, with the order of activity being Fe2P<CoP<MoP<WP<Ni2P. The best catalyst,
Ni2P/SiO2, was superior to the standard HDS catalyst, sulfided Ni–Mo/Al2O3. In another
study, it was also reported that supported transition metal phosphide catalysts, Ni2P/SiO2
and CoP/SiO2, were better than the noble metal catalyst, Pd/Al2O3 in the HDO of
gamma-valerolactone (GVL), which is used as a model compound of bio-oil [30].
Previous studies have reported that nickel phosphide had high activity for the HDO
of various compounds such as anisole [31], guaiacol [32,33], 2-methyltetrahydrofuran
[34,35], 2-methylpyran [36], and actual bio-oil [37], and that the activity was affected by
the method of synthesis [38]. A previous study [35] on the HDO of 2-
methyltetrahydrofuran which was used as a model compound of bio-oil, showed that
Ni2P/SiO2 had as high activity for HDO as noble metal catalysts and that the total HDO
conversion was almost 100 %. This study also reported that on Ni2P/SiO2, the main
product was n-pentane (75%), following by n-butane (14%), 2-pentanone (4.9%), 2-
pentanol (2.6%) and 1-pentanol (3.6%) and that the amount of n-pentane was much higher
than in other catalysts. This result means that Ni2P/SiO2 is highly active only for
4

hydrodeoxygenation process, not for C-C cracking. The studies reported here address the
HDO of 2-methylfuran, a precursor mainly derived from hemicellulose, whose
concentration in bio-oil is higher than that of 2-methyltetrahydrofuran [39]. The
compounds are shown in Fig. 1. For use as catalysts, two samples of Ni2P/SiO2 were
prepared by two different methods. The methods were temperature-programmed
reduction (TPR) of a phosphate precursor (Ni2P/SiO2-TPR) and liquid-phase (LP)
phosphidation of an acetylacetonate precursor (Ni2P/SiO2-LP). The liquid-phase
method is relatively new, and previous work will be described in the Results and
Discussion section. Furthermore, another transition metal phosphide, Cu3P/SiO2,
prepared by TPR of a hydroxide precursor is compared. Copper has similar properties
as nickel catalysts and has been used for the steam reforming of methanol and catalytic
hydrogenation [40], but has not been reported for HDO reactions. The only studies of
Cu3P have been in the electrocatalysis field [41,42], so the activity of Cu3P/SiO2 was also
tested.
Experimental
Materials
A fumed silica support (Cab-osil ® EH5) provided by Cabot Corp. was used as the
support material. Chemicals used for the catalyst syntheses were Ni(OH)2 (Sigma-
Aldrich), Cu(OH)2 (Sigma-Aldrich), H3PO3 (Sigma-Aldrich, 99.0%), trioctylphosphine
(TOP, Aldrich, 95%), nickel acetylacetonate [Ni(acac)2, 97%] 1-octadecene (Aldrich,
90%). The chemicals used for the reaction studies were 2-methylfuran (TCl, >98.0%),
and toluene (Wako, >99.0%). The chemicals used for calibration of the gas
chromatograph were 2-methyltetrahydrofuran (TCl, >98.0%), 2-pentanone (TCl,
>97.0%), n-pentane (TCl, >99.5%), 1-pentanol (Sigma-Aldrich, >99.0%) and 2-pentanol
(Sigma-Aldrich, 99.0%).
Preparation of Catalysts
Both Ni2P/SiO2-TPR and Cu3P/SiO2 were prepared by a temperature-programed
reduction (TPR) method from supported phosphate precursors. The phosphate precursors
were made by the impregnation of the dried silica support to the incipient wetness point
5

(about 2 cm3 of solution per gram of support) with a mixed solution with Ni(OH)2 for
Ni2P/SiO2 and Cu(OH)2 for Cu3P/SiO2, and H3PO3. For 2 grams of silica, the amounts of
Ni(OH)2 (0.207 g) or Cu(OH)2 (0.218 g) corresponded to 1.116 mmol/g-support and the
amount of H3PO3 (0.366 g) was 2.232 mmol/g-support. The precursors were obtained
after drying the impregnated solid at 100˚C overnight, followed by pelletization and
sieving to a size of 650-1180 µm. Quantities of 1.50 grams of the precursor were reduced
in a flow of 1500 ml/min H2 (1000 ml/min per gram of precursor) from room temperature
to 650˚C for Ni2P/SiO2 and 580˚C for Cu3P/SiO2 at a heating rate of 3 ˚C/min and
maintained at the final temperature for 2 hours. Large flow rates were used in the
reduction because previous studies had shown that product water vapor inhibited the
synthesis of the phosphides [38,43].The prepared catalysts were cooled to room
temperature in N2 and then passivated in a 0.5 vol% O2/N2 flow for 2 hours.
The Ni2P/SiO2–LP was prepared by a liquid-phase method. First a precursor was
obtained by mixing with 1.156 mmol of nickel acetylacetonate (Ni(acac)2), 8 mL (17.3
mmol) of trioctylphosphine (TOP) and 10 mL of 1-octadecene (solvent), and then heating
the components in Ar to 60 ˚C to dissolve the solids under strong stirring to form a green
solution. Following this, 1g of silica (EH-5) was added and the mixture stirred for 1 h,
after which the temperature was raised to 340 ˚C (in other experiments to 330 and 317
˚C) always under Ar. On heating, the suspension turned black in several minutes.
After reaction for 2 h the mixture was cooled to 50 ˚C, and was mixed with excess ethanol.
The solid product was separated with a high pressure filter. The sample was sonicated in
hexane to remove organics bound to the surface and precipitated with excess ethanol. This
sonication-precipitation cycle was repeated at least two times to remove as much of the
bound organics as possible from the system. The majority of the excess phosphorus was
removed from the sample by the filtering and washing procedure, as the use of smaller
amounts of TOP resulted in incomplete phosphidation. Then, the solid was dried at 80
˚C overnight. The samples were not passivated after synthesis and were characterized
by x-ray diffraction measurements after removal from the drying oven. They were
reduced in hydrogen for the chemisorption and reactivity measurements to be described
below.
Characterization of Catalysts
6

The reduction properties of the samples were characterized by temperature-
programed reduction (TPR) (TPD-1-AT, BEL Japan). After pretreatment in nitrogen at
300 ˚C for 2 hours and cooling down, 1.0 gram of the precursor was heated up to 700 ˚C
at a heating rate 2 ˚C/min with a H2 flow (1000 ml/min). The H2O product was monitored
by a quadrupole mass spectrometer.
X-ray diffraction (XRD) measurements were conducted to determine the crystal
structure of the catalysts with a XRD (RINT-2700, Rigaku) diffractometer operated at
40 kV and 100 mA, using Cu Kα monochromatized radiation (λ = 0.154178 nm).
Crystallite size was calculated by the Scherrer equation;
퐾휆
퐷푐 =
훽cos(휃)
In this equation K is the shape factor with a value of 0.9, λ is the X-ray wavelength with
a value of 0.154178 nm, β is the line broadening at half the maximum intensity, and 2θ is
the Bragg angle.
The quantity of active sites on the catalysts was estimated by CO uptake
measurements for the Ni2P catalysts and H2 uptake for the Cu3P catalyst. After reducing
the catalysts in H2 at 450 ˚C for 2 hours and cooling them in He, calibrated pulses of 3%
CO/He or H2 were injected over the catalysts. The CO uptake and H2 uptake were
calculated by the diminished area of the peak of the signal of mass 28 (CO) or mass 2
(H2) on a mass spectrometer caused by adsorption of CO or H2 onto the catalysts surface.
The X-ray photoelectron spectroscopy (XPS) measurements were carried out by a
PHI 5000 Versaprobe with a MgKα X-ray source (1486.7 eV). The XPS spectra were
collected using a pass energy of 117.4 eV for wide-scanning measurements and 23.5 eV
for narrow-scanning measurements. For XPS measurement, the samples were ground to
powders and then pelletized by pressurizing to 30 MPa for 20 min with 0.05 g of sample.
The samples denoted as “fresh” were measured after reduction of the prepared samples
in H2 at 450 ˚C (in the same manner as for the reactivity measurements) followed by
passivation in 0.5 % O2/N2. The samples denoted as “spent” were simply removed from
the reactor after the reactivity measurements. In all cases the samples were handled in
laboratory air so were subject to oxidation and to exposure to carbon contamination.
7

However, the goal of the measurements was to assess the phosphorus content, not the
oxidation state of the nickel or the coverage with organic compounds, so the contact with
the atmosphere was immaterial.
Catalyst Reactivity Studies
Hydrodeoxygenation of 2-methylfuran was carried out in a continuous-flow reactor
in the gas phase and a schematic of the reaction system is shown in Figure 2. A mixture
of 95 vol% of 2-methylfuran as reactant and 5 vol% of toluene as an internal standard
was vaporized at 35˚C in a bubbler through which H2 was passed. For the utilized flow
rate of 100 cm3min-1 this gave a concentration of 2-MF 32.6 mol% and a molar flow rate
of 24 μmol s-1. The reaction gas was introduced to a reactor bed and then the product gas
was analyzed using an on-line gas chromatograph (GC; GC-2014, Shimazu) equipped
with a flame ionization detector (FID) and gas chromatograph mass spectrometer (GC-
MS; GCMS-QP2010Ultra, Shimazu). The passivated Ni2P/SiO2 or Cu3P/SiO2 catalysts
were pelletized to a size of 650-1180 µm and loaded in the middle of the reactor where a
sheathed thermocouple was also located. The amount of catalyst was set to give a
predetermined contact time decided in advance. The contact time was calculated by the
following equation:
[
]
Active sites μmol × Catalyst weight [g]
Contact time [s] =
⁄
Reactant molar flowrate [μmol s]
Again the active sites were estimated by CO uptake measurements for the Ni2P
catalysts and H2 uptake for the Cu3P catalyst. The passivated Ni2P/SiO2 or Cu3P/SO2
catalysts were pretreated in a H2 flow (150 ml/min) at 450˚C for 2 hours before reaction
tests. The conversion of 2-methylfuran (X2MF), the selectivity to products (Si), and the
TOF were defined as follows:
푛푀퐹
푋2푀퐹 = (1 −
) × 100 %
) × 100 %
푛푀퐹,0
푛ꢀ
푆ꢀ = (
푛푀퐹,0 − 푛푀퐹
8

(
⁄
)
2푀퐹
푅푒푎푐푡푎푛푡 푓푙표푤 푟푎푡푒 [휇푚표푙 푠] 푋
푇푂퐹 =
(
)( )
퐴푐푡ꢀ푣푒 푠ꢀ푡푒푠 [휇푚표푙/푔] 퐶푎푡푎푙푦푠푡 푤푒ꢀ푔ℎ푡 [푔]
in which 푛푀퐹,0 and 푛푀퐹 are the moles of 2-methylfuran in the feed and product,
respectively, and 푛ꢀ is the moles of product i.
To test the temperature dependence of the catalysts and compare the activity of the
catalysts, quantities of 0.35 g of Ni2P/SiO2-TPR (corresponding to 19 μmol of
chemisorption sites) or 0.40 g of Cu3P/SiO2, (corresponding to 14 μmol of chemisorption
sites) were loaded in the reactor. After pretreatment, the pressure was fixed at 0.5 MPa
and the H2 flow rate was set to 0.78 s contact time, which was 100 ml/min for Ni2P/SiO2–
TPR and 76 ml/min for Cu3P/SiO2. The reaction was conducted in the range of 250-
450˚C, with the temperature varied as follows: 450 °C → 350 °C → 250 °C → 300 °C
→ 400 °C. Each temperature was held for 2-3 hours. The temperature was cycled down
and up again to demonstrate that a smooth conversion curve was obtained, and thus, that
the catalyst was stable during the measurements.
For activity testing of the Ni2P/SiO2–LP sample, an amount of catalyst of 0.543 g
(corresponding to 19 μmol of CO uptake sites) were loaded in the reactor. As before,
the pressure was fixed at 0.5 MPa and the H2 flow rate was 100 ml/min to give a contact
time of 0.78 s, and the same temperature variation schedule was used.
For the effect of reactants partial pressure, reaction experiments were conducted
changing H2 or 2-MF partial pressures using N2 as the balance. A quantity of 0.362 g of
Ni2P/SiO2-LP was used and the total pressure and the temperature were fixed at 0.5 MPa
and 350 °C. To check the effect of H2 partial pressure (contact time 1.1 s), the 2-MF flow
9

rate and partial pressure were fixed 7.9 μmol/s and 35 kPa and the H2 partial pressure was
varied in the order: 470, 376, 282, 235, 188 , 94 kPa by changing H2 flow rate. To check
the 2-MF partial pressure effect, H2 flow rate and partial pressure were fixed at 18 ml/min
and 282 kPa, and the 2-MF partial pressure was varied in the order; 18.5, 23.1, 28.7, 35.3
(contact time 1.1 s), and 43.0 kPa. N2 was used as balance to keep the total pressure at
0.5 MPa. The kinetic data was fit to rate expressions using the non-linear fitting
program in POLYMATH software.
For contact time studies, a certain amount of Ni2P/SiO2-TPR catalyst was loaded in
the reactor. After pretreatment, the pressure was fixed at 0.5 MPa and the temperature
was set to 300 ˚C. The contact time was varied in a range of 0.37-1.5 s.
Results and Discussion
Catalyst synthesis and characterization
The synthesis of phosphide catalysts by temperature-programmed reduction was first
reported by the group of Oyama [24]. For the synthesis using TPR the reduction
temperatures of the phosphide precursors were determined on small amounts of samples
by carrying out the reduction to high temperature. A reduction feature appears at around
600˚C for Ni2P/SiO2-TPR and around 550 ˚C for Cu3P/SiO2, and from these the reduction
temperatures were set to 650 ˚C to prepare Ni2P/SiO2-TPR and 580 ˚C to form Cu3P/SiO2.
The TPR profile for Ni2P/SiO2-TPR has been reported before [32,33,48] and just the
reduction to form Cu3P/SiO2 is shown in Fig. 3, which displays the mass 18 (H2O) signal
as a function of temperature.
The XRD patterns for Ni2P/SiO2-TPR and Cu3P/SiO2 are shown in Fig. 4. . The broad
feature at at around 22˚ is due to the amorphous SiO2 support. The Ni2P/SiO2-TPR shows
weak signals at around 41˚, 44˚, 47˚ and 55˚, which match the diffraction lines of Ni2P.
The pattern for Cu3P/SiO2 had peaks at around 36˚, 39˚, 41˚, 45˚, 46˚ and 47˚, and those
correspond to the lines of Cu3P. The lines were sharper for Cu3P/SiO2 than for
10

Ni2P/SiO2-TPR indicating higher crystallinity and crystallite size for the former.
Crystallite sizes were calculated by Scherrer’s equation and gave sizes of 14 nm for
Ni2P/SiO2-TPR and 53 nm for Cu3P/SiO2.
The synthesis of phosphides by liquid phase methods has been studied extensively as
an alternative to the TPR method. Qian and co-workers demonstrated the production of
CoP, Co2P, Ni2P, Cu3P [44], and FeP [45] by a solvothermal route. However, the
method was difficult to control and produced aggregated nanoparticles with a large degree
of polydispersity. Yao, et al. reported the synthesis of transition metal phosphide
nanoparticles including MoP, WP, Ni2P, FeP, Fe2P, and CoP by reduction of phosphates
[46], and the problem of aggregation was solved by the group of Brock using supports
[47]. It has been reported that trioctylphosphine (TOP) is an effective reagent for liquid
phase synthesis because it can act as a coordinating solvent as well as a source of
phosphorus [48,49]. Different morphologies such as FeP nanorods have been obtained by
injecting iron carbonyl at high temperature into a mixture of TOP and trioctylphosphine
oxide (TOPO) [50,51]. Hyeon and coworkers have obtained uniform nanorods of Fe2P,
MnP, Co2P, FeP, and Ni2P [52] by using a syringe pump. In the present study a simple
one-pot method was developed to prepare supported Ni2P.
The XRD patterns of the nickel phosphide samples prepared in the liquid phase by
carrying out the phosphidation reaction at different temperatures are shown in Fig. 5.
The sample prepared at 340 ˚C shows the diffraction lines corresponding to Ni2P, while
the samples prepared at 330 and 317 ˚C show additional lines due to Ni12P5, a phase that
is slightly phosphorus deficient. For further experiments the phase-pure sample
obtained in the liquid phase at 340 ˚C was used and is denoted as Ni2P/SiO2-LP. The
coexistence of the Ni12P5 phase in the other samples indicates that the conditions were
barely sufficient for complete phosphidation, and may indicate some phosphorus
deficiency even in the sample prepared at 340 ˚C. This will be confirmed later with X-
ray photoelectron spectroscopy.
The samples were characterized by chemisorption of CO and H2 to give estimates of
11

the amount of active sites on the Ni2P/SiO2 and Cu3P/SiO2 catalysts. The quantities are
summarized in Table 1. From these, the particle sizes and the active surface areas of
nickel and copper were calculated by the following equations:
푁퐴퐿
S =
푛̅푓
6
D =
푆휌
In these equations S is the theoretical surface area, D is the particle size, 푛̅ is the density
of metal atoms from the crystal structure, 푓 is the fractional weight loading of the
samples, 푁퐴 is Avogadro constant, L is the metal site density of the solids, and 휌 is 7.09
g 푐푚−3 for Ni2P [29] and 7.20 g 푐푚−3 for Cu3P, which was calculated from the lattice
parameters (a=b=6.996 and c=7.232 γ = 120˚) [53]. The value of L for metal phosphides
is known to be close to 1× 1015 atoms 푐푚−2 [54].
The calculated results are shown in Table 1. The active surface area (S) is 42 m2 g-1 for
Ni2P/SiO2-TPR, 26 m2 g-1for Ni2P/SiO2-LP, and 55 m2 g-1 for Cu3P/SiO2 . The theoretical
particle sizes (D) from the chemisorption values are 7 nm for Ni2P/SiO2-TPR, 25 nm for
Ni2P/SiO2-LP, and 40 nm for Cu3P/SiO2. The crystallite sizes from the X-ray diffraction
line-broadening calculated with the Scherrer equation are 14 nm for Ni2P/SiO2-TPR, 30
nm for Ni2P/SiO2-LP, and 53 nm for Cu3P/SiO2. These are in reasonable agreement with
the chemisorption particle sizes.
Reactivity results
Figure 6 shows the conversion of 2-methylfuran on Ni2P/SiO2-TPR at 250-450 ˚C at
0.5 MPa with 19 μmol of sites loaded in the reactor and a contact time of 0.78 s. The
conversion on Ni2P/SiO2 increased steadily with rising temperature and reached more
than 99% at 450 ˚C. The main products were 2-pentanone and n-pentane, with small
amounts of 1-pentanol and 2-pentanol formed. The selectivity to 2-pentanone started
high at 70 mol% went through a shallow maximum and then decreased at higher
temperature. The selectivity to n-pentane started low, dipped to a slight minimum, and
12

then increased at higher temperature. The selectivity to the alcohols was very low,
below 3 mol% and passed through a maximum at intermediate temperatures. The
conversion on Cu3P/SiO2 was much lower and reached only 3% at 450 ˚C and is not
shown.
Figure 7 shows the conversion of 2-methylfuran on Ni2P/SiO2-LP at identical
conditions as for the TPR sample of 250-450 ˚C at 0.5 MPa with 19 μmol of sites loaded
in the reactor and a contact time of 0.78 s. In this case the LP sample showed a marked
loss of activity with time, and stable conversions were obtained after 12-18 h. The data
shown in Fig. 7 for the LP sample were obtained after this stabilization period. The
conversions at this point were notably lower than in the case of the Ni2P prepared by the
TPR method. The main products were still 2-pentanone, n-pentane, 1-pentanol, and 2-
pentanol, but the selectivity patterns were markedly different. For the Ni2P/SiO2-LP, the
selectivity to 2-pentanone rose with increasing temperature while that of Ni2P/SiO2-TPR
had decreased. Also for the Ni2P/SiO2-LP the selectivity to n-pentane fell with rising
temperature while that of Ni2P/SiO2-TPR had decreased. Meanwhile, the selectivities
to the alcohols, although also low, decreased through the temperature interval instead of
going through a maximum.
It should be noted that the reactivity of 2-methylfuran (2-MF) was considerably lower
than that of the related saturated compound 2-methyltetrahydrofuran (2-MTHF), whose
structure was shown earlier (Fig. 1). Activity studies demonstrated that at the same
pressure of 0.5 MPa full conversion was achieved at 325 oC over Ni2P/KUSY (21 μmol
of sites) [55] and at 350 ˚C over Ni2P/SiO2 (16 μmol of sites) [32]. The samples above
used for the 2-methyltetrahydrofuran studies were prepared by the TPR method.
The
method was identical to that employed in the present study. For Ni2P/SiO2-TPR (19 μmol
of sites) full conversion of 2-MF was achieved at 450 ˚C, a full 100 ˚C higher than for 2-
MTHF. This indicates that the first steps in the transformation of 2-methylfuran, likely
the hydrogenation steps, are difficult.
The differences in activity and selectivity indicate substantial divergence of catalytic
properties between the Ni2P samples prepared by the TPR and LP methods, and there are
two possible explanations for this. One is an ensemble effect, or structure-sensitivity,
13

due to the different crystallite sizes of the samples, and the other is a composition effect
due to differences in the P content of the samples. The differences in activity are likely
not due to an ensemble effect, which would result from particle size changes producing
different exposed surface planes. This has been shown not to be important for the HDO
of 2-MTHF [49]. Rather than surface structure, it is likely that the differences in activity
are due to the differences in P content. In the preparation of the Ni2P/SiO2-TPR sample,
excess phosphorus (Ni:P = 1:2) was used and was retained in the sample, while in the
Ni2P/SiO2-LP sample, excess phosphorus was also used (Ni:P = 1:15), but most of it was
washed away. Preparation of LP samples with smaller amounts of phosphorus (Ni:P =
1:10) resulted in mixtures, indicating that phosphorus takeup is not efficient in the liquid-
phase synthesis. The presence of oleylamine produces Ni5P4, as also reported recently
[49]. In the TPR sample it is likely that the extra phosphorus was retained on the support,
as there was no washing, and this extra P served as a replenishment source. The high P
levels indicated by XPS support this interpretation. In the LP sample a notable loss of
initial activity was shown, with stable activity obtained only after 12-18 h. The lack of
high and stable initial activity in the LP sample is probably due to a phosphorus deficiency
in the material. The surface probably needs to develop an active configuration and
composition as a result of interactions with the reactant and products.
In terms of selectivity, both the TPR and LP catalysts form the same products, the
main ones being 2-pentanone and n-pentane. The different temperature dependencies
in the formation of the products indicate that the activation energies of the various
reactions are different. The reaction network will be discussed in a later section.
Figure 8 shows the XRD patterns before and after reaction for all three catalysts. There
are no significant changes in the line positions and no new peaks. The results suggest
that the lower activity of Ni2P/SiO2-LP than Ni2P/SiO2-TPR is not due to some sintering
or phase change process. They also indicate that the very low activity of Cu3P/SiO2 is an
innate property of the material. The latter is deduced from the fact that the Cu3P is
phase pure and has comparable chemisorption capacity as the Ni2P samples (Table 1).
In order to obtain further insight on the reason for the reactivity difference between
14

the Ni2P prepared by the TPR method and the LPmethod, examination of the near-surface
of the samples by x-ray photoelectron spectroscopy (XPS) was undertaken. Since the
hypothesis was that it was the phosphorus content that was responsible for the activity
differences, measurements of the P/Ni ratio before and after reaction were made. This
ratio would not be affected very much by carbon contamination on the outer layer of the
samples because the photoelectron mean free path is 0.6 nm at the binding energy of
phosphorus of 133 eV, while the mean free path is 1.5 nm at the binding energy of Ni of
853 eV.
The actual XPS spectra can be seen in the Supplementary Information section. A
summary of the calculated elemental composition is provided in Table 2.
Examining the individual entries, it can be seen that the O signal is the highest followed
by Si, C, Ni, and P. This is reasonable as all the elements can be oxidized and the samples
were exposed to the atmosphere. Comparing the fresh and spent samples it can be seen
that for both the TPR and LP samples the O signal decreases slightly after reaction. On
the other hand, the C signal increases. This is also reasonable, as adsorbed species from
reaction are expected to be present on the spent samples to block oxygen and enhance
carbon. The Si signal does not change appreciably so it’s likely that most adsorbates are
on the Ni.
Indeed, the Ni signal goes down after reaction. The P levels are higher
In
in the TPR samples because the TPR method of preparation uses extra phosphorus.
the last column the P/Ni ratio indicates that after reaction the TPR samples also have an
enhanced phosphorus level compared to the LP samples.
The higher absolute and proportional P levels in the TPR samples are conjectured to
be responsible for their higher activity. Such enhancements have been reported in earlier
15

studies of phosphides for hydroprocessing reactions [56]. In addition, for the related
HDO of 2-methyltetrahydrofuran, studies of NiFeP alloys indicated that the reaction was
structure-insensitive, with only one surface Ni atom involved in the rate-determining step
[49], so effects of particle size are not likely to be responsible for the differences in activity.
These results suggest that the method of preparation is important, and procedures that
allow extra phosphorus to be present on the support to replenish the phosphide
nanoparticles, such as the TPR method, are superior.
The kinetics of the reaction were studied on the Ni2P/SiO2-LP sample by varying the
H2 and the 2-MF partial pressures at 350 ˚C and a total pressure of 0.5 MPa (Figs. 9, 10).
For the H2 partial pressure there was an S-shaped curve in the conversion (Fig. 9a), and
weak dependencies of selectivity on the H2 levels (Fig. 9b). The unusual shape of the
conversion curve may have been due to competition for sites by adsorbed 2-MF and
hydrogen, with the hydrogen becoming favored at high partial pressure of H2, and
resulting in increased reaction rate.
For the 2-MF partial pressure there was a simple monotonous decrease in conversion
with increasing 2-MF partial pressure (Fig 10a). This indicates that high levels of
adsorbed 2-MF might have inhibited the reaction. The selectivities again showed weak
dependencies on reactant partial pressure (Fig. 10b).
The partial pressure experimental data (Figs. 9 and 10) were fitted to Langmuir-
Hinshelwood and power-rate law expressions, and the rate laws and fitted constants are
displayed below. Plots of the data and calculated curves are given in the Supplementary
Information file. The rates are turnover frequencies, so in contrast to the results of
16

conversion, TOF goes up with increasing 2-MF partial pressure as shown in Fig. S3 and
S4. The TOF rises because its definition has the product of inlet 2-MF molar flow rate
and conversion in the numerator. To attain higher 2-MF molar flow rate the 2-MF partial
pressure and feed flow rate were increased, and the resulting increase in flow rate more
than compensated for the decrease in conversion.
푘퐾2푀퐹푃2푀퐹 퐾 푃
√
퐻
2
퐻
2
푟푒푎푐푡푖표푛 푟푎푡푒 =
(1 + 퐾2푀퐹푃2푀퐹 + 퐾 푃 )푛
√
퐻
2
퐻
2
In the Langmuir-Hinshelwood expression the large equilibrium constant for
adsorption of 2-MF in the denominator accounts for the negative dependence of the
conversion on the 2-MF partial pressure. The data were obtained at isothermal
conditions (350˚C) so the rate parameters cannot be separated into energy and entropy
terms. The data were also limited giving rise to a limited value of regression coefficient
(perfect fit: R2 = 1). The exponent of 0.4 in the denominator of the expression results in
partial cancellation of the partial pressure terms in the numerator, and indicate that the
orders in 2-MF and H2 will be fractional positive numbers.
푏
(
)푎(
)
푟푒푎푐푡푖표푛 푟푎푡푒 = 푘 × 푃2푀퐹 푃퐻2
As expected from the discussion above, the rate parameters here show moderately
positive exponents for both 2-MF and H2. The regression coefficient is also moderate,
and of the same value as the Langmuir-Hinshelwood analysis, indicating that both models
are adequate for describing the data.
17

Figure 11 shows the conversion and selectivity as a function of contact time in the
HDO of 2-MF at 300 ˚C and 0.5 MPa. The conversion increases with contact time as
expected. Fig. 11a) shows that the main products are 2-pentanone and n-pentane, with
small amounts of 1-pentanol and 2-pentanol also produced. The selectivity to 2-
pentanone starts high and decreases at long contact time identifying it as a primary
product. The selectivity to n-pentane increases with contact time suggesting that it is a
final product. Fig. 11b) displays the selectivity to the pentanols in an expanded scale.
The selectivity to 1-pentanol and 2-pentanol go through a maximum, indicating that they
are intermediary compounds.
Based on the above contact time results, a reaction network can be proposed. As
shown in Fig. 12 it is thought that a first reaction is the hydrogenation of 2-methylfuran
(2-MF) to 2-methyltetrahydrofuran (2-MTHF), which is slow. As discussed earlier, this
is deduced from the high temperatures needed for the conversion of 2-MF as compared
to the conversion of 2-MTHF. Subsequently, the reaction sequence follows the network
established previously for 2-MTHF [33]. A ring-opening step occurs where ring-
opening can occur on either side of the oxygen to produce 2-pentanone or pentanal. This
gives rise to two branches in the reaction. In the top pathway 2-pentanone is formed,
which is hydrogenated to 2-pentanol, which undergoes dehydration to 2-pentene, and
which is hydrogenated to n-pentane. In the bottom pathway pentanal is formed, which
is hydrogenated to 1-pentanol, which undergoes dehydration to 1-pentene, and which is
also hydrogenated to n-pentane. Both 2-MTHF and pentanal as well as the olefins are
very reactive and cannot be observed at the high temperatures of this study. The
presence of the two branches are deduced from the observation of 1-pentanol and 2-
pentanol (Fig. 11b).
18

19

Conclusions
Samples of Ni2P/SiO2-TPR and Cu3P/SiO2 catalysts were successfully synthesized
by a temperature-programed reduction (TPR) method from phosphate precursors.
Cu3P/SiO2 had much lower activity in the HDO of 2-methylfuran than Ni2P/SiO2, and this
appears to be an innate property of Cu3P/SiO2. A sample of Ni2P/SiO2-LP was also
prepared by a liquid-phase method using trioctylphosphine as phosphiding agent. The
Ni2P/SiO2-TPR was formed in smaller phosphide particle size and had higher activity
than the Ni2P/SiO2-LP. The higher activity was probably due to a higher phosphorus
content in the phosphide prepared by TPR as indicated by X-ray photoelectron
spectroscopy.
The hydrodeoxygenation (HDO) of 2-methylfuran (2-MF) required higher
temperatures than the HDO of 2-methyltetrahydrofuran. Contact time measurements on
the Ni2P/SiO2-TPR showed that in the 2-MF reaction network 2-pentanone was an initial
product, 1-pentanol and 2-pentanol were intermediate compounds, and n-pentane was a
final product. The n-pentane could be produced at 69% selectivity at 450˚C. Kinetic
measurements on the Ni2P/SiO2-LP indicated that the turnover frequency depended on
fractional orders of 0.6 in 2-MF and 0.5 in H2.
Acknowledgments
This work was supported by a Grant-in-Aid for Scientific Research (B) (no. 26289301) of JSPS,
Japan, and Fuzhou Univ., China.. STO received support from the National Science Foundation under
grant CHE-1361842 and the US Department of Energy under grant DE-FG02-96ER14669
20

References
1) Ota, E. L. Kunkes, J. Kröhnert, M. Schmal, M. Behrens, Appl. Catal. A: Gen
452 (2013) 203-213.
2) F. Souza Toniolo, R. Newton S.H. Magalhães, C. A. C. Perez, M. Schmal,
Appl. Catal. B: Env. 117-118 (2012) 156-166.
3) N. Mosier, C. Wyman, B. Dale, R. Elander, Y.Lee, M. Holtzapple, M. Ladisch,
Bioresour. Technol. 96 (2005) 673-686.
4) A.Demirbas, Appl. Energy 88 (2011) 17-28.
5) P. M. Mortensen, J. D. Grunwaldt, P. A. Jensen, K. G. Knudsen, A. D. Jensen,
Appl. Catal. A: Gen. 407 (2011) 1-19.
6) M. F. Demirbas, Appl. Energy 86 (2009) S151-S161.
7) Y. Wang, T. He, K. Liu, J. Wu, Y. Fang, Bioresour. Technol. 108 (2012) 280-
284.
8) Y. Lin, C. Zhang, M. Zhang, J. Zhang, Energy Fuels 24 (2010) 5686-5695
9) Z. Su-Ping, Energy Sources 25 (2003) 57-65.
10) J. S. Moon, E. G. Kim, Y. K. Lee, J. Catal. 311 (2014) 144-152.
11) V.O.O. Gonçalves, P. M. D. Souza, V. T. D. Silva, F. B. Noronha, F. Richard,
Appl. Catal. B: Environ. 205 (2017) 357-367.
12) J. R. Hayes, R. H. Bowker, A. F. Gaudette, M. C. Smith, C. E. Moak, C. Y.
Nam, T.K. Pratum, M.E. Bussell, J. Catal. 276 (2010) 249-258.
13) X. Li, Z. Sun, A. Wang, X. Yang, Y. Wang, Appl. Catal. A: Gen. 417–418
(2012) 19–25.
14) R. H. Bowker, B. Ilic, A. C. Bo, M. A. Reynolds, B. D. Murray, M. E, Appl.
Catal. A: Gen. 482 (2014) 221-230.
21

15) M. Toba, Y. Abe, H. Kuramochi, M. Osako, T. Mochizuki, Y. Yoshimura,
Catal. Today 164 (2011) 533-537.
16) O. I. Şenol, E. M. Ryymin, T. R. Viljava, A. O. Krause, J. Mol. Catal. A 277
(2007) 107-112
17) S. F. Moya, R. L. Martins, A. Ota, E. L. Kunkes, M. Behrens, M. Schmal,
Appl. Catal. A: Gen. 411–412 (2012) 105-113.
18) A.Ota, E. L. Kunkes, J. Kröhnert, M. Schmal, M. Behrens, Appl. Catal. A:
Gen. 452 (2013) 203-213.
19) W. Xu, S. D. Senanayake, J. Llorca, J. C. Hanson, R. Si, H. Idriss, D.
Stacchiola, J. A. Rodriguez, J. Catal. 291 (2012) 117-126.
20) J. Wildschut, F. H Mahfud, R. H. Venderbosch, H. J. Heeres, Ind. Eng. Chem.
Res. 48 (2009) 10324-10334.
21) A.Gutierrez, R. K. Kaila, M. L. Honkela, R. Slioor, A. O. Krause, Catal. Today
147 (2009) 239-246.
22) S. F. Zaman, K.J. Smith, Appl. Catal. A: Gen. 378 (2010) 59-68.
23) R. Prins, M.E. Bussell, Catal. Lett. 142 (2013) 1413-1436.
24) S. T. Oyama, J. Catal. 216 (2003) 343-352.
25) S. T. Oyama, X. Wang, Y. K. Lee, W. J. Chun, J. Catal. 221 (2004) 263-273.
26) W. Li, B. Dhandapani S. T. Oyama, Chem. Lett. 3 (1998) 207-208.
27) S. T. Oyama, T. Gott, H. Zhao, Y. K. Lee, Catal. Today 143 (2009) 94-107.
28) E. Furimsky, Appl. Catal. A: Gen. 199 (2000) 147-190.
29) V. N. Bui, D. Laurenti, P. Delichère, C. Geantet, Appl. Catal. B: Environ. 101
(2011) 246-255.
22

30) V. N. Bui, D. Laurenti, P. Delichère, C. Geantet, Appl. Catal. B: Environ. 101
(2011) 246-255.
31) K. Li, R. Wang, J. Chen, Energy Fuels 25 (2011) 854-863.
32) J. S, Moon, E. G. Kim, Y. K. Lee, J. Catal. 311 (2014) 144-152.
33) H. Y. Zhao, D. Li, P. Bui, S. T. Oyama, Appl. Catal. A: Gen. 391 (2011) 305-
310.
34) P. Bui, J. A. Cecilia, S. T. Oyama, A. Takagaki, A. Infantes-Molina, H. Zhao,
D. Li, E. Rodríguez-Castellón, A. J. López, J. Catal. 294 (2012) 184-198.
35) A.Iino, A. Cho, A. Takagaki, R. Kikuchi, S. T. Oyama, J. Catal. 311 (2014)
17-27.
36) P. Bui, B. Callow, K. Nozaki, A. Takagaki, S. T. Oyama, ACS Catal. 6 (2016)
4549-4558.
37) N. Koike, S. Hosokai, A. Takagaki, S. Nishimura, R. Kikuchi, K. Ebitani, S.
T. Oyama, J. Catal. 333 (2016) 115-126.
38) S. T. Oyama, X. Wang, Y. K. Lee, K. Bando, F. G. Requejo, J. Catal. 210
(2002) 207-217.
39) M. S. A. Bakar, J. O. Titiloye, J. Anal. Appl. Pyrolysis 103 (2013) 362-368.
40) W. Wang, S. Wang, X. Ma, J. Gong, Chem. Soc. Rev. 40 (2011) 3703-3727.
41) Z. Ma, F. Zhang, H. Meng, M. Wang, X.-Q. Bao, B. Tang, X. Wang,
Electrochim. Acta 220 (2016) 193-204.
42) K. Zhang, Z. Xiong, S. Li, B. Yan, J. Wang, Y. Du, J. Alloys Comp. 706
(2017) 89–96.
43) S.J. Sawhill, K.A. Layman, D.R. Van Wyk, M.H. Engelhard, C. Wang, M.E.
Bussell, J. Catal. 231 (2005) 300-313.
23

44) Y. Xie, H. L. Su, X. F. Qian, X. M. Liu, Y. T. Qian, J. Solid State Chem. 149
(2000) 88-91.
45) Y. Gu, G. Fan, Y. Qian, H. Zheng, Z. Yang, J. Am. Chem. Soc. 37 (2002)
1101-1105.
46) Z. Yao, M. Li, X. Wang, X. Qiao, J. Zhu, Y. Zhao, G. Wang, J. Yin, H. Wang,
Dalton Trans. 44 (2015) 5503-5509.
47) S. L. Brock, S. C. Perera, K. L. Stamm, Chem. Eur. J. 10 (2004) 3364-3371.
48) K. Y. Yoon, Y. Jang, J. Park, Y. Hwang, B. Koo, J. G. Park, T. Hyeon, J. Solid
State Chem. 181 (2008) 1609-1613.
49) D. Li, K. Senevirathne, L. Aquilina, S.L. Brock, Inorg. Chem. 54 (2015) 7968-
7975.
50) J. H. Chen, M. F. Tai, K. M. Chi, J. Mater. Chem. 14 (2004) 296-298.
51) Qian, F. Kim, L. Ma, F. Tsui, P. Yang, J. Liu, J. Am. Chem. Soc. 126 (2004)
1195-1198.
52) J. Park, B. Koo, Y. Hwang, C. Bae, K. An, J. G. Park, H. M. Park, T. Hyeon,
Angew. Chem. 43 (2004) 2282-2285.
53) Mauvernay, M. P. Bichat, F. Favier, L. Monconduit, M. Morcrette, M. L.
Doublet, Ionics 11 (2005) 36-45.
54) X. Wang, P. Clark, S. T. Oyama, J. Catal. 208 (2002) 321-331.
55) A.Cho, J. Shin, A. Takagaki, R. Kikuchi, S. T. Oyama, Top. Catal. 55 (2012)
969–980.
56) X. Wang, Y. Lee, S. T. Oyama, K. Bando, F. G. Requejo, J. Catal. 210 (2002)
207-217.
24

Figure 1 Structural formulas of 2-methylfuran and 2-methyltetrahydrofuran
25

Figure 2 Schematic of reactor for the HDO of 2-methylfuran
26

300
400
500
600
700
800
Temperature / oC
Figure 3.
Reduction profile of precursor for Cu3P/SiO2
27

Ni2P/SiO2-TPR
Cu3P/SiO2
SiO2
Ni2P
Cu3P
20
30
40
50
60
70
80
2 theta / Degrees
Figure 4. XRD patterns of Ni2P/SiO2-TPR and Cu3P/SiO2
28

340 C
330C
317 C
Ni2P-PDF 00-003-0953
Ni12P5-PDF 01-074-1381
20
30
40
50
60
70
80
2Degree
Figure 5. XRD patterns of Ni2P/SiO2-LP prepared at different temperatures
29