Applied Catalysis A: General 391 (2011) 305310

Contents lists available at ScienceDirect

Applied Catalysis A: General

journal homepage: www.elsevier.com/locate/apcata

Hydrodeoxygenation of guaiacol as model compound for pyrolysis oil on

transition metal phosphide hydroprocessing catalysts
H.Y. Zhao a , D. Li a,b , P. Bui a , S.T. Oyama a,c,
a
Department of Chemical Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, United States
b
State Key Lab of Heavy Oil Processing, China University of Petroleum, Beijing 102249, PR China
c
Department of Chemical Systems Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan

a r t i c l e i n f o a b s t r a c t

Article history: The gas phase hydrodeoxygenation (HDO) of guaiacol, as a model compound for pyrolysis oil, was tested
Received 26 March 2010 on a series of novel hydroprocessing catalysts transition metal phosphides which included Ni2 P/SiO2 ,
Received in revised form 14 July 2010 Fe2 P/SiO2 , MoP/SiO2 , Co2 P/SiO2 and WP/SiO2 . The turnover frequency based on active sites titrated by
Accepted 21 July 2010
the chemisorption of CO followed the order: Ni2 P > Co2 P > Fe2 P, WP, MoP. The major products from
Available online 30 July 2010
hydrodeoxygenation of guaiacol for the most active phosphides were benzene and phenol, with a small
amount of methoxybenzene formed. Kinetic studies revealed the formation of reaction intermediates
Keywords:
such as catechol and cresol at short contact times. A commercial catalyst 5% Pd/Al2 O3 was more active
Hydrodeoxygenation
Guaiacol
than the metal phosphides at lower contact time but produced only catechol. A commercial CoMoS/Al2 O3
Biooil deactivated quickly and showed little activity for the HDO of guaiacol at these conditions. Thus, transition
Transition metal phosphides metal phosphides are promising materials for catalytic HDO of biofuels.
Ni2 P 2010 Elsevier B.V. All rights reserved.
CoMoS

1. Introduction of a reducing gas such as hydrogen [10], and in the hydrogena-

Pyrolysis oil from thermal cracking of biomass is attracting
attention as an alternative liquid fuel because of the depletion
of petroleum deposits and the increasing environmental concern
with the burning of nonrenewable resources [1]. However, oxygen
removal is required to upgrade pyrolysis oil because its high oxygen
content (2040%) leads to undesirable properties of the oil such as
low energy density, and thermal and chemical instability [2]. The
subjects of hydrodeoxygenation [3] and pyrolysis oil treatment [4]
have been reviewed. The average composition of pyrolysis oils is
5065 wt% organic components, that include organic acids, alde-
hydes, ketones, furans, phenolic compounds, guaiacols, syringols
and sugar based compounds, 1530 wt% water and 20 wt% colloidal
lignin fraction [5,6]. The phenolics content, a major part of the lignin
fraction may reach 30% of the organic component [68]. Guaiacol is
one of the most abundant of the lignin-derived products in biooil,
present at levels of approximately 0.180.51 wt% in switchgrass
and alfalfa derived pyrolysis oils [9].
Generally, there are two methods for removal of oxygen. In
the direct deoxygenation method, which is generally conducted at
atmospheric pressure, CO bonds are broken without the assistance
Corresponding author.
E-mail address: oyama@vt.edu (S.T. Oyama).
tion route aromatic rings are hydrogenated before removal of
oxygen [11]. The former has been reported on tungsten (IV) com-
pounds [12] and acidic zeolites such as HZSM-5 [13]. The latter
process is carried out at high pressure and temperature, and is
related to hydrotreating of petroleum feedstocks for removal of
sulfur and nitrogen. Thus, this hydrodeoxygenation (HDO) pro-
cess can potentially allow use of the existing petroleum rening
infrastructure for processing and transportation [14]. Conventional
sulde catalysts for petroleum hydroprocessing [15] and precious
metal catalysts [16] have been studied for their reactivity in gua-
iacol and model oxygenate compound HDO. Oxygenated groups
in pyrolysis oil such as ketones, aldehydes, and organic acids
require lower temperatures for elimination of the reactive func-
tionalities but guaiacol type molecules and other phenolic species
require higher temperature [17]. Phenyloxygen bonds are cleaved
at 500650 K using hydropressing catalysts under hydrogen pres-
sure, in which the oxygen is ultimately removed as water. However,
typical hydrodesulfurization catalysts, such as NiMoS/Al2 O3 and
CoMoS/Al2 O3 were found to quickly deactivate by coke deposi-
tion in model HDO reactions because of the acidity of the reactant
[18]. Suldes on neutral supports including carbon, silica and
alumina modied by K for HDO reactions have been reported
[1720]. Yet the effect of modication for alumina supported cat-
alysts has not been optimized. It was found that guaiacol coking
reactions were negligible with molybdenum sulde supported on

0926-860X/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2010.07.039
306 H.Y. Zhao et al. / Applied Catalysis A: General 391 (2011) 305310
activated carbon, and the catalyst showed good stability and poten-
tial for catalytic hydrotreating system [21]. Noble metal Pt as an
active component was added to a conventional CoMoS catalyst and
showed no signicant improvement [17]. According to this study,
monometallic and bimetallic noble metal catalysts supported on
zirconia have lower coke formation than a CoMoS/Al2 O3 catalyst
and in particular Rh-containing catalysts demonstrate potential in
biofuel upgrading. Given the propensity of biofuels to thermally
decompose to form coke and coke-like matter, catalysts with sup-
ports that are less active for coke formation or more hydrogenating
catalysts permitting a rapid transformation of dioxygenated reac-
tants (guaiacol, catechol) into less coke-forming products (phenol)
would be highly desirable.
Hydrodeoxygenation of actual biooils has been studied with
two stages because of the thermal instability of the oil [22]. In
a rst stage, a stabilization process was carried out at low tem-
perature to eliminate reactive compositions like ketones. In a
second stage, deoxygenation of the phenolic-type molecules was
carried at higher temperatures. Using diluted model oxygenated
compound solutions for hydrodeoxygenation study will give more
precise chemical information and avoid thermal polymerization
reactions [23]. In the present study, guaiacol (methoxyphenol)
is chosen as model compound for hydrodeoxygenation because
guaiacol and substituted guaiacols constitute a relatively high con-
centration of the lignin-derived fraction (up to 0.5 wt%) and these
have a high tendency to coke. Guaiacols contains two different
oxygenated functions (phenolic and methoxy groups), so are chal-
lenging molecules to completely deoxygenate. Transition metal
phosphides supported on neutral silica are a promising class of new
hydroprocessing catalysts [24,25], and it was of interest to investi-
gate them for guaiacol catalytic hydrodeoxygenation in comparison
to commercial catalysts such as CoMoS/Al2 O3 and 5% Pd/Al2 O3 .
2. Experimental
2.1. Materials
The 5% Pd/Al2 O3 commercial catalyst was provided by BASF
Catalysts, In. and the CoMo/Al2 O3 hydrotreating catalyst was pro-
vided by Haldor Topse. Transition metal phosphides Ni2 P/SiO2 ,
Fe2 P/SiO2 , MoP/SiO2 , Co2 P/SiO2 and WP/SiO2 were synthesized as
will be described below, using a fumed silica EH-5 support provided
by Cabot Corp. The chemicals used in the synthesis of the cata-
lysts were Ni(NO3 )2 6H2 O (Alfa Aesar, 99%), Fe(NO3 )3 9H2 O (Alfa
Aesar, 99%), (NH4 )6 Mo7 O24 4H2 O (Alfa Aesar, 99%), Co(NO3 )2 6H2 O
(Alfa Aesar, 99%), (NH4 )6 W12 O39 xH2 O (Aldrich, 99%), (NH4 )2 HPO4
(Aldrich, 99%). The chemicals used for the reactivity tests were
guaiacol (Alfa Aesar, 98%). The gases employed were H2 (Airco,
Grade 5, 99.99%), He (Airco, Grade 5, 99.99%), CO (Linde Research
Grade, 99.97%), 0.5% O2 /He (Airco, UHP Grade, 99.99%), O2 (Airco,
UHP Grade, 99.99%), N2 (Airco, Grade 5, 99.99%). Chemical stan-
dards for GC and mass spectrometry were benzene, phenol, toluene,
methoxybenzene (Alfa Aesar, 98%), cyclohexane (Alfa Aesar, 99%).
2.2. Metal phosphides synthesis
Ni2 P/SiO2 [2628], Fe2 P/SiO2 [27], MoP/SiO2 [29,30], Co2 P/SiO2
[27] and WP/SiO2 [31] were prepared by temperature-programmed
reduction (TPR), following procedures reported previously
[32,33]. Briey, the synthesis of the catalysts involved two
stages. First, solutions of the corresponding metal phos-
phate precursors were prepared by dissolving appropriate
amounts of Ni(NO3 )2 6H2 O, Fe(NO3 )3 9H2 O, (NH4 )6 Mo7 O24 4H2 O,
Co(NO3 )2 6H2 O, (NH4 )6 W12 O39 xH2 O, with ammonium phosphate
in distilled water, and these solutions were used to impregnate
silica by the incipient wetness method. The obtained samples were
dried and calcined at 500 C for 6 h, then ground with a mortar
and pestle, pelletized with a press (Carver, Model C), and sieved to
particles of 6501180 m diameter (16/20 mesh). Second, the solid
phosphates were reduced to phosphides at 2 C min1 in owing
H2 [1000 cm3 (NTP) min1 g1 ]. Reduction temperatures were
568 C for Ni2 P/SiO2 , 680 C for Fe2 P/SiO2 , 680 C for Co2 P/SiO2 ,
494 C for MoP/SiO2, and 527 C for WP/SiO2 . The samples were
kept at the reduction temperature for 2 h, followed by cooling to RT
under He ow [100 cm3 (NTP) min1 ], and then were passivated
at RT in a 0.5% O2 /He for 4 h. The Ni, Fe, Mo, Co, W molar loading
were all 1.6 mmol g1 (mmol per g of support), corresponding to
a weight loading of Ni2 P of 8.6% with an initial Ni/P ratio of 1/2,
Fe2 P of 8.3% with an initial Fe/P ratio of 1/2, MoP 12.8% with initial
Mo/P ratio of 1, Co2 P of 8.6% with initial Co/P ratio 1, WP 19.9%
with initial W/P ratio 1.
2.3. Characterization
Irreversible CO uptake measurements were used to titrate the
surface metal atoms and to provide an estimate of the active sites
on the catalysts for the noble metals and the transition metal phos-
phides. Usually, 0.3 g of a passivated sample was loaded into a
quartz reactor. Noble metal catalysts were reduced in H2 at 325 C
for 2 h while passivated transition metal phosphides were reduced
at 450 C for 2 h with H2 at 300 ml (NTP) min1 . After cooling in He,
pulses of CO in a He carrier at 43 mol s1 [65 cm3 (NTP) min1 ]
were injected at RT through a sampling valve, and the mass 28
(CO) signal was monitored with a mass spectrometer. CO uptake
was calculated by measuring the decrease in the peak areas caused
by adsorption in comparison with the area of a calibrated vol-
ume (19.5 mol). Low temperature O2 chemisorption was used for
CoMo/Al2 O3 applying the same technique. Prior to the measure-
ment the sample was sulded in a ow of 10% H2 S/H2 at 400 C.
Surface areas of the samples were obtained using the BET
method based on adsorption isotherms at liquid nitrogen temper-
ature, and using a value of 0.162 nm2 for the cross-sectional area of
a N2 molecule. The measurements were performed in a volumetric
adsorption unit (Micromeritics ASAP 2000). X-ray diffraction (XRD)
patterns of the samples were obtained with PANalytical Xpert Pro
powder diffractometer operated at 45 kV and 40 mA, using Cu K
monochromatized radiation ( = 0.154178 nm).
2.4. Reactivity studies
Hydrodeoxygenation activity was measured in a packed bed
reactor at atmospheric pressure. Guaiacol was introduced by means
of a saturator at 25 C with a concentration of 0.024 mol%. The
dimensions of the reactors were 1.5 cm i.d. 25.5 cm long, and were
loaded with 30 mol active sites. To start a reaction, catalysts were
placed in the catalytic reactor and pretreated at the same condi-
tions as used for chemisorption. After pretreatment, a ow mixture
of hydrogen and nitrogen at a ratio of 4 to 1 saturated with guaiacol
was introduced at 150 cm3 (NTP) min1 . The hydrogen to guaiacol
molar ratio was 33. The catalysts were stabilized for 6 h after the
feed was introduced. Then samples were taken every 23 h until the
conversion of guaiacol reached steady-state. Reactivity testing was
performed as a function of temperature, starting at the highest tem-
perature of 300 C and was varied downwards and upwards with
the initial temperature repeated at the end. Generally it took about
100 h of on-stream time to collect the rate data at several different
temperatures. The reaction products were analyzed using an online
gas chromatograph (HewlettPackard, 5890A) equipped with a
0.32 mm i.d. 50 m fused silica capillary column and a ame ioniza-
tion detector. The reactants and products were identied by their
retention time in comparison with commercially available stan-
 H.Y. Zhao et al. / Applied Catalysis A: General 391 (2011) 305310 307

dards and conrmed by gas chromatographymass spectrometry Table 1

Characterization results for catalysts including CO chemisorption, dispersion, par-
(GCMS) (HewlettPackard, 58905972A). Response factors were
ticle size, BET surface area.
determined experimentally using pure compounds. Low molecular
weight products (methane and methanol) could not be separated CO uptake Dispersion (%) Partical size BET surface
(mol/g) (nm) area (m2 /g)
with the present column, and were not analyzed. However, since
they arise as by products of guaiacol reaction to phenol, their omis- Ni2 P/SiO2 134 12 8 309
sion does not affect the calculation of guaiacol conversion. Fe2 P/SiO2 52 4 20 233
MoP/SiO2 214 19 5 207
N (guaiacol)in N (guaiacol)out Co2 P/SiO2 42 4 25 307
Conversion [%] = 100 WP/SiO2 70 6 15 147
N (guaiacol)in
Pd/Al2 O3 120 25 4 82
CoMo/Al2 O3 100a 12 7 224
The conversions of guaiacol and product distributions were cal-
a
culated based on the analyzed gas phase. The conversion of guaiacol O2 uptake for CoMo/Al2 O3 .
was calculated from the initial and nal amounts of guaiacol.

Turnover frequency [s1 ] Co2 P/SiO2 < Fe2 P/SiO2 < WP/SiO2 < Ni2 P/SiO2 < MoP/SiO2 assuming
that each active site adsorbs one CO molecule. The particle size
Reactant owrate [mol/s] conversion (d) was then calculated using equation d 0.9/D. It was found that
=
Quantity of sites [mol/g] Catalyst weight [g] nickel and molybdenum formed much smaller particles and were
better dispersed than the other transition metal phosphides. Except
The turnover frequency was calculated to compare the intrinsic for WP/SiO2 , all other transition metal phosphides show fairly high
activity of different catalysts. BET surface area.

3. Results and discussion 3.2. X-ray diffraction

3.1. CO chemisorption and BET areas
Table 1 reports uptakes of CO at room temperature for the metal
phosphide catalysts and the noble metal and uptakes of O2 at dry-
ice acetone temperature for CoMoS/Al2 O3 . Table 1 also provides BET
characterization results. Earlier studies have shown that uptakes
of the SiO2 and Al2 O3 were negligible [3436]. The CO chemisorp-
tion uptakes of the different samples varied in a wide range from
42 to 200 mol/g. The dispersion (D) of metal sites was estimated
from the CO uptakes and the known loading of the samples (in all
cases 1.16 mmol g1 of total metal). The order of dispersion was
X-ray diffraction (XRD) was used to ascertain the phase and
phase composition of the synthesized transition metal phos-
phides by comparing with the standard powder diffraction patterns
(Fig. 1). The diffraction pattern for silica supported nickel phos-
phide shows three major peaks at 40.5 , 44.8 and 47.5 , which
line up well with the standard pattern for Ni2 P. The peaks are
signicantly broadened indicating that small Ni2 P crystals were
formed. The three major peaks for cobalt phosphide line up well
with the standard Co2 P pattern suggesting that the major phase is
Co2 P in our synthesis, which is consistent with result reported by
Bussell that Co2 P is obtained by using a Co/P ratio of 1 [37]. The

Fig. 1. X-ray diffraction pattern for transition metal phosphides.

308 H.Y. Zhao et al. / Applied Catalysis A: General 391 (2011) 305310

Table 2
Activation energy for guaiacol reaction on transition metal phosphide catalysts.

Conversion (%) Temperature ( C)

200 225 250 275 300 Ea (kJ/mol)

Ni2 P/SiO2 31 52 65 85 93 40
Co2 P/SiO2 21 39 61 77 82 52
WP/SiO2 9 12 22 59 23
a
Fe2 P/SiO2 15 70
MoP/SiO2 10 14 19 50 63
a
Not calculated because of deactivation.

275 C the conversion decreased abruptly from 70% to 15%, and

Fig. 2. Turnover frequency of guaiacol on transition metal phosphides.
Fe2 P/SiO2 XRD pattern shows three strong peaks which are in line
with Fe2 P. Noticeably, the Fe2 P/SiO2 XRD pattern is similar to that
of Ni2 P/SiO2 but its peaks are much narrower than that of Ni2 P/SiO2 ,
indicating that the Fe2 P particle size is larger than that of the Ni2 P,
which is consistent with the CO chemisorption results (Table 1).
The WP/SiO2 XRD pattern shows strong narrow peaks lining up
well with the standard WP pattern. In the case of MoP/SiO2 , no
peaks were detected for its XRD pattern, which indicated that MoP
is well dispersed on the silica. Combined with the CO chemisorp-
tion results, it suggests that the MoP crystallites are smaller than
5 nm and are not detectable by XRD.
3.3. Reactivity
3.3.1. HDO activity of transition metal phosphides
The hydroprocessing of biooils required very long contact
times to remove the oxygen content [38]. The reactivity tests
here were carried out at high contact time 20.2 min. A gas phase
feed containing 0.024 mol% guaiacol carried out with a mixture of
hydrogen and nitrogen at a ratio of 4 to 1 was used to test the HDO
activity of the transition metal phosphides and commercial cata-
lysts. All the catalysts showed expected responses to temperature
with higher conversions at higher temperatures and reasonable
stability over the time course of the reactions. The catalytic activity
of transition metal phosphides are compared in Fig. 2 in terms of
turnover frequency (TOF) based on sites titrated by the adsorption
of CO. These measurements were carried out starting with the
highest temperature of 300 C and then lowering and raising the
temperature back to 300 C to determine whether catalyst deacti-
vation was occurring. As Fig. 2 shows, at the high temperature of
300 C, the TOF of the catalysts for the HDO of guaiacol follows the
order: Ni2 P/SiO2 > Co2 P/SiO2 > Fe2 P/SiO2 > WP/SiO2 > MoP/SiO2 .
For Ni2 P/SiO2 in the temperature range 200300 C, the con-
version of guaiacol varied from 31% to 93%, while for Co2 P/SiO2
the conversion ranged from 21% to 82%. The smooth variation
of the conversion for Ni2 P/SiO2 and Co2 P/SiO2 with temperature
indicates that these catalysts were not deactivating. However with
Fe2 P/SiO2 , as the reaction temperature was lowered from 300 to
the temperature was not lowered further. After 50 h at 300 C the
conversion of guaiacol decreased from 70% to 10%, so the Fe2 P/SiO2
catalyst obviously underwent deactivation, probably by coking,
which is commonly observed [39]. For WP/SiO2 in the temperature
range 200300 C, the conversion of guaiacol varied from 9% to
59%, while for MoP/SiO2 it varied from 10% to 50%. Here there was
some deactivation, especially for MoP/SiO2 . In summary, Ni2 P/SiO2
and Co2 P/SiO2 are much more active than Fe2 P/SiO2 , WP/SiO2 and
MoP/SiO2 , and are stable in the HDO of guaiacol.
The apparent activation energy for guaiacol on the different
catalysts were estimated based on the reactivity at different tem-
peratures and are reported in Table 2. To minimize the effect of
deactivation, the lower conversion data at lower temperature were
used to calculate the apparent activation energy (Table 2). The
activation energies obtained for the overall conversion of guaiacol
(2363 kJ/mol) are lower than expected for carbon-oxygen bond
rupture (>240 kJ/mol) [40]. These low activation energies may indi-
cate that the oxygen removal reactions occur by hydrogenation
of double bonds in the aromatic ring followed by elimination of
water [4144]. Although guaiacol has a high coking tendency, in our
continuous ow experiments, the activity of the catalysts was mea-
sured after a stabilization period. Thus in our case the low activation
energy reects the high activity of the transition metal phosphides.
Products distribution and total conversion for HDO of guaiacol
are reported in Table 3. With the catalyst Ni2 P/SiO2 , products were
composed of 30% phenol, 60% benzene and 10% methoxybenzene.
The major product benzene resulting from the complete HDO of
guaiacol is the desirable product. For Co2 P/SiO2 , products consisted
of 32% phenol, 52% benzene, 15% C3C5 and 1% methoxybenzene.
For Fe2 P/SiO2 , the products were mainly phenol up to 94% and 6%
methoxybenzene. With WP/SiO2 , the product for HDO of guaiacol
was essentially 100% phenol. Similarly to Ni2 P/SiO2 and Co2 P/SiO2 ,
the major products with MoP/SiO2 were 28% phenol, 53% benzene,
15% C3C5 and 4% toluene. Thus, the major products for HDO of
guaiacol on transition metal phosphides were benzene, and very
small amounts of toluene. Partial HDO products were phenol and
small amounts of methoxybenzene.
3.3.2. Effect of contact time
The reaction of guaiacol HDO has been suggested to proceed
by rst the hydrogenolysis of the methyloxygen bond of the
methoxy group to form catechol, followed by the elimination of
the hydroxyl groups to produce phenol and water and then hydro-

Table 3
Products distribution for HDO of guaiacol with a concentration of 0.024 mol% at 300 C, atmosphere pressure, at contact time 20.2 min and space velocity 1.4 h1 .

Total conversion (%) Product distribution (%)

Phenol Benzene Methoxybenzene Toluene C3C5

Ni2 P/SiO2 80 30 60 10 0 0
Co2 P/SiO2 70 32 52 1 0 15
Fe2 P/SiO2 64 94 0 6 0 0
WP/SiO2 60 100 0 0 0 0
MoP/SiO2 54 28 53 0 4 15
 H.Y. Zhao et al. / Applied Catalysis A: General 391 (2011) 305310 309

Table 4
Effect of contact time on conversion of guaiacol and products distribution at 300 C.

Catalyst Contact time (min) Space velocity (h1 ) Conversion (%) Product distribution (%)

Phenol Benzene Methoxybenzene Catechol Cresol C3C5

Ni2 P/SiO2 20.2 1.4 80 30 60 10 0 0 0

0.339 59 19 28 4 38 0 30 0
Co2 P/SiO2 20.2 1.4 70 32 52 1 0 0 15
0.339 59 35 1 0 0 99 0 0
WP/SiO2 20.2 1.4 60 100 0 0 0 0 0
0.339 59 12 12 0 0 88 0 0

carbons [4144]. This reaction sequence suggests that catechol is
the primary reaction product, which is transformed to phenol. Our
results did not show catechol as a product for HDO of guaiacol in
all cases. Thus reactivity tests with lower contact time were carried
out to identify the reaction intermediates and the possible reac-
tion pathway (Table 4). The contact time dened in the following
equation, was decreased from 20.2 to 0.339 min
Contact time [min]
Quantity of sites [mol/g] Catalyst weight [g]
= which shows the possible reaction pathway on transition metal
Reactant owrate [mol/min]
Three catalysts Ni2 P/SiO2 , Co2 P/SiO2 , WP/SiO2 were chosen for
the lower contact time reactivity tests because they had the high-
est activity and showed minimal deactivation. As the contact time
decreased from 20.2 to 0.339 min, the conversion of HDO guaiacol
decreased signicantly for all the catalysts, as expected. However,
the decrease of the contact time from 20.2 to 0.339 min was not
accompanied by a linear decrease of the conversion. Instead, the
conversion of guaiacol at the contact time of 0.339 min was higher
than expected, and this could be because the reactor operated in
integral manner (at high conversion) and there may have been
product inhibition by water.
In the case of Ni2 P/SiO2 a new product, cresol, was observed
and more methoxybenzene was formed while less benzene
was produced. Selectivity towards phenol was 28%, which was
almost the same as at higher contact time, but no catechol was
observed. This result might be due to a direct elimination of the
methoxy group by hydrogenolysis of the aromatic carbonoxygen
bond [17].
In the case of Co2 P/SiO2 , the products were 99% catechol and
only 1% phenol at lower contact time compared with 94% phe-
nol and 6% methoxybenzene at higher contact time. With WP/SiO2 ,
at lower contact time 88% catechol was observed instead of 100%
phenol. Thus, the intermediate catechol was observed for HDO of
guaiacol at lower contact time. The results indicate that the rst
bond to be broken is the bond between the oxygen and the methyl
carbon, which is depicted in the reaction scheme illustrated in Fig. 3
phosphides. The scheme was adapted from results of this study as
well as work from others. Cyclohexanol was proposed as an inter-
mediate in the HDO of phenol [45], while cresol was reported in
several studies [11,40].
3.3.3. HDO activity comparison
The commercial catalysts Pd/Al2 O3 and CoMo/Al2 O3 were also
tested for the HDO of guaiacol at the lower contact time 0.339 min
at 300 C to compare with the transition metal phosphides. The
results are summarized in Table 5. The catalyst Pd/Al2 O3 is more
active in terms of guaiacol conversion at 70% than the most active
transition metal phosphide Co2 P/SiO2 with 35% conversion. How-
ever, catechol was the only product for conversion of guaiacol with
Pd/Al2 O3 . Catechol easily leads to coking, and this may explain the
lower performance of the Pd catalyst. With CoMoS/Al2 O3 the con-
version rate of guaiacol decreased quickly and almost no conversion
was observed for guaiacol. The deactivation may have two origins

Fig. 3. Reaction network for HDO of guaiacol.

310 H.Y. Zhao et al. / Applied Catalysis A: General 391 (2011) 305310
Table 5
Activity comparison between transition metal phosphides and commercial catalysts at contact time 0.339 min and space velocity 59 h1 at 300 C.
Conversion (%) Product distribution (%)
Phenol
Ni2 P/SiO2 19 28
Co2 P/SiO2 35 1 0
WP/SiO2 12 12
5% Pd/Al2 O3 70 0 0
CoMo/Al2 O3 1 0
[23], coking or poisoning. Thus, either coke or other heavy prod-
ucts block the active sites, or the primary product catechol adsorbs
strongly on them [46]. The main superiority of Ni2 P/SiO2 at low
contact time over Pd/Al2 O3 comes from the phosphides ability to
form phenol in a higher proportion. Then, this catalyst is able to
form products less susceptible to coke formation. Thus, Ni2 P/SiO2
is better than Pd/Al2 O3 and CoMoS/Al2 O3 under the conditions used
in this study.
4. Conclusions
A group of transition metal phosphides were evaluated for the
hydrodeoxygenation of guaiacol. The activity for HDO of guaiacol
follows the order: Ni2 P > Co2 P > Fe2 P, WP, MoP. The major prod-
ucts for HDO of guaiacol are phenol, benzene, methoxybenzene,
with no catechol formed at higher contact time. At lower contact
time catechol is the major products for Co2 P and WP. No catechol
was observed for HDO of guaiacol with Ni2 P even at low contact
time. The commercial 5% Pd/Al2 O3 catalyst is more active than the
metal phosphides at lower contact time, but the major product is
catechol which is undesired. The commercial hydrotreating cat-
alyst CoMoS/Al2 O3 deactivated quickly and showed little activity
for the HDO of guaiacol at these conditions. These results indicate
that transition metal phosphides are promising catalysts for the
treatment of bio-derived feedstocks.
Acknowledgments
This work was supported by the US Department of Energy, Ofce
of Basic Energy Sciences, through Grant DE-FG02-963414669,
the National Renewable Energy Laboratory through Grant DE-
FG3608GO18214, and the Japan Ministry of Agriculture, Forestry,
and Fisheries (Norinsuisansho).
References
[1] E.G.W. Huber, NSF, Breaking the Chemical and Engineering Barriers to Ligno-
cellulosic Biofuels: Next Generation Hydrocarbon Bioreneries, 2008.
[2] A.V. Bridgewater, G. Grassi, Biomass Pyrolysis Liquids: Upgrading and Utilisa-
tion, Elsevier Applied Science, London, New York, 1991.
[3] E. Furimsky, Catal. Rev. Sci. Eng. 25 (1983) 421458.
[4] D.C. Elliott, Energy Fuels 21 (2007) 17921815.
[5] S. Fernandoet, Energy Fuels 20 (2006) 17271737.
[6] D. Mohan, C.U. Pittman, P. Steele, Energy Fuels 20 (2006) 848889.
Benzene Methoxybenzene Catechol Cresol
4 38 0 30
0 99 0
0 0 88 0
0 100 0
0 100 0 0
[7] L. Ingram, D. Mohan, M. Bricka, P. Steele, D. Strobel, D. Crocker, B. Mitchell, J.
Mohammad, K. Cantrell, C.U. Pittman, Energy Fuels 22 (2008) 614625.
[8] J. Zhang, H. Toghiani, D. Mohan, C.U. Pittman, R. Toghiani, Energy Fuels 21
(2007) 23732385.
[9] C.A. Mullen, A.A. Boateng, Energy Fuels 22 (2008) 21042109.
[10] J. Filley, C. Roth, J. Mol. Catal. A: Chem. 139 (1999) 245252.
[11] P. Grange, E. Laurent, R. Maggi, A. Centeno, B. Delmon, Catal. Today 29 (1996)
297301.
[12] B.K. Sharpless, T.B. Flood, J. Chem. Soc. Chem. Commun. (1972) 370371.
[13] Y.S. Prasad, N.N. Bakhshi, J.F. Mathews, R.L. Eager, Can. J. Chem. Eng. 64 (1986)
278284.
[14] G.W. Huber, S. Iborra, A. Corma, Chem. Rev. 106 (2006) 40444098.
[15] E. Laurent, B. Delmon, Appl. Catal. A: Gen. 109 (1994) 7796.
[16] A. Gutierrez, R.K. Kaila, M.L. Honkela, R. Sliorr, A.O.I. Krauser, Catal. Today 147
(2009) 239246.
[17] A. Centeno, E. Laurent, B. Delmon, J. Catal. 154 (1995) 288298.
[18] E. Laurent, A. Centeno, B. Delmon, in: Proc. 6th Inter. Symp. Catalyst Deactiva-
tion, Stud. Surf. Sci. Catal. 88 (1994) 573578.
[19] M. Ferrari, S. Bosmans, R. Maggi, B. Delomon, P. Grange, Catal. Today 65 (2001)
257264.
[20] M. Ferrari, B. Delmon, P. Grange, Carbon 40 (2002) 497511.
[21] G.D. Puente, A. Gil, J.J. Pis, P. Grange, Langmuir 15 (1999) 58005806.
[22] E.G. Baker, D.C. Elliot, Research in Thermochemical Biomass Conversion, Else-
vier, London, 1988, p. 883.
[23] E. Laurent, B. Delmon, Ind. Eng. Chem. Res. 32 (1993) 25162524.
[24] S.T. Oyama, J. Catal. 216 (2003) 343.
[25] S.T. Oyama, T. Gott, H. Zhao, Y.-K. Lee, Catal. Today 143 (2009) 94107.
[26] S.T. Oyama, X. Wang, F. Requejo, T. Sato, Y. Yoshimura, J. Catal. 209 (2002)
15.
[27] S.T. Oyama, X. Wang, Y. Lee, K. Bando, F.G. Requejo, J. Catal. 210 (2002)
207217.
[28] S.T. Oyama, X. Wang, Y.-K. Lee, W.-J. Chun, J. Catal. 221 (2004) 263273.
[29] W. Li, B. Dhandapani, S.T. Oyama, Chem. Lett. 3 (1998) 207208.
[30] S.T. Oyama, P. Clark, V.L.S. Teixeira da Silva, E.J. Lede, F.G. Requejo, J. Phys. Chem.
B 105 (2001) 49614966.
[31] P. Clark, W. Li, S.T. Oyama, J. Catal. 200 (2001) 140147.
[32] P.A. Clark, X. Wang, S.T. Oyama, J. Catal. 207 (2002) 256265.
[33] X. Wang, P.A. Clark, S.T. Oyama, J. Catal. 208 (2002) 321331.
[34] P.A. Clark, X. Wang, P. Deck, J. Catal. 210 (2002) 116126.
[35] P.A. Clark, S.T. Oyama, J. Catal. 218 (2003) 7887.
[36] S.T. Oyama, Y.-K. Lee, J. Catal. 258 (2008) 393400.
[37] A.W. Burns, K.A. Layman, D.H. Bale, M.E. Bussell, Appl. Catal. A: Gen. 343 (2008)
6876.
[38] E. Churin, R. Maggi, P. Grange, B. Delmon, in: A.V. Bridgwater, J.L. Kuester (Eds.),
Research in Thermochemical Biomass Conversion, Elsevier, London and New
York, 1988, pp. 896909.
[39] E. Laurent, B. Delmon, Appl. Catal. A 109 (1994) 7796.
[40] A.I. A, E. Chornet, R.W. Thring, R.P. Overend, Fuel 75 (1996) 509516.
[41] S.J. Hurff, M.T. Klein, Ind. Eng. Chem. Fundam. 22 (1983) 426430.
[42] J.B. Bredenbrerg, M. Huuska, P. Toropainen, J. Catal. 120 (1989) 401408.
[43] M. Huuska, J. Rintala, J. Catal. 94 (1985) 230238.
[44] J.B. Bredenberg, M. Huuska, J. Raty, M. Korpio, J. Catal. 77 (1982) 242247.
[45] C. Zhao, Y. Kou, A.A. Lemonidou, X. Li, J.A. Lercher, Angew. Chem. Int. Ed. 49
(2009) 39873990.
[46] J.B. Bredenberg, Z. Sarbak, J. Chem. Technol. Biotechnol. 42 (1988) 221234.