Selective vapor-phase hydrodeoxygenation of anisole to benzene on molybdenum carbide catalysts

Article abstract of DOI:10.1016/j.jcat.2014.07.025

Vapor-phase hydrodeoxygenation (HDO) of anisole over Mo₂C catalysts at low temperatures (420-520 K) and ambient pressure showed (1) remarkable selectivity for C-O bond cleavage, giving benzene selectivity >90% among C₆⁺pro

Full text of DOI:10.1016/j.jcat.2014.07.025

Journal of Catalysis 319 (2014) 44–53
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Selective vapor-phase hydrodeoxygenation of anisole to benzene
on molybdenum carbide catalysts
⇑
Wen-Sheng Lee, Zhenshu Wang, Ryan J. Wu, Aditya Bhan
Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, MN 55455, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Vapor-phase hydrodeoxygenation (HDO) of anisole over Mo2C catalysts at low temperatures (420–520 K)
and ambient pressure showed (1) remarkable selectivity for C–O bond cleavage, giving benzene
selectivity >90% among C+6 products, (2) high hydrogen efficiency for the HDO reaction as indicated by
low cyclohexane selectivity (<9%), and (3) good stability over ꢀ50 h. Methane selectivity increased at
the expense of methanol selectivity as anisole conversion increased, suggesting that the phenolic C–O
bond was cleaved preferentially. The concurrent near half-/zero-order dependence of benzene synthesis
rates on H2/anisole pressure, and the preferential inhibition of benzene synthesis rates upon introduction
of CO relative to isotopic HD exchange suggest that catalytic sites for H2 activation are distinct from those
required for the activation of anisole. The involvement of metallic sites on Mo2C catalysts for this reaction
was demonstrated by the nearly invariant benzene synthesis rate per CO chemisorption site.
Ó 2014 Elsevier Inc. All rights reserved.
Received 17 April 2014
Revised 30 July 2014
Accepted 31 July 2014
Keywords:
Lignin upgrading
Anisole
Benzene
Molybdenum carbide
Hydrodeoxygenation
1. Introduction
Anisole has been widely chosen as a model compound for
lignin-derived phenolics because it contains a methoxyl group
Utilization of biogenic sources as an alternative carbon feed-
stock to produce fuels and chemicals has drawn a lot of attention
because of sustainability and environmental concerns [1–3]. In this
context, the lignin fraction of lignocellulosic biomass represents
the most viable source to produce aromatic compounds. Hydro-
deoxygenation (HDO) of phenolic ethers derived from lignin
results in the synthesis of aromatic compounds such as benzene,
toluene, and xylene (BTX) [4–7]. High selectivity to cleavage of
the phenolic C–O bond, which gives aromatics products, is desired,
but concurrently challenging because of its strong bond strength
(422–468 kJ moleꢁ1) [1]. Severe reaction conditions with both high
reaction temperature (500–700 K), which facilitates phenolic C–O
bond cleavage [1,8–10] and high hydrogen pressure (1–30 MPa),
which alleviates catalyst deactivation by reducing coke formation
[8,11–13] are typically used for HDO of phenolic compounds to
give aromatic products. This accompanies potential issues such
as (i) excess hydrogen consumption and/or loss of aromaticity via
unwanted hydrogenation and/or (ii) reduction of the carbon chain
length via hydrogenolysis reactions, and/or (iii) low selectivity to
desired deoxygenated aromatics due to other side reactions such
as alkylation/transalkylation promoted by high reaction tempera-
tures [14,15].
(–OCH3), one of the major oxygen-containing functional groups
found in lignin-derived compounds [12,15–21]. Low selectivity
(<25%) to the desired deoxygenated aromatics (i.e., benzene) but
high selectivity (>40%) to phenol was found for anisole HDO with
conventional hydrotreating Co–Mo sulfide catalysts (either in the
liquid or in the vapor phase) at different conditions with tempera-
tures ranging from 523 to 560 K and total pressures ranging
from 1.5 to 50 MPa [18,22,23], suggesting that Co–Mo-based cata-
lysts preferentially break the weaker aliphatic C–O bond
(ꢀ339 kJ moleꢁ1), not the phenolic C–O bond, in anisole. Transition
metal phosphides, a potential new class of hydroprocessing cata-
lysts [24–26], have also been investigated (Ni2P/SiO2, MoP/SiO2
and NiMoP/SiO2) for anisole HDO in a fixed bed reactor [17]. While
the selectivity to deoxygenated products (benzene and cyclohex-
ane) can reach almost ꢀ100% at 573 K and 1.5 MPa, the benzene
selectivity was found to be less than 20%, suggesting inefficient
usage of H2 for the HDO reaction. Similarly, benzene selectivity
was found to be only ꢀ30–46% at 498–548 K under 50 atm H2
when using copper chromite as a catalyst for liquid-phase anisole
HDO [16]. Vapor-phase anisole HDO over noble metal catalysts
such as Pt/Al2O3 at 573 K and near ambient pressure (ꢀ140 kPa)
resulted in benzene selectivity as low as ꢀ4.4% and high phenol
selectivity (ꢀ65%) [21]. Pt–Sn bimetallic catalysts showed higher
benzene yield than either Pt or Sn monometallic catalysts for
vapor-phase guaiacol HDO reactions; however, these bimetallic
⇑
Corresponding author. Fax: +1 612 626 7246.
E-mail address: abhan@umn.edu (A. Bhan).
http://dx.doi.org/10.1016/j.jcat.2014.07.025
0021-9517/Ó 2014 Elsevier Inc. All rights reserved.

W.-S. Lee et al. / Journal of Catalysis 319 (2014) 44–53
45
formulations still resulted in aromatic yields less than 50% for
vapor-phase anisole HDO reactions at 673 K and atmospheric pres-
sure [12].
existing pure H2 flow to give the desired composition) to the final
target temperature (ꢀ863–978 K, typically 863 K) within 1.5 h and
held at the final temperature for 2–6 h (typically 2 h) and then
cooled down to RT in the CH4/H2 (15/85 vol%) gas mixture flow.
The resulting material was then treated in a flowing 1% O2/He mix-
ture (Matheson, Certified Standard Purity) at ꢀ1.67–3.33 cm3 sꢁ1
for at least 2 h before being removed from the reactor to passivate
the carbidic surface [33]. Thus, a typical protocol for the Mo2C
catalyst preparation (sample#8 in Table 1), ‘‘CH4/H2-RT(1.5 h)-
623 K[12 h](1.5 h)-863 K[2 h]-cool’’, represents that the ammo-
nium molybdate tetrahydrate in the reactor was exposed to a gas
environment of CH4/H2 mixture and heated up to 623 K within
1.5 h and held at 623 K for 12 h. After that, the temperature was
increased to 863 K within 1.5 h. Finally, the reactor was held at
863 K for 2 h and then cooled down to RT. Even though the prepa-
ration conditions were carefully controlled in this work, different
batches of Mo2C catalysts may have small variations in their phys-
ical/chemical properties because the degree of carburization as
well as the amount of carbon deposition was found to be sensitive
to the carburization conditions [34,35]. Therefore, samples #1-8
listed in Table 1 consisted of multiple batches of Mo2C synthesized
with the corresponding preparation conditions to create a repre-
sentative Mo2C catalyst that shares a set of physical/chemical
properties. Sample #9 in Table 1, however, was prepared by re-car-
burizing a mixture of multiple batches of Mo2C with different
amounts of CO chemisorption sites at 323 K (ꢀ6 g) in a CH4/H2
mixture (14/86 vol%, ꢀ9.66 cm3 sꢁ1) using a temperature protocol
shown in Table 1. We note that the fresh, passivated sample was
stored in a vial (ambient conditions) and exposed to air before
the kinetic measurements and aged <1 month prior to the charac-
terization discussed below.
High selectivity to deoxygenation products with selectivity to
benzene ꢀ50%, toluene ꢀ25%, and xylene ꢀ10% at ꢀ673 K under
1 atm, however, was reported when an acidic function (H-BEA)
was introduced together with a metal function (Pt) [15]. Hicks
and coworkers recently reported that high selectivity to benzene
among C+6 products (ꢀ90%) for anisole HDO reactions in the liquid
phase with toluene as a solvent could be obtained using bimetallic
FeMo phosphide (FeMoP) catalysts; however, high hydrogen pres-
sures ꢀ2.1 MPa and high reaction temperatures ꢀ673 K were
required to accomplish this selective deoxygenation [19]. MoO3,
which showed the highest specific rates for acetone HDO among
other reducible metal oxides such as V2O5, Fe2O3, CuO, and WO3
via a reverse Mars-van Krevelen mechanism, was also tested for
vapor-phase anisole HDO at 673 K under ambient pressure [27].
A mixture of deoxygenated aromatics was observed in which the
selectivity of benzene, toluene, xylene, and alkylbenzenes was
ꢀ60%, ꢀ20%, ꢀ6.5%, and ꢀ13%, respectively [27]. No observation
of sequential hydrogenation products of the aforementioned aro-
matics was reported, implying that this process is hydrogen effi-
cient [27].
Recently, transition metal carbides have been reported to selec-
tively remove oxygen from C2–C3 oxygenates [28,29], vegetable
oils [30], stearic acid [4], and guaiacol [31]. Herein, we report that
Mo2C is a selective HDO catalyst for phenyl ethers at low temper-
atures (ꢀ420–520 K) under atmospheric pressure with good cata-
lyst stability over the course of ꢀ50 h, in which benzene can be
formed almost exclusively with >90% selectivity among C+6 prod-
ucts. The high hydrogen efficiency of this process is evidenced by
the low selectivity (<9%) to cyclohexane in successive hydrogena-
tion reactions of benzene even at high hydrogen-to-oxygenate
molar ratios (ꢀ700) in the reactant feed. Plausible reaction mech-
anisms in which two distinct catalytic sites are involved for vapor-
phase anisole HDO on Mo2C catalysts are proposed based on
detailed kinetic measurements performed in this work. Metallic
sites on Mo2C catalysts are involved in the reaction as inferred
from the near invariance in benzene synthesis rates normalized
by the number of catalytic centers measured by ex situ CO chemi-
sorption, although the requirement of carbidic or oxycarbidic sites
for HDO chemistry remains ambiguous at this point.
Pd/Al2O3 (ꢀ1 wt% Pd) catalysts were prepared by an incipient
wetness method in which ꢀ1 g of alumina (Sasol North America
Inc., pretreated in a flow of dry air (ultrapure, Minneapolis Oxygen)
with a total flow rate ꢀ1.67 cm3 sꢁ1 at 723 K (0.012 K sꢁ1) for 4 h
prior to usage) was impregnated with an appropriate amount of
Pd solution (Pd(NO3)2ꢂ2H2O, ꢀ40% metal basis, Sigma Aldrich).
After impregnation, samples were dried in a static oven at
ꢀ363 K overnight, followed by treatment in a flow of dry air
(ꢀ1.67 cm3 sꢁ1) at 363 K (0.012 K sꢁ1) for 9 h and subsequently
heated at 823 K (0.025 K sꢁ1) for 4 h.
2.2. Materials characterization
2. Experimental methods
Unless otherwise mentioned, a representative fresh, passivated
Mo2C sample, sample#8 in Table 1, was used for the characterization
protocol discussed below. The bulk structures of both spent and
fresh samples were determined using X-ray diffraction (XRD, Bruker
D8 Discover 2D X-ray diffractometer with a two-dimensional
2.1. Catalyst synthesis
Mo2C catalysts were prepared based on a prior report with
ammonium molybdate tetrahydrate ((NH4)6Mo7O24ꢂ4H2O) as a
precursor [28,32]. Mo2C catalysts with varying number of CO
adsorption sites were obtained using varying temperature and flow
rate protocols that are described in detail below. An appropriate
amount (0.6–6 g) of ammonium molybdate tetrahydrate (sieved,
VÅNTEC-500 detector, Cu K X-ray radiation with a graphite
a
monochromator, and a 0.5 mm point collimator). The sample was
drop casted on a SiO2 zero-background holder and measured in
three measurement frames at 30° (2h), 60° (2h) and 90° (2h) with
a 600 s frameꢁ1 dwell (
D2h frame width of 35° (2h)). Two-
177–400
was loaded in a tubular quartz reactor (I.D. 10 mm) and purged
with typically
desired total flow rate (1.09–2.93 cm3 sꢁ1
l
m, Sigma, 99.98%, trace metal basis; typically ꢀ0.6 g)
dimensional images were then converted to one-dimensional
intensity vs. 2h for analysis. The crystallite size of the sample was
estimated using the Scherrer equation [35]. The particle morpholo-
gies of the sample were characterized via transmission electron
microscopy (TEM FEI Tecnai T12, 120 keV). For TEM sample prepara-
tion, the Mo2C catalyst was sonicated in dimethylformamide (DMF,
Sigma, 99.8% ACS reagent) for 2 h and a drop was placed onto a
holey-carbon Cu grid and then dried at 353 K in a vacuum oven
overnight prior to the TEM analysis. The BET surface area (Sg), pore
volume, and mesopore size distribution (obtained by Barrett–
Joyner–Halenda (BJH) analysis at desorption branch) were mea-
sured using N2 adsorption/desorption isotherms (Micromeritics
a
;
2.93 cm3 sꢁ1), consisting of either 15/85 vol% of CH4 (Matheson,
99.97%) and H2 (Minneapolis Oxygen, 99.999%) mixture or pure
H2 at room temperature (RT) for about 1–5 min (typically in CH4/
H2 mixture). The reactor was then heated in a three-zone split tube
furnace (Series 3210, Applied Test System) from RT to the first
target temperature (ꢀ618–623 K, typically 623 K) within 1.5 h,
and the temperature was then held for 12–24 h. Subsequently,
the sample was heated in a CH4/H2 (15/85 vol%) gas mixture (the
gas mixture either remained unchanged or CH4 was added to the

46
W.-S. Lee et al. / Journal of Catalysis 319 (2014) 44–53
Table 1
Preparation conditions for high surface area Mo2C samples with varying numbers of catalytic centers measured by CO chemisorption at 323 K.
Catalysts
AMTa (g)
CH4 flow rate (cm3 sꢁ1
)
H2 flow rate (cm3 sꢁ1
)
Temp profilec
CO uptake at 323 K
mole gꢁ1, STP)d
(
l
#1
#2
#3
#4
#5
#6
#7
#8
#9
#10
3.0
6.0
2.4
6.0
0.6
3.6
0.6
0.6
6.0b
1.5
0.44
0.44
0.44
0.44
0.44
0.44
0.29
0.44
1.33
0.42
0.65
2.49
2.49
2.49
2.49
2.49
1.66
2.49
8.33
2.33
CH4/H2-RT(1.5 h)-623 K[24 h](1.5 h)-873 K[6 h]-cool
CH4/H2-RT(1.5 h)-623 K[12 h](1.5 h)-978 K[3 h]-cool
H2-RT(1.5 h)-618 K[15 h]-CH4/H2(1.5 h)-863 K[2 h]-cool
H2-RT(1.5 h)-618 K[15 h]-CH4/H2(1.5 h)-863 K[2 h]cool
CH4/H2-RT(1.5 h)-623 K[12 h](1.5 h)-873 K[2 h]-cool
CH4/H2-RT(1.5 h)-623 K[12 h](1.5 h)-858 K[4 h]-cool
H2-RT(1.5 h)-618 K[15 h]-CH4/H2(1.5 h)-863 K[2 h]-cool
CH4/H2-RT(1.5 h)-623 K[12 h](1.5 h)-863 K[2 h]-cool
CH4/H2-RT(1.5 h)-623 K[0 h](1.5 h)-908 K[3 h]-cool
CH4/H2-RT(1.5 h)-628 K[5 h](1.5 h)-883 K[3 h]-cool
51
92
101
120
122
130
163
220
152
208
a
Ammonium molybdate tetrahydrate ((NH4)6Mo7O24ꢂ4H2O, sieved 177–400
l
m).
b
Samples were prepared by re-carburizing a mixture of multiple batches of Mo2C catalysts (ꢀ6 g) with different amounts of CO chemisorption sites using the conditions
shown in this table.
c
Detailed description for the preparation conditions can be found in Section 2.1. For the catalysts prepared from heating in the pure H2 environments first, such as sample
#3 (H2-RT(1.5 h)-618 K[15 h]-CH4/H2(1.5 h)-863 K[2 h]-cool), the switch of H2 to CH4/H2 was achieved by adding CH4 (0.44 cm3 sꢁ1) to the existing H2 flow (2.49 cm3 sꢁ1).
d
Except for sample #8 and #10 in which CO chemisorption were measured within one day of the Mo2C catalyst synthesis, the rest of the samples were stored in vials for a
few weeks prior to both CO chemisorption and the kinetic measurements. The CO chemisorption and the kinetic measurements were, however, carried out on the same day
for each sample listed above.
ASAP 2020). The sample was degassed (<6
l
mHg) at 523 K for at
were heated to at least 373 K via resistive heating to prevent
condensation of effluents sent to a mass spectrometer (MKS Cirrus
200 Quadrupole MS system) and a gas chromatograph (Agilent
6890) equipped with a methyl-siloxane capillary column (HP-1,
least 4 h before N2 adsorption. CO (Matheson, 99.5%) chemisorption
uptakes were measured at 323 K using a Micromeritics ASAP 2020
instrument. CO chemisorption measurements for all samples shown
in Table 1 were carried out following the procedure described below.
The cell loaded with the Mo2C catalyst (0.05–0.2 g) was first
50 m ꢃ 320
lm ꢃ 0.52
lm) connected to a flame ionization detec-
tor (FID). The concentration of analytes from reactor effluents was
quantified with an external standard method in which the
corresponding GC calibration curves with R2 values >0.96 and at
least 4 different concentrations ensured precise measurement of
the concentration of products of interest. The carbon balance was
typically better than 92%. The anisole conversion and product
selectivity are calculated as follows:
evacuated at 383 K (ꢀ2
l
mHg) for 0.5 h, and then treated in H2 at
723 K for 2 h, followed by degassing (ꢀ2
l
mHg) at 723 K for 2 h.
The sample was subsequently cooled to 323 K prior to the first
adsorption isotherm measurement (between 100 and 450 mmHg).
The cell was then degassed (ꢀ2
lmHg) to remove weakly adsorbed
species prior to the second isotherm. Both isotherms were
Anisole conversion ¼ ðsum of the carbon in productsÞout=ðmoles of carbon in anisoleÞin;
P
Cþ6 product selectivity ¼ ðmoles of C6þ product iÞ=ð moles of Cþ6 productsÞ:
C1 products selectivity ¼ ðmole of methane or methanolÞ=ðmole of methane þ mole of methanolÞ
extrapolated to zero pressure to determine the total (from the first
isotherm) and weakly adsorbed (from the second isotherm) species.
The amount of irreversibly or strongly adsorbed species was
obtained by the difference between two extrapolated values
[35,36]. The procedure for H2 chemisorption of 1 wt% Pd/Al2O3
catalysts used for H2–D2–CO titration experiments (Section 3.3) is
similar to that for the CO chemisorption of Mo2C catalysts except
the Pd sample was pretreated in H2 at 523 K for 1 h prior to H2
chemisorption at 323 K and irreversibly adsorbed H2 was used to
calculate the Pd metal dispersion by assuming a hydrogen-to-metal
stoichiometry of one.
Kinetic studies were conducted in a differential reactor bed in
which anisole conversion was controlled to be <15%. The absence
of heat and external mass transfer limitations was confirmed by
estimating Mears’ criteria (Table S1, SI). The benzene turnover rate
was independent of linear velocities of the reactants (Fig. S1, SI),
demonstrating that the reaction rate was not limited by external
mass transfer. Internal mass transfer limitations can be neglected
because the estimated Thiele modulus was significantly less than
1 (Table S2, SI).
H2ꢁD2 exchange reactions were performed by flowing a gas
mixture (ꢀ1.67 cm3 sꢁ1) consisting of anisole/D2/H2/Ar (vol%) =
ꢀ0.12/40/40/bal to a reactor packed with a mixture of ꢀ0.022 g
of Mo2C catalysts (sample #10 in Table 1) and ꢀ0.45 g quartz sand
(washed with 1 M HNO3 solution and treated in a flow of air at
1073 K for 2 h prior to usage), which had been tested under
standard reaction conditions (STD, reaction mixture consisting of
anisole/H2 (vol%) = 0.12/bal with total flow rate ꢀ1.67 cm3 sꢁ1 at
406 K) for ꢀ2 h. The net rates of H/D exchange were measured
by using online mass spectrometry (MS, MKS Cirrus 200 Quadru-
pole MS system). The mass spectrometer was calibrated by flowing
a gas mixture with varying H2/D2 compositions through a reactor
packed with a mixture of ꢀ0.2 g Pd/Al2O3 catalysts (Pd loading
ꢀ1 wt%, and metal dispersion ꢀ12% by H2 chemisorption
measured at 323 K) and ꢀ1 g quartz sand at ꢀ463 K to obtain the
2.3. Kinetic measurements of vapor-phase anisole HDO on Mo2C
catalysts
Vapor-phase anisole hydrodeoxygenation (HDO) rates were
measured using a tubular quartz reactor (I.D. 10 mm) with an outer
thermowell at the bottom of the reactor for holding a thermocou-
ple to monitor reaction temperature. Catalysts (0.03–0.14 g) were
treated in pure H2 (ꢀ1.67 cm3 sꢁ1) at ꢀ750 K for 1 h and cooled
down to a typical reaction temperature ꢀ423 K before introducing
the reaction mixture (ꢀ1.67 cm3 sꢁ1) consisting of ꢀ0.14/bal (vol%)
of anisole and H2. Anisole (Sigma, 99%, ReagentPlus) was added
using a syringe pump (KD Scientific, Model 100). All flow lines

W.-S. Lee et al. / Journal of Catalysis 319 (2014) 44–53
47
corresponding equilibrated H2/D2 mixture. The Pd catalysts were
activated by flowing H2 (ꢀ1.67 cm3 sꢁ1) at 573 K for 1 h with a
ramping rate ꢀ0.05 K sꢁ1 from RT before calibrating the mass spec-
trometer for HD. The error associated with HD quantification by
online mass spectrometry, calculated from 11 independent mea-
surements of the HD response factor, was found to be <10%. The
synthesis rates for sample #8 and #9 (ꢀ3 ꢃ 10ꢁ4 mole sꢁ1
ꢁ1
mole
, Table 2), which will be discussed in Section 3.3.
COsite
The crystallite size of b-Mo2C for the fresh, passivated Mo2C cat-
alyst (sample#8 in Table 1), estimated using the Scherrer equation
for the 2h peak at ꢀ52° (the peak intensity has contributions only
from b-Mo2C phase), was found to be ꢀ3.3 nm, which is compara-
ble to that reported by Boudart and coworkers for Mo2C catalysts
(ꢀ5.8 nm) prepared using MoO3 as the precursor [35]. The BET
surface area and the amount of irreversibly chemisorbed CO at
ꢀ323 K for a typical fresh, passivated Mo2C catalyst (sample #8
measured reversibility of H/D exchange, gHD, was estimated to be
ꢀ0.1. Rates of H/D exchange are reported as forward rates of H/D
exchange, which were obtained by dividing the net rate of H/D
exchange by (1 ꢁ
gHD).
in Table 1) were ꢀ116 m2 gꢁ1 and 220
l
mole gꢁ1 (STP) respec-
tively, which is in good agreement with results reported by
Thompson and coworkers [32]. Porous features of a typical fresh,
passivated Mo2C catalysts were evidenced by the following: (1)
the existence of a hysteresis loop observed in the relative pressure
(P/P0) range of 0.4–0.8, characteristic of multilayer adsorption
occurring inside mesopores in the N2 adsorption isotherms, as
shown in Fig. 2; (2) the observed contrast regions inside a repre-
sentative Mo2C particle from TEM analysis, as shown in Fig. 3.
Furthermore, the theoretical (100) interplanar distance matched
the spacing of the lattice fringes, confirming the presence of the
b-Mo2C structure. The average mesopore diameter and pore vol-
ume (at P/P0 ꢀ0.99) were found to be ꢀ6.4 nm (BJH desorption
branch) and 0.14 cm3 gꢁ1 respectively. A typical fresh, passivated
Mo2C catalyst was in polycrystalline form as evidenced by the
arrangement of discrete spots observed in the selected area
electron diffraction (SAED) of a typical Mo2C particle (not shown)
consistent with the results reported by Bell and coworkers [37].
The much larger particle size (Dp ꢀ70 nm, estimated by BET sur-
3. Results and discussion
3.1. Characterization of Mo2C
The absence of peaks assigned to the MoO3 crystalline phase, as
shown in Fig. 1, suggests that bulk carburization was achieved.
b-Mo2C (Joint Committee on Powder Diffraction Standards, No.
35-0787, 2h peaks at ꢀ34.355° (100), 37.979° (002), 39.393°
(101), 52.124° (102), 61.529° (110), 69.567° (103), 74.647°
(112), and 75.514° (201)) was identified as the major carbidic
phase and a-MoC1ꢁx as the minor phase based on the comparison
of the observed XRD peak intensities from both phases. The mate-
rials are, therefore, referred to as Mo2C formulations because the
Mo/C molar ratio is near two in both phases. An impurity phase
that contributed the signal at ꢀ26° (2h), however, was also found
in the fresh, passivated Mo2C catalyst (Fig. 1). Both MoO2 and
Mo(CO)6 might be the source of the impurity based on the compar-
ison between the observed XRD pattern and the references of
MoO2 (JCPDS No. 32-0671) and Mo(CO)6 (JCPDS No. 12-0691)
(see Fig. S2(a) in the SI). Although our XRD analysis does not allow
us to distinguish between Mo(CO)6 and MoO2, the impurity could
be residual MoO2 with poor crystallinity since the formation of
MoO2 was observed during the carburization process with ammo-
nium molybdate as the Mo2C precursor by Djéga-Mariadassou and
coworkers [34]. The observation of two different phases, b-Mo2C
face area (Sg) using the equation Dp ꢀ6 (
q
ꢃ Sg)ꢁ1 where
q is the
bulk density of the Mo2C catalyst, reported to be around
ꢀ760 kg mꢁ3 [38]), as compared to the crystallite size (ꢀ3.3 nm)
determined by XRD, also suggests the synthesized Mo2C catalyst
is polycrystalline.
3.2. Kinetics of vapor-phase anisole HDO on Mo2C catalysts
and
a
-MoC1ꢁx, and the impurity phase at ꢀ26° (2h) for the Mo2C
Fig. 4 shows the conversion and selectivity for vapor-phase ani-
sole HDO on a Mo2C catalyst (sample #8 in Table 1) over a temper-
ature range of 420–520 K at ambient pressure. The time-on-stream
behavior of the catalyst showed that <20% deactivation occurred
after two temperature cycles (420 K–520 K–472 K) over a time per-
iod of ꢀ50 h, implying Mo2C catalysts are significantly more stable
than Pt- or Co–Mo-based catalysts, which typically showed >20%
deactivation within ꢀ5 h [12,15,18]. The selectivities of all major
products (benzene, cyclohexane (Fig. 4(b)), methane, and methanol
(Fig. 4(c))) at a given temperature (e.g., 472 K) during the second
temperature cycle, were found to be the same as those measured
during the first temperature cycle, indicating that catalyst deacti-
vation was caused by a change in the number and not the chemis-
try or identity of the active sites.
There are two possible reasons for catalyst deactivation: (i) car-
bon deposition, which can directly block the active sites and/or
occlude catalyst mesopores resulting in the loss of the surface area
(and hence the active sites) [15,16,18], and (ii) slow oxidation of
the Mo2C catalyst to form an oxycarbide and/or oxide phases at
the surface or the sub-surface upon exposure to the oxygen-con-
taining reactant. A cause of catalyst deactivation is the formation
of carbonaceous species occurred during high temperature cycles
because no sign of catalyst deactivation was found for another
independent catalyst stability test at 420 K for more than 15 h.
(Fig. S3(a), SI); however, oxidation of the carbidic surface at higher
reaction temperatures cannot be excluded. Our research does not
distinguish between these deactivation mechanisms and instead
focuses on understanding the kinetics, mechanism, and site
requirements of vapor-phase anisole HDO.
catalyst synthesized in this work is consistent with those reported
by Thompson and coworkers [32]. The peak at ꢀ26° (2h), however,
was not observed for Mo2C catalysts prepared by re-carburization
(sample #9 in Table 1, passivated prior to XRD measurement) in a
CH4/H2 mixture flow (14/86 vol%, ꢀ9.66 cm3 sꢁ1) as shown in
Fig. S2(b) in the SI. The presence of this impurity phase, however,
was found to have a negligible effect on the catalytic performance
in vapor-phase anisole HDO, as inferred from similar benzene
1600
(101)
(002)
1400
1200
1000
(100)
(112)
(110)
800
600
(201)
(103)
(102)
400
200
0
15
30
45
60
75
90
2θ
Fig. 1. X-ray diffraction pattern for the fresh, passivated Mo2C catalyst (sample #8
in Table 1).

48
W.-S. Lee et al. / Journal of Catalysis 319 (2014) 44–53
Table 2
Catalytic performance (benzene turnover rate (TOR), benzene selectivity and apparent Ea) of vapor-phase anisole HDO for high surface area Mo2C samples with varying numbers
of catalytic centers measured by CO chemisorption at 323 K.
Catalystsa
CO uptake at 323 K
mole gꢁ1, STP)b
Benzene TOR at 423 K
Benzene selectivity
(%)c
Apparent activation energy,
( ꢃ 10ꢁ4 mole sꢁ1 mole
)
Ea (kJ moleꢁ1
)
ꢁ1
COsite
c
(
l
#1
#2
#3
#4
#5
#6
#7
#8
#9
#10
51
92
4.2
3.1
2.7
2.9
3.0
4.3
5.7
3.0d
2.7d
2.6e
93
91
91
91
90
94
94
92
91
92
75.8 1.8
72.7 0.8
71.5 1.3
69.3 4.5
71.0 0.4
67.8 1.8
69.3 1.8
ND
101
120
122
130
163
220
152
208
ND
ND
ND: not determined.
a
Detailed preparation conditions for the listed samples can be found in Table 1.
Except for sample #8 and #10 in which CO chemisorption were measured within one day of the Mo2C catalyst synthesis, the rest of the samples were stored in vials for a
b
few weeks prior to both CO chemisorption and the kinetic measurements. The CO chemisorption and the kinetic measurements were carried out on the same day for each
sample listed above.
c
The reported rate represents an average value at 423 K between 1–2 h time-on-stream with anisole conversion <14%. Reaction conditions: anisole/H2 = ꢀ0.14%/bal
(1.67 cm3 sꢁ1) at ꢀ423 K; catalyst loading: 80–250 mg.
d
Reaction temperature was ꢀ420 K. Apparent activation energy ꢀ70 kJ moleꢁ1 was used to calculate the rate at 423 K by extrapolation.
e
The reported rate represents an average value between 1–2 h time-on-stream with reaction conditions: anisole/H2 = ꢀ0.14%/bal (1.67 cm3 sꢁ1) at 406 K prior to H2–D2–
CO titration experiment shown in Fig. 9 with catalyst loading: ꢀ0.022 g. Apparent activation energy ꢀ70 kJ moleꢁ1 was used to calculate the rate at 423 K by extrapolation.
120
100
80
60
40
20
0
0.0
0.2
0.4
0.6
0.8
1.0
Relative Pressure (P/Po)
Fig. 2. N2 adsorption ( )/desorption ( ) isotherm at 77 K for the fresh, passivated
Mo2C catalyst (sample #8 in Table 1).
An independent kinetic study was conducted to assess the
structural evolution of Mo2C formulations during anisole HDO.
Steady-state rates of anisole HDO were measured on a Mo2C cata-
lyst (sample #9 in Table 1) at ꢀ420 K and ambient pressure
(Fig. S3(b), SI). The catalyst was then cooled down to RT in flowing
He (ꢀ3.33 cm3 sꢁ1); subsequently treated in a flowing 1% O2/He
mixture (ꢀ3–4 cm3 sꢁ1) for 1 h prior to XRD analysis. The catalyst
bed temperature increased from RT to ꢀ333 K upon exposure to
the 1% O2/He mixture suggesting that the catalyst, after exposure
to the reactant mixture, was still pyrophoric in nature. No notice-
able change was observed in the XRD pattern for the spent Mo2C
catalysts (Fig. S2(b), SI), suggesting the bulk structure of the cata-
lyst remained in carbidic form after the HDO reaction.
Fig. 3. A representative TEM image of the fresh, passivated Mo2C catalyst (sample
#8 in Table 1). Lattice fringes (indicated by arrow) corresponding to the (100) plane
of b-Mo2C were identified.
(523–673 K), but is comparable to the performance of bimetallic
FeMo phosphides (FeMoP) catalysts for liquid-phase anisole HDO
reaction reported by Hicks and coworkers [19]. While the overall
selectivity to deoxygenated products for vapor-phase anisole HDO
reaction was found to be similar (>90%) between the Mo2C catalysts
reported here and the MoO3 catalysts reported by Roman-Leshkov
and coworkers [27], negligible transalkylation activity was
observed for the Mo2C catalysts with the sum of the selectivity to
toluene and styrene being less than 2% at all anisole conversions.
No C2–C5 products were observed at all reaction temperatures,
suggesting that Mo2C catalysts are very selective in cleaving C–O
bonds instead of C–C bonds. The ratio of the sum of C1 synthesis
rates (methanol + methane, denoted as C1 rates) to the sum of C6
synthesis rates (benzene + cyclohexane, denoted as C6 rates) for
the data shown in Fig. 4 was found to be near 1 (Fig. S5 in the
Benzene selectivity was found to be >90% based on C+6 products
over anisole chemical conversions ranging from 10% to 100% for the
Mo2C formulations we report, which contrasts with the low
selectivity (<50%) to deoxygenated aromatics (i.e., benzene)
obtained using commercial hydrotreating (Co–Mo-based)
[18,22,23], Pt-based [12,15,21], transition metal phosphide (Ni2P,
MoP, and NiMoP) [17], or copper chromite [16] catalysts at
high hydrogen pressures (1–5 MPa) and high temperatures

W.-S. Lee et al. / Journal of Catalysis 319 (2014) 44–53
49
benzene–cyclohexane interconversion) was found to be much
smaller than 1.
(a)
550
500
450
400
100
80
60
40
20
0
Two possible explanations for the observed low selectivity to
cyclohexane during vapor-phase anisole HDO reaction are as fol-
lows: (i) the coverage of benzene decreases as the reaction temper-
ature increases, resulting in lower probability of the sequential
hydrogenation reaction, and (ii) the surface of Mo2C catalysts is
influenced by the presence of oxygen when using an oxygenate
as a reactant. Mo2C catalysts have comparable or even higher turn-
over rates compared to Ru catalysts for benzene hydrogenation at
near ambient temperatures and pressures [34,39,40]. A passivated
Mo2C sample, however, was found to be completely inactive for
benzene hydrogenation by Djéga-Mariadassou and coworkers
[34]. Fig. 5 shows benzene hydrogenation on Mo2C catalysts before
and after vapor-phase anisole HDO reaction at atmospheric pres-
sure. The cyclohexane specific rate was significantly suppressed
in the presence of the anisole, suggesting that the presence of oxy-
gen altered the hydrogenation functionality of the Mo2C catalysts.
Similar suppression in the benzene hydrogenation activity was
observed when replacing anisole with methanol or water (Fig. S4,
SI). This experimental observation is consistent with the inhibitive
relationship between the benzene hydrogenation turnover rate
and the residual oxygen content of the Mo2C catalysts reported
by Djéga-Mariadassou and coworkers [34]. We propose that the
remarkably high hydrogen efficiency of Mo2C catalysts in anisole
HDO is therefore a consequence of the presence of oxygen-contain-
ing species from the reactants, which significantly changes the
hydrogenation functionality of Mo2C catalysts. Along the same
lines, different selectivities (decomposition vs. dehydrogenation)
have been reported for cyclohexene reactions on C/W(110) pre-
treated with O2 at different temperatures, suggesting the possibil-
ity of manipulating the catalytic selectivity of transition metal
carbides by exposure to oxygen [41].
520 K
520 K
472 K
472 K
420 K
25
420 K
0
50
75 100 125 150 175 200
Time / ks
(b)
100
80
60
40
20
0
15
10
5
420K
520K
472K
420K
520K
472K
0
0
50
100
150
200
Time / ks
(c)
Methane selectivity increased at the expense of methanol selec-
tivity as the reaction temperature and anisole conversion increased
(Fig. 4(c) and Fig. 6), suggesting that the phenolic C–O bond
(ꢀ422 kJ moleꢁ1) was cleaved first to form methanol and benzene
as the primary products and methanol was subsequently hydro-
deoxygenated to form methane and water. This contrasts with
the high selectivity to phenol and the lack of methanol production
during anisole HDO over Co–Mo-based [22,23], transition metal
420K
520K
472K
420K
520K
472K
120
100
80
60
40
20
0
120
100
80
60
40
20
0
1400
1200
1000
800
600
400
200
0
210
180
150
120
90
Anisole HDO
0
50
100
150
200
Time / ks
Fig. 4. (a) Conversion of anisole, (b) measured selectivity for benzene
cyclohexane ( ) and sum of toluene, styrene and phenol ( ) and (c) measured
( ),
Benzene
hydrogenation
Benzene
selectivity for methane (I) and methanol ( ) on Mo2C (sample #8 in Table 1) as a
hydrogenation
60
function of temperature and time-on-stream at 1 atm. Reaction conditions: anisole/
benzene
ꢁ1
H
2 = ꢀ0.14%/bal (total flow rate ꢀ1.67 cm3 sꢁ1), space velocity ꢀ17 cm3 sꢁ1
g
and
cat
30
catalyst loading ꢀ97 mg.
cyclohexane
0
35
0
5
10
15
20
25
30
SI), suggesting that C1 species are formed in hydrogenolysis reac-
tions of the Ar–OCH3 bond in anisole, rather than dissociation of
C+6 products. C1 species are therefore, not included in the C+6 selec-
tivity calculation.
Low cyclohexane selectivity (<9%; no cyclohexene was
observed) showed that vapor-phase anisole HDO on Mo2C catalysts
is hydrogen efficient. The observed lower cyclohexane selectivity
at the higher reaction temperatures, however, was not thermody-
namically, but kinetically controlled because the calculated
Time / ks
Fig. 5. Cyclohexane production rate measured from either benzene hydrogenation
reaction (j, shaded areas) or vapor-phase anisole HDO reaction ( ) on a Mo2C
catalyst. Benzene synthesis rate ( ) indicates deoxygenation activity. The catalyst
was first tested for benzene hydrogenation for ꢀ10 ks (feed composition: benzene/
H
2 = 0.23%/bal with total flow rate ꢀ1.67 cm3 sꢁ1) and then switched to vapor-
phase anisole HDO reaction (feed composition: anisole/H2 = 0.14%/bal with total
flow rate ꢀ1.67 cm3 sꢁ1), followed by restoring the reaction conditions to benzene
hydrogenation. The reaction temperature was maintained at ꢀ423 K under ambient
pressure during the switch of aforementioned two different reactions. Catalyst
loading was ꢀ150 mg.
approach to equilibrium (
g
, defined as [Pcyclohexane]/([Pbenzene] ꢃ
[PH2]3 ꢃ Keq); where Keq is the equilibrium constant for

50
W.-S. Lee et al. / Journal of Catalysis 319 (2014) 44–53
100
80
60
40
20
0
100
Anisole pressure / kPa
(a)
-16.5
80
8
6
4
2
0
-17.0
-17.5
-18.0
-18.5
60
40
20
0
0
20
40
60
80
100
Anisole conversion / %
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
)
Ln(H2 pressure / atm
Fig. 6. Selectivity of methane (I) and methanol ( ) as a function of anisole
conversion; a range of anisole conversion was achieved by using different amounts
of catalyst or varying the reaction temperature. Reaction conditions: anisole/
H2 = ꢀ0.14%/bal (total flow rate ꢀ1.67 cm3 sꢁ1), and catalyst loading ꢀ11 to
ꢀ140 mg. The dashed lines are drawn to guide the eyes.
(b)
1.0
0.8
0.6
0.4
0.2
0.0
sulfide [17,19], and Pt-based catalysts [21], wherein the weaker
aliphatic C–O bond (ꢀ339 kJ moleꢁ1) was preferentially cleaved.
The occurrence of demethoxylation, instead of demethylation, for
vapor-phase anisole HDO on Mo2C was further supported by the
very low selectivity to phenol, <1%, at all anisole conversions
(0.6–100%).
ꢁ1
Benzene synthesis rate (mole sꢁ1
g
) was independent of ani-
Cat
sole chemical conversion (0.6–14%) achieved using different
amounts of catalysts (sample #8 in Table 1 and 0.03–0.14 g) as
shown in Fig. S6 in the SI. This experimental result shows there
is no product inhibition and/or the effects of product concentration
gradients on the benzene synthesis rate can be neglected at the
current reaction conditions.
Fig. 7(a) shows that the rate of benzene synthesis does not
depend on anisole pressure (0.06–1.1 kPa), indicating the surface
is predominantly covered by anisole-derived intermediates. The
concurrent near half-order (ꢀ0.6) dependence of benzene synthe-
sis rates on H2 pressure (Fig. 7(a)) at constant anisole pressure
(ꢀ0.14 kPa) suggests that dissociative adsorption of hydrogen
occurs, however, on sites that are distinct from those required
for anisole adsorption because if competition between anisole
and molecular hydrogen existed for the same site, we would
observe inhibition of catalytic rates by anisole. The hydrogen order
was also found to be independent of the anisole concentration
(0.12–0.96 kPa) as shown in Fig. 7(b), confirming that distinct sites
are required for hydrogen and anisole activation.
It was suggested by Ko and coworkers that CO can molecularly
chemisorb on exposed on-top sites of Mo on a carbided Mo (100)
surface, while dissociative H2 chemisorption can occur across four-
fold hollow sites of the carbide surfaces [42]. These surface science
studies are consistent with positive reaction orders for both H2 and
CO in CO hydrogenation reactions, suggesting that CO adsorption
does not inhibit H2 chemisorption [43]. Similarly, we postulate ani-
sole adsorption can occur on sites resembling those required for CO
chemisorption (CO chemisorption sites are involved in vapor ani-
sole HDO as we discuss in Section 3.3), while another type of site,
such as fourfold hollow sites, may be responsible for H2 activation.
The apparent activation energy was found to be ꢀ76
1.3 kJ moleꢁ1 (Fig. S7, SI). Fig. 8(a) and (b) show that both the
hydrogen order and the apparent activation energy were invariant
with anisole conversion (ꢀ0.6–15%, achieved by varying catalyst
weight over a range of 0.03–0.14 g), which confirms that the
products of anisole HDO do not have measurable kinetic effects
on the benzene synthesis rate under these reaction conditions.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Anisole pressure / kPa
Fig. 7. (a) Effect of H2 pressure ((j) varied from ꢀ0.1 to ꢀ1 atm (balance He) at
ꢀ0.14 kPa anisole at 423 K) and anisole pressure (( ) varied from ꢀ0.06 to ꢀ1 kPa,
balance H2 at 423 K) and (b) hydrogen order measured at different anisole
concentrations. Experimental data obtained at 423 K, total pressure = 1 atm.
A plausible reaction mechanism involving the addition of disso-
ciated hydrogen to adsorbed anisole is proposed to account for the
experimentally observed reaction order dependencies: zero order
for anisole, ꢀ1/2 order for hydrogen and no kinetic effects of
anisole HDO products on the benzene synthesis rate. Hydrogen
molecules undergo dissociative adsorption on site 1 (S1), while
anisole molecules (R represents a C6H5 functional group) adsorb
on a distinct site 2 (S2) [Eqs. (1) and (2)]. The dissociated hydrogen
first binds to the oxygen atom of the adsorbed anisole to give an
anisole-derived intermediate [Eq. (3)] that facilitates the cleavage
of the phenolic C–O bond, presumably because of the strong
interaction between the carbidic surface [44–46] and the benzene
ring, to generate methanol (4), followed by a sequential addition of
dissociated hydrogen to form benzene adsorbed on S2 [Eq. (5)]. The
adsorbed benzene can then desorb [Eq. (6)] to complete the
catalytic cycle.
H2 þ 2S1 $ 2H ꢁ S1
ð1Þ
ð2Þ
ð3Þ
ð4Þ
ð5Þ
ð6Þ
R ꢁ O ꢁ CH3 þ S2 $ R ꢁ O ꢁ CH3 ꢁ S2
R ꢁ O ꢁ CH3 ꢁ S2 þ H ꢁ S1 $ R ꢁ OH ꢁ CH3 ꢁ S2 þ S1
R ꢁ OH ꢁ CH3 ꢁ S2 ! R ꢁ S2 þ CH3OH
R ꢁ S2 þ H ꢁ S1 $ R ꢁ H ꢁ S2 þ S1
R ꢁ H ꢁ S2 $ R ꢁ H þ S2
Based on the assumption that (1) cleavage of the strong pheno-
lic C–O bond is the rate-determining step [Eq. (4)], (2) all other
steps are at quasi-equilibrium, and (3) the most abundant reactive

W.-S. Lee et al. / Journal of Catalysis 319 (2014) 44–53
51
is the rate-determining step [Eq. (5)], (2) adsorbed anisole
(R–O–CH3–S2) and empty sites are the MARI for S2 and S1 respec-
tively, the same benzene synthesis rate dependencies with zero
order in anisole and ꢀ1/2 order in H2 can be derived.
1.0
0.8
0.6
0.4
0.2
0.0
(a)
The rate equations for benzene synthesis derived from the pro-
posed mechanisms described above can give zero order in anisole
and half order in H2. If we consider that (i) there are two types of
sites involved in this reaction, one (S1) for H2 dissociation and the
other (S2) for anisole activation [Eqs. (1) and (2)], (ii) a non-uni-
form surface of S1 sites and a uniform surface of S2 sites, and (iii)
considering step [Eq. (3)] to be the rate-determining step, the frac-
tional order of 0.6 in H2 can be derived, as briefly shown below (A
detailed derivation can be found in Section 2 in the SI):
We consider [j] such ensembles for the sites S1 and the concen-
tration of the dissociatively adsorbed hydrogen in ensemble j can
be derived from site balance in ensemble j and the equilibrium
of [Eq. (1)]:
0
3
6
9
12
15
18
Anisole conversion / %
1
ꢀ
ꢁ
2
½L1ꢄj
1
k1
2
½H ꢁ S1ꢄj ¼
ꢂ ½H2ꢄ ꢂ
ð8Þ
ꢂ
ꢃ
1
2
100
80
60
40
20
0
1
2
k
(b)
ꢁ1
k1
1 þ ½H2ꢄ
k
ꢁ1
The turnover rate, Vj,t in each ensemble j for benzene synthesis
can be expressed as
1
1
2
2
k3k1 ꢂ ½R ꢁ O ꢁ CH3 ꢁ S2ꢄ ꢂ ½H2ꢄ
Vj;t
¼
ð9Þ
1
2
1
2
1
2
kꢁ1 þ k1½H2ꢄ
According to Boudart and Djega-Mariadassou [49], the rate
constants can be related to an equilibrium constant and can be
expressed as
k1 ¼ ko expð
a
ðt ꢁ t0ÞÞ
ð10Þ
ð11Þ
ð12Þ
1
k3 ¼ ko expðð
a
ꢁ 1Þðt ꢁ t0ÞÞ
3
¼ koꢁ1 expðð
ꢁ 1Þðt ꢁ t0ÞÞ
a
0
3
6
9
12
15
18
k
ꢁ1
Anisole conversion / %
where k01, kꢁ0 1, and k03 are rate constants in the limit of full coverage;
a
Fig. 8. (a) Hydrogen order and (b) apparent activation energy as a function of
anisole conversion. Experimental data obtained at 423 K, H2 pressure varied from
ꢀ0.1 to ꢀ1 atm (balance He) at ꢀ0.14 kPa anisole, total pressure = 1 atm. Anisole
conversion was adjusted by using different amounts of Mo2C catalysts (0.03–0.14 g)
is a so-called transfer coefficient, which correlates the forward
and reverse rate constant of an elementary step to its equilibrium
constant; t is dimensionless affinity defined as positive in the direc-
A0
RT
with total flow rate ꢀ1.67 cm3 sꢁ1
.
tion of adsorption in which t ¼ and t0, t1 represent the upper
bound and lower bound respectively.
An average turnover rate, vt can be obtained by integrating the
rates on each ensemble over the entire surface:
intermediate (MARI) for site 2 (S2) is anisole (R–O–CH3–S2) and site
1 (S1) is predominantly empty, a rate equation for benzene synthe-
sis consistent with the experimentally observed zero-order depen-
dence in anisole and half-order dependence in H2 can be derived.
Density functional theory (DFT) calculations by Ren et al. have
suggested that addition of dissociated hydrogen to an adsorbed
propanol molecule on Mo2C catalysts occurs with an energy barrier
exceeding 1 eV, which is similar to the step shown in [Eq. (3)]. If we
consider this step [Eq. (3)] to be the rate-determining step, MARI
for site 1 (S1) is empty sites, and the coverage of the adsorbed ani-
sole (R–O–CH3–S2) species to be much higher than that for all other
adsorbates on S2 sites, a rate equation for benzene synthesis that
can account for experimentally observed zero order in anisole
and half order in H2 can also be derived.
Alternatively, considering that oxygen can bind strongly to the
surface of Mo2C catalysts [28,47,48], the cleavage of the phenolic
C–O bond in surface-bound anisole might be facilitated by an adja-
cent empty site [Eq. (7)], instead of reacting with dissociated
hydrogen to form the active intermediate (R–OH–CH3–S2, [Eq.
(4)]).
Z
t0
1
L1
vt
¼
Vj;tdL0
ð13Þ
j
t1
A continuous site distribution is assumed [49] with the form:
½L1ꢄ
c
dL0j ¼
expðꢁ
c
ðt ꢁ t0ÞÞdt
ð14Þ
expð fÞ ꢁ 1
c
where
c
is a constant characterizing the site distribution and dL0j is
the concentration of sites with a standard affinity for adsorption of
dissociated hydrogen between A0 and A0 + dA0 in the ensemble j and
f = t0 ꢁ t1.
The integrated solution is (the detailed integration procedure
can be found in the Section 2 in the SI):
"
#
2mꢁ1
1
2
0
kꢁ1
2
2
p
021
1
1
2
1
1
2
k01 ꢂ½H2
ꢄ
c
k03k ꢂ ½R ꢁ O ꢁ CH3 ꢁ S2ꢄ ꢂ ½H2ꢄ
vt
¼
021
expð
cfÞ ꢁ 1
sinðð2m ꢁ 1Þ Þ
p
k
ꢁ1
ð15Þ
R ꢁ O ꢁ CH3 ꢁ S2 þ S2 ! R ꢁ S2 þ CH3O ꢁ S2
ð7Þ
where m =
Therefore;
a c
ꢁ
Based on the assumptions that (1) the addition of dissociated
hydrogen to the benzene ring intermediate adsorbed on S2 (R–S2)
1ꢁm
vt / ½R ꢁ O ꢁ CH3 ꢁ S2ꢄ ꢂ ½H2ꢄ
ð16Þ

52
W.-S. Lee et al. / Journal of Catalysis 319 (2014) 44–53
The S2 sites would then be treated with uniform surface kinetics
and if we assume the MARI on S2 is adsorbed anisole species, the
order in anisole would be zero and a fractional order of 0.6 for H2
can be obtained with m = 0.4.
1.0
0.9
1.0
0.9
STD
3.3. Site requirements of molybdenum carbide catalysts for vapor-
phase anisole HDO
0.4
0.3
0.2
0.1
0.0
0.4
0.3
0.2
0.1
0.0
The existence of basic, acidic, and metallic catalytic sites on
Mo2C catalysts has been noted in temperature programmed
desorption measurements of probe molecules such as NH3, CO2,
and CO, together with probe reaction studies [34,35,50]. The
involvement of metallic sites in vapor-phase anisole HDO reaction
was demonstrated by the near invariance in the turnover rate
(TOR) of catalytic hydrodeoxygenation at 420 K (average value
between 1–2 h time-on-stream) normalized by the number of cat-
alytic centers measured by CO chemisorption (ex situ measure-
ment using Micromeritics 2020) at 323 K as shown in Table 2.
This is also observed for the sample (#9 in Table 2) prepared by
re-carburization of a mixture of multiple batches of Mo2C catalysts,
which reinforces the idea that the number of CO chemisorption
sites is a relevant descriptor to assess the number of catalytic cen-
ters involved in the vapor-phase HDO of anisole on Mo2C. Further-
more, the presence of the species that give a diffraction peak at
ꢀ26° (2h) has no measurable effects on HDO catalytic rates or ben-
0.00
0.05
0.10
0.15
Co-fed CO flow rates / cm3
s
-1
Fig. 9. Effects of CO co-feed during vapor-phase anisole hydrodeoxygenation on
normalized benzene synthesis ( ) and H/D exchange forward rates ( ). Forward
rates of H/D exchange and benzene synthesis rates with a CO co-feed (from 0.0125
to 0.125 cm3 sꢁ1) were normalized to those measured under standard conditions
(STD, in the absence of CO co-feed) prior to the CO titration. See Table S3 in the SI
for measured rates of H/D isotopic exchange and benzene synthesis. CO was co-fed
into the reactant mixture comprising anisole/D2/H2/Ar (vol%) = ꢀ0.12/40/40/bal
with a total flow rate ꢀ1.67 cm3 sꢁ1 at reaction temperature ꢀ406 K under ambient
pressure; Mo2C loading ꢀ0.022 g (sample #10 in Table 1). The dashed lines are
drawn to guide the eyes.
zene selectivity as the normalized benzene synthesis rate
ꢁ1
(mole sꢁ1 mole
) and benzene selectivity for sample #8 (with
COsite
with kinetics studies and proposed mechanisms discussed in Sec-
tion 3.2 above to rationalize the measured concurrent zero-order
dependence in anisole and near half-order dependence in H2
observed experimentally.
It is challenging to compare Mo2C formulations we report with
other catalysts used for vapor-phase anisole HDO because (i) the
rates reported are typically at varying conversions [15,19,27], (ii)
fast catalyst deactivation is encountered in some cases [28], (iii)
product inhibition may become relevant for other catalysts, and
(iv) the identity of the active sites is not known. Since benzene is
not necessarily the dominant product from anisole HDO reactions,
the specific rate for deoxygenation products per gram of catalyst
and/or the number of active centers assessed by chemical adsorp-
the XRD peak at ꢀ26 (2h), Fig. 1) and #9 (without the XRD peak
ꢀ26 (2h), Fig. S2(b), SI) were almost identical (ꢀ3 ꢃ 10ꢁ4 mole sꢁ1
ꢁ1
mole
and ꢀ91%, respectively). The near invariance in mea-
COsite
sured apparent activation energies (ꢀ70 kJ moleꢁ1) and benzene
selectivity (ꢀ92%) for samples reported in Table 2 suggests that
the identity of the active sites and/or the HDO reaction mechanism
are similar on all tabulated Mo2C samples. These experimental
results also confirm the absence of heat and mass transfer limita-
tions because the benzene synthesis rate normalized per CO
chemisorption site was found to be similar, which satisfies the
Koros-Nowak criterion [51].
The existence of multiple active sites on Mo2C catalysts was
further tested by measuring forward rates of H/D exchange and
benzene synthesis with and without the presence of a CO co-feed
during vapor-phase anisole HDO reactions (Fig. 9). The measured
forward rates of H/D exchange and benzene synthesis for Mo2C
´
tion experiments is compared. Roman-Leshkov and coworkers
reported specific rates for deoxygenation of anisole on MoO3
ꢁ1
catalysts to be ꢀ7.5 ꢃ 10ꢁ6 mole sꢁ1
g
with benzene, toluene,
Cat
xylene and alkylbenzenes as major products at 673 K and 1 atm
catalysts (sample #10 in Table 1) atꢁ4106 K in the absence of CO were
ꢁ1
found to be ꢀ3.4 mole
s
moleCOsite and ꢀ8.7 ꢃ 10ꢁ5 mole sꢁ1
[27] and that for a 1 wt% Pt/H-BEA bifunctional catalyst to be
HD
ꢁ1
3.3 ꢃ 10ꢁ1 mole sꢁ1 mole
with benzene, toluene, xylene
ꢁ1
mole
, respectively. The forward rate of H2–D2 exchange for
surface Pt
COsite
and C+9 aromatics as major deoxygenation products at 673 K at
1 atm [15]. Hicks and coworkers reported that ꢀ92% anisole
conversion with deoxygenation product selectivity ꢀ96% (90% for
benzene and 6% for cyclohexane) could be achieved by a FeMo
bimetallic phosphides catalyst (FeMoP) in 0.5 h in a liquid-phase
noble metal based catalysts was reported to be 90–500 mol sꢁ1
ꢁ1
mole
for SiO2 supported noble metal catalysts by Goel
surf metal
et al. [52]. The much lower forwardꢁr1ates of H/D exchange for
Mo2C catalysts (ꢀ3.4 moleHD sꢁ1 moleCOsite) compared to those for
noble metal based catalysts might be attributed to the presence
of anisole, an oxygen-containing molecule, which inhibits the sur-
face hydrogenation functionality of Mo2C formulations as discussed
in Section 3.2 and Fig. 5.
Fig. 9 shows that the extent of inhibition by a CO co-feed
(ꢀ0.0125 to 0.125 cm3 sꢁ1) for the forward rates of H/D exchange
and benzene synthesis was different. This experimental observa-
tion suggests that CO inhibits benzene synthesis, presumably via
blocking the sites responsible for anisole adsorption and/or activa-
tion, and that H2 activation likely occurs on a distinct site, since
otherwise, the normalized forward rates of H/D exchange would
be suppressed to the same extent as the normalized benzene syn-
thesis rates upon the introduction of CO (see Table S3 in the SI for
additional information). The inference of the involvement of dis-
tinct catalytic sites for the activation of H2 and oxygen-containing
molecules on Mo2C catalysts from this titration study is consistent
batch reactor, which corresponds to a specific deoxygenation rate
ꢁ1
of ꢀ1.7 ꢃ 10ꢁ3 mole sꢁ1 mole
at 673 K under 4.2 MPa H2 [19].
metal
The deoxygenation rate based on the rate of benzene synthesis
(excluding cyclohexane) for Mo2C formulations (ꢀ3 ꢃ 10ꢁ4 at
423 K, Table 2) extrapolated to 673 K based on the apparent Ea
ꢀ70 kJ moleꢁ1 (Table 2) and under ambient pressure is ꢀ4.8 ꢃ
10ꢁ1 mole sꢁ1 mole
.
ꢁ1
COsites
Density functional theory (DFT) calculations by Rodriguez and
coworkers have suggested that Mo oxycarbide is formed on the
surface of a Mo2C catalyst during water gas shift (WGS) and is
responsible for the experimentally observed high WGS activity
[47]. The authors proposed that chemisorbed oxygen from water
dissociation could be destabilized by the C-terminated Mo2C
(001) surface, which facilitates intermediate strength binding of
reaction intermediates and therefore enhances the rate of the

W.-S. Lee et al. / Journal of Catalysis 319 (2014) 44–53
53
[2] J. Zakzeski, P.C.A. Bruijnincx, A.L. Jongerius, B.M. Weckhuysen, Chem. Rev. 110
(2010) 3552–3599.
[3] G.W. Huber, J.A. Dumesic, Catal. Today 111 (2006) 119–132.
[4] R.W. Gosselink, D.R. Stellwagen, J.H. Bitter, Angew. Chem. Int. Ed. 52 (2013)
5089–5092.
[5] J.Y. He, C. Zhao, J.A. Lercher, J. Am. Chem. Soc. 134 (2012) 20768–20775.
[6] N. Yan, Y.A. Yuan, R. Dykeman, Y.A. Kou, P.J. Dyson, Angew. Chem. Int. Ed. 49
(2010) 5549–5553.
[7] C. Zhao, Y. Kou, A.A. Lemonidou, X.B. Li, J.A. Lercher, Angew. Chem. Int. Ed. 48
(2009) 3987–3990.
WGS reaction. In this work, since the oxygen in anisole can be
removed as either water or methanol, depending on anisole chem-
ical conversion, the possibility of in situ formation of oxycarbide
phase during vapor-phase HDO, therefore, cannot be excluded.
We also note that the existence or involvement of carbidic or oxyc-
arbidic phases [53] cannot be inferred from the CO adsorption
studies we report above and the identification of working active
phase during HDO reaction requires additional characterization.
Although the identity of the two active sites involved in vapor-
phase anisole HDO on Mo2C catalysts cannot be inferred from
our results, we report that Mo2C is a stable and selective catalyst
for deoxygenation of aromatic ethers at ambient pressure and
low temperatures for the synthesis of aromatics and report kinetic
and chemical titration studies to infer the involvement of (i) two
distinct sites in this chemistry, and (ii) metallic sites in this chem-
istry that can be titrated by CO to rigorously calculate catalytic
rates.
[8] D.C. Elliott, Energy Fuels 21 (2007) 1792–1815.
[9] H.Y. Zhao, D. Li, P. Bui, S.T. Oyama, Appl. Catal. A 391 (2011) 305–310.
[10] A. Centeno, E. Laurent, B. Delmon, J. Catal. 154 (1995) 288–298.
[11] B. Donnis, R.G. Egeberg, P. Blom, K.G. Knudsen, Top. Catal. 52 (2009) 229–240.
[12] M.A. Gonzalez-Borja, D.E. Resasco, Energy Fuels 25 (2011) 4155–4162.
[13] D. Meier, J. Berns, O. Faix, U. Balfanz, W. Baldauf, Biomass Bioenergy 7 (1994)
99–105.
[14] J.Q. Bond, D.M. Alonso, D. Wang, R.M. West, J.A. Dumesic, Science 327 (2010)
1110–1114.
[15] X.L. Zhu, L.L. Lobban, R.G. Mallinson, D.E. Resasco, J. Catal. 281 (2011) 21–29.
[16] K.L. Deutsch, B.H. Shanks, Appl. Catal. A 447 (2012) 144–150.
[17] K.L. Li, R.J. Wang, J.X. Chen, Energy Fuels 25 (2011) 854–863.
[18] C.V. Loricera, B. Pawelec, A. Infantes-Molina, M.C. Alvarez-Galvan, R. Huirache-
Acuna, R. Nava, J.L.G. Fierro, Catal. Today 172 (2011) 103–110.
[19] D.J. Rensel, S. Rouvimov, M.E. Gin, J.C. Hicks, J. Catal. 305 (2013) 256–263.
[20] R.C. Runnebaum, R.J. Lobo-Lapidus, T. Nimmanwudipong, D.E. Block, B.C.
Gates, Energy Fuels 25 (2011) 4776–4785.
4. Conclusions
In summary, molybdenum carbide catalysts can convert anisole
to benzene in the vapor phase with unprecedented selectivity
(>90% among C+6 products), high hydrogen efficiency, (<9% selectiv-
ity for the undesired sequential hydrogenation product, cyclohex-
ane) and stable rates at low reaction temperatures (420–520 K)
and ambient pressure. The stronger phenolic C–O bond, instead
of the weaker aliphatic C–O bond (ꢀ422 vs. ꢀ339 kJ moleꢁ1) in ani-
sole was preferentially cleaved during the HDO reaction. The prod-
ucts of anisole HDO showed no measurable kinetic inhibition
effects on the benzene synthesis rate. At least two distinct sites
are required and metallic sites are involved in vapor-phase anisole
HDO on Mo2C catalysts. The extent of inhibition in benzene syn-
thesis rates was found to be different from that in forward rates
of H/D exchange upon co-feeding CO in the reaction mixture, sug-
gesting that the catalytic sites for H2 activation are distinct from
those for anisole adsorption and/or activation. Taken together with
the remarkable selectivity for benzene, the relatively low reaction
temperature, and low H2 pressure required to break the phenolic
C–O bond, these initial findings show molybdenum carbide is a
promising catalyst for catalytic deoxygenation with particular rel-
evance to lignin-derived phenolic ethers.
[21] R.C. Runnebaum, T. Nimmanwudipong, D.E. Block, B.C. Gates, Catal. Lett. 141
(2011) 817–820.
[22] S.J. Hurff, M.T. Klein, Ind. Eng. Chem. Fundam. 22 (1983) 426–430.
[23] A.L. Jongerius, R. Jastrzebski, P.C.A. Bruijnincx, B.M. Weckhuysen, J. Catal. 285
(2012) 315–323.
[24] S.T. Oyama, T. Gott, H.Y. Zhao, Y.K. Lee, Catal. Today 143 (2009) 94–107.
[25] S.T. Oyama, X. Wang, F.G. Requejo, T. Sato, Y. Yoshimura, J. Catal. 209 (2002) 1–
5.
[26] Y.Y. Shu, S.T. Oyama, Chem. Commun. (2005) 1143–1145.
[27] T. Prasomsri, T. Nimmanwudipong, Y. Roman-Leshkov, Energy Environ. Sci. 6
(2013) 1732–1738.
[28] H. Ren, W. Yu, M. Salciccioli, Y. Chen, Y. Huang, K. Xiong, D.G. Vlachos, J.G.G.
Chen, ChemSusChem 6 (2013) 798–801.
[29] W.T. Yu, Z.J. Mellinger, M.A. Barteau, J.G.G. Chen, J. Phys. Chem. C 116 (2012)
5720–5729.
[30] J.X. Han, J.Z. Duan, P. Chen, H. Lou, X.M. Zheng, H.P. Hong, ChemSusChem 5
(2012) 727–733.
[31] A.L. Jongerius, R.W. Gosselink, J. Dijkstra, J.H. Bitter, P.C.A. Bruijnincx, B.M.
Weckhuysen, ChemCatChem 5 (2013) 2964–2972.
[32] J.A. Schaidle, A.C. Lausche, L.T. Thompson, J. Catal. 272 (2010) 235–245.
[33] J. Patt, D.J. Moon, C. Phillips, L.T. Thompson, Catal. Lett. 65 (2000) 193–195.
[34] J.S. Choi, G. Bugli, G. Djega-Mariadassou, J. Catal. 193 (2000) 238–247.
[35] J.S. Lee, S.T. Oyama, M. Boudart, J. Catal. 106 (1987) 125–133.
[36] F.H. Ribeiro, R.A.D. Betta, G.J. Guskey, M. Boudart, Chem. Mater. 3 (1991) 805–
812.
[37] G.S. Ranhotra, G.W. Haddix, A.T. Bell, J.A. Reimer, J. Catal. 108 (1987) 24–39.
[38] E.A. Blekkan, P.H. Cuong, M.J. Ledoux, J. Guille, Ind. Eng. Chem. Res. 33 (1994)
1657–1664.
[39] J.S. Lee, M.H. Yeom, K.Y. Park, I.S. Nam, J.S. Chung, Y.G. Kim, S.H. Moon, J. Catal.
128 (1991) 126–136.
Acknowledgments
[40] C. MarquezAlvarez, J.B. Claridge, A.P.E. York, J. Sloan, M.L.H. Green, Stud. Surf.
Sci. Catal. (1997) 485–490.
This research was supported by Office of Basic Energy Sciences,
the U.S. Department of Energy under award number no.
DE-SC0008418 (DOE Early Career Program). Part of this work was
carried out in the Characterization Facility of the University of Min-
nesota, which receives partial support from the National Science
Foundation through the MRSEC program. We thank Professor Fabio
Ribeiro of Purdue University for helpful technical discussions and
Mr. Minje Kang for preparing the Pd/Al2O3 catalyst used in this
research.
[41] N. Liu, S.A. Rykov, J.G. Chen, Surf. Sci. 487 (2001) 107–117.
[42] E.I. Ko, R.J. Madix, Surf. Sci. 100 (1980) L449–L453.
[43] G.S. Ranhotra, A.T. Bell, J.A. Reimer, J. Catal. 108 (1987) 40–49.
[44] X.C. Liu, A. Tkalych, B.J. Zhou, A.M. Koster, D.R. Salahub, J. Phys. Chem. C 117
(2013) 7069–7080.
[45] A.S. Rocha, A.B. Rocha, V.T. da Silva, Appl. Catal. A 379 (2010) 54–60.
[46] B.J. Zhou, X.C. Liu, J. Cuervo, D.R. Salahub, Struct. Chem. 23 (2012) 1459–1466.
[47] P. Liu, J.A. Rodriguez, J. Phys. Chem. B 110 (2006) 19418–19425.
[48] A.J. Medford, A. Vojvodic, F. Studt, F. Abild-Pedersen, J.K. Norskov, J. Catal. 290
(2012) 108–117.
[49] M. Boudart, G. Djéga-Mariadassou, Kinetics of Heterogeneous Catalytic
Reactions, Princeton University Press, Princeton, 1984.
[50] S.K. Bej, L.T. Thompson, Appl. Catal. A 264 (2004) 141–150.
[51] S. Koso, Y. Nakagawa, K. Tomishige, J. Catal. 280 (1967) 221–229.
[52] S. Goel, Z.J. Wu, S.I. Zones, E. Iglesia, J. Am. Chem. Soc. 134 (2012) 17688–
17695.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2014.07.025.
[53] H.H. Hwu, M.B. Zellner, J.G.G. Chen, J. Catal. 229 (2005) 30–44.
References
[1] H.M. Wang, J. Male, Y. Wang, ACS Catal. 3 (2013) 1047–1070.