Chemical titration and transient kinetic studies of site requirements in Mo2C-catalyzed vapor phase anisole hydrodeoxygenation

Article abstract of DOI:10.1021/acscatal.5b00713

The turnover frequency (TOF) of benzene synthesis from vapor phase anisole hydrodeoxygenation (HDO), estimated via in situ CO titration, was found to be invariant (1.1 ± 0.3 × 10^-3 s^-1) over molybdenum carbide (Mo₂C) catalysts with varying CO chemisorption uptakes (～70 to ～260 μmol g^-1, measured ex situ at 323 K). Accumulation of oxygen (～0.29 monolayer) over Mo₂C catalysts was determined by an oxygen mass balance during the transient of anisole HDO at 423 K under ambient pressure (H₂/anisole molar ratio ～ 110). Similar product selectivity, apparent activation energy, and TOF of benzene synthesis for an oxygen treated (with oxygen incorporation: O/Mo_bulk (molar ratio) = 0.075) and freshly prepared Mo₂C catalysts (no exposure to air prior to kinetic measurements) demonstrate that the effect of oxygen at these low concentrations is solely to reduce the number of active sites for anisole HDO, resulting in a lower (～3 times) benzene synthesis rate per gram of catalyst for the oxygen-modified material. The observed benzene synthesis rates per CO chemisorption site for bulk molybdenum oxide (MoO_x) catalysts were found to be ～10 times lower than those for Mo₂C catalysts, suggesting that bulk molybdenum oxide phases are not associated with the dominant active sites for anisole HDO at 423 K under ambient pressure.

Full text of DOI:10.1021/acscatal.5b00713

Research Article
pubs.acs.org/acscatalysis
Chemical Titration and Transient Kinetic Studies of Site
Requirements in Mo2C‑Catalyzed Vapor Phase Anisole
Hydrodeoxygenation
Wen-Sheng Lee,† Anurag Kumar, Zhenshu Wang, and Aditya Bhan*
Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota 55455, United States
S
* Supporting Information
ABSTRACT: The turnover frequency (TOF) of benzene
synthesis from vapor phase anisole hydrodeoxygenation
(HDO), estimated via in situ CO titration, was found to be
invariant (1.1
0.3 × 10−3 s−1) over molybdenum carbide
(Mo2C) catalysts with varying CO chemisorption uptakes
(∼70 to ∼260 μmol g−1, measured ex situ at 323 K).
Accumulation of oxygen (∼0.29 monolayer) over Mo2C
catalysts was determined by an oxygen mass balance during the
transient of anisole HDO at 423 K under ambient pressure
(H2/anisole molar ratio ∼ 110). Similar product selectivity, apparent activation energy, and TOF of benzene synthesis for an
oxygen treated (with oxygen incorporation: O/Mobulk (molar ratio) = 0.075) and freshly prepared Mo2C catalysts (no exposure
to air prior to kinetic measurements) demonstrate that the effect of oxygen at these low concentrations is solely to reduce the
number of active sites for anisole HDO, resulting in a lower (∼3 times) benzene synthesis rate per gram of catalyst for the
oxygen-modified material. The observed benzene synthesis rates per CO chemisorption site for bulk molybdenum oxide (MoOx)
catalysts were found to be ∼10 times lower than those for Mo2C catalysts, suggesting that bulk molybdenum oxide phases are not
associated with the dominant active sites for anisole HDO at 423 K under ambient pressure.
KEYWORDS: molybdenum carbide, molybdenum oxide, hydrodeoxygenation (HDO), oxidation, anisole, benzene synthesis,
in situ CO titration, oxygen treatment, turover frequency (TOF)
adsorption sites are associated with sites responsible for the
activation of anisole. In this work, we use in situ CO titration to
assess the turnover frequency (TOF) of benzene synthesis rates
over Mo2C catalysts with different CO chemisorption uptakes
measured ex situ at 323 K (∼70 to ∼260 μmol g−1) and report an
average TOF of 1.1 0.3 × 10−3 mol s−1 molCOsite−1, implying
that only a fraction of the metallic sites probed by ex situ CO
chemisorption studies is active for HDO catalysis on Mo2C
surfaces.
The oxophilicity of molybdenum carbide is evidenced from its
pyrophoric nature4−7 and the experimental observation of a
Mo2C1.04O0.88 composition, even after treating an air-exposed
molybdenum carbide catalyst in a 3/1 (vol %) CH4/H2 mixture
at 623 K for 1 h by Bell and co-workers.6 DFT calculations by
Rodriguez and co-workers also suggest that O2 can bond strongly
to both C and Mo sites of molybdenum carbides with
corresponding adsorption energies of ∼520 and ∼786 kJ
mol−1.8 The possibility of in situ oxidation of Mo2C catalysts
by oxygen-containing reactants during HDO, therefore, cannot
be excluded. Transient kinetic measurements combined with
temperature-programmed surface reaction with H2 (TPSR)
1. INTRODUCTION
Lignin, a natural amorphous polymer consisting of phenolic
compounds that comprise 15−30% by weight of biomass and
40% by energy, can be regarded as an alternative sustainable
source for aromatic feedstock production, specifically, benzene,
toluene, and xylene (BTX).1,2 The conversion from phenolic
compounds to aromatic hydrocarbons, however, requires the
selective removal of oxygen heteroatoms and the concurrent
preservation of aromaticity. Benzene, for example, can be
produced from anisole, a model compound representing
phenolic monomers derived from lignin, via selective cleavage
of the phenolic C−O bond. Catalytic hydrodeoxygenation
(HDO) can remove the oxygen as water; however, undesired
CC bond hydrogenation or C−C bond cleavage by hydro-
genolysis can also concurrently occur. We have previously
reported that molybdenum carbide (Mo2C) catalysts can
selectively convert anisole to benzene in the vapor phase at
low reaction temperatures (423−520 K) with high benzene
selectivity >90% (C6+ basis), no C2−C5 products, and very low
cyclohexane selectivity (<9%, C6+ basis).3
Metallic sites on Mo2C catalysts were identified to be involved
in vapor phase anisole HDO based on the observed near
invariance in benzene synthesis rates normalized by CO
chemisorption sites measured ex situ at 323 K (average value:
Received: April 5, 2015
Revised: May 15, 2015
Published: June 9, 2015
3.4
1.0 × 10−4 mol s−1 molCOsite−1),3 suggesting that CO
© 2015 American Chemical Society
4104
DOI: 10.1021/acscatal.5b00713
ACS Catal. 2015, 5, 4104−4114

ACS Catalysis
Research Article
experiments of a spent Mo2C catalyst after vapor phase anisole
HDO were employed in this work to show that oxygen
accumulation on/in Mo2C catalysts occurs during HDO, even
at a high H2-to-anisole molar ratio (∼110) in the reactant feed at
423 K under ambient pressure.
Mo2C catalysts.4,7,16 After the sample was cooled to RT (typically
20−30 min), the passivated Mo2C catalyst was removed from the
reactor. Multiple batches of Mo2C synthesized using the same
preparation conditions were mixed in order to (i) create a
representative Mo2C catalyst that shares a set of physical/
chemical properties and (ii) perform in situ CO titration
experiments (section 2.4) and transient kinetic measurements
(section 2.5), which required large amounts of sample (0.43 to
8.5 g) for quantitative analysis using online mass spectrometry
(MKS Cirrus 200 Quadrupole MS system).
Boudart and co-workers9−11 reported that chemisorbed
oxygen on/in tungsten carbide catalysts can introduce acidic
sites, which increases the rate of alkane isomerization but inhibits
the rate of alkane hydrogenolysis reactions, resulting in high
alkane isomerization selectivity (70−99%). An oxycarbide phase
of molybdenum was proposed to be responsible for high
selectivity (∼85%) of n-heptane isomerization by Ledoux and co-
workers.12 Residual oxygen retained in molybdenum carbide
catalysts, achieved by using different carburization temperatures
for MoO3 catalysts, was also found to suppress benzene
hydrogenation rates.3,13,14 Rodriguez and co-workers8 have
employed density functional theory (DFT) calculations to
suggest that a Mo oxycarbide can be formed during the water gas
shift reaction on the surface of Mo2C in which chemisorbed
oxygen, produced from water dissociation, is involved. These
reports demonstrate that the presence of heteroatoms on/in
transition metal carbide catalysts can alter their reactive
functionality by introducing acidic sites or by changing the
number of active centers available for catalysis.
A Mo2C catalyst, which was not exposed to air (i.e., the
passivation step was omitted), was also prepared according to the
temperature protocol described above, except that the sample
was cooled in a flow of H2 (total flow rate of 2.3 cm3 s−1) from
873 to 423 K. Once the temperature reached 423 K, the gas
stream was switched from pure H2 to the reactant mixture
(anisole/H2 = 0.16/balance, vol %, total flow rate ∼1.67 cm3 s−1)
to perform vapor phase anisole HDO under ambient pressure.
An oxygen-treated Mo2C catalyst was prepared using a
method similar to that reported by Ribeiro et al.9,11,17 A freshly
synthesized Mo2C catalyst (from ∼1.6 g AMT), following a
temperature protocol for the typical Mo2C synthesis described
above, was cooled to RT in H2 flow (∼2.3 cm3 s−1), subsequently
treated in a flow of 1% O2/He (3.75 cm3 s−1) at RT for 2 h and
thereafter at ∼423 K for 30 min (ramping rate ∼0.07 K s−1). After
oxygen treatment, the catalyst was quenched to RT in a flow of
H2/Ar (87.5/12.5, vol %, total flow rate ∼2.67 cm3 s−1), followed
by heating at ∼723 K for 1 h with a ramping rate ∼0.12 K s−1
from RT (Figure S1, Table S1 in the Supporting Information
(SI)), and then cooled in the same H2/Ar mixture flow to 423 K
prior to reaction rate/apparent activation energy measurements
for vapor phase anisole HDO. An independent experiment was
performed in which the oxygen-treated Mo2C catalyst, after
being treated in a flow of H2/Ar (87.5/12.5, vol %, total flow rate
∼2.67 cm3 s−1) at 723 K for 1 h and cooled to RT, was subjected
to a second temperature-programmed surface reaction with H2/
Ar (87.5/12.5, vol %, total flow rate ∼2.67 cm3 s−1) at 773 K for
∼5 h with a ramping rate 0.19 K s−1 (Figure S2, Table S1 in the
SI) to assess the amount of oxygen incorporated and retained on
or in the molybdenum carbide catalyst before vapor phase anisole
HDO. The reactor effluent during catalyst treatments was
monitored and quantified by using an online mass spectrometer
with calibrated response factors.
In this study, we investigate the effect of oxygen incorporation
on/in molybdenum carbide catalysts on the kinetics of vapor
phase anisole HDO. An oxygen-treated molybdenum carbide
catalyst with oxygen uptake O/Mobulk (molar ratio) = 0.075
showed an ∼3 times lower benzene synthesis rate per gram of
catalyst (2.4 × 10−8 vs 6.8 × 10−8 mol s−1 g−1) as compared with a
freshly prepared molybdenum carbide catalyst in which the
catalyst was not exposed to oxygen before reaction. A similar
TOF of benzene synthesis, obtained by in situ CO titration, was
found for both the oxygen-treated and freshly prepared
molybdenum carbide formulations, suggesting that the effect of
oxygen is to solely reduce the number of active sites for vapor
phase anisole HDO.
2. EXPERIMENTAL METHODS
2.1. Catalyst Synthesis. Mo2C catalysts were prepared by a
temperature-programmed reduction method similar to those
reported previously.3,7,15 Briefly, ∼1.6 g of ammonium
molybdate tetrahydrate (AMT, (NH4)6Mo7O24·4H2O sieved,
177−400 μm, Sigma, 99.98%) was loaded into a tubular quartz
reactor with a thermocouple attached to a thermowell (i.d. 10
mm) and purged with a flow of a CH4 (Matheson, 99.97%)/H2
(Minneapolis Oxygen, 99.999%) gas mixture (15/85, vol %, total
flow rate ∼2.75 cm3 s−1) for 1−5 min at room temperature
(hereafter denoted as RT) and then heated in a three-zone split
tube furnace (series 3210, Applied Test System) with a typical
temperature protocol (15% CH4/H2, RT (1.5 h) to 623 K [5 h]
(1.5 h) to 873 K [3 h], cool) that is described in detail below. The
reactor was heated from RT to the first target temperature (∼623
K) within 1.5 h, and the temperature was then held for 5 h.
Subsequently, the sample was heated to the second target
temperature (∼873 K) within 1.5 h and held at this temperature
for 3 h prior to being cooled to RT in the same CH4/H2 gas
mixture flow. The resulting material (Mo2C) was then treated in
a flow of 1% O2/He mixture (Matheson, Certified Standard
Purity) at ∼1.67−3.33 cm3 s−1 to passivate the carbidic
surface.4,7,15 The reactor temperature was found to increase
from RT to ∼333−373 K upon the introduction of a 1% O2/He
gas mixture, confirming the pyrophoric feature of as-synthesized
MoO3 (Sigma, 99.5+ % ACS Reagent) catalysts were used as
received and sieved (177−400 μm) prior to kinetic measure-
ments of vapor phase anisole HDO.
2.2. Materials Characterization. The bulk structures of the
samples were characterized by X-ray diffraction with three
measurement frames at 30° (2θ), 60° (2θ), and 90° (2θ) at a 600
s frame−1 dwell (Δ2θ frame width of 35° (2θ)) (XRD, Bruker D8
Discover 2D X-ray diffractometer with a two-dimensional
VÅNTEC-500 detector, Cu Kα X-ray radiation with a graphite
monochromator, and a 0.5 mm point collimator). Two-
dimensional images were then converted to one-dimensional
intensity vs 2θ for analysis. CO chemisorption experiments were
performed using a Micromeritics ASAP 2020 analyzer. An
appropriate amount of sample (0.1−0.5 g) was loaded into a
chemisorption cell, followed by evacuation at 383 K (∼2 μm Hg)
for 0.5 h, H2 treatment at 723 K for 2 h, and degassing (∼2 μm
Hg) at 723 K for 2 h, and then cooled to 323 K. Subsequently, the
first CO adsorption isotherm (total adsorbed species, between
100−450 mmHg) was measured at 323 K. The cell was then
degassed (∼2 μm Hg) to remove weakly adsorbed species prior
4105
DOI: 10.1021/acscatal.5b00713
ACS Catal. 2015, 5, 4104−4114

ACS Catalysis
Research Article
to the second isotherm. Both isotherms were extrapolated to zero
pressure, and the amount of strongly adsorbed CO was obtained
by the difference between the two extrapolated values.4,18 Unless
otherwise mentioned, the BET surface area measurements
(Micromeritics ASAP 2020) were carried out immediately after
the chemisorption measurements. The samples, after chem-
isorption, were passivated by flowing 1% O2/He gas mixture for
at least 0.5 h prior to being removed from the chemisorption cell
and subsequently loaded in physisorption cells for N2 adsorption
experiments without any degassing pretreatments.
2.3. Steady State Kinetic Measurements of Vapor
Phase Anisole HDO. A passivated Mo2C sample (∼0.25 g with
CO chemisorption uptake ∼79 μmol g−1) was placed in a tubular
quartz reactor (i.d. 10 mm) and pretreated in a pure H2 flow
(∼1.67 cm3 s−1) at ∼750 K for 1 h with a ramping rate ∼0.125 K
s−1 and cooled to reaction temperature ∼423 K before vapor
phase anisole HDO at ambient pressure.
to calculate the average residence time of CO over the catalyst
bed, which was used to determine the amount of CO adsorbed
and/or reacted over the surface of the Mo2C during titration. The
turnover frequency (TOF) of benzene synthesis, by assuming
that one CO molecule titrates one active site, was therefore
obtained (see section 3.1 for a detailed description). Replicates of
CO titration experiments with the same or different CO flow
rates were conducted to determine the error associated with the
TOF calculation in which the reported error for TOF is a
standard deviation calculated from the multiple measurements.
2.5. Transient Measurements of Vapor Phase Anisole
HDO. Approximately 2.1 g of the Mo2C catalyst with CO
chemisorption uptake ∼199 μmol g−1 was loaded into a reactor
and pretreated in pure H2 (∼1.67 cm3 s−1) at ∼723 K for 1 h
(∼0.12 K s−1), followed by cooling the sample in the H2 flow to
∼423 K prior to the introduction of a reaction mixture consisting
of ∼0.75/16.7/bal (vol %) of anisole/Ar/H2 (total flow rate ∼2
cm3 s−1). Ar, cofed in the reaction feed, served as an internal
standard for quantification of the reactor effluents during
transient measurements using a mass spectrometer (MKS Cirrus
200 Quadrupole MS system). The reaction reached steady state
after ∼6 ks time-on-stream (Figure S3 in the SI). A mass balance
of oxygen-containing species (anisole, methanol, water, CO and
CO2) from 0 to 6 ks or carbon species from 0 to 8 ks was used to
assess the amount of oxygen or carbon accumulated on/in the
Mo2C catalyst during this transient. All transient kinetic
measurements were conducted at anisole chemical conversions
<20%.
2.6. Temperature-Programmed Surface Reaction with
H2 (TPSR) Study of Mo2C Catalysts after Anisole HDO. A
temperature-programmed surface reaction with H2 (TPSR) was
performed on the Mo2C sample used for transient kinetic
analysis (section 2.5) after ∼43 ks in vapor phase anisole HDO at
423 K under ambient pressure (Figure S3 in the SI). The feed
was first switched from the reaction mixture consisting of 0.75/
16.7/bal (vol %) of anisole/Ar/H2 (total flow rate ∼2 cm3 s−1) to
pure He (total flow rate ∼5 cm3 s−1) for 0.72 ks to remove weakly
bound surface species at ∼423 K; subsequently, the feed was
switched to a H2/Ar mixture (83.3/16.7, vol %, ∼2 cm3 s−1) for
0.3 ks, followed by heating the sample from 423 to 730 K at a
ramping rate of ∼0.17 K s−1. The reactor effluents were
monitored and quantified (using Ar as an internal standard)
using a mass spectrometer.
MoO3 samples (∼0.42 g of MoO3 per batch) were pretreated
in a flow of pure H2 (∼1.67 cm3 s−1) at ∼675 K for 1−6 h with a
ramping rate ∼0.1 K s−1, instead of 750 K, prior to vapor phase
anisole HDO at 423 K under ambient pressure.
Vapor phase anisole HDO reactions were carried out by
feeding a reaction mixture (∼1.67 cm3 s−1) consisting of ∼0.15−
0.75/bal (vol %) of anisole/H2. Anisole (Sigma, 99%,
ReagentPlus) was added using a syringe pump (KD Scientific,
model 100). All flow lines were heated to at least 373 K via
resistive heating to prevent condensation of anisole and reactor
effluents. The products for anisole HDO were quantified using a
flame ionization detector (FID) of a gas chromatograph (Agilent
6890 equipped with a methyl-siloxane capillary column (HP-1,
50 m × 320 μm × 0.52 μm)) with calibrated response factors.
The apparent activation energy was measured after the rates had
reached steady state by varying the reaction temperature between
408 to 438 K. The carbon balance was typically better than 90%.
The anisole conversion and product selectivity were calculated as
follows:
anisole conversion
(sum of moles of carbon in products)out
=
(moles of carbon in anisole)in
C6+ products selectivity (C+6 basis)
moles of C+6 producti
∑ moles of C+6 products
=
3. RESULTS AND DISCUSSION
3.1. In Situ CO Titration of Mo2C Catalysts for Vapor
Phase Anisole HDO. Molybdenum carbide catalysts were first
allowed to reach steady state for vapor phase anisole HDO prior
to introducing a CO/Ar cofeed (section 2.4). Benzene synthesis
rates were found to be significantly suppressed in the presence of
CO, but they could be fully restored upon the removal of CO,
suggesting that CO is a reversible titrant, as shown in Figure 1.
Multiple CO titration experiments with different CO flow rates
(∼0.005 to ∼0.028 cm3 s−1) were then performed to obtain the
TOF of benzene synthesis per CO chemisorption site. Figure 2
shows the mass spectrometer signals of benzene, CO, and Ar as a
function of time during a typical CO titration experiment. Since a
synchronic breakthrough of both Ar and CO mass spectrometer
signals was observed in a blank experiment in which CO (0.005−
0.025 cm3 s−1)/Ar (0.033 cm3 s−1) gas mixtures were cofed in an
empty reactor with pure H2 gas flow (1.67 cm3 s−1) (Figure S4 in
the SI), the apparent residence time of CO in the reactor (the
delayed breakthrough time for CO as compared with that of Ar)
C1 products selectivity (C1 basis)
moles of methane or methanol
moles of methane + moles of methanol
=
2.4. In Situ CO Titration of Mo2C Catalysts for Vapor
Phase Anisole HDO. In situ CO titration was carried out after
steady state benzene synthesis rates were observed in which
0.43−8.5 g of Mo2C (with equivalent 180−560 μmol of CO
chemisorption sites in the reactor) was employed. A gas mixture
of CO (∼0.005 to ∼0.053 cm3 s−1)/Ar (∼0.03 cm3 s−1) was
introduced into the reactant stream (anisole/H2 (vol %) = ∼0.36
to ∼0.75/bal) with a total flow rate ∼1.67 cm3 s−1, which
corresponds to CO partial pressures ranging from ∼0.3 to ∼3.1
kPa. An online mass spectrometer (MKS Cirrus 200 Quadrupole
mass spectrometer system) was used to track the resulting
transient response of Ar, CO, and the synthesis rates of benzene
during each titration experiment. The Ar flow was used as a tracer
4106
DOI: 10.1021/acscatal.5b00713
ACS Catal. 2015, 5, 4104−4114

ACS Catalysis
Research Article
contributes negligibly to the observed CO residence time,
which will be discussed later. Alternatively, the average residence
time of CO, τ, can be estimated by calculating the area between
the Ar and CO curves as shown in Figure 2 in which both Ar and
CO signals were normalized to their corresponding values
obtained at steady state.19 The number of CO molecules
adsorbed on the molybdenum carbide catalyst could be obtained
by multiplying the average CO residence time and the known
CO molar flow rate. The residence times of CO obtained by the
two different methods outlined above were found to be in
agreement (with error typically <15%), as shown in Table 1. The
TOF of benzene synthesis per metallic site on molybdenum
carbide catalysts (with the assumption that one CO molecule
poisons one active site for benzene synthesis) can therefore be
rigorously calculated on the basis of the ratio of the reduction of
benzene synthesis rate due to the presence of CO and the
amount of CO adsorbed on the catalyst. The resulting values
were found to be 1.1 0.3 × 10−3 mol s−1 molCOsite−1 (Table 1).
Nearly invariant benzene TOF values with different CO flow
rates ranging from 0.005 to 0.053 cm3 s−1 were also obtained, as
shown in Figure 3 (the detailed data for the three sets of titration
experiments used to plot Figure 3 can be found in Table 1 and
Tables S2, S3 in the SI), suggesting (i) the absence of CO mass
transfer limitations during the in situ titration, (ii) a negligible
effect of CO hydrogenation on the TOF calculation, and (iii)
little or no diversity in the active sites titrated by CO under
conditions employed in this work. The number of operating
active sites obtained via (benzene synthesis rates in the absence
of CO)/(measured TOF) was found to be 34 7 μmol g−1
(average value shown in Table 1 with catalyst loading 1.11 g),
which corresponds to ∼13% of the total potential active sites
measured by ex situ CO chemisorption (∼240 μmol g−1).
Since molybdenum carbide catalysts have been shown to be
active for CO hydrogenation and water gas shift,6−8,20 the
observed average CO residence time (τ in Figure 2) can be
contributed from both CO competitive adsorption with anisole
on active sites for HDO and CO adsorption/reaction on sites
active for CO hydrogenation or water gas shift. No detectable
amount of CO2 was observed (TCD, with He as the GC carrier
gas) during in situ CO titration, suggesting that the effect of the
water gas shift on the estimation of CO adsorption sites can be
neglected.
The molar ratio between C1 and C6 species during vapor phase
anisole HDO (1 atm and 423 K) was found to be stoichiometric
(C1/C6 ∼1), as shown in our previous report.3 The observed
excess in stoichiometric values of C1 formation rates (i.e., relative
to benzene synthesis rates) must therefore come from CO
hydrogenation (Table 1). Shou and Davis20 employed steady-
state isotopic transient kinetic analysis to assess the number of
active intermediates (i.e., operating active sites) over supported
Mo2C catalysts and concluded that <1% of the surface Mo sites
were involved in CO hydrogenation reactions at 573 K under 1.2
bar with a CO/H2 molar ratio = 1. We note that although the
coverage of active intermediates for CO hydrogenation at our
reaction temperature (423 K) is expected to be higher, the much
lower CO partial pressure (<2 kPa) employed during in situ CO
titration experiments would have a counter effect on the
coverage. The contribution of CO adsorption associated with
hydrogenation to the observed average CO residence time (τ) is,
therefore, assumed to be negligible.
Figure 1. Benzene synthesis rates vs time-on-stream during the course
of in situ CO titration. The catalyst (∼1.11 g, CO chemisorption uptake
∼240 μmol g−1) was stabilized for ∼10 ks before in situ CO titration.
Shaded areas indicate CO/Ar mixtures (∼0.005 to ∼0.028 cm3 s−1/
∼0.036 cm3 s−1) were cofed into the reactant feed comprising anisole/
H2 (vol %) = ∼0.36/bal with a total flow rate ∼1.67 cm3 s−1 at reaction
temperature ∼423 K under ambient pressure. The dashed line is shown
as a guide to the eye.
Figure 2. Normalized Ar (m/z = 40), CO (m/z =28) and benzene mass
spectrometer signals (m/z = 78, corrected by subtracting the signal
contributed from anisole) during the course of in situ CO titration. τB, τ,
and τE, represent the breakthrough time of CO, average residence time
of CO determined by integration of the normalized Ar and CO
breakthrough curves (shaded area) and extrapolated residence time of
CO, which was used to calculate the amount of CO required to shut
down benzene synthesis (see section 3.1). CO/Ar was cofed as a titrant/
inert tracer with a flow rate of ∼0.0126 cm3 s−1/∼0.036 cm3 s−1 into a
reactant feed comprising anisole/H2 (vol %) = ∼0.36/bal (total flow
rate ∼1.67 cm3 s−1) at a reaction temperature ∼423 K under ambient
pressure; Mo2C loading ∼1.11 g (CO chemisorption 240 μmol g−1).
Benzene synthesis rates in the presence/absence of CO were measured
by GC before 0 s and after 400 s.
was attributed to adsorption or reactions (CO hydrogenation
and/or water gas shift) over the catalysts.
The number of active sites occupied by CO can be estimated
by multiplying the known CO molar flow rate and the CO
residence time (τB, the difference of the breakthrough times
between the CO and Ar signals) if we consider that CO
hydrogenation or CO consumed via the water gas shift
The amount of CO required to completely deactivate vapor
phase anisole HDO rates (i.e., the number of operating sites) can
also be obtained by multiplying the known CO flow rate and the
4107
DOI: 10.1021/acscatal.5b00713
ACS Catal. 2015, 5, 4104−4114

ACS Catalysis
Research Article
Table 1. Experimental Conditions and Measured Kinetics for in Situ CO Titration of Mo2C Catalysts with Catalyst Loading ∼1.11
a
g in Vapor Phase Anisole HDO shown in Figures 1, 3, and 7
benzene
TOF based on
extrapola−ti1on
rate (×
methane
methanol
CO mol flow
avg CO
apparent amt
CO adsorbed
TOF (10−3
no. of operating sites
−1
exptl
set
10−8 mol rate (× 10−8 rate (× 10−8 rate (× 10−6 residence
mol s
(mol s−1
during anisole HDO
b
b
b
c
d
e
f
g
−1
cond
s−1
)
mole s−1
)
mole s−1
)
mol s−1
)
time τ (s) (× 10−5 mol)
molCOsite
)
molCOsite
)
reaction (μmol)
1
2
3
4
5
6
7
8
9
no CO
4.3
1.8
0.8
1.3
3.8
4.2
with
CO
0.3
68 (54)
54 (48)
27 (23)
16 (17)
35 (37)
20 (20)
31 (20)
45 (43)
19 (23)
2.1
1.7
2.8
3.4
1.8
3.1
2.2
1.8
3.3
1.2
1.3
32
26
37
43
25
40
31
28
44
no CO
4.4
1.9
0.8
1.3
3.9
4.3
with
CO
0.3
1.0
2.1
0.5
1.5
0.7
0.4
1.7
1.5
1.1
1.0
1.6
1.0
1.3
1.4
0.9
1.4
0.7
0.4
1.1
0.5
0.8
1.4
0.5
no CO
4.4
1.3
0.8
1.8
3.5
4.7
with
CO
no CO
4.4
1.1
0.9
2.2
3.6
4.6
with
CO
no CO
4.4
1.5
0.9
1.6
3.8
4.1
with
CO
no CO
4.4
1.1
0.9
2.0
3.4
4.6
with
CO
no CO
4.3
1.4
0.9
1.7
3.5
4.1
with
CO
no CO
4.3
1.7
0.9
1.5
3.6
4.0
with
CO
no CO
4.2
1.1
0.9
2.0
3.5
4.1
with
CO
a
b
CO chemisorption uptake ∼ 240 μmol g−1. Reaction conditions: anisole/H2 (vol %) = ∼0.36%/bal with total flow rate ∼1.67 cm3 s−1 at reaction
temperature ∼423 K, anisole conversion and benzene selectivities (steady state values in the absence of CO) during the course of titration were
c
∼19% and ∼95% respectively. Obtained by integrating the area between normalized Ar and CO signals as shown in Figure 2; the number in the
d
parentheses represents the CO breakthrough time in seconds (τB as shown in Figure 2). Obtained via multiplying CO molar flow rate by average
e
CO residence time. Obtained by ((benzene rates without CO) − (benzene rates with CO))/(apparent amount of CO adsorbed on the catalysts
during CO titration). Obtained by extrapolation method; see section 3.1 for detailed information. Obtained by (benzene rates without CO)/(the
corresponding measured TOF).
f
g
CO residence time obtained by linearly extrapolating the initial
slope of the benzene synthesis rates to zero21 by assuming that (i)
CO adsorption is irreversible and far away from equilibrium at
low CO coverage and (ii) all active sites behave the same in the
linear regime, as demonstrated in Figure 2. The TOF can
therefore be calculated on the basis of the ratio of observed
benzene synthesis rates in the absence of CO and the measured
amount of operating sites determined via the extrapolation
method. No significant difference was found in the TOF values
obtained from the two different methods as shown in Table 1,
which suggests that the lag of the CO breakthrough curve relative
to Ar (Figure 2) was dominated by CO competitive adsorption
with anisole on the active sites for HDO, not on the sites for CO
hydrogenation.
3.2. Transient Kinetic Measurements of Vapor Phase
Anisole HDO over Molybdenum Carbide Catalysts. An
oxygen mass balance during the transient of anisole HDO and a
temperature-programmed surface reaction with H2 of a sample
after being tested for ∼43 ks for anisole HDO (section 3.3) were
performed to examine the extent of in situ oxidation of
molybdenum carbide catalysts by anisole during HDO at 423
K under ambient pressure with a high H2 to anisole ratio (∼110)
in the reactant feed. Figure 4a shows mass spectrometer signals of
anisole and the corresponding products as a function of time
during the transient (<6 ks) in which time zero indicates the
Figure 3. Turnover frequency (TOF) of benzene synthesis determined
by in situ CO titration as a function of CO space time, defined as
(amount of catalyst)/(CO cofeed flow rate employed in the
corresponding titration experiments). Different CO space times were
achieved using both different amounts of catalyst (∼1.77, ∼1.11, ∼0.43
g) and different CO cofed flow rates (∼0.0076−0.053 cm3 s−1). Data
shown in Tables 1, S2, and S3 were used to plot this figure. Reaction
conditions: feed composition, anisole/H2 (vol %) = 0.36−0.75/bal;
reaction temperature, 423 K; ambient pressure; anisole conversion ∼10
to ∼19% with benzene selectivity ∼95%.
4108
DOI: 10.1021/acscatal.5b00713
ACS Catal. 2015, 5, 4104−4114

ACS Catalysis
Research Article
hexane in independent studies.3,13,14 The observation of
cyclohexane prior to benzene formation shown in Figure 4a
can therefore be explained via sequential hydrogenation of
benzene to cyclohexane by the catalyst that is not yet oxidized.
Figure 4b shows the cumulative oxygen accumulation as a
function of time, calculated by performing an oxygen mass
balance (considering anisole, water, and methanol as the only
oxygen-containing species) every 500 s (between 0 and 3.5 ks).
The total oxygen retained in the molybdenum carbide catalyst,
after ∼3.5 ks as shown in Figure 4b, was found to be ∼982 μmol
(∼468 μmol g−1), which corresponds to ∼0.29 monolayer of
oxygen, on the basis of the measured BET surface area of the
sample ∼96 m2 g−1 (BET surface area was obtained from an
independent experiment before vapor phase anisole HDO) and a
Mo site density of 1 × 1015 sites cm−2 (Table 2).14 The observed
in situ oxidation of molybdenum carbide during HDO is similar
to that reported by Prasomsri et al.22 in which the authors
concluded that the surface of MoO3 can be partially carburized
during HDO to generate oxycarbide or oxycarbohydride phases,
as evidenced by XPS (observation of Mo5+ species) and XRD
(observation of oxycarbohydride phase).
Figure 4c shows the calculated carbon mass balance as a
function of time during the transient, for which only anisole,
benzene, cyclohexane, methane, and methanol were considered.
The time required for the carbon balance to close (∼8 ks) was
found to be longer than that for the oxygen balance (∼4 ks).
Carbon accumulation between 0 and 8 ks time-on-stream was
found to be ∼8100 μmol g−1 (equivalent to ∼1350 μmol g−1 of
C6 species), which corresponds to C1 and C6 species coverage of
∼510% and ∼85%, respectively, on the basis of a BET surface
area of 96 m2 g−1 and a Mo site density of 1 × 1015 sites cm−2
(Table 2).14 The carbon mass balance suggests that carbon
deposition also occurred in the transient (before ∼8 ks) prior to
steady state rates for vapor phase anisole HDO being attained; C1
species are inferred to be preferentially deposited on the surface
of molybdenum carbide catalysts because the calculated C1/C6
molar ratio was significantly less than 1 during this time period, as
shown in Figure S5.
3.3. Temperature-Programmed Surface Reaction with
H2 of Spent Molybdenum Carbide Catalysts. A temper-
ature-programmed surface reaction with H2 of the molybdenum
carbide catalyst tested for anisole HDO for 43 ks (the same
sample used for the transient kinetic study discussed in section
3.2 and shown in Figure S3 in the SI) was directly performed
(without exposure to air) to determine the amount of oxygen
accumulated during the transient of the vapor phase anisole
HDO; the data are shown in Figure 5. Benzene (mass
spectrometer signal, m/z = 78) was observed upon switching
the gas flow from pure He to the H2/Ar gas mixture. This
observation supports the reaction mechanism proposed in our
previous work wherein anisole decomposes to form C6H5
intermediates that can desorb from the surface only by reaction
with H* species derived from molecular H2 and in the absence of
H2 persist on the surface.3 After purging the catalyst in a flow of
H2/Ar for ∼0.3 ks, the temperature was ramped to 730 K at a
ramping rate 0.17 K s−1, and both oxygen-containing species
(water, methanol, CO and CO2) and carbon species (C1: CO,
CO2, methane, and methanol; C6: benzene, cyclohexane, and
anisole) were observed during the temperature ramp. Benzene
evolution occurred at earlier times-on-stream (∼1.2 ks) than
oxygen-containing species (except for methanol), which suggests
that most of the oxygen in the molybdenum carbide catalyst
could not be removed at ∼420 K, and consequently, the
Figure 4. (a) Transient mass spectrometer (MS) signals of benzene
(C6H6, m/z = 78, corrected by subtracting the signal contributed from
anisole), anisole (C7H8O, m/z = 108), cyclohexane (C6H12, m/z = 84),
methane (CH4, m/z = 16), methanol (CH3OH, m/z = 31), and water
(H2O, m/z = 18) for 2.1 g of Mo2C (BET surface area ∼ 96 m2 g−1, CO
chemisorption ∼ 199 μmol g−1) tested in the vapor phase anisole HDO
reaction at ∼423 K under ambient pressure with reaction feed
composition: anisole/Ar/H2 (vol %) = 0.75/16.7/bal (total flow rate
∼ 2 cm3 s−1); (b) cumulative oxygen accumulation in the Mo2C catalyst
obtained by calculating an oxygen mole balance (considering only H2O,
CH3OH, C7H8O) every 0.5 ks from 0 to 3.5 ks, during the transient of
anisole HDO shown in part (a); (c) carbon mole balance during the
transient of anisole HDO shown in part (a). Each data point represents
the carbon mole balance starting from 0 ks time-on-stream. Carbon
mole balance was defined as (mole of carbon measured at the outlet)/
(mole of anisole fed in × 7).
switch of the reactant feed from bypass to the reactor. Upon the
introduction of the reaction feed, methane was observed
immediately, followed by cyclohexane at time-on-stream ∼0.5
ks. No oxygen-containing products (water and methanol) were
observed before ∼1.5 ks, suggesting that oxygen accumulation
(in situ oxidation of the molybdenum carbide catalyst) occurred
prior to steady state rates of anisole HDO being attained.
Benzene was found to increase at the expense of cyclohexane,
with a concurrent increase in oxygen-containing products, water
and methanol, at time-on-stream ∼1.2 ks. The presence of
oxygen in molybdenum carbide catalysts was previously shown
to significantly suppress hydrogenation of benzene to cyclo-
4109
DOI: 10.1021/acscatal.5b00713
ACS Catal. 2015, 5, 4104−4114

ACS Catalysis
Research Article
Table 2. Amount of Oxygen and Carbon Species Retained on/in the Mo2C Catalyst during the Transient of Vapor Phase Anisole
HDO as Shown in Figure 4
a
b
cat. amt BET surface area
CO chemisorption uptake (323
oxygen
equiv C6
d
oxygen coverage obtained by C6 coverage obtained by
c
e
f
(g)
(m2 g−1
96
)
K, μmol g−1
)
coverage (%)
coverage (%)
TPSR (%)
TPSR (%)
2.1
199
29
85
25
14 (110)
a
b
No H2 pretreatment was performed prior to the BET measurement. Average value of two independent CO chemisorption measurements: 204 and
c
194 μmol g−1. Oxygen accumulation (water, methanol, and anisole were considered) during the transient of vapor phase anisole HDO shown in
Figure 4a. Oxygen accumulation between 0 and 3.5 ks time-on-stream (Figure 4b) was found to be ∼468 μmol g−1. Oxygen coverage was calculated
d
based on a site density of ∼1 × 1015 sites cm−2 and the measured BET surface area, ∼96 m2 g−1. Carbon accumulation (only methane, methanol,
benzene, cyclohexane, and anisole were considered) between 0 and 8 ks time-on-stream (Figure 4c) was found to be ∼8100 μmol g−1, which is
equivalent to ∼1350 μmol of C6 per gram of catalyst. C6 coverage was calculated on the basis of the assumption described in footnote c, above.
e
Amount of oxygen-containing species (water, methanol, anisole, CO, and CO2 were considered) measured during the TPSR experiment shown in
f
Figure 5 was found to be ∼402 μmol g−1. Oxygen coverage was calculated on the basis of the assumption described in footnote c, above. Amount of
carbon (methane, methanol, CO,CO2, benzene, cyclohexane and anisole were considered) measured during the TPSR experiment shown in Figure 5
was found to be ∼1655 μmol g−1. The amount of C6 species, including benzene, cyclohexane, and anisole, was found to be ∼225 μmol g−1, which is
used to calculate the surface C6 species coverage on the basis of the assumption described in footnote c, above. The number in parentheses
represents the C6 coverage on the basis of CO chemisorption uptake measured at 323 K shown in this table.
suggesting that some residual carbonaceous species were
retained on the catalyst and could not be removed upon the
thermal treatment in H2 to ∼730 K. This mild coke formation, in
addition to in situ oxidation of molybdenum carbide catalysts by
anisole, could cause the initial deactivation of benzene synthesis
rates (<3 ks shown in Figure S3 in the SI), Three different types
of carbonaceous speciesnear surface carbide, polymeric, and
graphitic carboninherited from different carburization con-
ditions during molybdenum carbide synthesis, have been
identified and found to be removed as CH4 at temperatures
∼763, ∼963 K, and ∼1083 K, respectively, during temperature-
programmed surface reaction with H2 (TPSR) by Oyama and co-
workers,23 which suggests the difficulty of removing carbona-
ceous species. C6 species (only benzene, cyclohexane, and anisole
were considered) eluted from the reactor during the TPSR
experiment (Figure 5) were found to be ∼225 μmol g−1, which
corresponds to ∼110% coverage on the basis of the total number
of potential active sites as assessed by ex situ CO chemisorption
at 323 K (199 μmol g−1), implying that C6 species are potentially
the dominant surface species, consistent with the observed zero-
order dependence in anisole pressure of HDO rates reported
previously.3
Figure 5. Temperature-programmed surface reaction with H2 (TPSR)
of the Mo2C catalyst after being tested in vapor phase anisole HDO for
∼43 ks, as shown in Figure S3. The catalyst was purged in a flow of He
(∼5 cm3 s−1) at ∼423 K for ∼720 s, followed by switching the gas from
He to H2/Ar (= 83.3/16.7, vol %, total flow rate ∼2 cm3 s−1) gas mixture
at ∼423 K for ∼300 s; subsequently, the temperature was ramped from
∼423 to ∼730 K with a ramp rate ∼0.17 K s−1. The MS signal of m/z =
78 was directly assigned to benzene because the contribution from
anisole was negligible.
3.4. Temperature-Programmed Surface Reaction with
H2 of Passivated Mo2C Catalysts. Figure 6 shows the
sequential hydrogenation of benzene to cyclohexane was
significantly inhibited before ∼1.2 ks (∼420 K). Once oxygen-
removal in the molybdenum carbide catalyst started (∼1.5 ks,
corresponding to ∼460 K in Figure 5), a notable increase in
cyclohexane formation was observed concurrently, as expected.
Quantification of the amount of oxygen retained in the
molybdenum carbide (only water, methanol, anisole, CO, and
CO2 were considered) during the TPSR experiment gave ∼0.25
monolayer of oxygen based on a BET surface area of 96 m2 g−1
and a Mo site density of 1 × 1015 sites cm−2 (Table 2)14, which (i)
is in agreement with the value obtained from the oxygen balance
during the transient of anisole HDO (∼0.29 monolayer) shown
in Figure 4 and confirms in situ oxidation of molybdenum
carbide during vapor phase anisole HDO and (ii) suggests
negligible further oxidation of the molybdenum carbide catalyst
occurred during steady state vapor phase anisole HDO.
Figure 6. Temperature-programmed surface reaction with H2 (TPSR)
of the passivated Mo2C catalyst with CO chemisorption uptake ∼79
μmol g−1, in which two consecutive segments (shaded region (RT to
723 K) and unshaded regions (723 to 1056 K), heating rate ∼0.16 K s−1)
were employed. Catalyst loading ∼0.25 g; H2 flow = 1.67 cm3 s−1.
The amount of carbon evolved during the TPSR experiment
was found to be ∼1655 μmol g−1 (Table 2), which is significantly
lower than that obtained from the carbon balance during the
transient of vapor phase anisole HDO (∼8100 μmol g−1),
4110
DOI: 10.1021/acscatal.5b00713
ACS Catal. 2015, 5, 4104−4114

ACS Catalysis
Research Article
Table 3. Quantification of Water Evolved during Temperature Programmed Surface Reaction with H2 of the Passivated Mo2C
Catalyst Shown in Figure 6 and the Corresponding Benzene Synthesis Rates/Selectivity for Vapor Phase Anisole HDO at 423 K at
1 atm
cat.
formulation
after 1st
segment of
TPSR
H2O removed in H2O removed in
CO
oxygen coverage
benzene
1st segment of
2nd segment of BET surf chemisorption after 1st segment
synthesis rate
TPSR
TPSR
area
uptake
of TPSR
(monolayer)
(× 10−8 mol
benzene
h
a
b
c
d
e
f
g
h
temp protocol
(μmol g−1
)
(μmol g−1
)
(m2 g−1
)
(μmol g−1
)
s
−1 g−1
3.2
)
sel. (%)
90
RT (45 min)−723 K (1 h)−1056 K
5988
1416
64
79
1.3
Mo2CO0.3
(40 min) (0.5 h)
(4.0 × 10−4
)
a
b
See section 2.1 for the nomenclature of the temperature protocols. Catalyst loading: ∼0.25 g. Amount of water removed during a typical catalyst
c
activation process employed in this work (the first segment of the temperature protocol, indicated as the shaded region shown in Figure 6). Amount
of water removed during the second segment of the TPSR experiment, as shown in Figure 6. Obtained by using the sample right after CO
chemisorption measurement; the sample was passivated by 1% O2/He prior to BET measurement. Obtained by an independent measurement in
d
e
f
which the sample was pretreated in H2 at 723 K for 2 h prior to the CO chemisorption measurement at 323 K. CO and CO2 shown in Figure 6 were
not included for calculating the oxygen coverage. BET surface area (64 m2 g−1) and a Mo site density (∼1 × 1015 sites cm−2) were used to calculate
g
the oxygen coverage. CO and CO2 shown in Figure 6 were not included for the catalyst formulation calculation. Mo2COy was used to estimate the
stoichiometry of the catalyst formulation by assuming that all of the oxygen was removed after the second segment of the TPSR experiment (see SI
h
for detailed information). Reaction was carried out at ∼423 K under ambient pressure with reaction feed comprising anisole/H2 (vol %) = 0.16/bal
(total flow rate ∼1.67 cm3 s−1). The number in parentheses represents benzene synthesis rates per CO chemisorption site. Data reported as an
average value of 1−2 h time-on-stream, and the corresponding benzene synthesis rate/benzene selectivity vs time-on-stream is shown in Figure S6.
temperature-programmed surface reaction with H2 (TPSR) of a
representative passivated molybdenum carbide catalyst in which
the benzene synthesis rate per CO chemisorption site (by ex situ
experimental result suggests that oxycarbidic sites exist during
vapor phase anisole HDO because (i) more than one monolayer
of oxygen is retained in the molybdenum carbide catalyst after H2
treatment at ∼723 K for 1 h prior to reaction, and (ii) the catalyst
can be further oxidized by the reactant at 423 K, even at a high
H2-to-anisole molar ratio (∼110) under ambient pressure.
3.5. Effect of Oxygen Treatment of Molybdenum
Carbide Catalysts on Vapor Phase Anisole HDO. Iglesia
et al.8,25 have shown that oxygen treatment can significantly
increase the isomerization selectivity of n-hexane (from 0 to
99%) at 520 K, which was attributed to the formation of acidic
sites, generated via oxygen incorporation, in tungsten carbide
catalysts. Oxygen treatment has also been shown to suppress
hydrogenolysis and promote the isomerization of n-hexane over
molybdenum carbide catalysts.5 Molybdenum carbide catalysts
prepared at higher carburization temperatures were found to
have a higher density of CO chemisorption sites and a higher
turnover frequency per CO chemisorption site for benzene
hydrogenation,13 indicating that the presence of oxygen in
molybdenum carbide catalysts can simultaneously affect both the
number and reactivity of the active sites (i.e., different TOF) for a
given reaction.
Freshly prepared molybdenum carbide, without exposure to
air, was subjected to oxygen treatment (section 2.1) to further
investigate the effects of oxygen incorporation in molybdenum
carbide on the kinetics of vapor phase anisole HDO. The results
presented in Table 4 show a ∼3 fold reduction in benzene
synthesis rates per gram of catalyst for an oxygen-treated sample
(2.4 × 10−8 mol s−1 g−1) compared with a freshly prepared
catalyst (6.8 × 10−8 mol s−1 g−1). We note that the reported
benzene synthesis rates per gram catalyst for both freshly
prepared and oxygen-treated samples were normalized by the
weight of the spent catalyst in which the catalysts were passivated
in a flow of 1% O2/He before being removed from the reactor.
The weights of the spent catalysts were found to be within 3% of
the theoretical weight of Mo2C (based on AMT amount),
indicating that bulk carburization of the precursor (AMT) was
complete.
CO chemisorption at 323 K), ∼4 × 10−4 mol s−1 molCOsite
−1
(Table 3), was found to be within the average value of those
reported previously (average value 3.4 1.0 × 10−4 mol s−1
molCOsite−1);3 the corresponding benzene synthesis rate as a
function of time is shown in Figure S6. The temperature profile
consisted of two consecutive segments in which the first one was
intended to simulate the typical catalyst activation process
(catalysts treated in a flow of H2 at ∼723 K for 1 h; see section
2.3), and the second segment attempted to remove the residual
oxygen, if any, in the molybdenum carbide catalysts via a
temperature ramp to 1056 K. Oxygen was found to be removed
predominantly in the form of water (∼7400 μmol g−1, Table 3),
which is at least ∼20 times in excess of the amount removed as
CO (∼150 μmol g−1) or CO2 (∼340 μmol g−1) during the entire
TPSR process (Figure 6).
The observed water profile shown in Figure 6 is similar to that
reported by Leary et al.24 in which a sharp water peak was
observed at 518 K, followed by two close peaks at 598 and 658 K.
Quantification of the amount of oxygen evolved during the TPSR
experiment, in which water was considered as the only oxygen-
containing eluent, revealed that only ∼81% of oxygen in the
passivated molybdenum carbide catalyst could be removed
during the typical activation process (H2 treatment at 723 K for 1
h, the first segment shown in Figure 6) if we assume that 100%
oxygen removal can be achieved at 1056 K in the presence of H2
flow. This experimental observation is in line with the difficulty of
completely removing oxygen in tungsten carbide catalysts
reported by Ribeiro et al.18 in which 50% of oxygen was found
to be still retained after thermal treatment in H2 at 1150 K. The
BET surface area of this molybdenum carbide sample was found
to be ∼64 m2 g−1 (after CO chemisorption as described in
section 2.2). The amount of oxygen retained in a molybdenum
carbide catalyst after the typical activation process (segment A,
Figure 6) was found to be ∼1.3 monolayers on the basis of BET
surface area and a surface Mo site density of 1 × 1015 sites cm−2,14
which corresponds to a formulation of Mo2CO0.3 if (i) no excess
carbon in the original molybdenum carbide is considered and (ii)
100% oxygen removal after the whole TPSR process (Table 3;
see SI for the estimation of the catalyst formulation). This
The presence of oxygen in the oxygen-treated molybdenum
carbide sample could be inferred from the suppressed rate of
cyclohexane formation during the transient of vapor phase
anisole HDO as compared with that for a freshly prepared
4111
DOI: 10.1021/acscatal.5b00713
ACS Catal. 2015, 5, 4104−4114

ACS Catalysis
Research Article
synthesis. The unperturbed chemistry per operating site in the
oxygen-treated molybdenum carbide catalysts was further
Table 4. Vapor Phase Anisole HDO over Oxygen-Treated and
Freshly Prepared (without exposure to air) Molybdenum
Carbide Catalysts at 423 K under Ambient Pressure
evidenced by the similar benzene TOF obtained by in situ CO
titration for the two samples (∼1.5 × 10−3 mol s−1 molCOsite
,
−1
oxygen-treated
Table 4). These experimental results show that although oxygen
treatment can successfully introduce oxygen in a molybdenum
carbide catalyst, the presence of oxygen, at least at the level of 734
μmol g−1 (0.075 O/Mobulk (molar ratio)), alters the number and
not the identity or reactivity of the active sites for the HDO
reaction.
measured kinetics
fresh Mo2C
Mo2C
a
b
benzene synthesis rate (mol s−1g−1
)
6.8 × 10−8
1.7 × 10−3
96.7
2.4 × 10−8
1.5 × 10−3
benzene TOF (mol s−1 molCOsite
benzene selectivity (%)
cyclohexane selectivity (%)
phenol selectivity (%)
)
−1
92.7
4.5
3.1
0.2
2.0
During the transient, the fresh Mo2C formulation can
accomplish both Ar−O bond hydrogenolysis and benzene
hydrogenation, as shown in Figure 4a; however, at steady state,
the working catalyst that incorporates a certain amount of oxygen
primarily accomplishes only Ar−O bond cleavage reactions. We
cannot probe the reactivity of fresh Mo2C samples when oxygen
is present in one of the reactants and, hence, emphasize the point
that oxygen accumulation after the initial transient changes only
the number of sites and not their identity.
We have previously reported that CO chemisorption uptake
for molybdenum carbide catalysts decreases as a function of aging
time on the shelf, presumably as a result of slow oxidation via
inevitable air exposure (Figure S9 in the SI).27 Figure 7 shows
methane selectivity (%)
methanol selectivity (%)
11.3
20.5
79.5
1
88.7
apparent activation energy, Ea
71
1
65
c
(kJ mol−1
)
a
Steady state values taken as an average of 3−4 h time on-stream. The
rate per gram catalyst was normalized by the weight of the spent
catalyst in which the catalyst was passivated in 1% O2/He before being
removed from the reactor. Catalysts were prepared starting from ∼1.6
b
g AMT. An average value of two independent CO titration
experiments in which CO (0.035 cm3 s−1)/Ar (0.035 cm3 s−1) gas
mixtures were co-fed in the reactant feed; the effect of CO
hydrogenation was not considered for benzene TOF calculation.
c
The corresponding Arrhenius plot is shown in Figure S8 in the SI.
molybdenum carbide catalyst (Figure S7 in the SI). The amount
of oxygen incorporated in the oxygen treated molybdenum
carbide samples, based on the water evolved during the TPSR
experiments for the two oxygen treated Mo2C samples (section
2.1, Table S1 in the SI), was assessed to be ∼734 μmol g−1
(equivalent to 0.075 O/Mobulk (molar ratio)). We note that the
amount of water evolved during the H2/Ar treatment at 723 K for
1 h was found to be essentially the same, with an error of <1%, for
the two oxygen-treated Mo2C samples (Table S1 in the SI),
suggesting that the oxygen treatment was reproducible and the
quantification of the oxygen incorporated in the Mo2C catalysts
was representative. The TOF of benzene synthesis considering
one incorporated oxygen atom to occupy one active site for
anisole HDO was found to be 6 × 10−5 mol s−1 molOsite −1. This
TOF value is ∼16 times lower than that obtained with CO as a
titrant (1.1 0.3 × 10−3 mol s−1 molCOsite −1, Table 1), suggesting
that oxygen can either adsorb on sites that are not related to
anisole HDO chemistry or incorporate into the bulk structure of
molybdenum carbide, resulting in a much higher uptake of
oxygen compared with that of CO.16,25 Oyama and co-workers
have also shown that oxygen chemisorption uptake was ∼6 times
higher than CO chemisorption uptake on Mo2C catalysts,
resulting in a much lower turnover rate per oxygen
chemisorption site for cumene hydrogenation as compared
with that per CO chemisorption site.26 CO chemisorption
uptake is, therefore, a more relevant descriptor to gauge benzene
synthesis rates of molybdenum carbide catalysts for vapor phase
anisole HDO.
Figure 7. Benzene TOF determined by in situ CO titration for Mo2C
samples with different CO chemisorption uptakes achieved via different
sample pretreatment conditions (Table S4). Data shown represent the
average of multiple (7−31 times) individual titration experiments with
different CO co-fed flow rates and different catalyst loadings, as reported
in Tables 1, S2, S3. Error bars represent one standard deviation of
multiple titration experiments in the corresponding experiment set.
that benzene synthesis TOF measured by in situ CO titration
(1.1 0.3 x10−3 mol s−1 molCOsite−1) is invariant over a series of
molybdenum carbide samples, with varying amounts of CO
chemisorption uptake (measured by ex situ CO chemisorption,
323 K) achieved by aging or treating the catalysts under different
pretreatment conditions (see Table S4 for detailed information
on pretreatment conditions), which is consistent with the
experimental observation that the net effect of oxygen
incorporation, at least above the amount of oxygen required
for Mo2C catalysts to reach steady state anisole HDO rates (i.e.,
exclusive Ar−O bond cleavage; see section 3.2), is to alter the
observed benzene synthesis rates per gram of catalyst but not the
benzene synthesis rate per active site.
The benzene selectivity, however, was found to be similar, on
both oxygen -treated and freshly prepared molybdenum carbide
+
catalysts, at a level of ∼90% on a C6 basis (Table 4). No
significant difference in methane/methanol selectivities was
observed for the two samples, either. The apparent activation
energy for benzene formation was found to be similar for both
samples, ∼70 kJ mol−1 (Table 4, and Figure S8 in the SI for the
corresponding Arrhenius plot). The similarity in product
selectivity as well as apparent activation energy suggests that
the effect of incorporated oxygen in molybdenum carbide
catalysts is to reduce the number of active sites for benzene
CO has been generally accepted as a probe molecule to titrate
metallic sites through its π-bond donation from the antibonding
4112
DOI: 10.1021/acscatal.5b00713
ACS Catal. 2015, 5, 4104−4114

ACS Catalysis
Research Article
Table 5. CO Chemisorption Uptake and Benzene Synthesis Rate of Vapor Phase Anisole HDO for MoO3 Samples Pretreated in
Either Pure H2 or CH4/H2 Mixtures Using Different Temperature Protocols
CO
benzene synthesis rate (×
x for
MoOx
chemisorption benzene synth rate (×
10−5 mol−1
a
c
c
sample
cat. pretreatment
H2, RT (35 min)−675 K (1 h)−423 K
H2, RT (35 min)−675 K (3 h)−423 K
H2, RT (35 min)−673 K (6 h)−423 K
(μmol g−1
)
10−10 mol s−1 g−1
)
s−1 molCOsite
)
b
1
2
3
4
2.8
2.5
1.7
9.4
2.6
2.8
b
33
4.6
1.4
87.1
23.1
1110
2.7
d
15% CH4/H2, 423 K (30 min)−648 K−973 K (150 min) (2 h)−423 K
19−37
a
b
See section 2.1 for the nomenclature of the temperature protocols. Catalyst loading was ∼0.42 g for all the experiments. An average value of two
independent CO chemisorption measurements: 7.9 and 10.8 μmol g−1 for MoO2.8, and 34.7 and 31.3 μmol g−1 for MoO2.5. CO chemisorption values
reported represent independent experiments performed by loading ∼0.42 g of MoO3 into a chemisorption cell, followed by the corresponding H2
c
treatment, as shown in this Table. Data reported as an average value between 1 and 2 h time-on-stream. Figure S11 shows benzene synthesis rates vs
time-on-stream for sample no. 3. Reactions were carried out at ∼423 K under ambient pressure. Reaction feed composition: anisole/H2 (vol %) =
d
∼0.15/bal with total flow rate ∼1.67 cm3 s−1. Catalyst mass ∼0.42 g was used to calculate benzene synthesis rates. Sample no. 4 was obtained by in
situ carburization of sample no. 1 after it was tested for anisole HDO for ∼25 ks. Carburization was carried out in a flow of 15% CH4/H2 at ∼2.11
cm3 s−1 with a designated temperature protocol. After carburization, sample no. 4 was first tested for vapor phase anisole HDO at 423 K at 1 atm for
∼30 ks. Sample no. 4 was then recarburized using the same carburization conditions described above, followed by an independent measurement of
vapor phase anisole HDO rates for ∼11 ks. Data reported as an average value between 1 and 2 h time-on-stream for the two independent anisole
HDO measurements. Rates were normalized to ∼0.306 g of Mo2C by considering a stoichiometric conversion from MoO3 to Mo2C. CO
chemisorption uptake values of ∼300−600 μmol g−1 were used to normalize benzene synthesis rates per CO chemisorption site shown in this table.
orbital to metals; however, CO can also be used to titrate Lewis
acid sites, which can accept the lone pair from CO. The latter case
would be similar to an “oxygen vacancy” on the surface of Mo
oxycarbide.28 The potential active sites involved in vapor phase
anisole HDO can be, therefore, Mo metallic sites or surface
oxygen vacancy sites, although we cannot distinguish between
these two scenarios on the basis of the in situ CO titration results.
3.6. Vapor Phase Anisole HDO of MoOx Catalysts.
Recently, Prasomsri et al.29 showed that MoO3 can also
selectively cleave the phenolic C−O bond to convert anisole to
benzene in the vapor phase with ∼39% yield at ∼673 K under
ambient pressure. In this work, we assess the rate of vapor phase
anisole HDO on bulk molybdenum oxide (MoOx) at lower
reaction temperatures (∼423 K) and compare oxide and carbide
formulations for low temperature HDO. Table 5 shows both CO
chemisorption uptake and the corresponding benzene synthesis
rate for MoOx formulations prepared by treating MoO3 under
different H2 treatment conditions prior to vapor phase anisole
HDO at ∼423 K under ambient pressure. MoO3 (identified by
XRD, JCPDS 035-0609, Figure S10 in the SI) treated in H2 at
∼675 K for 1, 3, and 6 h was progressively reduced to MoO2.8,
MoO2.5, and MoO1.7, respectively, in which the stoichiometry of
the reduced formulations was assessed on the basis of
quantification of the water removed during the H2 treatment.
The reduction of MoO3 was further confirmed by XRD analysis
in which the diffraction pattern of the spent MoO1.7 catalyst was
found to match that of MoO2 (JCPDS 032−0671), as shown in
Figure S10. Both the benzene synthesis rate per gram of catalyst
and the CO chemisorption uptake were found to increase as the
extent of reduction of the MoO3 catalysts increases (Table 5),
presumably because of the increasing number of Lewis acid sites
(oxygen vacancies).22
normalized to the CO chemisorption uptake, and it was found to
be nearly invariant (2.3
0.8 × 10−5 mol s−1 molCOsite−1),
suggesting that metallic sites associated with H2-treated MoO3
catalysts are involved in vapor phase anisole HDO (Table 5).
Benzene synthesis rates per CO chemisorption site for the MoOx
catalysts (2.3 0.8 × 10−5 mol s−1 molCOsite−1), however, were
found to be at least 10 times lower than that for typical Mo2C
catalysts (∼4 × 10−4 mol s−1 molCOsite−1; Table 3).
In situ carburization of the MoO2.8 catalyst, after ∼25 ks time-
on-stream for vapor phase anisole HDO, was carried out prior to
kinetic measurements (denoted as sample no. 4 in Table 5) to
demonstrate the difference in the benzene synthesis rates over
carbidic and MoOx sites (see Scheme S1 in the SI for the
evolution of MoO3 catalysts under different pretreatment
conditions). The benzene synthesis rate per gram of catalyst
(average value of two independent measurements) for the
“oxygen-exposure-free” carburized sample (sample no. 4 in Table
5) was found to be ∼1.1 × 10−7 mol s−1 g−1, which is ∼50 times
higher than that for the MoO2.8 catalyst (sample no. 1 in Table
5), and the selectivity of methane (C1 basis), methanol (C1
+
basis), and benzene (C6 basis) were found to be 11, 89, and 97%
respectively, which are similar to those obtained using
molybdenum carbide catalysts.3 If we use the CO chemisorption
uptake reported by Lee et al.,5 ∼300−600 μmol g−1, to normalize
the benzene synthesis rate for this freshly carburized catalyst, the
corresponding benzene synthesis rates per CO chemisorption
site were ∼1.9 × 10−4 to ∼3.7 × 10−4 mol s−1 molCOsite−1. These
rates are significantly higher (∼7−14 times) than those for MoOx
catalysts (Table 5), demonstrating that metallic sites associated
with bulk molybdenum oxide catalysts (MoOx), are much less
active for vapor phase anisole HDO at low temperature (∼423
K) and ambient pressure as compared with molybdenum carbide
catalysts.
A typical plot of benzene synthesis rate per gram of catalyst as a
function of time for the MoO1.7 catalyst is shown in Figure S11 in
which the catalyst was found to suffer from continuous
deactivation over the course of the investigation (∼20 ks).
Benzene and methanol were the only products observed for
samples 1−3 in Table 5. Because metallic sites in Mo2C catalysts
assessed by ex situ CO chemisorption were found to correlate
with measured rates of vapor phase anisole HDO,3 the average
benzene synthesis rate between 1 and 2 h time-on-stream was
4. CONCLUSIONS
Site requirements of molybdenum carbide catalysts for vapor
phase anisole HDO were investigated via in situ CO titration,
transient kinetic studies, and temperature-programmed surface
reaction with H2. Approximately 1.3 monolayers of oxygen were
found to be still retained in a molybdenum carbide catalyst,
which was passivated by 1% O2/He, even after hydrogen
4113
DOI: 10.1021/acscatal.5b00713
ACS Catal. 2015, 5, 4104−4114

ACS Catalysis
Research Article
(7) Schaidle, J. A.; Lausche, A. C.; Thompson, L. T. J. Catal. 2010, 272,
235−245.
treatment at 723 K for 1 h prior to reaction. The further
accumulation of ∼0.29 monolayer of oxygen (based on BET
surface area) during the observed transient in anisole HDO was
evidenced from an oxygen mass balance, demonstrating that in
situ oxidation of molybdenum carbide by the reactant (anisole)
during HDO could not be avoided, even at a high hydrogen-to-
anisole molar ratio (∼110) in the reactant feed. The TOF for
benzene synthesis estimated by in situ CO titration was found to
be 1.1 0.3 × 10−3 mol s−1 molCOsite−1 with an assumption that
one CO molecule titrates one active site for benzene synthesis,
suggesting that <20% of total potential active sites (assessed by ex
situ CO chemisorption) are operating during anisole HDO at
423 K under ambient pressure. An oxygen treatment could
successfully incorporate oxygen (734 μmol g−1, O/Mobulk (molar
ratio) = 0.075) in molybdenum carbide. The corresponding
benzene synthesis rates per gram of catalyst were found to be ∼3
times lower in these formulations compared with a freshly
prepared molybdenum carbide catalyst. Similar product
selectivity, apparent activation energy for benzene formation,
and TOF measured by in situ CO titration for the oxygen treated
and freshly prepared molybdenum carbide catalysts suggest that
the effect of oxygen, at a level of 734 μmol g−1, is to reduce the
number of the active sites for anisole HDO. The observed
benzene synthesis rates per CO chemisorption site for bulk
MoOx catalysts were found to be ∼10 times lower than those for
Mo2C catalysts, suggesting that the dominant active sites for
anisole HDO at 423 K under ambient pressure are associated
with molybdenum carbide, not bulk molybdenum oxide.
(8) Liu, P.; Rodriguez, J. A. J. Phys. Chem. B 2006, 110, 19418−19425.
(9) Iglesia, E.; Ribeiro, F. H.; Boudart, M.; Baumgartner, J. E. Catal.
Today 1992, 15, 307−337.
(10) Ribeiro, F. H.; Boudart, M.; Dallabetta, R. A.; Iglesia, E. J. Catal.
1991, 130, 498−513.
(11) Ribeiro, F. H.; Betta, R. A. D.; Boudart, M.; Baumgartner, J.;
Iglesia, E. J. Catal. 1991, 130, 86−105.
(12) Blekkan, E. A.; Cuong, P.-H.; Ledoux, M. J.; Guille, J. Ind. Eng.
Chem. Res. 1994, 33, 1657−1664.
(13) Choi, J.-S.; Krafft, J.-M.; Krzton, A.; Djega-Mariadassou, G. Catal.
Lett. 2002, 81, 175−180.
(14) Choi, J.-S.; Bugli, G.; Djega-Mariadassou, G. J. Catal. 2000, 193,
238−247.
(15) Ren, H.; Yu, W.; Salciccioli, M.; Chen, Y.; Huang, Y.; Xiong, K.;
Vlachos, D. G.; Chen, J. G. G. ChemSusChem 2013, 6, 798−801.
(16) Ranhotra, G. S.; Haddix, G. W.; Bell, A. T.; Reimer, J. A. J. Catal.
1987, 108, 24−39.
(17) Iglesia, E.; Ribeiro, F. H.; Boudart, M.; Baumgartner, J. E. Catal.
Today 1992, 15, 455−458.
(18) Ribeiro, F. H.; Dalla Betta, R. A.; Guskey, G. J.; Boudart, M. Chem.
Mater. 1991, 3, 805−812.
(19) Shannon, S. L.; Goodwin, J. G. Chem. Rev. 1995, 95, 677−695.
(20) Shou, H.; Davis, R. J. J. Catal. 2013, 306, 91−99.
(21) DeWilde, J. F.; Chiang, H.; Hickman, D. A.; Ho, C. R.; Bhan, A.
ACS Catal. 2013, 3, 798−807.
(22) Prasomsri, T.; Shetty, M.; Murugappan, K.; Roman-Leshkov, Y.
Energy Environ. Sci. 2014, 7, 2660−2669.
(23) Miyao, T.; Shishikura, I.; Matsuoka, M.; Nagai, M.; Oyama, S. T.
Appl. Catal., A 1997, 165, 419−428.
(24) Leary, K. J.; Michaels, J. N.; Stacy, A. M. J. Catal. 1986, 101, 301−
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acscatal.5b00713.
313.
■
(25) Sajkowski, D. J.; Oyama, S. T. Appl. Catal., A 1996, 134, 339−349.
(26) St. Clair, T. P.; Dhandapani, B.; Oyama, S. T. Catal. Lett. 1999, 58,
169−171.
(27) Lee, W.-S.; Wang, Z.; Zheng, W.; Vlachos, D. G.; Bhan, A. Catal.
Sci. Technol. 2014, 4, 2340−2352.
(28) Delporte, P.; Meunier, F.; Pham-Huu, C.; Vennegues, P.; Ledoux,
M. J.; Guille, J. Catal. Today 1995, 23, 251−267.
(29) Prasomsri, T.; Nimmanwudipong, T.; Roman-Leshkov, Y. Energy
Environ. Sci. 2013, 6, 1732−1738.
S
Estimation of catalyst formulation of passivated Mo2C
catalysts from TPSR, 4 tables of data, 1 scheme, and 11
figures showing results (PDF)
AUTHOR INFORMATION
Corresponding Author
*Fax: (+1) 612-626-7246. E-mail: abhan@umn.edu.
■
Present Address
†(W.-S.L.) Core R&D, The Dow Chemical Company, 2030 Dow
Center, Midland, Michigan 48674, USA.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This research was supported by Office of Basic Energy Sciences,
the U.S. Department of Energy. under Award No. DE-
SC0008418 (DOE Early Career Program).
REFERENCES
■
(1) Zakzeski, J.; Bruijnincx, P. C. A.; Jongerius, A. L.; Weckhuysen, B.
M. Chem. Rev. 2010, 110, 3552−3599.
(2) Rensel, D. J.; Rouvimov, S.; Gin, M. E.; Hicks, J. C. J. Catal. 2013,
305, 256−263.
(3) Lee, W.-S.; Wang, Z.; Wu, R. J.; Bhan, A. J. Catal. 2014, 319, 44−53.
(4) Lee, J. S.; Oyama, S. T.; Boudart, M. J. Catal. 1987, 106, 125−133.
(5) Oyama, S. T. The Chemistry of Transition Metal Carbides and
Nitrides; Oyama, S. T., Ed.; Springer Science & Business Media:
London, New York, 1996; pp 398−413.
(6) Ranhotra, G. S.; Bell, A. T.; Reimer, J. A. J. Catal. 1987, 108, 40−49.
4114
DOI: 10.1021/acscatal.5b00713
ACS Catal. 2015, 5, 4104−4114