Role of the Support and Reaction Conditions on the Vapor-Phase Deoxygenation of m-Cresol over Pt/C and Pt/TiO2 Catalysts

Article abstract of DOI:10.1021/acscatal.5b02868

The catalytic deoxygenation of biomass fast pyrolysis vapors offers a promising route for the sustainable production of liquid transportation fuels. However, a clear understanding of the mechanistic details involved in this process has yet to be achieved, and questions remain regarding the role of the catalyst support and the influence of reaction conditions. In order to gain insight into these questions, the deoxygenation of m-cresol was investigated over Pt/C and Pt/TiO₂ catalysts using experimental and computational techniques. The performance of each catalyst was evaluated in a packed-bed reactor under two conditions (523 K, 2.0 MPa and 623 K, 0.5 MPa), and the energetics of the ring hydrogenation, direct deoxygenation, and tautomerization mechanisms were calculated over hydrogen-covered Pt(111) and oxygen vacancies on the surface of TiO₂(101). Over Pt(111), ring hydrogenation to 3-methylcyclohexanone and 3-methylcyclohexanol was found to be the most energetically favorable pathway. Over TiO₂(101), tautomerization and direct deoxygenation to toluene were identified as additional energetically favorable routes. These calculations are consistent with the experimental data, in which Pt/TiO₂ was more active on a metal site basis and exhibited higher selectivity to toluene at 623 K relative to Pt/C. On the basis of these results, it is likely that the reactivity of Pt/TiO₂ and Pt/C is driven by the metallic phase at 523 K, while contributions from the TiO₂ support enhance deoxygenation at 623 K. These results highlight the synergistic effects between hydrogenation catalysts and reducible metal oxide supports and provide insight into the reaction pathways responsible for their enhanced deoxygenation performance.

Full text of DOI:10.1021/acscatal.5b02868

Research Article
pubs.acs.org/acscatalysis
Role of the Support and Reaction Conditions on the Vapor-Phase
Deoxygenation of m‑Cresol over Pt/C and Pt/TiO Catalysts
2
†
†
Michael B. Griffin, Glen A. Ferguson, Daniel A. Ruddy, Mary J. Biddy, Gregg T. Beckham,*
and Joshua A. Schaidle*
National Bioenergy Center, National Renewable Energy Laboratory, 15013 Denver West Parkway, Golden, Colorado 80401, United
States
*
S Supporting Information
ABSTRACT: The catalytic deoxygenation of biomass fast pyrolysis vapors offers a
promising route for the sustainable production of liquid transportation fuels. However,
a clear understanding of the mechanistic details involved in this process has yet to be
achieved, and questions remain regarding the role of the catalyst support and the
influence of reaction conditions. In order to gain insight into these questions, the
deoxygenation of m-cresol was investigated over Pt/C and Pt/TiO catalysts using
2
experimental and computational techniques. The performance of each catalyst was
evaluated in a packed-bed reactor under two conditions (523 K, 2.0 MPa and 623 K,
0.5 MPa), and the energetics of the ring hydrogenation, direct deoxygenation, and
tautomerization mechanisms were calculated over hydrogen-covered Pt(111) and
oxygen vacancies on the surface of TiO (101). Over Pt(111), ring hydrogenation to 3-
2
methylcyclohexanone and 3-methylcyclohexanol was found to be the most energetically favorable pathway. Over TiO (101),
2
tautomerization and direct deoxygenation to toluene were identified as additional energetically favorable routes. These
calculations are consistent with the experimental data, in which Pt/TiO was more active on a metal site basis and exhibited
2
higher selectivity to toluene at 623 K relative to Pt/C. On the basis of these results, it is likely that the reactivity of Pt/TiO and
2
Pt/C is driven by the metallic phase at 523 K, while contributions from the TiO support enhance deoxygenation at 623 K.
2
These results highlight the synergistic effects between hydrogenation catalysts and reducible metal oxide supports and provide
insight into the reaction pathways responsible for their enhanced deoxygenation performance.
KEYWORDS: hydrodeoxygenation, catalytic fast pyrolysis, TiO , m-cresol, bio-oil, DFT, hydrogen coverage, platinum
2
1
. INTRODUCTION
are catalytically deoxygenated in the presence of hydrogen.
Supported noble metals are attractive catalysts for ex situ CFP,
since they can activate molecular hydrogen and are stable in
high-steam environments. Many studies have demonstrated the
utility of supported noble-metal catalysts for deoxygenation of
aromatic oxygenates (e.g., phenol, cresol, guaiacol), which are
representative of the lignin-derived components contained
Biomass deconstruction using fast pyrolysis offers a promising
route for the production of renewable bio-oil.
high oxygen content of bio-oil contributes to a number of
undesirable characteristics, and bio-oil must be upgraded before
it is suitable for use as a drop-in transportation fuel or
blendstock. Ex situ catalytic fast pyrolysis (CFP), outlined in
Figure 1, has been identified as an economically favorable
approach for the production and upgrading of bio-oil. In this
process biomass undergoes fast pyrolysis, and the resulting
vapors are sent to a secondary downstream reactor, where they
1
−4
However, the
5
,6
10−21
within bio-oil.
Although much has been learned from these
studies, a clear understanding of the mechanistic details, role of
the support, and influence of the reaction conditions has yet to
be achieved.
6
−9
Molecules such as m-cresol are ideal model compounds to
gain insight into the deoxygenation of substituted phenols,
which is particularly difficult due to the high dissociation energy
22
of the aryl−oxygen bond. From a number of mechanistic
investigations over a variety of catalysts, three primary reaction
pathways have been proposed for this process, including ring
2
2,23
hydrogenation followed by dehydration (HDO),
direct
2
4,25
deoxygenation (DDO),
and tautomerization
10,11,26,27
3,6,8,10,14,15,19,22,25,28−31
(
TAU),
as shown in Figure 2.
Received: December 15, 2015
Revised: March 2, 2016
Figure 1. Process flow diagram for ex situ CFP and subsequent
upgrading steps. Adapted from Ruddy et al.
6
©
XXXX American Chemical Society
2715
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727

ACS Catalysis
Research Article
11
deoxygenation over Pt/SiO2. Additionally, intermediates in
the TAU pathway such as 2-cyclohexen-1-one have been
identified using in situ diffuse reflectance infrared Fourier
transform spectroscopy (DRIFTS) during deoxygenation
10,11
studies of phenol over noble-metal catalysts.
For m-cresol
deoxygenation over a Pt/SiO2 catalyst, DFT calculations
demonstrated that the TAU pathway was more energetically
favorable than DDO on Pt(111), in agreement with the
experimental observation that selectivity to 3-methylcyclohex-
20
anone was greater than that to toluene. However, over a more
oxophilic metal such as Ru(0001), DFT calculations indicated
that DDO was favored, and toluene was the dominant reaction
2
0
product observed experimentally. The importance of
oxophilicity on the TAU pathway is consistent with a study
4
6,47
by Hensley et al.,
which reported that the keto−enol
isomerization of phenol over the oxophillic base metal Fe(110)
surface proceeds with a high barrier of 1.83 eV.
4
6
Figure 2. Proposed pathways for deoxygenation of aromatic
compounds over noble-metal catalysts.
Development of an overarching mechanism is also
complicated by varying reaction conditions employed during
the deoxygenation of pyrolysis vapors. For example, ex situ
CFP requires higher temperatures (573−723 K) and lower
pressures (0.1−1 MPa) than typical deoxygenation via
The HDO pathway, shown in Figure 2a, has been proposed
as the major pathway for deoxygenation of aromatic
compounds (including for m-cresol) over silica- and alumina-
6
hydrotreating (HT; 473−573 K, 1−20 MPa). The resulting
23
differences in surface coverage of hydrogen and substrate as
well as thermodynamic and kinetic limitations will likely
influence the dominant reaction pathways. Indeed, recent
reports demonstrate that the reaction temperature plays an
important role in determining the product distribution from
deoxygenation of bio-oil model compounds over noble-metal
catalysts. At temperatures ≤523 K, ring hydrogenation is
favorable and saturated products are produced with high
selectivity; however, at temperatures above 523 K, hydro-
supported platinum catalysts. The HDO pathway requires a
bifunctional catalyst that can hydrogenate the ring and
dehydrate the resulting alcohol. While this pathway is widely
6
,14,19,22,25,27,28,32
proposed,
computational studies using density
functional theory (DFT) suggest that significant barriers exist
for the initial ring hydrogenation of guaiacol and catechol over
1
0,12,23
Pt(111),
leading to a disparity between experimental
results and model predictions.
The DDO pathway, shown in Figure 2b, has been proposed
14,33,34
24
for guaiacol deoxygenation over Pt/γ-Al O
Computational results from Chiu et al.
and Pt/C.
indicate that DDO
genation activity decreases and the production of aromatic
2
3
3
5,36
13−15,18,23,48−50
products increases.
Shifts in the thermodynamic
is most likely to occur at step sites for a Ru catalyst. This result
is consistent with findings of Lu et al., who used microkinetic
modeling to show that the barrier for DDO over the Pt(111)
equilibrium at higher temperature may contribute to these
6
differences. However, the limited amount of directly
comparable data makes definitive conclusions difficult, and a
better understanding of catalyst behavior under these two
conditions is critical to understand the product distribution
obtained during the deoxygenation of pyrolysis vapors.
Previous experimental and computational studies have not
yet arrived at a clear mechanism for aromatic deoxygenation,
and it is likely that multiple routes contribute to the observed
catalyst reactivity. Further investigation is necessary to gain a
better understanding of the influence of reaction conditions,
mechanistic details, and role of the support. To that end, here
we present a combined experimental−computational inves-
tigation into the deoxygenation of the model compound m-
12
surface was kinetically unfavorable. The authors proposed
that either corner and edge sites or the support contribute to
deoxygenation of phenol and catechol. Other reports have also
suggested that the support has the potential to influence the
activity and selectivity observed during the deoxygenation of
bio-oil model compounds over noble-metal cata-
1
1,14,15,23,27,31,32,37−43
lysts.
Particularly interesting results have
been observed over reducible metal oxides such as TiO , CeO ,
2
2
1
0,17
and ZrO2,
where deoxygenation is hypothesized to involve
44
oxygen vacancies. These studies suggest that reducible metal
oxide supports may contribute to the deoxygenation of model
bio-oil compounds via a DDO mechanism at an oxygen defect
site.
cresol over Pt/C and Pt/TiO catalysts. Each catalyst was
2
evaluated in a gas-phase flow reactor under temperatures and
pressures relevant to ex situ CFP (623 K, 0.5 MPa) and HT
(523 K, 2.0 MPa). Computational modeling was conducted
over a hydrogen-covered Pt(111) surface to better understand
the transformations that occur over the metallic phase and over
The TAU pathway was recently proposed by Nie and
1
0,11,20,26,45
Resasco
and is shown in Figure 2c. This pathway
involves a keto−enol tautomerization, which occurs between
the hydroxyl group and the adjacent carbon−carbon double
bond on the aromatic ring. The first step is the hydrogen shift
from the hydroxyl group, resulting in the keto intermediate.
This intermediate can then undergo hydrogenation to form
either the saturated ketone or a reactive alcohol. The ketone
can undergo hydrogenation−deoxygenation, while the reactive
alcohol can be deoxygenated to form toluene. Experimental
support for this mechanism includes the development of a
kinetic model based on the TAU pathway that accurately
predicts the product distribution observed during m-cresol
an oxygen vacancy on the TiO (101) surface to gain insight
2
into energetically accessible pathways over reducible metal
oxide supports. A combination of the experimental and
computational results was used to develop a reaction network
that incorporates the influence of reaction conditions and the
role of the support during m-cresol deoxygenation. The
computational results indicate that the most energetically
favorable pathway is ring hydrogenation to 3-methylcyclohex-
anone and 3-methylcyclohexanol over Pt(111), while deoxyge-
2
716
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727

ACS Catalysis
Research Article
nation via tautomerization and direct deoxygenation may
proceed over TiO (101) with slightly higher barriers. These
2
findings, which are consistent with the experimentally observed
product selectivities, suggest that the reactivity of Pt/TiO and
2
Pt/C is driven by the metallic phase at 523 K while
contributions from the TiO support enhance deoxygenation
2
at 623 K.
2. RESULTS
2.1. Catalyst Synthesis and Characterization. The Pt
catalysts were prepared by typical incipient wetness methods
and reduced at 723 K prior to catalytic characterization. The X-
ray diffraction (XRD) patterns of the reduced Pt/C and Pt/
TiO are given in Figure 3. In both cases, the characteristic
2
Figure 4. TEM images and particle size distributions for Pt/TiO (a,
2
b) and Pt/C (c, d). A high-magnification inset in (a) shows the
measured lateral spacing between Pt atoms.
uptake, dispersion, average crystallite size, and average particle
diameter according to chemisorption, XRD, and TEM are
shown in Table 1. CO chemisorption carried out over TiO and
2
Table 1. Pt/TiO and Pt/C Metal Loading, CO Uptake,
2
Dispersion, Average Crystallite Size, and Average Particle
Diameter Determined by CO Chemisorption, XRD, and
TEM
Figure 3. XRD patterns of (a) Pt/TiO and (b) Pt/C with reference
2
peaks for Pt, anatase TiO , and graphite. The small peaks at 27 and
2
36° in (a) are attributed to diffractions from rutile TiO2.
metal loading, wt % CO uptake, μmol gcat−
1
dispersion, %
catalyst
diffraction pattern of crystalline Pt is evident. The titania used
in this report is a mixed-phase material consisting of primarily
the anatase phase, with a lesser portion of rutile (small peaks at
Pt/C
4.32
4.8
87.9
26.7
39.4
10.9
Pt/TiO2
particle size, nm
2
7 and 36°, Figure 3a). In addition to Pt diffractions, the Pt/C
XRD, crystallite
size
TEM, surface area
weighted
TEM, volume
weighted
material consists of graphitic carbon and minor contributions
from other crystalline phases associated with the carbon
support (Figure 3b). A diffractogram of the carbon support
in the absence of Pt is presented in Figure S1 in the Supporting
Information.
chemisorption
4
.1
26.4
24.4
28.2
17.5
29.9
26.7
1
5.2
Transmission electron microscopy (TEM) images and the
corresponding particle size distribution for Pt/TiO and Pt/C
carbon in the absence of Pt exhibited no detectable CO uptake.
The average crystallite sizes determined by XRD line
broadening agree reasonably well with the volume-weighted
TEM measurements and are primarily influenced by con-
tributions from the largest particles. The average particle sizes
determined by CO chemisorption are lower than the surface-
weighted TEM measurements, particularly for the Pt/C
catalyst. This apparent discrepancy may be due to the presence
of small particles (<2 nm) that contribute to CO uptake but are
unable to be observed by TEM and are expected to minimally
influence the crystallite size determined by XRD, which is
volume-weighted. The increased dispersion of Pt/C relative to
2
are shown in Figure 4. The majority of the particles observed
were spherical, although a small number of elongated particles
were present on both catalysts. A micrograph showing the
spherical shape of a platinum particle on titania is provided in
Figure S2 in the Supporting Information. In comparison to the
Pt/TiO catalyst, Pt/C exhibited lower polydispersity in size in
2
the vicinity of the average (0−5 nm) but contained a greater
number of large particles (>20 nm). An inset provided in
Figure 4a shows a high-magnification micrograph of Pt/TiO2.
The lateral spacing of Pt atoms in this image was 2.84 Å, which
is in agreement with the surface lattice constant of Pt(111)
determined via scanning tunneling microscopy (2.80 Å) and
Pt/TiO according to CO chemisorption is consistent with a
2
5
1
density functional theory calculations (2.83 Å). Carbon-
supported Pt particles are also known to adopt a truncated-
octahedral geometry with predominately (111) sites exposed at
previous study and is likely a result of the high surface area of
17
the carbon support. Although the Pt particle sizes are
different on the carbon- and TiO -supported catalysts, previous
2
5
2,53
diameters >2 nm.
The BET surface areas of the carbon and
reports demonstrated that metal dispersion does not
2
−1
TiO supports were determined to be 960 and 95 m g ,
significantly affect the product distribution observed during
2
10,45
respectively. The NH -TPD profile from each support is
the deoxygenation of phenol and m-cresol over Pd/ZrO2.
It
3
provided in Figure S3 in the Supporting Information and
has also been demonstrated that the presence of reducible
metal oxide supports increases the intrinsic deoxygenation
activity of noble metals regardless of their dispersion. While it
−1
indicates an acid site density of 53 μmol g for carbon and 490
μmol g− for TiO . The metal loading, carbon monoxide
1
45
2
2
717
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727

ACS Catalysis
Research Article
remains possible that particle size effects influence the reactions
in this report, their contributions are expected to be minimal.
activity may be affected by product inhibition at 35%
conversion, and the data in Figure 5 are not intended to
directly reflect intrinsic catalytic properties. Instead, they
provide a comparative assessment of the activity and selectivity
observed over each catalyst at a similar level of conversion.
Toluene was the dominant product for both catalysts under ex
situ CFP conditions, accounting for 78% and 46% of the carbon
2.2. Catalyst Testing. The carbon selectivity and turnover
frequency (TOF) for each catalyst is compared at 35% (±3%)
conversion in Figure 5. It is possible that the selectivity and
selectivity over Pt/TiO and Pt/C, respectively. Under HT
2
conditions, 3-methylcyclohexanone and 3-methylcyclohexanol
were highly favored and accounted for >95% of the carbon
selectivity. Methylcyclohexane was detected under both
conditions but was a minor product formed with <11% of
the carbon selectivity in all cases. These data are in agreement
with the thermodynamic concepts summarized by Ruddy et
6
al., which suggests that ring hydrogenation reactions are
favorable at low temperature (473−523 K) while other
reactions such as DDO and acid-catalyzed transformations are
6
activated above 523 K. In terms of TOF, both catalysts were
more active under HT conditions, and Pt/TiO exhibited more
2
than double the activity of Pt/C under comparable conditions.
The carbon selectivities and conversions at weight hourly
space velocities (WHSV) from 200 to 700 h− are presented in
Figure 6. The data shown for each catalyst and set of conditions
were collected during a single experiment and are based on an
average of two samples taken at 50 min increments. The
variance between the duplicate samples is not expected to be a
function of WHSV, and the error associated with these
1
Figure 5. Carbon selectivity and TOF for each catalyst compared at
3
5% conversion (±3%) under ex situ CFP (623 K, 0.5 MPa) and HT
(
523 K, 2.0 MPa) conditions.
Figure 6. Carbon selectivities (colored bars) and conversions (black circles) as a function of WHSV for (a) Pt/C−CFP, (b) Pt/C-HT, (c) Pt/TiO -
2
CFP, and (d) Pt/TiO -HT. Ex situ CFP conditions correspond to 623 K and 0.5 MPa, while HT conditions correspond to 523 K and 2.0 MPa.
2
2
718
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727

ACS Catalysis
Research Article
measurements was estimated by calculating the pooled standard
deviation of the six samples collected within a single
experiment. In all cases, the pooled standard deviation was
were calculated over an oxygen vacancy on the TiO (101)
surface.
2.3.1. Pt(111) Surface. The HDO mechanism over the
Pt(111) surface (Pt-HDO) is shown in Scheme 2a and results
2
<6%. The results of separate experiments shown in Figure S4 in
the Supporting Information suggest that both catalysts are
highly stable, and any effect of catalyst deactivation on the data
presented in Figure 6 is expected to be minimal. The selectivity
to 3-methylcyclohexanone increased as conversion decreased
for both catalysts and conditions. This result suggests that 3-
methylcyclohexanone is a primary product. This assignment is
consistent with a previous report in which 3-methylcyclohex-
anone was identified as a primary product during the
Scheme 2. (a) Pt-HDO, (b) Pt-TAU, and (c) Pt-TAU-HDO
Mechanisms over the Hydrogen-Covered Pt(111) Surface
a
deoxygenation of m-cresol over Pt/SiO at 573 K and
2
20
atmospheric pressure. Unlike the case for 3-methylcyclohex-
anone, the selectivity to 3-methylcyclohexanol increased with
increasing conversion under HT conditions. This result is
consistent with the sequential HDO reaction pathway shown in
Figure 2a, in which the aromatic ring is hydrogenated to form a
6
,48
ketone, followed by hydrogenation to a saturated alcohol.
The selectivity toward the deoxygenated products methyl-
cyclohexane and toluene was greater over Pt/TiO2 in
comparison to Pt/C under both conditions, particularly at
the highest conversion. For example, under HT conditions the
combined selectivity to methylcyclohexane and toluene at a
a
All energies are in eV and are referenced to R .
1
−1
WHSV of 200 h was 16 and 4.4 for Pt/TiO and Pt/C,
2
in the formation of 3-methylcyclohexanol (structure TH4),
which has a significantly weakened C−O bond energy relative
to m-cresol. The potential energy surface (PES) for the Pt(111)
reaction steps is shown in Figure S5 in the Supporting
Information. The barrier relative to the starting structure R1
over the hydrogen-covered Pt(111) surface is 0.69 eV for the
formation of 3-methylcyclohexenol, denoted RH -RH . The
respectively.
2.3. Calculated Mechanisms. To examine the aromatic
deoxygenation reaction mechanisms, calculations were per-
formed to identify energetically favorable pathways over the
Pt(111) and anatase TiO (101) surfaces (Scheme 1). The
2
1
2
Scheme 1. Reaction Pathways for m-Cresol Hydrogenation
and Deoxygenation That Were Computationally
RH structure can undergo tautomerization to the TH
2
3
structure. A similar tautomerization reaction, forming a slightly
different intermediate, would also be possible starting at RH1.
However, the difference in the barriers between RH -RH and
Investigated over Pt and TiO Surfaces
2
1
2
RH to TH is only 0.06 eV. Therefore, this change is unlikely
2
3
to affect the overall reaction. While a third hydrogenation step
is possible from RH , this reaction results in an endothermic
2
structure and would require rearrangement of the hydroxyl
group from the axial to the lower energy equatorial position.
The TH and TH structures are nearly degenerate, with a
3
4
negligible interconversion barrier. Therefore, the TH structure
3
can convert to the TH alcohol on the basis of hydrogen
4
availability and thermodynamic equilibrium. The final TH to
4
P2 barrier is the highest in this pathway (and all pathways in
this report) at 1.23 eV and is endothermic by 1.14 eV. An
alternative route for this C−O bond breaking is a reaction at a
defect, corner, edge, or step site on the Pt surface, which has
35,36
been demonstrated for Ru surfaces.
In addition, reactions
that involve the support may also be possible (vide infra). The
structures and reaction profiles for Pt-HDO are shown in
Figures S6−S10 in the Supporting Information.
The Pt-TAU pathway is presented in Scheme 2b. The first
step in this reaction is simultaneous hydrogen transfers from
the surface to the β-carbon (relative to the hydroxyl group) and
from the hydroxyl group to the Pt(111) surface. A schematic
for the enol to keto intermediate is shown in Figure 7. Stepwise
transfer of hydrogen between the β-carbon and the surface was
not found to be stable during optimization with the hydrogen-
covered surface. The highest barrier in this pathway is the
removal of the hydroxyl group by the Pt(111) surface, step T2-
P1, with a barrier of 1.13 eV. The first diffusion for this pathway
corresponds to a correction to account for differences in the
contributions from carbon are expected to be minimal, as it is
considered to be an inert support for deoxygenation reactions
under the conditions explored here. Pathways over the
Pt(111) surface include coadsorbed hydrogen for the ring
hydrogenation steps of the HDO and TAU mechanisms to
account for surface crowding. To explore the effect of the
reducible metal oxide support, the DDO and TAU mechanisms
54
2
719
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727

ACS Catalysis
Research Article
Scheme 3. (a) TiO -DDO (b) and TiO -TAU Mechanisms
2
2
a
over the TiO (101) Surface
2
Figure 7. Hydrogen shift for the tautomerization of m-cresol over a
hydrogen-covered Pt(111) surface.
hydrogen number between this state and the reference. This
correction corresponds to the addition of 1/2 H and the H-
2
defect formation energy, while the second corrects only for H-
defect formation present in the reactant. The relatively high C−
O bond dissociation barrier and endothermic reaction energy
are consistent with the low percentage of deoxygenation seen in
Figures 5 and 6 for the Pt/C catalyst. The structures and
reaction profiles for these steps are shown in Figures S11−S13
in the Supporting Information.
a
All energies are in eV and are referenced to R1.
The Ti-TAU mechanism can occur in the same manner as
the Pt-TAU mechanism. In the Ti-TAU mechanism, shown in
Scheme 3b, tautomerization occurs entirely over the TiO (101)
surface with the same three steps of the Pt-TAU mechanism:
namely, tautomerization, alcohol formation, and deoxygenation.
The first diffusion step, shown in Scheme 3b, T to T1a, is the
addition of a hydrogen atom to the surface with the energy of
the hydroxyl group removed. The highest barrier to
tautomerization over the TiO (101) surface is shown in
Scheme 3b, step T1a to T , with an energy of 0.95 eV and is
higher than the Pt(111) surface by 0.21 eV. The energy barrier
is essentially degenerate with the Pt-TAU-HDO mechanism.
2
The HDO mechanism is also possible starting from
intermediates in the tautomerization pathway denoted Pt-
1
1
TAU-HDO. In this pathway, after the initial tautomerization
1
step, the ring is hydrogenated, resulting in TH . As can be seen
2
in Scheme 2c, the highest barrier for the first ring hydro-
genation, step T -TH , is 1.09 eV. The second hydrogenation,
2
1
2
TH -TH , is endothermic by 0.22 eV relative to the reactants.
2
2
3
The deoxygenation barrier is high and exothermic as
mentioned above. The structures and reaction profiles for
these steps are shown in Figures S14 and S15 in the Supporting
Information.
The final step in Scheme 3b, T to P , is shown as a single step
2a
1
but occurs in two steps. Since the barrier of the second step is
below the reaction energy of the first step, only the barrier for
the first step is shown. The final step is the C−O bond
2
.3.2. TiO (101) Surface. Deoxygenation of m-cresol over
2
TiO (101) is hypothesized to be an inverse Mars−van Krevelen
2
44
breaking. Since this step starts from the reactive alcohol T , the
(
MvK) process at oxygen-vacancy defects. The experimental
2
barrier is 0.58 eV lower than for the aryl−oxygen bond of m-
cresol and 0.66 eV lower than the C−O bond breaking of 3-
methylcyclohexanol. The structures and reaction profiles for
these steps are shown in Figures S18−S21 in the Supporting
Information.
The reaction network and PES for the overall reaction is
shown in Figure 8. The reaction network indicates all structures
results reported here indicate that no activity is observed over
anatase TiO in the absence of Pt. This is likely due to
2
passivation of the oxygen vacancies during the initial phase of
the reaction. As such, a mechanism for regenerating the
4
4,55
vacancies must be present.
Spillover of hydrogen from the
55,56
metallic phase is the most likely source of hydrogen.
While
the barrier for spillover was not calculated due to uncertainty
associated with the morphology of the interface, the barriers for
hydrogen activation at high coverage over Pt(111) and
and barriers relative to the reactant R , and the surface state for
1
each set of reactions is indicated by color. The PES for the
entire reaction set indicates the most likely observed products
under HT conditions and the routes for deoxygenation under
CFP conditions.
hydrogen diffusion over the TiO (101) surfaces were explored.
2
The barrier for hydrogen dissociation is less than the
physisorption of molecular hydrogen on the Pt(111) surface
even at high coverage (7:8 ratio of hydrogen to surface sites).
3
. DISCUSSION
Diffusion of atomic hydrogen over the anatase TiO (101)
2
surface was calculated to be 0.6 eV. This value is in agreement
3.1. Reactions over Pt. The reaction network in Figure 8
with previous studies on the diffusion of hydrogen on rutile
suggests that the minimum energy pathway over Pt(111) leads
to formation of 3-methylcyclohexanone and 3-methylcyclohex-
anol via the Pt-HDO mechanism. This result is consistent with
the experimental results shown in Figures 5 and 6 collected
under HT conditions. Deoxygenation of 3-methylcyclohexanol
would not be expected over the Pt(111) surface, as it has a high
overall barrier of 1.23 eV. However, it is possible that
deoxygenation of the saturated alcohol to form methylcyclo-
hexane could occur at undercoordinated sites, as has been
55
TiO that reported a similar barrier of 0.56 eV. The PES for
2
the reactions of m-cresol over the TiO (101) surface is shown
2
in Figure S16 in the Supporting Information.
The inverse MvK mechanism for the DDO of m-cresol at the
oxygen vacancy sites (Ti-DDO) occurs by directly breaking the
3
8,44,57,58
aryl−oxygen bond as shown in Scheme 3a.
This
mechanism was investigated over the TiO (101) surface with a
2
single oxygen vacancy by directly breaking the aryl−oxygen
bond while eliminating the oxygen defect. The hydrogen
addition to the aryl carbon comes from either a surface
hydroxyl group or from intramolecular transfer from the
hydroxyl group. The reference state for this reaction contains
an oxygen defect. The structures and reaction profiles for these
steps are shown in Figure S17 in the Supporting Information.
35,36
previously demonstrated over Ru catalysts.
In addition to the reaction barriers, hydrogen availability and
surface crowding are important variables that may affect the
accessible reaction pathways. Previous studies predict that the
Pt(111) surface would be covered with a full monolayer (ML)
of hydrogen under HT conditions, while under ex situ CFP
2
720
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727

ACS Catalysis
Research Article
Figure 8. Reaction network (top) and PES (bottom) for reaction steps occurring over the Pt(111) metallic phase and the TiO (101) support. The
2
barrier heights are labeled with ‡, and energies without labels are reaction energies. All energies are in eV and are referenced to R1.
conditions the hydrogen coverage could vary from 1 to 0.5 ML
on the basis of the error in the calculated energies and
is low and the barriers are accessible at higher temperatures,
then the Pt-TAU mechanism may occur, resulting in
deoxygenation to form toluene. These findings are consistent
with the experimental results in Figures 5 and 6, which show
high selectivity to toluene under ex situ CFP conditions.
The interconversion between cycloalkanes and aromatics
over Pt catalysts is well documented, and it is possible that
some toluene is produced from the dehydrogenation of
59,60
uncertainty of entropic contributions.
The reduction in
hydrogen coverage under ex situ CFP conditions may reduce
the favorability of the Pt-HDO pathway due to its high
hydrogen requirement. This is supported by previous computa-
tional work in which the hydrogenation barriers of catechol
1
0
61
(
1.23 eV) and phenol (0.94 eV) over a Pt(111) surface with
2
8,79−81
a single hydrogen adsorbate were shown to be high and
endothermic. In this case, other reaction pathways over Pt(111)
may become significant. Included in these pathways is Pt-TAU,
shown in Scheme 2b and Figure 8. This mechanism is similar to
methylcyclohexane.
In order to understand how
thermodynamic limitations affect this process, the equilibrium
constant for the reaction of methylcyclohexane to toluene and
hydrogen was determined as a function of temperature, as
shown in Figure 9. At 523 K, methylcyclohexane is
thermodynamically favored, while at 623 K toluene becomes
thermodynamically favorable. The approach to equilibrium
based on the molar ratios of toluene to methylcyclohexane
observed during each experiment is presented in Table 2.
Values in excess of 1 are likely due to error associated with the
20
that proposed by Tan et al. in which m-cresol deoxygenation
proceeds through a tautomerization step over bare Pt and Ru
catalysts, resulting in the formation of 3-methylcyclohexanone.
In the current work, deoxygenation via the Pt-TAU mechanism
is higher in energy than ring hydrogenation through the Pt-
HDO mechanism by 0.43 eV. However, if hydrogen availability
2
721
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727

ACS Catalysis
Research Article
limited under HT conditions. However, under ex situ CFP
conditions, no thermodynamic constraint is imposed. The
formation of toluene via the Pt-HDO pathway under ex situ
CFP conditions is consistent with the results presented in
Figure 6, which show high selectivity to toluene coupled with
increasing selectivity to 3-methylcyclohexanone at lower
conversion, suggesting that it is a primary product. Con-
sequently, it is possible that both the Pt-HDO and Pt-TAU
pathways contribute to the production of toluene under ex situ
CFP conditions.
The tautomerization of m-cresol is a step in both the Pt-TAU
and Pt-TAU-HDO mechanisms, but the reaction mechanism
presented here differs significantly from previous computational
reports on Pt(111) and Fe(110) surfaces. The barrier resulting
from gas-phase intramolecular hydrogen transfer (no surface)
has been reported to be 2.29 eV. In a separate study on
61
Pt(111), in which the surface was included in the hydrogen
transfer step, the barrier was significantly reduced to 1.17 eV. A
reduction in the tautomerization barrier with surface partic-
Figure 9. Equilibrium constant for the reaction of methylcyclohexane
to give toluene and hydrogen plotted as a function of temperature.
The Keq values were determined by integrating the van’t Hoff equation
from the standard state to each temperature. Standard state enthalpies,
standard molar entropies, and heat capacity data as a function of
temperature fit to third-order polynomials were obtained from the
NIST. A log10 Keq value equal to 0 corresponds to a Keq value of 1.
20
ipation was also observed in the work by Tan et al. Thus,
surface participation in the hydrogen transfer reaction
significantly lowers the barrier. In the Pt-TAU mechanism
presented in this study, which includes both surface
participation and surface hydrogen, the tautomerization barrier
was found to be 0.74 eV. This low barrier indicates that the
presence of coadsorbed hydrogen would significantly enhance
the favorability of m-cresol tautomerization over Pt(111).
Table 2. Approach to Equilibrium Based on the Ratio of
Toluene to Methylcyclohexane Observed during Each
Experiment
3
.2. Reactions over TiO . The experimental results
2
approach to equilibrium
presented in Figure 5 show that Pt/TiO exhibits greater
2
catalyst/support, conditions
Pt/C, HT
200 h−
1
400/450 h−1
700 h−1
activity and higher selectivity to toluene than does Pt/C. The
increased TOF and enhanced selectivity toward deoxygenated
0.89
1.1
0.81
1.4
Pt/TiO , HT
0.67
1.1
products observed over Pt/TiO relative to Pt/C are consistent
2
2
Pt/C, CFP
0.034
0.049
0.022
0.047
0.029
0.067
with a previous report in which Ru/TiO exhibited activity and
2
Pt/TiO , CFP
selectivity to aromatic products greater than that of Ru/SiO2,
Ru/Al O , and Ru/C for guaiacol deoxygenation at atmos-
2
2
3
17
quantification of small amounts of toluene, which accounted for
1.0% of the carbon selectivity under HT conditions. At these
concentrations, small discrepancies in the analytical methods
can have a large effect on the resulting ratios. Despite the
uncertainties in this measurement, it is clear that the conversion
of methylcyclohexane to toluene may be thermodynamically
pheric pressure and 673 K. The Ru-based results were
<
attributed to synergistic metallic phase−support interactions in
which Ru facilitated the reduction of TiO , creating oxygen
2
vacancies that were highly active for deoxygenation. Similar
results have been reported for other metal/reducible oxide
44,45
catalysts and metal-free reducible oxide catalysts.
For
Figure 10. m-Cresol deoxygenation pathways accessible for Pt/C and Pt/TiO catalysts under HT (523 K, 2.0 MPa) and CFP conditions (623 K,
2
0.5 MPa) at high hydrogen coverage.
2
722
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727

ACS Catalysis
Research Article
example, during the deoxygenation of phenol at 573 K and
atmospheric pressure, benzene production was promoted over
Pd/ZrO , while Pd/Al O and Pd/SiO favored the formation
enhanced activity of Pt/TiO under HT conditions observed in
2
this work, as shown in Figure 5. Thus, it is possible that under
conditions which favor ring hydrogenation a more reducible
metal oxide could drive deoxygenation. However, a more
detailed theoretical and experimental understanding of the
active phase−support interface is necessary to determine if this
possibility is feasible.
2
2
3
2
of cyclohexanone. The increased benzene selectivity was
4
+
attributed to oxophilic Zr sites on the surface of the
1
0
support. Similarly, Schimming et al. correlated the conversion
of guaiacol with the concentration of oxygen vacancies over a
44
ceria−zirconia catalyst. Other calculations over 10-atom Ru
clusters with no hydrogen on a rutile h-TiO (110) indicate an
4. CONCLUSIONS
2
interesting alternate mechanism related to the amphoteric
The activity and selectivity observed during the deoxygenation
of m-cresol over Pt catalysts is dependent on the choice of
support and reaction conditions. Computational modeling
identified ring hydrogenation to 3-methylcyclohexanone and 3-
methylcyclohexanol as the most energetically favorable pathway
62
nature of the support.
These experimental data are supported by the computational
results presented in this report. Reaction pathways for DDO
and TAU shown in Scheme 3 over the TiO (101) surface
2
provide additional pathways for deoxygenation. The barriers for
these reactions are 1.06 eV for Ti-TAU (step T -T ) and 1.15
over Pt(111). Over TiO (101), tautomerization and direct
2
1
a
2
deoxygenation to toluene were identified as additional
energetically favorable routes. These findings are consistent
eV for Ti-DDO (step R -P ). These barriers are similar to
1
1
previous calculations at sulfur-defect sites in sulfided MoS and
2
with the experimental results, in which Pt/TiO was more
2
CoMoS catalysts, with reported barriers of 1.04 and 1.14 eV,
active on a metal site basis and exhibited higher selectivity to
toluene than did Pt/C. The influence of reaction conditions
was demonstrated by the increased selectivity to toluene
observed under ex situ CFP conditions (623 K, 0.5 MPa)
relative to HT conditions (523 K, 2.0 MPa), which is thought
to be associated with thermodynamic limitations imposed on
the conversion of cycloalkanes to aromatics under HT
conditions. The combined experimental−computational results
reported here highlight the synergistic effects between hydro-
genation catalysts and reducible metal oxide supports and
provide insight into the reaction pathways responsible for their
enhanced deoxygenation performance. Additional investiga-
tions into reactions at the metal−oxide interface, while beyond
the scope of this study, could provide significant insight and
further improvement in catalytic performance.
5
7
respectively. The Ti-DDO and Ti-TAU barriers are higher
than the Pt-HDO mechanism barrier (0.69 eV, step RH -RH )
1
2
proposed to occur over the metallic phase. These results
suggest that the reactivity of Pt/TiO and Pt/C is likely driven
2
by the metallic phase at low temperature (HT conditions), as
discussed above. At the higher temperature of ex situ CFP,
these barriers become accessible, resulting in increased
deoxygenation from the TiO support. It is important to note
2
that the barrier for oxygen vacancy formation via a spillover
mechanism was not directly calculated, and therefore it is not
possble to determine if the oxygen defect formation is the
highest barrier in this pathway. Our results suggest that the Ti-
DDO and Ti-TAU mechanisms occur if these defects are
present.
3
.3. Combined Reaction Mechanism. A reaction
mechanism is proposed in Figure 10 and suggests that the
5
. EXPERIMENTAL MATERIALS AND METHODS AND
COMPUTATIONAL METHODS AND MODELS
Pt-HDO pathway is accessible to Pt/TiO and Pt/C under
2
both conditions. Under ex situ CFP conditions, Pt can facilitate
deoxygenation to methylcyclohexane and dehydrogenation to
toluene. The Pt-TAU pathway may also be accessible under ex
situ CFP conditions and results in the production of toluene.
5
.1. Materials. TiO (Aeroxide P90) was obtained from
2
Evonik, and carbon (DARCO activated charcoal) was obtained
from Sigma-Aldrich. Both supports were used as received.
Pt(NH ) (NO ) was obtained from Strem Chemicals and
3
4
3 2
Additional pathways over the TiO -supported catalyst also
2
used as received. Crushed quartz (150−250 and 300−425 μm,
Powder Technologies Inc.) and silicon carbide (177−250 μm,
McMaster Carr) were used as diluent materials for reactor
testing. m-Cresol (99%) was obtained from Sigma-Aldrich and
used as received. Certified gas blends (5% argon/95% hydrogen
and 1% oxygen/99% helium) were obtained from Air Liquide.
Hydrogen (UHP, General Air), and nitrogen (house supply)
were also utilized.
contribute to the observed reactivity, especially under ex situ
CFP conditions. These include DDO and TAU, which result in
the production of toluene.
It is also possible that deoxygenation reactions can occur in
two steps with one occurring over the metallic phase and the
other occurring over the support, in anaology to ring
hydrogenation−dehydration over hydrogen-activating metals
6
on acidic supports: e.g., Pt/Al O . In this work the barrier for
2
3
5
.2. Catalyst Synthesis. Carbon- and TiO -supported Pt
2
DDO of 3-methylcyclohexanol at an oxygen vacancy on the
were prepared from Pt(NH ) (NO ) via standard incipient-
3
4
3 2
TiO (101) surface is 0.71 eV, as shown in Scheme 4. This
2
wetness impregnation methods. For Pt/TiO , 14.25 mL of an
2
possibility is consistent with a previous study in which the
aqueous solution of Pt(NH ) (NO ) corresponding to a 5.0
3
4
3 2
enhanced deoxygenation activity of Ru/TiO was attributed to
2
wt % metal loading was added dropwise to 9.5 g of TiO . For
17,38
2
an interface effect.
Similar effects may contribute to the
Pt/C, 23.75 mL of an aqueous solution of Pt(NH ) (NO )
3
4
3 2
corresponding to a 5.0 wt % metal loading was added dropwise
to 9.5 g of carbon. The impregnated materials were dried at 323
K overnight in an oven and reduced in the reactor at 723 K
under flowing molecular hydrogen. Catalyst samples used for
characterization were also reduced at 723 K prior to analysis.
These conditions were selected on the basis of literature values
and have been shown to achieve complete reduction of
Scheme 4. Intrinsic Barriers of 3-Methylcyclohexanol DDO
at Defect Sites in the TiO (101) Surface
2
63,64
supported Pt.
2
723
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727

ACS Catalysis
Research Article
5
.3. Catalyst Characterization. Powder XRD data were
geometry with no correction for surface wetting. The surface-
weighted average particle diameter was calculated using
collected using a Rigaku Ultima IV diffractometer with a Cu Kα
source (40 kV, 44 mA). Diffraction patterns were collected in
the 2θ range of 20−80° at a scan rate of 4° min . Samples
10−20 mg) were supported on a glass sample holder with a
.2 mm recessed sample area and were pressed into the
recession with a glass slide to obtain a uniform z axis height.
Data were compared to reference card files from the
International Center for Diffraction Data (Pt, 00-004-0802;
graphite, 00-008-0415; TiO -anatase, 00-001-0562, TiO -rutile,
3
2
∑D /∑D
and the volume-weighted average particle
P,i
P,i
−1
4
3
diameter was calculated using ∑D /∑D . High-resolution
P,i
P,i
(
0
TEM images of Pt/TiO were collected at 300 kV using a FEI
2
Tecnai super twin instrument equipped with a thermionic
emitter. Samples for high-resolution TEM were dispersed in
hexane and distributed on a Si window. Nitrogen physisorption
data were collected at 77 K using a Quantachrome Quadrasorb
SI instrument. Samples were pretreated under vacuum for 20 h
at 473 K. Surface areas were determined using the Brunauer−
2
2
0
0-001-1292) to confirm the identity and phase of the sample.
The average crystallite size, which is analogous to the volume-
weighted apparent particle size, was calculated from XRD peak
broadening of the supported catalysts using the Scherrer
equation on the basis of the Pt(111) diffraction.
Active site density and metal dispersion for Pt/C and Pt/
TiO were measured from CO pulse chemisorption performed
using an Altamira AMI-390 microflow reactor system equipped
with a thermal conductivity detector (TCD). Samples (ca. 180
mg) were loaded in a quartz U-tube reactor and heated under
0% H /Ar to 723 K at 5 K min with a hold time of 2 h. After
the reduction step, catalyst samples were flushed with He (50
mL min ) for 30 min to remove adsorbed hydrogen. The
samples were then cooled to 303 K and dosed with 500 μL
pulses of a 10% CO/He mixture. Saturation of the sample was
achieved when the peak areas of two consecutive pulses were
within ±4% and was generally within three to eight pulses. A
sample loop of known volume (500 μL corresponding to 1.65
μmol of CO under the operating conditions) was used to
calibrate the TCD response for CO. The site density was
determined by assuming a stoichiometry of one CO molecule
adsorbed per active site and used to calculate the average
particle size, which is analogous to the surface-weighted
Emmett−Teller (BET) method with a P/P range of 0.05−0.3.
0
5.4. Catalytic Testing. Approximately 0.02 g of catalyst was
diluted with 177−250 μm diameter particles of silicon carbide
to a volume of 4 mL and added to the isothermal zone of a 20
mL downward-flow tubular packed-bed reactor. The temper-
ature of the isothermal zone was monitored using a four-point
thermocouple inserted into the catalyst bed. Approximately 2
mL of 150−250 μm diameter crushed quartz was added to each
side of the catalyst bed, and the remainder of the reactor was
packed with 300−425 μm diameter crushed quartz. To begin
an experiment, a gas mixture of hydrogen (95%) and argon
(5%) was introduced at the desired WHSV, and the reactor
pressure was increased to the desired set point (0.50 or 2.0
MPa). Next, the isothermal zone was heated to the reaction
2
−1
1
2
−1
−1
temperature (523 or 623 K) at 5 K min . Once the reactor
temperature was stable, m-cresol was introduced from an Eldex
Optos 1LMP HPLC pump. An 8:1 molar ratio of molecular
hydrogen to m-cresol was maintained for the duration of the
reaction period, which is twice the stoichiometric requirement
for complete hydrogenation to methylcyclohexane and water.
Condensable products from the reaction were collected in a
hot trap controlled at 393 K. These products were analyzed
using an Agilent Technologies 7890A gas chromatograph
equipped with a flame ionization detector and mass
spectrometer. Uncondensed products in the gas phase were
analyzed online using a separate Agilent Technologies 7890B
gas chromatograph modified by Wasson ECE Instrumentation.
Products were identified through retention time comparison
with known standards and confirmed using mass spectrometry.
Quantitative analysis was carried out using flame ionization and
thermal conductivity detectors, which had been calibrated with
standards of known concentrations. In the event that a standard
was unavailable, response factors were extrapolated from similar
compounds on the basis of carbon number. Argon was used as
an internal standard for the gas-phase analysis, and all values
were adjusted to account for changes in total molar flow rate.
Gas chromatography was used to determine the molar
composition and formula weight of the condensed products.
The mass flow rate of the condensed products was determined
by dividing the sample mass by the collection time. Using the
mass flow rate and formula weight, the total molar flow rate of
the condensed phase was calculated. From the total molar flow
rate and the molar composition, the individual molar flow rate
of each product was determined. In all cases the mass balance
closure, based on theoretical inlet flowrates and measured
outlet flowrates, was >90%. Both catalysts were highly stable
and exhibited <3% change in activity, as determined by guaiacol
65
apparent particle size. Measurements carried out over carbon
and TiO in the absence of Pt exhibited no CO uptake.
The total number of acid sites was determined by NH3
temperature-programmed desorption (NH -TPD) using an
2
3
Altamira Instruments AMI-390 system with gas flow rates of 25
−1
mL min . Samples of each support (ca. 100 mg) were loaded
into a 0.5 in. quartz U-tube reactor and held as a fixed bed
between plugs of quartz wool. Oxide samples were pretreated
under flowing 10% O /Ar at 5 K min to 773 K and then held
at this temperature for 4 h. Carbon samples were pretreated
under flowing Ar at 5 K min to 773 K and then held at this
temperature for 4 h. The samples were cooled to 393 K under
flowing He and then saturated with flowing 10% NH /He for
0 min. Excess and/or physisorbed NH was removed by
−1
2
−1
3
3
3
holding the samples at 393 K under flowing He for 1 h. TPD of
NH was performed by heating the sample from 393 to 773 K
3
−1
at 30 K min and then holding for 30 min under flowing He.
Desorbed NH was measured with a thermal conductivity
detector, and calibration was performed after each experiment
by introducing 10 pulses of 10% NH /He from a 5.0 mL
3
3
sample loop into a stream of flowing He.
Galbraith Laboratories (Knoxville, TN) performed elemental
analysis using inductively coupled plasma optical emission
spectroscopy (ICP-OES). Samples for TEM were dropcast
onto carbon-coated copper grids (Ted Pella part no. 01824)
from chloroform or hexanes suspensions. Imaging was
performed using a FEI G20 Tecnai operating at 200 keV.
The particle size distributions were obtained by the manual
measurement of >100 particles using ImageJ software. The
average particle diameter was calculated by assuming spherical
conversion, during the reaction period. Carbon and TiO , in
2
the absence of Pt, exhibited no measurable m-cresol conversion
during experiments carried out at 623 K. To check for the
absence of external and internal mass transfer limitations, the
Weisz−Prater and Mears criteria were calculated. The resulting
2
724
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727

ACS Catalysis
Research Article
values were significantly below 0.1, indicating that mass transfer
limitations did not affect the observed reaction rates.
was optimized at a constant volume to find the optimal
parameters. Energies were calculated, including the Tkatchenko
76,77
The conversion of m-cresol was calculated according to eq 1
and is defined as the percent decrease in the molar flow rate of
m-cresol between the feed and outlet of the reactor. The carbon
selectivity was calculated according to eq 2 and is defined as the
molar flow rate of carbon contained in a product divided by the
total molar flow rate of carbon in all products multiplied by
and Scheffler dispersion correction, at U values of 2.5.
The
final values of a = 3.863 Å and c = 9.520 Å were close to the
experimental values of a = 3.782 Å and c = 9.502 Å. The
optimized lattice vectors in Å for the slab model were (10.2738,
0.0000, 0.0000 × 0.0000, 11.5904, 0.0000 × 0.0000, 0.0000,
33.5747).
1
00. The TOF is defined as the molecules of m-cresol
Previous work using ab initio thermodynamics and DFT
indicate that the Pt(111) surface would exhibit significant
converted per Pt active site as measured by CO chemisorption.
hydrogen coverage at the experimental temperatures and
m‐cresol conversion
59,60,78−80
pressures used in this study.
While the surface is
⎛
⎜
⎝
⎞
[F
]
[F
− [F
]
likely to contain a significant amount of adsorbed aromatic
species, this high-coverage effect may be modeled primarily by
m‐cresol feed
m‐cresol outlet
=
⎟ × 100
]
⎠
m‐cresol feed
(1)
81
coverage of a simple adsorbate such as hydrogen. For these
where Fm‑cresol = molar flow rate of m-cresol.
reasons, all of the calculations were started using a surface with
a complete hydrogen monolayer or a single surface vacancy in
the hydrogen monolayer, as shown in Figure 11 at the atop
⎛
⎝
⎞
FC
carbon selectivity = ⎜
⎟ × 100
82
⎜
n
i=1
⎟
sites. All Pt(111) models used in these calculations were
∑
FC
⎠
(2)
started with hydrogen atoms at atop sites.
n
i = 1
where F = molar flow rate of carbon in product i and ∑ (Fi)
i
=
total molar flow rate of carbon in all products.
5.5. Computational Methods. Calculations were per-
formed using the Vienna Ab initio Simulation Package (VASP)
6
6−69
5.3.
The periodic density functional calculations used the
projector augmented wave (PAW) potentials and an energy
cutoff of 400 eV for the plane wave basis set for all
7
0,71
72
calculations.
The Perdew−Burke−Ernzerhof (PBE)
generalized gradient corrected functional was used for all
periodic calculations with Monkhorst−Pack 5 × 5 × 1 k-point
sampling for metals and 3 × 3 × 1 k-point sampling for metal
oxides. First-order Methfessel−Paxton smearing with σ = 0.2
eV was used to determine partial occupancies for metallic
systems. For TiO , a small Gaussian smearing of σ = 0.01 eV
2
was used to assist in convergence. All reported total energies
are extrapolated to zero broadening. As TiO is a strongly
2
correlated material, i.e., one with highly localized electrons, the
on-site Coulomb and exchange interactions (Hubbard U
correction) were included using the simplified method of
Figure 11. Hydrogen-covered Pt(111) and anatase TiO (101) models
7
3
2
Dudarev et al. In this method, only the difference between the
on-site values is included. The difference used for these
calculations was U = 2.5 eV. This value was found to be optimal
for barriers and reaction energies for CO reacting on a
(side and top-down views) used in this study. The small white spheres
represent hydrogen atoms, the orange spheres platinum atoms, the
silver spheres titanium atoms, and the red spheres oxygen atoms.
74,75
TiO (101) surface by Sorescu et al.
Optimizations were
2
carried out until the forces were below 0.05 eV/Å for geometry
optimizations and below 0.07 eV/Å for nudged elastic band
calculations. All optimizations used the quasi-Newton method.
The van der Waals (vdW) forces were calculated using the
method of Tkatchenko and Scheffler as implemented in
The most common defect in the TiO (101) surface is an
2
55,83
oxygen vacancy.
These vacancies create undercoordinated
sites that are the most likely source of reactivity with
44
oxygenated aromatics. Oxygen vacancies in the TiO (101)
2
7
6,77
VASP.
.6. Computational Models. Slab models were used in the
lattice were formed by removing a two-coordinate oxygen atom
from the surface, as illustrated in Figure 11. This vacancy was
5
55,83
calculations of the Pt(111) and anatase TiO (101) surfaces. Pt
previously found to have the lowest formation energy.
The
2
slab models were formed from the optimized crystal unit cell
for platinum. The optimized lattice vectors for the slab were
energies were determined relative to the reactant in all cases.
Barriers were determined using nudged elastic band calcu-
lations with 7−11 images between the reactants and products.
Potential energy surfaces (PES) were made from combining
calculations of elementary steps. The energies were corrected
for intermediates by adding corrections to ensure each step had
the same number and type of atoms as well as the same number
of electrons. In most instances, the difference for the energy of
a structure calculated using corrected and uncorrected
references was below 0.05 eV. Cases where this was not true
correspond to differences in molecular configuration only.
(
0
11.2478, −5.6239, 0.0000 × 0.0000, 9.7409, 0.0000 × 0.0000,
.0000, 30.0000). The final vector allows for sufficient distance
between slabs to avoid unphysical interactions. The surface
model is five layers thick with the stoichiometry of Pt . The
80
upper three layers were relaxed, while the lower layers were
fixed at their ideal crystal positions. The TiO (101) crystals
2
were optimized by minimizing the energy as a function of
volume fit to the Birch−Murnaghan equation of state. After
volume optimization, the ratio of the unique lattice vectors c/a
2
725
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727

ACS Catalysis
ASSOCIATED CONTENT
Research Article
(
10) de Souza, P. M.; Rabelo-Neto, R. C.; Borges, L. E. P.; Jacobs, G.;
Davis, B. H.; Sooknoi, T.; Resasco, D. E.; Noronha, F. B. ACS Catal.
■
* Supporting Information
S
2
(
015, 5, 1318−1329.
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acscatal.5b02868.
11) Nie, L.; Resasco, D. E. J. Catal. 2014, 317, 22−29.
12) Lu, J. M.; Behtash, S.; Mamun, O.; Heyden, A. ACS Catal. 2015,
(
XRD patterns of carbon and Pt/C, micrograph of Pt/
5, 2423−2435.
13) Gutierrez, A.; Kaila, R. K.; Honkela, M. L.; Slioor, R.; Krause, A.
O. I. Catal. Today 2009, 147, 239−246.
14) Runnebaum, R. C.; Nimmanwudipong, T.; Block, D. E.; Gates,
B. C. Catal. Lett. 2011, 141, 817−820.
15) Nimmanwudipong, T.; Aydin, C.; Lu, J.; Runnebaum, R. C.;
(
TiO , NH -TPD profiles of carbon and TiO , time on
2
3
2
stream reaction data for Pt/TiO and Pt/C, and detailed
2
(
steps in the reaction pathways for m-cresol deoxygena-
tion over the TiO (101) and Pt(111) surfaces, including
2
(
energy profiles and important bond lengths and distances
Brodwater, K. C.; Browning, N. D.; Block, D. E.; Gates, B. C. Catal.
Lett. 2012, 142, 1190−1196.
(
PDF)
(16) Nimmanwudipong, T.; Runnebaum, R. C.; Block, D. E.; Gates,
AUTHOR INFORMATION
B. C. Energy Fuels 2011, 25, 3417−3427.
17) Boonyasuwat, S.; Omotoso, T.; Resasco, D. E.; Crossley, S. P.
■
(
Corresponding Authors
E-mail for G.T.B.: gregg.beckham@nrel.gov.
E-mail for J.A.S.: joshua.schaidle@nrel.gov.
Catal. Lett. 2013, 143, 783−791.
*
(
(
18) Elliott, D. C.; Hart, T. R. Energy Fuels 2009, 23, 631−637.
*
19) Nimmanwudipong, T.; Runnebaum, R. C.; Block, D. E.; Gates,
Author Contributions
These author contributed equally.
B. C. Catal. Lett. 2011, 141, 779−783.
†
(20) Tan, Q.; Wang, G.; Nie, L.; Dinse, A.; Buda, C.; Shabaker, J.;
Resasco, D. E. ACS Catal. 2015, 5, 6271−6283.
Notes
(21) Griffin, M.; Baddour, F.; Habas, S.; Ruddy, D.; Schaidle, J. Top.
The authors declare no competing financial interest.
Catal. 2016, 59, 124−137.
(
(
22) Furimsky, E. Appl. Catal., A 2000, 199, 147−190.
23) Foster, A. J.; Do, P. T. M.; Lobo, R. F. Top. Catal. 2012, 55,
ACKNOWLEDGMENTS
■
1
18−128.
24) Gao, D.; Schweitzer, C.; Hwang, H. T.; Varma, A. Ind. Eng.
Chem. Res. 2014, 53, 18658−18667.
25) Saidi, M.; Samimi, F.; Karimipourfard, D.; Nimmanwudipong,
T.; Gates, B. C.; Rahimpour, M. R. Energy Environ. Sci. 2014, 7, 103−
29.
26) Nie, L.; de Souza, P. M.; Noronha, F. B.; An, W.; Sooknoi, T.;
Resasco, D. E. J. Mol. Catal. A: Chem. 2014, 388-389, 47−55.
27) Li, Y.; Xu, X.; Zhang, P.; Gong, Y.; Li, H.; Wang, Y. RSC Adv.
013, 3, 10973−10982.
28) Zhu, X.; Nie, L.; Lobban, L. L.; Mallinson, R. G.; Resasco, D. E.
Energy Fuels 2014, 28, 4104−4111.
29) Massoth, F. E.; Politzer, P.; Concha, M. C.; Murray, J. S.;
This work was supported by the U.S. Department of Energy’s
Bioenergy Technologies Office, Contract No. DE-AC36-
(
0
8GO28308, at the National Renewable Energy Laboratory.
(
Computer time was provided by the Texas Advanced
Computing Center under the National Science Foundation
Extreme Science and Engineering Discovery Environment
Grant MCB-090159 and by the National Renewable Energy
Laboratory Computational Sciences Center. The authors thank
NREL researchers Mayank Behl (chemisorption measure-
ments), Matt Yung (chemisorption measurements), Susan
Habas (TEM imaging), Jeffery Aguiar (TEM imaging), and
Erick White (reactor operation). Helpful discussions with
Samuel Dull, Connor Nash, and Vassili Vorotnikov are also
gratefully acknowledged. The U.S. Government retains and the
publisher, by accepting the article for publication, acknowledges
that the U.S. Government retains a nonexclusive, paid up,
irrevocable, worldwide license to publish or reproduce the
published form of this work, or allow others to do so, for U.S.
Government purposes.
1
(
(
2
(
(
Jakowski, J.; Simons, J. J. Phys. Chem. B 2006, 110, 14283−14291.
(30) Runnebaum, R. C.; Lobo-Lapidus, R. J.; Nimmanwudipong, T.;
Block, D. E.; Gates, B. C. Energy Fuels 2011, 25, 4776−4785.
(31) Nimmanwudipong, T.; Runnebaum, R. C.; Ebeler, S. E.; Block,
D. E.; Gates, B. C. Catal. Lett. 2012, 142, 151−160.
(32) Li, K.; Wang, R.; Chen, J. Energy Fuels 2011, 25, 854−863.
(33) Runnebaum, R. C.; Nimmanwudipong, T.; Block, D. E.; Gates,
B. C. Catal. Sci. Technol. 2012, 2, 113−118.
34) Lee, K.; Gu, G. H.; Mullen, C. A.; Boateng, A. A.; Vlachos, D. G.
ChemSusChem 2015, 8, 315−322.
35) Chiu, C.-c.; Genest, A.; Borgna, A.; Roesch, N. ACS Catal. 2014,
, 4178−4188.
36) Chiu, C.-c.; Genest, A.; Borgna, A.; Roesch, N. Phys. Chem.
Chem. Phys. 2015, 17, 15324−15330.
37) Ghampson, I. T.; Sepulveda, C.; Garcia, R.; Fierro, J. L.;
Escalona, N.; DeSisto, W. J. Appl. Catal., A 2012, 435-436, 51−60.
38) Omotoso, T.; Boonyasuwat, S.; Crossley, S. P. Green Chem.
014, 16, 645−652.
39) Chang, J.; Danuthai, T.; Dewiyanti, S.; Wang, C.; Borgna, A.
ChemCatChem 2013, 5, 3041−3049.
40) Corpuz, A. R.; Pang, S. H.; Schoenbaum, C. A.; Medlin, J. W.
Langmuir 2014, 30, 14104−14110.
41) Nakagawa, Y.; Nakazawa, H.; Watanabe, H.; Tomishige, K.
ChemCatChem 2012, 4, 1791−1797.
42) Wan, H.; Chaudhari, R. V.; Subramaniam, B. Top. Catal. 2012,
(
REFERENCES
■
(
(
(
1) Czernik, S.; Bridgwater, A. V. Energy Fuels 2004, 18, 590−598.
2) Yaman, S. Energy Convers. Manage. 2004, 45, 651−671.
3) Mortensen, P. M.; Grunwaldt, J. D.; Jensen, P. A.; Knudsen, K.
G.; Jensen, A. D. Appl. Catal., A 2011, 407, 1−19.
4) Kersten, S.; Swaaij, W. v.; Lefferts, L.; Seshan, K. Catalysis for
Renewables: From Feedstock to Energy Production; Centi, G., van Santen,
R. A., Eds.; Wiley-VCH: Weinheim, Germany, 2007; pp 119−145.
5) Bridgwater, A. V. Environ. Prog. Sustainable Energy 2012, 31,
61−268.
6) Ruddy, D. A.; Schaidle, J. A.; Ferrell, J. R., III; Wang, J.; Moens,
L.; Hensley, J. E. Green Chem. 2014, 16, 454−490.
7) Dutta, A.; Schaidle, J.; Humbird, D.; Baddour, F.; Sahir, A. Top.
Catal. 2016, 59, 2−18.
8) Venkatakrishnan, V. K.; Delgass, W. N.; Ribeiro, F. H.; Agrawal,
R. Green Chem. 2015, 17, 178−183.
9) Talmadge, M. S.; Baldwin, R. M.; Biddy, M. J.; McCormick, R. L.;
(
4
(
(
(
(
2
(
(
2
(
(
(
(
(
(
(
Beckham, G. T.; Ferguson, G. A.; Czernik, S.; Magrini-Bair, K. A.;
Foust, T. D.; Metelski, P. D.; Hetrick, C.; Nimlos, M. R. Green Chem.
014, 16, 407−453.
55, 129−139.
(43) Zhao, H. Y.; Li, D.; Bui, P.; Oyama, S. T. Appl. Catal., A 2011,
391, 305−310.
2
2
726
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727

ACS Catalysis
Research Article
(
44) Schimming, S. M.; LaMont, O. D.; Koenig, M.; Rogers, A. K.;
D’Amico, A. D.; Yung, M. M.; Sievers, C. ChemSusChem 2015, 8,
073−2083.
45) de Souza, P. M.; Nie, L.; Borges, L. E. P.; Noronha, F. B.;
Resasco, D. E. Catal. Lett. 2014, 144, 2005−2011.
46) Hensley, A. J. R.; Wang, Y.; McEwen, J.-S. ACS Catal. 2015, 5,
23−536.
47) Sun, J.; Karim, A. M.; Zhang, H.; Kovarik, L.; Li, X. S.; Hensley,
A. J.; McEwen, J.-S.; Wang, Y. J. Catal. 2013, 306, 47−57.
48) Zhao, C.; He, J.; Lemonidou, A. A.; Li, X.; Lercher, J. A. J. Catal.
011, 280, 8−16.
49) Do, P. T. M.; Foster, A. J.; Chen, J.; Lobo, R. F. Green Chem.
012, 14, 1388−1397.
50) Zhu, X.; Lobban, L. L.; Mallinson, R. G.; Resasco, D. E. J. Catal.
011, 281, 21−29.
51) Krupski, K.; Moors, M.; Joz
Materials 2015, 8, 2935.
52) Shao, M. H.; Peles, A.; Shoemaker, K. Nano Lett. 2011, 11,
714−3719.
53) Schweitzer, N. M.; Schaidle, J. A.; Ezekoye, O. K.; Pan, X.; Linic,
(78) Jin, H.; Wang, L.; Searles, D. J.; Sun, C. Chin. Sci. Bull. 2014, 59,
2175−2180.
2
(
(79) Neurock, M.; Pallassana, V.; van Santen, R. A. J. Am. Chem. Soc.
2000, 122, 1150−1153.
(80) Wang, S.; Vorotnikov, V.; Vlachos, D. G. ACS Catal. 2015, 5,
104−112.
(
5
(
(81) Burkholder, L.; Stacchiola, D.; Tysoe, W. Surf. Rev. Lett. 2003,
10, 909−916.
(82) Saeys, M.; Reyniers, M. F.; Neurock, M.; Marin, G. B. J. Phys.
Chem. B 2005, 109, 2064−2073.
(
2
(
2
(
2
(
(83) Vittadini, A.; Casarin, M.; Selloni, A. Theor. Chem. Acc. 2007,
117, 663−671.
́
́
wik, P.; Kobiela, T.; Krupski, A.
(
3
(
S.; Thompson, L. T. J. Am. Chem. Soc. 2011, 133, 2378−2381.
(
54) Lam, E.; Luong, J. H. T. ACS Catal. 2014, 4, 3393−3410.
(
55) Conradie, J.; Gracia, J.; Niemantsverdriet, J. W. J. Phys. Chem. C
2
(
(
012, 116, 25362−25367.
56) Prins, R. Chem. Rev. 2012, 112, 2714−2738.
57) Badawi, M.; Paul, J. F.; Payen, E.; Romero, Y.; Richard, F.;
Brunet, S.; Popov, A.; Kondratieva, E.; Gilson, J. P.; Mariey, L.;
Travert, A.; Mauge, F. Oil Gas Sci. Technol. 2013, 68, 829−840.
(
58) Bui, V. N.; Toussaint, G.; Laurenti, D.; Mirodatos, C.; Geantet,
C. Catal. Today 2009, 143, 172−178.
59) Chizallet, C.; Bonnard, G.; Krebs, E.; Bisson, L.; Thomazeau, C.;
Raybaud, P. J. Phys. Chem. C 2011, 115, 12135−12149.
60) Chizallet, C.; Digne, M.; Arrouvel, C.; Raybaud, P.; Delbecq, F.;
Costentin, G.; Che, M.; Sautet, P.; Toulhoat, H. Top. Catal. 2009, 52,
005−1016.
61) Li, G.; Han, J.; Wang, H.; Zhu, X.; Ge, Q. ACS Catal. 2015, 5,
009−2016.
62) Nelson, R. C.; Baek, B.; Ruiz, P.; Goundie, B.; Brooks, A.;
(
(
1
(
2
(
Wheeler, M. C.; Frederick, B. G.; Grabow, L. C.; Austin, R. N. ACS
Catal. 2015, 5, 6509−6523.
(
63) Romanmartinez, M. C.; Cazorlaamoros, D.; Linaressolano, A.;
Delecea, C. S. Carbon 1993, 31, 895−902.
64) Ebitani, K.; Hattori, H. Bull. Chem. Soc. Jpn. 1991, 64, 2422−
427.
65) Karakaya, C.; Deutschmann, O. Appl. Catal., A 2012, 445-446,
21−230.
66) Kresse, G.; Furthmuller, J. Phys. Rev. B: Condens. Matter Mater.
(
2
(
2
(
Phys. 1996, 54, 11169−11186.
(
67) Kresse, G.; Furthmuller, J. Comput. Mater. Sci. 1996, 6, 15−50.
(
68) Kresse, G.; Hafner, J. Phys. Rev. B: Condens. Matter Mater. Phys.
1
(
1
(
1
(
1
(
993, 47, 558−561.
69) Kresse, G.; Hafner, J. Phys. Rev. B: Condens. Matter Mater. Phys.
994, 49, 14251−14269.
70) Blochl, P. E. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50,
7953−17979.
71) Kresse, G.; Joubert, D. Phys. Rev. B: Condens. Matter Mater. Phys.
999, 59, 1758−1775.
72) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77,
865−3868.
73) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.;
Sutton, A. P. Phys. Rev. B: Condens. Matter Mater. Phys. 1998, 57,
505−1509.
74) Sorescu, D. C.; Al-Saidi, W. A.; Jordan, K. D. J. Chem. Phys.
011, 135, 124701.
75) Sorescu, D. C.; Civis, S.; Jordan, K. D. J. Phys. Chem. C 2014,
18, 1628−1639.
76) Tkatchenko, A.; DiStasio, R. A.; Car, R.; Scheffler, M. Phys. Rev.
Lett. 2012, 108, 236402.
77) Tkatchenko, A.; Scheffler, M. Phys. Rev. Lett. 2009, 102, 073005.
3
(
1
(
2
(
1
(
(
2
727
DOI: 10.1021/acscatal.5b02868
ACS Catal. 2016, 6, 2715−2727