Multiproduct steady-state isotopic transient kinetic analysis of the ethanol coupling reaction over hydroxyapatite and magnesia

Article abstract of DOI:10.1021/cs502023g

The Guerbet coupling of ethanol into butanol was investigated using multiproduct steady-state isotopic transient kinetic analysis (SSITKA) in a comparative study between stoichiometric hydroxyapatite (HAP) and magnesia (MgO) catalysts at 613 and 653 K, respectively. The steady-state catalytic reactions were conducted in a gas-phase, fixed-bed, differential reactor at 1.3 atm total system pressure. Multiproduct SSITKA results showed that the mean surface residence time of reactive intermediates leading to acetaldehyde was significantly shorter than that of intermediates leading to butanol on both HAP and MgO. This finding may suggest that the dehydrogenation of ethanol to acetaldehyde is fast on these surfaces compared with C-C bond formation. If adsorbed acetaldehyde is a key reaction intermediate in the Guerbet coupling of ethanol into butanol, then SSITKA revealed that the majority of adsorbed acetaldehyde produced on the surface of MgO desorbs into the gas-phase, whereas the majority of adsorbed acetaldehyde on HAP likely undergoes sequential aldol-type reactions required for butanol formation. Adsorption microcalorimetry of triethylamine and CO₂ showed a significantly higher number of acid and base sites on the surface of HAP compared with those on MgO. Diffuse reflectance infrared Fourier transform spectroscopy of adsorbed ethanol followed by stepwise temperature-programmed desorption revealed that ethoxide is more weakly bound to the HAP surface compared with MgO. A high surface density of acid-base site pairs along with a weak binding affinity for ethanol on HAP may provide a possible explanation for the increased activity and high butanol selectivity observed with HAP compared with MgO catalysts in the ethanol coupling reaction.

Full text of DOI:10.1021/cs502023g

Research Article
pubs.acs.org/acscatalysis
Multiproduct Steady-State Isotopic Transient Kinetic Analysis of the
Ethanol Coupling Reaction over Hydroxyapatite and Magnesia
Sabra Hanspal, Zachary D. Young, Heng Shou,† and Robert J. Davis*
Department of Chemical Engineering, University of Virginia, 102 Engineers’ Way, Charlottesville, Virginia 22904-4741, United States
S
* Supporting Information
ABSTRACT: The Guerbet coupling of ethanol into butanol
was investigated using multiproduct steady-state isotopic
transient kinetic analysis (SSITKA) in a comparative study
between stoichiometric hydroxyapatite (HAP) and magnesia
(MgO) catalysts at 613 and 653 K, respectively. The steady-state
catalytic reactions were conducted in a gas-phase, fixed-bed,
differential reactor at 1.3 atm total system pressure. Multi-
product SSITKA results showed that the mean surface residence
time of reactive intermediates leading to acetaldehyde was
significantly shorter than that of intermediates leading to butanol
on both HAP and MgO. This finding may suggest that the dehydrogenation of ethanol to acetaldehyde is fast on these surfaces
compared with C−C bond formation. If adsorbed acetaldehyde is a key reaction intermediate in the Guerbet coupling of ethanol
into butanol, then SSITKA revealed that the majority of adsorbed acetaldehyde produced on the surface of MgO desorbs into the
gas-phase, whereas the majority of adsorbed acetaldehyde on HAP likely undergoes sequential aldol-type reactions required for
butanol formation. Adsorption microcalorimetry of triethylamine and CO2 showed a significantly higher number of acid and base
sites on the surface of HAP compared with those on MgO. Diffuse reflectance infrared Fourier transform spectroscopy of
adsorbed ethanol followed by stepwise temperature-programmed desorption revealed that ethoxide is more weakly bound to the
HAP surface compared with MgO. A high surface density of acid−base site pairs along with a weak binding affinity for ethanol on
HAP may provide a possible explanation for the increased activity and high butanol selectivity observed with HAP compared
with MgO catalysts in the ethanol coupling reaction.
KEYWORDS: hydroxyapatite, magnesia, Guerbet reaction, ethanol, butanol, microcalorimetry, DRIFTS,
steady-state isotopic transient kinetic analysis (SSITKA), acid−base bifunctional catalyst
to form acetaldehyde, which then undergoes an aldol
condensation reaction, followed by hydrogenation of the
resulting unsaturated condensation products to give butanol.5−7
Alternative reaction pathways for ethanol coupling have been
proposed by researchers in the field, primarily those involving a
“direct” condensation route without the participation of gas-
phase acetaldehyde.3,8−11 This “direct” bimolecular path was
verified by Iglesia and co-workers through studies of 13C-
labeled acetaldehyde reacting with ethanol over K−Cu/
MgCeOx catalysts at short contact times, which resulted in a
portion of the coupled products being unlabeled, suggesting
that ethanol coupling can proceed without the presence of gas-
phase acetaldehyde.10 Recently, Scalbert et al.11 investigated the
role of acetaldehyde self-aldolization in the ethanol coupling
reaction over a hydroxyapatite catalyst using a thermodynamic
approach and also found that the reaction primarily proceeds
through a “direct” biomolecular route and that aldol
condensation of acetaldehyde is a minor pathway in the
sequence leading to butanol. The details regarding the “direct”
1. INTRODUCTION
The use of corn-based bioethanol as a domestic and renewable
transportation fuel alternative has led to a growth of ethanol
production in the United States. This increased availability
combined with several drawbacks posed by ethanol as a blend
fuel has called attention to its catalytic transformation into a
higher-value fuel or chemical such as butanol.1 Butanol is an
important industrial chemical and has recently generated
interest as a potential gasoline fuel additive due to its higher
energy density relative to ethanol and, thus, ability to improve
fuel economy for biofuel blended gasoline.1,2
Butanol is commonly synthesized by hydroformylation via
the oxo process. This method uses fossil fuel-derived feedstocks
and often requires expensive catalytic materials and extreme
reaction pressures, leading to high energy input and substantial
operating costs.3,4 An alternative route to butanol synthesis is
the catalytic conversion of bioderived ethanol via the so-called
Guerbet reaction.
The Guerbet reaction is a well-known industrial route for
higher alcohol synthesis that ultimately couples two short-chain
alcohols to produce a longer-chain saturated alcohol. The most
commonly accepted reaction sequence for the Guerbet
coupling of ethanol involves an initial dehydrogenation step
Received: December 16, 2014
Revised: February 2, 2015
© XXXX American Chemical Society
1737
DOI: 10.1021/cs502023g
ACS Catal. 2015, 5, 1737−1746

ACS Catalysis
Research Article
condensation route, as to whether the reaction mechanism
involves the direct self-coupling of two ethanol molecules or
the condensation of ethanol and acetaldehyde, remain unclear,
and therefore, mechanistic studies are still ongoing. For a more
comprehensive discussion of Guerbet chemistry, including
details regarding the reaction mechanism, please refer to the
recent review paper by Kozlowski and Davis.12
The ethanol coupling reaction has been extensively studied
over a wide range of catalytic materials, such as solid base metal
oxides,13,14 both unpromoted and modified by transition metal
or alkaline earth metal components;9,10 hydrotalcite-derived
Mg−Al mixed oxides;15,16 and alkali metal ion-exchanged
zeolites.8,17 These catalysts typically require high reaction
temperatures (>673 K) and produce butanol with poor
selectivities at relatively low rates. Recent studies have
demonstrated unusually high activity and high butanol
selectivity for the Guerbet coupling of ethanol over
hydroxyapatite catalysts.18,19
The stoichiometric form of hydroxyapatite (HAP,
Ca10(PO4)6(OH)2) has a Ca/P molar ratio equal to 1.67;
however, this ratio can vary substantially in synthetic materials.
The surfaces of HAP can be acidic or basic in nature, depending
on the Ca/P molar ratio of the material, which in turn
influences the catalytic behavior of HAP.20−22 The active sites
on these materials have not been identified, and therefore, a
detailed understanding of the reaction mechanism is lacking. In
this study, we used steady-state isotopic transient kinetic
analysis (SSITKA) of the ethanol coupling reaction over a
stoichiometric calcium hydroxyapatite catalyst and compared
the results with those obtained with MgO, a standard solid base
catalyst, in an effort to gain insight into the reaction mechanism
at the active site level.
The SSITKA technique is a well-established and powerful
method that allows quantification of important kinetic
parameters, such as surface coverages of adsorbed reactant
species and surface reaction intermediates, average surface
lifetimes of those intermediates, and an upper bound of the
turnover frequency.23−25 For additional details regarding the
SSITKA technique, see the comprehensive review by Shannon
and Goodwin.26 Recently, our group studied the gas-phase
conversion of ethanol to butanol over MgO at 673 K using
SSITKA; however, acetaldehyde could not be followed as an
intermediate with the reaction system.27 In the present work,
we have performed a comparative study between stoichiometric
hydroxyapatite (HAP) and MgO, using a modified SSITKA
reactor system that allows the monitoring of multiple products
formed during the reaction.
200 cm3 of 0.3 M diammonium phosphate ((NH4)2HPO4,
Aldrich, >99.99%). These compositions corresponded to a
stoichiometric molar ratio of Ca/P (1.67) in the resulting
mixed solution. Both solutions, previously adjusted with
aqueous ammonia to pH = 10.5, were simultaneously added
dropwise to 100 cm3 of distilled deionized water (DDI) at 353
K. Sufficient aqueous ammonia was added continuously during
precipitation to maintain a pH of 10.5. The resulting
suspension was stirred for 24 h at 353 K. The precipitate was
recovered by vacuum filtration, washed 3 times with DDI water,
and dried in air at 400 K overnight.
High-purity and ultrafine (500 Å) MgO was obtained
commercially (Ube Material Industries, Ltd.). The MgO and
HAP powders were calcined at 873 K for 2 h in flowing air
using a 10 K min−1 ramp rate, then pressed, crushed, and sieved
into pellets of 106−180 μm.
2.2. Catalyst Characterization. Crystalline phases of the
catalysts were confirmed by powder X-ray diffraction (XRD) on
a PANalytical X’Pert Pro diffractometer using monochromatic
Cu Kα radiation (λ = 1.54 Å). Scans were collected at 2θ = 20−
90° with a 0.05° step size.
Specific surfaces areas were obtained by N2 adsorption
measured at 77 K using the BET method on a Micromeritics
ASAP 2020 automated analyzer.
Elemental analysis of the HAP catalyst was performed by
Galbraith Laboratories (Knoxville, TN) using inductively
coupled plasma optical emission spectroscopy (ICP-OES) for
calcium and phosphorus content in the bulk material. The
chemical composition on the surface of the HAP sample was
analyzed by X-ray photoelectron spectroscopy (XPS) using a
ThermoFisher ESCALab 250 apparatus. The signals were
referenced to adventitious carbon (C(1s)) at a binding energy
of 285.09 eV.
2.3. Ethanol Coupling Reactions. Guerbet coupling of
ethanol was performed in a downward flow, fixed-bed, stainless
steel, tubular reactor (i.d.: 0.46 cm) at 1.3 atm total system
pressure. The catalyst pellets (0.2 g of MgO or 0.063 g of HAP)
rested upon a packed region of quartz wool in the reactor tube
with a thermocouple positioned at the center of the catalytic
bed. Prior to reaction, the catalysts were heated in situ to 773 K
at 10 K min−1 in flowing He (50 cm3 min−1) and held at 773 K
for 1 h. The reactor effluent was analyzed by an online SRI
8610C gas chromatograph equipped with a flame ionization
detector (FID). Reactants and products were quantified using a
Restek MXT-Q-Bond column (0.53 mm i.d., 30 m length)
connected to the FID.
The peak areas of reactants and products identified by GC
were used to determine the ethanol conversion and selectivity
of products. The conversion of ethanol was calculated as
follows:
To relate reactivity results to catalyst properties, adsorption
sites on the catalysts were characterized using diffuse
reflectance FT-IR spectroscopy (DRIFTS) during stepwise
temperature-programmed desorption (STPD) of adsorbed
ethanol as well as adsorption microcalorimetry of carbon
dioxide, triethylamine, and ethanol. Results from surface
characterization, reactivity testing, and isotopic transient studies
were used to propose key structural and compositional
properties that facilitate the Guerbet coupling reaction.
⎛
⎜
⎝
⎞
⎟
⎠
ΣniMi
2
conversion (C%) =
× 100
where ni is the number of carbon atoms in product i and Mi is
the molar ratio of product i detected to the initial moles of
ethanol.
The selectivity toward product i was calculated on the basis
of the total number of carbon atoms in the product and is
therefore defined as
2. EXPERIMENTAL METHODS
2.1. Catalyst Preparation. The calcium hydroxyapatite
catalyst was prepared using a controlled coprecipitation method
based on the procedure described by Tsuchida et al.28 First, two
aqueous solutions were prepared: 200 cm3 of 0.5 M calcium
nitrate tetrahydrate (Ca(NO3)2·4H2O, Acros Organics) and
niMi
selectivity (C%) =
× 100
ΣniMi
1738
DOI: 10.1021/cs502023g
ACS Catal. 2015, 5, 1737−1746

ACS Catalysis
Research Article
2.4. Multiproduct SSITKA. A schematic of the reactor
system used for the multiproduct SSITKA experiments is
shown in Figure 1. Details regarding the system were described
tion and overlapping issues that are often associated with mass
spectrometric (MS) analysis of higher alcohols, the gas samples
were passed to a hydrogenolysis reactor held at 673 K and
converted into methane over 5 wt % Pt/Al2O3 (Sigma-Aldrich)
after separation by GC and prior to entering the MS (Pfeiffer
Vacuum). In the MS, the ion signals for m/z = 15 (12CH4) and
17 (13CH4) were continuously monitored to determine the
isotope content of the original gas sample. On the basis of the
fragmentation pattern for methane, a portion of the 12CH4
signal is attributed to 13CH4; therefore, the MS 12CH4
responses were all corrected by subtracting the 13CH4
contribution.
Figure 2 shows an example set of normalized transient
response curves for Ar, acetaldehyde, butanol, and ethanol that
Figure 1. Reaction system for multiproduct SSITKA. This figure is
adapted from Shou and Davis.29
in previous work from our group,29 so only a brief description is
provided here. After achieving steady-state conversion (mini-
mum of 16 h on stream), an isotopic switch was performed
using a Valco 2-postion pneumatic valve, from unlabeled 12C
ethanol (anhydrous, Sigma-Aldrich, 99.5%) to doubly labeled
13C ethanol (Cambridge Isotopes Laboratories, Inc.; 1,2-13C,
99%). The unlabeled/labeled ethanol were each contained in
two identical saturators that were submerged in a heated water
bath maintained at 299 K. The 13C-labeled ethanol was received
with a substantial amount of water (5.89 wt %); therefore, 3A
molecular sieves (Sigma-Aldrich) that were previously treated
at 523 K for several hours in flowing He, were added to the
saturator to dehydrate the ethanol. Molecular sieves were also
added to the unlabeled ethanol so that the liquid level in both
saturators was the same.
Helium (GTS-Welco, 99.999%) was used as the carrier gas
and flowed through the saturators with a mole fraction of
ethanol in the gas phase equal to 6.2%. The 12C ethanol gas
feed was mixed with an inert argon (GTS-Welco, 99.999%)
tracer (2 vol % of the total flow rate) that was used to correct
for the gas-phase holdup in the reactor. The SSTIKA
experiments were conducted at three different total gas flow
rates (30, 50, and 75 cm3 min−1) to investigate reactant and
product readsorption effects.
Figure 2. Normalized isotopic transient response curves following the
switch from unlabeled ethanol to doubly labeled 13C-labeled ethanol
with a total flow of 50 cm3 min−1 at 613 K during the coupling of
ethanol over HAP: ( ) argon, ( ) acetaldehyde, ( ) ethanol, (
butanol.
⧫
■
●
▲
)
were obtained following the isotopic switch from 12C ethanol to
13C ethanol during the steady-state reaction of ethanol over
HAP at 613 K at a total flow rate of 50 cm3 min−1. Transient
responses were normalized by the difference between the initial
and final ion signals. The argon decay curve was used to
determine the gas-phase holdup of the reactor system since we
assumed that the inert gas did not adsorb or react on the
surface of the catalyst. Therefore, the difference in area under
the normalized transient response of each species (Fi) from that
of the inert Ar tracer (FAr) is equal to the overall mean surface
residence time associated with that species (τi):
τ = ∞ (F − FAr) dt
∫
i
i
0
The surface coverage of reactant or reactive intermediates
that lead to a specific product (Ni) can then be determined as
follows:
The 12C/13C ethanol switch was achieved without disrupting
the steady-state of the reaction by maintaining the reaction
temperature as well as the total system pressure at 1.3 atm, with
the use of two back pressure regulators positioned at the end of
the reactor and the vent line. Following the isotope switch, 16
gas samples of the reactor effluent were collected at various
time intervals throughout the transient, using an automated
Valco 34-port sampling valve. The samples were injected and
separated by gas chromatography (GC). To avoid fragmenta-
N = Ri × τ
i
i
where Ri refers to the steady-state flow rate of reactant or the
reaction rate to form product, i.
2.5. Adsorption Microcalorimetry. Adsorption sites on
HAP and MgO were characterized using adsorption micro-
calorimetry of carbon dioxide, triethylamine (TEA), and
ethanol. The experiments were conducted at 303 K using a
1739
DOI: 10.1021/cs502023g
ACS Catal. 2015, 5, 1737−1746

ACS Catalysis
Research Article
heat-flow microcalorimeter. Experimental procedures27 as well
as a detailed description of the apparatus30 used for the
calorimetry measurements have been reported in previous
work.
In summary, samples were first outgassed at 773 K for 16 h
under vacuum to a pressure below 10−3 Torr then cooled to
room temperature. Prior to adsorption, the pretreated sample
cell was inserted into an isothermal heat block (maintained at
303 K) for 2 h and allowed to thermally equilibrate with the
system. Incremental doses of the gas probe molecule (carbon
dioxide, triethylamine, ethanol) were introduced to the catalyst
via a volumetric dosing system. Liquid TEA and ethanol were
purified by several freeze−pump−thaw cycles prior to use.
Adsorption isotherms and differential enthalpies of adsorption
were obtained by measuring the amount of adsorbed species on
the catalytic surface and the heats evolved for each dose.
2.6. Diffuse Reflectance Infrared Fourier Transform
Spectroscopy (DRIFTS). Stepwise temperature-programmed
desorption (STPD) of adsorbed ethanol was investigated in the
diffuse reflectance mode on a Bio-Rad (FTS-60A) FTIR
spectrometer equipped with a liquid-nitrogen-cooled MCT
detector. The DRIFTS experiments were conducted using a
high-temperature gas reaction chamber (Harrick Scientific)
positioned onto a Praying Mantis diffuse reflectance sample
accessory. All spectra were obtained by coadding and averaging
100 scans at a spectral resolution of 4 cm−1.
Figure 3. X-ray diffraction patterns of (a) MgO and (b) stoichiometric
HAP (catalysts were calcined at 873 K for 2 h in air). Patterns are
offset for clarity.
Table 1. Specific Surface Areas of Stoichiometric
Hydroxyapatite and Magnesia Catalysts
catalyst
BET surface area (m2 g−1
)
HAP
MgO
35
35
The STPD measurements were carried out according to the
experimental procedure described in detail in previous work by
Birky et al.27 Catalyst samples, diluted in KBr powder at 1 wt %
HAP and 5 wt % MgO, were loaded into the DRIFTS cell and
pretreated in situ at 773 K for 1 h in flowing He (30 cm3
min−1). The DRIFTS cell was exposed to anhydrous ethanol
(Sigma-Aldrich) at 303 K for 15 min by passing He (30 cm3
min−1) through an ethanol saturator followed by purging under
He flow (30 cm3 min−1) for 15 min. The catalyst sample was
then heated stepwise to 673 at 10 K min−1 with IR spectra
collected after waiting 15 min at each temperature. The
DRIFTS spectrum of the catalyst sample taken at each
temperature prior to ethanol adsorption was used as back-
ground for each measurement.
Table 2. Product Distribution during the Catalytic
Conversion of Ethanol over MgO at 653 K
selectivity (C%)
total flow
rate
ethanol
conv
(%)
rate ethanol
conv
(cm3 min−1
)
(mol m−2 s−1
)
ethene acetaldehyde butanol
30
50
75
7.9
4.5
3.7
1.5 × 10−8
1.4 × 10−8
1.7 × 10−8
11
12
13
49
67
67
40
21
20
Catalyst loading: 0.2 g; T = 653 K; total system pressure = 1.3 atm.
conversion (<5%). As conversion increased, the product
distribution shifted toward the coupled product, butanol, as
expected for a sequential reaction network in which
acetaldehyde is a reaction intermediate in the conversion of
ethanol into butanol. The dehydration reaction of ethanol to
ethene, which is an undesired side reaction, was also observed
over MgO at 653 K. The ethene selectivity remained constant
(∼10%) at the reaction conditions tested. Constant ethene
selectivity was also observed in previous work by Birky et al.
over MgO at 673 K over a wide range of ethanol conversions
(7−23%).27
3. RESULTS
3.1. Catalyst Characterization. Powder X-ray diffraction
patterns of the investigated materials are shown in Figure 3.
The magnesia catalyst (Figure 3a) had an XRD pattern that is
characteristic of periclase MgO, the cubic form of magnesium
oxide. The XRD pattern of the stoichiometric HAP material,
prepared via coprecipitation (Figure 3b), confirmed that the
sample was composed of crystalline hydroxyapatite and that no
other phases were present.
The BET surface areas of the HAP and MgO catalysts used
for this work are summarized in Table 1. Surface analysis by
XPS revealed that the HAP material had a lower Ca/P surface
molar ratio (1.46) than that measured in the bulk by ICP-OES
(1.66). The surface deficiency in calcium is consistent with
prior works and is likely due to its susceptibility to lattice
substitutions and an ability to assemble in nonstoichiometric
forms.31−33
3.2. Steady-State Conversion of Ethanol. The reactivity
results obtained during the steady-state conversion of ethanol
over MgO at 653 K at three different total gas flow rates are
presented in Table 2. Acetaldehyde, formed via dehydrogen-
ation of ethanol, was the primary product at low ethanol
Table 3 presents the analogous results obtained from the
steady-state reaction of ethanol over stoichiometric hydrox-
yapatite at 613 K at three different total flow rates. Butanol was
the major product observed at all three conditions, with
selectivities toward the coupled product >60% even at
conversions as low as 3.2%. The dehydrated side product
ethene was a minor product observed during the coupling of
ethanol over HAP with a selectivity of only 1% at the reaction
conditions investigated. It should also be noted that the
temperature of the reaction over HAP (613 K) was 40 K lower
than that over MgO, and even at the lower temperature, the
HAP catalyst was ∼300% more active in the ethanol coupling
reaction than MgO on a surface area basis (Tables 2,3).
1740
DOI: 10.1021/cs502023g
ACS Catal. 2015, 5, 1737−1746

ACS Catalysis
Research Article
results obtained previously in our lab on MgO at 673 K.27
Significantly higher surface coverages of ethanol were observed
on HAP at 613 K. Since a lower temperature and a smaller
HAP sample mass loading relative to MgO were required to
maintain low conversion, the interaction of ethanol with the
reactor system walls and quartz wool plug would be relatively
more important in the HAP experiments. A blank SSITKA
experiment was thus performed at the two highest flow rates
(50 and 75 cm3 min−1) to determine the surface residence time
of the ethanol in the system. This time constant associated with
the blank reactor was then subtracted from the measured τethanol
to give a “corrected” τethanol associated with the catalyst at each
flow rate (50 and 75 cm3 min−1), which was used to calculate a
“corrected” value of ethanol surface coverage accounting for
ethanol adsorption in the reactor system (Table 4). Evidently,
the surface coverage of ethanol was similar on both MgO and
HAP, at ∼5.0 × 10−6 mol m−2.
Table 3. Product Distribution during the Catalytic
Conversion of Ethanol over Stoichiometric HAP at 613 K
selectivity (C%)
total flow
rate
ethanol
conv
(%)
rate ethanol
conv
(cm3 min−1
)
(mol m−2 s−1
)
ethene acetaldehyde butanol
30
50
75
6.6
4.3
3.2
4.1 × 10−8
4.4 × 10−8
5.0 × 10−8
1
1
1
24
32
36
75
67
63
Catalyst loading: 0.06 g; T = 613 K; total system pressure = 1.3 atm.
Ethanol conversions obtained during the steady-state
Guerbet coupling of ethanol over HAP and MgO are plotted
in Figure 4 as a function of inverse volumetric flow rate, which
Figure 5 compares the normalized transient decays for 12C-
unlabeled butanol obtained over HAP with that obtained over
Figure 4. Conversion of ethanol as a function of inverse reactant flow
rate during the coupling of ethanol over MgO and stoichiometric HAP
at 653 and 613 K, respectively. The observed linear dependence
confirms differential reactor conditions.
Figure 5. Normalized isotopic transient response curves for butanol
following the switch from unlabeled ethanol to doubly labeled 13C-
labeled ethanol with a total flow of 50 cm3 min−1 during the coupling
of ethanol over MgO at 653 K and HAP at 613 K. Curves have been fit
using a two-term exponential decay function.
is proportional to reactor space time. Linearity of the results
confirms that the reactor was operated differentially with
respect to ethanol conversion.
3.3. Multiproduct SSITKA. Multiproduct SSITKA meas-
urements during the Guerbet coupling of ethanol over MgO
and HAP allowed for the quantification of important kinetic
parameters of the reaction, such as surface concentrations of
reaction intermediates (Ni), mean surface residence times of
adsorbed species (τ), and an approximation for the intrinsic
turnover frequencies of the catalyic cycle.
The average surface residence time and surface coverage of
ethanol on MgO and HAP with varying flow rate are
summarized in Table 4. The surface coverage of ethanol on
MgO was ∼5.0 × 10−6 mol m−2 (between 4.5 and 5.7 × 10−6
mol m−2), regardless of total flow rate, which agrees well with
MgO, during the steady-state reaction of ethanol following a
switch from 12C-labeled ethanol to 13C-labeled ethanol. The
butanol response curves presented in Figure 5 were fit with a
two-term exponential decay function. The 12C butanol signal
observed with the HAP catalyst exhibited a slow decay that did
not fully reach background level until after 900 s. The butanol
transient over MgO, on the other hand, was completed after
only 300 s.
Mean surface residence times as well as coverages of surface
intermediates that led to acetaldehyde and butanol are
summarized in Tables 5 and 6, respectively. For both MgO
Table 4. Time Constants and Surface Coverages of Ethanol (Nethanol) during the Steady-State Guerbet Coupling of Ethanol over
MgO and HAP at 653 and 613 K, Respectively
τethanol (s)
coverage of ethanol, Nethanol (mol m−2
)
total flow rate (cm3 min−1
)
MgO (653 K)
HAP (613 K)
MgO (653 K)
HAP (613 K)
HAP (corrected) (613 K)
30
50
75
25
18
10
43
4.8 × 10−6
5.7 × 10−6
4.5 × 10−6
2.7 × 10−5
9.1 × 10−6
1.1 × 10−5
8.8
7.1
3.9 × 10−6
5.0 × 10−6
1741
DOI: 10.1021/cs502023g
ACS Catal. 2015, 5, 1737−1746

ACS Catalysis
Research Article
Table 5. Time Constants and Surface Coverages of Reactive
Intermediates Leading to Acetaldehyde (NAcH) during the
Steady-State Guerbet Coupling of Ethanol over MgO and
HAP at 653 and 613 K, Respectively
coverage of intermediates to
acetaldehyde, NAcH
τAcH (s)
(mol m−2
)
total flow rate
MgO
(653 K)
HAP
(613 K)
MgO
(653 K)
HAP
(613 K)
(cm3 min−1
)
30
50
75
15
13
11
4.4
4.6
4.6
1.1 × 10−7
1.2 × 10−7
1.2 × 10−7
4.3 × 10−8
6.5 × 10−8
8.2 × 10−8
Table 6. Time Constants and Surface Coverages of Reactive
Intermediates Leading to Butanol (Nbutanol) during the
Steady-State Guerbet Coupling of Ethanol over MgO and
HAP at 653 and 613 K, Respectively
Figure 6. DRIFTS spectra of adsorbed ethanol at 303 K on MgO
(left) and HAP (right) collected after heating to various temperatures:
(a) 303, (b) 373, (c) 473, (d) 573, and (e) 673 K. Spectra are offset
for clarity.
coverage of intermediates to
τbutanol (s)
butanol, Nbutanol (mol m−2
)
total flow rate
MgO
HAP
(613 K)
MgO
HAP
(613 K)
(cm3 min−1
)
(653 K)
(653 K)
30
50
75
93
53
27
310
117
69
2.8 × 10−7
8.1 × 10−8
4.6 × 10−8
4.8 × 10−6
1.7 × 10−6
1.1 × 10−6
(υCH3), and 2847 (υCH2) cm−1 are attributed to C−H
stretches of ethoxide, and the bands observed at 1125 and 1063
cm−1 correspond to the two C−C−O stretching modes of the
adsorbed ethoxide species.27 Features of the surface ethoxide
intermediate in the C−H stretching region remain evident on
the MgO surface up to 673 K.
The DRIFTS spectra collected of adsorbed ethanol on HAP
at 303 K followed by stepwise heating to 673 K are displayed in
Figure 6 (right). The adsorption of ethanol revealed three
peaks in the 3100−2800 cm−1 region that can be assigned to
the C−H stretches of a surface ethoxide species: 2975 (υCH3),
2932 (υCH2), and 2904 cm−1 (υCH3). Lower wavenumber
bands associated with ethoxide could not be detected because
of strong absorption by the HAP material in this region. The
intensity of the C−H bands attributed to ethoxide significantly
decreased after heating the sample to just 373 K. Upon further
heating to 573 K, the bands were not observed, indicating
ethoxide had completely desorbed from HAP at this temper-
ature. This result is consistent with previous reports on
hydroxyapatite materials that also showed the disappearance of
C−H ethoxide bands by 573 K.34
3.5. Adsorption Microcalorimetry of Carbon Dioxide,
Triethylamine and Ethanol. Adsorption sites on the catalytic
materials were also characterized by adsorption microcalorim-
etry at 303 K. The experimental results, obtained from the
adsorption of carbon dioxide, triethylamine, and ethanol, are
summarized in Figures 7, 8, and 9, respectively.
Carbon dioxide was used to probe the surface base properties
of the catalysts. Isotherms obtained from the adsorption
microcalorimetry of CO2 on MgO and HAP are given in Figure
7. The total adsorption capacity, or uptake of each probe
molecule on the catalytic surface, was determined by
extrapolating the high pressure, horizontal portion of the
isotherm (corresponds to saturation) to zero pressure. For CO2
on MgO and HAP, the capacity was 1.0 and 2.5 μmol m−2,
respectively, suggesting a significantly higher base site density
on the surface of stoichiometric hydroxyapatite compared with
that observed on MgO.
and HAP, the average surface lifetime of reactive intermediates
leading to acetaldehyde was much shorter than that leading to
butanol. Stated another way, it took a significantly longer time
for butanol to exit the reactor compared with acetaldehyde. The
surface coverage of reaction intermediates leading to
acetaldehyde relative to butanol on the two catalysts also
provides valuable information. On MgO, there was a higher
number of intermediates that led to acetaldehyde compared
with those that led to butanol (NAcH > NBuOH), whereas the
opposite trend was observed over HAP (NBuOH ≫ NAcH).
Moreover, Table 6 indicates the surface density of adsorbed
intermediates leading to butanol was orders of magnitude
greater on HAP than on MgO at all three flow rates
investigated.
The readsorption of reactants or products in the reaction can
have a significant influence on the time constant measured by
SSITKA. The possible effect of readsorption can be seen in the
variation of τ for ethanol and butanol with respect to flow rate
listed in Tables 4 and 6, respectively. The mean surface
residence times of adsorbed intermediates to acetaldehyde
(τAcH) (Table 5), however, were relatively independent of total
flow rate, which suggests acetaldehyde readsorption was
negligible. Because butanol is the product of a sequential
surface reaction, the contribution of readsorption to the butanol
formation time constant cannot be determined by simply
varying the flow rate. This concept will be explored in the
Discussion section.
3.4. STPD of Adsorbed Ethanol Monitored by DRIFTS.
To probe the interaction of ethanol on the catalytic surfaces,
DRIFTS of preadsorbed ethanol on MgO and HAP followed
by STPD was performed. The spectra obtained from these
experiments are presented in Figure 6.
The left plot in Figure 6 shows IR bands observed in the
3200−2600 and 1300−1000 cm−1 regions after ethanol
adsorption on MgO at 303 K at various temperatures. The
bands are characteristic of adsorbed ethoxide species formed on
the surface of the catalyst. The peaks at 2954 (υCH3), 2917
Figure 7b presents the differential heats of CO2 adsorption
on MgO and HAP as a function of surface coverage. The initial
differential heat of CO2 adsorption on MgO was ∼20 kJ mol−1
1742
DOI: 10.1021/cs502023g
ACS Catal. 2015, 5, 1737−1746

ACS Catalysis
Research Article
Figure 7. Adsorption microcalorimetry of carbon dioxide on (Δ)
Figure 8. Adsorption microcalorimetry of triethylamine on (Δ) MgO
●
MgO and ( ) stoichiometric hydroxyapatite (HAP) catalysts at 303
●
and ( ) stoichiometric hydroxyapatite catalysts at 303 K; (a)
K; (a) adsorption isotherms of carbon dioxide and (b) differential
heats of adsorption as a function of coverage.
adsorption isotherms of triethylamine and (b) differential heats of
adsorption as a function of coverage.
consistent with the surface coverages of ethanol on MgO and
HAP determined by SSITKA (Table 4).
higher than that on HAP, which implies MgO exposes stronger
base sites that interact with CO2 compared with those on HAP.
Adsorption microcalorimetry of ammonia has been em-
ployed extensively as a probe of surface acidity; however, there
can be multiple interactions with ammonia and solid surfaces.
These additional adsorption states include weak hydrogen
bonding interactions with the surface as well as deprotonation
of ammonia by strong base sites to form surface NH2 species.35
Triethylamine (TEA) is harder to deprotonate and is
therefore less susceptible to dissociation on the surface. It
also has a stronger proton affinity than ammonia, suggesting it
is a stronger gas-phase base and should therefore be more
selective in probing surface acid sites.36
Adsorption isotherms of TEA obtained on MgO and HAP at
303 K are presented in Figure 8a. The low observed uptake and
linear variation of TEA surface coverage with pressure on MgO
indicate a very weak interaction with the surface. In contrast,
TEA was chemisorbed onto the HAP catalytic surface with an
overall uptake of 1.7 μmol m−2, which is illustrated by its
Langmuirian adsorption isotherm. A weak surface interaction
between TEA and MgO is confirmed by the relatively low
differential heats of TEA adsorption on MgO compared with
those observed on HAP (Figure 8b). Evidently, HAP exposes a
considerably higher acid site density than MgO.
Differential heats of ethanol adsorption as a function of
surface coverage for the catalysts are provided in Figure 9b. The
initial differential heats observed on MgO and HAP were ∼118
and ∼90 kJ mol−1, respectively. This ∼30 kJ mol−1 difference in
adsorption energy suggests that the ethanol interacts more
weakly with HAP compared with MgO, which is consistent
with the results obtained from STPD of preadsorbed ethanol
(Figure 6) that reveal desorption of ethanol from HAP at a
significantly lower temperature than from MgO. It is also worth
mentioning that the ethanol adsorption sites on both MgO and
HAP appear to be fairly uniform in strength, indicated by the
invariance in ΔHads as a function of coverage up to about 4
μmol m−2 (Figure 9b).
4. DISCUSSION
The SSITKA measurements allowed for the direct quantifica-
tion of critical kinetic parameters, such as mean surface
residence times, as well as surface coverages of intermediates
leading to products formed in the reaction. Transient results
revealed that for both MgO and HAP, the residence time for
intermediates leading to acetaldehyde (τAcH) was much shorter
than the residence time of reaction intermediates leading to
butanol (τBuOH) at all three flow rates investigated. Moreover,
the residence times of ethanol and butanol depended strongly
on flow rate, which suggests that the alcohols readsorbed and
desorbed as they passed through the reactor. It is tempting to
extrapolate the SSITKA results to infinite flow rate in an
Figure 9 shows the results from adsorption microcalorimetry
of ethanol on MgO and HAP at 303 K. The adsorption
isotherms for the two catalysts in Figure 9a were nearly
identical, resulting in an overall ethanol uptake of 5.1 μmol m−2
on MgO and 5.2 μmol m−2 on HAP. These results are
1743
DOI: 10.1021/cs502023g
ACS Catal. 2015, 5, 1737−1746

ACS Catalysis
Research Article
adsorbed enolate species. The isotopic transient results
obtained during the steady-state conversion of ethanol over
MgO revealed a significantly higher surface coverage of reactive
intermediates leading to acetaldehyde than to butanol (Tables
5, 6: NAcH > NBuoH) at all flow rates. These results suggest a
higher fraction of surface acetaldehyde produced during the
reaction of ethanol on MgO desorbed rather than coupled form
butanol. The coverages NAcH and NBuOH are consistent with the
higher selectivity to acetaldehyde compared with butanol
observed over MgO.
In contrast, the coverage of intermediates that led to butanol
on HAP was higher than that leading to acetaldehyde by
roughly 1−2 orders of magnitude (NBuOH ≫ NAcH), regardless
of flow rate (Tables 5, 6). Evidently, the majority of
acetaldehyde formed from ethanol dehydrogenation on HAP
remained on the surface to undergo sequential reactions leading
to butanol formation.
A previous study from our group on ethanol coupling over
MgO investigated the relationship between butanol rate and
acetaldehyde concentration over a broad range of ethanol
conversions (7−23%) and observed a first-order dependence of
butanol formation on gas-phase acetaldehyde concentration.27
In the current work on MgO, a 36% decrease in gas-phase
acetaldehyde concentration at the reactor exit (from 57.8 μmol
L−1 to 37.2 μmol L−1 at 7.9% and 3.7% ethanol conversion,
respectively) corresponded to a 44% decrease in butanol
production rate (3.0 × 10−9 to 1.7 × 10−9 mol s−1 m−2), which
is consistent with earlier work and confirms a strong
dependence of butanol formation on acetaldehyde concen-
tration. Interestingly, a 28% decrease in gas-phase acetaldehyde
concentration at the reactor exit during ethanol coupling over
HAP was not accompanied by a significant change in the
production rate of butanol. These results suggest that most of
the butanol formed over HAP did not involve the participation
of gas-phase acetaldehyde, whereas the butanol formation over
MgO might involve a surface that is closer to equilibrated with
gas-phase acetaldehyde.
Figure 9. Adsorption microcalorimetry of ethanol on (Δ) MgO and
●
( ) stoichiometric hydroxyapatite (HAP) catalysts at 303 K; (a)
adsorption isotherms of ethanol and (b) differential heats of
adsorption as a function of coverage.
attempt to remove the effects of readsorption on the calculated
time constants. Unfortunately, the selectivity to butanol during
Guerbet coupling can depend on ethanol conversion because of
the sequential nature of the reaction network to produce
butanol. Therefore, the discussion of SSITKA results will be
focused on those determined at the highest flow rate, 75 cm3
min−1, which is a compromise between achieving a measurable
conversion level and minimizing readsorption effects. From
Table 4, the values of τethanol for MgO and HAP were 10 and 7.1
s, respectively, and provide a reasonable estimation of the time
constant for an alcohol to desorb from the catalyst bed and exit
the reactor. The values of τbutanol reported in Table 6 for MgO
and HAP were 27 and 69 s, respectively, at 75 cm3 min−1 total
flow rate. The substantially longer time constants associated
with butanol formation compared with ethanol desorption over
both materials suggests that the butanol time constants are
determined primarily by reaction kinetics instead of adsorp-
tion/desorption artifacts. Moreover, since the ratio of ethanol
to butanol in the gas phase at the reactor exit is >50:1 over both
catalysts, the competitive adsorption of ethanol over butanol
should also help to minimize readsorption effects in the time
constant associated with butanol formation.
Adsorption microcalorimetry of CO2 was performed to
investigate the number and nature of base sites on the catalytic
surfaces. Microcalorimetry results revealed that the coverage of
CO2 on the HAP surface was 2.5 times that of MgO. However,
differential heats of CO2 adsorption on HAP were lower than
on MgO. Recent IR studies on HAP have shown that CO2
interacts with the OH− groups and the O2− atoms of surface
PO4 groups.37 Adsorption microcalorimetry suggests that
3−
relative to MgO, HAP exposed a higher number density of base
sites but that the sites were weaker in adsorption binding
energy. Adsorption microcalorimetry of TEA also revealed a
higher acid site density on the surface of HAP compared with
MgO. Since the Guerbet coupling of ethanol likely occurs on
acid−base site pairs, the high reactivity of ethanol on HAP
compared with MgO appears to be related to high number of
acid−base site pairs with appropriate binding affinity on HAP.
The higher density of these preferred site pairs may also explain
why a significantly higher surface coverage of intermediates
leading to butanol was observed on HAP relative to MgO.
Stepwise temperature-programmed desorption of pread-
sorbed ethanol at 303 K on MgO and HAP was studied
using DRIFTS. The IR spectra revealed the formation of a
surface ethoxide intermediate on both of the catalytic materials.
Ethanol dissociatively adsorbs on the surface forming ethoxide
coordinated to a Lewis acid site (Ca2+ on HAP, Mg2+ for MgO)
and proton-like hydrogen coordinated to a Brønsted base site.
A commonly hypothesized reaction path for the Guerbet
coupling of ethanol involves the aldol condensation of
intermediate acetaldehyde, which likely proceeds through a
surface enolate species that produces coupled products that
desorb as butanol. In this coupling path, surface acetaldehyde
produced from ethanol dehydrogenation may desorb or
undergo base-catalyzed abstraction of the α-H to form an
1744
DOI: 10.1021/cs502023g
ACS Catal. 2015, 5, 1737−1746

ACS Catalysis
Research Article
The STPD results show that the ethoxide species on HAP
mostly desorbed from the surface by 473 K, whereas a
temperature of 673 K was required on MgO. This 200 K
difference in desorption temperature indicates that the
dissociated ethanol is much more weakly held to HAP relative
to MgO, which is consistent with the results obtained from
adsorption microcalorimetry of ethanol where the initial
differential heat of ethanol adsorption on HAP was ∼30 kJ
mol−1 lower than that on MgO.
NAcH > NBuOH. Given the generally accepted mechanism for
Guerbet coupling that involves aldol condensation of
acetaldehyde, it appears that a greater fraction of the
acetaldehyde produced during the reaction proceeds toward
coupling products on HAP relative to MgO.
Adsorption microcalorimetry of CO2 showed a higher surface
density of base sites on HAP compared with that on MgO, but
the CO2 adsorption binding energy was weaker on HAP.
Moreover, adsorption of triethylamine revealed significant
Lewis acidity on HAP and negligible acidity on MgO.
It is likely that the high activity and selectivity observed
during the Guerbet coupling of ethanol over HAP involves the
proper balance of acid−base site pairs to facilitate all of the
steps in the sequence, including alcohol dehydrogenation, aldol
condensation and aldehyde hydrogenation. The relatively
strong basicity of MgO retains adsorbed ethanol at higher
temperatures compared with HAP, which is consistent with the
idea that Guerbet coupling is facilitated by weak acid−base
bifunctional catalysts.
The TOF based on ethanol adsorption is a lower bound
because not all of the adsorbed ethanol proceeds to product. A
better estimate of the TOF can be derived from the isotopic
transient results as the reciprocal of the mean surface residence
time (TOF = τ−1). Unfortunately, the readsorption of alcohols
during the transient increased the measured surface residence
time. If we try to minimize the effects of readsorption by using
the surface residence time of intermediates to butanol at the
highest flow rate in the study (75 cm3 min−1), we have τBuOH
=
27 and 69 s for MgO (653 K) and HAP (613 K), respectively.
Moreover, if we assume that the τEtOH at high flow rate is a
reasonable approximation of the effect of readsorption, then we
can simply subtract the value for τEtOH (Table 4) from τBuOH
(Table 6), as reported by Birky et al. to get a “corrected” τBuOH.
The inverse of the “corrected” τBuOH provides a better estimate
of the TOF associated with intermediates that form butanol,
denoted as TOFSSITKA. The values derived from Tables 4 and 6
are 0.059 s−1 for MgO (653 K) and 0.016 s−1 for HAP (613 K).
The values of TOFSSITKA for acetaldehyde production can be
estimated as the inverse of the τAcH at the highest flow rate
because readsorption appears to be insignificant; thus, the
TOFSSITKA for AcH is 0.091 s−1 for MgO at 653 K and 0.22 s−1
for HAP at 613 K. The higher selectivity to butanol over HAP
compared with MgO is apparently the consequence of a much
higher coverage of surface intermediates leading to butanol
during the steady-state reaction.
ASSOCIATED CONTENT
* Supporting Information
The following file is available free of charge on the ACS
■
S
Publications website at DOI: 10.1021/cs502023g.
Normalized transient response curves are provided from
additional SSITKA experiments that were conducted at
various flow rates (PDF)
AUTHOR INFORMATION
Corresponding Author
*E-mail: rjd4f@virginia.edu.
■
Present Address
†(For H.S.) SABIC Technology Center, 1600 Industrial Blvd.,
Sugar Land, TX 77478, USA
Ogo et al.38 used isotopic exchange reactions to show that
ethanol dehydrogenation occurs rapidly on strontium-sub-
stituted hydroxyapatite catalysts and that aldol condensation is
the rate-determining step. The very short residence time of
intermediates leading to acetaldehyde (τAcH) relative to τBuOH
observed here by SSITKA would support their kinetic
mechanism. It is possible that the kinetically relevant step of
the reaction is related to enolate formation via α-H abstraction
of adsorbed acetaldehyde. Results from adsorption micro-
calorimetry of CO2 (a general probe of base sites) and STPD of
adsorbed ethanol suggest HAP has a weaker surface affinity
than MgO for those probe molecules. Furthermore, adsorption
of triethylamine revealed the presence of weak Lewis acid sites
on HAP that were not found on MgO. Thus, under steady-state
reaction conditions, more surface acid−base pairs on HAP may
be available for enolate formation compared with MgO, which
would lead to a higher formation rate of coupled products.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by the Chemical Sciences, Geo-
sciences and Biosciences Division, Office of Basic Energy
Sciences, Office of Science, U.S. Department of Energy, Grant
No. DE-FG02-95ER14549. The authors thank Dmitry Pestov
at Virginia Commonwealth University for assistance with XPS.
REFERENCES
■
(1) Dowson, G. R. M.; Haddow, M. F.; Lee, J.; Wingad, R. L.; Wass,
D. F. Angew. Chem., Int. Ed. 2013, 52, 9005−9008.
(2) Savage, N. Nature 2011, 474, S9−S11.
(3) Tsuchida, T.; Sakuma, S.; Takeguchi, T.; Ueda, W. Ind. Eng.
Chem. Res. 2006, 45, 8634−8642.
(4) Koda, K.; Matsu-ura, T.; Obora, Y.; Ishii, Y. Chem. Lett. 2009, 38,
838−839.
5. CONCLUSIONS
(5) O’Lenick, A. J. J. Surfact. Deterg. 2001, 4, 311−315.
(6) Carlini, C.; Di Girolamo, M.; Macinai, A.; Marchionna, M.;
Noviello, M.; Raspolli Galletti, A. M.; Sbrana, G. J. Mol. Catal. A:
Chem. 2003, 200, 137−146.
(7) Veibel, S.; Nielsen, J. Tetrahedron 1967, 213, 1723−1733.
(8) Yang, C.; Meng, Z. Y. J. Catal. 1993, 142, 37−44.
(9) Ndou, A. S.; Plint, N.; Coville, N. J. Appl. Catal. A 2003, 251,
337−345.
Isotopic transient studies were performed during the steady-
state Guerbet coupling of ethanol to butanol over stoichio-
metric hydroxyapatite and MgO at 613 and 653 K, respectively.
The HAP catalyst was ∼3 times more active in the reaction
than MgO on a surface area basis, even at the lower TOF. The
selectivity over HAP was as high as 75% to butanol.
The surface coverage of reactive intermediates leading to
(10) Gines, M. J. L.; Iglesia, E. J. Catal. 1998, 176, 155−172.
(11) Scalbert, J.; Thibault-Starzyk, F.; Jacquot, R.; Morvan, D.;
Meunier, F. J. Catal. 2014, 311, 28−32.
butanol (NBuOH) relative to that leading to acetaldehyde (NAcH
)
was very high on HAP (NBuOH ≫ NAcH), whereas on MgO,
1745
DOI: 10.1021/cs502023g
ACS Catal. 2015, 5, 1737−1746

ACS Catalysis
Research Article
(12) Kozlowski, J. T.; Davis, R. J. ACS Catal. 2013, 3, 1588−1600.
(13) Ueda, W.; Kuwabara, T.; Ohshida, T.; Morikawa, Y. J. Chem.
Soc., Chem. Commun. 1990, 1558−1559.
(14) Ueda, W.; Ohshida, T.; Kuwabara, T.; Morikawa, Y. Catal. Lett.
1992, 12, 97−104.
(15) Di Cosimo, J.; Apesteguía, C.; Gines
2000, 190, 261−275.
(16) Leon, M.; Díaz, E.; Ordon
442.
́
, M.; Iglesia, E. J. Catal.
́
́
ez, S. Catal. Today 2011, 164, 436−
̃
(17) Gotoh, K.; Nakamura, S.; Mori, T.; Morikawa, Y. Stud. Surf. Sci.
Catal. 2000, 130, 2669−2674.
(18) Tsuchida, T.; Kubo, J.; Yoshioka, T.; Sakuma, S.; Takeguchi, T.;
Ueda, W. J. Catal. 2008, 259, 183−189.
(19) Ogo, S.; Onda, A.; Yanagisawa, K. Appl. Catal. A 2011, 402,
188−195.
(20) Kibby, C. L.; Hall, W. K. J. Catal. 1973, 29, 144−159.
(21) Joris, S. J.; Amberg, C. H. J. Phys. Chem. 1971, 76, 3167−3171.
(22) Bett, J. A. S.; Christner, L. G.; Hall, W. K. J. Am. Chem. Soc.
1967, 486, 5535−5541.
(23) McClaine, B. C.; Davis, R. J. J. Catal. 2002, 211, 379−386.
(24) Gao, J.; Mo, X.; Goodwin, J. G. J. Catal. 2010, 275, 211−217.
(25) Ledesma, C.; Yang, J.; Chen, D.; Holmen, A. ACS Catal. 2014,
4, 4527−4547.
(26) Shannon, S. L.; Goodwin, J. G. Chem. Rev. 1995, 95, 677−695.
(27) Birky, T. W.; Kozlowski, J. T.; Davis, R. J. J. Catal. 2013, 298,
130−137.
(28) Tsuchida, T.; Kubo, J.; Yoshioka, T.; Sakuma, S.; Takeguchi, T.;
Ueda, W. J. Japan Pet. Inst. 2009, 52, 51−59.
(29) Shou, H.; Davis, R. J. J. Catal. 2013, 306, 91−99.
(30) Bordawekar, S.; Doskocil, E.; Davis, R. Langmuir 1998, 14,
1734−1738.
̌ ́
(31) Stosic, D.; Bennici, S.; Sirotin, S.; Calais, C.; Couturier, J.-L.;
Dubois, J.-L.; Travert, A.; Auroux, A. Appl. Catal. A 2012, 447, 124−
134.
(32) Tsuchida, T.; Kubo, J.; Yoshioka, T.; Sakuma, S.; Takeguchi, T.;
Ueda, W. J. Catal. 2008, 259, 183−189.
(33) Tanaka, H.; Watanabe, T.; Chikazawa, M. J. Chem. Soc. Faraday
Trans. 1997, 93, 4377−4381.
(34) Faria, R. M. B.; Cesar, D. V.; Salim, V. M. M. Catal. Today 2008,
133−135, 168−173.
(35) Auroux, A.; Gervasini, A. J. Phys. Chem. 1990, 94, 6371−6379.
(36) Cardona-Martinez, N. C.; Dumesic, J. A. Adv. Catal. 1992, 38,
149−244.
(37) Diallo-Garcia, S.; Ben Osman, M.; Krafft, J.-M.; Casale, S.;
Thomas, C.; Kubo, J.; Costentin, G. J. Phys. Chem. C 2014, 118,
12744−12757.
(38) Ogo, S.; Onda, A.; Iwasa, Y.; Hara, K.; Fukuoka, A.; Yanagisawa,
K. J. Catal. 2012, 296, 24−30.
1746
DOI: 10.1021/cs502023g
ACS Catal. 2015, 5, 1737−1746