Oxygen and hydroxyl species induce multiple reaction pathways for the partial oxidation of allyl alcohol on gold

Article abstract of DOI:10.1021/ja502347d

Partial oxidation of alcohols is a topic of great interest in the field of gold catalysis. In this work, we provide evidence that the partial oxidation of allyl alcohol to its corresponding aldehyde, acrolein, over oxygen-precovered gold surfaces occurs via multiple reaction pathways. Utilizing temperature-programmed desorption (TPD) with isotopically labeled water and oxygen species, reactive molecular beam scattering, and density functional theory (DFT) calculations, we demonstrate that the reaction mechanism for allyl alcohol oxidation is influenced by the relative proportions of atomic oxygen and hydroxyl species on the gold surface. Both atomic oxygen and hydroxyl species are shown to be active for allyl alcohol oxidation, but each displays a different pathway of oxidation, as indicated by TPD measurements and DFT calculations. The hydroxyl hydrogen of allyl alcohol is readily abstracted by either oxygen adatoms or adsorbed hydroxyl species on the gold surface to generate a surface-bound allyloxide intermediate, which then undergoes α-dehydrogenation via interaction with an oxygen adatom or surface hydroxyl species to generate acrolein. Mediation of a second allyloxide with the hydroxyl species lowers the activation barrier for the α-dehydrogenation process. A third pathway exists in which two hydroxyl species recombine to generate water and an oxygen adatom, which subsequently dehydrogenates allyloxide. This work may aid in the understanding of oxidative catalysis over gold and the effect of water therein.

Full text of DOI:10.1021/ja502347d

Article
pubs.acs.org/JACS
Oxygen and Hydroxyl Species Induce Multiple Reaction Pathways for
the Partial Oxidation of Allyl Alcohol on Gold
Gregory M. Mullen,† Liang Zhang,‡ Edward J. Evans, Jr.,‡ Ting Yan,‡ Graeme Henkelman,‡
and C. Buddie Mullins*,†,‡
‡
†McKetta Department of Chemical Engineering and Department of Chemistry, and Center for Nano and Molecular Science and
Technology, Texas Materials Institute, and Center for Electrochemistry, University of Texas at Austin, Austin, Texas 78712-0231,
United States
S
* Supporting Information
ABSTRACT: Partial oxidation of alcohols is a topic of great interest in the field of
gold catalysis. In this work, we provide evidence that the partial oxidation of allyl
alcohol to its corresponding aldehyde, acrolein, over oxygen-precovered gold surfaces
occurs via multiple reaction pathways. Utilizing temperature-programmed desorption
(TPD) with isotopically labeled water and oxygen species, reactive molecular beam
scattering, and density functional theory (DFT) calculations, we demonstrate that
the reaction mechanism for allyl alcohol oxidation is influenced by the relative
proportions of atomic oxygen and hydroxyl species on the gold surface. Both atomic
oxygen and hydroxyl species are shown to be active for allyl alcohol oxidation, but
each displays a different pathway of oxidation, as indicated by TPD measurements
and DFT calculations. The hydroxyl hydrogen of allyl alcohol is readily abstracted by
either oxygen adatoms or adsorbed hydroxyl species on the gold surface to generate a
surface-bound allyloxide intermediate, which then undergoes α-dehydrogenation via
interaction with an oxygen adatom or surface hydroxyl species to generate acrolein. Mediation of a second allyloxide with the
hydroxyl species lowers the activation barrier for the α-dehydrogenation process. A third pathway exists in which two hydroxyl
species recombine to generate water and an oxygen adatom, which subsequently dehydrogenates allyloxide. This work may aid in
the understanding of oxidative catalysis over gold and the effect of water therein.
selectivity toward the corresponding aldehydes. Despite the
breadth of investigation on the subject, the reaction
mechanisms and active sites for oxidation of alcohols over
gold catalysts remain topics of debate. Model studies performed
on single-crystal surfaces in ultrahigh vacuum (UHV) have
helped elucidate the mechanisms for partial oxidation of many
simple alcohols over gold surfaces.22−28 However, little
investigation into the partial oxidation of alcohol species with
multiple functional groups has been carried out.25,29
Furthermore, studies have shown that on gold surfaces
oxidation mechanisms for simple alcohols vary as chain length
is increased.24 Accordingly, the presence of multiple functional
groups may also prove to influence oxidative reaction pathways
over gold surfaces.
Corma et al. have shown that allylic alcohols do not display
the same trends in reactivity as simple alcohols for catalytic
partial oxidation processes carried out over palladium−gold
systems.30 These catalysts induce isomerization, hydrogenation,
and polymerization pathways in addition to catalyzing the
partial oxidation reaction.30 Model palladium surfaces have
been shown to catalyze the partial oxidation of allyl alcohol to
acrolein, the corresponding aldehyde, in addition to inducing
1. INTRODUCTION
Gold was long thought to be an inert material with no catalytic
use, but in 1973, Bond et al. showed that olefin hydrogenation
could be facilitated by gold catalysts.1 In the mid-1980s, gold
catalysts were demonstrated to display high activity for the CO
oxidation reaction by Haruta et al.2 and for hydrochlorination
of ethylene by Hutchings.3 These discoveries led to an
explosion of interest in the field of gold catalysis. Gold has
been demonstrated to catalyze both oxidation and hydro-
genation reactions, displaying exceptional activity and selectiv-
ity for a host of processes of catalytic interest at low
temperatures.4−7 Studies of model gold surfaces have guided
catalyst development and enhanced the fundamental under-
standing of gold−adsorbate interactions.8−15
Several studies have shown that gold displays exceptional
activity and selectivity for catalytic partial oxidation reactions of
alcohols,16−19 a class of transformation of interest to several
industries including pharmaceuticals, cosmetics, and agro-
chemicals. Current practices for carrying out such reactions
utilize hazardous and expensive metal oxides as stoichiometric
reactants. A “green” alternative for carrying out such processes
utilizing a catalytic material is desirable. Catalysts composed of
gold,16,18,19 palladium,20 and gold−palladium bimetallic mix-
tures21 have been shown to display activity for the partial
oxidation of a number of alcohols, demonstrating high
Received: March 7, 2014
Published: April 6, 2014
© 2014 American Chemical Society
6489
dx.doi.org/10.1021/ja502347d | J. Am. Chem. Soc. 2014, 136, 6489−6498

Journal of the American Chemical Society
Article
parallel reaction processes.31,32 The side reactions observed
over palladium−gold catalysts30 and model palladium surfa-
ces31,32 limit the selectivity toward the partial oxidation
pathway. In contrast, gold catalysts supported on nano-
particulate ceria have demonstrated high activity and selectivity
for the partial oxidation of allylic alcohols.30,33 It is of interest to
understand the fundamental reactive processes that occur
between allylic alcohols and oxygen on the gold surface.
rather than that of the aldehyde, but this process has been
suggested to first proceed through the production of the
corresponding aldehyde.17 Evidence has been provided to
suggest that production of the aldehyde is the rate-limiting step
of the partial oxidation process to generate carboxylic acids.49
The incorporation of base in the reactant solution, therefore,
appears to increase the rate of alcohol oxidation to the resulting
aldehyde. An increase in the presence of surface hydroxyl
species when in basic solutions may be responsible for this
enhancement in partial oxidation activity. Several studies have
supported this conclusion, as summarized in the excellent
review by Ide et al.35 Of particular interest is the observation
that the effect of gold particle size on activity for glycerol and
CO oxidation reactions is far less significant for systems in basic
aqueous media than in neutral, acidic, or vapor-phase
conditions.50 In fact, even bulk gold powder has been shown
to display activity for these reactions in basic solutions.50
Here we report on the systematic study of allyl alcohol
oxidation over the Au(111) surface precovered with various
exposures of oxygen and hydroxyl species. We utilize TPD with
isotopically labeled water (D2O) and oxygen (18O), reactive
molecular beam scattering (RMBS), and density functional
theory (DFT) methods to elucidate the mechanistic details of
the selective oxidation of allyl alcohol to acrolein. This reaction
can occur through multiple pathways, proceeding via both
reaction-limited and desorption-limited kinetic processes. The
choice of pathway is dictated by the arrangement of oxygen
adatoms, hydroxyl species, and allyloxide intermediates on the
gold surface. Our results suggest that both oxygen adatoms and
hydroxyl species can oxidize allyl alcohol to acrolein, depending
on the relative proportions of surface species. These
observations may aid in the general mechanistic understanding
of oxidative alcohol catalysis and enhance the fundamental
understanding of oxidation reactions over gold surfaces.
Some studies have investigated the oxidation of allylic
alcohols over the oxygen-covered Au(111) surface.25,29 In an
UHV study, Deng et al. showed that the Au(111) surface
precovered with oxygen adatoms by exposure to ozone induced
the partial oxidation of allyl alcohol to acrolein in the
subsequent temperature-programmed desorption (TPD) spec-
trum at 255 K in addition to producing several other oxidation
products (CO2, acrylic acid, and carbon suboxide).29 Based on
analogy to TPD studies of isotopically labeled propene on O/
Au(111), which displayed a similar product distribution, they
proposed a mechanism for this reaction that proceeds through a
surface-bound allyloxide intermediate.29 Liu et al. showed that
cyclohexanol (a saturated alcohol) and 2-cyclohexen-1-ol (an
allylic alcohol) exhibit differences in reactivity on O/Au(111).25
Both species were oxidized to yield their corresponding ketones
and CO2, but 2-cyclohexen-1-ol also underwent secondary
oxidation pathways, resulting in the production of 2-cyclo-
hexene-1,4-dione and phenol.25 The observation of additional
pathways for 2-cyclohexen-1-ol was attributed to the higher
relative acidity of the allylic C−H bonds in 2-cyclohexen-1-ol vs
C−H bonds in cyclohexanol.25 These works present interesting
insights into the oxidation of allylic alcohols over gold surfaces.
Studies have explored the effects of water and hydroxyl
species on the catalytic activity of gold surfaces toward a host of
reactions.34,35 Water has been shown to significantly enhance
the activity of gold catalysts for CO oxidation.36−39 The
mechanistic reason for this enhancement remains a topic of
debate. Some studies have argued that water aids in the
activation of molecular oxygen,40,41 while others suggest that it
is associated with the decomposition of reactive intermediates
on the gold surface.42 Yet another possible explanation is
provided by Ide et al. in a review of the role of hydroxyl species
in gold catalysis.35 The authors provide interesting evidence
linking CO oxidation experiments carried out in the vapor
phase to experiments performed in the liquid phase in both
classical catalytic and electrocatalytic systems. Their conclu-
sions suggest that hydroxyl species on the gold surface are
responsible for oxidizing CO, while O2 acts to scavenge
electrons from the surface and regenerate hydroxyl species.
Water has also been shown to affect the activity of gold
surfaces for other reactions. Propene epoxidation43−45 and the
reduction of nitric oxide with propene46 have been shown to be
enhanced by the presence of water when carried out over gold
catalysts. Additionally, water plays an important role in the
selective oxidation of alcohols on gold catalysts.35,47 Water is a
component in every system for the selective oxidation of
alcohols to aldehydes, as a product of the reaction at the very
least, but often as a solvent as well.16,17,48 Differences have been
observed for alcohol oxidation carried out in water versus in the
absence of solvent.17 Furthermore, changing the pH of the
solution has been shown to significantly affect the oxidation
activity for various alcohol species over gold catalysts, with the
reaction occurring more readily in basic solutions.35,48 When
carrying out partial oxidation of alcohols in alkaline media,
production of the corresponding carboxylic acid dominates
2. EXPERIMENTAL AND COMPUTATIONAL METHODS
2.1. UHV Experiments. All experiments in this study were
performed in an UHV chamber that has been described in detail
previously.51 A thin Au(111) single-crystal sample, cut into a circular
disk ∼11 mm in diameter and 1.5 mm thick, is mounted to a tantalum
plate that can be resistively heated to 900 K via a DC power supply
regulated by a proportional-integral-differential controller. The gold
sample is also in thermal contact with a liquid nitrogen reservoir for
cooling to a minimum temperature of 77 K. The sample temperature
was monitored with a K-type thermocouple spot-welded to the
tantalum plate. Prior to each experiment, the gold surface was cleaned
by exposure to atomic oxygen at 77 K, followed by annealing to 700 K
unless otherwise specified. Cleanliness was routinely verified by Auger
electron spectrometry (AES). Periodically, the surface was sputtered
by Ar+ ion bombardment at room temperature to remove impurities,
followed by annealing to 800 K for 10 min to restore the (111)
structure.
Reactants were delivered to the sample via a supersonic molecular
beam dosing apparatus consisting of four differentially pumped
subchambers. Analyses were carried out in a scattering chamber
(base pressure of less than 1 × 10−10 Torr) equipped for quadrupole
mass spectrometry (QMS), AES, and low-energy electron diffraction.
TPD spectra were obtained by heating the Au(111) sample at a rate
of 2 K/s while monitoring desorption of various species via QMS. For
oxidation experiments, oxygen adatoms were dosed onto the sample
first, followed by allyl alcohol. Isotopic oxygen was employed in these
experiments to observe the production of water in TPD spectra
without interference from H216O displaced from the chamber walls by
allyl alcohol (see Supporting Information for further discussion). We
populated the Au(111) surface with 18O and monitored m/z = 20
(H218O), which was not affected by allyl alcohol desorption in the
6490
dx.doi.org/10.1021/ja502347d | J. Am. Chem. Soc. 2014, 136, 6489−6498

Journal of the American Chemical Society
Article
chamber. Additionally, the use of 18O allowed us to determine whether
oxygen associated with the surface was exchanged with oxygen in allyl
alcohol or incorporated into the resulting products.
Oxygen adatoms were dosed onto the gold sample via a radio
frequency (RF)-powered plasma-jet source using a gas mixture of 8%
(v/v) O2 in argon.52,53 Oxygen surface coverages were estimated by
AES following a method employed in a previous study.38 All reactant
species were dosed onto the sample through the same nozzle to
promote coincident exposure, but the RF power was only engaged
while dosing the atomic oxygen. Deionized water was used for TPD
experiments employing H2O. Allyl alcohol (ACROS, 99%), D2O
(Spectra, 99.9%), 16O2 (Matheson Trigas, 99.99%), 18O2 (Isotec,
99%), and Ar (Matheson Trigas, 99.9%) were used as received,
without further purification.
RMBS spectra were obtained by monitoring product and reactant
species via QMS while impinging allyl alcohol onto the O/Au(111)
surface held at constant temperature. Atomic oxygen was dosed onto
the Au(111) surface at 77 K. The surface was then ramped to the
specified temperature in vacuum prior to impingement of pure allyl
alcohol. Allyl alcohol was impinged first onto an inert flag held in front
of the gold sample to obtain baseline signals for each spectrum and
then impinged onto the sample itself to observe product generation
signals. Under these conditions, we estimate the allyl alcohol
translational energy to be approximately 0.1 eV and that virtually
every allyl alcohol molecule is adsorbed to both the clean and
adsorbate-covered surfaces.54−56
2.2. DFT Calculations. Calculations were performed with the
Vienna ab initio simulation package.57−60 The interaction between the
ionic core and the valence electrons was described by the projector-
augmented wave method,61 and the valence electrons were described
with a plane-wave basis up to an energy cutoff of 280 eV.62,63 The
exchange correlation contribution to the total energy functional was
determined using the PBE generalized gradient approximation
functional.64 Corrections for van der Waals interactions were made
using the optB86b-dvW density functional as implemented by Klimes
et al.65 The location and energy of the transition states were calculated
with the climbing-image nudged elastic band method.66,67 Spin
polarization was tested and applied when necessary. The gold surface
was modeled as a 3×3 Au(111) slab with four atomic layers and 14 Å
of vacuum. The Brillouin zone was sampled using a 4×4×1
Monkhorst−Pack k-point mesh.68 The convergence criteria for the
electronic structure and the atomic geometry were 10−4 eV and 0.01
eV/Å, respectively.
Figure 1. TPD spectra for allyl alcohol (m/z = 57) desorption from
Au(111). Allyl alcohol was adsorbed on the Au(111) surface at a
sample temperature of 77 K.
behavior for allyl alcohol TPD from the Rh(111) surface;73
however, the reasons for such behavior were not discussed in
that work. Ehlers et al. studied the interactions of methanol
with the Pt(111) surface via IR spectroscopy and observed that
three distinct layers were formed on this surface.71 The
methanol monolayer interacted strongly with the surface and
did not undergo hydrogen-bonding interactions. The multi-
layers of methanol, on the other hand, did hydrogen bond, as
evidenced by the width and position of the O−H stretch. At the
dosing temperature of 90 K, the methanol multilayers were
amorphous. A significant portion of the multilayer methanol
underwent a phase transition to generate an α-ice phase upon
annealing the surface, while some of the multilayer remained
amorphous due to stronger interaction with the surface or with
the methanol monolayer. Similar observations were made in the
study of ethanol on highly ordered pyrolytic graphite,72 which
displayed a TPD spectrum very similar to that we have
observed for allyl alcohol desorption from the Au(111) surface
in this study, suggesting that the formation of an amorphous
second layer interacting with the monolayer or weakly with the
surface may be a general phenomenon displayed in a number of
alcohol/surface interactions. Additionally, both methanol70 and
2-butanol24 have been shown to display similar features for
desorption from the Au(111) surface, in further support of this
phenomenon.
The α1 feature displayed a peak at ∼160 K. This feature
appeared after the α2 feature had saturated and grew
indefinitely as allyl alcohol coverage was increased, displaying
zero-order desorption kinetics. These characteristics are
consistent with desorption of condensed phase multilayers
physisorbed on top of the α2 species.
No decomposition, coupling, or isomerization products were
detected during allyl alcohol TPD from the Au(111) surface.
AES taken after allyl alcohol desorption tests did not detect the
presence of decomposition products (carbon and oxygen) on
the Au(111) surface. Furthermore, TPD spectra of allyl alcohol
desorption from the clean Au(111) surface taken repeatedly
without cleaning the surface in between experiments were
highly reproducible, suggesting that the surface was unchanged
before and after each experiment. These observations suggest
that allyl alcohol did not react during TPD experiments on the
clean Au(111) surface.
̌
3. RESULTS AND DISCUSSION
3.1. Allyl Alcohol Desorption from the Clean Au(111)
Surface. TPD spectra were taken for allyl alcohol desorption
from the clean Au(111) surface for coverages ranging from 0.5
to 7.7 monolayers (ML). These spectra, depicted in Figure 1,
were obtained by monitoring m/z = 57 (the primary mass
fragment for allyl alcohol) via QMS. The desorption spectra
displayed three features, which we label α1, α2, and β. The β
feature displayed a peak at 195 K. This feature was populated
first and saturated before the α1 and α2 features began to
appear. First-order desorption behavior was observed for this
feature, with a desorption energy estimated to be 50.1 kJ/mol
by Redhead analysis.69 We attribute the β feature to desorption
of the allyl alcohol monolayer interacting directly with the
Au(111) surface.
The α2 feature displayed a peak at 167 K. This feature began
to populate after the β feature had saturated and also displayed
first-order desorption kinetics. The desorption energy for this
feature was estimated to be 42.7 kJ/mol by Redhead analysis.69
The α2 peak saturated before the α1 feature began to populate.
The desorption of alcohol species from a number of different
surfaces has demonstrated features similar to α2 in the allyl
alcohol TPD spectrum.24,70−73 Brown et al. observed similar
6491
dx.doi.org/10.1021/ja502347d | J. Am. Chem. Soc. 2014, 136, 6489−6498

Journal of the American Chemical Society
Article
3.2. Acrolein Desorption from the Clean Au(111)
Surface. TPD spectra were taken for acrolein desorption,
monitored by m/z = 56 (the parent mass of acrolein), from the
clean Au(111) surface for coverages ranging from 0.9 to 2.9
ML, as displayed in Figure 2. These desorption spectra
Figure 2. TPD spectra for acrolein (m/z = 56) desorption from
Au(111). Acrolein was adsorbed on the Au(111) surface at a sample
temperature of 77 K.
exhibited two features. The high-temperature feature displayed
a peak at 166 K. This feature was populated first and saturated
before the low-temperature feature began to appear, displaying
first-order desorption behavior, with a desorption energy
estimated to be 42.4 kJ/mol by Redhead analysis.69 We
attribute this high-temperature feature to desorption of the
acrolein monolayer interacting directly with the Au(111)
surface.
Figure 3. TPD spectra for acrolein (m/z = 56), allyl alcohol (m/z =
57), and H218O (m/z = 20) following adsorption of 3.6 ML of allyl
alcohol on the Au(111) surface populated with (a) 0.02 ML of 18O
and (b) 0.15 ML of 18O. Allyl alcohol and 18O were adsorbed at a
sample temperature of 77 K.
The low-temperature feature displayed a peak at ∼133 K.
This feature appeared after the high-temperature feature had
saturated and grew indefinitely as acrolein coverage was
increased, displaying zero-order desorption kinetics. These
characteristics are consistent with desorption of condensed-
phase multilayers physisorbed on top of the monolayer.
3.3. Multiple Reaction Pathways for Oxidation of Allyl
Alcohol on O/Au(111). Allyl alcohol oxidation was carried
out on the gold surface predosed with 0.02 and 0.15 ML
coverages of 18O adatoms, as shown in Figure 3. In these
experiments each specified coverage of oxygen was dosed onto
the Au(111) surface, followed by 3.6 ML of allyl alcohol prior
to TPD. The spectra taken at each oxygen coverage displayed
some similar features. Allyl alcohol multilayers desorbed at 160
K in the spectra for both oxygen coverages (Figure 3),
displaying features characteristic of the α1 feature as observed
over the clean Au(111) surface (Figure 1). This observation
suggests that the multilayer allyl alcohol species was unaffected
by the presence of oxygen adatoms on the gold surface.
Peaks characteristic of the α2 feature observed over the clean
Au(111) surface were no longer evident in the allyl alcohol
spectra in Figure 3, suggesting that the interaction resulting in
this feature had been disrupted by the presence of oxygen
adatoms on the gold surface at both precoverages. Attenuation
of the α2 feature may result from differences in the adsorption
structure of the allyl alcohol monolayer or changes to the
electronic structure of surface gold atoms induced by oxygen.
Relative to the β feature for allyl alcohol desorption from the
clean Au(111) surface, which displayed a peak at 195 K, the
peak of the β feature was shifted upward in temperature to 205
and 210 K for the 0.02 ML O/Au(111) (Figure 3a) and 0.15
ML O/Au(111) (Figure 3b) surfaces, respectively. A portion of
this feature may result from desorption of unreacted allyl
alcohol in the monolayer, but the shift to higher temperature
suggests that some allyl alcohol associated with the monolayer
interacted with the oxygen-precovered gold surface more
strongly than with the clean Au(111) surface. The formation
of an allyl alcohol−oxygen complex or a surface allyloxide
species might explain this observation. Similar effects observed
in the study of various alcohol species over O/Au(111) have
been suggested to result from alkoxide formation.22−24,70 Deng
et al. have also suggested that the partial oxidation of allyl
alcohol to acrolein proceeds through an allyloxide intermediate
by analogy to experiments with isotopically labeled propene.29
We carried out DFT studies to explore the energetics
associated with allyloxide formation from allyl alcohol on the
oxygen-precovered Au(111) surface. Removal of the hydroxyl
hydrogen from allyl alcohol via surface oxygen, resulting in the
formation of an allyloxide and a surface hydroxyl species, as
depicted in Scheme 1a, occurs with a very low activation barrier
6492
dx.doi.org/10.1021/ja502347d | J. Am. Chem. Soc. 2014, 136, 6489−6498

Journal of the American Chemical Society
Article
3b). The lowest-temperature feature (γ1) displayed a peak at
125 K. To probe the process responsible for generating this
feature, we carried out the RMBS experiment depicted in
Figure 4, in which acrolein production was monitored while
Scheme 1. Proposed Mechanistic Pathways for
Dehydrogenation of Allyl Alcohol To Generate Allyloxide on
the O/Au(111) Surface via (a) an Oxygen Adatom and (b) a
a
Hydroxyl Species
a
Activation barriers (Ea) were determined via DFT calculations.
of 1.9 kJ/mol. Hydroxyl species on the Au(111) surface can
also readily promote removal of the hydroxyl hydrogen from
allyl alcohol, as depicted in Scheme 1b, via a barrier-less process
(Ea = 0.1 kJ/mol). Reaction energy diagrams for these
processes are included in Figure S1. As shown in Figure 3,
the water production (H218O, m/z = 20) spectrum for each
oxygen coverage displayed a feature peaked at 155 K,
characteristic of molecular desorption of water.74 These results
are also in good agreement with a DFT study in which the
interactions of methanol with the O/Au(111) and OH/
Au(111) surfaces were explored.75
Figure 4. RMBS of allyl alcohol on 0.15 ML of O/Au(111). Allyl
alcohol was dosed onto the sample at a surface temperature of 150 K
while monitoring acrolein (m/z = 56) production and allyl alcohol
(m/z = 57) desorption via QMS. Atomic oxygen was adsorbed at a
sample temperature of 77 K. The flux of allyl alcohol on the surface
was 0.17 ML/s.
allyl alcohol was dosed onto the 0.15 ML O/Au(111) surface
held at 150 K (above the temperature at which the γ1 feature
appeared, but below the temperature at which the γ2 feature
appeared). In this experiment, a molecular beam of allyl alcohol
was impinged onto an inert flag held in front of the O/Au(111)
sample for 20 s (time = 20−40 s) to establish baseline signals
for acrolein and allyl alcohol and then onto the O/Au(111)
sample for 20 s (time = 70−90 s) to observe acrolein
production dynamics. An induction period of ∼6 s was
observed, as shown in Figure 4, prior to acrolein production.
At the given allyl alcohol flux rate of 0.17 ML/s, this exposure
roughly corresponds to the dosing equivalent required to
saturate the β feature (monolayer) of allyl alcohol on the clean
Au(111) surface (Figure 1). This observation may suggest that
acrolein produced on the O/Au(111) surface during the
experiment diffused away from the monolayer and desorbed
from a multilayer-like state once a sufficient coverage of allyl
alcohol had been dosed. Furthermore, the temperature at which
the γ1 feature appeared in Figure 3b coincides with the
desorption feature of acrolein multilayers on the clean Au(111)
surface (Figure 2). These observations suggest that the γ1
acrolein production feature in Figure 3b was the result of
acrolein desorption from a multilayer-like state, and the process
responsible for its production was desorption-limited.
We suggest that, for multilayer coverages of allyl alcohol on
the 0.15 ML O/Au(111) surface, acrolein was produced at or
below the temperature region associated with the γ1 feature,
and diffused into the overlying multilayers of allyl alcohol,
desorbing from a multilayer-like state. To test this mechanism
further, we carried out an experiment in which 0.6 ML of
acrolein was dosed on the clean Au(111) surface, followed by
3.6 ML of allyl alcohol before carrying out TPD (Figure S2).
Despite having been dosed directly onto the gold surface,
acrolein desorbed at a lower temperature than allyl alcohol in a
In addition to the molecular water desorption feature, a small
shoulder at 165 K is also evident in each of the water
production spectra in Figure 3. When oxidized over O/
Au(111) surfaces, simple alcohols have been shown to display
water production features at ∼175 K, which are characteristic of
hydroxyl recombination.22−24 The shoulders observed in the
spectra in Figure 3 may correlate to water produced via
hydroxyl recombination on this surface. In addition to the
features at 155 and 165 K, each water production spectrum in
Figure 3 displayed an additional feature at higher temperature.
For the 0.02 ML O/Au(111) surface, only one acrolein
feature was observed, with a peak at 215 K, as displayed in
Figure 3a. This peak shifted upward in temperature on the 0.15
ML oxygen surface to 225 K, as observed in the feature which
we have labeled γ3 in Figure 3b (discussion of γ1 and γ2 features
to follow). The acrolein production feature displayed in Figure
3a and the γ3 feature in Figure 3b coincided with the high-
temperature desorption features for allyl alcohol and water.
These acrolein production features appeared higher in
temperature than both the multilayer (peaked at 133 K) and
the monolayer (peaked at 166 K) features for acrolein
desorption from the clean Au(111) surface (Figure 2). These
observations indicate that acrolein production associated with
the feature in Figure 3a and the γ3 feature in Figure 3b occurs
via a reaction-limited kinetic process on the gold surface.
Similar TPD features observed in the selective oxidation of
simple alcohols to aldehydes over O/Au(111) have been
attributed to the decomposition of an alkoxide intermedi-
ate.22,23
Two additional features, which we label γ1 and γ2, became
evident in the TPD spectrum for acrolein production when the
oxygen coverage was increased from 0.02 to 0.15 ML (Figure
6493
dx.doi.org/10.1021/ja502347d | J. Am. Chem. Soc. 2014, 136, 6489−6498

Journal of the American Chemical Society
Article
single feature characteristic of multilayer desorption. Monolayer
acrolein desorption was not observed during this TPD
experiment, supporting our assertion that acrolein associated
with the γ1 production feature diffused into the allyl alcohol
multilayers prior to desorption.
presence of water and hydroxyl species has been shown to
significantly affect catalytic processes over gold surfaces.34,35
We performed DFT calculations to further investigate the
influence of oxygen and various intermediates on acrolein
production over the Au(111) surface. As mentioned previously,
allyl alcohol is readily dehydrogenated to yield allyloxide via
removal of its hydroxyl hydrogen by either atomic oxygen or
hydroxyl species on the gold surface. The second dehydrogen-
ation step (removal of an α-hydrogen from the allyloxide) is
facile if atomic oxygen is present on the surface, occurring with
an activation energy of 10.6 kJ/mol, whereas the process is less
facile if a surface hydroxyl species is the oxidizing agent (Ea =
36.5 kJ/mol). The reaction energy diagrams for these processes
are depicted in Figure S1. We expect that dehydrogenation of
allyloxide to generate acrolein and surface hydrogen would be a
highly unfavorable process, as the instability of the gold−
hydride has been well documented.80,81 Furthermore, this
dehydrogenation process was shown to be unfavorable for
methoxide on the Au(111) surface.75 Therefore, we suggest
that the γ1 acrolein production feature is the result of α-
hydrogen removal from allyloxide via surface oxygen adatoms,
as shown in Scheme 2a. The low activation barrier of only 10.6
The intermediate-temperature acrolein production feature,
denoted γ2, displayed a peak at 163 K as shown in Figure 3b.
This peak temperature roughly coincided with the desorption
peak of the acrolein monolayer on the clean Au(111) surface
(Figure 2). The γ2 feature may result from acrolein produced at
lower temperature desorbing from a monolayer-like state or
from acrolein production due to the decomposition of reactive
intermediates on the surface. The results displayed in Figure S2
would seem to suggest the latter because the clean Au(111)
surface populated with acrolein and allyl alcohol did not display
the monolayer feature in the acrolein desorption spectrum.
Therefore, we would expect acrolein produced at low
temperature to desorb from a multilayer-like state, as suggested
for the γ1 feature.
The observation of both reaction-limited and desorption-
limited kinetic processes indicates that the partial oxidation of
allyl alcohol to acrolein proceeds via multiple reaction pathways
over the O/Au(111) surface. Propagation of low-temperature
features (γ1 and γ2) occurred as the oxygen coverage was
increased from 0.02 to 0.15 ML, as displayed in Figure 3. Only
the γ3 acrolein production feature was observed at the oxygen
adatom coverage of 0.02 ML (Figure 3a). The γ1 and γ2 features
grew as oxygen coverage was increased, while the size of the γ3
feature did not significantly change (Figure 3b). These
observations suggest that the surface arrangement of oxygen
adatoms is instrumental in dictating the oxidation kinetics of
allyl alcohol on O/Au(111).
Scheme 2. Proposed Mechanistic Pathways for α-
Dehydrogenation of Allyloxide To Generate Acrolein on the
a
O/Au(111) Surface
DFT and UHV studies have shown that oxygen adatoms on
the Au(111) surface can induce significant changes to the gold
surface morphology.76,77 Several types of oxygen−gold
interactions are known to occur on this surface, resulting in
variations in the reactivity of the oxygen. The most reactive
form of oxygen on the Au(111) surface, chemisorbed oxygen,
binds in the 3-fold coordination sites.78 Chemisorbed oxygen
dominates for exposures at low coverages and low surface
temperatures.78 The other forms of oxygen on gold, surface
oxide and bulk oxide, are less reactive and result from stronger
interactions between gold and oxygen atoms. Generation of
gold oxide causes significant restructuring of the (111) surface,
resulting in the release of gold atoms into nanostructures that
assemble on the surface.79 High oxygen coverages and high
surface temperatures have been shown to favor the formation of
gold oxides.78 The allyl alcohol oxidation experiments carried
out in this study all utilized relatively low oxygen coverages
dosed at low surface temperature. It is unlikely that significant
amounts of gold oxide were generated on the surface in our
experiments. A study from Baker et al. showed that, even with
dosing at 200 K, only chemisorbed oxygen was detected on the
Au(111) surface until the oxygen coverage was increased
beyond 0.5 ML.77 We suggest that under the conditions
employed in our study, only chemisorbed oxygen adatoms were
present on all of the oxygen-precovered Au(111) surfaces.
Initial oxygen coverages may prove to direct relative
populations of reactive intermediates on the Au(111) surface,
and adsorbed allyloxide species may react differently with
atomic oxygen adatoms and hydroxyl species on the Au(111)
surface, causing changes to the patterns of reactivity as surface
oxygen coverage is increased. It should be noted that the
a
Activation barriers (Ea) were determined via DFT calculations.
kJ/mol for this process would allow it to occur at temperatures
consistent with desorption-limited acrolein production. Also,
this process would not be expected to occur at very low oxygen
coverages, as observed in Figure 3a. At low oxygen coverages, a
single allyl alcohol species would be unlikely to encounter a
cluster of oxygen adatoms on the surface. Each oxygen would
be converted to a hydroxyl species via removal of a hydroxyl
hydrogen from allyl alcohol in the allyloxide formation process
prior to interaction with allyloxide on the surface.
We suggest that the γ2 acrolein production feature is also due
to allyloxide dehydrogenation via surface oxygen; however, in
6494
dx.doi.org/10.1021/ja502347d | J. Am. Chem. Soc. 2014, 136, 6489−6498

Journal of the American Chemical Society
Article
this case the surface oxygen is generated by the recombination
of two hydroxyl species, producing H2O and an oxygen adatom,
as depicted in Scheme 2b. We calculated the activation barrier
for this process to be 17.3 kJ/mol (reaction energy diagram in
Figure S1). After hydroxyl recombination, the resulting oxygen
adatom would be free to induce α-hydrogen removal via the
mechanism suggested in Scheme 2a. In this sense, the hydroxyl
species acts to poison the surface toward the low-temperature
oxidation pathway, inhibiting this process until sufficient energy
has been provided to cause hydroxyl recombination and
regenerate oxygen adatoms. The temperature at which the γ2
feature appeared in Figure 3b is the approximate onset
temperature for hydroxyl recombination on the Au(111)
surface.38
the surface hydroxyl O−H bond from 1.00 Å on the clean
Au(111) surface to 1.05 Å in the transition state for this
dehydrogenation process. This elongation increases the
oxidizing potential of the hydroxyl species and allows
abstraction the α-hydrogen from the reactant allyloxide to
occur more readily. For this process, we would expect
production of both water and allyl alcohol to be observed in
addition to acrolein since hydrogen transfer between the
resulting surface species would be facile. Processes resulting in
the generation of water and allyloxide (Ea = 0.1 kJ/mol) or allyl
alcohol and a hydroxyl species (Ea = 1.5 kJ/mol) both occur
readily. Reaction energy diagrams for this process are displayed
in Figure S1.
To probe the γ3 acrolein production process in our UHV
system, we carried out allyl alcohol oxidation on the Au(111)
surface populated with deuterated hydroxyl (d-hydroxyl)
species. Ojifinni et al. showed that D2O readily forms d-
hydroxyl species upon interaction with oxygen adatoms on the
Au(111) surface.38 The authors of that study showed that, upon
heating the Au(111) surface populated with 0.08 ML of D216O
and 0.18 ML of 18O, a signal for m/z = 22 (D218O) was
observed in the corresponding TPD spectrum, indicating that
deuterium had transferred from the water that was impinged on
the surface to the surface oxygen adatoms.38 Furthermore,
significant swapping of surface oxygen atoms with oxygen from
the water was demonstrated by monitoring the TPD signals for
m/z = 32 (16O2), m/z = 34 (16O18O), and m/z = 36 (18O2). We
dosed 0.15 ML of D216O on the Au(111) surface precovered
with 0.15 ML of 18O to generate a surface populated with d-
hydroxyl species. Upon this surface, we dosed 3.6 ML of allyl
alcohol, and then we carried out TPD while monitoring the
production of acrolein, D218O, and HD18O (Figure 6). In the
Occurrence of the hydroxyl recombination pathway is further
substantiated by TPD of allyl alcohol carried out on H2O/O/
Au(111), displayed in Figure 5. Several studies have suggested
Figure 5. TPD spectra for acrolein (m/z = 56) following adsorption of
3.6 ML of allyl alcohol on the Au(111) surface populated with 0.15
ML of atomic oxygen and 0.15 ML of atomic oxygen followed by
∼0.15 ML of H2O. All species were adsorbed at a sample temperature
of 77 K.
that hydroxyl formation from water and oxygen atoms on the
Au(111) surface is a facile process that occurs at temperatures
as low as 77 K;38,82 therefore, we would expect the H2O/O/
Au(111) surface to be populated with hydroxyl species rather
than oxygen adatoms at the time of allyl alcohol dosing. When
allyl alcohol oxidation was carried out over this surface, the γ1
acrolein production feature was inhibited and the γ2 feature was
enhanced. These observations suggest that atomic oxygen is
required for the low-temperature (γ1) pathway and that
hydroxyl species are associated with the intermediate-temper-
ature (γ2) pathway, in support of our proposed mechanism.
The γ3 feature is suggested to result from removal of an α-
hydrogen from allyloxide via a hydroxyl species in a process
mediated by a second allyloxide, as shown in Scheme 2c. As
mentioned previously, the DFT-derived activation barrier for
removal of an α-hydrogen from allyloxide via a single hydroxyl
species is relatively high (36.5 kJ/mol); however, if this process
is mediated by the interaction of a second allyloxide, the
activation barrier is decreased to 29.2 kJ/mol. Reaction energy
diagrams for these processes are displayed in Figure S1. The
allyloxide mediator species partially abstracts hydrogen away
from the surface hydroxyl species, as suggested by elongation of
Figure 6. TPD spectra for acrolein (m/z = 56), HD18O (m/z = 21),
and D218O (m/z = 22) following adsorption of 3.6 ML of allyl alcohol
on the Au(111) surface populated with 0.15 ML of 18O and 0.15 ML
of D216O. All species were adsorbed at a sample temperature of 77 K.
temperature region correlating to γ3 acrolein production,
generation of HD18O (m/z = 21) was observed, while D218O
(m/z = 22) was not. These observations are consistent with the
mechanism depicted in Scheme 2c. Because the allyl alcohol
dosed onto the surface was not deuterated, we would expect
HD18O to be produced via α-hydrogen abstraction from the
resulting allyloxide, while D218O would not be generated since
the allyloxide was not deuterated. Furthermore, acrolein
6495
dx.doi.org/10.1021/ja502347d | J. Am. Chem. Soc. 2014, 136, 6489−6498

Journal of the American Chemical Society
Article
production associated with the γ3 feature was less pronounced
for the D2O/O/Au(111) surface (Figure 6) than for the H2O/
O/Au(111) surface (Figure 5), suggesting a kinetic isotope
effect (KIE) had occurred when d-hydroxyl species were
populated on the gold surface. The observation of a KIE would
be expected since abstraction of the O−H/D bond is associated
with the rate-limiting step of our proposed mechanism for the
γ3 acrolein production pathway.
Oxygen swapping from surface adatoms into the acrolein
product or the unreacted allyl alcohol was not observed in any
of the spectra taken in this study (data not shown), suggesting
that the C−O bond was not broken in the oxidation of allyl
alcohol on O/Au(111). It has been suggested that the presence
of an intact C−O bond facilitates α-hydrogen cleavage of
alkoxide species over various surfaces,83,84 further supporting
our proposed mechanisms.
Allyl alcohol and 1-propanol have similar structures but
display different TPD features upon oxidation on the O/
Au(111) surface. Although the mechanism of partial oxidation
which we have determined for allyl alcohol is similar to that
suggested for 1-propanol on the O/Au(111) surface (hydroxyl
hydrogen removal to generate a surface alkoxide followed by α-
hydrogen removal from the alkoxide),23 allyl alcohol displayed
three aldehyde production features (γ1, γ2, and γ3) when
oxidized on the O/Au(111) surface, whereas 1-propanol
displayed only one feature.23 The temperature at which the
peak for propanal, the aldehyde product of 1-propanol
oxidation, occurred was ∼200 K,23 which is analogous to the
γ3 feature we observed in Figure 3 for acrolein production. This
observation may suggest that 1-propanol and other simple
alcohols that display similar TPD features for oxidation on the
O/Au(111) surface22,23 do not observe the same set of reaction
pathways as allylic alcohols with respect to surface oxygen
adatoms and hydroxyl species on the Au(111) surface. These
differences may begin with alkoxide formation. The observation
of predominantly molecular H2O desorption for allyl alcohol
TPD on O/Au(111) rather than hydroxyl recombination as
observed for TPD of simple alcohols on O/Au(111)22−24
suggests that the allyloxide species can be formed via hydroxyl
hydrogen abstraction via either oxygen adatoms or surface
hydroxyl species. Our DFT results support this conclusion.
Simple alcohols may only generate alkoxides via hydrogen
abstraction by oxygen adatoms on the surface and not by
hydroxyl species, and the resulting differences in surface
intermediate populations may drive reaction kinetics toward
different pathways. However, this prospect is not supported in
the DFT study carried out by Chang et al. for methanol
interaction with O/Au(111), which suggested that methoxide
can be readily formed via hydroxyl hydrogen removal from
methanol by either oxygen adatoms or hydroxyl species on the
gold surface.75
The oxidation of 2-propanol to acetone on O/Au(111) was
demonstrated by Gong et al. to display features with peaks at
∼180 and 210 K.23 Elaboration as to the reason for the
observation of multiple features was not provided in this work.
It is possible that these features resulted from the oxidation of
the 2-propyl alkoxide via different surface species on the gold
surface. Unless hindered from desorption by interaction in a
surface complex, it seems unlikely that a reaction process
analogous to the γ1 pathway observed for allyl alcohol oxidation
occurred for 2-propanol oxidation. Multilayer desorption of
acetone on the clean Au(111) surface has been shown to occur
at 132 K,85 well below the onset of the acetone production
feature in the study by Gong et al.23 The observation of
multiple production features for 2-propanol oxidation may
suggest that multiple reaction pathways occurred; however,
potential differences in reaction pathway brought on by the
presence of the carbon−carbon double bond cannot be ruled
out, and the observation of multiple reaction pathways may be
unique to allylic alcohols.
The differences we observed in the reactivity of hydroxyl
species and oxygen adatoms on the Au(111) surface are of
particular interest. Zope et al. showed that hydroxyl species are
directly involved in the partial oxidation of ethanol and glycerol
to their corresponding carboxylic acids over gold catalysts in
basic aqueous media.47 These processes were suggested to
proceed first through gold-catalyzed partial oxidation to
generate the corresponding aldehydes.47 Via a combination of
DFT studies and experiments with isotopically labeled water
(H218O) and molecular oxygen (18O2), the authors showed that
oxygen from the water rather than from O2 was incorporated
into the resulting product species and suggested that O2 acted,
instead, to scavenge electrons from the gold particles after
oxidation of alcohol had occurred via a hydroxyl species,
regenerating the hydroxyl and closing the catalytic cycle.47
Our results show that oxidation of allyl alcohol via hydroxyl
species is possible but occurs with a higher activation barrier
than hydroxyl recombination on the gold surface followed by
oxidation of allyloxide via a surface oxygen adatom. The
activation barrier for oxidation of allyloxide via a hydroxyl
species is lowered by mediation from a second surface
allyloxide; therefore, we suggest that the relative proportions
of reactive intermediates on the gold surface dictate the
mechanism for allyl alcohol oxidation. For conditions under
which large concentrations of hydroxyl species are expected to
reside on the gold surface, hydroxyl recombination followed by
oxidation of allyloxide via surface oxygen would be the most
likely pathway. For conditions with low surface hydroxyl
concentrations, dehydrogenation of allyloxide via a surface
complex consisting of a hydroxyl species interacting with a
second allyloxide would be expected to dominate. Such
pathways may be broadly applicable to other alcohol species
as well.
The alkoxide intermediate present on the gold surface during
oxidation of 1-propanol may interact strongly with hydroxyl
species in a surface complex, hindering recombination and
inhibiting a potential γ2 production pathway. Alternatively,
differences in the reactivity of the α-hydrogen due to the lack of
allylic nature may inhibit the activity of the 1-propyl alkoxide
toward the γ1 production pathway. It is possible that 1-propanol
(and other simple alcohols) can undergo multiple reaction
pathways on the Au(111) surface due to the relative
populations of surface intermediates, but these pathways simply
cannot be observed in TPD spectra because the required
surface ensembles are not sampled.
4. CONCLUSIONS
Allyl alcohol adsorbs and desorbs molecularly from the clean
Au(111) surface without undergoing any chemical trans-
formations. Oxidation of allyl alcohol to acrolein occurs over
oxygen-precovered Au(111) surfaces and proceeds via a
mechanism similar to those suggested for other alcohol species
over gold catalysts: dehydrogenation of the hydroxyl hydrogen
to form a surface alkoxide followed by α-dehydrogenation of
the alkoxide to selectively generate the aldehyde product.
6496
dx.doi.org/10.1021/ja502347d | J. Am. Chem. Soc. 2014, 136, 6489−6498

Journal of the American Chemical Society
Article
(2) Haruta, M.; Kobayashi, T.; Sano, H.; Yamada, N. Chem. Lett.
1987, 405.
(3) Hutchings, G. J. J. Catal. 1985, 96, 292.
(4) Hashmi, A. S. K.; Hutchings, G. J. Angew. Chem., Int. Ed. 2006,
Oxidation of allyl alcohol to acrolein occurs via three
different reaction pathways over the Au(111) surface, dictated
by the relative populations of atomic oxygen and hydroxyl
species on the surface. Similar results have not been observed in
the TPD spectra for oxidation of simple alcohols on the
Au(111) surface, which may suggest that allylic alcohols and
simple alcohols react differently in the presence of various
surface intermediates or that aldehydes produced on the surface
during oxidation of simple alcohols by an analogous mechanism
are retained on the surface in structures that desorb at higher
temperatures.
Our observations may provide insight into the pathways for
oxidation of alcohols over gold surfaces, in both vapor- and
liquid-phase systems. Our results suggest that allyl alcohol is
most readily oxidized to acrolein via oxygen adatoms on the
Au(111) surface. If a single allyl alcohol molecule encounters
multiple surface oxygen adatoms, then this process can occur at
very low temperatures. A hydroxyl species can also oxidize the
reactive intermediate, allyloxide, but does so via a process with
a higher activation barrier. Yet another oxidative pathway
involves the generation of surface oxygen by hydroxyl
recombination, followed by oxidation via the low-temperature
pathway. This process may dominate in the high-pH conditions
under which alcohol oxidation has been shown to occur readily
over gold catalysts.
Our work may also aid in the general understanding of the
reactivity of oxygen and hydroxyl species on gold surfaces. Such
reactions are of great interest, considering the enhancement
effect many gold catalysts display upon interaction with water.
Water is present to some extent in most catalytic systems,
whether as an impurity, reactant, product, or solvent.
Accordingly, water may have unknown/unexplored effects on
a host of catalytic reactions. Therefore, thorough consideration
of the effects of water is necessary when assessing a catalytic
system, particularly when oxidation reactions are involved.
45, 7896.
(5) Bond, G. C.; Thompson, D. T. Catal. Rev.−Sci. Eng. 1999, 41,
319.
(6) Haruta, M.; Date, M. Appl. Catal. A Gen. 2001, 222, 427.
́
(7) Haruta, M. Chem. Rec. 2003, 3, 75.
(8) Gong, J. Chem. Rev. 2012, 112, 2987.
(9) Gong, J.; Mullins, C. B. Acc. Chem. Res. 2009, 42, 1063.
(10) Pan, M.; Brush, A. J.; Pozun, Z. D.; Ham, H. C.; Yu, W.-Y.;
Henkelman, G.; Hwang, G. S.; Mullins, C. B. Chem. Soc. Rev. 2013, 42,
5002.
(11) Pan, M.; Gong, J.; Dong, G.; Mullins, C. B. Acc. Chem. Res.
2014, 47, 750.
(12) Madix, R. J.; Friend, C. M.; Liu, X. J. Catal. 2008, 258, 410.
(13) Min, B. K.; Friend, C. M. Chem. Rev. 2007, 107, 2709.
(14) Rodriguez, J. A.; Senanayake, S. D.; Stacchiola, D.; Liu, P.;
Hrbek, J. Acc. Chem. Res. 2014, 47, 773.
(15) Rodriguez, J. A. Catal. Today 2011, 160, 3.
(16) Prati, L.; Rossi, M. J. Catal. 1998, 176, 552.
(17) Davis, S. E.; Ide, M. S.; Davis, R. J. Green Chem. 2013, 15, 17.
(18) Abad, A.; Concepcion
Int. Ed. 2005, 44, 4066.
́
, P.; Corma, A.; García, H. Angew. Chem.,
(19) Biella, S.; Rossi, M. Chem. Commun. 2003, 378.
(20) Mori, K.; Hara, T.; Mizugaki, T.; Ebitani, K.; Kaneda, K. J. Am.
Chem. Soc. 2004, 126, 10657.
(21) Enache, D. I.; Edwards, J. K.; Landon, P.; Solsona-Espriu, B.;
Carley, A. F.; Herzing, A. A.; Watanabe, M.; Kiely, C. J.; Knight, D. W.;
Hutchings, G. J. Science 2006, 311, 362.
(22) Gong, J.; Mullins, C. B. J. Am. Chem. Soc. 2008, 130, 16458.
(23) Gong, J.; Flaherty, D. W.; Yan, T.; Mullins, C. B.
ChemPhysChem 2008, 9, 2461.
(24) Yan, T.; Gong, J.; Mullins, C. B. J. Am. Chem. Soc. 2009, 131,
16189.
(25) Liu, X.; Friend, C. M. Langmuir 2010, 26, 16552.
(26) Liu, X.; Xu, B.; Haubrich, J.; Madix, R. J.; Friend, C. M. J. Am.
Chem. Soc. 2009, 131, 5757.
(27) Xu, B.; Madix, R. J.; Friend, C. M. J. Am. Chem. Soc. 2010, 132,
16571.
(28) Xu, B.; Liu, X.; Haubrich, J.; Friend, C. M. Nat. Chem. 2010, 2,
61.
(29) Deng, X.; Min, B. K.; Liu, X.; Friend, C. M. J. Phys. Chem. B
2006, 110, 15982.
(30) Abad, A.; Almela, C.; Corma, A.; García, H. Chem. Commun.
2006, 3178.
ASSOCIATED CONTENT
* Supporting Information
■
S
Displacement of H216O from walls of UHV chamber by allyl
alcohol, acrolein and allyl alcohol desorption from the clean
Au(111) surface, and DFT-calculated reaction energy diagrams
for acrolein production on the Au(111) surface. This material is
available free of charge via the Internet at http://pubs.acs.org.
(31) Davis, J. L.; Barteau, M. A. J. Mol. Catal. 1992, 77, 109.
(32) Shekhar, R.; Barteau, M. A. Surf. Sci. 1994, 319, 298.
(33) Abad, A.; Corma, A.; García, H. Pure Appl. Chem. 2007, 79,
1847.
(34) Mullen, G. M.; Gong, J.; Yan, T.; Pan, M.; Mullins, C. B. Top.
Catal. 2013, 56, 1499.
AUTHOR INFORMATION
Corresponding Author
mullins@che.utexas.edu
■
Notes
The authors declare no competing financial interest.
(35) Ide, M. S.; Davis, R. J. Acc. Chem. Res. 2014, 47, 825.
́
(36) Date, M.; Okumura, M.; Tsubota, S.; Haruta, M. Angew. Chem.,
Int. Ed. 2004, 43, 2129.
ACKNOWLEDGMENTS
■
(37) Ojeda, M.; Zhan, B.-Z.; Iglesia, E. J. Catal. 2012, 285, 92.
(38) Ojifinni, R. A.; Froemming, N. S.; Gong, J.; Pan, M.; Kim, T. S.;
White, J. M.; Henkelman, G.; Mullins, C. B. J. Am. Chem. Soc. 2008,
130, 6801.
(39) Kim, T. S.; Gong, J.; Ojifinni, R. A.; White, J. M.; Mullins, C. B.
J. Am. Chem. Soc. 2006, 128, 6282.
(40) Gao, F.; Wood, T. E.; Goodman, D. W. Catal. Lett. 2009, 134, 9.
(41) Bond, G. C.; Thompson, D. T. Gold Bull. 2000, 33, 41.
(42) Schubert, M. M.; Venugopal, A.; Kahlich, M. J.; Plzak, V.; Behm,
R. J. J. Catal. 2004, 222, 32.
(43) Huang, J.; Akita, T.; Faye, J.; Fujitani, T.; Takei, T.; Haruta, M.
Angew. Chem., Int. Ed. 2009, 48, 7862.
We are thankful for the generous support of the Department of
Energy (DE-FG02-04ER15587 [C.B.M.] and DE-FG02-
13ER16428 [G.H.]) and the Welch Foundation (Grants F-
1436 [C.B.M.] and F-1841 [G.H.]). G.M.M. acknowledges the
National Science Foundation for a Graduate Research Fellow-
ship. Cara R. Touretzky and Richard C. Pattison are gratefully
acknowledged for aid with instrument optimization.
REFERENCES
■
(1) Bond, G. C.; Sermon, P. A.; Webb, G.; Buchanan, D. A.; Wells, B.
J. Chem. Soc., Chem. Commun. 1973, 5.
6497
dx.doi.org/10.1021/ja502347d | J. Am. Chem. Soc. 2014, 136, 6489−6498

Journal of the American Chemical Society
Article
́
(44) Lee, S.; Molina, L. M.; Lopez, M. J.; Alonso, J. A.; Hammer, B.;
Lee, B.; Seifert, S.; Winans, R. E.; Elam, J. W.; Pellin, M. J.; Vajda, S.
Angew. Chem., Int. Ed. 2009, 48, 1467.
(45) Ojeda, M.; Iglesia, E. Chem. Commun. 2009, 352.
(46) Ueda, A.; Oshima, T.; Haruta, M. Appl. Catal. B Environ. 1997,
12, 81.
(47) Zope, B. N.; Hibbitts, D. D.; Neurock, M.; Davis, R. J. Science
2010, 330, 74.
(48) Carrettin, S.; McMorn, P.; Johnston, P.; Griffin, K.; Hutchings,
G. J. Chem. Commun. 2002, 696.
(49) Jorgensen, B.; Christiansen, S. E.; Thomsen, M. L. D.;
Christensen, C. H. J. Catal. 2007, 251, 332.
(50) Ketchie, W. C.; Fang, Y.-L.; Wong, M. S.; Murayama, M.; Davis,
R. J. J. Catal. 2007, 250, 94.
(51) Wheeler, M. C.; Seets, D. C.; Mullins, C. B. J. Chem. Phys. 1996,
105, 1572.
(52) Pollard, J. E. Rev. Sci. Instrum. 1992, 63, 1771.
(53) Wheeler, M. C.; Seets, D. C.; Mullins, C. B. J. Chem. Phys. 1997,
107, 1672.
(54) Kang, H. C.; Mullins, C. B.; Weinberg, W. H. J. Vac. Sci. Technol.
A 1990, 8, 2538.
(55) Mullins, C. B.; Weinberg, W. H. J. Chem. Phys. 1990, 92, 3986.
(56) Rettner, C. T.; Mullins, C. B.; Bethune, D. S.; Auerbach, D. J.;
Schweizer, E. K.; Weinberg, W. H. J. Vac. Sci. Technol. A 1990, 8, 2699.
(57) Kresse, G.; Hafner, J. Phys. Rev. B 1993, 47, 558.
(58) Kresse, G.; Hafner, J. Phys. Rev. B 1994, 49, 14251.
(59) Kresse, G.; Furthmuller, J. Phys. Rev. B 1996, 54, 11169.
̈
(60) Kresse, G.; Furthmiiller, J. Comput. Mater. Sci. 1996, 6, 15.
(61) Blochl, P. Phys. Rev. B 1994, 50, 17953.
̈
(62) Hohenberg, P.; Kohn, W. Phys. Rev. 1964, 136, B864.
(63) Kohn, W.; Sham, L. J. Phys. Rev. 1965, 140, A1133.
(64) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77,
3865.
(65) Klimes,
195131.
(66) Henkelman, G.; Uberuaga, B. P.; Jon
2000, 113, 9901.
(67) Henkelman, G.; Jon
̌
J.; Bowler, D. R.; Michaelides, A. Phys. Rev. B 2011, 83,
́
sson, H. J. Chem. Phys.
́
sson, H. J. Chem. Phys. 2000, 113, 9978.
(68) Monkhorst, H. J.; Pack, J. D. Phys. Rev. B 1976, 13, 5188.
(69) Redhead, P. A. Vacuum 1962, 12, 203.
(70) Gong, J.; Flaherty, D. W.; Ojifinni, R. A.; White, J. M.; Mullins,
C. B. J. Phys. Chem. C 2008, 112, 5501.
(71) Ehlers, D. H.; Spitzer, A.; Luth, H. Surf. Sci. 1985, 160, 57.
̈
(72) Burke, D. J.; Wolff, A. J.; Edridge, J. L.; Brown, W. A. J. Chem.
Phys. 2008, 128, 104702.
(73) Brown, N. F.; Barteau, M. A. J. Am. Chem. Soc. 1992, 114, 4258.
(74) Kay, B. D.; Lykke, K. R.; Creighton, J. R.; Ward, S. J. J. Chem.
Phys. 1989, 91, 5120.
(75) Chang, C.-R.; Yang, X.-F.; Long, B.; Li, J. ACS Catal. 2013, 3,
1693.
(76) Min, B. K.; Alemozafar, A. R.; Biener, M. M.; Biener, J.; Friend,
C. M. Top. Catal. 2005, 36, 77.
(77) Baker, T. A.; Xu, B.; Liu, X.; Kaxiras, E.; Friend, C. M. J. Phys.
Chem. C 2009, 113, 16561.
(78) Baker, T. A.; Liu, X.; Friend, C. M. Phys. Chem. Chem. Phys.
2011, 13, 34.
(79) Min, B. K.; Alemozafar, A. R.; Pinnaduwage, D.; Deng, X.;
Friend, C. M. J. Phys. Chem. B 2006, 110, 19833.
(80) Hammer, B.; Nørskov, J. K. Nature 1995, 376, 238.
(81) Tierney, H. L.; Baber, A. E.; Kitchin, J. R.; Sykes, E. C. H. Phys.
Rev. Lett. 2009, 103, 246102.
(82) Quiller, R. G.; Baker, T. A.; Deng, X.; Colling, M. E.; Min, B. K.;
Friend, C. M. J. Chem. Phys. 2008, 129, 064702.
(83) Xu, X.; Friend, C. M. Surf. Sci. 1992, 260, 14.
(84) Davis, J. L.; Barteau, M. A. Surf. Sci. 1988, 197, 123.
(85) Syomin, D.; Koel, B. E. Surf. Sci. 2002, 498, 53.
6498
dx.doi.org/10.1021/ja502347d | J. Am. Chem. Soc. 2014, 136, 6489−6498