Partially oxidized gold nanoparticles: A catalytic base-free system for the aerobic homocoupling of alkynes

Article abstract of DOI:10.1016/j.jcat.2014.04.003

The mechanism of alkyne homocoupling over gold nanoparticles and clusters, isolated and supported on CeO_2, has been theoretically investigated by means of periodic DFT calculations. The theoretical study indicates that O₂ dissociation on gold generates basic O atoms able to abstract the proton of the alkyne, and cationic Au^δ+ and Au⁺ species that decrease the activation barrier for the CC bond forming step. Kinetic results show that the base-free homocoupling of alkynes is effectively catalyzed by gold nanoparticles supported on different solids, and confirm the theoretical prediction that the dissociation of oxygen on the gold nanoparticle is the controlling step of the global reaction.

Full text of DOI:10.1016/j.jcat.2014.04.003

Journal of Catalysis 315 (2014) 6–14
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Partially oxidized gold nanoparticles: A catalytic base-free system
for the aerobic homocoupling of alkynes
Mercedes Boronat a, Siris Laursen a,b, Antonio Leyva-Pérez a, Judit Oliver-Meseguer a,
Diego Combita a, Avelino Corma a,
⇑
a Instituto de Tecnología Química, Universidad Politécnica de Valencia, Consejo Superior de Investigaciones Científicas, Av. de los Naranjos, s/n, 46022 Valencia, Spain
b Department of Chemical and Biomolecular Engineering, University of Tennessee, Knoxville, TN 37996-2200, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
The mechanism of alkyne homocoupling over gold nanoparticles and clusters, isolated and supported on
CeO2, has been theoretically investigated by means of periodic DFT calculations. The theoretical study
indicates that O2 dissociation on gold generates basic O atoms able to abstract the proton of the alkyne,
and cationic Aud+ and Au+ species that decrease the activation barrier for the CAC bond forming step.
Kinetic results show that the base-free homocoupling of alkynes is effectively catalyzed by gold nanopar-
ticles supported on different solids, and confirm the theoretical prediction that the dissociation of oxygen
on the gold nanoparticle is the controlling step of the global reaction.
Received 20 February 2014
Revised 2 April 2014
Accepted 3 April 2014
Keywords:
Gold
Heterogeneous catalysis
DFT
Ó 2014 Elsevier Inc. All rights reserved.
Oxidative coupling of alkynes
Reaction mechanism
CAH activation
1. Introduction
(PA) on Au/CeO2 catalysts and found that PA can be activated at
cationic gold sites at the metal-support interface, producing a
The synthesis of 1,3-diynes, which are important building
blocks in fine chemistry, pharmaceuticals and bioactive com-
pounds, has been traditionally carried out via oxidative homocou-
pling of terminal alkynes. Copper salts assisted by a base, as well as
catalytic systems formed by a combination of palladium and
copper salts have been extensively used in homogeneous phase
[1–4]. The development of heterogeneous catalysts for the homo-
coupling of alkynes, with the advantages of easy separation and
recycling, is interesting from an economic and environmental
points of view, and some examples exist in the literature reporting
the catalytic activity of copper-containing hydrotalcites [5,6],
zeolites [7,8], and Cu(OH)x supported on different transition metal
oxides [9,10]. It has also been reported that alkyne homocoupling
occurs under typical conditions for Sonogashira cross-coupling
reactions, both in homogeneous phase using Pd organocatalysts
[11,12] and over heterogeneous Au/CeO2 [13] and Au/La2O3 [14]
catalysts. We recently investigated the mechanism of the Sono-
gashira reaction between iodobenzene (IB) and phenylacetylene
non-negligible amount of diphenyldiacetylene (DPDA), i.e., the
product of PA homocoupling [15]. In this work, we combine DFT
calculations with kinetic experiments to firstly determine the
mechanism of PA homocoupling on heterogeneous gold catalysts
and, consequently, to identify the nature of the gold active sites
required to prepare an optimized heterogeneous catalyst for the
homocoupling of alkynes.
2. Experimental section
2.1. Models and methods
The mechanism of PA homocoupling was investigated by means
of periodic density functional theory, using the Perdew–Wang
(PW91) exchange–correlation functional within the generalized
gradient approach (GGA) [16,17] as implemented in the VASP code
[18,19]. The valence density was expanded in a plane wave basis
set with a kinetic energy cutoff of 500 eV, and the effect of the core
electrons in the valence density was taken into account by means
of the projected augmented wave (PAW) formalism [20].
⇑
Corresponding author. Fax: +34 9638 77809.
Integration in the reciprocal space was carried out at the
C k-point
E-mail addresses: boronat@itq.upv.es (M. Boronat), slaursen@utk.edu (S. Laursen),
anleyva@itq.upv.es (A. Leyva-Pérez), joliverm@itq.upv.es (J. Oliver-Meseguer),
diecmmer@posgrado.upv.es (D. Combita), acorma@itq.upv.es (A. Corma).
of the Brillouin zone. Transition states were located using the
DIMER [21,22] and NEB [23] algorithms, and stationary points
http://dx.doi.org/10.1016/j.jcat.2014.04.003
0021-9517/Ó 2014 Elsevier Inc. All rights reserved.

M. Boronat et al. / Journal of Catalysis 315 (2014) 6–14
7
were characterized by pertinent frequency analysis calculations.
Charge distributions were estimated using the theory of atoms in
molecules (AIM) of Bader using the algorithm developed by
Henkelman [24,25]. In the case of modeling CeO2, the Hubbard
repulsion term (U) with a value of 5 eV was applied to the cerium
atoms in all calculations [26].
Neutral gold clusters and nanoparticles were modeled by
means of an Au3 system formed by three gold atoms and by an
Au38 system having a typical cuboctahedral shape and ꢀ1 nm
diameter, respectively. A partially oxidized gold nanoparticle was
obtained by adding 16 O atoms to the surface of the Au38 system,
generating a Au38O16 model previously described [27]. These mod-
els were placed in a 25 Â 25 Â 25 Å3 cubic box, large enough as to
avoid interactions between periodically repeated nanoparticles or
adsorbates, and during the geometry optimizations, only the posi-
tions of the adsorbates and of the O atoms directly involved in the
mechanism were allowed to fully relax.
different pressures of oxygen or at different concentrations of
alkyne. The points of the kinetic curves were obtained from inde-
pendent batch experiments by stopping the reaction at different
reaction times. For that, the solid catalyst (Au/C or Au/CeO2:
41 mg, Au/ZnO: 65 mg; 2 mol% Au) was placed in a thick double-
walled 2.5-ml glass reactor equipped with a magnetic stirrer. Then,
1,3-dichlorobenzene (0.5 ml) and the corresponding amount of
ortho-tolylacetylene were added and the vial capped with the
pressure system. Neat molecular oxygen gas was loaded through
a needle until the manometer indicated the corresponding pres-
sure. Then, the reactor was magnetically stirred in a pre-heated
oil bath at 170 °C for the desired time. After this time, the reactor
was cooled with water and the gas released. Diethyl ether (1 ml)
was added and the whole mixture was filtered and submitted to
GC and GC–MS analysis after dodecane (22 ll, 0.1 mmol) was
added as an external standard. The yield of homocoupling of
ortho-tolylacetylene was plotted versus time and the slope at low
A stoichiometric CeO2 (111) surface was modeled by means of a
(4 Â 3) supercell slab containing 64 O atoms and 32 Ce atoms, and
then, four different models were generated to simulate Au nano-
particles supported on CeO2: (a) AuOx/CeO2 model: a two atomic
layer thick Au nano-rod containing 20 Au atoms was placed on
the stoichiometric CeO2 (111) surface and six additional oxygen
atoms were added between the Au nano-rod and the CeO2 surface
Ce4+ sites. This model contains both metallic Au0 and cationic Aud+
species at the metal–support interface. (b) Au2O3/CeO2 model: a
gold oxide strip consisting of 6 Au atoms and 9 O atoms was placed
on the stoichiometric CeO2 (111) surface, generating a system con-
taining cationic Au+ and Au3+ atoms. (c) Au10/CeO2 model: a small
cluster consisting of ten Au atoms arranged in two layers having
7 and 3 Au atoms, respectively, was placed on the stoichiometric
CeO2 (111) surface, resulting in a supported Au nanoparticle con-
taining metallic low coordinated Au0 atoms, and (d) Au9O7/CeO2
model: a small cluster consisting of nine Au atoms was placed on
the stoichiometric CeO2 (111) surface, and 7 O atoms were added
at the metal-support interface, generating a system containing
three low-coordinated metallic Au0 atoms on top of the particle
and four cationic Au+ sites at the metal-support interface. To obtain
the optimized structures of these models, the Ce and O atoms in
the bottom layer of each system were kept fixed in their bulk opti-
mized positions, while the atomic positions of the rest of the atoms
were fully relaxed. In the study of the mechanism of PA homocou-
pling, only the positions of the adsorbates and of the O atoms
directly involved in the mechanism were allowed to fully relax
during the geometry optimizations.
yields gives the initial rate in hÀ1
.
2.2.3. Leaching studies
Experiments were carried out following the procedure above
detailed but at different reaction times. After representing these
data in a plot-time yield graphics, a second set of experiments
was carried out as follows: at c.a. 25% conversion, the gas was
released and the solid catalyst was filtered at the reaction temper-
ature collecting the filtrates in a second reactor equipped with a
magnetic bar. This reactor was closed and placed under the same
magnetically stirred bath oil at 170 °C after oxygen atmosphere
was re-loaded (0.75 mmol). This protocol was followed for each
data point. After the desired time, diethyl ether (1 ml) was added
and the mixture was filtered and submitted to GC and GC–MS
analysis after dodecane (5.6 ll, 0.025 mmol) was added as external
standard. Comparison among the curves gives estimation about
the presence of catalytically active species in solution. The results
obtained for the batch reactions are similar to those corresponding
to single reactions in which aliquots were periodically taken after
release of the oxygen atmosphere and re-load.
2.2.4. Reusability studies
Experiments were carried out following the procedure above
detailed but recovering the solid catalyst by decantation instead
of filtration after diethyl ether (1 ml) addition. The solid was then
washed three times with diethyl ether (1 ml), dried, and used in a
second run.
3. Results and discussion
2.2. Experimental
3.1. Theoretical study of reaction mechanism on gold catalyst models
2.2.1. Catalyst screening and scope
The solid catalyst (Au/Carbon and Au/CeO2: 41 mg; Au/TiO2, Au/
ZnO, Au/Al2O3 and Au/Fe2O3: 65 mg, 2 mol% Au) was placed in a
thick double-walled 2.5-ml glass reactor equipped with a magnetic
stirrer. Then 1,3-dichlorobenzene (0.5 ml) and the corresponding
alkyne (0.25 mmol) were added and the vial capped with the pres-
sure system. Molecular oxygen gas was loaded through a needle
until the manometer indicated c.a. 5 bars (0.75 mmol). Then, the
reactor was placed stirred in a pre-heated oil bath at 170 °C and
magnetically stirred for 18 h. After this time, the reactor was
cooled with water and the gas released. Diethyl ether (1 ml) was
added and the mixture was filtered and submitted to GC and
In
a first step, we investigated the mechanism of PA
homocoupling on a small Au3 cluster and on two different models
of isolated gold nanoparticles, that is, a Au38 system containing
only metallic low coordinated Au0 sites (see Table 1), and a
Table 1
Calculated Bader charges on selected Au atoms in different gold catalyst models.
Model
Au0
Aud+
Au+
Au3+
Au3
Au38
ꢀ0
ꢀ0
–
–
–
–
–
–
–
–
GC–MS analysis after dodecane (22
external standard.
ll, 0.1 mmol) was added as
Au2O, Au2O3
Au38O16
AuOx/CeO2
Au2O3/CeO2
Au10/CeO2
Au9O7/CeO2
0.47
1.19
ꢀ0
0.21
0.33
0.47
0.65
ꢀ0.76
–
ꢀ0
ꢀ1.14
2.2.2. Initial rate measurements
Initial rates were calculated from the slope of the first-order
kinetic curves of the homocoupling of ortho-tolylacetylene under
ꢀ0
ꢀ0
ꢀ0.64

8
M. Boronat et al. / Journal of Catalysis 315 (2014) 6–14
partially oxidized gold nanoparticle containing three types of sites:
metallic Au0 centers, slightly positively charged Aud+ sites directly
bonded to one O atom, and cationic Au+ centers in which the Au
atom is directly bonded to two O atoms forming a linear OAAuAO
structure, as described in previous work [27]. The calculated
adsorption, activation and reaction energies, schematized in
Fig. 1, are summarized in Table 2 and the optimized geometries
of the structures involved in the mechanism are depicted in
Figs. 2–4.
PA adsorption and dissociation on isolated neutral gold nano-
particles was investigated as one of the elementary steps in the
Sonogashira reaction between PA and IB [15]. It was found that
PA adsorbs strongly on low coordinated neutral Au0 atoms at
corner or edge sites, but the dissociation process is endothermic
by 20 kcal/mol and involves an activation energy of almost
40 kcal/mol (see Table 2). As previously described, the presence
of a base (carbonate) lowers the barrier to 9.0 kcal/mol, and the
process becomes exothermic by 7.7 kcal/mol. After dissociation
on an isolated Au38 nanoparticle, the phenylacetylenyl fragment
is placed on top of a low coordinated Au atom, while the H atom
occupies a bridge position between two Au atoms (see Fig. 2).
The surface coupling step occurs through a transition state in
which the CAC bond is formed while the two organic fragments
are still quite strongly bonded to the Au atoms, as indicated by
the optimized values of the CAAu distances, 2.04 Å. The calculated
activation energy for this elementary step is 25 kcal/mol and DPDA
formation and desorption releases 8.5 kcal/mol.
It has been recently reported that small gold clusters are highly
efficient in the activation of the C„C bond in alkynes [28,29] and,
in order to check the relevance of this effect on the mechanism of
alkyne homocoupling, PA adsorption and dissociation over a small
Au3 cluster was also investigated. As depicted in Fig. 3, PA interacts
strongly with the low coordinated atoms of the Au3 cluster produc-
ing an activation and weakening of the C„C bond reflected in an
increase in the CC bond length of 0.05 Å. However, this interaction
does not favor the desired dissociation of the CAH bond, which is
endothermic by 21 kcal/mol and involves an activation barrier of
30 kcal/mol. Moreover, the calculated activation energy for the
coupling of two phenylacetylenyl fragments attached to the Au3
cluster is as high as 27.5 kcal/mol, indicating that such small clus-
ters should not be active for alkyne homocoupling.
calculated reaction energy for PA deprotonation, 10.8 kcal/mol, is
not as endothermic as on the Au38 model, but we should take into
account that it also includes the energy gain due to formation of a
OAH bond. To compare the stability of phenylacetylenyl fragments
directly bonded to metallic or cationic Au sites, a Au38O2 model
was used that contains metallic Au0 and cationic Aud+ and Au+ sites
(see Fig. 5 and Table 1). Phenylacetylenyl interacts much stronger
with metallic Au0 atoms in Au38 and Au38O2 models than with cat-
ionic Aud+ centers and Au+ sites, following the linear trend depicted
in Fig. 5. This result suggests a higher reactivity of phenylacetyle-
nyl fragments bonded to cationic Aud+ and Au+ sites and, indeed,
the calculated activation energies for the surface coupling step
are in this line. The barrier obtained on the Au38O16 model, in
which the two reactant phenylacetylenyl fragments are directly
interacting with cationic Aud+ sites, is 18 kcal/mol, 7 kcal/mol
lower than on the metallic Au38 system.
Besides PA deprotonation and surface coupling steps, O2 disso-
ciation generating adsorbed O atoms and cationic Aud+ and Au+
sites is necessary to close the catalytic cycle. It should be remarked
at this point that the activation barriers for O2 dissociation over
gold nanoparticles vary considerably with particle size and shape
[27,30,31]. Thus, for instance, the activation barriers for successive
O2 dissociation over the most favorable site of cuboctahedral Au38
NPs leading to formation of a gold oxide over layer are around
8 kcal/mol or less [27]. But when the process takes place at edge
sites also present in Au38 and in larger, smaller or irregular gold
nanoparticles, the activation barriers increase by 10–20 kcal/mol
[30,31], suggesting that O2 dissociation might be the rate deter-
mining step for the global homocoupling reaction.
Next, we considered four different models for the Au/CeO2
catalysts containing different types of neutral and cationic sites:
(a) a gold nanorod on a partially oxidized CeO2(111) surface
(AuOx/CeO2 model, see Computational Details section) that con-
tains both Au0 and Aud+ species; (b) a Au2O3 strip on stoichiometric
CeO2(111) that only contains cationic Au+ and Au3+ atoms (Au2O3/
CeO2 model); (c) a Au10 cluster on stoichiometric CeO2(111) that
only contains metallic low coordinated Au0 sites (Au10/CeO2
model); and (d) a Au9 cluster on a partially oxidized CeO2(111)
surface (Au9O7/CeO2 model) that contains low coordinated metal-
lic Au0 and cationic Au+ species, as indicated by the Bader charges
listed in Table 1.
The effect of cationic gold sites and basic oxygen atoms is more
relevant. PA adsorption at slightly positive Aud+ sites and at
cationic Au+ centers present in the partially oxidized Au38O16
model is energetically favored (À7.3 and À4.1 kcal/mol at Aud+
and Au+, respectively) and the activation barrier for its deprotona-
tion (14.4 and 12.6 at Aud+ and Au+, respectively) is considerably
lower than on the pure metallic Au38 nanoparticle, because it is
assisted by a basic O atom from the gold oxide over layer. After dis-
sociation, the phenylacetylenyl fragment is occupying a bridge
position between a Au0 and a Aud+ (or Au+) site (see Fig. 4). The
PA was initially adsorbed at neutral Au0 and cationic Aud+, Au+
and Au3+ atoms present in the different Au/CeO2 models, and the
resulting optimized structures and calculated adsorption energies
are given in Fig. 6 and Table 2, respectively. It was found that PA
cannot adsorb on cationic Au3+ atoms present in the Au2O3/CeO2
model, but interacts with Au+ sites in the same system, with calcu-
lated adsorption energies ranging between À6.7 and À9.5 kcal/
mol. A weaker interaction was found with slightly positive Aud+
sites present in AuOx/CeO2 and Au9O7/CeO2 models, with calcu-
lated adsorption energies around À3 kcal/mol. And, as found for
isolated gold nanoparticles, the stronger interaction is obtained
for PA adsorption at low coordinated metallic Au0 sites present
in Au10/CeO2 and Au9O7/CeO2 models, with calculated adsorption
energies ranging between À9 and À13 kcal/mol.
PA dissociation on metallic Au0 sites was not investigated on
Au/CeO2 models based on the high activation barrier obtained on
the isolated Au38 nanoparticle, but was calculated on all cationic
Au sites. In all cases, the activation energies obtained were lower
than 15 kcal/mol, and the dissociation process was found to be
exothermic. However, it is not possible to directly compare the
stability of the phenylacetylenyl fragments from calculated
DEPA
reaction energies on the different models because the dissociation
process always involves formation of a OAH bond whose stability
depends on the basicity of the O atom involved. To avoid this prob-
lem in the calculation of the activation barriers for the bimolecular
Fig. 1. Schematic representation of energy profiles for PA dissociation (left) and for
bimolecular coupling step (right) in gold catalyzed PA homocoupling.

M. Boronat et al. / Journal of Catalysis 315 (2014) 6–14
9
Table 2
Adsorption, activation and reaction energies (in kcal/mol) calculated for the elementary steps schematized in Fig. 1 on neutral Au0 and cationic Aud+ and Au+ sites on different gold
catalyst models.
Model
Au site
Eads
Eact
Eact
CC
PA
PA
Au3
Au38
Au38O16
Au38O16
AuOx/CeO2
Au2O3/CeO2
Au10/CeO2
Au9O7/CeO2
Au9O7/CeO2
Au0
Au0
Aud+
Au+
Aud+
Au+
Au0
Au0
Au+
À37.0
À14.8
À7.3
À4.1
À2.7
À9.5
À9.0
À12.5
À3.5
30.0
39.9
14.4
12.6
8.7
27.5
25.0
18.0
12.4
38.3
16.1
33.0
9.7
18.2
2.5
Fig. 2. Optimized geometries of the species involved in the mechanism of PA homocoupling on an isolated neutral Au38 nanoparticle. Distances in Å.
atoms involved in the formation of the new CAC bond are still
strongly attached to the Au atoms, and the distance between them
is ꢀ2.2 Å. But this similarity in the optimized geometries is not
reflected in the calculated activation energies. The highest Eact CC
values are obtained on the AuOx/CeO2 and Au10/CeO2 models, 38
and 33 kcal/mol, respectively, and the reason is that the phenyl-
acetylenyl fragments are in both cases initially coordinated to
three metallic Au0 atoms, which is the most stable situation as
shown in Fig. 5. When the phenylacetylenyl fragments are initially
attached to cationic Au+ sites in the Au2O3/CeO2 model, the activa-
tion energy is considerably lower, 16.1 kcal/mol. And the most
favorable situation is found on the Au9O7/CeO2 model. In this case,
not only the reactants are less stable because they are bonded to
cationic Au+ sites, but there is also an extra stabilization in the
transition state due to formation of a new AuAC bond (see
Fig. 7d, right). As a result, the calculated activation energy is only
2.5 kcal/mol.
Fig. 3. Optimized geometries of the species involved in the mechanism of PA
activation on an isolated neutral Au3 cluster. Distances in Å.
surface coupling step, the H atoms resulting from PA dissociation
were placed in all cases on the CeO2 support, so that the stabiliza-
tion due to formation of this bond is equivalent in all systems. The
optimized geometries of the reactants and transition states for the
surface coupling step on the four different Au/CeO2 models
considered are depicted in Fig. 7, and the calculated activation
energies are listed in Table 2. In all transition states, the two carbon
O2 dissociation generating adsorbed O atoms and cationic gold
sites in supported gold catalysts has been widely studied in the lit-
erature, mainly in relation to the mechanism of low temperature
CO oxidation [32–35]. Despite the large variety of models
and methods employed, there is general consensus that molecular
O2 preferentially adsorbs at the interface between the gold

10
M. Boronat et al. / Journal of Catalysis 315 (2014) 6–14
Fig. 4. Optimized geometries of the species involved in the mechanism of PA homocoupling on an isolated partially oxidized Au38O16 nanoparticle. Distances in Å.
Fig. 5. Interaction of phenylacetylenyl radical with different metallic Au0 and cationic Aud+ and Au+ sites on isolated Au nanoparticles, and relationship between calculated
interaction energies (values in brackets, in kcal/mol) and net atomic charge on Au atom. Distances in Å.
nanoparticle and the metal oxide support, and that the presence of
either O vacancy defects in reduced surfaces or extra O atoms in
oxidized surfaces enhances reactivity by decreasing the activation
barriers for O2 dissociation[36,37].
It can be concluded from the theoretical study that PA adsorbs
at metallic Au0 and cationic Aud+ and Au+ sites, and that its depro-
tonation is energetically feasible in all systems having O atoms
able to abstract the proton, suggesting that a base might not be

M. Boronat et al. / Journal of Catalysis 315 (2014) 6–14
11
Fig. 6. PA adsorption at neutral and cationic Au sites in different Au/CeO2 models.
necessary to perform the homocoupling PA provided that the gold
nanoparticles are partially oxidized. The phenylacetylenyl frag-
ments directly attached to cationic Au species are less stable and
consequently more reactive than those bonded to metallic Au0
atoms and, as a consequence, the activation barriers for the bimo-
lecular surface coupling step are lower on the catalysts containing
cationic gold. Finally, in the absence of a base, the protons that are
transferred from PA to the oxygen atoms at the metal-support
interface will form H2O that will desorb from the catalyst, decreas-
ing the concentration of cationic Aud+ and Au+ active sites, which
should be regenerated by reaction with O2.
A remarkable conclusion from this part of the study is the double
role of O2 on the reaction mechanism. On one hand, O2 is the final
acceptor of electrons and protons generated in the dissociation of
the CAH bond of PA. But on the other hand, its dissociation is
necessary to generate the cationic Aud+ and Au+ active sites on the
surface of isolated gold nanoparticles or at the gold-support inter-
face in Au/CeO2 catalysts. As previously discussed, the activation
energy for O2 dissociation depends on the size and shape of the gold
nanoparticles and the interaction with the support, and might
become the rate determining step of the global process. In particu-
lar, it suggests that small gold clusters that are able to efficiently
activate the alkyne C„C bond but cannot dissociate molecular O2
[29], will probably be inactive toward alkyne homocoupling.
stabilize surface Aud+ species. As shown in Fig. 8, good catalytic
activities were found with all materials. To confirm that the catal-
ysis was heterogeneous, the reaction was stopped after ꢀ20% con-
version and the solid catalyst was filtered in hot. Although some
gold was leached out from Au/CeO2, Au/C and Au/Fe2O3 (1.0%,
0.1% and 0.2% of the initial gold, respectively), the gold species in
solution gave only a marginal activity in the presence of air (see
Figs. S1–S3), which is consistent with the fact that when AuCl or
AuNaCl4 were introduced as catalyst in solution, no activity was
observed. The scope of the homocoupling reaction under O2-pro-
moted base-free conditions was examined for different alkynes in
the presence of catalytic amounts of Au/C (see Table 3). Aromatic
(entries 1–9) and aliphatic alkynes (entries 10–11) can be both
coupled with yields ranging from 60% to 90%. These yields are
good, taking into consideration that alkynes can undergo other
reactions in the presence of supported gold catalysts [38]. The
reusability of the AuAC catalyst was also tested and complete con-
version and selectivity for the homocoupling product was obtained
after four uses (entry 2).
At this point, it is confirmed that the coupling of alkynes is cat-
alyzed by gold nanoparticles in the absence of a base. The second
proposal from the theoretical part of the work is that O2 dissocia-
tion on the gold nanoparticles generating the oxidized Aud+ and
Au+ active sites is, most probably, the rate determining step of
the global reaction. To confirm this hypothesis, the influence of
O2 pressure on the initial reaction rate was investigated. As shown
in Fig. 9 for Au/CeO2, Au/C and Au/ZnO, the reaction practically
does not occur under N2 atmosphere, while a maximum in activity
is obtained at intermediate O2 pressures (between 2 and 3 equiva-
lents of O2, depending on the catalyst), and further increasing O2
pressure suppresses activity. On the other hand, no influence of
the alkyne concentration on the reaction rate is observed in the
range studied. The above results lead to the conclusion that the
controlling step during the alkyne coupling on gold nanoparticles
under our reaction conditions is the dissociation of oxygen
generating the cationic Aud+ and Au+ active sites. To further clarify
whether the decrease in catalytic activity observed at high O2
pressure is due to competitive adsorption of reactants or involves
3.2. Catalytic activity of supported gold nanoparticles
According to the theoretical study, homocoupling of alkynes in
the presence of O2 should be effectively catalyzed by gold nanopar-
ticles without the need of an added base, provided that these gold
nanoparticles are able to dissociate molecular O2 and become
partially oxidized, generating basic O atoms able to abstract the
proton from the alkyne as well as cationic Aud+ and Au+ sites that
favor the CAC bond forming step.
To test the above hypothesis, we carried out the homocoupling
of ortho-tolylacetylene in the presence of O2 and in absence of any
base on a series of supported gold catalysts, including Au/CeO2, Au/
C, Au/TiO2, Au/ZnO, Au/Al2O3 and Au/Fe2O3. All of them are able to

12
M. Boronat et al. / Journal of Catalysis 315 (2014) 6–14
Fig. 7. Optimized structures of reactants (left) and transition states (right) involved in the bimolecular surface coupling step on (a) Au/ox-CeO2, (b) Au2O3/CeO2, (c) Au10/CeO2
and (d) Au9/ox-CeO2 models. Distances in Å.
an irreversible modification of the catalyst, a sample of Au/ZnO
that had been used at 6 bar O2 pressure and found to be inactive
was reused at 2 bar O2 pressure. Under these new conditions, it
showed the same activity that the as-prepared Au/ZnO samples,
indicating that the catalyst did not undergo any irreversible deac-
tivation or transformation under reaction conditions. Moreover,

M. Boronat et al. / Journal of Catalysis 315 (2014) 6–14
13
Fig. 8. Base-free Au-supported-catalyzed oxidative homocoupling of ortho-tolylacetylene. GC yields.
DFT adsorption energy values obtained for PA and O2 interaction
with
Table 3
a
Au38 nanoparticle are À14.8 kcal/mol (Table 2) and
Results for the base-free oxidative homocoupling of alkynes catalyzed by AuAC under
an oxygen atmosphere.
À20 kcal/mol [31] respectively, suggesting the possibility of com-
petitive adsorption at the active sites.
Au-C (2-5 mol%)
R
R
R
The results obtained suggest that the alkyne homocoupling
reaction should only occur on those gold particles able to dissoci-
ate oxygen and to form the ‘‘partially oxidized’’ species. If this is
so, then gold clusters with few gold atoms should not be active
since it has been shown [29] that they are not able to dissociate
oxygen. Indeed, when the homocoupling reaction was carried out
under oxygen in a solution containing gold clusters with 3–9 atoms
prepared as described in the literature [39,40], no conversion was
found (see Fig. S4), in agreement with the predictions made by the
theoretical calculations.
O2 (0.75 mmol)
1,3-dichlorobenzene (0.5 M),
170 ºC, 18 h
(0.25 mmol)
Run
R
Catalyst (mol%)
Yield (%)a
1
2
3
4
5
6
7
8
9
H
2
2
5
5
2
2
5
5
2
2
2
70
85b
90c
80
85
77
67
80
65
60
65
o-Tolyl
o-Tolyl
m-Tolyl
o-Chlorophenyl
m-Chlorophenyl
o-Bromophenyl
m-Bromophenyl
m-Anisyl
4. Conclusion
10
11
Decyl
Dodecyl
The mechanism of alkyne homocoupling over gold nanoparti-
cles and clusters, isolated and supported on CeO2, has been theo-
retically investigated by means of periodic DFT calculations.
Phenylacetylene adsorbs at metallic Au0 and cationic Aud+ and
Au+ sites, and its deprotonation is energetically feasible in all sys-
tems having basic O atoms able to abstract the proton. On the other
a
b
c
GC yield, products isolated by column chromatography after reaction.
After 4 uses the yield is similar (81%).
Conventional glassware, fitted with a septum rubber and coupled to an O2
balloon.
Fig. 9. Dependence of the initial rate vs. oxygen and alkyne concentration for Au/CeO2 (a and b), Au/C (c and d), and Au/ZnO (e and f) for the base-free Au-catalyzed oxidative
homocoupling of ortho-tolylacetylene. For reaction conditions see Fig. 8 GC yields.

14
M. Boronat et al. / Journal of Catalysis 315 (2014) 6–14
hand, the alkynyl fragments directly attached to cationic Aud+ and
Au+ species are more reactive than those bonded to metallic Au0
atoms, and as a consequence, the activation barriers for the bimo-
lecular surface coupling step are lower on the catalysts containing
cationic gold.
Experimental results show that the base-free homocoupling of
different alkynes is catalyzed by gold nanoparticles supported on
a large variety of solids, and confirm the theoretical prediction that
the dissociation of oxygen on the gold nanoparticle is the control-
ling step of the global reaction. The easy recovering of the catalyst,
the use of air as terminal oxidant, the generation of water as only
by-product, and the isolation of the homocoupling products after
simple solvent removal makes this CAH activation gold-catalyzed
system an extremely practical and environmentally friendly
process.
[4] E. Negishi, A. Alimardanov, in: E. Negishi (Ed.), Handbook of Organopalladium
Chemistry for Organic Synthesis, Wiley, NewYork, 2002, p. 989.
[5] S.M. Auer, M. Schneider, A. Baiker, J. Chem. Soc. Chem. Commun. (1995) 2057.
[6] B.C. Zhu, X.Z. Jiang, Appl. Organomet. Chem. 21 (2007) 345.
[7] P. Kuhn, A. Alix, M. Kumarraja, B. Louis, P. Pale, J. Sommer, Eur. J. Org. Chem.
(2009) 423.
[8] P. Kuhn, P. Pale, J. Sommer, B. Louis, J. Phys. Chem. C 113 (2009) 2903.
[9] T. Oishi, T. Katayama, K. Yamaguchi, N. Mizuno, Chem. Eur. J. 15 (2009) 7539.
[10] T. Oishi, K. Yamaguchi, N. Mizuno, ACS Catal. 1 (2011) 1351.
[11] K. Sonogashira, in: E. Negishi (Ed.), Handbook of Organopalladium Chemistry
for Organic Synthesis, Wiley, New York, 2002, p. 493.
[12] K. Sonogashira, J. Organomet. Chem. 653 (2002) 46.
[13] C. Gonzalez-Arellano, A. Abad, A. Corma, H. Garcıa, M. Iglesias, F. Sanchez,
Angew. Chem. Int. Ed. 46 (2007) 1536.
[14] S.K. Beaumont, G. Kyriakou, R.A. Lambert, J. Am. Chem. Soc. 132 (2010) 12246.
[15] M. Boronat, D. Combita, P. Concepción, A. Corma, H. García, R. Juárez, S.
Laursen, J.D. Lopez-Castro, J. Phys. Chem. C 116 (2012) 24855.
[16] J.P. Perdew, J.A. Chevary, S.H. Vosko, K.A. Jackson, M.R. Pederson, D.J. Singh, C.
Fiolhais, Phys. Rev. B 48 (1993) 4978.
[17] J.P. Perdew, Y. Wang, Phys. Rev. B 45 (1992) 13244.
[18] G. Kresse, J. Furthmüller, Phys. Rev. B 54 (1996) 11169.
[19] G. Kresse, J. Hafner, Phys. Rev. B 47 (1993) 558.
[20] P.E. Blöchl, Phys. Rev. B 50 (1994) 17953.
Acknowledgements
[21] A. Heyden, A.T. Bell, F.J. Keil, J. Chem. Phys. 123 (2005) 224101.
[22] G. Henkelman, H. Jónsson, J. Chem. Phys. 111 (1999) 7010.
[23] G. Henkelman, H. Jónsson, J. Chem. Phys. 113 (2000) 9978.
[24] E. Sanville, S.D. Kenny, R. Smith, G. Henkelman, J. Compos. Chem. 28 (2007)
899.
[25] G. Henkelman, A. Arnaldsson, H. Jónsson, Comput. Mater. Sci. 36 (2006) 254.
[26] C. Loschen, J. Carrasco, K.M. Neyman, F. Illas, Phys. Rev. B. 75 (2007) 035115.
[27] L. Alves, B. Ballesteros, M. Boronat, J.R. Cabrero-Antonino, P. Concepción, A.
Corma, M.A. Correa-Duarte, E. Mendoza, J. Am. Chem. Soc. 133 (2011) 10251.
[28] J. Oliver-Meseguer, J.R. Cabrero-Antonino, I. Domínguez, A. Leyva-Pérez, A.
Corma, Science 338 (2012) 1452.
Financial support from Spanish MCIINN (Consolider Ingenio
2010-MULTICAT CSD2009-00050 and Subprograma de apoyo a
Centros y Universidades de Excelencia Severo Ochoa SEV 2012
0267
Projects)
is
acknowledged.
Red
Española
de
Supercomputación (RES) and Centre de Càlcul de la Universitat
de València are gratefully acknowledged for computational
facilities and technical assistance. A.L.-P. and J.O.-M. also thank
ITQ for a contract.
[29] M. Boronat, A. Leyva-Pérez, A. Corma, Acc. Chem. Res. 47 (2014) 834.
[30] A. Roldán, S. González, J.M. Ricart, F. Illas, ChemPhysChem 10 (2009) 348.
[31] M. Boronat, A. Corma, Dalton Trans. 9 (2010) 8538.
Appendix A. Supplementary material
[32] L.M. Molina, M.D. Rasmussen, B. Hammer, J. Chem. Phys. 120 (2004) 7673.
[33] I. Remediakis, N. Lopez, J. Norskov, Angew. Chem. Int. Ed. 44 (2005) 1824.
[34] J. Wang, B. Hammer, Phys. Rev. Lett. 97 (2006) 136107.
[35] I.X. Green, W. Tang, M. Neurock, J.T. Yates Jr., Science 333 (2011) 736.
[36] S. Laursen, S. Linic, J. Phys. Chem. C 113 (2009) 6689.
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2014.04.003.
[37] S. Laursen, S. Linic, Phys. Chem. Chem. Phys. 11 (2009) 11006.
[38] S. Carrettin, M.C. Blanco, A. Corma, A.S.K. Hashmi, Adv. Synth. Catal. 348 (2006)
1283.
[39] J. Zheng, C. Zhang, R.M. Dickson, Phys. Rev. Lett. 93 (2004) 077402/1.
[40] J. Oliver-Meseguer, A. Leyva-Pérez, A. Corma, ChemCatChem 5 (2013) 3509.
References
[1] J. Liu, J.W.Y. Lam, B.Z. Tang, Chem. Rev. 109 (2009) 5799.
[2] R.E. Martin, F. Diederich, Angew. Chem. Int. Ed. 38 (1999) 1350.
[3] P. Siemens, R.C. Livingston, F. Diederich, Angew. Chem. Int. Ed. 39 (2000) 2632.