Silver nanoparticles for olefin production: New insights into the mechanistic description of propyne hydrogenation

Article abstract of DOI:10.1002/cctc.201300569

The gas-phase partial hydrogenation of propyne was investigated over supported Ag nanoparticles (2-20 nm in diameter) prepared by using different deposition methods, activation conditions, loadings, and carriers. The excellent selectivities to propene attained over the catalysts, exceeding 90 %, are independent of the particle size but the activity is maximal over approximately 4.5 nm Ag particles. Certain kinetic fingerprints of Ag, such as the positive dependence on the alkyne pressure, the relatively low reaction order in H ₂, and the low apparent activation energy, deviate from those of conventional hydrogenation metals such as Pd and Ni, questioning the applicability of the classical Horiuti-Polanyi scheme. Periodic dispersion-corrected density functional theory (DFT-D) calculations and microkinetic analysis demonstrate the occurrence of an associative mechanism, which features the activation of H₂ on the adsorbed propyne at structural step sites. By using the atomistic Wulff model, the number of B5 sites available on the Ag nanoparticles was estimated to be maximal in the size range of 3.5-4.7 nm. The rate of propene production correlates with the density of B5 sites, which suggests that the latter are potential active centers for the reaction. This alternative pathway broadens the mechanistic diversity of hydrogenation reactions over metal surfaces and opens new directions for understanding metals that do not readily activate H₂. The money metal! Silver selectively catalyzes the hydrogenation of propyne to propene, and the activity is maximal over 4.5 nm nanoparticles. The rate of propene production correlates well with the density of B5 sites, which suggests that the latter are potential active centers in the reaction. The hydrogenation follows an associative scheme, featuring the activation of H₂ directly on the propyne-silver-surface intermediates. Copyright

Full text of DOI:10.1002/cctc.201300569

CHEMCATCHEM
FULL PAPERS
DOI: 10.1002/cctc.201300569
Silver Nanoparticles for Olefin Production: New Insights
into the Mechanistic Description of Propyne
Hydrogenation
Gianvito Vilꢀ,[a] David Baudouin,[b] Ioannis N. Remediakis,[c] Christophe Copꢀret,*[b]
Nfflria López,*[d] and Javier Pꢀrez-Ramírez*[a]
The gas-phase partial hydrogenation of propyne was investi-
gated over supported Ag nanoparticles (2–20 nm in diameter)
prepared by using different deposition methods, activation
conditions, loadings, and carriers. The excellent selectivities to
propene attained over the catalysts, exceeding 90%, are inde-
pendent of the particle size but the activity is maximal over
approximately 4.5 nm Ag particles. Certain kinetic fingerprints
of Ag, such as the positive dependence on the alkyne pressure,
the relatively low reaction order in H2, and the low apparent
activation energy, deviate from those of conventional hydroge-
nation metals such as Pd and Ni, questioning the applicability
of the classical Horiuti–Polanyi scheme. Periodic dispersion-cor-
rected density functional theory (DFT-D) calculations and mi-
crokinetic analysis demonstrate the occurrence of an associa-
tive mechanism, which features the activation of H2 on the ad-
sorbed propyne at structural step sites. By using the atomistic
Wulff model, the number of B5 sites available on the Ag nano-
particles was estimated to be maximal in the size range of 3.5–
4.7 nm. The rate of propene production correlates with the
density of B5 sites, which suggests that the latter are potential
active centers for the reaction. This alternative pathway broad-
ens the mechanistic diversity of hydrogenation reactions over
metal surfaces and opens new directions for understanding
metals that do not readily activate H2.
Introduction
The selective hydrogenation of targeted functional groups
over heterogeneous catalysts is one of the most studied fami-
lies of reactions in catalysis and plays a central role in multiple
industrial sectors.[1,2] One example is the Pd-catalyzed partial
hydrogenation of unsaturated hydrocarbons in the gas or
liquid phase, which is a crucial step for the purification of
olefin streams and for the manufacture of fine chemicals.[3] The
practical relevance of these processes has led to intense efforts
to improve the fundamental understanding of the reaction,
and many catalytic systems have been studied with the pur-
pose of maximizing the alkene selectivity.[4–12] The prototypical
mechanism describing the heterogeneously-catalyzed hydro-
genation of unsaturated hydrocarbons, referred to as the Hori-
uti–Polanyi (HP) scheme, involves the dissociation of molecular
H2 on the metal surface followed by the addition of H atoms
to the adsorbed organic moiety.[13] Originally proposed for the
hydrogenation of ethylene and benzene over nickel, this
scheme was extended to alkynes and dienes over Pd and Ni[10]
and served as a guide for understanding double bond isomeri-
zation and deuterium scrambling.[14] The HP scheme has also
been applied to metals with limited H2 splitting ability, as
shown for propyne hydrogenation over gold and copper nano-
particles.[10] Over these metals, the dissociation of H2 is im-
proved at steps and kinks, owing to electronic and geometric
effects,[10,15] which minimizes over-hydrogenation, leading to
a remarkable olefin yield.
Ag shows the lowest reactivity toward H2;[16–18] its dissocia-
tion features large activation barriers even at low-coordinated
surface sites.[19] Nevertheless, Sµrkµny and Rꢀvay studied the
hydrogenation of ethyne over SiO2- and TiO2-supported Ag
nanoparticles (4–8 nm) and reported full selectivity to ethene
at a low degree of ethyne conversion (3–13%).[20] Although the
work demonstrated the highly selective character of Ag in
alkyne hydrogenation, no mechanistic and kinetic understand-
ing exists thus far and the difficulty to dissociate H2 on Ag
questions the validity of the HP scheme.
[a] G. Vilꢀ, Prof. J. Pꢀrez-Ramꢁrez
Department of Chemistry and Applied Biosciences
Institute for Chemical and Bioengineering
ETH Zurich
Wolfgang-Pauli-Strasse 10, CH-8093 Zurich (Switzerland)
E-mail: jpr@chem.ethz.ch
[b] Dr. D. Baudouin, Prof. C. Copꢀret
Laboratory of Inorganic Chemistry
Department of Chemistry and Applied Biosciences
ETH Zurich
Wolfgang-Pauli-Strasse 10, CH-8093 Zurich (Switzerland)
E-mail: ccoperet@inorg.chem.ethz.ch
[c] Prof. I. N. Remediakis
Department of Materials Science and Technology
University of Crete
GR-71003 Heraklion (Greece)
[d] Prof. N. Lꢂpez
Institute of Chemical Research of Catalonia, ICIQ
Av. Paꢃsos Catalans 16, E-43007 Tarragona (Spain)
E-mail: nlopez@iciq.es
An alternative mechanism, observed in organometallic
chemistry, homogeneous catalysis, and isolated metal centers
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201300569.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&1
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
on oxide surfaces, involves the cleavage of the HꢀH bond via
the direct activation of molecular H2 on the adsorbed hydro-
carbon–metal complex (hereafter referred to as associative
mechanism).[21,22] In heterogeneous catalysis, this scheme was
speculated for the hydrogenation of norbornadiene and proto-
[23]
adamantanone over Au/SiO2
but no supporting evidence
was provided. The weak interaction between H2 and Ag justi-
fies a detailed mechanistic and kinetic analysis on this alterna-
tive pathway.
Herein, we study the partial hydrogenation of propyne to
propene over supported Ag nanoparticles with variable parti-
cle size (2–20 nm), attained by modification of different prepa-
ration variables. Olefin production over the catalysts is favored
(up to 93% selectivity), even at a high degree of alkyne con-
version. By combining a large set of systematic steady-state
catalytic tests over well-characterized samples, DFT calcula-
tions, and microkinetic analyses, we demonstrate that H2 split-
ting on the alkyne adsorbate is the key elementary step of the
selective hydrogenation of alkyne on Ag. The activity is maxi-
mal over approximately 4.5 nm Ag particles, which correlates
with the highest amount of B5 sites on such particles evaluat-
ed by using the atomistic Wulff method. The development of
this pathway provides new views on hydrogenation mecha-
nisms over metal surfaces beyond the classical dissociative
pathway proposed by Horiuti and Polanyi nearly 80 years ago.
Results and Discussion
Figure 1. Transmission electron micrographs and particle size distributions
of two representative 1 wt% Ag/SiO2 catalysts with different Ag dispersion.
The catalysts were prepared by the spray deposition method followed by
(a) activation in H2 flow at 473 K for 0.5 h or (b) calcination in static air at
573 K for 2 h.
Catalyst characterization
The characterization data of selected Ag-based catalysts are
listed in Table 1. The silver content in the samples matches the
nominal loading. The BET surface area of the catalysts after the
deposition of 0.5–5 wt% Ag is extremely close to the specific
surface area of the corresponding support. Transmission elec-
tron microscopy (TEM) was used to assess the distribution of
Ag nanoparticles and to estimate an average value of metal
dispersion. The representative transmission electron micro-
graphs of exemplary 1 wt% Ag/SiO2 catalysts with different Ag
dispersions as a consequence of H2 activation after spray depo-
sition of the Ag precursor on the carrier are shown in Figure 1.
Calcination in static air leads to a large average Ag particle size
of approximately 20 nm (Ag dispersion=5%) in contrast to
the much smaller particles of 4.4 nm (Ag dispersion=17%) if
the same spray-deposited catalyst is activated in flowing H2.
The impregnation of 1 wt% Ag on silica followed by conven-
tional drying and H2 flow activation generates a cata-
lyst with an Ag dispersion amounting to 24% and an
average Ag particle size of 4.6 nm. The higher disper-
Table 1. Characterization data of selected catalysts.
sion value over the impregnated sample than over
the spray-deposited sample is associated with the
[d]
Catalyst
Method Activation Ag content[b] SBET
dAg
D[d]
[%]
[wt%]
[m2 gꢀ1
]
[nm]
more uniform size distribution of the Ag nanoparti-
cles in the impregnated sample. Ag particles of ca.
2 nm (Ag dispersion=75%) were obtained in the
1 wt% Ag/SiO2 catalyst prepared through impregna-
tion followed by freeze drying and H2 flow activation.
In fact, the rapid cooling of the precursor phase
during freeze drying reduces the mobility of the
liquid, minimizes particle segregation, and often pro-
vides a homogeneous powder of fairly uniform parti-
cle size.[24,25] 1 wt% Ag/Al2O3 and 1 wt% Ag/TiO2 ob-
tained through spray deposition and H2 flow activa-
tion also demonstrate highly dispersed nanoparticles
0.5 wt% Ag/SiO2 SD
1 wt% Ag/SiO2
H2 flow
static air
H2 flow
H2 flow
H2 flow
H2 flow
H2 flow
H2 flow
0.5
1.3
1.3
1.0
1.0
4.9
0.9
1.1
199 (197)[c]
196
196
198
195
186
79 (81)
28 (32)
–
–
5
SD
21.3ꢁ8.3
4.4ꢁ3.1 17
2.0ꢁ0.3 75
4.6ꢁ1.6 24
6.2ꢁ3.3 14
2.3ꢁ0.8 49
2.8ꢁ0.2 45
FD
CD
SD
SD
SD
5 wt% Ag/SiO2
1 wt% Ag/Al2O3
1 wt% Ag/TiO2
[a] CD=impregnation followed by conventional drying; FD=impregnation followed
by freeze drying; SD=spray deposition. [b] Determined from inductively coupled
plasma optical emission spectroscopy analysis. [c] BET surface area of the supports
within parentheses. [d] Determined from TEM analysis.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
2
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
centered at 2.3 and 2.8 nm, respectively. The Ag loading mod-
erately affects the dispersion: the average Ag particle size in-
creases from 4.4 to 6.2 nm if the Ag content increases from
1 to 5 wt% over the Ag/SiO2 catalyst prepared through spray
deposition and H2 flow activation. Thus, it can be concluded
that the stabilization of small Ag nanoparticles on the carriers
requires rapid drying and flow activation conditions whereas
the nature of the support, the silver loading, and the deposi-
tion method play a secondary role. The systematic characteri-
zation of the catalysts after the hydrogenation tests enabled
us to conclude that Ag loading, catalyst textural properties,
and Ag dispersion are not altered by the reaction (Figure S1).
favor CꢀC coupling reactions at the expense of the lower
alkene selectivity. Regarding the effect of the reactants ratio,
upon increasing the inlet partial H2 pressure, the conversion of
propyne increases quasi-linearly; in addition, the selectivity to
propene increases, which suggests that H2 activation is the
rate-limiting step of the reaction. The selectivity to propane is
very low (<5%), even at H2/C3H4 =25, and oligomers are pro-
duced at the lowest H2-to-hydrocarbon ratios.[26] Despite differ-
ent Ag dispersions in the two catalysts, the selectivity patterns
are analogous; in contrast, the catalyst calcined in static air
(Figure 2c and d) shows a significantly low propyne conversion
because of its larger (and less active) Ag particles.
All remaining catalysts were tested at T=473 K and H2/
C3H4 =25; Table 2 summarizes the effect of different prepara-
tion variables on the catalytic performance. The conversion of
propyne increases with an increased Ag content, which reach-
es more than 90% at 1 wt% Ag. The plateau at higher load-
ings is due to the formation of larger Ag particles, which
expose a minor number of surface atoms responsible for the
catalytic process. The selectivity to propene is remarkably high
(>90%) regardless of the particle size, and the selectivity to
propane and oligomers is less than 10% in all catalysts. It
should be stressed that the high selectivity to propene is at-
tained at a full degree of conversion, being significant for in-
dustrial purposes. The activation conditions during preparation
also affect the catalyst performance. The increased Ag particle
size upon calcination in static air leads to low alkyne conver-
sion (31%) at an alkene selectivity of 91%. In contrast, the
Propyne hydrogenation
The effect of temperature (373–573 K) and H2/C3H4 ratio (5–25)
on the performance of propyne hydrogenation was investigat-
ed over two representative Ag-based catalysts with different
particle sizes (Figure 2). Figure 2a and b shows the effect of
temperature and H2/C3H4 ratio over 1 wt% Ag/SiO2 prepared
through spray deposition and H2 flow activation (Figure 1a),
whereas Figure 2c and d refers to the SiO2-supported 1 wt%
catalyst prepared through spray deposition and calcination in
static air (Figure 1b). Both the materials show an increase in
propyne conversion with an increase in temperature. The se-
lectivity to propene is practically unchanged (ꢂ90%) up to
473 K, and the selectivity to the undesired products (propane
and oligomers) does not exceed 10%. Higher temperatures
Figure 2. Propyne hydrogenation performance versus temperature (at H2/C3H4 =25; left) and H2/C3H4 ratio (at T=473 K; right) over 1 wt% Ag/SiO2 prepared
by spray deposition followed by (a,b) activation in H2 flow at 473 K for 0.5 h or (c,d) calcination in static air at 573 K for 2 h. Other reaction conditions: t=1 s
and P=1 bar.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
3
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Table 2. Influence of different preparation variables on the performance
of the supported Ag catalysts in propyne hydrogenation.[a]
Catalyst
Method Activation X(C3H4) S(C3H6) S(C3H8) S(OL)
[–]
[–]
[–]
[–]
Ag loading
0.5 wt% Ag/SiO2 SD
H2 flow
H2 flow
H2 flow
0.68
0.92
0.90
0.91
0.93
0.091
0.01
0.03
0.04
0.08
0.04
0.05
1 wt% Ag/SiO2
5 wt% Ag/SiO2
SD
SD
Deposition method and activation conditions
1 wt% Ag/SiO2
SD
SD
SD
SD
SD
FD
FD
FD
FD
FD
CD
static air
He flow
NO flow
O2 flow
H2 flow
static air
He flow
NO flow
O2 flow
H2 flow
H2 flow
0.31
0.82
0.81
0.87
0.92
0.41
0.79
0.74
0.76
0.70
0.89
0.91
0.86
0.99
0.84
0.93
0.93
0.89
0.87
0.86
0.91
0.90
0.06
0.08
0.00
0.08
0.03
0.03
0.05
0.06
0.08
0.05
0.08
0.03
0.06
0.01
0.08
0.04
0.04
0.06
0.07
0.06
0.04
0.02
Figure 3. Rate of propene production (solid symbols) and selectivity to pro-
pene (open symbols) as a function of the average Ag particle size in the
supported catalysts containing 1 wt% Ag. Reaction conditions: T=473 K,
H2/C3H4 =25, t=1 s, and P=1 bar.
ure 1b) are shown in Figure 4. The apparent activation energy
for propyne hydrogenation is approximately 30 kJmolꢀ1 (Fig-
ure 4a) and is independent of the Ag particle size. This value is
significantly lower than what has been generally reported for
Pd, Pt, Cu, and Ni (50–70 kJmolꢀ1).[27–29] Upon increasing the
partial H2 pressure, the normalized reaction rate increases over
catalysts with extremely different Ag particles, with a reaction
order in H2 between 0 and 1 (Figure 4b). For highly active
metals such as Pd, Pt, and Ni, the reaction order in H2 is gener-
ally greater than 1.[27–29] In contrast, upon increasing the partial
pressures of propyne, the reaction rate increases (Figure 4c),
which leads to a positive reaction order in propyne in the
range of 0–0.3. The positive dependence of the reaction rate
on the partial alkyne pressure is unusual in the partial hydro-
genation of alkynes, for which negative values are
obtained.[27–29]
Carrier
1 wt% Ag/Al2O3 SD
1 wt% Ag/TiO2 SD
H2 flow
H2 flow
0.58
0.70
0.91
0.87
0.06
0.07
0.03
0.06
[a] Reaction conditions: T=473 K, H2/C3H4 =25, t=1 s, and P=1 bar.
small and highly dispersed Ag nanoparticles after activation in
a flow leads to high propyne conversions (81–92%) at similar
levels of propene selectivity (84–99%). The specific activation
atmosphere (He, NO, O2, or H2) has a relatively minor effect on
the performance, as long as the treatment is performed under
flowing conditions. Varying the carrier, the Al2O3- or TiO2-sup-
ported catalysts exhibit lower propyne conversions (58–70%)
than the Ag/SiO2 catalysts, probably owing to the extremely
small Ag nanoparticles (2.3–2.8 nm), which are less active in
propyne hydrogenation (vide infra). Many other factors, such as
strong metal–support interactions (Ag over TiO2), may also be
involved in the loss of activity; however, such an investigation
is beyond the scope of this article.
Hydrogenation mechanism on Ag: Computational approach
To obtain a molecular-level understanding of the atypical be-
havior of Ag in alkyne hydrogenation, periodic DFT calculations
were performed on the stepped (211) surface of the metal.
Two possible reaction pathways comprising the activation of
H2 on the metal surface (HP or dissociative mechanism) and
that on the adsorbed alkyne (associative mechanism) were
compared (Figure 5a and Table S1).
The reaction rate and propene selectivity versus the average
Ag particle size is displayed in Figure 3. The selectivity to pro-
pene is size independent andis stable at approximately 90%.
No marked selectivity differences were observed over the
broad range of conversions (20–90%), which emphasizes the
extremely selective character of Ag in propyne hydrogenation.
The rate of olefin production per mole of total Ag reached
a maximum for the catalyst with 4.5 nm particles (this impor-
tant point is elaborated below). Notably, the same catalysts
were tested at three times lower contact times to decrease the
degree of conversion and the volcano dependence was analo-
gous, which demonstrates the kinetic relevance of the results
independent of conducting the tests under more integral or
differential conditions.
At steps and kinks, the elementary steps of the dissociative
hydrogenation mechanism are as follows [Steps (D1)–(D7)]:
*
*
C3H4 þ Ð C3H4
ðD1Þ
ðD2Þ
ðD3Þ
ðD4Þ
ðD5Þ
ðD6Þ
ðD7Þ
*
*
H2 þ þ # Ð H þ H#
*
*
*
*
C3H4 þ H Ð C3H5
þ
þ
*
*
*
*
C3H5 þ H Ð C3H6
The apparent activation energy and reaction orders were de-
termined over SiO2-supported Ag nanoparticles with variable
particle size to understand whether these kinetic parameters
are particle size dependent. The results for the catalysts with
an average particle size of 4.4 nm (Figure 1a) and 21 nm (Fig-
*
*
C3H6 Ð C3H6 þ
*
*
*
*
C3H6 þ H Ð C3H7
þ
*
*
H# þ Ð H þ #
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
4
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Figure 5. Energy profiles for the hydrogenation of propyne on Ag(211) (a).
Lateral view of the stepped Ag(211) surface and transition state for H2 acti-
vation via the dissociative (b) and associative (c) mechanisms.
Figure 4. Arrhenius plot for propyne hydrogenation at T=373–473 K, H2/
C3H4 =25, t=1 s, and P=1 bar (a). Normalized reaction rate as a function of
the inlet partial pressure of H2 (b) and the inlet partial pressure of C3H4 (c) at
T=473 K, t=0.3 s, and P=1 bar. Solid circles correspond to 1 wt% Ag/SiO2
prepared by spray deposition followed by activation in H2 flow at 473 K for
0.5 h; open squares correspond to 1 wt% Ag/SiO2 prepared by spray deposi-
tion followed by calcination in static air at 573 K for 2 h.
adsorbed H atoms: one at the edge and the other one on the
lower surface of the step. This splitting is endoenergetic
(+0.53 eV) and has an energy barrier of 1.33 eV. The barrier re-
duces to 1.06 eV upon introduction of the zero-point vibration-
al energy (ZPVE) contributions, which are required considering
the small mass of H2. This large barrier suggests that H2 split-
ting is the rate-limiting step of this scheme. Notably, this value
is much higher for Ag than for Au clusters (0.84 eV)[10] and
Cu(211) (0.50 eV). The transfer of the first H atom (Step D3) to
the adsorbed alkyne, forming an alkenyl surface species, is en-
doenergetic (+0.44 eV), with an energy barrier of 0.88 eV,
whereas the transfer of the second H atom (Step D4), forming
the adsorbed propene, is highly exoenergetic (ꢀ2.82 eV), with
an energy barrier of 0.37 eV. The adsorbed propene sits parallel
to the surface and can easily desorb (Step D5) with an energy
of only 0.14 eV. The over-hydrogenation reaction (Step D6) is
marginally exoenergetic (ꢀ0.19 eV) but is hindered by a large
energy barrier of 1.01 eV; hence, propene desorption is pre-
ferred.
The two active sites identified are on the edge of the step
(*) and on the lower terrace (#). The adsorption of propyne
(D1) is slightly exoenergetic (by ꢀ0.17 eV) on step sites and is
less favored (ꢀ0.05 eV) on flat surfaces. Therefore, propyne ad-
sorbs more strongly on low-coordinated surface sites. This
result points to the steps as the active sites of the reaction.
Considering the similar adsorption energies of propyne
(ꢀ0.17 eV) and propene (ꢀ0.14 eV), the preferential adsorption
of alkynes and the lack of adsorption of alkenes (thermody-
namic selectivity)[8] cannot be invoked to explain the high pro-
pene selectivity observed experimentally. Instead, a sizable ki-
netic barrier hindering the over-hydrogenation reaction of the
olefin is expected. The dissociation of H2 (Step D2; Figure 5b)
occurs at the steps of the Ag(211) surface and leads to two
The elementary steps of the associative mechanism are as
follows [Steps (A1)–(A8)]:
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
5
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Hydrogenation mechanism on Ag: Kinetic analysis and
structural aspects
*
*
C3H4 þ Ð C3H4
ðA1Þ
ðA2Þ
ðA3Þ
ðA4Þ
ðA5Þ
ðA6Þ
ðA7Þ
ðA8Þ
*
*
C3H4 þ H2 Ð C3H4 : H2
By following the elementary steps D1–D7 of the HP scheme, it
is possible to derive the mathematical expressions for the rate
and reaction orders (complete derivations are given in the
Supporting Information). Considering that H2 dissociation is
the most possible rate-limiting step, the initial rate of the reac-
tion is expressed in Equation (1):
*
*
C3H4 : H2 þ # Ð C3H5 þ H#
*
*
C3H5 þ H# Ð C3H6 þ #
*
*
C3H6 Ð C3H6 þ
*
*
C3H6 þ H2 Ð C3H6 : H2
k2pðH2Þ
*
ð1Þ
r ꢂ k2pðH2Þqð Þqð#Þ ꢂ
*
*
C3H6 : H2 þ # Ð C3H7 þ H#
1 þ K1pðC3H4Þ
*
*
H# þ Ð H þ #
The reaction orders in H2 and propyne can be calculated by
using Equations (2) and (3):
The coadsorption of propyne (Step A1) and H2 (Step A2) is
isoenergetic with respect to the individual reactants (Fig-
ure 5c). The transition state involves a metastable intermediate
(C3H4ꢀHꢀHꢀAg complex) formed at the step edge of the
Ag(211) surface. The HꢀH and CꢀH bond lengths are 0.927
and 1.453 Š, respectively, whereas the CꢀHꢀH and CꢀCꢀH
bond angles are 173 and 1408, respectively. The Bader charges
are 0.05 jej for the H atom in contact with the alkyne and
ꢀ0.25 jej for the H atom in contact with the Ag atoms of the
lower terrace, which confirms the polarization of the HꢀH
bond observed in heterolytic splitting reactions.[21,22] This reac-
tion leads to the formation of an alkenyl and a H-surface spe-
cies sitting on the terrace (Step A3) and is an exoenergetic
(ꢀ0.43 eV) process associated with an energy barrier of
1.03 eV. Upon ZPVE correction, the energy barrier decreases to
0.74 eV. Notably, it is not possible to directly compare the cal-
culated value and the experimental value, because the appar-
ent activation energy depends on the specific set of reaction
conditions.[30] However, the large energy required for this step
confirms that H2 splitting is hindered on Ag; hence, the hydro-
genation requires that an excess of H2 (H2/C3H4 =25) should be
applied. The difference in the C3H5 +H level between the asso-
ciative and the dissociative mechanism is due to the corre-
sponding configuration for each step; in the dissociative
scheme, the H atoms are in the top terrace, and in the associa-
tive case, the H atoms are on the lower terrace. The transfer of
H atoms to the alkenyl species leading to propene is both
thermodynamically (ꢀ1.00 eV) and kinetically (0.36 eV) favored.
The formed propene sits perpendicular to the Ag surface and
desorbs isoenergetically. Regarding the over-hydrogenation re-
actions (Steps A6 and A7), desorbed propene needs to read-
sorb in a flat configuration. The new H2 bond breaking
(Step A6) is marginally exoenergetic (ꢀ0.04 eV). In contrast, the
H addition forming C3H7* is highly endoenergetic (+0.98 eV)
and hinders the production of propane. CꢀC coupling reac-
tions are also characterized by high energy barriers,[26] which
explain the low degree of oligomerization observed over Ag.
By comparing the energy profiles in Figure 5a, it is clear that
the associative mechanism requires lower activation barriers
and no high-energy intermediates, which makes this process
a much more efficient hydrogenation pathway than the typical
HP mechanism encountered on Pd and Ni.
@ ln r
nðH2Þ ¼ pðH2Þ
¼ 1
ð2Þ
ð3Þ
@ pðH2Þ
@ ln r
K1pðC3H4Þ
@ pðC3H4Þ 1 þ K1pðC3H4Þ
nðC3H4Þ ¼ pðC3H4Þ
¼
The latter varies between 0 (with p(C3H4)=0) and ꢀ1 (with
p(C3H4)!1).
The initial rate of the reaction that is derived assuming the
first H2 addition (step D3) to be the rate-limiting step of the HP
scheme is expressed in Equation (4):
k3K1K2pðC3H4ÞpðH2Þ
r ꢂ k3qðC3Hꢃ4ÞqðHꢃÞ ꢂ
ð4Þ
2
½1 þ K1pðC3H4Þ þ K2pðH2Þꢄ
The reaction orders can be written as shown in Equations (5)
and (6):
1 þ K1pðC3H4Þ ꢀ K2pðH2Þ
1 þ K1pðC3H4Þ þ K2pðH2Þ
ð5Þ
ð6Þ
nðH2Þ ¼
1 ꢀ K1pðC3H4Þ þ K2pðH2Þ
nðC3H4Þ ¼
1 þ K1pðC3H4Þ þ K2pðH2Þ
Thus, n(H2)ꢅ1 and ꢀ1ꢅn (C3H4)ꢅ0. In both cases, the HP
mechanism cannot account for the positive reaction orders in
Figure 4.
By following the elementary steps A1–A8 of the associative
mechanism and assuming step A3 to be the rate-determining
step, the initial rate of the reaction for the associative mecha-
nism can be calculated by using Equation (7):
k3K1K2pðC3H4ÞpðH2Þ
r ꢂ k3qðC3H4: H2Þqð#Þ ꢂ
1 þ K1pðC3H4Þ þ K1K2pðC3H4ÞpðH2Þ
ð7Þ
Thus, the reaction orders are expressed as given in
Equations (8) and (9):
1 þ K1pðC3H4Þ
ð8Þ
nðH2Þ ¼
1 þ K1pðC3H4Þ þ K1K2pðC3H4ÞpðH2Þ
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
6
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
showing the classical 1/d2 decrease observed in other metal
nanoparticles.
1
nðC3H2Þ ¼
ð9Þ
1 þ K1pðC3H4Þ þ K1K2pðC3H4ÞpðH2Þ
By using the results in Table S2 and the particle size distribu-
tion histograms of selected Ag samples (Table 1 and Figure S4),
it was possible to re-evaluate the data in Figure 3. The rate of
propene production increases as a function of the amount of
B5 sites, which indicates the preeminence of B5 sites in the re-
action (Figure 6). This result has been confirmed by fitting of
Thus, the reaction orders are as follows: 0ꢅn(H2)ꢅ1 and
0ꢅn(C3H4)ꢅ1. In contrast to the HP scheme in which H2 and
C3H4 compete for site adsorption and no positive reaction
order in hydrocarbon can be observed, in the associative
mechanism the cooperative effects due to H2 cleavage on the
hydrocarbon are responsible for these positive dependences.
The experimental data in Figure 4 are in line with the microki-
netic model derived from an associated reaction pathway,
which confirms that propyne hydrogenation on Ag occurs
through direct H2 splitting on the adsorbed propyne.
Although DFT calculations in the previous subsection sug-
gest that the reaction occurs at step-like positions, which are
highly concentrated on small nanoparticles, experimentally the
activity is maximal over approximately 4.5 nm Ag whereas
smaller nanoparticles are less active (Figure 3). To rationalize
this aspect, we have used the atomistic Wulff construction to
model Ag nanoparticles of different sizes and determine which
sites are active in propyne hydrogenation. The investigation
has revealed the key role of B5 sites.
Figure 6. Rate of propene production (solid symbols) as a function of the
number of active B5 sites for selected supported catalysts containing 1 wt%
Ag. The line represents the fitting of experimental data obtained through
linear regression (equation: y=0.004x+0.452; correlation coefficient,
R2 =0.92). The inset shows a Wulff-shaped Ag nanoparticle with a size of
4.3 nm and, in dark gray, the location of the active B5 sites. Reaction condi-
tions: T=473 K, H2/C3H4 =25, t=1 s, and P=1 bar.
In a nanoparticle, B5 sites alleviate the stress in the forma-
tion of line defects between two faces. Two types of B5 struc-
tures can be identified on the Ag nanoparticles: at the edge
between (111) and (211) faces and within two (211) surfaces
(Figure S3a). The presence of the B5 site is identified by char-
acteristic coordination numbers (z) of the Ag atoms around it.
This site is surrounded by two step-edge atoms (z=7), two ter-
race atoms (z=9), one bottom step atom (z=10), and two
bulk atoms (z=12).
the experimental results shown in Figure 4b. The Hertz–Knud-
sen equation [Eq. (12)] was used to model the adsorption of H2
and propyne. This equation contains the area of one free
2
By using the atomistic Wulff model, we found that Ag nano-
particles with a core diameter smaller than 3 nm cannot ac-
commodate any (211) faces. Therefore, they contain no B5
sites, which is in agreement with the low activity measured in
Figure 3. The smallest particle with B5 sites is shown in Fig-
ure S3b, and its average size, defined as the average value be-
tween the smallest and the largest size of the particle, is
3.0 nm; this particle contains four-atom-long B-type steps in
which 24 B5 atoms are located (Table S2). Nevertheless, the
presence of fivefold Ag atoms (shown in red in Figure S3b)
could affect the concentration of these active B5 sites because
fivefold atoms can easily dislocate and diffuse toward the
lowest-energy sites, destroying the particle shape. A larger Ag
cluster with 2875 atoms and a particle size of 4.3 nm is shown
in Figure S3c. This nanoparticle contains five-atom-long B-type
steps in which 48 B5 sites are situated, and the smallest edge
corresponds to two sixfold atoms belonging to hexagonal
(111) and (110) faces. Compared to the structure discussed
previously, this nanoparticle is significantly more stable under
conditions typically used in the hydrogenation experiments. As
summarized in Table S2, the highest density of B5 sites is ob-
tained for particle sizes comprised between 3.5 and 4.7 nm. Be-
cause B5-type structures on Ag are step sites within the faces
of the nanoparticle and not only on its edges, the density of
active sites decreases rather slowly for dAg >4.7 nm, without
active site (Acat). From DFT calculations, a value of 7.8”10ꢀ20
Š
is obtained, which corresponds to the area of a site at the step
of the nanoparticle. However, the total surface energy is the
average between the number of these sites and the number
of total surface atoms, and the total surface that can be as-
signed to these sites is 10ꢀ18 Š2. This value does not account
for the real number of active sites. Propyne adsorption is en-
hanced only at specific step-like (B5) positions. Thus, from the
ratio between the constants fitted experimentally (Figure 7)
and the DFT values (Table S3), we can identify the amount of
B5 sites that are located on this Ag catalyst (1 wt% Ag/SiO2
prepared by spray deposition and activated in H2 flow, with
dAg =4.4 nm), as shown in Equation(10):
kDFT
kexp
ð10Þ
nðB5Þ ¼
A B5 site density of 228 mmol B5 g Agꢀ1 was obtained, which
is in excellent agreement with the atomistic simulation data
(Figure 6).
Conclusions
We have studied the gas-phase hydrogenation of propyne
over supported Ag nanoparticles with average sizes in the
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
7
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
was performed on a Schlenk line at 393 K, 10ꢀ6 bar, and under
0.50 cm3 minꢀ1 of dry air. The samples were activated in static air
(at 573 K for 2 h) or by flowing (42 cm3 minꢀ1) He, 1 vol% NO/Ar,
5 vol% O2/He, and 5 vol% H2/He (at 473 K for 0.5 h). In all cases,
the ramp rate was 5 Kminꢀ1. The flow treatments were performed
in situ before the hydrogenation tests.
Catalyst characterization
The Ag content in the solids was determined by inductively cou-
pled plasma optical emission spectroscopy in a Horiba Ultra 2 in-
strument equipped with a photomultiplier tube detector. Powder
XRD patterns were recorded on a PANalytical X’Pert PRO-MPD Dif-
fractometer using Bragg–Brentano geometry and Ni-filtered CuKa
radiation (l=0.1541 nm). Data were recorded at 2q=3–608, with
Figure 7. Microkinetic fitting of the data shown in Figure 4b, representing
the normalized reaction rate as a function of the inlet partial pressure of H2
at T=473 K, p(C3H4)=25 mbar, t=1 s, and P=1 bar. The data correspond to
1 wt% Ag/SiO2 prepared by spray deposition and activation in H2 flow at
473 K for 0.5 h.
an angular step size of 0.058 and a counting time of 0.75 sstepꢀ1
.
TEM was performed by using a Philips CM12 microscope operated
at 100 kV. The catalysts were analyzed in their powdered form. The
particle size distribution was estimated directly from the transmis-
sion electron micrographs by assuming a Gaussian behavior. The
dispersion of silver, D, defined as the percentage of surface Ag out
of the total Ag, was estimated from the TEM particle size distribu-
tion by assuming the absence of a silver oxide shell and consider-
ing particles to be truncated octahedrons with cubic symmetry.[33]
N2 isotherms at 77 K were measured in a Quantachrome Quadra-
sorb SI analyzer. Before the measurement, the samples were de-
gassed in vacuum at 473 K for 2 h. The thermogravimetric analysis
(TGA) of the catalyst after propyne hydrogenation was performed
range of 2–20 nm. The catalysts were intrinsically selective
toward propene (ꢂ90%) regardless of the particle size. Mecha-
nistic and kinetic studies demonstrated that the hydrogenation
of propyne on Ag follows an associative scheme, which in-
volves propyne adsorption on step sites and heterolytic split-
ting of H2 on the adsorbed hydrocarbon. The reaction rate pre-
sented a maximum over the catalysts with Ag particles of size
approximately 4.5 nm. As shown by the atomistic Wulff model,
these nanoparticles possess the highest density of B5 sites,
which are likely the active centers in the reaction. These find-
ings broaden the mechanistic views of hydrogenation reac-
tions over solid catalysts, going beyond the classical Horiuti–
Polanyi scheme, and open new directions for understanding
related poor H2-splitting hydrogenation catalysts such as Au,
Cu, and Fe–Al.
by using
a Mettler Toledo TGA/DSC 1 Star system under
40 cm3 minꢀ1 of air by ramping the temperature from 298 to
1173 K (ramping rate=10 Kminꢀ1).
Catalyst testing
The gas-phase hydrogenation of propyne was studied at ambient
pressure in a continuous-flow fixed-bed microreactor (internal di-
ameter=12 mm) using 0.2 g of catalyst (particle size=0.2–
0.4 mm). Catalyst screening was performed at different tempera-
tures (373–523 K) and H2/C3H4 ratios (5–25), with a contact time of
t=1 s. For this purpose, the partial H2 pressure (p(H2)) was varied
in the range of 125–625 mbar, with a fixed partial C3H4 pressure of
25 mbar, and using He as the balance gas. To evaluate the formal
kinetic dependence of the reaction rate on the reactants, p(C3H4)
was varied in the range of 25–125 mbar (with p(H2)=625 mbar)
and p(H2) was varied in the range of 125–625 mbar (with p(C3H4)=
25 mbar) at T=473 K and t=0.3 s. Under these conditions, the
conversion of propyne was less than 20%. All the catalytic data
were measured in steady state, and the performance was stable
during at least 5 h on stream. Tests with different sieve fractions in
the particle size range of 0.15–0.6 mm and different total volumet-
ric flows (3–84 cm3 minꢀ1) at constant t=1 s were conducted to
discard internal and external mass transport limitations. The tem-
perature gradients along the bed were less than 5 K. The composi-
tion of the gas at the reactor outlet was analyzed with an online
gas chromatograph (Agilent GC7890A) equipped with a GS-GasPro
column and a flame ionization detector. The conversion of pro-
pyne, X(C3H4), was determined as the amount of reacted alkyne di-
vided by the amount of alkyne at the reactor inlet. The selectivity
to propene (S(C3H6)) and propane (S(C3H8)) was calculated as the
amount of product formed divided by the amount of alkyne con-
verted. The selectivity to oligomers (OL) was obtained by using the
carbon balance, S(OL)=1ꢀS(C3H6)ꢀS(C3H8). It was assumed that all
missing carbon atoms in the balance comprised oligomers. TGAs
Experimental Section
Catalyst preparation
Commercial SiO2 (Evonik, 99.8%), g-Al2O3 (Merck, 99%), and TiO2
(anatase, Sigma–Aldrich, 99.7%) carriers were used as received.
Supported catalysts with a nominal Ag loading of 0.5–5 wt% were
prepared with AgNO3 (Sigma–Aldrich, 99.9%) as the Ag precursor.
The spray deposition method was performed with a Büchi Mini
Spray Dryer B-290 equipped with a two-fluid nozzle of 1.4 mm in
diameter. This technique, pioneered by Yara International,[31] ena-
bled the deposition of metal (oxide) particles with a high degree
of dispersion.[32] Silver nitrate (0.02–0.2 g) was dissolved in deion-
ized water (20 cm3) under magnetic stirring at RT, followed by the
addition of the support (2 g). The resulting suspension was
pumped at 3 cm3 minꢀ1 into the two-fluid nozzle, together with
a spray air flow of 0.5 m3 hꢀ1, which created droplets of 20–30 mm.
The inlet temperature was set at 493 K, the aspiration rate at
35 m3 hꢀ1, and the outlet temperature at 383 K. The dried particles
were separated by using a cyclone. The impregnation method was
performed by dissolving silver nitrate (0.011 g) in deionized water
(1.5 cm3) and by adding the solution to SiO2 (1.1 g). The paste was
dried under freezing and conventional conditions. Freeze drying
was performed on a Labconco FreeZone Plus 2.5 L Cascade Bench-
top Freeze Dry System at 40 K and 0.02 bar. Conventional drying
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
8
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
of the catalysts after propyne hydrogenation enabled the closing
of the carbon balance to 97 wt%, and the transmission electron
micrograph of these samples did not reveal the presence of coke
deposits (Figure S1).
a given quantity of matter will attain a shape that minimizes the
total surface energy of the system, which is the shape in which the
distance of each face from the center is proportional to the surface
tension of the respective surface. The original version of the model
is continuous and size independent. Therefore, the method might
indicate the formation of highly active surfaces for structures for
which there is not enough space to generate an open facet. The
atomistic Wulff construction[42–44] mitigates this problem, because it
further required that the enclosed volume had to fit the individual
Ag atoms. This approach has been used to predict the equilibrium
shape of Ru[42] and Au[43,44] nanoparticles interacting with the
surrounding environment.
Density functional theory
DFT calculations based on slabs were performed with the Vienna
Ab initio Simulation Package code[34] and the revised Perdew–
Burke–Ernzerhof (RPBE) functional.[35] To account for dispersion in-
teractions, the Grimme-based correction was used.[36–38] This ap-
proach ensured a correct treatment of the gas-phase reference
while including dispersion contributions. Core electrons were re-
placed by projector-augmented wave (PAW) pseudopotentials,[39]
whereas valence electrons were expanded in plane waves with a ki-
netic cutoff energy of 400 eV. As the planar Ag(111) surface does
not interact with reactants, a stepped (211) facet was used to rep-
resent the active sites of the Ag nanoparticles. The slab contained
nine metal layers, in a p(2”1) supercell for the adsorption of the in-
dividual molecules and in a p(4”1) supercell when assessing hydro-
genation and oligomerization paths. In all cases, the active site was
surrounded by empty positions, thus excluding lateral interactions.
The k-point sampling was derived from the Monkhorst–Pack
scheme.[40] Transition states were located through the climbing-
image version of the nudged elastic band (CI-NEB) algorithm.[41]
The vibrational analysis of the potential transition state structures
was performed to fully characterize the saddle points. For this
reason, the numerical Hessian matrix, obtained with a displacement
of 0.02 Š for each degree of freedom, was diagonalized to obtain
the corresponding eigenvalues and eigenmodes, and it showed
a single imaginary frequency.
Acknowledgements
This study was financially supported by ETH Zurich, the Minister-
io de Economꢁa y Competitividad (Spain) (CTQ2012-33826), and
the ICIQ Foundation. We are grateful to Dr. S. Mitchell and Dr. F.
Krumeich for performing electron microscopic analyses. The Elec-
tron Microscopy Centre of the Swiss Federal Institute of Technolo-
gy (EMEZ) is acknowledged for the use of their facilities. BSC-RES
is thanked for providing generous computational resources.
Keywords: B5 sites
· H2 activation · hydrogenation ·
mechanistic studies · nanoparticles · olefins · silver
[1] H. F. Rase in Handbook of Commercial Catalysts: Heterogeneous Catalysts,
CRC Press, Boca Raton, 2000, p. 110.
[2] S. Nishimura in Handbook of Heterogeneous Catalytic Hydrogenation for
Organic Synthesis, Wiley, New York, 2001, p. 63.
[3] M. L. Derrien in Catalytic Hydrogenation, Stud. Surf. Sci. Catal., Vol. 27
ˇ
´
(Ed.: L. Cerveny), Elsevier, Amsterdam, 1986, p. 613.
[4] D. Teschner, J. Borsodi, A. Wootsch, Z. Rꢀvay, M. Hävecker, A. Knop-Ger-
icke, S. D. Jackson, R. Schlçgl, Science 2008, 320, 86.
Kinetic analysis and structural models
The rate of the elementary steps was determined by multiplying
the coverage of reactants by the corresponding rate coefficient, k
(sꢀ1), which was computed according to the transition state theory
of Eyring, Evans, and Polanyi [Eq. (11)]:
[5] J. A. Anderson, J. Mellor, R. P. K. Wells, J. Catal. 2009, 261, 208.
[6] D. Mei, M. Neurock, C. M. Smith, J. Catal. 2009, 268, 181.
[7] M. Crespo-Quesada, A. Yarulin, M. Jin, Y. Xia, L. Kiwi-Minsker, J. Am.
Chem. Soc. 2011, 133, 12787.
ꢀ
ꢁ
ꢀ
ꢁ
[8] F. Studt, F. Abild-Pedersen, T. Bligaard, R. Z. Sørensen, C. H. Christensen,
J. K. Nørskov, Science 2008, 320, 1320.
[9] M. Armbrüster, K. Kovnir, M. Behrens, D. Teschner, Y. Grin, R. Schlçgl, J.
Am. Chem. Soc. 2010, 132, 14745.
Ea
kBT
Ea
kBT
k ¼ k0 exp ꢀ
ꢂ 1013 exp ꢀ
ð11Þ
Here, k0 is the preexponential factor, taken as 1013 for all the reac-
tions on the surface; Ea is the ZPVE-corrected activation energy; kB
is the Boltzmann constant; and T is the temperature. The rate of
adsorption steps (Steps A1, A2 and D1, D2) was calculated by
using the Hertz–Knudsen equation [Eq. (12)].
[10] B. Bridier, N. López, J. Pꢀrez-Ramírez, Dalton Trans. 2010, 39, 8412.
[11] G. Kyriakou, M. B. Boucher, A. D. Jewell, E. A. Lewis, T. J. Lawton, A. E.
Baber, H. L. Tierney, M. Flytzani-Stephanopoulos, E. C. H. Sykes, Science
2012, 335, 1209.
[12] M. Armbrüster, K. Kovnir, M. Friedrich, D. Teschner, G. Wowsnick, M.
Hahne, P. Gille, L. Szentmiklósi, M. Feuerbacher, M. Heggen, F. Girgsdies,
D. Rosenthal, R. Schlçgl, Y. Grin, Nat. Mater. 2012, 11, 690.
[13] J. Horiuti, M. Polanyi, Trans. Faraday Soc. 1934, 30, 1164.
[14] R. L. Augustine, J. F. van Peppen, Ann. N. Y. Acad. Sci. 1970, 172, 244.
[15] B. Hammer, J. K. Nørskov, Nature 1995, 376, 238.
[16] G. Lee, E. W. Plummer, Phys. Rev. B 1995, 51, 7250.
[17] A. Montoya, A. Schlunke, B. S. Haynes, J. Phys. Chem. B 2006, 110,
17145.
[18] A. B. Mohammad, K. H. Lim, I. V. Yudanov, K. M. Neyman, N. Rçsch, Phys.
Chem. Chem. Phys. 2007, 9, 1247.
[19] S. Klacar, H. Grçnbeck, Catal. Sci. Technol. 2013, 3, 183.
[20] A. Sµrkµny, Z. Rꢀvay, Appl. Catal. A 2003, 243, 347.
[21] D. Balcells, E. Clot, O. Eisenstein, Chem. Rev. 2010, 110, 749.
[22] C. Copꢀret, Chem. Rev. 2010, 110, 656.
*
pðiÞqð Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
rads
¼
Acat
ð12Þ
2pmðiÞkBT
Here, p(i) and m(i) are the partial pressure and the mass of the gas
phase species i, respectively, and Acat is the area of one free active
site. Site balances were used to ensure that the number of sites on
a catalyst is constant; hence, all coverages add up to unity. The mi-
crokinetic fitting was performed by using a plug-flow reactor
model. This is reasonable because the Pꢀclet number, defined as
the ratio between the rates of axial advection and radial diffusion,
was always greater than 20.
By using the atomistic Wulff method, we constructed structural
models for the equilibrium shape of small and large Ag nanoparti-
cles. The Wulff construction was based on the simple geometrical
criterion that under thermodynamic equilibrium conditions,
[23] V. Amir-Ebrahimi, J. J. Rooney, Catal. Lett. 2009, 127, 20.
[24] T. Vergunst, F. Kapteijn, J. A. Moulijn, Appl. Catal. A 2001, 213, 179.
[25] T. M. Eggenhuisen, P. Munnik, H. Talsma, P. E. de Jongh, K. P. de Jong, J.
Catal. 2013, 297, 306.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
9
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[26] DFT predicts that oligomers are mostly formed via the CꢀC coupling of
propene. The coupling of two propenes (2C3H6*QC6H12**, with DE=
ꢀ0.51 eV and an energy barrier of 0.51 eV) is favored over the coupling
of two propynes (2C3H4*QC6H8**, with DE=ꢀ0.61 eV and an energy
barrier of 1.25 eV). The involvement of propene in the oligomerization
reaction is experimentally confirmed, as shown in Figure S12.
[27] G. C. Bond, P. B. Wells, J. Catal. 1966, 5, 65.
[35] B. Hammer, L. B. Hansen, J. K. Nørskov, Phys. Rev. B 1999, 59, 7413.
[36] S. Grimme, J. Comput. Chem. 2006, 27, 1787.
[37] T. Bucko, S. Lebegue, J. G. Angyan, J. Hafner, J. Phys. Chem. A 2010, 114,
11814.
[38] V. G. Ruiz, W. Liu, E. Zojer, M. Scheffler, A. Tkatchenko, Phys. Rev. Lett.
2012, 108, 146103.
[39] P. E. Blçchl, Phys. Rev. B 1994, 50, 17953.
[28] R. S. Mann, S. C. Naik, Can. J. Chem. 1968, 46, 623.
[29] G. C. Bond in Metal-catalysed Reactions of Hydrocarbons, Springer, New
York, 2005, p. 421.
[40] H. J. Monkhorst, J. D. Pack, Phys. Rev. B 1976, 13, 5188.
[41] G. Henkelman, B. P. Uberuaga, H. Jonsson, J. Chem. Phys. 2000, 113,
9901.
[30] D. Teschner, G. Novell-Leruth, R. Farra, A. Knop-Gericke, R. Schlçgl, L.
Szentmiklósi, M. Hevia, H. Soerijanto, R. Schomäcker, J. Pꢀrez-Ramírez,
N. López, Nat. Chem. 2012, 4, 739.
[31] A. H. Øygarden, J. Pꢀrez-Ramírez, D. Waller, K. Schçffel, D. Brackenbury,
WO2004/110622, 2004.
[42] K. Honkala, A. Hellman, I. N. Remediakis, A. Logadottir, A. Carlsson, S.
Dahl, C. H. Christensen, J. K. Nørskov, Science 2005, 307, 555.
[43] G. D. Barmparis, I. N. Remediakis, Phys. Rev. B 2012, 86, 085457.
[44] G. D. Barmparis, K. Honkala, I. N. Remediakis, J. Chem. Phys. 2013, 138,
064702.
[32] M. Santiago, A. Restuccia, F. Gramm, J. Pꢀrez-Ramírez, Microporous Mes-
oporous Mater. 2011, 146, 76.
Received: July 15, 2013
[33] R. van Hardeveld, F. Hartog, Surf. Sci. 1969, 15, 189.
[34] G. Kresse, J. Hafner, Phys. Rev. B 1993, 47, 558.
Published online on && &&, 0000
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&10
&
ÞÞ
These are not the final page numbers!

FULL PAPERS
The money metal! Silver selectively cat-
alyzes the hydrogenation of propyne to
propene, and the activity is maximal
over 4.5 nm nanoparticles. The rate of
propene production correlates well with
the density of B5 sites, which suggests
that the latter are potential active cen-
ters in the reaction. The hydrogenation
follows an associative scheme, featuring
the activation of H2 directly on the pro-
pyne–silver-surface intermediates.
G. Vilꢀ, D. Baudouin, I. N. Remediakis,
C. Copꢀret,* N. Lꢂpez,* J. Pꢀrez-Ramꢁrez*
&& – &&
Silver Nanoparticles for Olefin
Production: New Insights into the
Mechanistic Description of Propyne
Hydrogenation
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&11
&
ÞÞ
These are not the final page numbers!