From the lindlar catalyst to supported ligand-modified palladium nanoparticles: Selectivity patterns and accessibility constraints in the continuous-flow three-phase hydrogenation of acetylenic compounds

Article abstract of DOI:10.1002/chem.201304795

Site modification and isolation through selective poisoning comprise an effective strategy to enhance the selectivity of palladium catalysts in the partial hydrogenation of triple bonds in acetylenic compounds. The recent emergence of supported hybrid materials matching the stereo- and chemoselectivity of the classical Lindlar catalyst holds promise to revolutionize palladium-catalyzed hydrogenations, and will benefit from an in-depth understanding of these new materials. In this work, we compare the performance of bare, lead-poisoned, and ligand-modified palladium catalysts in the hydrogenation of diverse alkynes. Catalytic tests, conducted in a continuous-flow three-phase reactor, coupled with theoretical calculations and characterization methods, enable elucidation of the structural origins of the observed selectivity patterns. Distinctions in the catalytic performance are correlated with the relative accessibility of the active site to the organic substrate, and with the adsorption configuration and strength, depending on the ensemble size and surface potentials. This explains the role of the ligand in the colloidally prepared catalysts in promoting superior performance in the hydrogenation of terminal and internal alkynes, and short-chain alkynols. In contrast, the greater accessibility of the active surface of the Pd-Pb alloy and the absence of polar groups are shown to be favorable in the conversion of alkynes containing long aliphatic chains and/or ketone groups. These findings provide detailed insights for the advanced design of supported nanostructured catalysts. Hybrid nanocatalysts: The classical Lindlar and the newly developed NanoSelect^TM catalysts are confronted in the semi-hydrogenation of alkynes (see figure). Systematic testing under continuous-flow three-phase conditions, coupled with detailed characterization analyses and molecular simulations, enable the understanding of the structure of the catalysts and the associated activity and selectivity patterns for a wide range of acetylenic compounds.

Full text of DOI:10.1002/chem.201304795

DOI: 10.1002/chem.201304795
Full Paper
&
Heterogeneous Catalysis
From the Lindlar Catalyst to Supported Ligand-Modified
Palladium Nanoparticles: Selectivity Patterns and Accessibility
Constraints in the Continuous-Flow Three-Phase Hydrogenation
of Acetylenic Compounds
Gianvito Vilꢀ,[a] Neyvis Almora-Barrios,[b] Sharon Mitchell,[a] Nfflria López,*[b] and Javier Pꢀrez-
Ramírez*[a]
Abstract: Site modification and isolation through selective
poisoning comprise an effective strategy to enhance the se-
lectivity of palladium catalysts in the partial hydrogenation
of triple bonds in acetylenic compounds. The recent emer-
gence of supported hybrid materials matching the stereo-
and chemoselectivity of the classical Lindlar catalyst holds
promise to revolutionize palladium-catalyzed hydrogena-
tions, and will benefit from an in-depth understanding of
these new materials. In this work, we compare the per-
formance of bare, lead-poisoned, and ligand-modified palla-
dium catalysts in the hydrogenation of diverse alkynes. Cata-
lytic tests, conducted in a continuous-flow three-phase reac-
tor, coupled with theoretical calculations and characteriza-
tion methods, enable elucidation of the structural origins of
the observed selectivity patterns. Distinctions in the catalytic
performance are correlated with the relative accessibility of
the active site to the organic substrate, and with the adsorp-
tion configuration and strength, depending on the ensemble
size and surface potentials. This explains the role of the
ligand in the colloidally prepared catalysts in promoting su-
perior performance in the hydrogenation of terminal and in-
ternal alkynes, and short-chain alkynols. In contrast, the
greater accessibility of the active surface of the Pd–Pb alloy
and the absence of polar groups are shown to be favorable
in the conversion of alkynes containing long aliphatic chains
and/or ketone groups. These findings provide detailed in-
sights for the advanced design of supported nanostructured
catalysts.
Introduction
maximize the alkene selectivity involves the modification of
palladium by the addition of promoters.[4–13] The Lindlar cata-
lyst (5 wt.% palladium deposited on calcium carbonate or
barium sulfate and treated with various forms of lead, yielding
a total weight Pd/Pb ratio of 1.5),[14] is the prototypic example
of modified Pd catalysts. The widespread use of the Lindlar
catalyst for over more than six decades derives from its unique
ability to stereo- and chemoselectively transform CꢀC into C=C
bonds with a predominantly cis configuration under mild reac-
tion conditions. Density functional theory (DFT) calculations of-
fered a solid understanding of the selectivity of the Lindlar cat-
alyst,[15] unraveling the role of Pd in alkyne adsorption, H2 split-
ting, and H addition, the positive poisoning of Pb in reducing
hydride formation, and the function of quinoline, generally
added to the reaction mixture, in preventing oligomerization
by isolating the adsorption sites. At the present time, econom-
ic considerations associated with the high palladium loading
and stringent environmental regulations regarding the use of
lead provide incentives to seek alternatives to the Lindlar cata-
lyst for alkyne hydrogenations.
The partial hydrogenation of acetylenic compounds in the
liquid phase is a crucial step in the synthesis of Z alkenes,
which are indispensable building blocks in the manufacture of
polymers, vitamins, fragrances, and agrochemicals.[1] This
family of reactions is almost universally carried out over
heterogeneous palladium-based catalysts.[2] Major challenges
in the Pd-catalyzed partial hydrogenation of alkynes are to pre-
vent the over-hydrogenation of alkenes to the corresponding
alkanes and/or the oligomerization of unsaturated moieties.
Some degree of control can be achieved by adjustment of the
reaction conditions, but the effectiveness is substrate specific,
requiring individual tuning.[3] A more common approach to
[a] G. Vilꢀ, Dr. S. Mitchell, Prof. J. Pꢀrez-Ramꢁrez
Institute for Chemical and Bioengineering
Department of Chemistry and Applied Biosciences, ETH Zurich
Vladimir-Prelog-Weg 1, 8093 Zurich (Switzerland)
E-mail: jpr@chem.ethz.ch
[b] Dr. N. Almora-Barrios, Prof. N. Lꢂpez
Institute of Chemical Research of Catalonia, ICIQ
Av. Pa¨ısos Catalans 16, 43007 Tarragona (Spain)
E-mail: nlopez@iciq.es
Many catalytic systems have been developed in the last
decade, such as: 1) Pd nanoparticles prepared with novel dep-
osition techniques;[16–20] 2) Pd-containing bimetallic alloys;[9,13]
and 3) catalysts based on alternative active phases.[21–26] In rela-
tion to the preparation strategies, colloidal approaches have
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201304795.
Chem. Eur. J. 2014, 20, 5926 – 5937
5926
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
proven to be the most versatile tool for synthesizing unsup-
ported nanoparticles with tailored size and shape.[27–34] These
materials are prepared from colloidal suspensions of metal
nanoparticles and exploit both reducing (i.e., hydrazine or
NaBH4) and stabilizing (i.e., surfactants, ionic liquids, polymers,
or dendrimers) agents as ligands.[27] Industrially, the use of un-
supported Pd nanoparticles prepared through colloidal routes
is typically hampered by the use of volatile, toxic, and expen-
sive organic reagents,[35] the absence of a carrier limiting their
application as heterogeneous catalysts, and the need for ex-
pensive separation techniques (i.e., centrifugation).
Recently, a novel platform of supported ligand-modified pal-
ladium catalysts (0.5 wt.% palladium supported on activated
carbons or titanium silicate; NanoSelectTM) has been developed
by BASF and is now commercialized.[35] The catalysts are pre-
pared by using the non-toxic and water-soluble hexadecyl-2-
hydroxyethyl-dimethyl ammonium dihydrogen phosphate
(C20H44NO5P, HHDMA) ligand, which elegantly integrates both
Figure 2. Setup for the evaluation of catalysts in three-phase hydrogenation.
The main unit (1) is equipped with an isothermal micro-reactor (zoomed in
(2)), pressure and flow controllers, and an electrolytic cell for in situ H2 gen-
eration. The liquid reactant is fed by an HPLC pump (3). The CatChangerTM
(4), which consists of six additional isothermal microreactors, enables high-
throughput testing and kinetic studies. The XYZ autosampler (5) allows reac-
tant feeding and product collection.
reducing and stabilizing functions in
a single molecule
(Figure 1). The intrinsic stereo- and chemoselectivity of these
improved understanding of the
performance of colloidal nano-
particles in alkyne hydrogena-
tion, opening new pathways for
the design of highly selective
catalysts with tailored properties.
Figure 1. Representation of the NanoSelectTM catalyst (adapted from Ref. [36]) (a) and molecular structure of crys-
talline hexadecyl(2-hydroxyethyl)dimethylammonium dihydrogen phosphate (HHDMA) determined by DFT calcula-
tions (b). The dashed line represents the cell used for the calculations. Color codes: P in pink, O in red, N in dark
blue, C in grey, and H in white. The crystallographic information file is provided in the Supporting Information.
Results and Discussion
Characterization of the Pd-
based catalysts
materials have been demonstrated in the hydrogenation of se-
lected alkynes and alkynols,[35,36] and the results are consistent
with earlier findings on supported Pd catalysts stabilized by
polyvinylpyrrolidone (PVP).[37] However, a systematic evaluation
of the hydrogenation performance of the NanoSelectTM cata-
lysts, tackling a wider library of acetylenic compounds, investi-
gating kinetic fingerprints, and leading to an improved under-
standing of the role of HHDMA are of foremost importance to
rationalize their selective behavior. Besides, a comparison of
the operation mode of NanoSelectTM-type catalysts with Lin-
dlar-type catalysts is of great fundamental relevance.
Four supported palladium catalysts (bare Pd/Al2O3, Lindlar-like
Pd-Pb/CaCO3, and the two ligand-modified c-Pd/C and c-Pd/
TiS) are characterized by multiple techniques in order to estab-
lish the relative loading and dispersion of palladium, the mor-
phology and porosity of the support, and the content and
thermal stability of organic molecules adsorbed on the ligand-
modified materials.
The Pd/Al2O3 catalyst has been characterized extensively in
prior works.[6,39] The material contains approximately 1 wt.% Pd
(Table 1); uniformly sized palladium nanoparticles of about
4 nm average diameter (Figure S1a in the Supporting Informa-
tion) decorate the Al2O3 support. The catalyst exhibits a Bruna-
uer–Emmett–Teller (BET) surface area of 181 m2 gÀ1 and a pore
volume of 0.90 cm3 gÀ1 (Table 1). Comparatively, the Lindlar cat-
alyst contains 4.53 wt.% Pd and, in addition, 3.41 wt.% Pb. The
surface atomic Pd/Pb ratio, determined by XPS, is about 1.3, in
agreement with early reports.[40] Microscopic examination re-
veals that the Pd–Pb nanoparticles are well-distributed over
the support (Figure S1b), but are significantly larger (ca. 14 nm
in diameter) than in the case of Pd/Al2O3. This is expected con-
sidering that the Lindlar catalyst contains a five times higher
palladium loading. The available BET surface area is limited
(10 m2 gÀ1), because the basic CaCO3 support is composed of
large, nonporous particles (Vpore =0.03 cm3 gÀ1). Particularly, the
Revisiting the development of catalytic materials for alkyne
hydrogenation, this work compares the selectivity of bare and
alloyed Pd-based catalysts with emerging HHDMA-modified
materials prepared by colloidal routes. In view of the obvious
benefits in moving from batch- to continuous-mode opera-
tion,[38] the hydrogenation performances of terminal or internal
alkynes, short- and long-chain alkynols, and acetylenic species
containing ketonic groups are evaluated in a continuous-flow
reactor (Figure 2). To rationalize the catalytic results, we have
carried out density functional theory (DFT) calculations and
classical molecular dynamics (MD) simulations, investigating
the atomic structure of the NanoSelectTM catalysts and deriving
selectivity descriptors, which can be generalized to other
ligand-modified palladium catalysts. These results lead to an
Chem. Eur. J. 2014, 20, 5926 – 5937
www.chemeurj.org
5927
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
observation of a previous study
on the metal distribution in c-
Pd/C, which suggested that the
greater surface area of the acti-
Table 1. Characterization data of the palladium catalysts.
[c]
[d]
Catalyst
Pd[a]
Pb[a]
N[b]
SBET
Vpore
D[e]
[%]
[wt.%]
[wt.%]
[wt.%]
[m2 gÀ1
]
[cmÀ3 gÀ1
]
vated carbon support (SBET
=
Pd/Al2O3
Pd-Pb/CaCO3
c-Pd/C
1.00
4.53
0.56
0.52 (0.51)[f]
-
–
–
181
10
336
231 (137)[f]
0.90
0.03
0.41
22
6
12
14 (17)[f]
336 m2 gÀ1, Table 1) is hardly uti-
lized since an egg-shell distribu-
tion of palladium on the support
is obtained.[36] The denser ar-
rangement of nanoparticles in
the case of c-Pd/TiS can be as-
cribed to the larger particle size
3.41
–
–
0.72
0.49 (0.02)[f]
c-Pd/TiS
0.16 (0.12)[f]
[a] ICP-OES. [b] Elemental analysis. [c] BET method. [d] Volume adsorbed at p/p0 =0.99. [e] CO pulse chemisorp-
tion. [f] After UV–ozone treatment for 120 min between brackets.
absence of microporosity for the Pd/Al2O3 and Lindlar catalysts
is beneficial in order to avoid diffusion limitations.
of the support, although the BET surface area of the sample
(SBET =231 m2 gÀ1) is relatively high. The microporosity of the c-
Pd/C catalyst (Vmicro =0.03 cm3 gÀ1) is also confirmed by the
higher uptake of N2 at low relative pressures (Figure S2 in the
Supporting Information). The degree of Pd dispersion calculat-
ed from the microscopy-derived particle size distributions
(19% for c-Pd/C and 16% for c-Pd/TiS) is slightly higher than
that determined by CO chemisorption (12% and 14%, respec-
tively), indicating that some of the palladium atoms are inac-
cessible. XPS measurements (Figure S3 in the Supporting Infor-
mation) evidence a single peak for the Pd 3d5/2 core level (at
335.0 eV for c-Pd/C and 335.3 eV for c-Pd/TiS) corresponding
to Pd0,[41] which indicates that
The catalysts prepared by colloidal routes (c-Pd/C and c-Pd/
TiS) contain about 0.5 wt.%. Examination of the catalysts by
high-angle annular dark-field scanning transmission electron
(HAADF-STEM) and SEM (Figure 3a and b) reveals that the sup-
ported nanoparticles possess a uniform, spherical shape, exhib-
iting average diameters of around 6 and 7 nm for c-Pd/C and
c-Pd/TiS, respectively. Despite distinctions in the morphology
and porosity of the activated carbons and titanium silicate sup-
ports, the nanoparticles appear discretely dispersed over the
external surface of both materials. These findings confirm the
the surface of the Pd nanoparti-
cles is entirely reduced.
The presence of HHDMA mol-
ecules on the surface of the c-Pd
catalysts is confirmed by the ob-
servation of pronounced bands
between 3000 and 2800 cmÀ1 in
the infrared spectra of the sam-
ples (Figure 4a).[42,43] Despite the
similar preparation protocols,[35]
quantitative elemental analysis
of N reveals that the residual
amounts are higher for c-Pd/C
than for c-Pd/TiS. This is associat-
ed with an enhanced adsorption
of HHDMA on the surface of the
activated carbon support,[36]
which prevents a reliable estima-
tion of the amount of stabilizer
molecules per surface palladium
atom. The stability of the
HHDMA molecules is compared
by thermogravimetric analysis in
air (Figure 4b and c). Both sam-
ples show a pronounced weight
loss at 340 K, associated to the
removal of physisorbed water
from the support, and a further
loss at 570–580 K, which is as-
signed, based on evolved-gas
Figure 3. SEM (left) and HAADF-STEM (middle) micrographs and respective particle size distributions of palladium
of c-Pd/C (a), c-Pd/TiS (b), and c-Pd/TiS after UV–ozone treatment for 120 min (c). The 20 nm scale bar applies to
all micrographs.
analysis by mass spectrometry,
to the decomposition of the or-
Chem. Eur. J. 2014, 20, 5926 – 5937
www.chemeurj.org
5928
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Structure of HHDMA-modified Pd
From the characterization data, it appears that the ligand
HHDMA covers the surface of the colloidally prepared catalysts.
Therefore, dispersion-corrected DFT calculations have been ap-
plied in order to attain an atomic understanding of the struc-
ture of the c-Pd. So far, this has been impeded by factors such
as: 1) the need to account for all interactions including van der
Waals forces, which determine the stabilizing contribution of
the aliphatic tails; 2) the necessity for unit cells sufficiently
large to represent the surfactant structure; and 3) the need to
input the crystal structure of the surfactant, which in many
cases remains unknown.
The first step involves the definition of an appropriate crystal
structure for the HHDMA molecule in the presence of the dihy-
drogen phosphate anion. Although the lattice of the precursor,
(HHDMA)2PdCl4, has been reported in the literature,[35] this data
cannot be directly employed because of the different identity
of the anions. Nevertheless, the crystal lattice that we have cal-
culated (Figure 1b) shares common features with the structure
of its precursor. In fact, the two aliphatic chains of the HHDMA
molecule are arranged linearly by van der Waals interactions.
Electrostatic forces hold the quaternary ammonium head-
À
groups of HHDMA in close proximity to the H2PO4 anion, and
this interaction is further reinforced by a hydrogen bond be-
À
tween the CH2OH side tail and the H2PO4 group. The struc-
tures yield N–P contacts of 4.225 Š.
The anionic dihydrogen phosphate group strongly adsorbs
on the surface of Pd(111),[44] and the long-chain cationic coun-
terion adapts to the configuration of the anion by electrostatic
interactions. The adsorption is exothermic by 4.59 eV per
HHDMA molecule and the final structure fits naturally in the
p(2”2) supercell without any further restriction. The three
oxygen atoms of the phosphate group reside close to the
metal sites, with an average PdÀO distance of 2.524 Š, and the
cationic headgroups are located 7.593 Š from the Pd surface.
Figure 4. CÀH stretching region of the DRIFT spectra recorded in He at
473 K (a). Thermogravimetric profiles in air with the corresponding MS sig-
nals of water (m/e 18) and carbon dioxide (m/e 44) of c-Pd/C (b) and c-Pd/
TiS (c). In (a), the reduced absorbance in the near-infrared region of the
darker carbon-supported catalyst compared with c-Pd/TiS precludes quanti-
tative analysis in the infrared spectra. In (b) and (c), the dashed line repre-
sents the first derivative curve of the thermogravimetric profile.
À
ganic capping layers. The c-Pd/C catalyst also shows a weight
loss at ca. 800 K, associated with the decomposition of the
support.[36] In this case, the oxidative decomposition of the
support dominates the mass signal measured.
Upon adsorption, the H2PO4 species tend to dissociate (Fig-
ure S4, Table S1 in the Supporting Information), forming
2À
HPO4 and H+ and further releasing 0.29 eV. In this process,
the whole ligand moves closer to the surface. In fact, the aver-
To confirm the impact of the residual HHDMA species on
the properties of the supported catalysts, we have applied
controlled UV–ozone treatments to remove the organic
layer.[30,34] Herein, the simultaneous action of ozone and ultra-
violet light oxidizes carbon-containing compounds into carbon
dioxide and water. As shown in Figure 4a, the IR spectrum of
c-Pd/TiS shows negligible absorbance at 3000–2800 cmÀ1 after
UV–ozone treatment for 2 h. This confirms the almost com-
plete removal of the organic layer, as also evidenced by ele-
mental analysis of N (Table 1). Comparison of the SEM/HAADF-
STEM micrographs (Figure 3c) demonstrates that the UV–
ozone treatment does not significantly alter the size or mor-
phology of the Pd nanoparticles, and the Pd content deter-
mined by inductively coupled plasma optical emission spec-
troscopy (ICP-OES) is unaltered.
age PdÀO distance reduces to 2.365 Š, the PdÀN distance is
2À
7.469 Š, and the hydrogen bond between the HPO4 anion
and the CH2OH tail is 2.375 Š. As shown in Figure 5 and addi-
tionally displayed in Movie 1 (in the Supporting Information),
this structure very effectively limits the size of the Pd ensem-
ble, which can be expected to block potential oligomerization
pathways. DFT calculations indicate that the enlargement of
the ensemble size to accommodate a bulky adsorption config-
uration by displacement of a hydrogen phosphate on the sur-
face of palladium would incur a large energy penalty (1.19 eV)
due to the repulsive interactions between the anionic head
groups. The latter ultimately explains the relative inflexibility of
the [H2PO4]À–Pd layer. On the other hand, classical MD simula-
tions performed in large simulation cells demonstrate that the
aliphatic chains of the surfactant are highly mobile, opening
pores that are dynamically closed in a nanosecond time frame
Chem. Eur. J. 2014, 20, 5926 – 5937
www.chemeurj.org
5929
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
In the hydrogenation of linear, terminal alkynes (1-pentyne
and 1-hexyne; respectively, Entry 1 and 2 of Table 2), the non-
promoted Pd/Al2O3 catalyst is poorly selective to the corre-
sponding olefins (ca. 70%) and yields over-hydrogenated prod-
ucts and oligomers. This is typical of monometallic Pd, which
needs modification in order to suppress the formation of hy-
drides and carbides, thereby stabilizing a high alkene yield. As
expected, the Lindlar catalyst shows an excellent degree of se-
lectivity to the terminal olefin (nearly 100%), owing to the
presence of lead, which strategically deactivates the palladium
sites. Comparatively, the two ligand-modified catalysts (c-Pd/C
and c-Pd/TiS) are also intrinsically selective (100%) to the pro-
duction of terminal olefins. This outstanding degree of selectiv-
ity has been confirmed for the solvent-free hydrogenation of
1-hexyne (100% selectivity to 1-hexene at 36% 1-hexyne con-
version). Upon application of substrates with an internal triple
bond, as for the hydrogenation of 3-hexyne (Entry 3), the ma-
terials prepared through colloidal routes yield the cis-alkene.
This stereoselectivity, matching the performance of conven-
tional Pd-Pb alloys at much lower Pd loadings, is attractive for
the pharmaceutical and fine chemical industries.
On the other hand, in the hydrogenation of enynes and al-
kynols (Entries 4–7), the chemoselectivity patterns observed
over the Lindlar and ligand-modified catalysts depend on the
substrate. In the hydrogenation of valylene (Entry 4), the colloi-
dal and Lindlar catalysts predominantly yield isoprene, and the
selectivity to 3-methyl-1-butyne, obtained by the reduction of
the double bond, is only about 5%. ‘Bare’ palladium, in con-
trast, produces isoprene, 3-methyl-1-butyne, and other over-
hydrogenated products. Similarly, the lowest degree of selec-
tivity in the hydrogenation of 2-methyl-3-butyn-2-ol (Entry 5),
3-methyl-1-pentyn-3-ol (Entry 6), and 3-methyl 1-penten-4-yn-
3-ol (Entry 7) is observed over the Pd-only nanoparticles (S-
(alkenols)=70, 67, and 58%, respectively). In this case, the Lin-
Figure 5. Side (a) and top view (b) of the DFT-calculated adsorption configu-
ration of HHDMA molecules on Pd(111). The configuration of the adsorbed
anion is HPO42À. The blue spheres represent Pd sites, and the ball and stick
models represent the adsorbed HHDMA. The color codes are indicated in
Figure 1. A movie file displaying the 3D structure is provided as Supporting
Information.
(Figure S5 in the Supporting In-
formation). These pores present
Table 2. Hydrogenation of acetylenic compounds over palladium catalysts.
elliptical shapes with a maximum
diameter of less than 10 Š.
Entry
Substrate
Desired product
Selectivity[a] [%]
Pd/Al2O3
Pd-Pb/CaCO3
98
c-Pd/C
100
c-Pd/TiS
100
1
71
Performance of the Pd-based
catalysts
2
3
4
67
53
65
100
100
92
97
97
93
96
100
96
The selective properties of the
Pd catalysts were evaluated in
the reduction of acetylenic spe-
cies of varying complexity, under
5
6
70
67
85
83
100
100
96
97
continuous-flow
liquid-phase
conditions. In contrast to batch
and semi-batch setups, com-
monly employed for laboratory
testing, a continuous-flow three-
phase reactor can be efficiently
operated at steady state, offer-
ing several advantages for indus-
trial implementation.[38]
7
8
9
58
73
74
76
97
83
86
85
63
87
93
66
[a] Reaction conditions: T=293 K, P=1 bar, FG(H2)=18 cm3 minÀ1, FL(substrate+solvent)=0.3 cm3 minÀ1. Under
these conditions, the degree of alkyne conversion is between 24–35% over all catalysts.
Chem. Eur. J. 2014, 20, 5926 – 5937
www.chemeurj.org
5930
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
dlar catalyst shows a higher selectivity (85, 83, and 76%, re-
spectively) at the same degree of alkynol conversion; and no-
tably, the colloidally prepared catalysts are fully chemoselective
to the alkene, as the OH functionality is not reduced. This ex-
cellent result, exceeding those of conventional Pd systems,
should be connected with the presence of HHDMA on the sur-
face of Pd (vide infra). However, in the hydrogenation of al-
kynes containing a ketone group, such as 3-butyn-2-one
(Entry 8) and 3-hexyn-2-one (Entry 9), the novel c-Pd catalysts
are less selective to 3-buten-2-one and cis-3-hexen-2-one than
Pd-Pb/CaCO3. Considering that at elevated temperature (333 K)
and pressure (8 bar), the ligand-modified materials yield unsa-
turated alcohols (3-buten-2-ol and cis-3-hexen-2-ol) with high
selectivity (85%), it can be expected that the ketone group in-
teracts with the surface of Pd and/or with the polar and protic
functionalities of HHDMA.
dlar and ligand-modified catalysts over a broad degree of con-
version was also confirmed in the hydrogenation of 3-hexyne
(Figure 6b), and valylene (Figure 6c). The selectivity to the
alkene product over the c-Pd catalysts only drops when higher
temperatures (293-353 K) and pressures (1–9 bar) are applied,
and the contour plot in Figure 7 highlights the selectivity
window for the hydrogenation of 1-hexyne over the ligand-
modified catalysts (T<313 K and P<2 bar); outside this range,
n-hexane and 2-hexene are formed.
Focusing on the hydrogenation of 1-hexyne, Figure 6a dis-
plays the relation between the degree of alkyne conversion
and the selectivity to 1-hexene, at 293 K and 1 bar. Over the
monometallic Pd/Al2O3, the selectivity to 1-hexene decreases
with an increasing alkyne conversion. Here, the Lindlar catalyst
and c-Pd/TiS preferentially form the olefin (ca. 100%), even at
X(1-hexyne)>90%. The intrinsic selective character of the Lin-
Figure 7. Contour plots of the hydrogenation of 1-hexyne over c-Pd/TiS at
different temperatures and pressures. Reaction conditions:
FG(H2)=18 cm3 minÀ1, FL(substrate+solvent)=0.3 cm3 minÀ1
.
It should be emphasized that the equivalent selectivity evi-
denced over the two colloidally prepared Pd catalysts in all of
the catalytic tests suggests that the specific influence of the
support is minor (Figure S6 in the Supporting Information).
This result differs from previous findings, which reported
a lower degree of selectivity over c-Pd/C.[36] The contrasting
observation could relate to the different mode of operation
(batch vs. continuous-flow reactor) adopted in that study.[38]
Figure 6. Selectivity to 1-hexene (a), cis-3-hexene (b), and isoprene (c) as
a function of the conversion of 1-hexyne, 3-hexyne, and valylene over the
palladium catalysts. Reaction conditions: T=293 K, P=1 bar, FG(H2)=9–
60 cm3 minÀ1, FL(substrate+solvent)=0.3–3 cm3 minÀ1. The data at 100%
conversion were collected with double the amount of catalyst.
Chem. Eur. J. 2014, 20, 5926 – 5937
www.chemeurj.org
5931
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Molecular-level understanding of the selectivity patterns
catalyst is À0.54 eV per H atom, and the differential adsorption
energy in the subsurface is thermoneutral. Thus, the adsorp-
tion takes place on the whole surface and, when the first layer
is completely covered, on subsurface positions. The H species
accumulated in the subsurface of palladium are easily trans-
ferred to the adsorbed hydrocarbons,[4,6] yielding fully saturat-
ed species, in agreement with the results displayed in Table 2,
Figure 6, and Figure 7 for Pd/Al2O3. For the Pd–Pb system, the
exothermic adsorption energy of H on the surface (À0.52 eV)
is similar to that of Pd, although less active sites are available.
Moreover, the differential adsorption energy for the addition of
H layers in the subsurface is endothermic (0.23 eV). Therefore,
Pd alloys are less prone to form b-hydrides.[15] Interestingly, the
action of the ligand on the catalyst prepared by colloidal
routes is similar to that of Pb in the Lindlar catalyst, effectively
reducing the number of active sites (geometric effect) and the
adsorption energy of H in the first layer (À0.39 eV) (electronic
effect). This explains the intrinsic alkene selectivity of the
HHDMA-modified catalysts (Table 2 and Figure 6). However, the
adsorption of H in the subsurface is only slightly endothermic
(0.07 eV). Therefore, if high H2 pressures are employed, over-
hydrogenation reactions cannot be avoided, as observed in
Figure 7. The solvent might, to a certain extent, mitigate this
problem, since it may reduce the amount of H2 in contact with
the surface. Isomerization of the alkenes takes place through
the Horiuti-Polanyi mechanism, depending on the amount of
hydrogen that can be stored by the catalyst. As HHDMA-modi-
fied catalysts can keep a significant amount of hydrogen at
P>2 bar, this explains the experimental formation of 2-hexene
seen in Figure 7.
In order to rationalize the hydrogenation performance of the
catalysts presented in Table 2, DFT calculations using dispersion
interactions have been applied. We have considered the ad-
sorption of selected model reactants (specifically, ethyne,
ethene, formaldehyde, and methanol) in order to understand
the hydrogenation performance observed for the functional-
ized alkynes in Table 2. The adsorption energies are obtained
as: Eads =Ecat+molÀEcatÀEmol, where Ecat+mol is the energy of the
catalyst with the adsorbed reactant, Ecat is the energy of the
catalyst (eventually covered by the ligand, HHDMA), and Emol is
the energy of the specific reactant. The results are summarized
in Figure 8, and the corresponding structures are presented in
Figure 8. Adsorption energies of ethyne, ethene, formaldehyde, and metha-
nol on the Pd(111), Pb-Pd(111), and c-Pd(111) surfaces. For the c-Pd sample,
two different configurations of the adsorbed anion have been employed:
H2PO4À (c-Pd) and HPO42À (c-Pd-H).
Figure S7 in the Supporting Information. The theoretical results
for the alloyed Pd–Zn[9] and Pd–Ga,[25] which are also selective
hydrogenation catalysts, parallel those of Pd–Pb (Figure S8 in
the Supporting Information). Therefore, for the sake of concise-
ness, only the latter will be presented here.
The Lindlar and ligand-modified catalysts are intrinsically ste-
reoselective to cis alkenes in the hydrogenation of internal al-
kynes (Table 2). The stereoselectivity of the hybrid palladium
catalysts is related to the small ensembles generated by either
Pb alloying or by HHDMA adsorption. This limited ensemble
ensures a lower coverage of H on the catalyst surface, prevent-
ing the H addition to the C radical and the reorganization/iso-
merization of the radical species.
Terminal and internal alkynes (1-pentyne, 1-hexyne, and 3-
hexyne) are selectively hydrogenated over the Pb-poisoned
and ligand-modified Pd catalysts, whereas the monometallic
Pd nanoparticles generate alkanes and oligomers. This result
can be explained on the basis of the thermodynamic selectivity
concept (a selective catalyst offers preferential adsorption of
the alkyne and a low stability to the adsorbed alkene).[4,21] The
adsorption of ethyne on bare Pd is highly exothermic (1.6 eV),
and its hydrogenation produces the vinyl intermediate,[3]
which is the precursors for the formation of ethylene. However,
once formed, the alkene remains on the catalyst surface, as its
adsorption energy is À1.0 eV, and can be over-hydrogenated
or oligomerized. On the other hand, the desorption of the
alkene from Pb- and HHDMA-‘poisoned’ surfaces is much less
exothermic (0.23 and 0.41 eV, respectively), occurring sponta-
neously under the conditions studied, which justifies the ab-
sence of oligomers over the Lindlar and the colloidally pre-
pared catalysts (Table 2 and Figure 7).
The chemoselective hydrogenation of valylene to isoprene
occurs primarily due to the adsorption of the triple bond on
the surface. Due to the immobility of the phosphate groups
discussed earlier and the rigidity of valylene, the ensemble is
too small to simultaneously adsorb the double bond. Similarly,
in the case of 2-methyl-3-butyn-2-ol, 3-methyl-1-pentyn-3-ol,
and 3-methyl-1-pentyn-4-yn-3-ol (Table 2), the higher selectivity
observed over the ligand-modified than the bare or alloyed Pd
catalysts are related to the interaction of the hydroxyl group
with the Pd surface. The adsorption energy of methanol is
À0.41 eV on Pd and À0.13 eV on Pd–Pb. As the alkynols inves-
tigated also contain unsaturated bonds, these values are fur-
ther reduced by the contribution of CꢀC and C=C bonds.
Therefore, the molecules adsorb strongly on the surface (in
line with previous studies on the hydrogenation of enynes),[45]
further reacting with H species and reducing the process selec-
tivity. This effect is not encountered on the c-Pd catalysts, as
the adsorption energy of an alcohol group with the coated
surface is slightly endothermic (0.06 eV).
Similarly, b-hydride phases formed by the penetration of hy-
drogen into the Pd lattice can be explained by the adsorption
energy of H atoms on surface and subsurface positions. The
average adsorption energy of H on the surface of a clean Pd
Chem. Eur. J. 2014, 20, 5926 – 5937
www.chemeurj.org
5932
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Access and adsorption constraints in hybrid Pd catalysts for
alkyne semi-hydrogenation
Most of the alkynes employed in the pharmaceutical and fine
chemical industries comprise long-chained, highly branched
compounds containing an internal triple bond and additional
functionalities. Stereo- and chemoselective hydrogenation only
occurs if the CꢀC unsaturation can be adsorbed on the palladi-
um site. However, for c-Pd, the presence of the ligand imposes
severe access and/or adsorption constraints, which are con-
trolled by the chemical and physical nature of the HHDMA
layer. To unravel the effects of these limitations, the performan-
ces of the Lindlar and NanoSelectTM catalysts have been evalu-
ated in the hydrogenation of acetylenes of varying adsorption
length. Particularly, the adsorption length has been obtained
from the density isocontour[46] and is defined as the size of the
hydrocarbon when the triple bond is activated on palladium.
In fact, as shown in Figure S9 in the Supporting Information,
this length differs from that of the molecule when the CꢀC un-
saturation is not activated. Figure 10 shows that the c-Pd/TiS
catalyst provides a comparable degree of conversion to that of
the Lindlar catalyst in the hydrogenation of short-chain al-
kynes, and is inactive when the adsorption length of the
alkyne is increased to more than 8 Š. In fact, for larger alkynes,
only the Lindlar catalyst provides an optimal conversion. This
can be rationalized considering that the Lindlar catalyst is
a two-dimensional system in which the Pb atoms create 2D
isolated ensemble sites, while the HHDMA-modified catalyst is
an anisotropic three-dimensional architecture in which the
ligand creates hydrophobic ’pores’. These porous assemblies
Figure 9. Schematic representation of the surface of Pd (a), Pd-Pb (b), and c-
Pd (c). In the Pb-poisoned Pd catalyst (b), the superposition of the Pb exclu-
sion areas is indicated by green hexagons and the available active sites are
enclosed by triangular regions. This represents a two-dimensional catalytic
surface. In the three-dimensional colloidal-based architecture (c), the electric
field generated by the adsorption of HHDMA is represented by green
arrows.
Finally, for the hydrogenation
of 3-butyn-2-one and 3-hexyn-2-
one, the superior performance
of the Lindlar catalyst with re-
spect to c-Pd surfaces can be
traced back to the protic nature
of the c-Pd and the electric
fields presented at the surface
(Figure 9).
The
adsorption
energy of the carbonyl group is
À0.60 and À0.11 eV for Pd–Pb
and c-Pd, respectively. As the
phosphate anion is protic, it can
easily transfer one of the hydro-
gen atoms to the incoming
Figure 10. Influence of the adsorption length of different acetylenic compounds on the hydrogenation activity.
Details on the determination of the adsorption length are given in the text and in Figure S9 in the Supporting In-
formation. Reaction conditions: T=293 K, P=1 bar, FG(H2)=60 cm3 minÀ1, FL(substrate+solvent)=3 cm3 minÀ1
.
ketone,
establishing
further
acid–base equilibria where the phosphate group behaves as
a buffer for hydrogen atoms. This behavior, resembling some-
how that of enzymes and ionic-liquid-stabilized nanoparticles
with strong directing electric fields and proton shuttles,[28,32] is
not encountered on bare and alloyed Pd and explains the re-
duction of the ketone group, which impairs the selectivity to
the alkenones.
evolve over time (Figure S5 in the Supporting Information), al-
lowing the acetylenic molecules to penetrate the surfactant
layer (Figure 11), provided they are not too bulky. In fact, the
insertion of bulky alkynes implies a penalty in terms of config-
urational entropy, which is compensated by the van der Waals
interactions between the alkyne and the HHDMA tail. Once the
acetylenic species has navigated into the surfactant layer, it
needs to reorientate to be able to access the active site with
the correct configuration. In this case, the adsorption can be
Chem. Eur. J. 2014, 20, 5926 – 5937
www.chemeurj.org
5933
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Conclusion
We have correlated the alkyne
hydrogenation performance of
a representative set of Pd-based
catalysts with structural descrip-
tors derived from characteriza-
tion methods, DFT calculations,
and MD simulations. Supported
ligand-modified palladium, de-
veloped as a lead-free alternative
Figure 11. Proposed mechanism for the insertion and adsorption of a long-chained alkyne (represented in green)
on the Pd surface of the NanoSelectTM catalyst. The picture highlights the opening of the HHDMA chains, which
allows the adsorption to take place. The color codes are indicated in the captions of Figure 1 and Figure 5.
limited by the rigid distribution of the phosphate groups on
the surface. Indeed, the dynamic flexibility of the ligands gen-
erates rigid elliptical cavities in close proximity to these anions,
which have a maximum length of about 8 Š. Phenylacetylene
does not suffer from a similar restriction not only for its small
size but also because it interacts strongly with the Pd surface,
opening a large ensemble. On the other hand, for 7-decyn-1-
ol, 1,1-diphenyl-2-propyn-1-ol, and 9-hexadecyn-1-ol, the ad-
sorption length is far too large to ensure that these species
can interact with the palladium and react. This justifies the ab-
sence of conversion for the NanoSelectTM catalyst. These results
are in agreement with the data in Figure 10. The fact that ace-
tylenic species with adsorption lengths greater than 8 Š are
not hydrogenated by the NanoSelectTM catalyst suggests that
Figure 12. Important acetylenic intermediates in the pharmaceutical and
fine chemical industries whose semihydrogenation is currently catalyzed by
the Lindlar systems.
the ensemble size of Pd is constituted by approximately two
adjacent palladium atoms.
In order to corroborate whether the activity of the NanoSe-
lectTM catalyst can be enhanced with a controlled removal of
the HHDMA layer, the c-Pd/TiS catalyst has been subjected to
UV–ozone treatments and subsequently tested in the hydroge-
nation of 1-hexyne, 9-hexadecyn-1-ol, and 1-1,-diphenyl-2-
propyn-1-ol (Figure S10 in the Supporting Information). The re-
sults show that the optimal duration of UV–ozone treatment is
substrate specific, requiring careful fine tuning. In general, the
complete removal of adsorbed HHDMA molecules is associated
with a drop of the selectivity to the alkene.
to the archetypal Lindlar catalyst, overcomes the suboptimal
selectivity of Pd-only nanoparticles through the presence of an
HHDMA capping layer. Similar to the role of lead in the Lindlar
catalyst, the adsorbed ligand serves to isolate and tailor the ac-
cessibility of the active site and tune the energy landscape of
the catalyst. For these reasons, in the hydrogenation of acety-
lenic compounds featuring isolated unsaturations and hydroxyl
groups, these materials exceed the performance of the Pd–Pb
alloy with a ten times lower palladium loading. Nevertheless,
specific substrates, such as long-chain alkynols and alkynones,
are not compatible with the use of the HHDMA-modified Pd
catalysts, due to steric hindrance and acid–base interactions
on the surface. This adds an extra degree of complexity in the
performance of Pd nanoparticles prepared by colloidal routes.
These findings provide a wide perspective on the applicability
and limits of hybrid catalysts for hydrogenation, offering key
insights for the design of future alkyne hydrogenation catalysts
which are ultraselective to olefin production and do not suffer
from accessibility constraints.
Despite these findings, which limit the application of the
NanoSelectTM catalyst in a broad range of industrial application
(Figure 12), it is noteworthy that the catalytic systems stan-
dardly applied in industrial gas-phase catalyzed hydrogena-
tions contain less than 0.05 wt.% Pd. On the other hand, the
Lindlar catalyst applied in liquid-phase fine chemical manufac-
turing contains 5 wt.% Pd. In this case, the addition of Pb, in
the catalyst formulation, and of quinoline, in the reaction mix-
ture, block a large fraction of the active sites, leaving only one
tenth of the Pd atoms available for hydrogenation.[15] The
novel materials prepared by colloidal routes are formulated
with ca. 0.5 wt.% Pd. The electrostatic interactions between
the stabilizer HHDMA and the Pd nanoparticles blocks three
out of four surface atoms (Figure 9), leaving single active cen-
ters that are completely isolated from the neighboring active
sites. Therefore, in addition to providing a Pb-free alternative
to the Lindlar catalysts, these hybrid materials competitively
improve the metal utilization in liquid-phase alkyne hydroge-
nations by site modification and isolation.
Experimental and Computational Section
Materials
Four supported palladium catalysts were evaluated in this study:
Pd/Al2O3 (Sigma–Aldrich, ref: 20570-2), Pd–Pb/CaCO3 (Alfa Aesar,
ref: 043172), c-Pd/C (NanoSelect LF 100TM, Strem Chemicals, ref:
Chem. Eur. J. 2014, 20, 5926 – 5937
www.chemeurj.org
5934
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
46-1710) and c-Pd/TiS (NanoSelect LF 200TM, Strem Chemicals, ref:
46-1711). The prefix ‘c-’ designates supported Pd catalysts which
have been prepared by colloidal routes, and TiS denotes the titani-
um silicate support (compositional analysis has shown that the
latter is characterized by a molar Ti/Si ratio of 2.3). The residual or-
ganic layer present on the surface of the Pd nanoparticles in the
NanoSelect catalysts was removed by applying an UV–ozone treat-
ment using a Bulbtronics 16 W low-pressure mercury lamp (emit-
ting at 185 and 257 nm). The sample was positioned at a distance
of 5 mm from the UV lamp for 1, 2, 5, 10, 20, 40, 60, or 120 min.
The acetylenic reagents 1-pentyne (ABCR-Chemicals, 98%), 1-
hexyne (Acros Organics, 98%), 3-hexyne (TCI Deutschland, 98%), 2-
methyl-1-buten-3-yne (ABCR-Chemicals, 97%), 2-methyl-3-butyn-2-
ol (Acros Organics, 98%), 3-methyl-1-pentyn-3-ol (TCI Deutschland,
>98%), 3-methyl-1-penten-4-yn-3-ol (ABCR-Chemicals, 95%), 3-
butyn-2-one (ABCR-Chemicals, 97%), 3-hexyn-2-one (ABCR-Chemi-
cals, 97%), 1-octyne (ABCR-Chemicals, 98%), 4-octyne (ABCR-
Chemicals, 99%), 9-hexadecyn-1-ol (ABCR-Chemicals, 98%), phenyl-
acetylene (Acros Organics, 98%), 1,1-diphenyl-2-propyn-1-ol
(Sigma–Aldrich, 99%), toluene (Acros Organics, 99.9%), and ben-
zene (Sigma–Aldrich,>99.5%) were used without further
purification.
trometer equipped with high-temperature cell, ZnSe windows, and
a mercury-cadmium-telluride detector. The cell was filled with pow-
dered catalyst and carefully leveled to minimize reflection from the
sample surface. The spectra were recorded at 473 K in the range of
650–4000 cmÀ1, by co-addition of 200 scans with a nominal resolu-
tion of 4 cmÀ1. Thermogravimetric analysis (TGA) was performed in
a Mettler Toledo TGA/DSC 1 Star system connected to a Pfeiffer
Vacuum ThermoStar GSD 320 T1 Gas Analysis mass spectrometer.
The analysis was performed under an air flow (40 cm3 minÀ1), ramp-
ing the temperature from 298 to 1173 K at 5 KminÀ1. The signals of
water (m/z=18) and carbon dioxide (m/z=44) were continuously
monitored.
Catalytic tests
The hydrogenation of acetylenic compounds was carried out in
a fully-automatized continuous-flow flooded-bed reactor (Thales-
Nano H-Cube ProTM), in which the liquid hydrocarbon and the gas-
eous hydrogen flow concurrently upward through a fixed bed of
catalyst particles (Figure 2). Compared to the typical downward
flow of a trickle-bed reactor, a flooded-bed reactor is highly advan-
tageous for studying three-phase hydrogenation reactions since it
enables improved performance by prolonging the catalyst lifetime
and enhancing the heat transfer between the gas and the liquid
phases.[48] The pressure drop, calculated according to the Lock-
hart–Martinelli correlation for a two-phase flow,[48] was insignificant
(0.07 bar) along the bed and estimation of the axial dispersion cor-
roborated minor deviations from the plug-flow regime. The con-
tacting efficiency was within the trickle and flooded flow regimes,
ensuring the absence of liquid holdup.[48] Hydrogen was generated
in situ by the electrolysis of Millipore-filtered water and supplied to
the reaction chamber by a mass-flow controller. The liquid feed
was supplied by an HPLC pump. The reactor was equipped with
a two-zone heating jacket, a pressure controller, and three thermo-
couples, located at the inlet, in the center, and at the outlet of the
catalyst bed. In particular, during reaction, the temperature differ-
ence between these thermocouples was less than 2 K, confirming
isothermal operation. The catalyst (0.1 g, particle size: 0.2–0.4 mm)
was diluted and carefully mixed with silicon carbide (0.07 g, parti-
cle size: 0.2–0.4 mm) and loaded into a cartridge of approximately
30 mm length”3.5 mm internal diameter. The influence of silicon
carbide on the reactant conversion was excluded by estimation of
the dilution correlation.[49] Reaction solutions contained 1 vol.% of
the acetylenic substrate, toluene as the solvent, and benzene as
the internal standard. The catalytic tests were performed at various
temperatures (293–413 K), pressures (1–9 bar), and liquid (0.3–
3 cm3 minÀ1) and H2 flow rates (3–60 cm3 minÀ1). Importantly, under
these conditions, a laminar flow is expected in both the liquid and
the gas phases. The possibility of alkyne evaporation was ruled out
by circulating the alkyne with the toluene solvent and the benzene
internal standard for 3 h, in the absence of hydrogen, and the con-
centration of reactant at the reactor outlet remained constant. In
addition, leaching of active Pd species from the catalyst was ex-
cluded. In fact, the liquid at the reactor outlet, containing the
alkyne, alkene, toluene, and benzene, was fed to the reactor inlet
Characterization
Nitrogen isotherms at 77 K were measured in a Micromeritics Tri-
Star II instrument. Prior to measurement, the sample was degassed
in vacuum at 393 K for 10 h. The palladium and lead contents in
the solids were determined by inductively coupled plasma optical
emission spectroscopy (ICP-OES) using a Horiba Ultra 2 instrument
equipped with photomultiplier tube detection. The palladium dis-
persion (D) was determined by CO pulse chemisorption by using
a Thermo TPDRO 1100 unit. The samples (0.05–0.1 g) were pre-
treated at 393 K under flowing He (20 cm3 minÀ1) for 60 min, and
reduced at 348 K under flowing 5 vol.% H2/He (20 cm3 minÀ1) for
30 min. Thereafter, 0.344 cm3 of 1 vol.% CO/He was pulsed over
the catalyst bed at 308 K every 4 min. The interval between succes-
sive pulses was minimized to avoid desorption of CO. The palladi-
um dispersion was calculated from the amount of chemisorbed
CO, considering an atomic surface density of 1.26”1019 atoms mÀ2
and an adsorption stoichiometry of Pd/CO=2.[47] X-ray photoelec-
tron spectroscopy (XPS) was performed on a VG-Microtech Multilab
3000 spectrometer featuring a hemispheric electron analyzer with
9
channeltrons and non-monochromatized AlKa radiation at
1486.6 eV. The catalysts were pretreated at 393 K under flowing He
(20 cm3 minÀ1) for 60 min, and reduced in situ at 348 K under flow-
ing 5 vol.% H2/He (20 cm3 minÀ1) for 30 min. The spectra were col-
lected under ultra-high vacuum conditions (residual pressure of ca.
5”10À8 Pa), at a pass energy of 50 eV. All binding energies were
referenced to the C 1s level at 284.6 eV in order to compensate for
charging effects. The distribution and morphology of the support-
ed palladium nanoparticles were studied by secondary electron
scanning electron microscopy (SEM) and high-angle annular dark-
field scanning transmission electron (HAADF-STEM) imaging, which
was conducted in an FEI Magellan XHR 400 microscope using an
accelerating voltage of 20 kV. The catalysts were dispersed as dry
powders on holey carbon-coated copper grids. The particle size
distribution was assessed by analysis of 350 individual Pd particles
and the dispersion of palladium was estimated by considering par-
ticles to be truncated octahedrons with cubic symmetry and as-
suming the absence of an oxide shell. The N content in selected
catalysts was determined by infrared spectroscopy using a LECO
CHN-900 combustion furnace. Fourier transform infrared (FTIR)
spectroscopy was performed in a Bruker Optics Vertex 70 spec-
(at T=293 K, P=1 bar, FG(H2)=18 cm3 minÀ1
, FL(substrate+sol-
vent)=0.3 cm3 minÀ1), in a cartridge filled only with diluent, and no
additional hydrogenation occurred, thus evidencing no loss of the
active Pd phase, in agreement with elemental analyses of the used
catalysts. The reaction products were collected employing an auto-
sampler, after reaching (in ca. 10 min) steady-state operation, and
analyzed offline using a gas chromatograph (HP 6890) equipped
with a HP-5 capillary column and a flame ionization detector.
Helium was used as a carrier gas, flowing at 1 cm3 minÀ1. The
Chem. Eur. J. 2014, 20, 5926 – 5937
www.chemeurj.org
5935
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
column was kept at an initial temperature of 303 K for 12 min and
then the temperature was ramped from 303 to 473 K at 5 KminÀ1
273 K, the Nosꢀ-Hoover thermostat was employed,[66,67] and the
Verlet leapfrog algorithm[68] was used to integrate the equations of
motion with a time step of 0.1 fs. The length of the subsequent
production run for the system was 1 ns.
.
The conversion (X) of alkyne was determined as the amount of re-
acted alkyne divided by the amount of alkyne at the reactor inlet,
whereas the selectivity (S) to a given alkene or alkane was quanti-
fied as the amount of that compound divided by the total amount
of products.
Acknowledgements
Computational details
Financial support from ETH Zurich, the Spanish “Ministerio de
Economía y Competitividad” (CTQ2012-33826), and the ICIQ
Foundation is acknowledged. Dr. Yargo Bonetti and Luisa Petti
(ETH Zurich) are thanked for performing UV–ozone treatments.
The FIRST Laboratory and the Scientific Center for Optical and
Electron Microscopy ScopeM of the Swiss Federal Institute of
Technology are acknowledged for providing access to their fa-
cilities. BSC-RES is thanked for generous computational
resources.
Quantum-mechanical calculations based on density functional
theory with periodic boundary conditions were carried out with
the VASP code,[50–52] using a basis set of plane waves to solve the
Kohn–Sham equations, the generalized gradient approximation
(GGA), and the revised Perdew–Burke–Ernzerhof (RPBE) function-
al.[53] The interaction between valence and core electrons was de-
scribed by the projected augmented wave (PAW) method.[52] The
number of plane waves was controlled by setting the cutoff
energy to 450 eV. For the geometric optimization, a 5”5”1 k-
point mesh was used in order to sample the reciprocal space. DFT
calculations focused on the Pd(111) facet, characterized by a slab
with five atomic layers, which was periodically repeated in the di-
rection perpendicular to the surface and separated by a vacuum
gap of approximately 30 Š. Particularly, Pd(111) is known to be the
most exposed and active surface of palladium.[6] Parallel to the sur-
face, the supercell consisted of a primitive, p(2”2) arrangement. In
the optimizations, the two atomic layers at the bottom of the slab
were fixed at their bulk positions, while the upper three layers
were fully relaxed. To model the Lindlar catalyst, single Pd atoms
were substituted by Pb with a 0.25 monolayer coverage, consider-
ing that earlier experimental studies on the interaction of Pb with
Pd in the Lindlar catalyst suggested the formation of a surface
Pd3Pb alloy.[15,54] Pd–Zn[10] and Pd–Ga[25] alloys, which are also selec-
tive catalysts for alkyne hydrogenation, were represented similarly
to Pd–Pb. The crystal structure of the HHDMA ligand had been de-
rived from the structure of hexadecyltrimethylammomium bro-
mide.[55] This was modified using the Materials Studio software
(version 6.1, Accelrys), in order to contain two units of the cationic
Keywords: alkynes
·
density functional theory
·
flow
chemistry · hydrogenation · nanoparticles · palladium
[1] S. Nishimura, Handbook of Heterogeneous Catalytic Hydrogenation for
Organic Synthesis, Wiley, New York, 2001, p. 148.
[2] H. F. Rase, Handbook of Commercial Catalysts: Heterogeneous Catalysts,
CRC Press, Boca Raton, 2000, p. 110.
[3] . Molnµr, A. Sµrkµny, M. Varga, J. Mol. Catal. A 2001, 173, 185.
[4] D. Teschner, J. Borsodi, A. Wootsch, Z. Rꢀvay, M. Hävecker, A. Knop-Ger-
icke, S. D. Jackson, R. Schlçgl, Science 2008, 320, 86.
[5] J. A. Anderson, J. Mellor, R. P. K. Wells, J. Catal. 2009, 261, 208.
[6] M. García-Mota, B. Bridier, J. Pꢀrez-Ramírez, N. López, J. Catal. 2010, 273,
92.
[7] M. Crespo-Quesada, A. Yarulin, M. Jin, Y. Xia, L. Kiwi-Minsker, J. Am.
Chem. Soc. 2011, 133, 12787.
[8] D. Mei, P. A. Sheth, M. Neurock, C. M. Smith, J. Catal. 2006, 242, 1.
[9] F. Studt, F. Abild-Pedersen, T. Bligaard, R. Z. Sørensen, C. H. Christensen,
J. K. Nørskov, Science 2008, 320, 1320.
[10] F. Studt, F. Abild-Pedersen, T. Bligaard, R. Z. Sørensen, C. H. Christensen,
J. K. Nørskov, Angew. Chem. 2008, 120, 9439; Angew. Chem. Int. Ed.
2008, 47, 9299.
[11] D. Teschner, Z. Rꢀvay, J. Borsodi, M. Hävecker, A. Knop-Gericke, R.
Schlçgl, D. Milroy, S. D. Jackson, D. Torres, P. Sautet, Angew. Chem. 2008,
120, 9414; Angew. Chem. Int. Ed. 2008, 47, 9274.
[12] D. Mei, M. Neurock, C. M. Smith, J. Catal. 2009, 268, 181.
[13] N. López, C. Vargas-Fuentes, Chem. Commun. 2012, 48, 1379.
[14] H. Lindlar, Helv. Chim. Acta 1952, 35, 446.
[15] M. García-Mota, J. Gómez-Díaz, G. Novell-Leruth, C. Vargas-Fuentes, L.
Bellarosa, B. Bridier, J. Pꢀrez-Ramírez, N. López, Theor. Chem. Acc. 2011,
128, 663.
[16] A. V. Biradar, A. A. Biradar, T. Asefa, Langmuir 2011, 27, 14408.
[17] P. Conley, R. M. Drost, M. Baffert, D. Gajan, C. Elsevier, T. W. Franks, H. Os-
chkinat, L. Veyre, A. Zagdoun, A. Rossini, M. Lelli, A. Lesage, G. Casano,
O. Ouari, P. Tordo, L. Emsley, C. Copꢀret, C. Thieuleux, Chem. Eur. J.
2013, 19, 12234.
[18] S. Domínguez-Domínguez, . Berenguer-Murcia, B. K. Pradhan, . Li-
nares-Solano, D. Cazorla-Amorós, J. Phys. Chem. C 2008, 112, 3827.
[19] W. Long, N. A. Brunelli, S. A. Didas, E. W. Ping, C. W. Jones, ACS Catal.
2013, 3, 1700.
À
alkyl chains in alternate configuration and two H2PO4 anions (Fig-
ure S4 in the Supporting Information), following the methodology
described elsewhere.[56] For the adsorption of individual HHDMA
molecules on palladium, van der Waals interactions were intro-
duced with the DFT-D2 approach,[57,58] employing the parameters
refined by Ruiz et al.[59] In order to investigate the flexibility of the
ligand adsorbed on Pd(111), classical molecular dynamics simula-
tions were carried out with the DL_POLY Classic code.[60] This
method has been successfully employed in the past for long-time-
scale simulations, which offer significantly more information on the
process itself, and for large size simulations, permitting the applica-
tion to macromolecular chains. However, the code cannot describe
the reactivity on the surface and, thus, it is only suited to under-
stand the ensemble size and the limitations imposed by the surfac-
tant layer to the access of the reactants. In our case, the HHDMA–
Pd system was modeled in a hexagonal simulation cell with surface
dimensions of 27.5”27.5 Š2, which is three times larger than the
DFT simulation cell. The Pd surÀface and the HHDMA adsorbates
were relaxed, keeping the H2PO4 ions fixed at the optimized posi-
tions provided by the DFT calculations. A combination of compati-
ble interatomic potential models was employed for the description
of the interactions between the atoms in the systems. Particularly,
the generalized AMBER force field[61,62] for the HHDMA molecule
and METAL[63] for palladium. This combination has been previously
validated by Heinz and co-workers.[64,65] The equilibration period
was 0.1 ns in the NVT ensemble. In order to fix the temperature at
[20] Y. Yabe, Y. Sawama, Y. Monguchi, H. Sajiki, Catal. Sci. Technol. 2014, 4,
260.
[21] Y. Segura, N. López, J. Pꢀrez-Ramírez, J. Catal. 2007, 247, 383.
[22] B. Bridier, N. López, J. Pꢀrez-Ramírez, Dalton Trans. 2010, 39, 8412.
[23] G. Vilꢀ, B. Bridier, J. Wichert, J. Pꢀrez-Ramírez, Angew. Chem. 2012, 124,
8748; Angew. Chem. Int. Ed. 2012, 51, 8620.
[24] G. Vilꢀ, D. Baudouin, I. N. Remediakis, C. Copꢀret, N. López, J. Pꢀrez-
Ramírez, ChemCatChem 2013, 5, 3750.
Chem. Eur. J. 2014, 20, 5926 – 5937
www.chemeurj.org
5936
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[25] G. Wowsnick, D. Teschner, M. Armbrüster, I. Kasatkin, F. Girgsdies, Y.
Grin, R. Schlçgl, M. Behrens, J. Catal. 2014, 309, 221.
[26] B. Bridier, J. Pꢀrez-Ramírez, J. Am. Chem. Soc. 2010, 132, 4321.
[27] J. P. Boitiaux, J. Cosyns, S. Vasudevan, Appl. Catal. 1983, 6, 41.
[28] A. Roucoux, J. Schulz, H. Patin, Chem. Rev. 2002, 102, 3757.
[29] C. Burda, X. Chen, R. Narayanan, M. A. El-Sayed, Chem. Rev. 2005, 105,
1025.
[30] N. Semagina, L. Kiwi-Minsker, Catal. Rev. Sci. Eng. 2009, 51, 147.
[31] N. Semagina, L. Kiwi-Minsker, Catal. Lett. 2009, 127, 334.
[32] R. Venkatesan, M. H. G. Prechtl, J. D. Scholten, R. P. Pezzi, G. Machado, J.
Dupont, J. Mater. Chem. 2011, 21, 3030.
[33] H. Bçnnemann, W. Brijoux, R. Brinkmann, E. Dinjus, T. Joussen, B. Korall,
Angew. Chem. 1991, 103, 1344; Angew. Chem. Int. Ed. Engl. 1991, 30,
1312.
[34] M. Crespo-Quesada, J. M. Andanson, A. Yarulin, B. Lim, Y. Xia, L. Kiwi-
Minsker, Langmuir 2011, 27, 7909.
[48] C. N. Satterfield, AIChE J. 1975, 21, 209.
[49] R. J. Berger, J. Pꢀrez-Ramírez, F. Kapteijn, J. A. Moulijn, Chem. Eng. J.
2002, 90, 173.
[50] G. Kresse, J. Furthmuller, Phys. Rev. B 1996, 54, 11169.
[51] G. Kresse, J. Furthmuller, Comput. Mater. Sci. 1996, 6, 15.
[52] G. Kresse, D. Joubert, Phys. Rev. B 1999, 59, 1758.
[53] B. Hammer, L. B. Hansen, J. K. Nørskov, Phys. Rev. B 1999, 59, 7413.
[54] D. Chadwick, M. A. Karolewski, Surf. Sci. 1983, 126, 41.
[55] Z. B. Wei, X. L. Wei, X. H. Wang, Z. N. Wang, J. Liu, J. Mater. Chem. 2011,
21, 6875.
[56] N. Almora-Barrios, G. Novell-Leruth, P. Whiting, L. M. Liz-Marzµn, N.
López, Nano Lett. 2014, 14, 871.
[57] S. Grimme, J. Comput. Chem. 2006, 27, 1787.
[58] T. Bucko, J. Hafner, S. Lebegue, J. G. Angyan, J. Phys. Chem. A 2010, 114,
11814.
[59] V. G. Ruiz, W. Liu, E. Zojer, M. Scheffler, A. Tkatchenko, Phys. Rev. Lett.
2012, 108, 146103.
[60] I. T. Todorov, W. Smith, K. Trachenko, M. T. Dove, J. Mater. Chem. 2006,
16, 1911.
[35] P. T. Witte, P. H. Berben, S. Boland, E. H. Boymans, D. Vogt, J. W. Geus,
J. G. Donkervoort, Top. Catal. 2012, 55, 505.
[36] P. T. Witte, S. Boland, F. Kirby, R. van Maanen, B. F. Bleeker, D. A. M. de
Winter, J. A. Post, J. W. Geus, P. H. Berben, ChemCatChem 2013, 5, 582.
[37] Y. Zhao, L. Jia, J. A. Medrano, J. R. H. Ross, L. Lefferts, ACS Catal. 2013, 3,
2341.
[61] W. D. Cornell, P. Cieplak, C. I. Bayly, I. R. Gould, K. M. Merz, D. M. Fergu-
son, D. C. Spellmeyer, T. Fox, J. W. Caldwell, P. A. Kollman, J. Am. Chem.
Soc. 1995, 117, 5179.
[38] M. Irfan, T. N. Glasnov, C. O. Kappe, ChemSusChem 2011, 4, 300.
[39] M. A. G. Hevia, B. Bridier, J. Pꢀrez-Ramírez, Appl. Catal. A 2012, 439–440,
163.
[62] J. M. Wang, P. Cieplak, P. A. Kollman, J. Comput. Chem. 2000, 21, 1049.
[63] H. Heinz, R. A. Vaia, B. L. Farmer, R. R. Naik, J. Phys. Chem. C 2008, 112,
17281.
[40] R. Schlçgl, K. Noack, H. Zbinden, A. Reller, Helv. Chim. Acta 1987, 70, 3.
[41] D. Briggs, M. P. Seah, Practical Surface Analysis by Auger and X-ray Photo-
electron Spectroscopy, 1st ed., Wiley, Chichester, 1983, p. 26.
[42] B. H. Stuart, Infrared Spectroscopy - Fundamentals and Applications,
Wiley, Chichester, 2004, p. 45.
[64] J. Feng, J. M. Slocik, M. Sarikaya, R. R. Naik, B. L. Farmer, H. Heinz, Small
2012, 8, 1049.
[65] J. Feng, R. B. Pandey, R. J. Berry, J. M. Slocik, R. R. Naik, B. L. Farmer, H.
Heinz, Soft Matter 2011, 7, 2113.
[66] S. Nosꢀ, J. Chem. Phys. 1984, 81, 511.
[43] X. Yang, N. Yan, Z. Fei, M. Crespo-Quesada, G. Laurenczy, L. Kiwi-Minsker,
Y. Kou, Y. Li, P. Dyson, Inorg. Chem. 2008, 47, 7444.
[67] W. G. Hoover, Phys. Rev. A 1985, 31, 1695.
[68] L. Verlet, Phys. Rev. 1967, 159, 98.
[44] O. M. Magnussen, Chem. Rev. 2002, 102, 679.
[45] B. Bridier, D. Karhµnek, J. Pꢀrez-Ramírez, N. López, ChemCatChem 2012,
4, 1420.
[46] N. Mehio, S. Dai, D. Jiang, J. Phys. Chem. A 2014, 118, 1150.
[47] G. Fagherazzi, P. Canton, P. Riello, N. Pernicone, F. Pinna, M. Battagliarin,
Langmuir 2000, 16, 4539.
Received: December 6, 2013
Published online on April 17, 2014
Chem. Eur. J. 2014, 20, 5926 – 5937
www.chemeurj.org
5937
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim