A balancing act: Manipulating reactivity of shape-controlled metal nanocatalysts through bimetallic architecture

Article abstract of DOI:10.1039/c5ta09368f

Manipulating the electronic structure of metal nanocrystals is one way of altering their catalytic activities. This ability is demonstrated here by introducing a Au interior to shape-controlled Pd nanocrystals, producing core@shell Au@Pd nanoparticles with varying shell thicknesses. As revealed by X-ray photoelectron spectroscopy, the electronic structure of the Pd shell depends on its thickness. These core@shell nanocrystals were used to catalyze two model reactions: selective hydrogenation of 2-hexyne and oxidation of formic acid, where different reactivities were found also as a function of shell thickness. The comparison of particles with varying bimetallic architecture but identical geometric features provides insight into how electronic regulation in a catalytic reaction can be achieved. It is concluded that a balance in binding interaction between the molecular substrate and catalyst surface is necessary to design an efficient catalyst and can be achieved with shape-controlled core@shell nanocrystals.

Full text of DOI:10.1039/c5ta09368f

Journal of
Materials Chemistry A
View Article Online
View Journal
PAPER
A balancing act: manipulating reactivity of shape-
controlled metal nanocatalysts through bimetallic
architecture†
Cite this: DOI: 10.1039/c5ta09368f
M. Laskar and S. E. Skrabalak*
Manipulating the electronic structure of metal nanocrystals is one way of altering their catalytic activities.
This ability is demonstrated here by introducing a Au interior to shape-controlled Pd nanocrystals,
producing core@shell Au@Pd nanoparticles with varying shell thicknesses. As revealed by X-ray
photoelectron spectroscopy, the electronic structure of the Pd shell depends on its thickness. These
core@shell nanocrystals were used to catalyze two model reactions: selective hydrogenation of 2-
hexyne and oxidation of formic acid, where different reactivities were found also as a function of shell
thickness. The comparison of particles with varying bimetallic architecture but identical geometric
features provides insight into how electronic regulation in a catalytic reaction can be achieved. It is
concluded that a balance in binding interaction between the molecular substrate and catalyst surface is
necessary to design an efficient catalyst and can be achieved with shape-controlled core@shell
nanocrystals.
Received 18th November 2015
Accepted 21st January 2016
DOI: 10.1039/c5ta09368f
www.rsc.org/MaterialsA
when the d-band center is far below the Fermi level and is
Introduction
stronger as the d-band center shis towards it. The strength of
1
15
Noble metal nanocrystals are widely used as catalysts despite surface–adsorbate interactions regulates the rate of a reaction.
the scarcity and high expense of such metals. To make more With these principles in mind, we examined the effect of the
efficient use of these metals, the geometric and electronic core@shell architecture on the activity and selectivity of shape-
parameters that govern their performance can be manipulated controlled metal nanocatalysts. This study was undertaken with
through nanostructuring and introduction of secondary Au@Pd octahedra as a model system, where the thickness of the
2
–6
metals.
catalysts can be controlled to dene the relative distribution of with similarly sized Pd octahedra revealed a tunable electronic
For example, the size and shape of metal nano- Pd shell was tuned between roughly 1 and 9 nm. Comparison
7,8
different planes, edges, and corners expressed. As different effect that depends on shell thickness, which we attribute
6,16
reactants favor different sites for adsorption and subsequent primarily to a strained shell. We note that by studying simi-
reaction due to varying symmetry match, this geometric control larly sized octahedral nanocrystals, the inuence of different
can enhance rates of reactions through selection of appropriate faceting on reactivity is minimized. This experimental design
9,10
size- and shape-controlled nanocrystals.
The electronics of allows the role of bimetallic architecture on reactivity to be
a surface can also be tuned by introducing a second metal either explicitly elucidated in contrast to the study of quasi-spherical
through a core@shell architecture or formation of an alloy; in core@shell nanoparticles where faceting is poorly dened from
either case, the strength of surface–adsorbate interactions can one sample to the next or of alloyed nanoparticles where
1
1,12
be tuned by this change in electronic structure.
This inter- composition and active site are intertwined.
The possibility for strain-engineered nanocrystals is of
action depends on orbital lling, the energy and width of the
surface d-band, the energy levels of the adsorbate, and the current interest, with a number of signicant advances recently
geometric orbital matching between the adsorbate and reported. For example, Xia and Mavrikakis found Pd@Pt
13
surface. A linear relationship between the d-band center and nanocubes with ultrathin shells to be better and more stable
the binding energy of a molecular substrate (i.e., the molecule catalysts towards the oxygen reduction reaction (ORR)
17
undergoing reaction) to a surface exists according to the compared to Pt/C and Pt nanocubes. A combination of
14
Hammer–Nørskov model. Thus, adsorbate binding is weak compressive lateral strain and the ligand effect (i.e., charge
transfer from a metal with a lower work function to one with
a higher work function when in close contact) lowered the Pt d-
Department of Chemistry, Indiana University-Bloomington, 800 E. Kirkwood Avenue,
band center, weakening OH binding to the Pt(100) surface. This
Bloomington, IN 47405, USA. E-mail: sskrabal@indiana.edu
change in electronic structure facilitated an enhanced rate of
†
Electronic supplementary information (ESI) available: SEM and TEM images of
OH hydrogenation. A volcano-type relationship between the
nanocrystals aer catalysis. See DOI: 10.1039/c5ta09368f
This journal is © The Royal Society of Chemistry 2016
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
Pt shell thickness and ORR specic activity was obtained, with Octahedral Pd seeds (ꢀ23 nm)
a maximum activity for the Pd@Pt nanocubes at 2–3 Pt atomic
The synthesis of Pd octahedral seeds was achieved by a reported
method. 2 mL of a 156.1 mM aqueous citric acid solution and
layers. The same group also examined Pd@Pt icosahedra, which
12
exhibited 7–8 times higher specic catalytic activity towards
2
mL of a 170.3 mM of aqueous AA solution was added to 4 mL
18
ORR compared to a commercial Pt/C catalyst. This enhance-
ment is consistent with Pt(111) being more active than Pt(100)
towards ORR and the predominance of twin places in the
icosahedral structure. Tsung and coworkers also reported
Au@Pd nanocubes and octahedra as suitable catalysts for for-
mic acid oxidation (FMO), citing shiing of the d-band center
ꢁ
of a 200 mM aqueous CTAC solution and stirred at 100 C. 3 mL
2 4
of a 67.9 mM aqueous K PdCl solution was added to the above
ꢁ
mixture and kept at 100 C for 3 h. This solution was diluted 100
times for further use as seeds.
Pd octahedra (37 nm)
16
for Pd towards the Fermi level for enhanced activity. However,
their study did not analyze the relationships between shell 2 mL of a 156.1 mM aqueous citric acid solution was added to 5
thickness, strain, and catalytic performance as we provide here. mL of a 100 mM aqueous CTAC solution. 1 mL of the 100 times
Specically, the electronic structure of shape-controlled Pd diluted Pd octahedra (edge length 23 nm) solution was added to
nanocrystals is manipulated through the introduction of Au the above mixture, followed by the addition of 1 mL of an
cores, providing a platform to examine the effect of bimetallic aqueous 5 mM K
architecture on two model reactions: 2-hexyne hydrogenation thoroughly mixed by inversion and kept at 70 C for 20 h to
PdCl
solution. The reaction mixture was
2
4
ꢁ
and formic acid oxidation. With the lattice constant of Au (a ¼ synthesize 37 nm Pd nanooctahedra.
4
08.53 pm) being larger than Pd (a ¼ 389.07 pm), an expansion
of the Pd lattice constant is anticipated in a core@shell Au@Pd Synthesis of size-controlled Au@Pd octahedra
nanostructure. This tensile strain will decrease orbital overlap,
To prepare Au@Pd octahedra, Au octahedra were prepared,
which were used as seeds for the epitaxial growth of Pd on top of
the Au octahedra.
narrow the d-band, and shi its center closer to the Fermi level.
At the same time, the higher Fermi level of Au relative to Pd will
facilitate charge transfer to Pd shell, potentially acting against
19,20
the strain contribution.
Taken together, for nanostructures
Octahedral Au seeds (ꢀ33 nm)
with thin Pd shells, the Au core is expected to provide stronger
strain and ligand effects, with the ligand effect diminishing over Au octahedra were prepared according to a previously reported
2
1
24
a few atomic layers. As the Pd shells increase in thickness, the method. 9.7 mL of a 15.5 mM aqueous CTAB solution, 0.25 mL
22
4 2
strain effect should persist until released by defects. Thus, at of a 0.01 M aqueous HAuCl $3H O solution, and 0.05 mL of
very thick Pd shells, the catalytic properties of the bimetallic a 0.1 M aqueous sodium citrate dihydrate solution were added
ꢁ
structure should be bulk-like, as has been observed in studies of to a vial and kept at 110 C for 24 h. The reaction was stopped by
23
single crystal surfaces with overlayers.
cooling at room temperature. For collection of Au octahedra,
the mixture was centrifuged at 3900 rpm for 1 h, the superna-
tant was removed, and the nanocrystals were redispersed in 4
mL water and subsequently used as seeds for the synthesis of
Au@Pd octahedra.
Experimental
Chemicals
All the chemicals were used as purchased without further
purication. Chloroauric acid (HAuCl $3H O, 99.9%), potas-
sium tetrachloropalladate (K PdCl ), L-ascorbic acid or AA
C H O , 99%), sodium citrate dihydrate (Na C H O $2H O),
Au@Pd octahedra (ꢀ37 nm, 40 nm, and 42 nm)
4
2
2
4
2
mL of a 156.1 mM aqueous citric acid solution was added to 5
(
ꢁ
6
8
6
3
6
5
7
2
mL of a 100 mM aqueous CTAC solution kept at 90 C. 0.5 mL of
the Au octahedral seed solution was added to the above mixture
hexadecyltrimethylammonium bromide or CTAB (C H NBr,
1
9
42
9
8%), hexadecyltrimethylammonium chloride or CTAC
42NCl, 25 wt% solution in water), titanium(IV) oxide (TiO
anatase, 325 mesh, $99% metals basis), formic acid (HCOOH,
95%), and 2-hexyne (C 10, 99%) were purchased from Sigma
Aldrich. Citric acid anhydrous (C ) was obtained from
VWR International, sulfuric acid (H ) from Mallinckrodt,
perchloric acid (HClO , 1 M) from Fluka, and ethanol (C H OH)
followed by 0.3 mL of a 2 mM aqueous K
2
PdCl
4
solution. The
(C
19
H
2
,
ꢁ
reaction mixture was mixed by inversion and kept at 90 C for 4
days. Au@Pd octahedra having edge length 40 nm were
prepared by increasing the volume of K PdCl solution to 0.5
2 4
mL. To obtain 42 nm Au@Pd octahedra, the volume of Au
octahedral seed was reduced to 0.05 mL, and rest of the reagent
amount were the same as used for the synthesis of 40 nm
Au@Pd octahedra.
>
6
H
6
8 7
H O
2
SO
4
4
2
5
from Pharmco. Nanopure water with resistivity 18.2 MU cm
Millipore) prepared using an ultrapure water system (Millipore)
(
was used in all experiments.
Supporting nanocrystals on TiO2
The nanocrystals were supported on TiO by stirring a slurry of
2
Synthesis of size-controlled Pd octahedra
TiO in 120 mL of water and a solution of as prepared nano-
2
For the preparation of Pd octahedra, initially smaller Pd octa- crystals overnight at room temperature. 0.12 mM of TiO was
2
hedra (ꢀ23 nm) were prepared, which were used as seeds for the added to appropriate amount of Pd and Au@Pd nanocrystal
1
6
ꢃ6
deposition of more Pd and growth of larger Pd octahedra.
solution containing ꢀ2.5 ꢂ 10 surface atoms (102.36 ꢂ 10
g
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2016

View Article Online
Paper
Journal of Materials Chemistry A
ꢃ5
of Pd for 37 nm Pd octahedra, and 2.89 ꢂ 10 g Pd for 37 nm electron microscope (SEM) operated at 30 kV and a spot size of
ꢃ5
Au@Pd octahedra, 5.21 ꢂ 10 g Pd for 40 nm Au@Pd octa- 3. Nanocrystal composition was examined with an Oxford INCA
ꢃ5
hedra, and 8.26 ꢂ 10 g Pd for 42 nm Au@Pd nanoctahedra). Energy Dispersive X-ray operating at 30 kV. High magnication
The samples were collected by centrifugation at 3900 rpm for images were taken on a JEOL JEM 3200FS TEM at 300 kV and
9
0 min. The supernatant were discarded and the TiO
2
sup- a spot size of 1 with a Gatan 4k ꢂ 4k Ultrascan 4000. Energy
ported nanocrystals were dried in a vacuum oven at room dispersive X-ray spectra were obtained with an Oxford INCA
temperature. To compare the intrinsic catalytic activity the Dispersive X-ray system interfaced to the JEM 3200FS TEM,
number of surface atoms relative to 2-hexyne were kept operating at 300 kV. Samples for TEM analysis were prepared by
constant, assuming that different nanocrystals have similar drop-casting a dispersed particle solution onto a carbon-coated
degrees of aggregation. It was also assumed that one face of the copper grid and then washing the grid twice with water.
octahedra were unavailable for catalysis as it was attached to Samples for SEM and EDX analysis were prepared by drop-
the support.
casting a dispersed particle solution onto a silicon wafer and
then washing the wafer three times with methanol aer initial
solvent evaporation. The concentrations of Pd in all the samples
prepared above were measured using Perkin Elmer ELAN DRCII
ICP-MS. Pd nanocrystals were digested in aqua regia and stan-
dard solutions of known concentration of Pd were used to
obtain the correct concentration. The products from the catal-
ysis reactions were analyzed using an Agilent 6890 Gas Chro-
matograph and 5973 Inert Mass Selective Detector. X-ray
photoelectron spectra of the Pd nanocrystals were collected
using PHI Versa Probe II scanning X-ray microprobe under
ultra-high vacuum condition and with a monochromatic Al Ka
X-ray source. The carbon 1s peak was calibrated at 284.8 eV and
was used as an internal standard.
Hydrogenation of 2-hexyne
In a typical reaction, a 250 mL three-necked round bottom ask,
equipped with a stir bar, was evacuated for 30 min. The ask was
lled with hydrogen using a balloon. Next, a mixture of supported
Pd nanocrystals dispersed in 20 mL ethanol was introduced in the
reaction vessel along with 0.25 mL 2-hexyne. Aliquots were
collected at regular intervals and centrifuged at 15 000 rpm for 15
min to remove the TiO2 supported metals as non-volatile
substance may contaminate the column of the gas chromato-
graph used for analyzing samples. The products from the catal-
ysis reactions were separated and analyzed using an Agilent 6890
Gas Chromatograph and 5973 Inert Mass Selective Detector.
Electrochemical oxidation of formic acid
Results
A three electrode cell system was utilized to measure the elec-
trochemical properties. A glassy carbon electrode was used as
the working electrode. A saturated calomel electrode was used
as a reference electrode and a Pt wire was used as a counter
electrode. A dispersion of nanoparticles in nanopure water was
deposited on the glassy carbon electrode and dried. A blank
scan in Ar saturated 0.5 M H SO solution was performed within
Au@Pd octahedra with varying shell thicknesses of 1–2 nm, 5–6
nm, and 7–9 nm were generated by seed-mediated growth. First,
2
4
the potential range ꢃ0.2 V to 1.0 V for about 200 cycles to clean
the surface of the sample deposited on the working electrode.
For the electrochemical oxidation of formic acid, the cyclic
voltammograms were recorded in Ar saturated 0.5 M HClO
4
and
0
1
.5 M formic acid solution within the potential range ꢃ0.2 V to
ꢃ1
.0 V at a sweep rate of 100 mV s . The current was normalized
to the electrochemically active surface area of the catalysts in
the electrode. Traditionally, ECSA were estimated from the
integrated reduction charge of surface palladium oxide by
ꢃ2
assuming a charge of 0.42 mC cm is required for the reduc-
tion of palladium oxide monolayer for all surfaces. However, the
surface area calculation should account for the atom packing on
ꢃ2
the surface. We have calculated a charge of 0.489 mC cm for
ꢃ2
the (111) surface and 0.420 mC cm for the (100) surface to
reduce the palladium oxide monolayer. Current density
measurements using both the standard reduction charge and
surface-specic reduction charge are calculated here.
Characterization
Fig. 1 SEM images of (a) 33 nm Au octahedra, (b) 37 nm Au@Pd
octahedra (1–2 nm shell thickness), (c) 40 nm Au@Pd octahedra (5–6
nm shell thickness), and (d) 42 nm Au@Pd octahedra (7–9 nm shell
Images of the nanocrystals were taken via a JEOL JEM 1010
Transmission Electron Microscope (TEM) at 80 kV with a ROM
CCD camera and on a FEI Quanta 600F environmental scanning thickness). TEM images and size distributions are provided in the insets.
This journal is © The Royal Society of Chemistry 2016
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
Au octahedra were prepared with an average edge length of 32 present the activity and selectivity results for the Au@Pd
24
nm by standard methods. The SEM image in Fig. 1a conrms nanocrystals, followed by a discussion of the Pd-shell thickness
high quality Au octahedra, with good sample monodispersity dependence to performance observed in both systems.
(Fig. 1a, lower inset). They were then used as seeds for the
deposition of Pd shells. Specically, K PdCl was reduced with
2
4
Hydrogenation of 2-hexyne with shape-controlled Au@Pd
nanocrystals
citric acid in the presence of CTAC and the octahedral Au seeds.
The Au seeds are preferential sites for heterogeneous nucleation
and growth of Pd. Both the citrate ions and CTAC are {111} facet
stabilizers that aid in the formation of Au@Pd octahedra. As
expected, as the concentration of K PdCl is increased, larger
In the case of Pd-based catalysts for 2-hexyne semi-
hydrogenation, all nanocrystals were supported on TiO
2
microparticles (Fig. S2†). The supported catalysts were cleaned
by washing with water multiple times, as other methods used to
remove capping agents are destructive to nanocrystal shape and
2
4
Au@Pd octahedra, with thicker Pd shells are produced (Fig. 1b–
d). SEM and TEM images (Fig. 1b–d) reveal that the majority of
the crystals are octahedral in shape, with slight truncations at
the corners. The corner truncation exposes {100} facets, and the
extent of truncation increases with increasing shell thicknesses.
TEM images conrm the core@shell structure on account of Z-
contrast difference between Au and Pd, along with elemental
mapping by STEM-EDX coupled with line analysis (Fig. 2 and
S1†). The thickness of the Pd shell is tuned from 1–2 nm to 5–6
nm to 7–9 nm, with increasing Pd concentration in the over-
growth step. A small lattice mismatch and slow growth rate
2
surface composition. This approach removes the majority of
CTAB or CTAC as our previous XPS characterization was unable
ꢃ
ꢃ
to detect the Br 3d and Cl 2p peaks by routine analysis of Pd
2
nanocrystals treated in this manner. Without the support, the
nanocrystals agglomerated in the ethanol-based media from
which the reaction took place. These supported nanocrystals
were then used as catalysts for the hydrogenation of 2-hexyne,
with the number of Pd surface atoms being available for catal-
ysis held constant for all samples (see Experimental section).
Fig. 3 shows the percentage of 2-hexyne consumed as a function
of time for the Au@Pd octahedra with increasing shell thick-
nesses compared to similarly sized Pd octahedra. The intrinsic
25
facilitated conformal growth of Pd on the Au seeds.
These samples were then studied as catalysts for two model
reactions: 2-hexyne hydrogenation and formic acid oxidation.
These reactions were selected because earlier reports by us and
other groups demonstrated that these reactions are sensitive to
2,26–28
catalyst structure and composition.
Moreover, selective
hydrogenation is important to the pharmaceutical and food,
29
and petrochemical industries while formic acid oxidation is
30–32
a model for small molecule oxidation in fuel cells.
First, we
Fig. 3 (a) Plot of % of 2-hexyne remaining as a function of reaction
Fig. 2 TEM images (a, d and g), HAADF-STEM-EDX mapping (b, e and time for 37 nm Pd, 37 nm Au@Pd, 40 nm Au@Pd, and 42 nm Au@Pd
h), and cross-sectional compositional line profiles (c, f and i) of Au@Pd octahedra. (b) Plot of a relative conversion of reagent and products, as
octahedra with varying Pd shell-thickness.
a function of time for 37 nm Au@Pd octahedra.
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2016

View Article Online
Paper
Journal of Materials Chemistry A
Table 1 Average edge length, shell thickness, selectivity toward cis-2-
hexene (at the complete consumption of 2-hexyne) and turnover
frequency (TOF) (per surface atom) of Pd and Au@Pd octahedra
Formic acid oxidation with shape-controlled Au@Pd
nanocrystals
In the case of FAO, the Pd-based nanocrystals were deposited on
a glassy carbon electrode prior to electrocatalysis. Blank scans
were conducted to clean the surface of the nanocrystals, as
described in the Experimental section. FAO was conducted
using standard methods, with perchloric acid as a supporting
Edge
length
Shell
thickness
(nm)
Selectivity
(%)
TOF
ꢃ1
(nm)
(s
)
Pd octahedra
37
37
40
42
—
86.5 (ꢄ1.2)
94.5 (ꢄ0.8)
88.2 (ꢄ1.8)
91.3 (ꢄ0.7)
5.9 (ꢄ0.4)
33
Au@Pd octahedra
Au@Pd octahedra
Au@Pd octahedra
1–2
5–6
7–9
14.0 (ꢄ0.6) electrolyte as sulfuric acid is known to poison metal surfaces.
5.0 (ꢄ0.3)
3.5 (ꢄ0.2)
Fig. 4 compares the cyclic voltammograms (second cycle) of the
Pd and Au@Pd octahedra during FAO. The current densities
were normalized to the electrochemically active surface area
(ECSA). Traditionally, ECSA are estimated from the integrated
catalytic activity (i.e., activity normalized to the number of reduction-charge of surface palladium oxide, with an assump-
ꢃ2
surface atoms) is greater for the thin shell (1–2 nm) Au@Pd tion that a charge of 0.42 mC cm is required for the reduction
16
octahedra compared to similarly size Pd octahedra. However, of a palladium oxide monolayer for all surfaces. However, the
with shell thicknesses of Au@Pd octahedra of 5–6 nm, the surface area calculation should account for the atomic packing
catalytic activity is comparable to the Pd-only octahedra. With on the surface, which in the case of Pd-based octahedra are
even thicker shells (7–9 nm shell), the catalytic activity is encased in {111} facets. For the Pd(111) surface, a charge of
ꢃ2
decreased further, below that of the Pd-only octahedra. The 0.489 mC cm
is required to reduce the palladium oxide
results are summarized in Table 1. All of the catalysts are monolayer. Thus, current density measurements using both the
selective towards cis-2-hexene. The product distribution as standard reduction-charge and surface-specic reduction-
a function of time is shown in Fig. 3b for the 37 nm Au@Pd charge are provided here. Regardless of the value used, the
octahedra, with the others in Fig. S3.† cis-2-hexene is generated trends were the same. The maximum current densities in the
selectively, and only aer complete consumption of 2-hexyne forward anodic scan and the corresponding potentials relative
will isomerization occur to produce trans-2-hexene, followed by to the saturated calomel electrode are summarized in Table 2.
hydrogenation to hexane. The origin of these observations will As can be seen, the highest current density was obtained with
be discussed shortly.
the thinnest shelled Au@Pd octahedra, being higher than both
the Pd-only octahedra and the Au@Pd octahedra with thicker
shells. This nding suggests that the thinnest shelled Au@Pd
octahedra are active in a wider potential window and do not get
20
blocked by hydroxyl coverage until very high potentials.
Similarly sized thin-shelled (1–2 nm) Au@Pd octahedra are
twice as active compared to Pd-only octahedra.
Discussion
The results from both model reactions indicate that the Au core
effects the catalytic activity of Pd shells. This observation is to be
expected as the Au core should regulate the width and energy of
the Pd d-band, which results in altered catalytic activity by
varying the binding strength between the molecular substrate
and the catalyst surface. The ligand effect should not impact
21
surface reactivity over more than a few atomic layers. With this
distance-dependence in mind, the strain effect should be
dominant in our Au@Pd samples as the thinnest shell studied
is still $1 nm. Information from known reaction mechanisms
in conjunction with XPS analysis will shed light on the
Fig. 4 Current densities for formic acid oxidation (second scan) by 37
nm Pd, 37 nm Au@Pd, 40 nm Au@Pd, and 42 nm Au@Pd octahedra.
Table 2 Average edge length, shell thickness, formic acid oxidation potential and current density of Pd and Au@Pd octahedra
ꢃ2
ꢃ2
Edge length (nm)
Shell thickness (nm)
E (V) vs. SCE
J (mA cm ) (traditional)
J (mA cm ) (surface specic)
Pd octahedra
37
37
40
42
—
0.24
0.20
0.20
0.22
6.82 (ꢄ0.65)
14.40 (ꢄ1.44)
4.85 (ꢄ0.22)
5.25 (ꢄ0.26)
7.91 (ꢄ0.65)
16.69 (ꢄ1.44)
5.62 (ꢄ0.22)
6.56 (ꢄ0.26)
Au@Pd octahedra
Au@Pd octahedra
Au@Pd octahedra
1–2
5–6
7–9
This journal is © The Royal Society of Chemistry 2016
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
20
possibility for strain-engineered nanocatalysts as a function of variable shell thicknesses. We note, however, that the
shell thickness. expression of {100} truncations at larger nanocrystal sizes may
XPS results indicate that there is a shi towards lower also contribute to these changes in electronic structure, and its
binding energy for the Pd 3d doublet for both 1–2 nm shell and contribution to reactivity is discussed later in the manuscript.
5–6 nm shell Au@Pd octahedra relative to Pd-only octahedra,
As the d-band moves towards the Fermi level for the Au@Pd
with the Pd 3d doublet for the Pd-only octahedra shied to octahedra, the binding strength between the molecular substrate
slightly higher binding energy relative to bulk Pd likely on and catalyst surface would increase in accordance with the
34
account of partial surface oxidation (Fig. 5). The XPS spectra Norskov model. The enhanced interaction between the molec-
were recorded from the unsupported catalysts as the signals ular substrate and the Pd surface increases the activity when
obtained from the TiO2 supported catalysts were too weak comparing the thinnest shelled Au@Pd octahedra to Pd-only
owing to minimal catalyst loading on TiO . However, TiO
octahedra for both model reactions. For 2-hexyne hydrogenation,
2
2
supports oen partially oxidize Pd surfaces, reducing the active stronger binding between the Pd surface and hexyne promotes
sites for hydrogenation. Although the support may inuence the introduction of more double bond character into the C^C
36
activity, this inuence should be similar for all supported bond, which enhances the hydrogenation rate. For FAO, the
35
catalysts, allowing for a direct comparison. The shi toward reaction proceeds through formate ion formation on Pd surfaces,
lower binding energy for the Au@Pd core@shell nanostructures and stronger bonding between the catalyst surface and formate
3
7
2
is consistent with tensile strain from the Au core and has been activates C–H bond scission, with facile emission of CO .
observed previously. Interestingly, with thick shelled (7–9 nm
12
Interestingly, molecular substrate–surface binding events
shell) Au@Pd octahedra, the binding energy shis back to the should be even stronger for the thicker (5–6 nm) shelled octa-
Pd-only octahedra value. This shi is consistent with the strain hedra, as suggested by XPS data. However, this increased
effect being diminished at long distances and has been interaction does not manifest in a greater hydrogenation rate or
observed in case of hollow Au-core Pd-shell nanoparticles with current density in the case of FAO (Fig. 3 and 4). Rather,
decreased rates are observed for both systems. Instead, we
surmise that the surface is being poisoned due to very strong
binding between the {111} facets of the Pd-based octahedra and
molecular substrates. In the case of 2-hexyne hydrogenation,
this poisoning may arise due to stronger binding between the
alkyne and the Pd surface, which blocks the catalyst surface
from further reaction. In the case of FAO, this poisoning may
arise from adsorption of COOH generated from dehydrogena-
23
tion of HCOOH. The thicker shells may actually enhance the
binding strength of the COOH intermediate.
For the thickest shelled octahedra, the XPS datum indicates
that the activities should be similar to the Pd-only octahedra as
the binding energies are similar. In the case of 2-hexyne
hydrogenation, however, the activity is diminished below that of
the Pd-only octahedra. We attribute this inconsistency to the
emergence of {100} facets with increasing Pd shell thickness
(Fig. 2a versus d and g). In fact, previous work by our group and
others found that {111} Pd facets are more active than {100} Pd
2
facets for 2-hexyne hydrogenation. Unfortunately, efforts to
suppress this truncation during synthesis through the use of
additional {111}-stabilizing capping agents were unsuccessful.
In the case of FAO, the maximum current density obtained is
nearly identical to the Pd-only octahedra, although some slight
differences in the shape of the CV are evident.
In general, these results with Au@Pd octahedra support the
concept of a “volcano” plot proposed by Sabatier regarding the
relationship between the reaction rate and adsorption ener-
15
gies. That is, weakly adsorbing molecular substrates may not
bind sufficiently on the catalyst surface to facilitate the neces-
sary bond breaking and making steps. Whereas, strongly
adsorbing molecular substrates may poison the catalyst surface.
The strength of the molecular substrate–surface interaction
must be “just right” to maximize the rate of reaction, with our
results indicating that these interactions can be manipulated
through the core@shell architecture.
Fig. 5 XPS of the Pd 3d region for 37 nm Pd, 37 nm Au@Pd (thin shell),
0 nm (thicker shell), and 42 nm Au@Pd (thickest shell) octahedra. The
black line corresponds to the Pd d5/2 binding energy for bulk Pd(0), the
royal blue line corresponds to Pd(II), and the orange line corresponds
to Pd(IV).
4
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2016

View Article Online
Paper
We note that defects could also contribute toward perfor-
Journal of Materials Chemistry A
3 S. J. Hwang, S.-K. Kim, J.-G. Lee, S.-C. Lee, J. H. Jang, P. Kim,
T.-H. Lim, Y.-E. Sung and S. J. Yoo, J. Am. Chem. Soc., 2012,
134, 19508–19511.
4 K. An and G. A. Somorjai, ChemCatChem, 2012, 4, 1512–1524.
5 J. X. Wang, H. Inada, L. Wu, Y. Zhu, Y. Choi, P. Liu,
W.-P. Zhou and R. R. Adzic, J. Am. Chem. Soc., 2009, 131,
17298–17302.
6 P. Moseley and W. A. Curtin, Nano Lett., 2015, 15, 4089–4095.
7 R. Van Hardeveld and F. Hartog, Surf. Sci., 1969, 15, 189–230.
8 R. Van Hardeveld and F. Hartog, Adv. Catal., 1972, 22, 75.
9 J. Chen, Q. Zhang, Y. Wang and H. Wan, Adv. Synth. Catal.,
2008, 350, 453–464.
mance as they have different electronic states than either face
and edge atoms. However, the model systems and the catalysts
were chosen so that their contribution would be minimal. Both
semi-hydrogenation and FAO prefer extended facets for reac-
tions, with defects providing poor geometric match and lower
activity. Moreover, high resolution images of Pd nanocubes and
octahedra of similar size to those used in the study here do not
undergo large structural rearrangements during catalysis and
can be viewed as at surfaces, and the core@shell architecture
is preserved aer use in these experiments (Fig. S4†). Thus,
while the contribution of defects to the catalytic results cannot
be fully ruled out, the systems were designed to minimize their 10 S. Mostafa, F. Behafarid, J. R. Croy, L. K. Ono, L. Li,
contribution and reveal the relationship between shell thick-
ness and performance in shape-controlled core@shell
nanocrystals.
J. C. Yang, A. I. Frenkel and B. R. Cuenya, J. Am. Chem.
Soc., 2010, 132, 15714–15719.
11 S. Cheong, L. Graham, G. L. Brett, A. M. Henning, J. Watt,
P. J. Miedziak, M. Song, Y. Takeda, S. H. Taylor and
R. D. Tilley, ChemSusChem, 2013, 6, 1858–1862.
Conclusions
1
2 J. W. Hong, D. Kim, Y. W. Lee, M. Kim, S. W. Kang and
The introduction of a Au core to shape-controlled Pd nano-
S. W. Han, Angew. Chem., Int. Ed., 2011, 50, 8876–8880.
crystals can enhance catalytic activity through precisely control- 13 W. Tang and G. Henkelman, J. Chem. Phys., 2009, 130,
ling the thickness of the Pd shell. This enhancement is attributed 194504.
to tensile strain which arises from the lattice mismatch between 14 B. Hammer and J. K. Nørskov, in Adv. Catal., Academic Press,
Au and Pd. Interestingly, this enhancement was only observed 2000, vol. 45, pp. 71–129.
with the thinnest shell thicknesses observed. With thicker shells, 15 R. I. Masel, Principles of adsorption and reaction on solid
activity diminishes due to poisoning of the catalyst surface. Yet
surfaces, John Wiley & Sons, 1996.
with even thicker shells, the tensile strain is released, as revealed 16 C.-H. Kuo, L. K. Lamontagne, C. N. Brodsky, L.-Y. Chou,
by XPS, and the catalyst performs similarly to Pd-only nano-
crystals. This study is consistent with the Sabatier principle,
J. Zhuang, B. T. Sneed, M. K. Sheehan and C.-K. Tsung,
ChemSusChem, 2013, 6, 1993–2000.
which outlines the need for balanced interaction strengths to 17 S. Xie, S.-I. Choi, N. Lu, L. T. Roling, J. A. Herron, L. Zhang,
achieve maximum rates of conversion. While demonstrated in
a model Au@Pd nanocrystal system for two model reactions, we
J. Park, J. Wang, M. J. Kim, Z. Xie, M. Mavrikakis and Y. Xia,
Nano Lett., 2014, 14, 3570–3576.
anticipate this ability to tune reactivity through bimetallic 18 X. Wang, S.-I. Choi, L. T. Roling, M. Luo, C. Ma, L. Zhang,
architecture will hold for other systems. Moreover, this approach
to tunable reactivity may prove more straight-forward than
M. Chi, J. Liu, Z. Xie, J. A. Herron, M. Mavrikakis and
Y. Xia, Nat. Commun., 2015, 6, 7594.
engineering alloyed nanoparticles, where the bimetallic archi- 19 D. J. Childers, N. M. Schweitzer, S. M. K. Shahri, R. M. Rioux,
tecture can modulate both the electronics (though strain and
charge transfer) and active site geometry in Au–Pd systems.
J. T. Miller and R. J. Meyer, Catal. Sci. Technol., 2014, 4, 4366–
4377.
38,39
20 C. Hsu, C. Huang, Y. Hao and F. Liu, Electrochem. Commun.,
2
012, 23, 133–136.
Acknowledgements
2
1 P. Strasser, S. Koh, T. Anniyev, J. Greeley, K. More, C. Yu,
Z. Liu, S. Kaya, D. Nordlund, H. Ogasawara, M. F. Toney
and A. Nilsson, Nat. Chem., 2010, 2, 454–460.
This work is supported by the U.S. Department of Energy (Basic
Energy Sciences) through an Early Career Award Grant (No.
205883). Professor Skrabalak is also supported through the 22 P. Dolle, R. Baudoing-Savois, M. De Santis, M. C. Saint-Lager,
Camille Dreyfus Teacher Scholar Program. We thank Prof.
David P. Giedroc and John P. Lisher for their help with the ICP-
M. Abel, J. C. Bertolini and P. Delich `e re, Surf. Sci., 2002, 518,
1–13.
MS instrument. We thank Dr Yaroslav Lozovyy for his assistance 23 L. A. Kibler, A. M. El-Aziz and D. M. Kolb, J. Mol. Catal. A:
with the XPS, which was made possible with NSF MRI Grant No. Chem., 2003, 199, 57–63.
126394, and Dr Jonathan A. Karty for his help with the GC-MS 24 C.-C. Chang, H.-L. Wu, C.-H. Kuo and M. H. Huang, Chem.
1
instrument. We also thank the IU Nanoscale Characterization
Facility for access to the necessary instrumentation.
Mater., 2008, 20, 7570–7574.
25 Y. Ding, F. Fan, Z. Tian and Z. L. Wang, J. Am. Chem. Soc.,
2010, 132, 12480–12486.
2
6 N. Semagina, A. Renken and L. Kiwi-Minsker, J. Phys. Chem.
C, 2007, 111, 13933–13937.
References
1
2
F. Zaera, ChemSusChem, 2013, 6, 1797–1820.
M. Laskar and S. E. Skrabalak, ACS Catal., 2014, 4, 1120–
27 A. S ´a rk ´a ny, O. Geszti and G. S ´a fr ´a n, Appl. Catal., A, 2008, 350,
157–163.
1128.
This journal is © The Royal Society of Chemistry 2016
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
2
2
3
3
3
3
3
8 X. Zhang, H. Yin, J. Wang, L. Chang, Y. Gao, W. Liu and 35 K. V. Kovtunov, D. A. Barskiy, O. G. Salnikov, D. B. Burueva,
Z. Tang, Nanoscale, 2013, 5, 8392–8397.
9 C. H. Bartholomew and R. J. Farrauto, Fundamentals of
industrial catalytic processes, John Wiley & Sons, 2011.
A. K. Khudorozhkov, A. V. Bukhtiyarov, I. P. Prosvirin,
E. Y. Gerasimov, V. I. Bukhtiyarov and I. V. Koptyug,
ChemCatChem, 2015, 7, 2581–2584.
0 M. Baldauf and D. M. Kolb, J. Phys. Chem., 1996, 100, 11375– 36 F. Zaera, Chem. Rev., 1995, 95, 2651–2693.
1
1381.
37 K. Tedsree, C. W. A. Chan, S. Jones, Q. Cuan, W.-K. Li,
X.-Q. Gong and S. C. E. Tsang, Science, 2011, 332, 224–228.
38 A. R. Wilson, K. Sun, M. Chi, R. M. Whitell, J. M. LeBeaull,
H. H. Lamb and B. J. Wiley, J. Phys. Chem. C, 2013, 117,
17557–17566.
1 M. J. Llorca, J. M. Feliu, A. Aldaz and J. Clavilier, J.
Electroanal. Chem., 1994, 376, 151–160.
2 C. Rice, S. Ha, R. I. Masel, P. Waszczuk, A. Wieckowski and
T. Barnard, J. Power Sources, 2002, 111, 83–89.
3 H. Miyake, T. Okada, G. Samjeske and M. Osawa, Phys. 39 F. Liu, D. Wechsler and P. Zhang, Chem. Phys. Lett., 2008,
Chem. Chem. Phys., 2008, 10, 3662–3669.
461, 254–259.
4 G. Collins, M. Schmidt, G. P. McGlacken, C. O'Dwyer and
J. D. Holmes, J. Phys. Chem. C, 2014, 118, 6522–6530.
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2016