Sie sind auf Seite 1von 19

REVIEW

Employing high-resolution materials characterization


to understand the effects of Pd nanoparticle structure
on their activity as catalysts for olefin hydrogenation
Marc R. Knecht & Dennis B. Pacardo
Received: 24 November 2009 / Revised: 25 January 2010 / Accepted: 26 January 2010 / Published online: 16 February 2010
# Springer-Verlag 2010
Abstract Recent developments in nanotechnology have led to
the production of new materials with a wide array of
applications, particularly in catalysis. Because of their small
size, nanoparticles have a maximized surface-to-volume ratio,
thus making them attractive targets for use as catalytic
structures; however, the number of analytical techniques
available to fully characterize materials on such a size scale is
quite limited. As a result, a complete understanding of the entire
nanoparticle structure remains unclear, especially when con-
sidering the active structural motif from which the specific
activity arises. Metallic Pd materials have been widely studied
due to their immense potential as catalysts for reactions such as
olefin hydrogenation and CC bond synthesis. These materials
require surface passivants to act as ligands and stabilize the
nanoparticles against aggregation and bulk formation. These
ligands have the added value to function as gates that
selectively allow reagents to reach the active surface of the Pd
nanoparticles for chemical turnover. This accounts for the
observed selectivities of the catalysts with the corresponding
changes in the turnover frequency values. Here we present a
broad overview of recent advances in the use of Pd nano-
particles for the industrially important hydrogenation reaction
with a focus on characterizing and understanding the base
structural effects that give rise to the catalytic activity.
Keywords Pd nanoparticles
.
Hydrogenation
.
Nanocatalysts
.
Materials characterization
.
Structurefunction relationship
Current trends in catalysis
The study of chemical catalysis has seen a renaissance in the
last decade, particularly because of the onset of nanotechnol-
ogy. The development of new synthetic methods to produce
functional materials with nanoscale dimensions paved the way
for the fabrication of nanoparticle catalysts that are highly
efficient, selective, environmentally friendly, and can be used in
a wide variety of reactions [15]. The nanoscale dimensions of
these materials make them attractive for catalytic applications
because of their large surface-to-volume ratio, which max-
imizes the amount of catalytically active materials exposed to
solution. For optimum materials, the nanoparticles must
maintain their structural stability while mediating the specific
catalytic reaction. This level of activity is typically controlled
by the ligands used to passivate the surface [6], which prevent
particle degradation while allowing for reagent interactions
with the catalytic particles [1, 5]. Different ligands or
stabilizers such as polymers [7, 8], dendrimers [1], and
peptides [2, 9] have been developed, all of which can be
used to fine-tune the surface structure of the materials for
the desired application. Beyond stability considerations, the
ligands can be used to determine the functionality of the
particles by acting as selective filters and mediators that allow
for specific molecular orientations for reactivity [1012]. This
enables the development of stable and selective catalysts,
which are desired for reactivity, selectivity, and for the design
Electronic supplementary material The online version of this article
(doi:10.1007/s00216-010-3516-z) contains supplementary material,
which is available to authorized users.
M. R. Knecht (*)
:
D. B. Pacardo
Department of Chemistry, University of Kentucky,
101 Chemistry-Physics Building,
Lexington, KY 40506-0055, USA
e-mail: mrknec2@email.uky.edu
Anal Bioanal Chem (2010) 397:11371155
DOI 10.1007/s00216-010-3516-z
of enhanced systems that are both energy-efficient and
environmentally friendly.
Surprisingly, there are limited analytical characterization
techniques that can be used to fully probe the structure of
nanomaterials, especially for particles within the 110 nm
size range, because the materials are too large to be studied
by many traditional molecular approaches and too small to
be probed by many solid-state methods. As a result, the
level of detail known about the surface structure, metallic
composition, and the interactions between these two levels
is incomplete. This level of information is important to
identify the structural characteristics that give rise to the
final activity. While methods are available, for example
electron microscopy [13], X-ray absorption spectroscopy
[1420], and diffraction techniques [21], they are unable,
individually, to fully probe the complete structure with
atomic level resolution. Instrumental advances are currently
being developed; however, the availability of such new
techniques is limited. Furthermore, it is fundamentally
important to use these characterization methods to probe
the ligand surface, as this region is likely to play a
significant role in the activity of the materials, especially
for catalytic capabilities.
Nanoparticle catalysts typically fall into a variety of
categories; however, two main groups are normally repre-
sented. First, homogeneous catalysts are those that are
dispersed into solution and remain independent species
before, during, and after the reaction [1, 2, 4, 12, 2226].
The second group, termed heterogeneous catalysts, is
usually attached to a solid support and remains undissolved
in the reaction medium [4, 2730]. While differences in
reactivity are prevalent for both types of catalytic species,
the advantage of heterogeneous catalysts lies in their facile
removal for product purification. Additionally, for these
insoluble species, modulations in their reactivity over their
dispersible counterparts can arise from interactions with the
solid support, which can change the electronic makeup of
the catalytic system. Beyond these classifications, the actual
size and structure of the nanocatalyst can have a dramatic
effect on their reactivity [5, 8, 28, 31]. By forcing the size
towards smaller particles, the surface-to-volume ratio is
greatly maximized, which enhances the reactive surface
area for the catalytic process [28, 31]; however, the degree
of difficulty of characterization at these smaller sizes
dramatically increases. Additionally, upon altering the
particle shape, changes in reactivity have been noted [5,
7, 8, 32]. This change is likely to arise from surface
coordinatively unsaturated metal atoms that exist in higher
energy states. Under such conditions, these metal atoms,
which typically lie at particle defect sites, are able to strongly
bind and interact with reagents in solution in attempts to
lower their individual energy states. This can result in
driving the specific chemical reactions at faster rates.
While the initial structures of the materials play important
roles in functionality, two possible catalytic mechanisms exist
that can alter the framework of the materials. In the first
mechanism, the catalytic reaction occurs directly on the
metallic surface of the materials [1, 31, 33]. As a result,
minimal changes to the overall particle structure are
expected, assuming that the capping ligands are able to
accommodate this functionality. In the second mechanism,
an atom-leaching effect is employed to drive the catalytic
reaction [3, 24, 34]. To that end, during the reaction, Pd
atoms are released from the nanoparticles into solution,
where they act as the catalytic species. Upon completion, the
nanoparticles are regenerated from the leached Pd atoms,
thus using the nanoparticles as a reservoir of catalytically
active species. This second mechanism, which is expected to
drastically alter the nanoparticle structure, is rather difficult
to characterize, especially for changes at the particle surface
and structure during the reaction.
While these general concepts tend to hold true for
nanocatalysts, different reactions and their specific mecha-
nisms can be altered by the composition and architecture of
the materials, which controls the electronic properties of the
nanosystem [35]. Below is a broad overview and descrip-
tion of the current state of nanocatalysts and how their
structure can directly alter their reaction capabilities for
olefin hydrogenation [36]. A significant focus is on
analytical characterization of the materials and how their
configuration affects the hydrogenation process. This
reaction was chosen because of its broad application in
both academic and industrial settings. Furthermore, the
hydrogenation reaction has also been used extensively as a
model system for characterizing and understanding nano-
catalytic effects to determine the efficiency of new
materials, thus increasing its relevance to basic research.
Synthesis of monometallic nanoparticle catalysts
Noble metals nanoparticles such as Pd [2, 3, 5, 10, 22, 24,
2729, 31, 34, 3739], Pt [33, 4042], and Au [4345]
have been widely studied as catalysts because of their
chemical reactivity. While many strategies have been
developed to fabricate these nanoparticles, ranging from
extraction-based methods to those using thermal decompo-
sition, most synthetic conditions currently employ the
reduction of metal-ions, obtained via inorganic salts, in
the presence of a ligand or stabilizing agent [1, 2, 4, 6, 46,
47]. In most cases, the dissolved metal salt binds to the
selected functional group of the ligand through a variety of
interactions, for example electrostatics or covalent bonding.
The molecular structure and composition of the metal ion
ligand complex can then control the composition, size,
shape, and structure of the final nanomaterial. Upon
1138 M.R. Knecht, D.B. Pacardo
reduction using NaBH
4
or comparable reagents, the metal
ions are fully reduced to the zero valent state, from which
nucleation and particle growth can ensue to produce the
catalytically reactive metallic materials. The final compos-
ite architecture of the materials is directly controlled by the
concentration of reagents in solution, the strength of the
reductant, the degree of the interaction between the metal
ions and the ligands, the metal ion-to-ligand ratio, and the
strength of the interaction between the ligands and the zero
valent metal surface for nanoparticle passivation [6]. While
most reactions typically result in the production of spherical
particles, which is thermodynamically preferred, as shown
in Fig. 1, production of other shapes such as rods, cubes,
tetrahedra, and wires have been achieved by modulating the
reaction conditions and through the use of seeding-based
reaction schemes [7, 8, 48, 49]. Furthermore, the shape of
the nanostructure can be controlled by the interactions of
the passivant ligands with the growing surface; some
ligands preferentially bind to different crystallographic
facets, which can alter their growth rates compared with
other, non-bound surfaces, thus resulting in shaped nano-
material fabrication.
For the basic synthesis of Pd nanomaterials, which is a
fundamentally important component for numerous catalytic
reactions, reduction of Pd
2+
or Pd
4+
in the presence of
appropriate monodentate coordinating ligands typically
produces spherical zero valent Pd particles [47, 50, 51].
The final structure of the materials is a metallic core that is
passivated on the surface with numerous ligands. The
identity of the binding moiety is chosen on the basis of
hardsoft acidbase theory [52] to produce optimally
passivated surfaces, which typically employ amines and
thiols for Pd. It is from the molecular structure of the ligand
that a variety of nanoparticle properties are derived,
including optical traits, solubility, and chemical reactivity.
Zamborini et al. have used this approach to prepare Pd
nanoparticles of diameters <2 nm that were extensively
characterized using
1
H NMR and TEM imaging [47]. In
this study, hexanethiolate (C
6
) and dodecanethiolate (C
12
)
ligands were employed and the size of the particles
generated was inversely proportional to the Pd-to-ligand
ratio used. To that end, as the specific ratio increased, the
size of the nanoparticles decreased; using a C
6
ligand-to-Pd
ratio of 1/2, 1, and 2, nanoparticles with dimensions of 3.0,
2.2, and 1.6 nm, respectively, were prepared [47]. To
further characterize this system,
1
H NMR spectroscopy was
employed. For this analysis, detection of resonances
associated with specific protons within the carbon chain
backbone was dependent upon their proximity to the
nanoparticle surface. For instance, for the C
6
materials
prepared with a ratio of 1/2 (3.0 nm), the terminal CH
3
and internal methylene protons were readily observed with
Fig. 1 TEM images of Pd
nanomaterials of various shapes
including (a) spherical DENs
(adapted with permission from
Ref. [23]; copyright 2004
American Chemical Society),
(b) nanorods (adapted with
permission from Ref. [49];
copyright 2007 American
Chemical Society), (c) nano-
cubes (adapted with permission
from Ref. [49]; copyright 2007
American Chemical Society),
and (d) nanowires (adapted with
permission from Ref. [48];
copyright 2009 American
Chemical Society)
Employing high-resolution materials characterization to understand the effects of Pd nanoparticle structure 1139
somewhat broad resonances; however, the protons bound to
the -carbon next to the surface-attached thiol could not be
observed. As the particle size decreased with increasing
ratios, these peaks tended to become less broad and the -
carbon protons could be observed [47]. This is likely to be
because of the small sizes of the clusters, which were not
observable using standard TEM methods. Furthermore,
some of these smaller particles may form organometallic-
based clusters, which would be small enough to minimize
the peak broadening effects observed with the larger
nanoparticles.
Beyond simple monodentate ligands, other polydentate
ligands have been explored to prepare Pd nanoparticles [8].
Using these species, chelate effects can be used to impart a
higher level of surface coordination between the individual
ligand and the nanoparticle. To that end, polymers such as
poly-(N-vinyl-2-pyrrolidone) (PVP) are optimum structures
as they present numerous amines that can bind and
passivate the surface [8, 34]. Using nearly identical
synthetic techniques as described above, but replacing the
ligands with appropriate multidentate species, nanoparticles
of 2.10.1 nm can readily be produced, as characterized by
extensive TEM measurements [34]. As further monitored
by Narayanan and El-Sayed using analytical TEM analysis,
changes to the particle size were noted for the materials
after multiple uses as catalysts for Suzuki C-coupling
reactions [34]. After the first reaction, the particle size
shifted to 2.90.3 nm, which suggests a change in the
particle morphology as a result of the catalytic reaction.
Furthermore, when these same particles were employed for
a second reaction, their diameters shifted to 2.20.2 nm
[34]. Based upon statistical analysis, these particle sizes
suggest that an atom leaching and deposition mechanism
may be responsible for the Suzuki reaction. The changes in
the diameter are likely to arise from the deposition of Pd
atoms on to remaining nanoparticles on completion of the
reaction. These atoms would likely be deposited on the
nearest particle, resulting in particle size shifts, rather than
evenly on to all particles in solution to maintain the initial
particle size. This level of information, which is highly
important in understanding catalytic mechanisms, is
obtained from TEM measurements. Furthermore, for such
an analysis, it is important to ensure that a statistically
relevant nanoparticle population is studied to ensure the
validity of the results. For instance, for the Suzuki-based
analysis, the diameters of 1800 particles per sample were
determined to guarantee that the changes in size were
statistically accurate [34].
Beyond the approach of employing ligands to passivate
the surface of Pd nanoparticles, other approaches have been
studied that employ templates to grow and protect nano-
particles within the structure [1, 46, 5355]. A unique
example of this fabrication method uses polyamidoamine
(PAMAM) dendrimers as a nanoreactor/template for the
production of nearly monodisperse nanoparticles with a
variety of metallic compositions, including Pd and Pt [16,
38, 46]. PAMAM dendrimers are monodisperse and highly
branched polymers that are grown radially from an ethyl-
enediamine core [56]. The final size of the polymer is
determined at the synthetic level, because different branch
lengths can be controllably grown using traditional organic
synthetic routes. The final structure can present chemical
functionalities, for example amines, alcohols, and carbox-
ylic acids, to the solution whereas the interior of the
dendrimer possesses numerous tertiary amine and amide
groups that are able to chemically coordinate metal ions
[1, 56]. Additionally, based upon the branched structure, an
internal void volume is present within the dendrimer that
can accommodate various chemical loads, including nano-
particles to prepare dendrimer-encapsulated nanoparticles
(DENs) [57].
PAMAM dendrimers are able to template the formation of
nanoparticles in a two-step manner [1, 15, 5861], which is
shown schematically in Fig. 2. First, metal ions are mixed in
the presence of the templates from which the ions can be
extracted from solution into the interior of the dendrimer.
This process is facilitated by the metal-ions covalently
binding to the tertiary amines to form a dendrimermetal
ion complex. Second, the metal ions are reduced by NaBH
4
,
which results in their coalescence within the dendrimer to
form the zerovalent nanoparticle. Based upon this synthesis,
a high level of control can be gained over the final size of the
materials. Using PAMAM dendrimers, a 1:1 ratio of Pd
2+
ions to tertiary amines can be achieved so that the number of
amines within the template controls the upper limit of
particle size [16, 37, 46, 62]. For instance, the sixth
Fig. 2 Schematic representation of the synthesis of dendrimer-
encapsulated nanoparticles
1140 M.R. Knecht, D.B. Pacardo
generation PAMAM dendrimer possesses 256 tertiary
amines; therefore, the structure is capable of producing metal
nanoparticles that possess 250 metal atoms [56]. Addition-
ally, the average number of atoms per particle can be
controlled by reaction stoichiometry, thus enabling produc-
tion of Pd nanoparticles with selected average numbers of
metal atoms at sizes <4 nm [31]. Furthermore, metallic
nanoparticles prepared in this manner are not passivated on
their exposed surface, because of the template-based struc-
ture [15, 63]; therefore, much of the metallic surface is
presented to the solution for catalytic studies.
An extensive number of characterization techniques have
been employed to confirm the complex structure of DENs
ranging from microscopy to spectroscopy [1, 15, 16, 31, 46,
58, 59]. Initial characterization of Pd DENs has used UV
visible spectroscopic measurements. From this, a ligand-to-
metal charge-transfer (LMCT) band is observed at 224 nm
[46]. This is consistent with the binding of Pd
2+
ions to the
interior tertiary amines to form the precursor complex.
Using a titration analysis, it has been determined that a 1:1
ratio of interior amines-to-Pd
2+
binding is possible, which
was elucidated using UVvisible; therefore, the upper
threshold of the number of atoms per particle is limited
by the number of amines [46]. Upon reduction, the size and
morphology of the metallic Pd materials has been exten-
sively characterized using TEM [31]. From this process,
nanoparticles ranging from 1.02.5 nm have been generated
where the stoichiometric ratio of Pd-to-dendrimer con-
trolled the final particle dimensions. To that end, the
particle size correlated quite well with a particle composed
of the number of atoms used in the stoichiometry [31].
While the above characterization methods possess the
ability to probe the metallic composition, elucidation of the
interactions of the dendrimer with the nanoparticle surface
required the use of high-resolution techniques such as
extended X-ray absorbance fine structure (EXAFS) [1519]
and
1
H NMR spectroscopies [63, 64]. Using these
techniques, atomic level information can be obtained
concerning the metallicligand surface, which is exceed-
ingly difficult to probe. From both spectroscopic methods,
it has been determined that minimal to no interactions
between the dendrimer template and the metallic surface,
specifically Au for these studies, are present [18, 63], which
suggests that the surface is able to interact with reagents in
solution. Furthermore, from the EXAFS studies, an AuAu
coordination number (CN) of 9.0 was achieved for Au
DENs with an average of 147 atoms [18], which corre-
sponds well with the expected theoretical value of 8.98 for
a cuboctahedral Au nanoparticle of 147 atoms [14].
Interestingly, contraction of the AuAu bond length was
determined spectroscopically using EXAFS2.832 for
the DENs versus 2.87 for bulk Au; this suggests that a
high degree of surface tension is present in the system and
causing these changes [18]. This evidence is important in
understanding the catalytic functionality of the materials on
the basis of the exposed reactive sites. For instance, these
results indicate that a significant fraction of the metallic and
catalytically active surface is not bound by the interior
dendrimer functionalities and is freely available for reac-
tivity. This is a unique feature associated with dendrimer-
based materials, which required highly sensitive and not
readily available spectroscopic techniques to confirm this
surface structural feature.
Monometallic olefin hydrogenation
Hydrogenation is an increasingly important reaction for
industrial applications, including the production of pharma-
ceuticals, polymers, and petroleum products. In this
reaction, H
2
gas is added to the system and used as a
reductant to produce the final product [36]. Traditional
chemical moieties and functional groups that are targeted
for hydrogenation reactivity include alkenes, alkynes,
carboxylic acids, aldehydes, and ketones [36]. A degree of
selectivity in these reactions is desired to produce enantio-
merically pure products, and to proceed at one reactive
moiety within a substrate molecule when multiple points of
reactivity exist in the same species. Targeting specific
moieties, rather than fully hydrogenating the entire chem-
ical compound, can reduce certain functional groups within
the molecule without removing necessary chemical struc-
tures and functionality. Here we focus on the use of Pd
nanocatalysts used in olefin hydrogenation and further
discuss how their structure alters their reactivity. This
reaction is an efficient probe of the catalytic capabilities
of nanomaterials and can be done using relatively simple
and inexpensive experimental analyses.
To achieve maximum hydrogenation reactivity, a high
surface area is desirable; however, to achieve such
conditions, the particle size should be approximately 1
10 nm [31]. Unfortunately, in this size range, the surface
energies of the Pd system are such that aggregation and
formation of bulk Pd black precipitate are expected. To
overcome this issue, surface passivants are required that
may lower the reactivity by binding and poisoning reactive
sites [6, 47]. To avoid this effect, DENs are ideal catalyst
candidates [1]. In this structure, the guest nanoparticles are
encapsulated within the dendrimer hosts where minimal
surface binding has been observed between the template
and the metallic surface [15, 63]. Using this method,
nanoparticles of <500 atoms are routinely prepared, which
places them in the optimized 13 nm diameter size range
for reactivity by maximizing the surface-to-volume ratio.
DENs demonstrate catalytic activity by allowing substrates
to penetrate the dendrimer periphery, to react at the exposed
Employing high-resolution materials characterization to understand the effects of Pd nanoparticle structure 1141
metallic surface, then letting the products diffuse back into
solution [10, 22, 23, 31, 33, 37, 38]. To differentiate between
DENs of different sizes, GX-OH(Pd
Y
) notation will be used
to describe Pd nanoparticles synthesized within X generation
of hydroxyl-terminated PAMAM dendrimers that contain, on
average, Y Pd atoms. Initial studies using G4-OH(Pd
40
)
DENs, which had an average particle size of 1.30.3 nm
measured by TEM analysis [38, 46], indicated high turnover
frequencies (TOFs) for the catalytic hydrogenation of allyl
alcohol to 1-propanol [38]. To analytically quantitate this
reaction, and obtain data that can be used for direct
comparison of reactivity, an aqueous DENs solution is
saturated with H
2
, to which the olefin-containing substrate,
allyl alcohol, is injected. By monitoring the change in gas
volume above the reaction, product yields can be calculated
to determine TOFs. For this reaction, a TOF of 218 mol
product (mol Pdh)
1
was observed [38]. Interestingly, as
the generation of the dendrimer increased, a decrease in the
TOF was observed. For instance when G6-OH(Pd
40
) or G8-
OH(Pd
40
) DENs were used, TOF values of 201 and 134 mol
product (mol Pdh)
1
, respectively, were determined [38].
These changes were attributed to the increased steric effects
of branch crowding at the dendrimer periphery, shown in
Fig. 3, which is a consequence of large dendrimers [38]. As a
result, the rate of substrate partitioning into the dendrimer is
slower for larger generations of dendrimers. These differ-
ences in TOFs can also be used to characterize the
nanoparticle structure, where the reactivity further supports
the encapsulated nanoparticle structure over a surface
passivated motif. This is a unique example where catalysis
can be used to probe the structure of the catalytic material.
This effect was further tested to determine the degree of
selectivity of DENs [10]. In this regard, rather than solely
changing the dendrimer generation, the molecular size of the
catalytic substrate was also modulated to determine if size-
based gating could be achieved using various generations of
dendrimers. This is attractive because of the cost of higher-
generation dendrimers [56], and the degree of selectivity for
the lower generations may allow for more reactivity. In this
study, G4-OH(Pd
40
), G6-OH(Pd
40
), and G8-OH(Pd
40
) were
all tested across five different allyl alcohol derivatives that
were increasingly larger in molecular size, as shown in
Table 1 [10]. Consistent with the previous study [38], as the
generation increased, the TOF values decreased, because of
the surface steric crowding at the dendrimer periphery
(Fig. 3); however, as the molecular size increased, the TOF
values again decreased for particles prepared using the same
dendrimer generation. To that end, when G4-OH(Pd
40
) was
employed as the catalyst, a TOF of 480 mol H
2
(mol Pdh)
1
was achieved using substrate 1, but when the largest
molecules were used as substrate, species 5, a TOF value of
100 mol product (mol Pdh)
1
was observed [10]. This
decrease is likely to have been caused by the larger molecular
size lowering its ability to penetrate and diffuse through the
dendrimer structure. Similar trends are observed for G6-OH
(Pd
40
) and G8-OH(Pd
40
), with lower TOFs compared with
the G4-OH(Pd
40
) species [10].
While the DENs possessed significant catalytic reactiv-
ity, their controlled structure at very small sizes (<4 nm)
Fig. 3 Effects of dendrimer generation on the catalytic activity of Pd
DENs, as studied by Niu et al. [10]. As the generation size gets larger,
steric crowding at the dendrimer periphery increases to act as a size-
selective gate to control the catalytic reactivity of the materials
1142 M.R. Knecht, D.B. Pacardo
enables more quantitative analysis of the hydrogenation
reaction mechanism. From an understanding of how the
reaction proceeds, optimized structures can be elucidated
and integrated into more complex systems to maximize the
reactivity for future catalysts. By being able to select the
average number of metal atoms per cluster by reaction
stoichiometry, the numbers of surface atoms, face atoms,
and defect atoms can be directly determined [31]. Here,
defect atoms refer to atoms along crystallographic edges
and vertices of the metallic structure. The numbers of these
specific atoms change with increasing particle sizes;
therefore, it can be analytically determined how these
substitutions alter the reactivity based upon the near
monodispersity of DENs of this size. For this analysis,
G6-OH dendrimers were used to prepare DENs with 55,
100, 147, 200, and 250 Pd atoms per cluster on average
[31]. The DENs were analyzed by TEM, which demon-
strated particle sizes of 1.3, 1.4, 1.5, 1.7, and 1.9 nm,
respectively, which correlated with the expected values
[31]. Figure 4a displays the calculated values of surface,
defect, and face atoms for these nanoparticles over sizes
between 1.0 and 2.0 nm [31]. Because the reaction is likely
to occur on the nanoparticle surface, normalization of TOF
values to the mols of the various types of atoms at
controlled positions can occur. Through this process,
identical TOFs are expected when the catalytic active site
is used to normalize the data. As presented in Fig. 4b, when
the TOF values for the various sized DENs are normalized
by the surface atoms, defect atoms, or number of particles
in solution, a linear trend with a positive slope is observed
[31]. This indicates that these values do not accurately
represent the active site; however, when the data are
normalized to the number of face atoms on the nanoparticle
surface, identical values are achieved for DENs with sizes
1.5 nm [31]. This suggests that the hydrogenation reaction
occurs preferentially at face atoms of the metallic compo-
nent. Interestingly, at sizes<1.5 nm, different normalization
trends are observed and are attributed to electronic effects
that are expected at these smallest of sizes [31].
In a unique recent study, Mizugaki et al. used Pd
nanoparticles as a core material to bind PAMAM dendrons
to the metallic surface [65]. The dendrons are grown from the
primary amine of 4-picolylamine, thus leaving the pyridine
ring available for nanoparticle surface binding. The dendrons
are grown from the first generation (G1), to the second
generation (G2), to the third generation (G3). Additionally,
each generation of dendrons are terminated with either C
6
or
C
12
aliphatic chains, thus rendering the materials soluble in
organic solvents [65]. Regardless of the surface function-
alities, when dissolved in solution employing organic
solvents, the dendrons self-assemble into micelle-like frame-
works with similar average diameters ranging between 9.4
and 11.8 nm, as measured by dynamic light scattering (DLS)
Table 1 TOF values for Gn-OH(Pd
40
) DENs using allyl alcohol and
structural analogues. (Adapted with permission from Ref. [10];
copyright 2001 American Chemical Society)
Fig. 4 Catalytic effects of particle size of Pd DENs. (a) Comparison of the ratio of atoms within the Pd nanoparticles. (b) Comparison of the TOF
values normalized to the respective Pd atoms. (Adapted with permission from Ref. [31]; copyright 2006 American Chemical Society)
Employing high-resolution materials characterization to understand the effects of Pd nanoparticle structure 1143
[65]. Size and assembly-based characterization of the
dendrons in solution is important to understanding the
synthetic mechanism for materials production. From this
information, a template-based synthetic scheme can be
developed, where comparison with sizes after nanoparticle
formation can be addressed to qualitatively describe the
effects of the ligand structural surface on the reactivity. At
the center of the micelle-like scaffold, Pd
2+
can be
sequestered; upon reduction, this produces nanoparticles that
are weakly bound by the pyridine ring. The formation of the
Pd core does not change the size of the micellular structure;
however, Pd nanoparticles with diameters of 6.82.0 nm,
5.31.7 nm, and 5.11.9 nm are grown employing the G1-
C6, G2-C6, and G3-C6 dendrons, respectively [65]. Inter-
estingly, similar effects are observed with the C
12
terminated
materials, but when using these dendrons, smaller Pd
nanoparticles are prepared with diameters of 4.61.9 nm,
3.61.3 nm, and 3.51.3 nm for the G1-C12, G2-C12, and
G3-C12 dendrons, respectively [65]. Based upon this
synthetic strategy, it is likely that the pyridine rings of the
core branching point interact with the Pd surface to stabilize
the nanomaterials.
Catalytic analysis of the materials for hydrogenation
produced size-dependent reaction effects; however, for
these materials, as the generation of the dendron increased,
the TOF of the materials for 1,3-cyclooctadiene to cyclo-
octene increased also. This effect is in contrast with the
size-dependent catalytic reactivity of the DENs discussed
above and is rationalized from the degree of surface binding
by the dendrons in the nanocomposite structure [65]. For all
of the dendrons, regardless of generation, G1, G2, or G3, or
surface functionality, C
6
or C
12
, nearly similar sized self-
assembled micelles are prepared, which was ascertained
using DLS characterization. As a result, it is likely to take
fewer G3 dendrons to produce the micelles compared with
the G1 materials. From this, fewer pyridine rings are
associated with the G3 micelle interior to bind to the Pd
nanoparticle; therefore, more exposed metallic surface is
available from these materials to drive the hydrogenation
reaction, which results in the higher TOF values. Identical
trends are observed by transitioning from C
6
to C
12
surface
functionalities, which can alter the number of dendrons
within the micelle structure. It is based upon the character-
ization of the materials structure, ascertained from a
combination of DLS and TEM studies, that a functional
mechanism to understand the catalytic properties of the
materials can be achieved. Without this level of structural
detail, determination of base causes for the different
reactivities would be challenging.
The naturally occurring cage-like protein ferritin also
functions as a template and stabilizer for metal nano-
particles similar to synthetic dendrimers [66]. The X-ray
crystal structure of ferritin, shown in Fig. 5, has 24 subunits
that self-assemble to form a caged framework with a cavity
for iron-storage in cells [66]. The structure of these proteins
includes hollow channels that allow metal ions to traverse
to the protein core; on reduction these produce nano-
particles of sizes selected by the metal-to-protein ratio [12,
66, 67]. By using this biomimetic method, spherical Pd
nanoparticles with sizes of 2.00.3 nm are formed at the
cavity of the protein, as characterized by traditional TEM
[12]. Further characterization of the nanoparticle synthesis
was conducted using gel electrophoresis. In this study, both
the apo-ferritin and nanoparticle containing protein migrat-
ed at the same rate, which indicated that minimal to no
changes in the protein structure/size occurred as a result of
encapsulation of the metallic nanoparticles. This was
corroborated using negative staining-based TEM where
the protein structure was clearly observed surrounding the
metallic nanoparticle to characterize the general structure of
the materials [12].
The catalytic activity of the Pd@ferritin nanoparticles for
hydrogenation was evaluated using increasingly larger
olefin substrates. As shown in Table 2, the TOF values of
the Pd@ferritin nanoparticles for the hydrogenation of
olefin containing compounds are lower than the values
Fig. 5 Analysis of the ferritin
protein used to prepare Pd
nanoparticles: (a) full crystal
structure, and(b) internal and
external dimensions of the
protein [12]. (Copyright
WileyVCH. Reproduced
with permission.)
1144 M.R. Knecht, D.B. Pacardo
generated from Pd-DENs [12], which may be an effect of
the surface passivant (protein versus dendrimer). Currently,
the interactions of the protein interior with the nanoparticle
surface have not been elucidated; therefore, no conclusions
can be drawn about the degree of Pd metallic surface
exposure. In this arrangement, however, the protein is likely
to cover the Pd surface to a higher degree than the template
PAMAM dendrimers, which is expected to reduce the
reaction rates. While the overall values are lower for the
Pd@ferritin materials, the system is again sensitive to
molecular size with larger substrates producing decreased
TOFs compared with smaller reagents [12]. This is likely to
be the effect of the protein allowing smaller molecules to
penetrate and interact with the metallic surface while
sterically preventing larger species from reacting based
upon pathways to the metallic catalyst; however, the charge
of the substrate species may also control the degree of
protein penetration. While the overall rates are lower for the
biomimetic system, the advantage here is that the templates
are prepared using biological techniques that are environ-
mentally friendly, over the stringent organic synthesis
required for PAMAM dendrimer fabrication [56].
While template-based methods to prepare homogeneous
Pd materials represent good strategies to probe the
structure-function relationship of nanocatalysts, the separa-
tion of the catalytic species from the product post-reaction
can be problematic. In many cases, both the catalysts and
product molecules are highly soluble in the reaction
medium, which requires the use of time-consuming
separation techniques, some of which can be complicated.
Furthermore, for industrial applications, such affects can
become cost prohibitive. By transitioning from a homoge-
neous to a heterogeneous system, such separation factors
can be minimized while maintaining the desired degree of
reactivity and selectivity from the catalyst.
Bruening and coworkers have fabricated heterogeneous
composite systems with high selectivity. As shown in Fig. 6,
Table 2 Olefin hydrogenation TOF values for ferritin-encapsulated
Pd nanoparticles employing variously sized substrate molecules [12].
(Copyright WileyVCH; reproduced with permission)
Substrate TOF
Pd@ferritin Pd particles
CH
2
= CHCONH
2
720.7 585.9
CH
2
= CHCOOH
2
6.31.1 122.6
CH
2
= CHCONH-iPr 516.5 150.3
CH
2
= CHCONH-tBu 315.9 6.10.6
CH
2
= CHCO-Gly-Ome 6.33.8 282.6
CH
2
= CHCO-D,L-Ala-OMe not detected 230.3
Fig. 6 Synthetic scheme for
production of Pd nanoparticles
embedded within a multilayer
polyelectrolyte film (MPF) on
the surface of alumina beads, as
developed by Bruening and
coworkers [29]
Employing high-resolution materials characterization to understand the effects of Pd nanoparticle structure 1145
for this system, multilayer polyelectrolyte films (MPFs) are
deposited on to the surface of alumina beads, from which the
Pd nanoparticles are grown within the polyelectrolyte film
[2729, 39]. The films are grown by alternating the
deposition of poly(acrylic acid) (PAA) and a Pd
2+
-containing
polyethylenimine (PEI) on to the alumina surface. Subse-
quent addition of NaBH
4
reduces the Pd
2+
to form zero
valent Pd nanoparticles embedded within the structure, which
can be fully separated from the solution by using filtration.
Based upon the materials structure, traditional TEM imaging
of the Pd nanoparticles on the alumina bead surface would be
difficult; the background associated with the bead would
prevent the ability to observe the Pd nanoparticles embedded
within the MPF. To overcome this issue, a model system was
employed in which the films and embedded nanoparticles
were grown directly on a TEM grid. This enabled observa-
tion of the Pd nanoparticles without interference from the
beads. Using this method, 13-nm Pd particles are prepared
within the polymeric film [39]. This arrangement of materials
encapsulates the catalytic particles within a polymeric
structure, similar to that observed with the DENs. Although
these are not exactly the structures used for catalysis, they
represent sufficient model systems of the likely structures
attached to the bead surface.
Initial catalytic analysis of the MPF materials for the
hydrogenation reaction demonstrated a high degree of
selectivity. These studies focused on the hydrogenation of
allyl alcohol to form 1-propanol. For this reaction employing
MPF materials coated with 3.5 bilayers of PAA/PEI, a TOF of
727128 mol product (mol Pdh)
1
was observed [39].
When the number of bilayers in the system was increased to
7.0, the TOF drastically reduced to a value of 14120 mol
product (mol Pdh)
1
[39]. This suggests that the thickness
of the polyelectrolyte layer controls the catalytic reactivity of
the materials by minimizing interactions with the Pd metallic
surface. Based upon the TEM evidence, the nanoparticles are
indeed embedded within the polymeric surface; therefore,
with a higher number of coatings, more Pd is buried deeper
within the framework, thus possibly being unavailable for
use for catalysis because of the steric effects of the additional
polymer layers. A control analysis of commercially available
5% Pd on alumina without a polymeric coating demonstrated
a TOF of 1300150 mol product (mol Pdh)
1
, which
supports the gating-based hypothesis derived from the
materials structure [39].
To further probe the size-selective capabilities of the
heterogeneous materials, 1-penten-3-ol and 3-methyl-1-
penten-3-ol were catalytically hydrogenated to form their
reduced products. For these materials, with 3.5 polyelectrolyte
bilayers, TOF values of 27823 mol product (mol Pdh)
1
and 6015 mol product (mol Pdh)
1
, respectively, were
observed [39]. These values are directly in line with those
that were expected, based upon a size gating effect suggested
by the structure of the materials observed via TEM
characterization. When the film thickness was further
increased to 7.0 bilayers, the TOF values decreased to 65
8 mol product (mol Pdh)
1
for 1-penten-3-ol, whereas the
3-methyl-1-penten-3-ol species demonstrated a low TOF
value of 61 mol product (mol Pdh)
1
[39]. These
decreasing values based upon substrate size and the poly-
electrolye film thickness are attributed to the size-gating
effects of the materials. Mixtures of these substrates
confirmed the selectivity of the reaction [39]. Interestingly,
when commercially available materials are used, significantly
higher TOF values are recorded; however, the degree of
selectivity is fully lost. The appropriate system should
therefore be selected before the reaction, on the basis of the
chemical functionality of the substrate and the desired
product.
While these materials demonstrate selectivity for the
hydrogenation of multiple substrates in solution, the ability
to reduce a single olefin within a molecular structure that
possesses multiple double bonds is desirable. This capability
has the potential to simplify synthetic schemes in which full
hydrogenation of all olefins within a compound is undesirable.
To study this ability, 1,5-heptadien-4-ol and 3,7-dimethyl-6-
octadien-3-ol were used [27]. Using the first substrate, a TOF
value of 183535 mol product (mol Pdh)
1
was observed,
with 97% conversion to 2-hepten-4-ol product over the fully
reduced product (4-heptanol) [27]. Studies of the second
compound resulted in 97% conversion to 3,7-dimethyl-6-
octen-3-ol; however, a much lower TOF was observed
compared with the first substrate (700140 mol product
(mol Pdh)
1
) [27]. From this study, two key facts became
evident about the reactivity of the MPFs. First, the more
exposed olefin was preferentially reduced. Considering the
structure of the nanomaterial, this suggests that the molecule
can penetrate the surface of the polyelectrolyte film and
reduce the exposed olefin in a more efficient manner
compared with the second double bond, which is sterically
protected. Second, the degree of substitution at the reducible
double bond affects the TOF of the reaction. When bulkier
substituents are in closer proximity, the ability of this region
of the molecule to penetrate the polymer film is reduced,
thereby reducing the reaction rate. Nevertheless, these results
suggest high selectivity for molecules of different sizes, and
the position of the olefin within the structure of the substrate,
all of which is controlled by the architecture of the composite
nanosystem that is mediated by the steric constraints of the
polymeric framework.
Most studies employing MPFs have focused on the effects
of film thickness; however, changes to the structure and size of
the individual catalytic particles embedded in the polymer film
can also affect the reactivity. When using the alumina-based
systemwith three PAA-PEI layers, increasing the concentration
of Pd
2+
used during materials production resulted in the
1146 M.R. Knecht, D.B. Pacardo
fabrication of larger nanoparticles, as extensively character-
ized employing TEM. For this analysis, deposition of the Pd
nanoparticle containing polymeric films on to TEM grids was
used, as described above, to observe any changes in the
particle size based upon the concentration of Pd
2+
in the
system. To that end, when the Pd
2+
concentration was 1, 4, 8,
or 15 mM, the average Pd particle size within the MPFs was
2.20.5, 2.90.5, 3.20.6, and 3.40.9 nm, respectively
[28]. While a distribution of particle sizes is observed, the
trend is maintained of larger dimensions as required for the
analysis. Further characterization of the system was achieved
by use of atomic-emission spectroscopy. In this method, the
amount of Pd associated with each alumina bead-based
system was quantitatively determined for use in the normal-
ization of the catalytic studies. This is an important value to
ensure validity of the comparison of the catalytic effects of the
different sized Pd nanoparticles. When employing these
materials for catalytic hydrogenation, three clear trends are
observed:
1. the TOF values for monosubstituted alcohols decrease
with increasing particle size;
2. the TOF values for disubstituted alcohols increase with
increasing particle size; and
3. the selectivity of the catalysts decreases for larger
particles [28].
As shown in Table 3, when hydrogenating allyl alcohol,
the TOF values decrease from 2900570 mol product
(mol Pdh)
1
to 76040 mol product (mol Pdh)
1
as the
particle size increases over the range studied [28]. Isomer-
ization of the products was also observed; however, this was
not incorporated in the presented data. Interestingly, when
employing the disubstituted alcohol 2-methyl-2-propenol as
the substrate with catalytic materials of increasingly size, the
TOF values increased from 121 mol product (mol Pdh)
1
to 29070 mol product (mol Pdh)
1
[28]. This is in direct
contrast to the trend observed for allyl alcohol, a mono-
substituted species. As a result, the selectivities of the
catalysts for mono versus disubstituted alcohols ranged from
240 for the smallest particles to 2.6 for the largest materials
[28]. An identical trend was observed when using crotyl
alcohol as the disubstituted species [28]. Furthermore, similar
trends were observed for polymer-coated homogeneous
catalysts prepared using the same polyelectrolytes in solu-
tion, suggesting that the observed effects were the result of
the particles and not the MPFs.
The changes in the reactivity and selectivity for the mono
and disubstituted alcohols were attributed to geometrical
aspects of the Pd nanoparticles, as determined via extensive
TEM characterization. It is known that the number of defect
atoms and sites on the nanoparticle surface increase for
smaller particles, whereas the number of face/terrace atoms
increases for larger particles [31]. It is rather difficult to
visually observe such defects using traditional microscopy
techniques; however, spectroscopic methods may enable
such observation. The authors suggest that the reactivity of
monosubstituted alcohols preferentially occurs at the defect
sites, which would result in the higher TOFs for smaller
particles, while the disubstituted alcohols react at the terrace
sites that would yield higher TOFs with larger particles
where the number of these atoms would increase [28].
Previous research by Wilson et al. has also indicated a
preference for the hydrogenation reaction to occur along
terrace/face atoms, which would support this hypothesis
[31]. As a result of the reaction site requirements, which
appear to be controlled directly by the metallic structure and
the ligand interface, the controlled reactivity and selectivity
of the system could be altered by particle size, thus yielding
another structural/functionality tool for hydrogenation cata-
lyst design.
In a variant of the MPF system, Kidambi and Bruening
prepared similar materials using alumina beads; however,
they grew the polyelectrolyte films without use of PAA
[29]. In the PAA/PEI system, PAA was the first polymer
deposited on to the alumina surface. In the new strategy,
PdCl
4
2
was bound directly on to the alumina surface,
which can subsequently bind PEI in a second step to form
the polymeric layer [29]. This process is repeated multiple
times (PdCl
4
2
binding followed by PEI deposition) to form
the structure. Upon reduction, Pd nanoparticles are gener-
ated within the deposited PEI film. The main difference
between these materials and those previously prepared is
the fact that no PAA is present within the system. To fully
characterize the new PAA-free system, a battery of
approaches was employed. First, UVvisible and X-ray
photoelectron spectroscopies (XPS) were used to ensure
that binding of the Pd
2+
ions to the subsequent PEI
polymeric layer was occurring [29]. This was evident from
the formation of the LMCT band in the UVvisible at
220 nm that increased proportionally to the number of
layers deposited on the alumina bead surface. Based upon
the XPS studies, the Pd-to-Cl ratio was <1; therefore, it is
Table 3 TOF values for MPF-encapsulated Pd nanoparticles of
various sizes employing allyl alcohol and structural analogues.
(Adapted with permission from Ref. [28]; copyright 2009 American
Chemical Society)
Employing high-resolution materials characterization to understand the effects of Pd nanoparticle structure 1147
likely that the PEI is able to displace Cl and bind directly to
the Pd ions to maintain the structure [29]. After reduction,
TEM analysis was used to quantitate nanoparticle produc-
tion. From this method, Pd nanoparticles are prepared with
diameters between 1 and 3 nm that were embedded within
the PEI film [29]. Further XPS studies of the materials after
reduction indicated the generation of a fully reduced
product, which was suitable for catalytic studies [29].
Catalytic hydrogenation employing these structures, how-
ever, produced surprising results. Although these materials
demonstrated size-dependent gating effects similar to those
of other MPF materials, the films produced with only PEI
demonstrated higher TOF values with increasingly thicker
films. To that end, for n values of 1, 2, 3, and 4, where n
represents the number of PdCl
4
2
/PEI layers deposited on
the alumina surface, TOF values of 820, 750, 1990, and
1220 mol product (mol Pdh)
1
were observed [29]. This
suggests that with thicker films, higher catalytic rates are
possible. The authors reasoned that this was a structural
effect that resulted from employing a single polymeric
system over the double polymeric system used previously
[29]. In this method, PdCl
4
2
is directly adsorbed on to the
alumina surface; therefore, a large fraction of the initially
deposited catalytic materials is likely to be buried within
the pores of the metal oxide bead. As a result, these buried
components may prove to be inactive, because it is unlikely
that the substrate molecules are able to penetrate the
polymer layers and the pores of the alumina to interact in
this region. While the absolute amount of Pd metal trapped
in the pores remains constant as the number of deposited
layers increases, the fraction of the total metal that is
entrapped within the pores will decrease for thicker films.
To that end, the fraction of inactive Pd metal will decrease
for the larger structures, thus increasing their TOF values.
Similar results are observed for 1-penten-3-ol and 3-
methyl-1-penten-3-ol; however, the relative TOFs between
these two molecules decrease with the larger molecule, as
expected [29].
As demonstrated above, the ability of the nanoparticle
catalyst to affect the selectivity of the hydrogenation
reaction is influenced by the substrate, support, and/or
overall structure of the nanosystem. In the case of DENs or
MPFs, the polymeric structure acts as a filter that regulates
which molecules can reach the catalytically active surface
resulting in a selective gate based upon size and olefin
position. Thus, a sterically hindered molecule will take a
longer time to diffuse through the network and reach the
reactive sites. Other solid supports, for example carbon
nanotubes (CNTs) have been employed; these are promis-
ing materials for the development of bulk heterogeneous
catalysts [30, 68, 69]. These materials are interesting as
they possess a large surface area, which would be amenable
to driving extensive catalytic reactions. Their selectivity
may be lower than for other materials; however, such
structures would be of interest for the hydrogenation of
sterically constrained olefins, and other species.
Karousis et al. demonstrated the synthesis of Pd nano-
particles on the surface of CNTs with the aid of a surfactant,
sodium dodecyl sulfate (SDS), which acts as both the
solubilizing agent for the CNTs in polar solvents and the
reductant of the metal ion, as shown in Fig. 7 [30]. The Pd/
CNT composite materials are fabricated by sonicating the
bare CNTs in an aqueous solution of SDS. This disperses the
materials and results in the interaction of SDS with the CNT
surface. This step is the followed by heating of the materials
under reflux in the presence of Pd(OAc)
2
for 6 h [30], which
results in thermal reduction of the Pd
2+
ions to form Pd metal
atoms that nucleate and grow into Pd nanoparticles on the
CNT surface. The newly formed composite materials were
then extensively characterized using TEM, UVvisible
spectroscopy, thermogravimetric analysis (TGA), and XPS
[30]. TEM images, presented in Fig. 8, reveal the deposition
of 24 nm Pd nanoparticles on the CNTs. The nanoparticles
seem to fully coat the tube surface, suggesting that
homogeneous modification of the materials is possible to
achieve a high density of catalytic materials [30]. Further
Fig. 7 Schematic representation of the production of Pd nanoparticles
on the surface of carbon nanotubes (CNTs), as developed by Karousis
et al. [30]
1148 M.R. Knecht, D.B. Pacardo
characterization by TGA confirmed that the surfactant and
degradation products remained bound to the CNT surface
after nanoparticle fabrication, and could be thermally
removed at temperatures >150 C. This supports the
hypothesized synthetic process and indicates that these
molecules are likely to interact with the catalytic surface of
the Pd nanomaterials once formed. Further analysis by XPS
established the production of zero valent Pd nanoparticles
and the interaction with the CNT surface [30]. Together,
these results aided in confirmation of the composite/
heterogeneous structure, which was catalytically active.
The catalytic activity of the Pd/CNT composites in the
hydrogenation of olefins was studied using alcohol and
aldehyde-containing olefins [30]. For methyl-9-octadecenoate,
2-methyl-3-buten-2-ol, and 2-methyl-2-pentenal, one equi-
valent of H
2
was consumed in conversion of the materials
into the expected final product; however when 3,7-dimethyl-
2,6-octadien-1-ol was used, which contains two olefin
species, two equivalents of H
2
were used to generate the
single 3,7-dimethyl octanol compound [30]. Interestingly,
when the aldehyde-containing molecule 3-phenyl-2-propenal
was used, a multitude of products were formed from reduction
of either the olefin or the aldehyde after consumption of
two H
2
equivalents. The TOF values from the conversion of
methyl-9-octadecenoate and 2-methyl-2-pentenal were deter-
mined to be 8163 and 8030 mol product (mol Pdh)
1
,
respectively [30]. The TOF values of Pd/CNT composites are
almost four times those of traditional Pd on C catalysts [30].
This effect is likely to be because of the large active surface
area of the Pd nanoparticles on the CNT surface that is used to
drive the reaction, which was extensively characterized, as
discussed above, to confirm this capability. Although the
Pd/CNT materials show high reactivity, they were not
selective for the hydrogenated moiety using the described
conditions.
Another important characteristic of the Pd/CNT materials
is that they are also recoverable by centrifugation after the
reaction and can be recycled for further use. Similar TOF
values were observed for some substrates even after seven
catalytic cycles; however, some substrates did have dimin-
ished TOF values [30]. Although it is unclear which sub-
strates maintained their TOFs after multiple use, or the extent
to which the TOF value decreased, this ability is quite useful
for catalyst use in multiple reactions over time. This suggests
that the reaction mechanism minimally alters the catalytic
surface for future use; however, complicated in-situ reaction-
characterization studies are required to confirm this possibil-
ity. Also, the ability to simply separate the Pd/CNT materials
by centrifugation ensures minimal loss of the precious metal
structures after the reaction.
In a separate method, Ye et al. deposited Pd nano-
particles on the surface of CNTs by using an organometallic
precursor [70]. In this method, the CNTs were acid washed
to functionalize the surface with COOH groups, which are
used for metal nanoparticle deposition. The organometallic
complexes were then reduced in a high-pressure system,
using H
2
, to form the Pd nanoparticles on the CNT surface.
Initial characterization of the materials by TEM analysis
revealed Pd nanoparticles on the exterior surface of the
CNTs. Employing the organometallic starting materials,
nanoparticles between 5 nm and 10 nm were generated on
the solid support of the CNT. Energy-dispersive spectro-
scopic (EDS) analysis of the composite materials indicated
that the particles were indeed composed of Pd metal, and a
selected area electron diffraction (SAED) pattern of a single
nanoparticle indicated that the materials were crystalline.
These results were further corroborated by XPS studies of
the materials that confirmed that the structures were
composed of zero valent Pd. Further time-based XPS
studies of the nanocomposites indicated that the Pd
Fig. 8 TEM images of the production of Pd nanoparticles on the
surface of CNTs. (a) CNTs before the reaction. (b) The free,
unattached Pd nanoparticles. (c) The CNTs with Pd nanoparticles on
the surface. (Reprinted with permission from Ref. [30]; copyright
2008 American Chemical Society)
Employing high-resolution materials characterization to understand the effects of Pd nanoparticle structure 1149
components were stable to oxidation over time, thus
suggesting that the CNT-based platform may be a unique
method to achieve active catalysts [70]. Unfortunately, it
remains unclear what is present/interacting on the metallic
surface, which is an important characteristic when consid-
ering the catalytic reactivity of the materials.
For catalytic analysis of the CNT-based structures, the
hydrogenation of trans-stilbene in liquid CO
2
was studied
[70]. These conditions were employed to enhance the green
chemistry attributes of the materials, an important goal for
future nanocatalytic studies. Under the reaction conditions,
which occurred at a pressure of 10
7
Pa, product conversion
of 80% and 96% were observed at 5 min and 10 min reaction
times, respectively, employing a 10% Pd by loading at a
temperature of 23 C. These yields are significantly enhanced
compared with those for commercially available Pd on C, for
which the yield was 10% for a time >20 min [70]. This effect
is likely to be because of the high metallic surface area
presented to the reaction solution on the CNT surface, and
which is maximized by the nanocomposite structure.
Tessonnier and coworkers tested the hydrogenation
selectivity of Pd/CNT composites in which the Pd was
embedded within the CNTs [71]. These materials were
prepared by dispersing the CNTs in a Pd-ion containing
solution, which was dried and calcined to form PdO [71].
The PdO was then thermally reduced using H
2
and the final
embedded structures were confirmed by TEM. From this
method, based upon high-resolution, analytical TEM
studies of the materials, Pd nanoparticles of diameters
between 4 nm and 8 nm were clearly evident on the interior
surface of the materials. Images of the outer surface
revealed a lack of modification; however, it was evident
that the Pd nanoparticles were interacting with the interior
walls of the CNTs to generate the composite materials [71].
Unfortunately, it is unclear what remains bound to the
displayed nanoparticle surface, but it is evident that the
structures are catalytically active in the hydrogenation of
cinnamaldehyde, which contains C=C and C=O function-
alities [71]. Possible products of this hydrogenation
reaction, as shown in Fig. 9, include hydrocinnamaldehyde,
which arises from the hydrogenation of the C=C moiety,
cinnamyl alcohol from the hydrogenation of C=O bond,
and phenylpropanol from the hydrogenation of both
functional groups. The catalytic results demonstrate selec-
tive hydrogenation of the olefin over the aldehyde to
produce hydrocinnamaldehyde as the lone product. When a
commercially available charcoal-supported Pd catalyst was
used for comparison, this species produced a mixture of
both hydrocinnamaldehyde and phenylpropanol [71].
The authors attribute the high selectivity of the embedded
Pd/CNT catalysts to electron transfer between the metal
and the carbon nanotube, which is possible based upon the
observed structure, thus increasing the charge density on
the metal surface. Also, the absence of microporosity in
the nanoparticles is likely to change the adsorption
properties toward the reactants, which may give rise to
the selectivity.
Many catalytic reactions, including hydrogenation, occur
in organic solvents; however, changing to more environmen-
tally friendly conditions is desirable. Chun et al. have
designed a system which uses ionic liquids, which may prove
to be important for green chemistry, to solubilize Pd/CNT
materials for the hydrogenation reaction [68]. In this sense,
the CNTs are functionalized with imidazolium bromide,
which makes the Pd/CNT composites soluble in both water
and ionic liquids. Pd nanoparticles of 100.5 nm diameter
are produced on the functionalized CNTs, as determined by
high-resolution TEM, and their catalytic potential for
hydrogenation was tested using trans-stilbene in methanol
[68]. The TOF values show very high efficiency of 2820 mol
product (mol Pdh)
1
[68]. The ease of separation of the Pd-
CNTs from the reaction mixture enables the recycling of the
catalyst, which can be used at least ten times without any
loss of catalytic reactivity.
Bimetallic nanoparticles for olefin hydrogenation
Beyond synthetic strategies to optimize the ligand surface
of monometallic nanoparticles, the incorporation of a
second metal into the inorganic component has resulted in
further enhancement of the catalytic properties of the
system [15, 23, 35, 37, 72, 73]. These enhancements
mainly arise from two key factors:
1. geometric arrangement of the metal atoms for specific
interactions with the substrate, and
2. changes to the electronic character of the metallic
system.
In the first case, the metal atoms are specifically
arranged within the structure to incorporate multiple
Fig. 9 Catalytic pathways for the hydrogenation of cinnamaldehyde
1150 M.R. Knecht, D.B. Pacardo
binding/interaction points with the reagents in solution to
facilitate the reaction, which can increase the rate of product
formation. In the second case, changes to the electronic make-
up of the system can occur on the basis of the electro-
negativities of the two metals. In this event, the more
electronegative atoms pull electron density from the second
component, making the second materials more electropositive
and thus more reactive in the catalytic reaction. This second
case is likely to directly affect the hydrogenation reaction. For
this to occur, the additional metal should be more electroneg-
ative than Pd, which would impart a slight positive character-
istic to the catalytic moieties; this is believed to increase their
interactions with the H
2
and olefin substrate to enhance the
reaction TOF. Typical examples of secondary metals include
Au and Pt, forming the PdAu [15, 23, 26, 67, 74] and PdPt
[37] systems, respectively. Both bimetallic modes of action,
however, result in substantial complexity in full confirmation
of the structures of the materials. From their nature,
bimetallic nanoparticles can be placed into three types of
categories: alloys, coreshell, or phase segregated. Because
the orientation of the atoms is important to overall activity, it
is critical to determine their individual arrangement within
the metallic nanoparticle to confirm the structural effects of
the reaction. To achieve such a level of information,
atomically-resolved and quantitative data concerning the
system must be acquired, which can be rather difficult to
achieve. To address these objectives, different spectroscopic
and catalytic methods have been used to probe the structure
of the materials, as discussed below.
Bimetallic nanoparticles are typically fabricated in two
different structural motifs in which the two metals are alloyed
together or arranged in a coreshell pattern, as shown in
Fig. 10. These structures are obtained via the synthetic
procedure used to prepare the materials. To achieve bimetallic
alloys, co-reduction of two different metal salts, in the
presence of an appropriate surface passivant, can be used.
This causes the nucleation and growth of nanoparticles
wherein a mixture of the two metals is generated [15, 23,
67, 74, 75]. For the coreshell arrangement, denoted by X@Y
to describe these structures, where X represents the core metal
and Y represents the shell metal, sequential reduction of the
two metals is normally used [19, 23, 26, 74]. For this method,
production of a monometallic nanoparticle is processed, from
which these materials can be used as a subsequent catalytic
seed to deposit the secondary metal atoms along the
nanoparticle surface. This process results in the formation of
the coreshell structure with a high degree of control over the
resulting materials because of the step-wise-based procedure.
Both methods have been extensively used for production of
alloyed and coreshelled materials for hydrogenation applica-
tions in which one of the metals was Pd.
By combining the synthetic strategy of bimetallic nano-
particles with the PAMAM dendrimer templates, bimetallic
PdAu alloy and coreshell nanoparticles have been readily
prepared [15, 19, 23]. This provides the unique modulation
of the electronic character of the system, with the open
metallic surface for reactivity. Using the co-reduction
method, PdAu alloy nanoparticles were prepared with
diameters <2 nm; where the metallic composition can be
directly modulated by the stoichiometric ratio of the two
metal ions used in synthesis [23]. EXAFS analysis of these
materials confirmed a quasi random alloy; however, the
core and shell of the structure was slightly enriched with
Au and Pd, respectively [15]. Furthermore, no interactions
with the dendrimer were noted, suggesting that the surface
is highly exposed to drive the reaction. When used as
catalysts for olefin hydrogenation, the bimetallic alloy
DENs were significantly more reactive than physical
mixtures of Pd and Au DENs, shown in Fig. 11 [23]. For
this analysis, both the number of particles and the Pd:Au
Fig. 10 Synthetic schemes for production of bimetallic nanoparticles.
The top route, co-reduction, results in the production of alloyed
materials whereas the bottom route, sequential reduction, results in the
fabrication of coreshell structures
Fig. 11 Bimetallic effects on the hydrogenation catalytic reactivity of
PdAu alloy DENs as a function of the mole % Pd within the
nanoparticle structure. For comparison, the catalytic TOF values for a
physical mixture of monometallic Pd DENs and Au DENs are
presented (Reprinted with permission from Ref. [23]; copyright 2004
American Chemical Society)
Employing high-resolution materials characterization to understand the effects of Pd nanoparticle structure 1151
ratio in the reaction solution were maintained; however, in
the alloyed particles, the Pd and Au atoms were chemically
combined whereas in the physical mixture no interactions
between the two metals is expected. The data were
subsequently plotted on the basis of the mole % of Pd in
the reaction, which was further confirmed by EXAFS and
EDS analysis [15, 23], which demonstrated the alloyed
particles had enhanced reactivity compared with the
physical mixture at all points within the system [23]. The
maximum activity was observed from the alloyed DENs
that contained 25% Au atoms within the structure, which
nearly doubled the observed TOF value as compared with
analysis of the physical mixture.
Using the dendrimer system, Au@Pd materials were also
prepared using the sequential reduction method. For this,
monometallic Au DENs were prepared that contained on
average 55 atoms; these materials were subsequently used
as seeds to deposit Pd on to the metallic surface [19, 23].
TEM analysis confirmed the increase in the particle
diameter upon shell generation and the particle sizes
corresponded quite well with the theoretical size as
determined by the stoichiometry employed in the materials
fabrication [23]. More recent EXAFS analysis verified the
synthetic method and the final structure of the Au@Pd
arrangement [19]. Using these materials, enhancement in
the TOF values compared with monometallic Pd materials
of nearly the same size was observed. A two-atom-thick
shell layer of Pd on the Au core was required to increase
the reaction rate 1.25 times [23]. The enhancement of the
reaction was less than that observed for the alloyed
materials, which may be because of the different metallic
structures; more intimate contact is likely to occur in the
alloyed materials to optimize the electronic variations
required for the catalytic activity compared with the
Au@Pd structures. As a result, higher reaction rates may
be observed with alloyed materials; however, further
scientific studies are required to confirm this hypothesis.
Although this method can be used to prepare Au@Pd
materials, the reverse arrangement of Pd@Au could not be
generated. When such a synthetic technique was employed,
in which monometallic Pd cores are prepared, followed by
Au shell deposition, inversion of the two metals occurs
resulting in Pd on the shell with an Au core [19]. These
results were quite unexpected, but were fully confirmed by
EXAFS, TEM, and UVvisible analysis [19]. From the
TEM results, particles of the expected sizes, based upon the
molecular stoichiometries used in the reaction, were
achieved; however, this information is unable to probe the
metallic arrangement of the atoms within the inorganic
particle. EXAFS analysis demonstrated that inversion of the
two metals occurred, based upon changes to the MM CNs
extracted from the X-ray absorbance data, which suggested
the fabrication a Au@Pd structure over the intended
Pd@Au arrangement [19]. At present, it is unclear why
such an inversion process would occur; however, such
events demonstrate the need to fully characterize the
materials at the atomic level to ensure that a proper
structural motif has been achieved that addresses the
observed reactivity.
PdAu alloy nanoparticles have also been used for
selective hydrogenation using a heterogeneous solid sup-
port for facile catalyst extraction. In the method reported by
Prvulescu et al., Pd
2+
and Au
3+
ions were co-reduced in
the presence of a surfactant stabilizer that produced zero
valent PdAu alloy nanoparticles 3 nm in diameter, as
confirmed by TEM [75]. Furthermore, the XRD patterns
were consistent with the expected PdAu structure. These
materials were then incorporated into a modified solgel
procedure to generate SiO
2
, which encapsulated the
materials within a porous metal oxide network. Based on
the composite materials processing employed, the pore
diameter of the SiO
2
materials ranged from 16.6 to
24.0 nm, as determined by BrunauerEmmettTeller
(BET) analysis; this is sufficiently large to enable substrate
molecules to penetrate the periphery and interact with the
alloyed bimetallic materials [75]. The composite structures
were then studied in the hydrogenation of cinnamaldehyde,
which has multiple reactive chemical functionalities (Fig. 9)
[75]. Using the bimetallic PdAu materials, selective
generation of cinnamyl alcohol occurred, over the preferred
phenylpropanol product [75]. This suggests that beyond
reactivity enhancements, PdAu materials may prove to be
selective for production of the desired organic materials.
This is a distinct advantage when multiple reactive groups
are present in a single substrate, which can significantly
complicate organic synthetic designs.
Similar results to those observed with the PdAu DENs
have been demonstrated using other materials that employ
traditional surface passivants bound to the nanoparticle
surface. For instance, Mizukoshi et al. have prepared
Au@Pd nanoparticles using a sonochemical-based ap-
proach employing a mixed solution of Au
3+
and Pd
2+
ions
[76]. UVvisible evidence indicates the Au
3+
ions are
reduced first, followed by the Pd
2+
species, producing an
Au@Pd coreshell motif, which results in particles of
8.0 nm in diameter, as measured via TEM analysis [76].
The diameter of the core and thickness of the shell can be
controlled by modifying the Pd:Au ratio used during
synthesis, adding a valuable level of synthetic control to
the system. When using these materials for the catalytic
hydrogenation of 4-pentenoic acid, maximum activity was
observed for the coreshell materials containing 20% Au
[76]. At this value, the materials are about 2.5 times more
efficient than their monometallic counterparts. Furthermore,
analogous Au@Pd materials with diameters of 6.32.2 nm
prepared by Okitsu et al. using similar synthetic techniques
1152 M.R. Knecht, D.B. Pacardo
demonstrated maximum hydrogenation activity of cyclo-
hexene at 30 mol% Au [77]. Interestingly, all of these
values correspond quite well to those observed with the
DEN-based bimetallic materials. This suggests that a Pd:Au
atomic ratio of 75:25 may be optimum for catalytic
reaction efficiency.
Similar to the studies that employed the biomimetic
monometallic Pd particles, the cage-like protein ferritin has
been used to synthesize bimetallic PdAu nanoparticles
within its hollow core. The recombinant L-chain apo-
ferritin from horse liver is used as a templating agent for
fabrication of alloy PdAu nanoparticles using the co-
reduction method and sequential-reduction methods were
used to produce the coreshell structure [67]. Using this
method both Au@Pd and Pd@Au materials could be
prepared. In the production of PdAu alloy nanoparticles,
Pd
2+
ions are loaded into the ferritin core that has been pre-
treated with Au
3+
ions [67]. Upon reduction of the metal ions
using NaBH
4
, the two species randomly mix to produce the
alloyed structure with an average metallic diameter of 2.2
0.2 nm. Interestingly, based upon the coordination environ-
ment of the two different metal ions prior to reduction, which
has been characterized using inductively coupled plasma-
optical emission spectroscopy to determine the number of Au
or Pd ions bound to the protein, fewer Au atoms than Pd
atoms can be introduced to the ferritin-based structure. As a
result, the ability to control the Pd:Au ratio in the final
materials can be limited by the number of ions able to be
bound by the protein. Coreshell nanoparticles, on the other
hand, are synthesized by loading the Au
3+
ions into the
protein; reduction then produces monometallic Au nano-
particles. Pd
2+
ions are then introduced and subsequently
reduced on to the Au core to produce an Au@Pd structure
with an average diameter of 2.40.3 nm [67]. The inverted
Pd@Au materials were produced using identical methods;
however, the Pd core was prepared first, followed by Au
deposition. These structural characteristics were confirmed
by using high-resolution TEM and EDS methods that
confirmed the composition of the materials. The catalytic
activities of the PdAu alloy and the Au@Pd nanoparticles in
the hydrogenation of acrylamide in water were then
determined. Using these two materials, TOF values of 510,
310, and 1300 were observed for monometallic Pd nano-
particles prepared in the ferritin protein, PdAu alloys, and the
Au@Pd coreshell materials, respectively [67]. Such results
were quite surprising, especially for the alloyed materials.
The TOF value for the Au@Pd nanoparticles is 2.5-times
higher than that for the Pd-monometallic species, suggesting
that the catalytic activity of Pd on the surface of the
nanoshell is enhanced by introduction of the Au core
because of the synergistic effect of bimetallic nanoparticles;
however, the TOF values for the PdAu alloys are substan-
tially lower than those for both the monometallic Pd and
Au@Pd materials [67]. This result is quite surprising, based
upon the above described alloy results and the expected
electronic configuration.
In a unique peptide-based study, Slocik and Naik have
demonstrated the production of PdAu nanomaterials in
which small Pd nanoparticles are deposited on the surface
of a larger, core, Au nanoparticle [26]. To achieve this
structure, a bifunctional peptide was generated that is able
to passivate the surface of Au nanoparticles while display-
ing a sequence known to bind Pd nanomaterials. Upon
reduction of Pd
4+
in the presence of the peptide-capped Au
particles, nucleation and growth of the surface Pd materials
occurs. This unique method was extensively studied using
UVvisible and high-resolution TEM analysis. The main
structural detail was confirmed by use of TEM, which is
able to display the Pd particles attached to the Au surface.
High-resolution images of the materials reveal the lattice
fringes associated with both systems in intimate contact, to
which electronic changes may be possible between the two
materials [26]. The catalytic capabilities of this system were
studied by hydrogenation of 3-buten-1-ol; this revealed a
TOF value of 1016 mol H
2
(mol Pdh)
1
, significantly
better than for comparable Pd nanoparticles, for which the
TOF value was 501.5 mol H
2
(mol Pdh)
1
[26]. These
results suggest that not all Pd atoms must be directly
attached to the Au atoms, but that the structure must
incorporate both species to result in the electronic shifts
required to drive the increased TOFs.
Conclusions
In conclusion, current research has demonstrated a prom-
ising future for use of nanomaterials as catalysts for olefin
hydrogenation and numerous other reactions, including C-
coupling and CO oxidation. A full comparison of the
catalytic TOF values of the materials presented herein,
compared with values for the Pd structure and substrate, is
presented in the Supporting Information. As future research
progresses towards understanding the catalytic reactivity of
these materials, it is important to remember that the
structure of the nanocatalyst is critically important to its
reactivity. Complete materials characterization at the atomic
level is required to fully understand the motifs responsible
for driving this unique functionality, especially for the
smallest of particles. Unfortunately, achieving full charac-
terization of the materials to ascertain motifs that drive the
reaction can be rather problematic. This is caused by the
lack of sufficient analytical methods to address the nano-
scale regime with a high degree of resolution. Development
of new techniques is required to observe such complicated
architectural changes, and these are undoubtedly being
developed. Many groups are actively researching this area,
Employing high-resolution materials characterization to understand the effects of Pd nanoparticle structure 1153
bringing hope to the design and fabrication of next-
generation catalytic materials that are functional under
ambient conditions with increased reaction rates.
Acknowledgements Acknowledgment is made to the Donors of the
American Chemical Society Petroleum Research Fund for partial
support of this research. Additionally, support from the University of
Kentucky is also acknowledged.
References
1. Scott RWJ, Wilson OM, Crooks RM (2005) J Phys Chem B
109:692704
2. Pacardo DB, Sethi M, Jones SE, Naik RR, Knecht MR (2009)
ACS Nano 3:12881296
3. Astruc D (2007) Inorg Chem 46:18841894
4. Astruc D, Lu F, Aranzaes JR (2005) Angew Chem Int Ed
44:78527872
5. Narayanan R, El-Sayed MA (2005) J Phys Chem B 109:
1266312676
6. Daniel M-C, Astruc D (2004) Chem Rev 104:293346
7. Xia Y, Xiong Y, Lim B, Skrabalak SE (2009) Angew Chem Int Ed
48:60103
8. Lim B, Jiang M, Tao J, Camargo PHC, Zhu Y, Xia Y (2009) Adv
Funct Mater 19:189200
9. Dickerson MB, Sandhage KH, Naik RR (2008) Chem Rev
108:49354978
10. Niu Y, Yeung LK, Crooks RM (2001) J Am Chem Soc 123:
68406846
11. Oh S-K, Niu Y, Crooks RM (2005) Langmuir 21:1020910213
12. Ueno T, Suzuki M, Goto T, Matsumoto T, Nagayama K,
Watanabe Y (2004) Angew Chem Int Ed 43:25272530
13. Pyrz WD, Buttrey DJ (2008) Langmuir 24:1135011360
14. Glasner D, Frenkel AI (2007) XAFS 13 Proc Int Conf X-ray
Absorpt Fine Struct 882:746748
15. Knecht MR, Weir MG, Frenkel AI, Crooks RM (2008) Chem
Mater 20:10191028
16. Knecht MR, Weir MG, Myers VS, Pyrz WD, Ye H, Petkov V,
Buttrey DJ, Frenkel AI, Crooks RM (2008) Chem Mater 20:
52185228
17. Myers VS, Frenkel AI, Crooks RM (2009) Chem Mater 21:
48244829
18. Petkov V, Bedford N, Knecht MR, WeirMG, Crooks RM,
Tang W, Henkelman G, Frenkel AI (2008) J Phys Chem C
Accepted, in press
19. Weir MG, Knecht MR, Frenkel AI, Crooks RM (2010) Langmuir
26:11371146
20. Frenkel AI, Hills CW, Nuzzo RG (2001) J Phys Chem B
105:1268912703
21. Suryanarayana C, Norton MG (1998) X-ray diffraction: a practical
approach. Plenum Press, New York
22. Garcia-Martinez J, Lezutekong R, Crooks RM (2005) J Am Chem
Soc 127:50975103
23. Scott RWJ, Wilson OM, Oh SK, Kenik EA, Crooks RM (2004) J
Am Chem Soc 126:1558315591
24. Diallo AK, Ornelas C, Salmon L, Aranzaes JR, Astruc D (2007)
Angew Chem Int Ed 46:86448648
25. Slocik JM, Govorov AO, Naik RR (2008) Angew Chem Int Ed
47:53355339
26. Slocik JM, Naik RR (2006) Adv Mater 18:19881992
27. Bhattacharjee S, Bruening ML (2008) Langmuir 24:29162920
28. Bhattacharjee S, Dotzauer DM, Bruening ML (2009) J Am Chem
Soc 131:36013610
29. Kidambi S, Bruening ML (2005) Chem Mater 17:301307
30. Karousis N, Tsotsou G-E, Evangelista F, Rudolf P, Ragoussis N,
Tagmatarchis N (2008) J Phys Chem C 112:1346313469
31. Wilson OM, Knecht MR, Garcia-Martinez JC, Crooks RM (2006)
J Am Chem Soc 128:45104511
32. Narayanan R, El-Sayed MA (2005) Langmuir 21:20272033
33. Ye H, Crooks RM (2005) J Am Chem Soc 127:49304934
34. Narayanan R, El-Sayed M (2003) J Am Chem Soc 125:83408347
35. Toshima N, Yonezawa T (1998) New J Chem 22:11791201
36. Maxted EB (2009) Catalytic hydrogenation and reduction. Read
Books, New York
37. Scott RWJ, Datye AK, Crooks RM (2003) J Am Chem Soc
125:37083709
38. Zhao M, Crooks RM (1999) Angew Chem Int Ed 38:364366
39. Kidambi S, Dai J, Li J, Bruening ML (2004) J Am Chem Soc
126:26582659
40. Bratlie KM, Lee H, Komvopoulos K, Yang P, Somorjai GA
(2007) Nano Lett 7:30973101
41. Tsung C-K, Kuhn JN, Huang W, Aliaga C, Hung L-I, Somorjai
GA, Yang P (2009) J Am Chem Soc 131:58165822
42. King JS, Wittstock A, Biener J, Kucheyev SO, Wang YM,
Baumann TF, Giri SK, Hamza AV, Baeumer M, Bent SF (2008)
Nano Lett 8:24052409
43. Yan W, Mahurin SM, Pan Z, Overbury SH, Dai S (2005) J Am
Chem Soc 127:1048010481
44. Yan W, Brown S, Pan Z, Mahurin SM, Overbury SH, Dai S
(2006) Angew Chem Int Ed 45:36143618
45. Chen M, Goodman DW (2008) Chem Soc Rev 37:18601870,
and references therein
46. Scott RWJ, Ye H, Henriquez RR, Crooks RM (2003) Chem Mater
15:38733878
47. Zamborini FP, Gross SM, Murray RW (2001) Langmuir 17:481488
48. Ksar F, Surendran G, Ramos L, Keita B, Nadjo L, Prouzet E,
Beaunier P, Hagge A, Audonnet F, Remita H (2009) Chem Mater
21:16121617
49. Xiong Y, Cai H, Wiley BJ, Wang J, Kim MJ, Xia Y (2007) J Am
Chem Soc 129:36653675
50. Son SU, Jang Y, Yoon KY, Kang E, Hyeon T (2004) Nano Lett
4:11471151
51. Garcia-Martinez JC, Scott RWJ, Crooks RM (2003) J Am Chem
Soc 125:1119011191
52. Lippard SJ, Berg JM (1994) Principals of bioinorganic chemistry.
University Science Books, Mill Valley, California
53. Jakhmola A, Pacardo DB, Knecht MR (2009) J Mater Chem.
accepted
54. Wang D, Luo H, Kou R, Gil MP, Xiao S, Golub VO, Yang Z,
Brinker CJ, Lu Y (2004) Angew Chem Int Ed 43:61696173
55. Song Y, Garcia RM, Dorin RM, Wang H, Qiu Y, Coker EN, Steen
WA, Miller JE, Shelnutt JA (2007) Nano Lett 7:36503655
56. Frechet JMJ, Tomalia DA (2002) Dendrimers and other dendritic
molecules. Wiley, New York
57. Jansen JFGA, de Brabander-van den Berg EMM, Meijer EW
(1994) Science 266:12261229
58. Knecht MR, Crooks RM (2007) New J Chem 31:13491353
59. Knecht MR, Garcia-Martinez JC, Crooks RM (2006) Chem Mater
18:50395044
60. Lang H, Maldonado S, Stevenson KJ, Chandler BD (2004) J Am
Chem Soc 126:1294912956
61. Lang H, May RA, Iversen BL, Chandler BD (2003) J Am Chem
Soc 125:1483214836
62. Zhao M, Sun L, Crooks RM (1998) J Am Chem Soc 120:
48774878
63. Gomez MV, Guerra J, Velders AH, Crooks RM (2009) J Am
Chem Soc 131:341350
64. Gomez MV, Guerra J, Myers VS, Crooks RM, Velders AH (2009)
J Am Chem Soc 131:1463414635
1154 M.R. Knecht, D.B. Pacardo
65. Mizugaki T, Murata M, Fukubayashi S, Mitsudome T, Jitsukawa
K, Kaneda K (2008) Chem Commun 241243
66. Uchida M, Klem MT, Allen M, Suci P, Flenniken M, Gillitzer E,
Varpness Z, Liepold LO, Young M, Douglas T (2007) Adv Mater
19:10251042
67. Suzuki M, Abe M, Ueno T, AbeS, Goto T, Toda Y, Akita T,
Yamada Y, Watanabe Y (2009) Chem Commun 48714873
68. Chun YS, Shin JY, Song CE, Lee S-G (2008) Chem Commun
942944
69. Tessonnier JP, Pesant L, Ehret G, Ledoux MJ, Pham-Huu C
(2008) Appl Catal A 288:203210
70. Ye X-R, Lin Y, Wang C, Engelhard MH, Wang Y, Wai CM (2004)
J Mater Chem 14:908913
71. Tessonnier J-P, Pesant L, Ehret G, Ledoux MJ, Pham-Huu C
(2005) Appl Catal A 288:203210
72. He J, Ichinose I, Kunitake T, Nakao A, Shiraishi Y, Toshima N
(2003) J Am Chem Soc 125:1103411040
73. Scott RWJ, Sivadinarayana C, Wilson OM, Yan Z, Goodman DW,
Crooks RM (2005) J Am Chem Soc 127:13801381
74. Dash P, Dehm NA, Scott RWJ (2008) J Mol Catal A 286:114119
75. Prvulescu VI, Prvulescu V, Endruschat U, Filoti G, Wagner FE,
Kubel C, Richards R (2006) Chem Eur J 12:23432357
76. Mizukoshi Y, Fujumoto T, Nagata Y, Oshima R, Maeda Y (2000)
J Phys Chem B 104:60286032
77. Okitsu K, Murakami M, Tanabe S, Matsumoto H (2000) Chem
Lett 13361337
Employing high-resolution materials characterization to understand the effects of Pd nanoparticle structure 1155

Das könnte Ihnen auch gefallen