Multiwalled carbon nanotubes drive the activity of metal@oxide core-shell catalysts in modular nanocomposites

Article abstract of DOI:10.1021/ja304398b

Rational nanostructure manipulation has been used to prepare nanocomposites in which multiwalled carbon nanotubes (MWCNTs) were embedded inside mesoporous layers of oxides (TiO₂, ZrO₂, or CeO₂), which in turn contained dis

Full text of DOI:10.1021/ja304398b

Article
pubs.acs.org/JACS
Multiwalled Carbon Nanotubes Drive the Activity of Metal@oxide
Core−Shell Catalysts in Modular Nanocomposites
§
†
†
‡
†
́
Matteo Cargnello, Marek Grzelczak, Benito Rodrıguez-Gonzalez, Zois Syrgiannis, Kevin Bakhmutsky,
́
Valeria La Parola,# Luis M. Liz-Marzan,‡ Raymond J. Gorte,§ Maurizio Prato,*,† and Paolo Fornasiero*,†
́
†Department of Chemical and Pharmaceutical Sciences, INSTM, Center of Excellence for Nanostructured Materials (CENMAT),
University of Trieste, Via L. Giorgieri 1, 34127 Trieste, Italy
‡Departamento de Química Física, Universidade de Vigo, 36310 Vigo, Spain
§Department of Chemical and Biomolecular Engineering, University of Pennsylvania, 311A Towne Building, 220 South 33rd Street,
Philadelphia, Pennsylvania 19104, United States
#Istituto per lo Studio dei Materiali Nanostrutturati (ISMN-CNR), Via Ugo La Malfa 153, Palermo I-90146, Italy
S
* Supporting Information
ABSTRACT: Rational nanostructure manipulation has been
used to prepare nanocomposites in which multiwalled carbon
nanotubes (MWCNTs) were embedded inside mesoporous
layers of oxides (TiO2, ZrO2, or CeO2), which in turn
contained dispersed metal nanoparticles (Pd or Pt). We show
that the MWCNTs induce the crystallization of the oxide layer
at room temperature and that the mesoporous oxide shell
allows the particles to be accessible for catalytic reactions. In
contrast to samples prepared in the absence of MWCNTs,
both the activity and the stability of core−shell catalysts is
largely enhanced, resulting in nanocomposites with remarkable performance for the water−gas-shift reaction, photocatalytic
reforming of methanol, and Suzuki coupling. The modular approach shown here demonstrates that high-performance catalytic
materials can be obtained through the precise organization of nanoscale building blocks.
thermal conductivity. CNTs can also be functionalized with
organic groups at different sites within their structure,6 thereby
modifying their solubility in organic solvents. Another
advantage is that electronic interactions between CNTs and
surrounding particles or layers can result in enhanced
performance.7,8
Three general approaches are commonly used to prepare
CNT−inorganic hybrids: (1) filling approaches, where species
are placed inside the nanotube channels; (2) ex-situ methods,
where preformed building blocks (e.g., metallic particles) are
attached to the nanotubes via chemical bonding interactions;
and (3) in situ methods, where the inorganic component is
simply synthesized in the presence of the pristine or
functionalized CNTs.5 Both of the latter approaches have
advantages and disadvantages. Ex-situ methods have the
advantage of producing materials with tailored properties
because they allow a better control on the size and the
morphology of the building blocks. With in situ approaches, it
is easier to manipulate the materials and the number of
synthetic steps is lower.
INTRODUCTION
■
The ability to build hierarchical structures by arranging different
building blocks with nanometer-scale precision is one of the
most useful aspects of nanotechnology.1,2 This concept can
produce novel materials with properties that are different from
those expected from the simple sum of the individual blocks.3
Indeed, the interactions between the constituent parts in
nanometer-scale ensembles can lead to novel electronic, optical,
or catalytic properties not available in the initial building
blocks.4 In the field of heterogeneous catalysis in particular, the
interactions between the active phase, supports, and promoters
is critical for obtaining high performance materials. These
interactions can be both electronic and geometric. Further-
more, the demands in terms of geometry and binding energy of
the active sites for different catalytic reactions can be very
different.
In particular, carbon nanotubes (CNTs) and multiwalled
carbon nanotubes (MWCNTs) have emerged as highly
beneficial building blocks for the preparation of various
composites. For example, hybrids of CNTs and inorganic
materials have received great attention for applications in gas
sensing, photovoltaics, and catalysis.5 The main advantages of
using CNTs as substrates for the formation of nanocomposites
are their large surface-to-volume ratio, excellent mechanical,
electrical, and optical properties, and good electrical and
For these reasons, composites of CNTs with catalytic metals
and metal oxides have been the subject of many studies.9−11
Received: May 7, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja304398b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Metals supported on CNTs with CeO2,12−18 TiO2,19−22 and
ZrO2 have been prepared through sol−gel, hydrothermal, or
electrodeposition procedures (see also ref 5 and references
therein). All these methods allow the production of materials
with an onion-type structure, in which the CNTs are covered
by oxide layers that in turn have metal particles on the external
surface. The resulting catalysts have shown promising proper-
ties in several applications.
Here, we describe the fabrication of MWCNTs covered with
layers of mesoporous oxides that in turn contain embedded
metal particles. As building blocks, we used oxidized
MWCNTs, functionalized metallic particles, and a metal
alkoxide, in a combination of ex-situ and in situ methods
(Scheme 1). Specifically, the resulting hierarchical nano-
tetrakis(decyloxide) were prepared according to procedures described
in detail elsewhere.25,27 Oxidized MWCNTs were prepared as follows.
Pristine MWCNTs (Nanocyl 7000) (150 mg) were dispersed in a
concn H2SO4/concn HNO3 mixture 3:1 v/v (100 mL) by sonication
(6 h at 30−50 °C) and magnetic stirring (12 h at 50 °C). The
suspension was washed six times by filtration; twice with water (250
mL), twice with NaOH (0.1 M, 250 mL), once with DMF (250 mL),
and once with THF (250 mL). Finally, MWCNTs were redispersed in
THF/DMF 5:1 v/v to a final concentration of MWCNTs of 0.5 mg
mL−1 as carbon.
General Procedure for the Preparation of MWCNTs@M/
oxide Materials. M/oxide structures were prepared by dissolving the
appropriate amount of the metal alkoxide (titanium butoxide,
zirconium butoxide, or cerium(IV) tetrakis(decyloxide)) in 25 mL of
THF. MUA-protected Pd or Pt particles (as THF solution) were then
added dropwise while stirring to avoid the precipitation of M-alkoxide
networks.25 The solution of M@oxide was then vigorously sonicated
and the THF/DMF solution of oxidized MWCNTs was added
simultaneously. A precipitate was formed almost immediately for all
the cases, leaving a colorless supernatant, which was taken as evidence
that the M@oxide structures had completely adsorbed/reacted with
the MWCNTs. A mixture of H2O (1 mL) and THF (10 mL) was
added to ensure the complete hydrolysis of the alkoxide precursor, and
the mixture was sonicated for an additional 15 min. The solids were
then recovered by centrifugation (4500 rpm for 15 min) and washed
with THF (30 mL portions) three times. The final materials were then
dried at 120 °C overnight prior to being used.
23
Scheme 1. Schematic Representation of the Preparation of
MWCNTs@M/Oxide
Characterization Techniques. TEM measurements were per-
formed on a TEM Philips EM208, using an acceleration voltage of 100
kV. Samples were prepared by drop casting from the dispersion onto a
TEM grid (200 mesh, copper, carbon only). High resolution TEM
(HRTEM) and analytical TEM studies were performed with a JEOL
JEM 2010 FEG-TEM operating at an acceleration voltage of 200 kV.
X-ray energy-dispersive spectra (EDS) were acquired using an Inca
Energy 200 TEM system from Oxford Instruments, and electron
energy loss spectroscopy (EELS) filtered TEM mappings were
performed with the help of a Quantum Gatan imaging filter (GIF).
Background subtraction was carried out prior to jump ratio mapping
overlap. The X-ray photoelectron spectroscopy (XPS) analyses were
performed with a VG Microtech ESCA 3000Multilab, equipped with a
dual Mg/Al anode. The spectra were excited by the unmonochrom-
atized, Mg Kα source (1253.6 eV) run at 14 kV and 15 mA. The
analyzer was operated in the constant analyzer energy (CAE) mode.
For the individual peak energy regions, a pass energy of 20 eV was set
across the hemispheres. Survey spectra were measured with a 50 eV
pass energy. The sample powders were analyzed as pellets, mounted
on a double-sided adhesive tape. Contact of the samples with air was
minimized during sample loading. The pressure in the analysis
chamber was of the order of 10−8 Torr during data collection. The
charging of the samples was removed by referencing all the energies to
the C 1s set at 285.1 eV, arising from the adventitious carbon. The
invariance of the peak shapes and widths at the beginning and at the
end of the analyses ensured absence of differential charging. Analyses
of the peaks were performed with the software provided by VG, based
on nonlinear least-squares fitting program using a weighted sum of
Lorentzian and Gaussian component curves after background
subtraction according to Shirley and Sherwood.28,29 Atomic
concentrations were calculated from peak intensity using the sensitivity
factors provided with the software. The binding energy values are
quoted with a precision of 0.15 eV and the atomic percentage with a
precision of 10%. UV−vis characterization was carried out with a
Cary 5000 spectrophotometer using 10 mm path length quartz
cuvettes. TGA of approximately 1 mg of each compound was recorded
on a TGA Q500 (TA Instruments) under air, by equilibrating at 100
°C, and following a ramp of 10 °C min−1 up to 1000 °C. Raman
spectra were recorded with an Invia Renishaw microspectrometer (50)
equipped with He−Ne laser at 633 nm. Powders were directly placed
onto a glass slide and at least 5 spectra per sample were recorded in
order to check the uniformity of the materials. N2 physisorption
experiments were carried out on a Micromeritics ASAP 2020C. The
composite is formed of core−shell metal@oxide particles24−26
supported on conducting MWCNTs.5 The approach is highly
flexible and versatile, allowing for the tuning of the composition
and concentration of all the components. As a demonstration of
this versatility, we prepared materials with Pd or Pt as the
metallic phase, and TiO2, ZrO2, or CeO2 as the metal-oxide
coatings. Furthermore, we show that the thickness of the oxide
layer surrounding the MWCNTs can be varied. Finally, we
demonstrate that these modular composites are active and
stable heterogeneous catalysts in several reactions. MWCNTs
dramatically promote the activity of the materials when
compared to the same metal/oxide combinations prepared in
the absence of MWCNTs, for reasons that are believed to be
both geometrical/morphological and electronic in nature.
These materials demonstrate the utility of combining several
building blocks to prepare functional nanostructures with
improved performance.
EXPERIMENTAL SECTION
■
Materials. Titanium butoxide (97%), zirconium butoxide solution
(80 wt % in 1-butanol), concentrated H2SO4 (95−97%), and
concentrated HNO3 (65%) were purchased from Sigma-Aldrich.
MWCNTs, Nanocyl 7000, were purchased from Nanocyl s.a., Belgium.
Preparation of the Building Blocks. Pd and Pt nanoparticles
protected by 11-mercaptoundecanoic acid (MUA) and cerium
B
dx.doi.org/10.1021/ja304398b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 1. Representative TEM images of MWCNTs@(1 wt %)Pd−oxide nanocomposites with titania (a−c), zirconia (d−f) and ceria (g−i) shells.
The weight concentration of MWCNTs is 5% (g), 10% (a, d, h), 15% (b, e, i), and 20% (c, f). All scale bars represent 100 nm.
samples were first degassed in vacuum at 120 °C overnight prior to N2
adsorption at liquid nitrogen temperature.
For the water−gas shift-reaction (WGSR), all catalytic tests were
conducted at atmospheric pressure. The fresh catalysts were pretreated
in flowing Ar at 40 mL min−1 for 30 min at 450 °C, after heating from
room temperature at 10 °C min−1. No other activation procedures
(e.g., reduction) were necessary since the catalyst was reduced in situ
under the WGS reaction conditions employed. Typically, 22 mg of
fresh sample were placed in a U-shaped, quartz microreactor with
internal diameter of 4 mm. The total gas flow rate under reaction
conditions was 27.2 mL min−1 to ensure a gas hourly space velocity
(GHSV) of ∼75 000 mL g−1 h−1. The feed gas for the WGS reaction
was 1.0 vol % CO and 3.0 vol % H2O, diluted in Ar. This gaseous
mixture was introduced into the reactor at 250 °C for 2 h. Aging
treatments were performed by heating the sample in the WGS
environment at temperatures up to 450 °C for 6 h. Subsequently, the
sample was cooled down to 250 °C in the WGS environment, and the
activity was measured again. All heating and cooling rates were 2 °C
min−1. Reactants and products were analyzed using a mass
spectrometer (Hyden Analytical HPR20).
Catalytic Experiments. The photocatalytic activity for H2
production was evaluated using a 250 mL Pyrex discontinuous batch
reactor with an external cooling jacket. A nominal 10 mg of catalyst
was loaded in the reactor, sonicated, and dispersed under vigorous
stirring into 80 mL of a water/methanol mixture (1/1 by volume). All
experiments were carried out at 20 °C, and 15 mL min−1 of Ar were
passed through the solution to transport the gaseous/volatile products
to a GC for quantitative analysis. A 125 W medium pressure mercury
lamp (Helios Italquartz) with Pyrex walls was used for UV−vis
excitation. Gaseous products were analyzed by GC analysis using a
thermal conductivity detector (TCD) and a flame ionization detector
(FID) for the quantification of H2 and CO2.
Suzuki coupling experiments were performed as follows. Iodoben-
zene (1.0 mmol), phenylboronic acid (1.3 mmol), NaOH (1.5 mmol),
and naphthalene (0.5 mmol as internal standard) were codissolved in 5
mL of ethanol. The mixture was stirred and heated up to 75 °C, then
the catalyst (0.05 mol % as Pd with respect to iodobenzene) was
added, at which point the reaction started. No biphenyl yield was
obtained in the absence of the catalyst. Reagent conversions and
product yields were determined by GC−MS analysis using
naphthalene as internal standard. The mixtures were analyzed using
an Agilent 7890 GC mounting a J&W DB-225 ms column (60 m, ID
0.25 mm, 250 μm) and coupled with an Agilent 5975 MS.
RESULTS AND DISCUSSION
■
The synthesis of the MWCNTs@M/oxide structures (M = Pd,
Pt; oxide = TiO2, ZrO2, CeO2) involved the use of three
building blocks: oxidized MWCNTs, metal nanoparticles
functionalized with 11-mercaptoundecanoic acid (MUA), and
C
dx.doi.org/10.1021/ja304398b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
either Ti, Zr, or Ce alkoxide (Scheme 1). The essential idea is
that the functionalized metal particles and the oxidized
MWCNTs are able to bind to the metal alkoxides due to the
presence of carboxyl groups that displace the weak alkoxide
ligands. In the first step, the M/oxide building blocks are
prepared by reacting the metal particles with the metal alkoxide.
Next, the M/oxide particles are connected to the MWCNTs by
reacting the MWCNTs with the alkoxide linkages that remain
available on the M/oxide particles. Then this supramolecular
ensemble is subjected to a controlled hydrolysis of the
remaining alkoxide groups in solution to obtain uniform
MWCNTs@M/oxide composites. The method is easy to
implement by simply placing the three building blocks in the
appropriate solvent mixtures (pure THF for combining the
metal particles and the alkoxides; THF/DMF in a 5:1 ratio to
react MWCNTs with M/oxide modules). DMF is necessary
because of the lower solubility of MWCNTs in pure THF. To
ensure proper mixing of the M/oxide particles with oxidized
MWCNTs, this step was performed under sonication. Effective
incorporation of all the components into the final materials was
confirmed by TGA (Supporting Information, Figure S1). The
characterization presented here and below is limited to Pd-
based systems but similar results were obtained for the Pt-based
nanocomposites.
Information). The oxide layer was observed to be smooth in
the case of titania and rougher for ceria, possibly because the
ceria film is composed of larger nanocrystallites (see below). In
each case, a porous architecture was formed around the
MWCNTs, a morphology that is expected to be advantageous
for catalytic applications. Interestingly, the TEM images of the
ceria-based composites demonstrated that the thickness of the
ceria layer decreased as the concentration of MWCNTs was
increased from 5 wt % to 15 wt % (Figure 1g−i), in agreement
with the increasing availability of surface area on the
MWCNTs.
A better understanding of the composite nanostructure was
obtained through HRTEM characterization of the ceria-based
materials (Figure 2). The nanocrystalline nature of the ceria
Reference samples, in which metal particles were embedded
inside oxide layers, were prepared in the manner described
above but in the absence of MWCNTs. MUA−Pd or MUA−Pt
nanoparticles were reacted with the alkoxides and the resulting
systems hydrolyzed in THF/DMF mixtures. These materials
are referred to hereafter as M@oxide. It is expected that the
metal particles are located under deeper layers with respect to
the MWCNTs composites, as previously observed elsewhere in
materials prepared using similar procedures.30 Physisorption of
N2 revealed that all the final materials were highly porous, with
surface areas of at least ∼120 m2 g−1 (Table S1, Supporting
Information). BET isotherms and BJH pore size distributions
(Figure S2, Supporting Information) of oxidized MWCNTs
show a large N2 uptake at pore sizes of ∼10 nm which can be
attributed to the filling of the MWCNTs cavities and of the
spaces in between the bundles of nanotubes.31 After deposition
of the Pd/oxide layers, this large uptake dramatically decreases,
likely because the oxide layer seals the MWCNTs caps, limiting
the filling by N2, and because the nanotubes are kept much
more separated. However, additional mesoporosity at small
pore sizes (2−5 nm) and micropores (<2 nm) are created,
which can be attributed to the presence of the oxide layer. This
mesoporosity accounts for the accessibility of metal particles, as
evidenced by the catalytic activity displayed by the nano-
composites (see later).
The composition in the final structure could be varied easily
because the solution-phase methods used to prepare the
building blocks were simple. For example, the concentration of
the building blocks in the composites could be adjusted by
modifying the initial ratios of the starting materials during the
synthesis, for example, by changing the loading of metal
nanoparticles or the MWCNTs/alkoxide ratio. The thickness of
the oxide layer could also be varied, as described below.
Figure 1 shows representative transmission electron micros-
copy (TEM) images of composites comprising 1 wt % Pd as
metal, with titania, zirconia, and ceria as the oxides. The images
indicate the presence of a uniform oxide layer covering the
MWCNTs. The oxide thickness depended on the material but
was generally 5 to 10 nm thick (Figure S3, Supporting
Figure 2. (a) Representative HRTEM image of MWCNTs(10 wt %)
@Pd(1 wt %)/CeO2 composites showing a partially covered
MWCNT. (b) Higher magnification image showing a single CeO2
crystallite on top of the ceria layer. (c) Electron diffraction pattern
corresponding to the image in panel a, showing the reflections and the
indexation on the basis of Fm3m CeO .
̅
2
film was clearly visible, with the nanotubes being covered by
small crystallites (∼4 to 5 nm, Figure 2b) of ceria, as confirmed
by electron diffraction patterns (Figure 2c). Even for the
thinnest ceria layer, the thickness of the oxide coating was at
least 6−7 nm, on average, for the sample MWCNTs(15%)@
Pd/CeO2 (Figure S3, Supporting Information). In one instance
where a partially exposed MWCNT was found (Figure 2a), the
TEM images demonstrate that the rest of the MWCNT is
located inside the ceria layer. The images also demonstrate that
the structural characteristics of MWCNTs are maintained after
the preparation of the composites, indicating that the integrity
of the MWCNTs had not been altered during the formation of
the composites. This fact is particularly important for retaining
the electronic and mechanical properties of the MWCNTs in
the final composites. A further demonstration of the core−shell
nature of the MWCNTs@Pd/CeO2 composites was provided
by EELS mapping, which showed that the carbon skeleton is
surrounded by a ceria layer (Figure 3).
Although the oxide layer was sometimes rough and
composed of small crystallites, the layers almost always
completely covered the MWCNTs. HRTEM characterization
did not reveal the presence of Pd or Pt nanoparticles. This may
be due to their small size (∼ 2 nm), low concentration (1 wt
%), or to the fact that they were completely encapsulated by
ceria, which has a rather high electron density. Indirect
evidence of the complete encapsulation of all Pd in the case
of unsupported Pd@ceria was obtained by EELS mapping.25 In
D
dx.doi.org/10.1021/ja304398b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
layer were critical for incorporating the metal particles into the
final composite. In the absence of these interactions, the metal
nanoparticles remained in the supernatant solution after the
synthesis.
Characterization by Raman spectroscopy also confirmed the
contact of the oxides with the surface of MWCNTs. In the
Raman spectra, the intensity ratio between the D band
(indicative of disorder in the structure) and the G band
(indicative of the ordered graphitic structure) is essentially
unaffected by the presence of the oxide (from 1.53 to 1.55),
suggesting that the incorporation of carboxyl groups on the
outer graphitic shell did not seriously compromise the overall
structure (Figure S5, Supporting Information). Subsequent
deposition of M/oxide particles slightly decreased the D/G
ratio (Figure S5, Supporting Information). Such “healing” of
MWCNT by an oxide shell has been recently explained by
further perturbation of outer shells with reacting oxides, leading
to a decrease in the contribution of the outer graphitic shell to
Raman fingerprint32−34 Therefore, the overall Raman signal
arises from the unmodified inner shell of the MWCNT, which
explains the decreased D/G ratio.
In the final composites, the Raman spectra also showed peaks
related to TiO2 and CeO2, indicating the formation of
crystalline oxides, prior to any thermal treatments. In the case
of ZrO2, the absence of a Raman signal for a crystalline phase of
the oxide suggests that the crystallites observed in TEM (Figure
1) might be too small and amorphous to give rise to Raman
contributions. It is noteworthy that none of the M/oxide
samples prepared in the absence of MWCNTs produced a
Figure 3. TEM image of a selected area of the sample (a) and
corresponding EELS mapping of the C and Ce content in CNTs(10
wt %)@Pd(1 wt %)/CeO2 composites (b) showing that the CNTs are
embedded inside the ceria layer. Palladium is not visible in this image
due to its very low loading.
addition, the problem of visualizing small metal particles deeply
embedded within oxide layers is well-known in the case of Pt
which has an even higher electron density than Pd.24 Indeed,
the presence of the metals in our samples was confirmed by
EDS analysis (Figure S4, Supporting Information in the case of
Pd) and XPS (see later), demonstrating that the metal−core/
oxide−shell nanoparticles were effectively incorporated into the
final product. In a parallel synthesis with nonfunctionalized
metal particles (e.g., dodecanethiol-capped Pd or Pt particles),
the metal particles were not incorporated into the nano-
composites, demonstrating that the specific interactions
between the protected metal particles and the growing oxide
Figure 4. Catalytic activity of the different combinations of materials. (a) CO conversion during WGS reaction at 250 °C over MWCNTs@M(1%)/
CeO2. (b) photocatalytic H2 production in the presence of methanol as sacrificial agent over MWCNTs@M(1%)/TiO2.
E
dx.doi.org/10.1021/ja304398b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Raman signal associated with the crystalline oxides, showing
that crystallization of the oxide layer was induced by the
presence of oxidized MWCNTs. This suggests that the
MWCNTs may exhibit a templating effect which induces the
hydrolysis of the oxide precursors and confines condensation
into small crystallites around the walls of the MWCNTs. The
low temperature formation of metal oxides is of great advantage
in catalysis since nanostructured catalysts have been shown to
exhibit enhanced reaction rates.32 The presence of nanocrystal-
line ceria was further demonstrated by UV−vis experiments
(Figure S6, Supporting Information).
X-ray photoelectron spectroscopy (XPS) demonstrated that
M/oxide and MWCNTs@M/oxide made with TiO2 and ZrO2
contain only Ti(IV) and Zr(IV) species (Supporting
Information, Table S2 and Figures S7−S8).35 By contrast, the
XPS spectra of CeO2-based materials revealed the coexistence
of Ce(III) and Ce(IV). The content of Ce3+ was appreciably
higher in the MWCNTs@Pd/CeO2 nanocomposites with
respect to the reference Pd@CeO2 sample (Supporting
Information, Table S2). It is known that formation of surface
carbonates can lead to the presence of Ce(III). However, the
higher surface area of the reference Pd@CeO2 sample with
respect to the MWCNTs@Pd/CeO2 (Supporting Information,
Table S1) suggests that the origin of the higher Ce(III)
concentration might be related to a preferential formation of
oxygen vacancies after embedding of the MWCNTs with ceria.
This observation of interactions between the ceria and the
MWCNTs agrees with previous reports of strong interactions
between ceria particles deposited on graphene oxide.36
Additionally, there was an appreciable signal for Pd in the
nanocomposites containing MWCNTs, which is noteworthy
because the Pd signal was absent in the Pd@CeO2 reference
material (Supporting Information, Figure S9).
We explored the catalytic behavior of the nanocomposites
(MWCNTs@M/oxide) and compared their activity with
reference materials (M@oxide) in selected reactions that have
important implications for energy-related applications. The
ceria-based structures were tested for the water−gas shift
(WGS) reaction,37 the titania-based nanocomposites were
investigated for the photocatalytic production of H2 using
methanol as a sacrificial agent,38 and the zirconia-based
architectures were tested for the Suzuki coupling reaction.39
It should be underlined that bare oxidized MWCNTs showed
no activity for any of the reactions that were studied under our
experimental conditions.
The WGS reaction (CO + H2O = CO2 + H2) is important
for tailoring the CO/H2 ratio in synthesis gas, and ceria-
supported group VIII metals have been shown to exhibit
significantly higher rates than the same metals on inert supports
such as Al2O3 or SiO2.40 The preparation of unsupported Pd@
CeO2 by a microemulsion procedure led to catalysts with poor
activity, apparently due to the formation of a dense ceria barrier
over the Pd that impeded the accessibility of metal particles.30
Alumina-supported Pd@ceria particles showed good initial
activity but were unstable, apparently because the ceria in the
shell becomes reduced in the WGS environment, due to redox
properties that are different from those of bulk ceria.41 In
agreement with these observations, Pd@CeO2 and Pt@CeO2
prepared in this work without the MWCNTs, as reference
samples, gave no activity at 250 °C under the conditions of our
present study (Figure 4, part a).
MWCNTs-containing samples (Figure 4, part a) and addition-
ally, unlike the alumina-supported particles,41 the MWCNT
nanocomposites demonstrated outstanding thermal stability,
even after accelerated aging under reaction conditions at 450
°C for 6 h. These results clearly indicate that the supporting
skeleton (MWCNTs vs alumina) plays an important role in the
stabilization of the catalytic performance.
The encapsulation of the Pd and Pt particles inside a thin and
porous ceria layer has also been shown to prevent metal
sintering in previous studies.26 Considering the medium-high
reaction temperature used here, the M@Ceria based systems
should be equally effective in this role.42 Therefore, the
supporting MWCNTs appear to play a major role in organizing
the core−shell units in a highly accessible nanoarchitecture.
The fact that the Pd- and Pt-based catalysts exhibit nearly the
same rates corroborates the mechanistic interpretation that the
role of the metal is to activate the CO in the reaction, while
ceria is responsible for water dissociation.37
From the photocatalytic activity standpoint, the rates for
hydrogen production on MWCNTs@M/TiO2 nanocompo-
sites, using methanol as a sacrificial agent, were constant at 8−
10 mmol h−1 g−1, with no deactivation for up to 15 h (Figure 4,
part b). MWCNTs@Pt/TiO2 materials were slightly more
active than MWCNTs@Pd/TiO2. The high rate indirectly
confirms that the titania shell must be crystalline for
photoactivity, which is somewhat surprising considering that
no thermal treatments have been applied to the titania shell.
This also indicates that the metal particles must be accessible to
protons in order to carry out the reduction to hydrogen. The
activity of the Pt-based nanocomposites showed a slight
dependence on the amount of MWCNTs present, whereas
the Pd-based ones were not affected by this parameter. The H2
evolution rate that we observed was higher than that found with
most semiconductor photocatalysts, although the use of
different conditions used for these experiments precludes a
definitive comparison of activities.43
The most interesting aspect of these results is that the
composites containing MWCNTs showed H2 evolution rates
that were 4 times higher than those of the reference Pd@TiO2
and Pt@TiO2 materials. Reference materials gave maximum
rates for H2 evolution of only up to 2.5 mmol h−1 g−1. The
higher activity of the MWCNT@M/TiO2 clearly demonstrates
the role of MWCNTs in enhancing the performance. This is
likely due to a combination of positive effects. First, the
MWCNT can delocalize the photogenerated electrons,7,22
enhancing the lifetime of the electron hole pairs and ultimately
the hydrogen evolution. Furthermore, the fact that the
mechanism of enhancement does not appear to depend on
the TiO2 thickness also suggests a direct participation of the
MWCNTs in light absorbance as a sensitizer.7
Regarding Suzuki coupling, the MWCNTs@Pd/ZrO2
materials showed high rates for biphenyl formation (Supporting
Information, Figure S10), with a quantitative yield after only a
few minutes of mild heating (75 °C). The amount and
thickness of the zirconia layer did not appear to affect the
observed activity, further demonstrating that Pd particles alone
are needed for the efficient coupling reaction. Hot filtration
tests suggested that Pd leaching is not responsible for the
observed activity, although it was not possible to completely
rule out the role of very small amounts of leached Pd as the true
active species.44 Tests performed with the reference Pd@ZrO2
sample showed much lower rates, indicating that the
organization of the particles around the MWCNTs might be
We examined the MWCNTs@ceria/Pd and MWCNTs@
ceria/Pt samples for WGS showing high activity of the
F
dx.doi.org/10.1021/ja304398b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(19) Moriguchi, I.; Hidaka, R.; Yamada, H.; Kudo, T.; Murakami, H.;
Nakashima, N. Adv. Mater. 2006, 18, 69.
(20) Eder, D.; Windle, A. H. Adv. Mater. 2008, 20, 1787.
(21) Liu, B.; Zeng, H. C. Chem. Mater. 2008, 20, 2711.
(22) Yu, Y.; Yu, J. C.; Yu, J. G.; Kwok, Y. C.; Che, Y. K.; Zhao, J. C.;
Ding, L.; Ge, W. K.; Wong, P. K. Appl. Catal., A 2005, 289, 186.
important for improving activity. Interestingly, the bare Pd
nanoparticles show a much lower activity under similar
conditions,45 implying that a synergic enhancement of the
activity is obtained through the combination of the MWCNTs
and the oxide layer.
(23) Lupo, F.; Kamalakaran, R.; Scheu, C.; Grobert, N.; Ruhle, M.
Carbon 2004, 42, 1995.
̈
ASSOCIATED CONTENT
■
S
(24) Yeung, C. M. Y.; Yu, K. M. K.; Fu, Q. J.; Thompsett, D.; Petch,
M. I.; Tsang, S. C. J. Am. Chem. Soc. 2005, 127, 18010.
(25) Cargnello, M.; Wieder, N. L.; Montini, T.; Gorte, R. J.;
Fornasiero, P. J. Am. Chem. Soc. 2010, 132, 1402.
(26) De Rogatis, L.; Cargnello, M.; Gombac, V.; Lorenzut, B.;
Montini, T.; Fornasiero, P. ChemSusChem 2010, 3, 24.
(27) Bakhmutsky, K.; Wieder, N. L.; Cargnello, M.; Galloway, B.;
Fornasiero, P.; Gorte, R. J. ChemSusChem 2012, 5, 140.
(28) Shirley, D. A. Phys. Rev. B 1972, 5, 4709.
* Supporting Information
TGA experiments; N2 adsorption−desorption isotherms and
corresponding BJH pore size distribution; representative HR-
TEM images and associated EDS profiles; Raman, UV−vis, and
XPS characterization; Suzuki cross-coupling results and addi-
tional references. This material is available free of charge via the
Internet at http://pubs.acs.org.
AUTHOR INFORMATION
(29) Sherwood, P. M. A. In Practical Surface Analysis; Briggs, D.,
Seah, M. P., Eds.; Wiley: New York, 1990; p 181.
■
Corresponding Author
(30) Cargnello, M.; Montini, T.; Polizzi, S.; Wieder, N. L.; Gorte, R.
J.; Graziani, M.; Fornasiero, P. Dalton Trans. 2010, 39, 2122.
(31) Bradley, R. H.; Cassity, K.; Andrews, R.; Meier, M.; Osbeck, S.;
Andreu, A.; Johnston, C.; Crossley, A. Appl. Surf. Sci. 2012, 258, 4835.
(32) Carrettin, S.; Concepcion, P.; Corma, A.; Lopez-Nieto, J. M.;
Puntes, V. F. Angew. Chem., Int. Ed. 2004, 43, 2538.
(33) Inoue, F.; Ando, R. A.; Corio, P. J. Raman Spectrosc. 2011, 42,
1379.
(34) Quintana, M.; Ke, X.; Van Tendeloo, G.; Meneghetti, M.;
Bittencourt, C.; Prato, M. ACS Nano 2010, 4, 6105.
(35) Camposeco, R.; Castillo, S.; Mejia, I.; Mugica, V.; Carrera, R.;
prato@units.it; pfornasiero@units.it
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Mr. Claudio Gamboz (University of Trieste) is acknowledged
for help with TEM characterization. Dr. Valentina Gombac and
Dr. Tiziano Montini (University of Trieste) are acknowledged
for help with catalytic characterization. We thank the University
of Trieste (FRA 2011 Project), INSTM, and MIUR (PRIN
contract No. 20085M27SS and Firb RBIN04HC3S) for
financial support. R.J.G. and K.B. acknowledge support from
AFOSR (MURI) Grant No. FA9550-08-1-0309.
Montoya, A.; Moran
Commun. 2012, 17, 81.
́ ́
-Pineda, M.; Navarrete, J.; Gomez, R. Catal.
(36) Joung, D.; Singh, V.; Park, S.; Schulte, A.; Seal, S.; Khondaker, S.
I. J. Phys. Chem. C 2011, 115, 24494.
(37) Bunluesin, T.; Gorte, R. J.; Graham, G. W. Appl. Catal., B 1998,
15, 107.
(38) Cargnello, M.; Gasparotto, A.; Gombac, V.; Montini, T.;
Barreca, D.; Fornasiero, P. Eur. J. Inorg. Chem. 2011, 2011, 4309.
(39) Miyaura, N.; Suzuki, A. Chem. Rev. 1995, 95, 2457.
(40) Hilaire, S.; Wang, X.; Luo, T.; Gorte, R. J.; Wagner, J. Appl.
Catal., A 2001, 215, 271.
(41) Wieder, N. L.; Cargnello, M.; Bakhmutsky, K.; Montini, T.;
Fornasiero, P.; Gorte, R. J. J. Phys. Chem. C 2011, 115, 915.
(42) Kim, J. S.; Wieder, N. L.; Abraham, A. J.; Cargnello, M.;
Fornasiero, P.; Gorte, R. J.; Vohs, J. M. J. Electrochem. Soc. 2011, 158,
B596−B600.
(43) Chen, X.; Shen, S.; Guo, L.; Mao, S. S. Chem. Rev. 2010, 110,
6503.
(44) Astruc, D. Inorg. Chem. 2007, 46, 1884.
REFERENCES
■
(1) Grzelczak, M.; Vermant, J.; Furst, E. M.; Liz-Marzan
Nano 2010, 4, 3591.
́
, L. M. ACS
(2) Murray, C. B.; Norris, D. J.; Bawendi, M. G. J. Am. Chem. Soc.
1993, 115, 8706.
(3) Nie, Z.; Petukhova, A.; Kumacheva, E. Nat. Nanotechnol. 2010, 5,
15.
(4) Murray, C. B.; Kagan, C. R.; Bawendi, M. G. Annu. Rev. Mater.
Sci. 2000, 30, 545.
(5) Eder, D. Chem. Rev. 2010, 110, 1348.
(6) Tasis, D.; Tagmatarchis, N.; Bianco, A.; Prato, M. Chem. Rev.
2006, 106, 1105.
(7) Woan, K.; Pyrgiotakis, G.; Sigmund, W. Adv. Mater. 2009, 21,
2233.
(45) Cargnello, M.; Wieder, N. L.; Canton, P.; Montini, T.;
Giambastiani, G.; Benedetti, A.; Gorte, R. J.; Fornasiero, P. Chem.
Mater. 2011, 23, 3961.
(8) Toma, F. M.; Sartorel, A.; Iurlo, M.; Carraro, M.; Parisse, P.;
Maccato, C.; Rapino, S.; Gonzalez, B. R.; Amenitsch, H.; Da Ros, T.;
Casalis, L.; Goldoni, A.; Marcaccio, M.; Scorrano, G.; Scoles, G.;
Paolucci, F.; Prato, M.; Bonchio, M. Nat. Chem. 2010, 2, 826.
(9) Wildgoose, G. G.; Banks, C. E.; Compton, R. G. Small 2006, 2,
182.
(10) Pan, X.; Bao, X. Chem. Commun. 2008, 6271.
(11) Banhart, F. Nanoscale 2009, 1, 201.
(12) Li, Y.; Ding, J.; Chen, J.; Xu, C.; Wei, B.; Liang, J.; Wu, D. Mater.
Res. Bull. 2002, 37, 313.
(13) Bai, J.; Xu, Z.; Zheng, Y. Chem. Lett. 2006, 35, 96.
(14) Zhang, D.; Yan, T.; Shi, L.; Pan, C.; Zhang, J. Appl. Surf. Sci.
2009, 255, 5789.
(15) Zhang, D.; Pan, C.; Zhang, J.; Shi, L. Mater. Lett. 2008, 62, 3821.
(16) Zhang, D.; Shi, L.; Fu, H.; Fang, J. Carbon 2006, 44, 2853.
(17) Fiala, R.; Khalakhan, I.; Matolinova, I.; Vaclavu, M.; Vorokhta,
M.; Sofer, Z.; Huber, S.; Potin, V.; Matolin, V. J. Nanosci. Nanotechnol.
2011, 11, 5062.
(18) Keyvanloo, K.; Mohamadalizadeh, A.; Towfighi, J. Appl. Catal.,
A 2012, 417−418, 53.
G
dx.doi.org/10.1021/ja304398b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX