Carbon nanotube-supported Pt-based bimetallic catalysts prepared by a microwave-assisted polyol reduction method and their catalytic applications in the selective hydrogenation

Article abstract of DOI:10.1016/j.jcat.2010.09.021

Platinum-based bimetallic catalysts supported on multi-walled carbon nanotube (CNT) were prepared by a facile microwave-assisted polyol reduction method (MAPR) and chemically probed in the selective hydrogenation of cinnamaldehyde. These bimetallic catalysts outperformed the catalysts prepared by the conventional impregnation method. CNT surface properties, selection of solvent and transition metal promoter were identified to be of great significance in tuning the catalytic performance. The best catalytic results were obtained on the Fe- and Co-modified Pt catalysts in the presence of ethyl acetate as the solvent after removing the oxygen-containing groups from CNT surfaces. Electrochemistry measurements including cyclic voltammetry and CO stripping, competitive hydrogenations of toluene and benzene, X-ray diffraction, X-ray photoelectron spectroscopy and transmission electron microscopy were employed to gain insights into the physicochemical properties of catalysts. Both catalytic reactions and characterizations revealed that the MAPR method afforded close contact between promoters and platinum active sites, leading to superior catalytic properties.

Full text of DOI:10.1016/j.jcat.2010.09.021

Journal of Catalysis 276 (2010) 314–326
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Carbon nanotube-supported Pt-based bimetallic catalysts prepared by
a microwave-assisted polyol reduction method and their catalytic applications
in the selective hydrogenation
Zhen Guo a, Yuanting Chen a, Lusi Li a, Xiaoming Wang b, Gary L. Haller b, Yanhui Yang a,
⇑
a School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore 637459, Singapore
b Department of Chemical Engineering, Yale University, New Haven, CT 06520, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 23 July 2010
Revised 21 September 2010
Accepted 22 September 2010
Available online 20 October 2010
Platinum-based bimetallic catalysts supported on multi-walled carbon nanotube (CNT) were prepared by
a facile microwave-assisted polyol reduction method (MAPR) and chemically probed in the selective
hydrogenation of cinnamaldehyde. These bimetallic catalysts outperformed the catalysts prepared by
the conventional impregnation method. CNT surface properties, selection of solvent and transition metal
promoter were identified to be of great significance in tuning the catalytic performance. The best catalytic
results were obtained on the Fe- and Co-modified Pt catalysts in the presence of ethyl acetate as the sol-
vent after removing the oxygen-containing groups from CNT surfaces. Electrochemistry measurements
including cyclic voltammetry and CO stripping, competitive hydrogenations of toluene and benzene, X-
ray diffraction, X-ray photoelectron spectroscopy and transmission electron microscopy were employed
to gain insights into the physicochemical properties of catalysts. Both catalytic reactions and character-
izations revealed that the MAPR method afforded close contact between promoters and platinum active
sites, leading to superior catalytic properties.
Keywords:
Bimetallic catalyst
Selective hydrogenation
Cinnamaldehyde
Close contact
Solvent effect
Carbon nanotubes
Ó 2010 Elsevier Inc. All rights reserved.
1. Introduction
performance of metallic nanoparticles may also be drastically
changed by modifying the surface chemistry of the support, e.g.,
Selective hydrogenation of unsaturated carbonyl compounds
is of paramount importance in pharmaceutical and fragrance
industries [1]. Selective hydrogenation of cinnamaldehyde (CALD)
to cinnamal alcohol (CALC) is a representative example of such
reactions. [2]. Nevertheless, hydrogenation of CALD forms other
by-products such as hydrocinnamaldehyde (HALD) and hydrocinn-
amal alcohol (HALC) due to the thermodynamically favorable
the activity of Pt nanoparticles was enhanced by a factor of 25 after
removing the surface oxygen-containing groups from a carbon
nanofiber (CNF) support [8]. Another promising approach to en-
hance the catalytic properties of Pt nanoparticles is to introduce
a second metal (or metal oxide) to form bimetallic catalysts. It
has been reported that metal promoters, such as Ga [8], Co [9]
and Ru [10], play a vital role in CALD hydrogenation in the pres-
ence of supported Pt nanoparticles due to the synergetic interac-
tions between Pt active sites and promoters.
In addition to support and promoter, other factors, e.g., the
preparation method, may also affect the outcome of selective
hydrogenation. Several methodologies have been developed to
synthesize Pt-based bimetallic catalysts. Although the procedures
of ion-exchange and impregnation (IMP) methods are simple, the
activities of as-synthesized catalysts are usually poorer when com-
pared to the catalysts prepared by more sophisticated approaches,
such as homogeneous deposition precipitation (HDP) and reduc-
tive deposition precipitation (RDP) [8]. For the RDP method, hydro-
gen is adsorbed on the pre-reduced parent metallic catalyst,
followed by depositing promoters on the pre-formed metal nano-
particles via the reduction of promoter precursors by the adsorbed
hydrogen [11]. The successive reduction steps in RDP result in
close contact between promoters and main active centers, which
hydrogenation of C@C double bonds of a,b-unsaturated aldehydes
[2,3]. In many cases, by-products such as acetals and other uniden-
tified high molecular weight compounds may also be produced in
large quantity [2,4]. Research efforts have been devoted to improve
the selectivity toward the desired unsaturated alcohol CALC. Sup-
ported platinum nanoparticles have been extensively investigated
and appear to be the most effective catalysts for the selective
hydrogenation of CALD [1,5,6].
Recent studies have focused on the effects of supports and pro-
moters on the activity and selectivity of Pt metallic nanoparticles
in CALD selective hydrogenation. Different support materials may
lead to distinctive hydrogenation results due to the specific inter-
actions between support and Pt nanoparticles [7]. The catalytic
⇑
Corresponding author. Fax: +65 6794 7553.
E-mail address: yhyang@ntu.edu.sg (Y. Yang).
0021-9517/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2010.09.021

Z. Guo et al. / Journal of Catalysis 276 (2010) 314–326
315
affords favorable catalytic properties in selective hydrogenation
[8]. Recently, a novel preparation method, i.e., microwave-assisted
polyol reduction (denoted as MAPR) technique has been proposed
to synthesize Pt-based electro-catalysts for fuel cells [12,13].
Microwave irradiation allows uniform heating of the metal precur-
sors to a high temperature within a few seconds, leading to a short-
er crystallization time and more homogenous nucleation when
compared to conventional preparation methods [13].
MAS-II). The slurry was agitated by a magnetic stirrer, and the tem-
perature inside the flask was monitored by an infrared sensor. The
suspension was heated to 438 K in 0.5 min and kept at the same
temperature for 1.5 min. After cooling to room temperature, the
powder was filtrated and washed with deionized water, followed
by drying at 373 K in vacuum. The as-synthesized catalyst was de-
noted as Pt/CNT and the expected Pt loading is 5 wt.% and con-
firmed by inductively coupled plasma (ICP) measurements. Pt/
CNT was pretreated at various temperatures (673, 773, 873 and
973 K) in a nitrogen flow (20 mL/min) for 2 h to remove the oxy-
gen-containing groups from CNT surfaces. The resulting catalysts
were denoted as Pt/CNT-673, Pt/CNT-773, Pt/CNT-873 and Pt/
CNT-973.
The preparations of bimetallic catalysts were similar to that of
Pt/CNT but in the presence of both Pt and transition metal precur-
sors. The Pt loading was kept at 5 wt.%, and the molar ratio of Pt to
promoter was 1:1. All the as-synthesized bimetallic catalysts were
pretreated at 973 K in nitrogen and were denoted as PtMn/CNT-
973, PtFe/CNT-973, PtCo/CNT-973, PtNi/CNT-973, PtCu/CNT-973,
PtZn/CNT-973 and PtSn/CNT-973. For comparison, Pt/CNT catalysts
promoted with Fe or Co were also prepared by a conventional IMP
method (denoted as PtFe/CNT-973-IMP and PtCo/CNT-973-IMP).
Aqueous solutions of H2PtCl6ꢀ6H2O and Fe(NO3)3ꢀ9H2O (Co(N-
O3)2ꢀ6H2O) were impregnated onto the oxidized CNT followed by
drying under ambient conditions overnight. The impregnated cat-
alysts were pre-reduced in a hydrogen flow (50 mL/min) at 473 K
for 2 h and subsequently exposed to a nitrogen flow (20 mL/min)
at 973 K for another 2 h.
In this contribution, we apply this facile MAPR method to pre-
pare Pt-based bimetallic catalysts with high activity and selectivity
for CALD hydrogenation, superior to those catalysts synthesized by
the conventional IMP method. The effects on the catalytic perfor-
mance of as-prepared bimetallic catalysts due to the pretreatment
of multi-walled carbon nanotube (CNT) support surfaces, solvents
and transition metal (metal oxide) promoters are examined. In
addition to the conventional characterization techniques such as
X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS)
and transmission electron microscopy (TEM), competitive hydro-
genations of toluene and benzene are employed to chemically
probe the electron density of Pt nanoparticles. We also apply the
electrochemistry measurements including cyclic voltammetry
(CV) and CO stripping to identify the nature of as-prepared bime-
tallic catalysts. It is revealed that different from IMP method, MAPR
retains the hydrophobic surface properties of CNT and affords close
contact between promoters and Pt nanoparticles, e.g., metal–metal
electronic interaction, which significantly improves the catalytic
performance of Pt active sites. Interestingly, other than generally
adopted solvents such as alcohols [1,2] and alkanes [14], the as-
prepared catalysts in this study showed remarkably high activity
and selectivity in the presence of ethyl acetate (EA) as the solvent.
2.3. Selective hydrogenation of CALD
Selective hydrogenation of CALD was carried out in a stainless
steel reactor with Teflon liner (Parr 4950, controller 4843). A cer-
tain amount of catalyst (0.01 g), 0.2 g of CALD (1.5 mmol) and
7.25 mL of solvent were added into the reactor. Tetradecane
2. Materials and methods
2.1. Chemicals
(38 lL, 0.146 mmol) was added as an internal standard. The air res-
H2PtCl6ꢀ6H2O (>37.5% Pt basis, Sigma–Aldrich), Mn(NO3)2ꢀ4H2O
(>97%, Sigma–Aldrich), Fe(NO3)3ꢀ9H2O (>98%, Sigma–Aldrich),
Co(NO3)2ꢀ6H2O (>98%, Sigma–Aldrich), Ni(NO3)2ꢀ6H2O (>98.5%,
Sigma–Aldrich), Cu(NO3)2ꢀ2.5H2O (98%, Sigma–Aldrich), Zn(NO3)2ꢀ
6H2O (>98%, Sigma–Aldrich), SnCl2 (98%, Sigma–Aldrich), HNO3
(65%, Fluka), KOH (99.99%, Sigma–Aldrich), CaH2 (95%, Sigma–Al-
drich), ethylene glycol (EG, 99.5%, Sinopharm. Chemical Reagent),
Nafion solution (5% in isopropanol and water, Sigma–Aldrich),
CALD (>99%, Aldrich), tetradecane (>99.5%, Fluka), cyclohexane
(99.5%, Sigma–Aldrich), 1-butanol (>99%, Sigma), ethanol (>99.5%,
Sigma–Aldrich), 2-propanol (>99%, Sigma), EA (>99.5%, Acros),
1,4-dioxane (99.8%, Sigma–Aldrich), benzene (>99.9%, Sigma–Al-
drich) and toluene (>99.9%, Sigma–Aldrich) were used as received
without any further purification.
idue inside the reactor was expelled by pressurizing and releasing
hydrogen several times. The reaction was performed at 333 K un-
der 10 atm of hydrogen for either 1 or 0.5 h depending on the cat-
alyst activity. In order to eliminate the external mass transfer
limitation, the effect of stirring speed on CALD conversion was
studied. The results show that CALD conversion is constant when
the stirring speed is above 600 rpm, denoting a kinetics-controlled
regime (see Supporting information Fig. S-1). Thus, the stirring
speed was set at 600 rpm for all the experiments in this study.
After the reaction, the catalyst powder was filtered off, and the fil-
trate was analyzed using a gas chromatograph (Agilent, 6890 N)
equipped with a HP-5 column (Agilent) and flame ionization detec-
tor (FID). The by-products were identified by a gas chromatogra-
phy–mass spectrometry (GC–MS, Agilent, GC 6890 N, MS 5973
inert).
2.2. Synthesis
The pristine CNT (>95%, Cnano) was treated with concentrated
and diluted nitric acids successively following the procedures as
described by Lordi et al. [15]. The concentrated nitric acid removed
the amorphous carbonaceous and metallic impurities; the diluted
nitric acid slowly oxidized the purified CNT to create abundant sur-
face functional groups, which facilitated the uniform deposition of
metal precursors. To deposit Pt precursors onto functionalized
2.4. Characterizations
Powder XRD patterns were recorded on a Bruker AXS D8Focus
diffractometer using a Ni filtered Cu K
a radiation (k = 0.15406
1 nm), operated at 40 kV and 40 mA. Diffraction data were col-
lected between 20° and 80° (2h) with a resolution of 0.02° (2h).
TEM measurements were taken on a JEOL JEM-2010 operated at
200 kV. The samples were suspended in ethanol and dried on holey
carbon-coated Cu grid. XPS measurements were taken on a VG
CNT, 0.1 g of CNT was immersed in 722
l
L of H2PtCl6ꢀ6H2O aque-
ous solution (20 mg/mL) and dried at 373 K in vacuum. EG
(40 mL) was added to the above-mentioned Pt–CNT composite fol-
lowed by sonicating for 10 min to afford a homogenous suspen-
sion. The said suspension was transferred into a three-neck flask
with a condenser and placed in a microwave reactor (Sineo,
Escalab 250 spectrometer equipped with an Al anode (Al Ka =
1846.6 eV). The background pressure in the analysis chamber
was lower than 1 ꢁ 10ꢂ7 Pa. Measurements were taken using 20
eV pass energy, 0.1 eV step and 0.15 min dwelling time. Binding

316
Z. Guo et al. / Journal of Catalysis 276 (2010) 314–326
with a HP-PONA column (Agilent) and FID detector. All the reaction
conversions were kept below 10%.
energy correction was carried out using the C1s peak of adventi-
tious C at 284.6 eV.
CV and CO stripping voltammetries were carried out in a 1.0 M
KOH solution at 50 mV sꢂ1 (Princeton Applied Research, VersaSTAT
3 Potentiostat/Galvanostat). The working electrode was prepared
3. Results
by dropping 30 lL of the catalyst ink onto a glassy carbon elec-
3.1. Effects of CNT surface oxygen-containing groups
trode. The ink was prepared by ultrasonically mixing 2 mg of cat-
alyst sample with 3 mL of 0.025 wt.% Nafion in ethanol solution.
Pt foil and Hg/HgO (1.0 M KOH) electrodes were used as the coun-
ter and reference electrodes, respectively. All the potentials in the
present study are given vs. the Hg/HgO (1.0 M KOH) electrode. The
potentials of CV test ranged between ꢂ0.8 and 0.3 V. CO stripping
was performed as follows: after purging the solution with nitrogen
for 20 min, gaseous CO was bubbled for 15 min to form the CO
adlayer on the catalysts while maintaining the potential at
ꢂ0.8 V. Excess CO in the solution was expelled by purging with
nitrogen. CO stripping voltammetry was also recorded in a range
between ꢂ0.8 and 0.3 V.
Catalyst support may influence hydrogenation via steric or elec-
tronic interaction, which is usually contributed by the surface
properties of support material [3]. The surface oxygen-containing
groups on a carbonaceous support including carboxylic, carbonyl
and hydroxyl moieties are initially indispensable to enhance the
loading and dispersion of metal precursors [13]. Interestingly, re-
moval of these oxygen-containing groups from CNF [18] surfaces
by a high-temperature treatment after loading metal particles
remarkably increases both selectivity and activity in hydrogena-
tion. Toebes et al. reported the correlation between the surface
oxygen density and the performance of CALD hydrogenation over
a CNF-supported Pt catalyst [19]. They suggested that the oxy-
gen-containing groups could be effectively removed from carbona-
ceous surfaces when the annealing temperature was higher than
573 K. Herein, the effects of CNT surface oxygen-containing groups
on hydrogenation performances over the catalysts prepared by
MAPR were examined. Pt/CNT catalysts were pretreated in a tem-
perature ranged from 673 to 973 K in nitrogen.
Various techniques, including XPS, Boehm titration, thermo-
gravimetric analysis (TGA) and chemical derivatization, have been
employed to characterize the functional groups bonded on CNF
[19] and CNT [20] surfaces. In this study, the density of surface
oxygen-containing groups on CNT was related to their electro-
chemical properties. CV measurements were taken on Pt/CNT cat-
alysts, and the results are shown in Fig. 1a. As the pretreatment
temperature increases, a remarkable decrease in the area enclosed
by CV curves, i.e., the background of CV signal is observed. Similar
result is also obtained for pure CNT before and after high-temper-
ature annealing in an inert atmosphere (see Supporting informa-
tion Fig. S-2). The decrease in area enclosed by CV curves has
been reported by Lawrence et al. [21]. In their study, CV profiles
of CNT prepared by different methods (including arc discharge
Competitive hydrogenations of toluene and benzene were car-
ried out to probe the electronic properties of Pt nanoparticles. By
measuring the relative hydrogenation rates of toluene and ben-
zene, an equilibrium adsorption constant (KT/B) can be derived
from Eq. (1), in which PT/PB is the ratio of toluene to benzene par-
tial pressures, r0B is the reaction rate of pure benzene, rB is the reac-
tion rate of benzene at PT/PB [16]. KT/B is temperature dependent,
and its value is related to the electron deficiency of Pt nanoparti-
cles; the larger the KT/B value, the more electron deficient is the
Pt [17].
rB0=rB ¼ 1 þ KT=BðPT=PBÞ
ð1Þ
The procedures for toluene and benzene hydrogenations were sim-
ilar to that reported by Poondi and Albert Vannice [16]. A liquid
chromatography pump (Shimadzu, LC-20AT) introduced toluene
and benzene into a mixer maintained at 423 K. The ratios of tolu-
ene/benzene ranged from 0 to 0.8. The gasified aromatic mixture
was carried by a H2 flow (40 mL/min) and passed through 0.02 g
of catalyst pre-loaded in a down-flow quartz reactor. The reactor
temperature was maintained at 308 K. The effluent gas was ana-
lyzed using an online gas chromatograph (Agilent, 6890 N) fitted
Fig. 1. (a) Cyclic voltammograms (CVs) of Pt/CNT catalysts in 1 M KOH solution with a scan rate of 50 mV sꢂ1; (b) magnification of the circled area in (a) showing shift of Pt–O
reduction peaks.

Z. Guo et al. / Journal of Catalysis 276 (2010) 314–326
317
CNT, i.e., the hydrophobicity of support material plays a more
important role in this particular reduction step.
CO stripping voltammetry is measured to characterize the cata-
lysts with and without high-temperature pretreatment. In CO
stripping, CO initially adsorbed on Pt nanoparticles will be oxidized
to CO2 at a negative potential along with the positively sweeping
potential of the working electrode. The potential for CO oxidization
relies on the electrochemical properties and chemical environment
of Pt nanoparticles. In addition, compared to CV measurement, CO
stripping is a more feasible method to determine the electrochem-
ical active surface (EAS) areas of Pt nanoparticles when Pt loading
is low, due to the low electrical signals attributed to the desorption
of hydrogen atoms bonded on Pt surfaces and the uncertainty
introduced by the background current subtraction [26]. Fig. 2
shows the CO stripping voltammograms of catalysts pretreated un-
der different temperatures. The main peak potential for CO oxidi-
zation remains constant for all the catalysts, whereas the onset
potential, at which CO stripping current attains 10% of the current
at main peak potential [26], negatively shifts when the pretreat-
ment temperature increases, suggesting that CO is more easily oxi-
dized on catalyst surfaces with fewer oxygen-containing groups. It
was reported that the oxidization of CO adsorbed on Pt–Ru nano-
particles supported on pristine CNT was easier when compared
to the partially oxidized CNT, and the promoted CO elimination
was assisted by the high conductivity of pristine CNT [27]. Vu
et al. reported that the conductivity of a pristine CNT pellet is 1.5
times higher than that of CNT pretreated with HNO3 [10]. Thus,
the negative shift of CO oxidization peak is contributed by the im-
proved conductivity of CNT after removing the surface oxygen-
containing groups. The EAS areas of Pt nanoparticles were calcu-
lated based on the CO stripping voltammetry, assuming that the
oxidation of a CO monolayer adsorbed on 1 cm2 of Pt requires
Fig. 2. CO stripping voltammograms of Pt/CNT catalysts treated with different
temperatures. The left and right dash lines indicate the onset and peak potentials of
Pt/CNT, respectively.
and chemical vapor deposition) were assessed, and they suggested
that a larger quantity of redox active sites (e.g., quinone) on CNT
surfaces led to a larger capacitance and CV background. A similar
trend was also observed on the carbon black pretreated in HNO3
with different concentrations and pretreatment time durations
[22]. Boiling carbon black in a concentrated HNO3 solution for
18 h resulted in the elimination of surface oxygen-containing
groups (such as quinone and hydroquinone) and consequently a
small area enclosed by the CV curves. Therefore, the decrease in
CV background upon high-temperature annealing observed in this
study suggests the removal of oxygen-containing groups from CNT
surfaces.
In these CV profiles, redox peak signals at ca. ꢂ0.7 and ꢂ0.5 V
are contributed by the adsorption and desorption of hydrogen
atoms bonded on Pt surface [23]. These peaks become less pro-
nounced at higher annealing temperatures, implying the decrease
in Pt surfaces areas, i.e., enlargement of the Pt particle sizes. Due
to the low Pt loading, the platinum oxide (Pt–O) reduction peaks
[24] are only discernable for Pt/CNT-773, Pt/CNT-873 and Pt/
CNT-973 with low backgrounds as shown in Fig. 1b, which is the
magnification of circled area in Fig. 1a. The reduction potential de-
pends on the thermodynamics of metal oxide, kinetics of all in-
volved reduction steps and the electrochemical environment
[25]. In an alkaline solution, Pt–O reduction follows the steps [23]:
420 lC [26]. The results show that the CO active surface areas of
Pt sites decrease (see Table 1), implying the sintering of small Pt
nanoparticles at high temperatures. These observations were fur-
ther evidenced by XRD and TEM measurements.
Fig. 3a shows the XRD patterns of Pt/CNT catalysts pretreated in
a nitrogen flow under different temperatures. The strong (0 0 2)
diffraction of hexagonal graphite at 2h of 25.8° suggests that the
graphite structure of CNT is retained after oxidizing CNT in
HNO3, depositing Pt precursors under microwave irradiation and
subsequently pretreating at high temperatures [28]. The diffraction
peaks at 2h of 39.7°, 46.2° and 67.4° are attributed to the cubic
platinum metal structure [29]. The average diameters of Pt nano-
particles (calculated based on the full width at half maximum of
Pt(1 1 1) reflection using Scherrer’s equation [28]) increase from
3 to 5 nm upon the high-temperature pretreatment, as listed in Ta-
ble 1. The CO stripping and XRD results are in good agreement with
direct TEM observations on Pt/CNT and Pt/CNT-973 as shown in
Fig. 4a and b. Pt/CNT shows a mean particle size at around
3.7 nm with a narrow size distribution. A noticeable particle size
increment is observed, and the particle size distribution is substan-
tially wider after pretreating Pt/CNT at 973 K. The aggregation of Pt
nanoparticles caused by the high-temperature pretreatment in this
study is more noticeable than that of the carbon-supported Pt
nanoparticles prepared by either IMP or HDP method [10,19]. For
MAPR, Pt is reduced in the liquid phase, the nucleation and growth
of Pt nanoparticles may take place either in the EG solution or on
the CNT surfaces. Although the latter case is preferable due to
the ionic interaction between CNT surface groups and Pt precursors
[30], it is still arbitrary to preclude the possibility that Pt particles
initially form in EG followed by adsorbing onto CNT surfaces, as
nanosized noble metal particles can be formed in EG solutions in
the absence of any support [12]. Meanwhile, the nucleation and
growth of Pt nanoparticles occur primarily on CNT surface for
IMP or HDP approach. It was reported that CNT-supported metal
Pt ꢂ O þ H2O þ eꢂ ! Pt ꢂ OH þ OHꢂ
Pt ꢂ OH þ eꢂ ! Pt þ OHꢂ
ð2Þ
ð3Þ
The Pt–O reduction potential decreases with increasing pretreat-
ment temperature, implying that low density of surface oxygen-
containing groups hampers the Pt–O reduction. Removal of surface
oxygen increases the hydrophobicity of CNT surfaces, which hinders
the contact of water with CNT support, resulting in a slow Pt–O
reduction. Although a low density of surface oxygen-containing
groups leads to a high conductivity of CNT [10], which may enhance
the current flow in response to the applied potential (i.e., accelerate
the diffusion of electrons to Pt particles) and result in a positive
shift of the Pt–O reduction potential, the water accessibility to

318
Z. Guo et al. / Journal of Catalysis 276 (2010) 314–326
Table 1
Hydrogenation results of CALD over Pt/CNT with different temperature treatments by using 2-propanol as solvent.
Catalysts
Particle sizea (nm)
EASb (m2 gꢂ1
)
Conversion (%)
Yield of products (mol.%)
TOFd (hꢂ1
)
CALC
HALD
HALC
Othersc
Pt/CNT
3.1
4.5
4.9
5.2
5.1
54.0
31.0
29.4
22.6
23.0
62.7
37.2
38.6
53.5
49.6
0
0
0
43.5
55.1
2.4
5.2
7.6
7.1
4.6
0
0
0
0
0
97.6
94.8
92.4
49.4
40.3
1213
1253
1371
2473
2288
Pt/CNT-673
Pt/CNT-773
Pt/CNT-873
Pt/CNT-973
a
b
c
Particle sizes were calculated from XRD patterns of Pt based on Scherrer’s equation.
EAS areas were deduced from data of CO stripping.
Based on the characterization of GC–MS, the other by-products include 1-(3-propoxyprop-1-enyl)benzene, cinnamyl formate, cinnamic acid, benzyl cinnamate, 4,4-
diphenylcyclohexa-1,5-dienyl acetate and some condensation products which cannot be identified by our GC–MS due to their large molecular weights.
d
TOF is defined as the number of converted CALD molecules in 1 h over one active site; the number of active sites (Pt atoms exposed on the surface of particles) is
determined by EAS value. Reaction time is 1 h.
Fig. 3. (a) XRD patterns of Pt/CNT treated with different temperatures; (b) XRD patterns of PtFe/CNT-973, PtCo/CNT-973, Pt/CNT-973, PtFe/CNT-973-IMP and PtCo/CNT-973-
IMP.
nanoparticles can catalyze the localized gasification of carbon
when the sample was pretreated at a high temperature, giving rise
to the formation of nanoscaled ridges and cavities (docking sta-
tions) on CNT surfaces, which may play an anchoring role in facil-
itating the immobilization of metal nanoparticles [31]. Therefore,
the mobility of Pt nanoparticles on CNT surfaces prepared by MAPR
is higher than that of CNT-supported Pt nanoparticles synthesized
by IMP or HDP, resulting in a remarkable increase in the Pt nano-
particle size under a high-temperature pretreatment. Nevertheless,
an optimized utilization of Pt active sites may be achieved under
such circumstances because the interaction between support and
Pt nanoparticles is minimized to a large extent [32,33].
To compare the catalytic properties of Pt/CNT pretreated at dif-
ferent temperatures, selective hydrogenation of CALD was con-
ducted in 2-propanol (a commonly used solvent), and the results
are listed in Table 1. For Pt/CNT pretreated at relatively low tem-
peratures (673 and 773 K), CALD conversion dramatically de-
creases when compared to the pristine Pt/CNT, which may be
attributed to the increased Pt particle size, i.e., diminishing of the
available Pt active sites evidenced by EAS area measurements. Fur-
ther increasing the annealing temperature leads to the elevation of
conversion though the Pt particle size slightly increases. The CNT
tion between aryl ring of CALD and the basic p sites on CNT sur-
faces [10]. Therefore, the local concentration of CALD on CNT
surfaces increases, resulting in an enhanced catalytic activity (as
shown in Scheme 1a). The turnover frequency (TOF, calculated
based on EAS areas) monotonically increases with the pretreat-
ment temperature, suggesting the beneficial effect of removing
surface oxygen from CNT support. TEM image shows that the Pt
nanoparticle sizes remain essentially constant after the hydrogena-
tion reaction (see Fig. 4c).
The effect of pretreatment temperature on selectivities is also
distinct. When the catalysts supported on CNT with relatively high
densities of surface oxygen-containing groups (Pt/CNT, Pt/CNT-673
and Pt/CNT-773) are tested, the product distributions are domi-
nated by undesired compounds which are not derived from CALD
hydrogenation but from side reactions, such as aldol condensation,
oxidization and esterification [2,19]. These by-products have been
identified using GC–MS and listed in the footnotes of Table 1. Toe-
bes et al. suggested that these side reactions were acid-catalyzed;
removing oxygen-containing groups from CNT surfaces decreased
the acidity of CNT surfaces, which suppressed the formation of
these by-products [19]. Unlike IMP and HDP preparation methods,
in which the surface oxygen-containing groups on carbon supports
may be partially removed after pre-reducing the catalyst in H2 at a
surface with a low oxygen density is beneficial for the p–p interac-

Z. Guo et al. / Journal of Catalysis 276 (2010) 314–326
319
Fig. 4. TEM images and particle size distributions of: (a) Pt/CNT, (b) Pt/CNT-973, (c) Recycled Pt/CNT-973 after hydrogenation of CALD in 2-propanol at 333 K under 10 atm of
hydrogen for 1 h; (d) PtFe/CNT-973, (e) PtCo/CNT-973, (f) PtFe/CNT-973-IMP and (g) PtCo/CNT-973-IMP, respectively.

320
Z. Guo et al. / Journal of Catalysis 276 (2010) 314–326
Scheme 1. (a) Hydrogenation of CALD on oxygen-containing CNT (left) and oxygen-free CNT (right) supported catalysts; (b) two hydrogenation pathways in EA; (c)
hydrogenation of CALD on bimetallic catalysts synthesized using IMP (left) or MAPR (right) methods.
high temperature, the catalysts prepared by MAPR method were
reduced in the liquid phase at a relatively low temperature. There-
fore, compared to the catalyst prepared by IMP method, more oxy-
gen-containing groups remain on CNT surfaces (evidenced by CV
measurements of monometallic Pt/CNT and the corresponding
impregnated catalyst Pt/CNT-IMP, as shown in Supporting infor-
mation Fig. S-3), which may seriously impede the hydrogenation
steps due to the acid sites on the CNT support surfaces.
Kinetic studies of CALD hydrogenation were investigated, and
the results are shown in Fig. 5. At the initial stage of reaction (with-
in 5 min), Pt/CNT-973 outperforms Pt/CNT-773, regardless of the
similar amount of by-products formed over both catalysts. The
superior activity of Pt/CNT-973 can be attributed to the removal
of oxygen-containing groups from CNT surfaces, which enhances
CNT-773, constantly producing a large amount of by-products
caused by the acid sites on CNT surfaces may exacerbate the cata-
lyst deactivation. These results reveal that side reactions dominate
the conversion of CALD over Pt/CNT-773, while the competition
between hydrogenation and side reactions is observed over Pt/
CNT-973.
Reasonable selectivity toward hydrogenation is obtained over
Pt/CNT-873 and Pt/CNT-973, and the major hydrogenation product
is CALC. These results further suggest that elimination of surface
oxygen from CNT can facilitate the hydrogenation steps, especially
the hydrogenation of C@O bond. Vu et al. suggested that the con-
ductivity of CNT was improved by removing electronegative oxy-
gen atoms from the CNT surface [10], which was further proved
by the negatively shifted onset potential for CO oxidation in this
study. The increased conductivity of CNT may enhance the electron
transfer from CNT to metal active sites [10]. The enriched electron
density surrounding the metallic nanoparticles reduces the proba-
bility of C@C bond coordinated to Pt active sites due to the elec-
tronic repulsion, thereby increases the selectivity toward CALC
[10]. It is worth mentioning that the product distribution is differ-
ent from the reported results over the Pt nanoparticles supported
on CNF, in which the CALC selectivity decreased with the density
of surface oxygen-containing groups on CNF, although both activ-
ity and selectivity toward total hydrogenation product were im-
proved [19]. The Pt nanoparticle size for Pt/CNT-973 (5 nm) is
the
p–p interaction between CALD and CNT. Nevertheless, the
reaction rate over Pt/CNT-973 decreases gradually, which may be
due to deactivation of catalytic sites caused by the deposition of
carbonaceous by-products. The formation of carbonaceous depos-
its in the liquid catalytic system is referred as fouling which has
been extensively investigated by Hajek et al. [34,35]. In a low-tem-
perature reaction, the transformation of hydrocarbons does not oc-
cur, and the carbonaceous deposits are mostly non-desorbed [35].
However, further formation of by-products is inhibited to a large
extent as a result of the hydrogenation steps and the rapid de-
crease in CALD concentration, as shown in Fig. 5. In case of Pt/

Z. Guo et al. / Journal of Catalysis 276 (2010) 314–326
321
Fig. 5. Kinetic studies on hydrogenation of CALD over Pt/CNT-773 and Pt/CNT-973 by using either 2-propanol or EA as a solvent. (a) Conversion of CALD as a function of time;
(b) percentage of CALD converted to by-products as a function of time.
larger than that of the catalyst reported by Toebes et al. (1–3 nm)
[19], and it has been generally accepted that a slightly larger Pt
nanoparticle size leads to a higher selectivity toward CALC due to
the steric effect [3,18]. Furthermore, the curved graphite surface
of CNT is beneficial to the charge transfer when compared to other
carbon structures [36]. CNF is of a fishbone-type structure with the
edges of graphite planes exposed at the surface [19]; the enhance-
ment of electron transfer by removing surface oxygen is less pro-
nounced for CNF due to the orientation of graphite layers and the
hydrogen atoms at the external edges.
are more polar than reactants [37]. The polarities of solvents in this
study are of the following sequence: cyclohexane < EA < 1-buta-
nol < ethanol < 2-propanol < dioxane. Both activity and selectivity
toward CALC increase with solvent polarity due to the higher
polarity of CALC than that of CALD [38]. No reaction took place in
dioxane under the current hydrogenation conditions, implying that
the catalyst may be poisoned by dioxane due to the solvent adsorp-
tion. A similar result was reported in a previous work; dioxane was
strongly adsorbed on Ru surfaces, which was attributed to the
interaction between surface noble metal atoms and the unshared
electron pair donated by oxygen atoms of dioxane [39]. It was also
suggested that Pt-based catalysts interacted more strongly with
solvent than Ru-based catalysts [39]. Significant improvements in
catalytic activity and selectivity were obtained when using EA as
the solvent. EA outperforms other solvents and the TOF increases
by a factor of 4 and 2 when compared to two commonly used sol-
vents, cyclohexane and 2-propanol, respectively.
3.2. Effects of solvents
Compared to the large number of reports on designing effective
catalysts, investigations into the solvent effect on CALD hydrogena-
tion are limited [3]. The solvent plays a crucial role in this particu-
lar hydrogenation reaction; some solvents such as alcohols may be
involved in the side reactions [19], and hydrocarbon solvents usu-
ally give a poor activity [3]. In this study, six solvents were exam-
ined for CALD hydrogenation over Pt/CNT-973, and the results are
listed in Table 2. The hydrogenation activities and selectivities
show a strong dependence on the selection of solvent. It was sug-
gested that a polar solvent enhances the formation of products that
Solubility of hydrogen in the solvent may play a significant role
in this catalytic reaction. The hydrogen solubility in EA is higher
than any other alcohol employed in this study [40], which may
partially contribute to the improved activity in EA. Although the
hydrogen solubility in alcohols is proportional to the molecular
weight of alcohols [40], a poor catalytic activity of hydrogenation
Table 2
Hydrogenation results of CALD over Pt/CNT-973 in different solvents.
Solvents
Conversion (%)
Yield of products (mol.%)
TOFb (hꢂ1
)
CALC
HALD
HALC
Othersa
Cyclohexane
Ethyl acetate
1-Butanol
Ethanol
2-Propanol
1,4-Dioxane
23.4
89.8
39.0
64.9
49.6
0
17.7
78.8
34.0
47.6
55.1
–
16.9
6.8
6.6
5.2
4.6
–
0
14.2
0
0
0
–
65.4
0.2
59.4
47.2
40.3
–
1062
4079
1771
2948
2288
–
a
Based on the characterization by GC–MS, the other by-products include 1-(3-propoxyprop-1-enyl)benzene, 1-((E)-3-propoxyprop-1-enyl)benzene, 1-(3-ethoxyhexa-1,5-
dienyl)benzene, cinnamyl formate, cinnamic acid, benzyl cinnamate, 4,4-diphenylcyclohexa-1,5-dienyl acetate, 2-(hydroxy(phenyl)methyl)hexanal and some condensation
products which cannot be identified by our MS due to their large molecular weights.
b
TOF is defined as the number of converted CALD molecules in 1 h over one active site; the number of active sites (Pt atoms exposed on the surface of particles) is
determined by EAS value. Reaction time is 1 h.

322
Z. Guo et al. / Journal of Catalysis 276 (2010) 314–326
is observed in 1-butanol, which may be due to the low polarity of
1-butanol and the catalyst deactivation derived from the high
selectivities toward by-products. High pressure increases the solu-
bility of hydrogen in EA [40]. In this study, CALD hydrogenation in
EA under different hydrogen pressures shows that high pressure
leads to a high conversion along with an improved selectivity to-
ward HALC rather than CALC (see Table 3). On the other hand,
hydrogenation of CALD in EA over bared CNT after high-tempera-
ture (973 K) annealing under the same reaction conditions
(10 atm of hydrogen, 333 K for 1 h) shows a conversion of 6.0%
and 100% selectivity toward HALD. Similar results (5.9% of CALD
conversion and 100% of HALD selectivity) are obtained when the
reaction is performed in the absence of any catalyst, implying the
lack of catalytic capability of CNT in this particular reaction. These
results suggest that CALD hydrogenation in EA may undergo a dis-
tinct reaction pathway besides the direct hydrogenation of CALD
by H atoms adsorbed on Pt nanoparticles.
Aiming at understanding the role of a solvent in the liquid phase
reaction, the chemistry involving common organic solvents, e.g.,
esters, ethers and alcohols on Pt surfaces in the presence of H2
has been investigated at room temperature using a reflection–
absorption infrared spectroscopy (RAIRS) [41]. A small amount of
CO adsorbed on Pt was detected due to the decomposition of eth-
anol. No CO adsorption was found when Pt was immersed in EA li-
quid film. As CO is notorious for its poison effect on Pt-based
catalysts, the high stability of EA on Pt surfaces may partially ex-
plain why Pt/CNT-973 exhibits a remarkably high activity in EA.
Investigations into EA hydrogenation revealed the formation of
hemiacetal bonded on noble metal surfaces as an intermediate
[42]. Herein, it is tentatively proposed that the hemiacetal interme-
diate may also form and participate in the CALD hydrogenation. As
illustrated in Scheme 1b, when EA adsorbs on Pt nanoparticle, dis-
sociated hydrogen adsorbed on Pt transfers to the carbonyl group
of EA. The oxygen of carbonyl group subsequently bonds to a va-
cant Pt site to form the hemiacetal intermediates [42]. Further
reduction of EA is not preferable because it occurs only at a sub-
stantially higher temperature [42]. Hydrogen transfer may occur
from the unstable hemiacetal intermediate to CALD, leading to
the transfer hydrogenation between CALD and the intermediate.
This proposed mechanism may explain why both selectivity and
activity remarkably increase in EA. Firstly, the competitive adsorp-
tion of CALD and EA solvent on Pt surfaces is not necessary for
hydrogenation because the intermediate can transfer hydrogen to
CALD. This may be the major factor accounting for the high TOF
in EA. Secondly, it is also not surprising to observe that the selec-
tivities toward by-products contributed by side reactions are ex-
tremely low (0.2%) due to the favorable hydrogenation of CALD
in EA. The C@C double bond hydrogenation may also be suppressed
due to the steric resistance between the intermediate and aryl ring
of CALD, resulting in a high selectivity toward CALC. In contrast to
the hydrogenation of CALD over Pt/CNT-973 in 2-propanol, de-
tailed studies on the time-dependent catalytic behavior of Pt/
CNT-973 in EA show that the CALD conversion is controlled by
the hydrogenation steps (see Fig. 5). By changing the solvent from
2-propanol to EA, the reaction rate remarkably increases, which
can be attributed to the enhanced hydrogenation rates and sup-
pressed formation of by-products. Thus, the possibility of deposit-
ing carbonaceous by-products on catalytic sites is minimized.
As the commercial EA contains less than 0.2 wt.% of water, the
influence of water content is also investigated by adding different
amount of water in pre-dried EA (which is prepared by distillation
of commercial EA on CaH2). The results in Table 4 show that the
activity over Pt/CNT-973 increases in the presence of a little
amount of water in EA (<0.2 wt.%), followed by a rapid decrease
due to the excess water content. Similar trend was also found in
a ethanol–water system, in which the best result was obtained in
the presence of 10 wt.% of water [43]. The remarkable difference
is that even a trace amount of water (<0.2 wt.%) can influence
the catalytic behavior dramatically in this study. Details of water
effect are still under investigation.
3.3. Effects of promoters and preparation methods
Adding a second metal (or metal oxide) to a Pt-based catalyst as
the promoter may influence the catalytic performance of this par-
ticular catalyst in CALD hydrogenation [3]. In this work, introduc-
ing the promoter was simply achieved by mixing Pt and promoter
precursor solutions with CNT support prior to the microwave-as-
sisted reduction. Several transition metals were screened, and
the results are summarized in Table 5. Most of the promoters
(Mn, Ni, Cu, Zn and Sn) show negative effects on CALD hydrogena-
tion, whereas introducing Fe and Co improves CALC selectivity and
CALD conversion, respectively. Fe- and Co-promoted Pt catalysts
were also prepared by an IMP method and subsequently pretreated
in nitrogen under a high temperature. As shown in Table 5, cata-
lytic conversions dramatically decrease upon the impregnation of
Fe and Co. The CALC selectivity remains constant for PtCo/CNT-
973-IMP, and it decreases in the presence of impregnated Fe.
XRD patterns of these catalysts are shown in Fig. 3b; diffraction
peaks assigned to the cubic platinum metal structure are clearly
discernable. No reflections from Fe or Co metallic and oxidized spe-
cies can be detected due to the low Fe and Co contents (1.47 and
1.56 wt.% for Fe and Co, respectively). The broadening of diffraction
peaks for PtCo/CNT-973-IMP suggests the formation of relatively
small Pt particles. Pt particle sizes for these samples are listed in
Table 5; no clear correlation between particle size and preparation
method can be derived. The mean particle sizes and size distribu-
tions were further measured by TEM observations, as shown in
Fig. 4d–g. For the bimetallic catalysts synthesized by MAPR, the
incorporation of Co affords a marginally larger Pt particle size
Table 3
Hydrogenation results of CALD over Pt/CNT-973 under different hydrogen pressures
in EA at 333 K, reaction time: 1 h.
Table 4
Hydrogenation results of CALD over Pt/CNT-973 under 10 atm of hydrogen at 333 K in
EA with different water contents, reaction time: 1 h.
Pressure (atm)
Conversion (%)
Yield of products (mol.%)
CALC
HALD
HALC
Othersa
Weight ratio of water
(wt.%)
Conversion
(%)
Yield of products (mol.%)
5
71.5
89.8
98.4
99.8
100
81.0
78.8
64.5
40.4
23.9
19.2
8.8
6.8
4.0
1.4
0.8
0.5
10.0
14.2
31.4
58.0
75.2
80.1
0.2
0.2
0.1
0.2
0.1
0.2
CALC HALD HALC Others
10
15
20
30
40
0.0a
<0.2b
0.2
0.4
0.8
66.6
89.8
76.3
73.2
63.0
51.2
83.4
78.8
82.6
82.4
84.9
84.4
8.8
6.8
7.9
8.9
10.7
10.9
7.6
14.2
9.3
8.6
4.3
0.2
0.2
0.2
0.1
0.1
0.1
100
a
Based on the characterization by GC–MS, the other by-products include cin-
1.6
4.6
namyl formate, cinnamic acid, 4,4-diphenylcyclohexa-1,5-dienyl acetate and some
condensation products which cannot be identified by our MS due to their large
molecular weights.
a
The moisture-free EA was prepared by distillation of commercial EA over CaH2.
Based on the catalogue of commercial EA.
b

Z. Guo et al. / Journal of Catalysis 276 (2010) 314–326
323
Table 5
Hydrogenation results of CALD over Pt-based bimetallic catalysts in EA.
Catalysts
Particle size (nm)
EAS (mꢂ2 gꢂ1
)
Conversion (%)
Yield of products (mol.%)
TOFb (hꢂ1
)
CALC
HALD
HALC
Othersa
Pt/CNT-973
5.1
4.5
4.5
5.0
5.3
5.1
4.8
5.2
4.9
3.6
23.0
–
15.1
19.0
–
–
–
–
28.6
35.7
59.8
45.6
56.5
66.2
53.4
38.1
41.7
20.0
11.9
17.4
80.9
71.6
87.7
73.8
76.5
59.5
77.1
50.1
66.3
80.4
12.8
28.0
11.5
18.4
20.4
40.4
21.3
49.8
33.6
19.5
6.2
0.2
0.7
7.7
3.0
0
1.4
0
0
0.1
0.2
0.1
0.1
0.1
0.1
0.2
0.1
0.1
0.1
5433
–
7818
7280
–
–
–
–
869
1018
PtMn/CNT-973
PtFe/CNT-973
PtCo/CNT-973
PtNi/CNT-973
PtCu/CNT-973
PtZn/CNT-973
PtSn/CNT-973
PtFe/CNT-973-IMP
PtCo/CNT-973-IMP
0
a
Based on the characterization by GC–MS, the other by-products include cinnamyl formate, cinnamic acid, 4,4-diphenylcyclohexa-1,5-dienyl acetate and some con-
densation products which cannot be identified by our MS due to their large molecular weighs.
b
TOF is defined as the number of converted CALD molecules in 0.5 h over one active site; the number of active sites (Pt atoms exposed on the surface of particles) is
determined by EAS value. Reaction time is 0.5 h.
and broader size distribution when compared to the Pt particles
promoted by Fe. Nevertheless, the trend inverses for the impreg-
nated catalysts; small particles with a narrow size distribution
are formed on the Co-modified catalyst. These results imply that
the interaction between Pt and promoter may be different due to
the distinctive preparation strategies.
Most of the promoter precursors are nitrate salts, and the pro-
moters should be present in the form of oxides if they are located
on Pt surface and exposed to air. The available surface areas of Pt
particles (represented by EAS areas derived from CO stripping
measurements) are summarized in Table 5. In general, the EAS area
is inversely proportional to the particle size of Pt monometallic cat-
alyst. The same trend also applies to PtFe/CNT-973-IMP and PtCo/
CNT-973-IMP; both impregnated catalysts have smaller particle
sizes and larger EAS areas than Pt/CNT-973. Although the particle
sizes of PtCo/CNT-973 (5.0 nm) and PtFe/CNT-973 (4.5 nm) are
smaller than that of Pt/CNT-973, the active surface areas of Pt for
these two catalysts are also lower than that of Pt/CNT-973 (see Ta-
ble 5). We suggest that the small surface area is contributed by the
coverage of Fe or Co oxides on Pt nanoparticles, namely a close
contact between promoters and catalytic active centers. The signif-
Fig. 6. CO stripping voltammograms of PtFe/CNT-973, PtCo/CNT-973, Pt/CNT-973,
PtFe/CNT-973-IMP and PtCo/CNT-973-IMP. The left and right dash lines indicate the
onset and peak potentials of Pt/CNT-973, respectively.
icantly high TOF suggests the superior activity of surface Pt atoms
for PtFe/CNT-973 and PtCo/CNT-973, while the relatively poorer
conversion for PtFe/CNT-973 is attributed to the lower EAS area,
i.e., fewer exposed Pt surface atoms. These observations imply that
C@O bond activation in CALD hydrogenation is enhanced by the
oxidized species attached on Pt surface as electrophilic or Lewis
sites [3] because the electropositive oxides may adsorb and acti-
vate C@O bond via the unoccupied electron pair of oxygen. For
the bimetallic catalysts prepared by IMP, promoters lack close con-
tact with Pt, and adding the second metal does not have substantial
influence on the EAS area, which can hardly contribute to the acti-
vation of C@O for hydrogenation (as illustrated in Scheme 1c). Fur-
thermore, the metal oxide layer (or small clusters) deposited on
remarkably higher than Pt/CNT-973, as shown in Fig. 6, suggesting
that impregnated Fe or Co mainly modifies the surface properties
of CNT rather than Pt nanoparticles. The promoting effect of metal
oxide on CO oxidization is based on the assumption that Pt parti-
cles are adjacent to the metal oxide promoters [45]. If the metal
oxide locates apart from Pt active sites and disperses primarily
on CNT surfaces, OHads can hardly participate in the CO oxidization
due to the segregated distribution of Pt and metal oxide clusters
(see Scheme 1c, left). The CNT conductivity may also be altered
due to the coverage of metal oxide with OHads. Therefore, the CO
oxidation is impeded by the decreased charge transfer capability
of CNT, resulting in the positive shift of oxidization potential for
impregnated catalysts.
CV profiles of these catalysts show that Pt–O reduction peak
potentials for Fe- and Co-promoted catalysts prepared by MAPR
shift negatively when compared to Pt/CNT-973, while Pt–O reduc-
tion peak potentials of corresponding impregnated bimetallic cat-
alysts either remain constant or shift positively (see Fig. 7). Metal
oxide islands on Pt nanoparticles may change the interaction be-
tween catalytic active sites and support, e.g., the electron flux from
CNT to Pt nanoparticles under certain given potential. The current
flow in response to applied potential may be diminished by metal
CNT surfaces may hinder the adsorption of CALD via p–p interac-
tion, resulting in the extremely low activities for impregnated
bimetallic catalysts.
The CO stripping measurements show that the onset potentials
for CO oxidation on PtFe/CNT-973 and PtCo/CNT-973 shift nega-
tively (see Fig. 6), implying the promoted CO oxidization. The hy-
droxyl groups generated by water dissociation can adsorb on the
surface of metal oxide species [44], and these adsorbed hydroxyls
(OHads) may be involved in CO stripping and facilitate the oxidation
of CO on a Pt surface [45]. Therefore, the negative shift of CO oxi-
dization potential further demonstrates the close contact between
metal oxide and Pt nanoparticles. On the contrary, the onset and
peak potentials of CO oxidation for the impregnated catalysts are

324
Z. Guo et al. / Journal of Catalysis 276 (2010) 314–326
active sites [10] the other is from promoter to Pt in the alloy
[7,46]. Herein, XPS was employed to investigate the electron and
oxidization states of Pt and promoters. To determine the peak po-
sition accurately, the XPS profiles were fitted by Gaussian–Lorentz-
ian sum function (The fitting details are summarized in Supporting
information Fig. S-4). As shown in Fig. 8a, Pt 4f7/2 binding energies
of impregnated catalysts shift toward higher value when compared
to those of Pt/CNT-973, suggesting that Fe or Co oxides dispersed
on CNT surfaces may hinder the electron transfer from CNT to Pt,
as donating electrons to Pt leads to the negative shift of Pt binding
energy [14]. For bimetallic catalysts prepared by the MAPR meth-
od, Pt 4f7/2 binding energies are lower than those of Pt/CNT-973,
which can be ascribed to the electron transfer from promoters to
Pt due to the formation of alloy. The XPS spectra (Fig. 8b) show that
nearly all the iron exists as metal oxide [47], and a shoulder peak at
707.85 eV is discernable for the Fe 2p spectrum of PtFe/CNT-973.
This binding energy is slightly larger than 2p3/2 binding energy of
metallic iron (706.8 eV) [47], which can be attributed to the forma-
tion of Pt–Fe alloy [48]. This shoulder peak is not obvious for PtFe/
CNT-973-IMP. In the XPS spectra of Co 2p (Fig. 8c), the main peak
centered at around 781 eV can be assigned to cobalt oxide species
[49]. The broadening of cobalt peak for PtCo/CNT-973 may be due
to the existence of metallic cobalt, which has the binding energy of
ca. 778 eV [49].
Fig. 7. Shift of Pt–O reduction peaks for PtFe/CNT-973, PtCo/CNT-973, PtFe/CNT-
973-IMP and PtCo/CNT-973-IMP, comparing to Pt/CNT-973.
In addition to the XPS measurements, an equilibrium adsorp-
tion constant of toluene and benzene (KT/B) is used to characterize
the electron density of Pt nanoparticles in the presence of promot-
ers. KT/B, which is derived from the relative hydrogenation rates of
toluene and benzene (see Eq. (1)), is only correlated with reaction
temperature and independent on other variables including particle
size and support [17]. As the electron donation is easier for toluene
compared to benzene, a larger KT/B value suggests a lower electron
density of Pt nanoparticle [17]. As shown in Fig. 9, the lower KT/B
value (3.24) of PtFe/CNT-973 when compared to that of Pt/CNT-
973 (4.28) implies that electrons are transferred from Fe to Pt in
this bimetallic catalyst. The increased electron density of Pt nano-
particles attenuates the binding energy, particularly that of C@C
bond, and favors the hydrogenation of C@O [3]. The promoted
CALC selectivity contributed by the electron transfer from iron to
platinum was also detected by the X-ray absorption measurements
oxide species enriched on Pt particles, which decreases the number
of available electrons for Pt–O reduction. For the impregnated
bimetallic catalysts, the addition of promoters mainly contributes
to the modification of support surface properties. Agglomeration
of metal oxides with surface-adsorbed OHads decreases the conduc-
tivity of CNT support (evidenced by CO stripping), which may also
impair the current flow toward Pt nanoparticles. Nevertheless, it
also makes the catalyst surface more hydrophilic and more acces-
sible to water, which facilitates the Pt–O reduction as interpreted
in Eqs. (2) and (3), i.e., leading to a less negative reduction
potential.
The existence of Co or Fe metallic phase alloyed with Pt after
reduction cannot be completely ruled out in this one-step prepara-
tion and reduction. Therefore, two kinds of electron transfer mech-
anisms should be considered: one is from CNT support to metal
Fig. 8. XPS spectra of: (a) Pt, (b) Fe and (c) Co on Pt/CNT-973, PtFe/CNT-973, PtCo/CNT-973, PtFe/CNT-973-IMP and PtCo/CNT-973-IMP.

Z. Guo et al. / Journal of Catalysis 276 (2010) 314–326
325
the close contact between promoters and Pt nanoparticles. The
concept of ‘‘close contact” has been suggested by Plomp et al.,
in which the promoter (tin or gallium) was reduced by the hydro-
gen dissociated on pre-formed Pt particles [8]. Similarly, Co-
decorated Pt nanoparticles were also synthesized by reducing
Co in a suspension of Pt particles prepared in advance; an organic
additive, 1,2-hexadecanediol, was employed as the reducing re-
agent [1]. Both examples require two successive reduction steps:
one for the formation of Pt nanoparticles and the other for the
reduction of promoters in the presence of Pt nanoparticles. In this
microwave-assisted synthesis, the bimetallic catalysts with close
contact between active sites and promoters can be prepared in
one step. The reduction potential of platinum (1.188 V) is higher
than that of iron (ꢂ0.440 V) and cobalt (ꢂ0.277 V) [50]; there is a
high probability that Pt can be readily reduced by EG prior to the
reduction of Fe or Co. It has been demonstrated that Co cannot be
reduced in the absence of Pt particles when diol was the reducing
agent [1]. Under a high temperature (433 K), EG may interact
with Pt ions, resulting in the oxidation of EG to aldehyde, glycolic,
oxalic acid or even CO2. The electrons donated from these oxida-
tion reactions lead to the reduction of Pt cations [12]. Once the Pt
particles form, the oxidation of EG may take place via an alterna-
tive or parallel reaction pathway that involves the hydrogen
abstraction from EG by platinum [12]. Thus, the hydrogen atoms
adsorbed on Pt nanoparticles may facilitate the reduction and
deposition of the second metal (Fe or Co) on Pt surfaces, which
is similar to the reduction steps in the RDP method reported by
Plomp et al. [8]. Further reductions of Pt and transition metal pre-
cursors afford the formation of alloys. Once the catalysts are ex-
posed to air, Fe or Co metal on the Pt surface will be rapidly
oxidized.
By employing the MAPR method, promoters are prone to closely
interact with Pt active sites, which are essentially evidenced by CO
stripping and CV measurements. The enrichment of promoters on
Pt catalytic sites attains the hydrophobic surface of CNT which is
beneficial to the adsorption of CALD reactants. In addition, the syn-
ergistic effects including activation of C@O bond and efficient me-
tal-to-metal electron transfer (evidenced by KT/B) remarkably
enhance the catalytic activity and selectivity due to the close con-
tact between Pt and promoter. For impregnated catalysts, promot-
ers primarily disperse on the surface of CNT, which blocks the
adsorption of CALD, resulting in low activities. Segregation of pro-
moter and Pt active site diminishes the electron transfer from pro-
moter to Pt, even hinders the electron donation from CNT to Pt. Pt
nanoparticles with low electron density cannot efficiently inhibit
the reduction of C@C bonds, resulting in a low selectivity toward
CALC.
Fig. 9. Relative benzene hydrogenation rate vs. ratio of its partial pressure at 308 K.
KT/B values of tested catalysts are slopes of lines.
on a carbon-supported Pt–Fe alloy [7]. The electron transfer from
Co to Pt is relatively poorer as evidenced by the larger KT/B value
for PtCo/CNT-973 (3.57). Less resistance toward the hydrogenation
of C@C bonds leads to the formation of more HALD and HALC,
which compromises the selectivity of CALC (see Table 5). For the
impregnated catalysts, electron transfer from Co to Pt is also de-
tected because of the lower KT/B value of PtCo/CNT-973-IMP
(3.67). The discrepancy regarding the Pt 4f binding energy may
be due to the low metal loadings. However, the electron transfer
in PtCo/CNT-973-IMP is less pronounced when compared to
PtCo/CNT-973, due to the lack of close contact between Co and
Pt. For PtFe/CNT-973-IMP, the KT/B value (4.49) is even larger than
that of Pt/CNT-973, implying that there is no electron transfer from
Fe to Pt. Additionally, Fe species enriched on the support surface
may hamper the electron transfer from CNT to Pt. The electron
deficiency of Pt nanoparticles on PtFe/CNT-973-IMP promotes
the formation of HALD and decreases the selectivity of CALC (see
Table 5).
4. Discussion
It has been demonstrated that electrochemistry measurements
including CV and CO stripping serve as complementary ap-
proaches to investigate the surface properties of CNT-supported
catalysts. Based on the characterization and chemically probed
reaction results, it can be concluded that surface oxygen-contain-
ing groups can be efficiently removed from CNT surfaces by the
high-temperature treatment. Elimination of these groups sup-
presses acid-catalyzed side reactions and enhances the adsorption
of CALD and electron transfer from CNT to Pt, resulting in sub-
stantially increased activity and selectivity to CALC. By screening
various kinds of solvents, a dramatic increase in TOF as well as
selectivity is observed when using EA as the solvent. It is pro-
posed that in addition to the direct hydrogenation by H atoms
adsorbed on Pt nanoparticles, transfer hydrogenation between
CALD and EA occurs, in which EA can be partially reduced to af-
ford a hemiacetal intermediate. Subsequent hydrogen transfer
from this intermediate to CALD contributes to the high activity
and selectivity.
5. Conclusions
In this study, Pt-based mono- and bimetallic catalysts were syn-
thesized by employing a facile and rapid MAPR method. These cat-
alysts were successfully applied to the selective hydrogenation of
CALD to CALC. The hydrogenation results highly depended on the
surface properties of the CNT support, solvent and promoter. Re-
moval of oxygen-containing groups from CNT surfaces elevated
both activity and selectivity, due to the suppressed side reactions
catalyzed by acid and enhanced electron transfer from CNT to Pt
nanoparticles. By screening solvents with different polarities, EA
was identified as the best candidate in this reaction, and the en-
hanced activity is hypothesized to result from the presence of par-
allel transfer hydrogenation steps. The catalytic properties can be
promoted by Fe and Co. Compared to IMP preparation, MAPR not
only served as a facile and time-saving synthesis methodology
but also afforded close contact between promoters and Pt nanopar-
ticles, which attains the affinity between CNT and CALD due to the
The superior activity and selectivity for CALD hydrogenation
over bimetallic catalysts synthesized by MAPR are ascribed to

326
Z. Guo et al. / Journal of Catalysis 276 (2010) 314–326
[19] M.L. Toebes, Y.H. Zhang, J. Hajek, T.A. Nijhuis, J.H. Bitter, A.J. van Dillen, D.Y.
p–p interaction and assists the metal–metal electron transfer.
These synergic effects contributed to the significantly improved
activity and high selectivity.
Murzin, D.C. Koningsberger, K.P. de Jong, J. Catal. 226 (2004) 215.
[20] K.A. Wepasnick, B.A. Smith, J.L. Bitter, D.H. Fairbrother, Anal. Bioanal. Chem.
396 (2010) 1003.
[21] N.S. Lawrence, R.P. Deo, J. Wang, Electroanal 17 (2005) 65.
[22] H. Wang, R. Cote, G. Faubert, D. Guay, J.P. Dodelet, J. Phys. Chem. B 103 (1999)
2042.
Acknowledgment
[23] C.H. Hamann, A. Hamnett, W. Vielstich, Electrochemistry, Wiley-VCH, New
York, 1998. p. 223.
[24] J.L. Cohen, D.J. Volpe, H.D. Abruna, Phys. Chem. Chem. Phys. 9 (2007) 49.
[25] F. Scholz, U. Schröder, R. Gulaboski, Electrochemistry of Immobilized Particles
and Droplets, Springer, Heidelberg, 2005. p. 79.
Funding from the Singapore Agency for Science, Technology and
Research (AꢃSTAR), SERC Grant No: 102 101 0020 in support of this
project is gratefully acknowledged.
[26] F. Maillard, M. Eikerling, O.V. Cherstiouk, S. Schreier, E. Savinova, U. Stimming,
Faraday Discuss. 125 (2004) 357.
[27] E. Yoo, T. Okada, T. Kizuka, J. Nakamura, J. Power Sources 180 (2008) 221.
[28] Z.T. Liu, C.X. Wang, Z.W. Liu, J. Lu, Appl. Catal. A – Gen. 344 (2008) 114.
[29] S. Chytil, W.R. Glomm, E.A. Blekkan, Catal. Today 147 (2009) 217.
[30] W.Z. Li, C.H. Liang, W.J. Zhou, J.S. Qiu, Z.H. Zhou, G.Q. Sun, Q. Xin, J. Phys. Chem.
B 107 (2003) 6292.
Appendix A. Supplementary materials
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2010.09.021.
[31] U.M. Graham, A. Dozier, R.A. Khatri, M.C. Bahome, L.L. Jewell, S.D. Mhlanga, N.J.
Coville, B.H. Davis, Catal. Lett. 129 (2009) 39.
References
[32] J.C. Duchet, E.M. van Oers, V.H.J. de Beer, R. Prins, J. Catal. 80 (1983) 386.
[33] Z.R. Li, Y.L. Fu, M. Jiang, T.D. Hu, T. Liu, Y.N. Xie, J. Catal. 199 (2001) 155.
[34] J. Hajek, J. Warna, D.Y. Murzin, Ind. Eng. Chem. Res. 43 (2004) 2039.
[35] J. Hajek, N. Kumar, T. Salmi, D.Y. Murzin, Catal. Today 100 (2005) 349.
[36] P.J. Britto, K.S.V. Santhanam, A. Rubio, J.A. Alonso, P.M. Ajayan, Adv. Mater. 11
(1999) 154.
[37] H. Rojas, J.L.G. Fierro, P. Reyes, J. Chil. Chem. Soc. 52 (2007) 1155.
[38] J. Hájek, N. Kumar, P. Mäki-Arvela, T. Salmi, D.Y. Murzin, J. Mol. Catal. A –
Chem. 217 (2004) 145.
[39] H. Takagi, T. Isoda, K. Kusakabe, S. Morooka, Energy Fuels 13 (1999) 1191.
[40] M.S. Wainwright, T. Ahn, D.L. Trimm, N.W. Cant, J. Chem. Eng. Data 32 (1987)
22.
[41] Z. Ma, F. Zaera, Catal. Lett. 96 (2004) 5.
[42] O.A. Ferretti, J.P. Bournonville, G. Mabilon, G. Martino, J.P. Candy, J.M. Basset, J.
Mol. Catal. 67 (1991) 283.
[43] Y.Z. Chen, S.W. Wei, K.J. Wu, Appl. Catal. A – Gen. 99 (1993) 85.
[44] Q. Yang, H. Choi, S.R. Al-Abed, D.D. Dionysiou, Appl. Catal. B – Environ. 88
(2009) 462.
[45] H. Yuan, D. Guo, X. Li, L. Yuan, W. Zhu, L. Chen, X. Qiu, Fuel Cell 9 (2009) 121.
[46] X. Wang, N. Li, L.D. Pfefferle, G.L. Haller, J. Phys. Chem. C (May) (2010),
doi:10.1021/jp102511k.
[47] P.C.J. Graat, M.A.J. Somers, Appl. Surf. Sci. 100–101 (1996) 36.
[48] T. Ma, Q. Fu, Y. Cui, Z. Zhang, Z. Wang, D. Tan, X. Bao, Chin. J. Catal. 31 (2010)
24.
[49] M. Hakamada, K. Tajima, K. Yoshimura, Y. Chino, M. Mabuchi, J. Alloy. Compd.
494 (2010) 309.
[1] S.C. Tsang, N. Cailuo, W. Oduro, A.T.S. Kong, L. Clifton, K.M.K. Yu, B. Thiebaut, J.
Cookson, P. Bishop, ACS Nano 2 (2008) 2547.
[2] X.X. Han, R.X. Zhou, B.H. Yue, X.M. Zheng, Catal. Lett. 109 (2006) 157.
[3] P. Gallezot, D. Richard, Catal. Rev. 40 (1998) 81.
[4] Y. Li, C.H. Ge, J. Zhao, R.X. Zhou, Catal. Lett. 126 (2008) 280.
[5] M.A. Vannice, B. Sen, J. Catal. 115 (1989) 65.
[6] M. Englisch, A. Jentys, J.A. Lercher, J. Catal. 166 (1997) 25.
[7] A.B. da Silva, E. Jordao, M.J. Mendes, P. Fouilloux, Appl. Catal. A – Gen. 148
(1997) 253.
[8] A.J. Plomp, D.M.P. van Asten, A.M.J. van der Eerden, P. Maki-Arvela, D.Y.
Murzin, K.P. de Jong, J.H. Bitter, J. Catal. 263 (2009) 146.
[9] W. Yu, Y. Wang, H. Liu, W. Zheng, J. Mol. Catal. A – Chem. 112 (1996) 105.
[10] H. Vu, F. Goncalves, R. Philippe, E. Lamouroux, M. Corrias, Y. Kihn, D. Plee, P.
Kalck, P. Serp, J. Catal. 240 (2006) 18.
[11] C. Carnevillier, F. Epron, P. Marecot, Appl. Catal. A – Gen. 275 (2004) 25.
[12] C. Bock, C. Paquet, M. Couillard, G.A. Botton, B.R. MacDougall, J. Am. Chem. Soc.
126 (2004) 8028.
[13] S.L. Knupp, W. Li, O. Paschos, T.M. Murray, J. Snyder, P. Haldar, Carbon 46
(2008) 1276.
[14] N. Mahata, F. Goncalves, M.F.R. Pereira, J.L. Figueiredo, Appl. Catal. A – Gen. 339
(2008) 159.
[15] V. Lordi, N. Yao, J. Wei, Chem. Mater. 13 (2001) 733.
[16] D. Poondi, M. Albert Vannice, J. Catal. 161 (1996) 742.
[17] G.L. Haller, J. Catal. 216 (2003) 12.
[50] J.A. Dean, Lange’s Handbook of Chemistry, McGraw-Hill, New York, 1999. p.
126.
[18] A.J. Plomp, H. Vuori, A.O.I. Krause, K.P. de Jong, J.H. Bitter, Appl. Catal. A – Gen.
351 (2008) 9.