Citethis:Phys. Chem. Chem. Phys.,2011,13 ,25902602 PAPERfulltext.calis.edu.cn/rsc/physical chemistry chemical physics/c0cp01832e.pdf · 2590 Phys. Chem. Chem. Phys., 2011, 13 , 25902602

2590 Phys. Chem. Chem. Phys., 2011, 13, 2590–2602 This journal is c the Owner Societies 2011

Cite this: Phys. Chem. Chem. Phys., 2011, 13, 2590–2602

Geometric and electronic effects on hydrogenation of cinnamaldehyde

over unsupported Pt-based nanocrystalsw

William O. Oduro,z Nick Cailuo, Kai Man K. Yu, Hongwei Yangy andShik Chi Tsang*

Received 16th September 2010, Accepted 16th December 2010

DOI: 10.1039/c0cp01832e

It is reported that catalytic hydrogenation of cinnamaldehyde to cinnamyl alcohol is a structural

sensitive reaction dependent on size and type of metal doper of unsupported platinum

nanocrystals used. Smaller sizes of platinum nanocrystals are found to give lower selectivity to

cinnamyl alcohol, which suggests the high index Pt sites are undesirable for the terminal aldehyde

hydrogenation. A plot of reaction selectivity across the first row of transition metals as dopers

gives a typical volcano shape curve, the apex of which depicts that a small level of cobalt on

platinum nanocrystals can greatly promote the reaction selectivity. The selectivity towards

cinnamyl alcohol over the cobalt doped Pt nanocrystals can reach over 99.7%, following the

optimization in reaction conditions such as temperature, pressure and substrate concentration.

Detailed studies of XRD, CO chemisorption (for FTIR), TEM, SEM, AES and XPS of the

nanostructure catalyst clearly reveal that the decorated cobalt atoms not only block the high index

sites of Pt nanocrystals (sites for Co deposition) but also exert a strong electronic influence on

reaction pathways. The d-band centre theory is invoked to explain the volcano plot of selectivity

versus metal doper.

Introduction

Hydrogenation of organic compounds plays a very important

role in chemical manufacturing processes. Among all

the hydrogenation reactions reported, the hydrogenation of

a,b-unsaturated aldehydes to their corresponding unsaturated

alcohols draws the most attention as the hydrogenation of

these compounds is of both fundamental and industrial

importance.1,2 There has been much recent interest in

synthesizing uniform metallic and bimetallic nanocrystals as

new heterogeneous catalysts because of appropriate metal

particle size and optimised geometric and/or electronic effects

in metallic and bimetallic nanostructures which may allow the

nanocrystals with tuneable catalytic properties 3,4 to overcome

thermodynamic favourable CQC hydrogenation over CQO

hydrogenation. Thus, this approach is especially important for

nowadays’ catalyst development for a high performance catalyst

material in terms of activity (to increase productivity), selectivity

(to reduce the needs in product separation) and energy

considerations (to reduce energy consumption).

We have investigated hydrogenation of cinnamaldehyde to

cinnamyl alcohol over unsupported Pt nanocrystals and its

transition metal doped (bimetallic) nanoparticles. The substrate

molecule contains three reducible groups (terminal aldehyde,

double bond at a-b carbon position and benzene ring) as a

chemical probe for this investigation. Cinnamaldehyde is one

interesting model compound for hydrogenation because a

number of partially hydrogenated products can be synthesized,

depending on the selectivity of the hydrogenation reaction

(see Scheme 1). In addition, the economic importance of

selective hydrogenation of a,b-unsaturated aldehyde is

particularly denoted,1,2 because the cinnamyl alcohol can be

used as pharmaceuticals, fragrances, and perfumes.5 From the

literature, selective hydrogenation of this compound is one of the

most widely studied reactions. A wide range of catalysts,

including promoted and unpromoted metals/alloys,6–8 metal

oxides,9,10 microporous supports,11 and polymer fibre catalysts12 in both liquid2,13–15 and vapour16,17 phases was systematically

investigated. It has been empirically shown that the selectivity of

the reaction can depend on some key parameters, including the

nature of the metal and particle size,18 catalyst support,19–21 and

type of promoters/additives21–23 used. There were postulations

on the importance of structural and electronic properties of

Wolfson Catalysis Centre, Department of Chemistry, University ofOxford, Oxford, UK OX1 3QR. E-mail: [email protected];Fax: +44 1865 272600; Tel: +44 1865 282610w Electronic supplementary information (ESI) available. See DOI:10.1039/c0cp01832ez Contact address: Institute of Industrial Research- CSIR. P. O. BoxLG 576 Legon, Accra, Ghana.y Contact address: State Key Laboratory of Physical Chemistry ofSolid Surfaces, National Engineering Laboratory for Green ChemicalProductions of Alcohols-Ethers-Esters, College of Chemistry ofChemical Engineering, Xiamen University, Xiamen 361005, Fujian,P. R. China.

PCCP Dynamic Article Links

www.rsc.org/pccp PAPER

http://dx.doi.org/10.1039/c0cp01832e

http://dx.doi.org/10.1039/c0cp01832e

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 2590–2602 2591

metal catalysts as the main underlying factors.1,4,24,25 The studies

of platinum particle size 1,4 and shape24 on activity and selectivity

also gave circumstantial evidence that the particle geometry plays

a key role in the hydrogenation reaction. However, a fundamental

understanding in reaction selectivity by these structural

parameters leading to tuneable properties has yet to be achieved.

Fourier transmission Infrared Spectroscopy, FTIR, is

commonly employed to study the surface chemistry of metal-

substrate interaction in catalytic systems, which provides a

means of observing the different types of adsorption emanating

from the different geometric arrangement of atoms on

the surface of a particle. Carbon monoxide-metal interaction

is one of such matrices that can act as useful interrogating

tools in elucidating surface properties because of the

strong CO-metal bond and also the extensive background

information in literature on the types of adsorbed

species.26–34 Three types of adsorption modes of CO on Pt

single crystal surfaces have been previously observed. A

vibration frequency u(CO) between 2090–2040 cm�1 is

assigned to linearly adsorbed CO mode on a Pt site and

1860–1780 cm�1 to bridged mode adsorption26 whilst a

strong vibration frequency u(CO) of ca. 1950–1925 cm�1

prevalent in small size particles (o5.0 nm) is attributed to

the stretching mode of multicarbonyl species.4,27 In additional,

the degree of back bonding of adsorbed CO on Pt gives

progressive red shift in adsorption frequencies that can

reflect the electron density of the metal nanoparticles. Thus,

in this paper, the technique is particularly employed to obtain

the surface feature relationship with respect to catalytic

performance and to provide a mechanistic understanding on

the substrate interaction with metal surface.

On the other hand, it should be noted that it is very challenging

to disentangle the complex interplays between geometric,

electronic, and steric effects in a working catalytic system (for

example, supported bimetallic catalysts with wide heterogeneities

in size, shape, composition and metal-support interfaces).

In order to shed light on the geometric and electronic

contributions on Pt and bimetallic nanoparticles in the

selective hydrogenation of cinnamaldehyde to cinnamyl

alcohol, we have employed unsupported Pt based nanocrystals

to eliminate the support effect in this paper. In addition, a

solution technique for controlled growth of metallic

nanocrystals of defined size and surface feature by chemical

reduction allowing tailoring of particle size has also been

used. Apart from the using CO-chemisorption (FTIR),

other spectroscopic techniques such X-ray Photoelectron

Spectroscopy (XPS), Auger Electron Spectroscopy (AES)

techniques have been employed to probe the electronic

properties. X-ray Powder Diffraction (XRD), Transmission

Electron Microscopy (TEM), Energy Dispersive X-ray analysis

(EDX) and Scanning Electron Microscopy (SEM) were

conducted in order to examine the structural and chemical

changes for the Pd nanocrystals before and after modification

with a second metal. Experimental parameters such as pressure

of hydrogen, concentration of cinnamaldehyde, temperature,

and reaction time were also studied.

Experimental

Synthesis of Pt nanostructures

The Pt and its bimetallic nanocrystals were synthesized by a

modified polyol process.35–37 Typically, a mixture of bis-(acetyl

acetonato) platinum(II), Pt(acac)2, (Pt 49.49%min. Alfa Aesar,

0.30 g), 1,2-hexadecanediol (90%, Aldrich, 0.20 g), oleic acid

(99+%, Aldrich, 100 mL) and oleylamine (98%, Aldrich,

100 mL) in 6.0 mL of dioctylether (99%, Aldrich) was

refluxed at 250 1C for 40 min in a three necked round

bottom flask in an inert environment by bubbling nitrogen

gas whilst ensuring continuous stirring with a magnetic stirrer.

After 40 min the reaction mixture was cooled and 4.2 nm Pt

nanocrystals were obtained. The effect of varying the amounts

of stabilisers (oleic acid and oleylamine) was investigated using

Scheme 1 Reaction pathways of the cinnamaldehyde hydrogenation.

Table 1 The quantities of metal precursors, surfactants and reducing agents used in the synthesis of unsupported Pt and bimetallic Pt basednanocrystals

Bimetal system Quantity of second metal precursor salt usedQuantity of Pt precursor salt, oleic acid, oleylamine,and hexadecanediol used

Pt Nil 0.30 g Pt(acac)2 ,100 ml, 100 ml and 0.2 gTiPt 0.22 g Ti(IV) isopropoxide (99.999%, Aldrich)MnPt 0.19 g Mn(acac)2 (Aldrich)FePt 0.13 g Fe(Ac)2 (95% Aldrich)CoPt 0.20 g Co(acac)2 (97% Aldrich)NiPt 0.20 g Ni(acac)2 (95% Aldrich)CuPt 0.20 g Cu(acac)2 AldrichZnPt 0.17 g Zn(Ac)2 dihydrate (98.5% BDH GPR)SnPt 0.18 g Sn(Ac)2 AldrichPbPt 0.29 g Pb(Ac)2 trihydrate (99.999%, Aldrich)

NB: acac = acetyl acetonate and Ac = acetate.


the same process which resulted in a newmethod of controlling

the Pt particle size in the range of 2.8–14 nm by the

modification of the polyol process (see Table 3).4

In the case of 4.2 nm Pt doped with other metallic element,

i.e cobalt, a separate glass vial containing a solution of

cobalt(II) acetyl acetonate, Co(acac)2, (min. 97%, Aldrich,

0.20 g), 5.0 mL of dioctylether (98%, Aldrich), 64 mL of oleic

acid (99+%, Aldrich) and 1,2-hexadecanediol (90%, Aldrich,

0.50 g) was first pre-heated to 100 1C to remove water. This

typical cobalt precursor solution was injected into the 4.2 nm

platinum sol in the refluxing set-up using a glass Pasteur

pipette at 200 1C. The nitrogen gas flow was increased as

a precautionary measure to maintain the oxygen free

environment during the injection process. The temperature

was maintained at 250 1C for 20 min. The reaction mixture

was then allowed to cool to room temperature (20 1C) and

repeatedly washed in portions of an ethanol and hexane

solvents before centrifugation. The collected particles after

the centrifugation were then re-dispersed and stored in

isopropanol (Fisher Analytical Reagent grade). The air dried

particles was directly used for catalysis studies.

The quantities of Pt(acac)2 for the Pt seed formation and

the second metal precursor salt used in the preparation of

the bimetallic systems are listed in Table 1 (metal mole ratio

of ca. 1).

Catalytic Testing of Pt nanostructures

The hydrogenation experiments were carried out under liquid

phase in a stainless steel autoclave with a glass liner fitted

tightly to its inner wall. The autoclave, equipped with a

magnetic stirrer, pressure gauge and a thermocouple, was

heated and regulated by a Parr 4842 temperature controller.

The size of reactor, weight of catalyst used and quantities of

testing mixture were clearly stated in the experiments. Initially,

for the comparison of selectivity, 37 mg of unsupported Pt

nanocrystals, 5 mL of 12.5% vol/vol cinnamaldehyde in

isopropanol (IPA), 20 bar H2, 100 1C for 2 h were placed

in 25 mL stainless steel autoclave. The liquid sample was

analyzed with GC-FID and GC-MS to identify and

quantify all products, which showed a total consumption of

cinnamaldehyde under the reaction conditions. Only two

main products, the unsaturated alcohol (cinnamyl alcohol)

and the saturated aldehyde, the 3-phenylpropionaldehyde

(hydrocinnamaldehyde) were obtained with only trace of

phenylpropanol detected. In order to verify the relationship

between conversion and selectivity, the reaction was prevented

from reaching a complete conversion. Thus, only 1.0 mg of

unsupported metal nanocrystals was then weighed and placed

in the autoclave together with a magnetic stirrer bar. 5.0 mL of

3.0 vol% of trans-cinnamaldehyde (99% Aldrich) in IPA was

then added. The autoclave was purged with pure hydrogen

(99.99%, BOC) for a minute to remove any traces of oxygen. It

was then charged to 20.0 bar of hydrogen (99.99%, BOC) at

20 1C and heated to 100 1C. The reaction was allowed to

process for 2 h. The autoclave was then cooled to room

temperature, depressurised and the liquid content was

analyzed by a Hewlett Packard 5890A GC - FID equipped

with Supelco COWAX 10 polar capillary column. The

parameters to be investigated i.e. reaction pressure, time and

concentration of substrate were varied.

TEM Characterisation of Pt nanostructures

A few drops of the as-synthesised nanoparticles sol was

dispersed in ethanol using an ultra-sonic bath for 15 min.

Three drops of the uniformly dispersed sol was then added

drop-wise onto a carbon film on a cupper grid (Lacey carbon

film, 300 Mesh Cu, Agar Scientific) and the solvent was dried

off using a 60 watts lamp. The cupper grid was then placed in

an oven at 100 1C to completely dry off the solvent. The

samples were analyzed by JEOL JEM-3000F FEG TEM,

JEOIL JEM-2010 (high resolution TEMs) and Philips CM

20 (analytical TEM).

SEM/EDX Characterisation of Pt nanostructures

The nanoparticle sol was dried on a watch glass and the

powder was then dispersed to cover the entire surface of the

25 mm diameter carbon disc mounted on an aluminium SEM

specimen stubs (Agar Scientific) by a double sided adhesive

tape. The samples were then arranged in the multiple stub

sample holder for analysis using a lower energy, o4 kV

Cambridge 360 Stereoscan, SEM which was equipped with

an Oxford Instruments INCA X-ray analysis system.

AES/XPS Characterisation of Pt nanostructures

The AES/XPS analyses were carried out using the ESCA Lab

II at Johnson Matthey technology centre, Sonning Common,

Reading, UK. The spectra were measured using an Al Ka(1486.6eV) X-ray source and a hemispherical electron energy

analyser. The acceleration voltage is 10 kV and the emission

current is 12 mA for the X-ray source. The spectra were

measured at around 90 degree emission angle. The electron

energy analyser operates in FAT mode (Fixed Analyzer

Transmission), with constant pass energy of 50 eV for survey

(wide) scans and 25 eV for high resolution scans. The overall

resolution of this XPS is around 1.1 eV. The peak fitting was

conducted using CasaXPS, with a Shirley background

subtraction, a lineshape of mixed Lorentzian (30%) and

Gaussian (70%) character. The typical FWHM values

obtained from the fitting are: C1s, O 1s, Pt 4f 7/2, Pt 4f 5/2,

Co 2p3/2, Co 2p 1/2, etc. (the exact number varies from sample

to sample but mostly close to these values).

XRD Characterisation of Pt nanostructures

A portion of the nanocrystals as unsupported nanocatalysts

(powder) were packed into the groove of an aluminium sample

holder and smoothened with a glass slid to ensure a very level

sample with diameter ca. 23 mm. The sample was then loaded

into a Philips DIFF 3, X-ray diffractometer with a Cu-anode

source operating at 40 kV and 30 mA generating the Cu-KaX-ray radiation at wavelengths Cu-Ka1 = 1.5406 A and

Cu-Ka2=1.54439 A. The sample was rotated through

angular acquisition range in 2y of 31–701, step size 0.021, step

speed 0.51 min�1 and at 1.25 s per step. The detector was

equipped with a graphite diffracted beam monochromator set

for Cu-Ka NaI scintillator with pulse height analysis. The


phase search and match were carried out using X’Pert High

Score software (version 1.0d).

FTIR surface characterisation of CO pre-adsorbed Pt

nanostructures

Pure carbon monoxide (obtained from BOC) at 1 atmosphere

was bubbled through the nanoparticles colloid in isopropanol

for 15 min. The colloid sample was placed onto the crystal

surface of the Smart Golden Gate for analysis. Excess solvent

was evaporated by heating the top plate to 60 1C, and

all spectra were obtained from an average of 512 scans using

a Nicolet 6700 FTIR spectrometer equipped with an MCT

detector at a resolution of 4 cm�1 over a range of wavenumber

from 4000 cm�1 to 650 cm�1.

Result and discussion

TEM Imaging

Typically, 4.2 nm Pt nanoparticles of a uniform size

distribution were obtained by our chemical reduction process

in solution, as shown in Fig. 1 (left). The high resolution TEM

images suggested that many of the particles were found to be

highly crystallized in cubooctahedron shape as nanocrystals

under careful examination of the images (Fig. 2, left). We

showed that the effect of varying the amounts of stabilisers

(oleic acid and oleylamine) can also result in controlling the Pt

particle size in the range of 2.8–14 nm (Table 3). Based on

geometric models of size effect of nanocrystals from Boronin

and Poltorak38 and Hardeveld and Hartog,39 smaller particle

sizes (2–4 nm) possess more corners and edges than larger

particle sizes, but the larger particles show more terrace faces

than those smaller particles (Fig. 2, right). As a result, different

particle sizes can give different catalytic properties if the

catalytic process depends on nature of the active sites on the

nanocrystal locations (faces, corners and edges). Experimental

evidence was indeed obtained by Schimpf and co workers40 in

1998, where their gold nanoparticles were characterized by

transmission electron microscopy (TEM) with reference to

catalytic activity. They showed that the increase in particle

size resulted in an increase in the proportion of the surface

atoms in the closely pack (111) plane with a dramatic decrease

in the edge and corner atoms.

Our 4.2 nm Pt nanoparticle sample was later decorated

(doped) with a second metal using subsequent deposition via

the same modified polyol process. It was reported that the

pre-formed metal seed is able to catalyse the reduction of a

second metal on the seed surface.41,42 Generally, surface atoms

with the lower coordination numbers (such as adatom, kink,

edge and step atoms) will be sites for deposition due to their

high surface energy and instability. This process was aimed

to modify the Pt nanocrystal surface with another metal in

order to alter the electronic structure of the Pt. Transmission

electron micrographs of the as-synthesized unsupported Pt

nanocrystals (seed) and the second metal doped Pt showed

no significant deviation in mean particle size within

experimental error (ca. 4.25 � 1.0 nm for Pt seed particles

and 3.40 � 0.42 nm for doped Pt particles), as shown in Fig. 1

(right). This implies that the extent of decoration was very

small degree perhaps only in atomic levels.

Elemental analysis

The extent of coverage of the secondmetal on the surface of the

Pt nanocrystals was determined using two analysis techniques,

namely the lower energy, high grazing angle SEM-EDX

(Fig. 3) and the XPS. The results summarized in Table 2

confirm that the decoration was indeed limited to selected

area on the Pt surface as both analyses still gave predominant

Pt signals. The exception was the CuPt which might have

formed bulk alloy under our reaction conditions. Also, in

general, the composition of Pt-M particles obtained from

SEM-EDX analysis is consistently lower than those of XPS

(more closer to near surface analysis). This attributes to the

difference in analytical depth of the two instruments (the

analysis depth for XPS is of about o5 nm while SEM-EDX

can be up to 20 nm) and the presence of small amount of pure

second metal nanoparticles from self-nucleation. It should be

noted that estimation of the surface coverage of sub-10 nm

particles by XPS is very challenging due to the similarity in

particle dimensions and photoelectron escape depths. If the

mean bimetallic nanoparticle dimensions arer4 nm then XPS

will sample the majority of the particles (likely to take place

in this case), hence this technique is not as surface sensitive

as when considering larger nanoparticles. The different

sampling depths for each dopant bimetal could be estimated

from associated inelastic mean free paths of the relevant

Fig. 1 (left) A transmission electron micrograph (magnification of 300 000 X and 200 kV) shows the as-synthesised Pt nanocrystals with particle

size distribution of 4.25� 1.0 nm; (right) TEM image of unsupported CoPt bimetallic nanoparticles before reaction with particle size distribution of

3.4 � 0.42 nm.


photoelectron excitation monitored. On possible way to assess

the degree of surface versus bulk dopant incorporation is to

compare the atomic% of dopants observed using two different

photon excitation energies, e.g. Mg vs. Al Ka sources. If the

resulting compositions are identical then the dopant is likely

uniformly distributed throughout these 4 nm NPs, while if the

Mg-Ka yields a significantly higher value then surface

segregation is proven and can be subsequently quantified.

Such detailed analysis is yet to be collected. Nevertheless,

taking the small degree of particle size variation and self-

nucleation into account, the determined compositions by

both techniques clearly suggest that the second metal content

on Pt was much lower than the recipe value. Thus, there was

only a small degree of deposition or decoration of the second

Fig. 2 (L) High resolution TEM of typical platinum nanocrystals synthesised using the polyol process, and the corresponding modelled surface

atoms ratio with particles size of cuboctahedal shaped nanocrystals.39

Fig. 3 An SEM micrograph (insert) and EDX spectrum showing the chemical composition of Pt decorated with equal mole of (a) Ni, (b) Co,

(c) Cu and (d) Zn in synthesis recipes.

Table 2 Surface elemental composition analysis of the as-synthesised platinum bimetallic catalysts by SEM-EDX and XPS techniques

Platinum bimetallic catalyst

Atomic % of Pt Atomic % of 2nd metal

SEM-EDX XPS SEM-EDX XPS

FePt 83 88.5 17 11.5CoPt 85 87.2 15 12.8NiPt 85 n.d. 15 n.d.CuPt 63 n.d. 37 n.d.ZnPt 85 93.9 15 6.1TiPt n.d. 85.9 n.d 14.1


metal on specific locations on Pt surface (but no extensive

encapsulation of Pt nanoparticle by the second metal). By

taking the atom response factors into account, the doping

metal contents were found to be below 14% with reference to

the near surface Pt atoms. We believe that these second metal

atoms on Pt nanocrystal surface play important geometric and

electronic roles in affecting catalytic performance of the

crystals.

XRD analysis

It is known that when a secondmetal is deposited (decorated) on

another metal of different lattice parameters, a slight expansion

or contraction of the lattice of top metal exerted by the

underlying metal lattice can be resulted. This could cause a

significant alteration in electronic properties of the metal due to

the structural change (isomorphic effect).43 XRD was therefore

carefully conducted (Fig. 4) in order to examine any possible Pt

lattice alteration before and after the second metal doping.

As seen from Fig. 4, we have only observed the typical Pt

diffraction peaks with no detectable diffraction peak due to

second metal (this may indicate the second metal was either

in amorphous state or as ultra-thin layers where the severe

peaks broadening rendered them indistinguishable from the

background). There is also no observable shift in diffraction

peaks to suggest any formation of M–Pt alloy or lattice

expansion / contraction of the Pt nanocrystal after the

modification of the second metal doper. On the other hand,

owing to the poor sensitivity of XRD to surface change, we still

cannot discount a small degree of geometrical disparity due to

surface alloy formation or lattice parameter alteration of either

the host or the gust metals at the interface. However, it is

interesting to observe from the figure that the diffraction peaks

of all second metal doped Pt samples becomes broadened,

which implies the ensembles of Pt (nanocrystals) are broken up

slightly by the presence of the metal doper.

Catalytic testing and optimisations

Table 3 summarises our earlier communication note4 that Pt

nanocrystals of different sizes made by modified polyol process

showed a very large difference in selectivity towards cinnamyl

alcohol (2.8 nm gave 24.8% selectivity and 14.4 nm gave

85.6%) at a complete cinnamaldehyde conversion, suggesting

that this reaction is very structural sensitive. Clearly, the

smaller Pt sizes with higher proportion of high index sites

(undesirable sites) gave lower selectivity to cinnamyl alcohol

(2.8 nm with 24.3% and 3.3 nm with 44.6%). The effect

appears to be less significant at larger Pt sizes (6.0 nm with

80.8% and 14.4 nm with 85.6%). It is interesting to note that

the selectivity to unsaturated alcohol converged to around

80–85%. Thus, it is consistent with the geometric models

from Boronin and Poltorak38 and Hardeveld and Hartog39

where the change in Pt coordination numbers is rather

insensitive to particle size greater than 5 nm owing to the

small variations. Thus, the supported Pt samples prepared by

conventional means with a generally smaller size than the

present unsupported particles should show greater

attenuation in selectivity. However, with the accurate control

in crystal size by the polyol method this work clearly underpins

the size effect on selectivity of this hydrogenation, which can be

separated from the effects of support and chemical promotion.

Detailed analysis also showed that the selectivity depends on

the coverage of the substrate molecule (the measured selectivity

was higher when higher concentration of the substrate

molecule was used. This effect implies that the higher degree

of surface coverage at higher substrate concentration favours

the terminal aldehyde hydrogenation by creating a steric

effect that inhibits hydrogenation of the CQC bond of

cinnamaldehyde, see Fig. 6).

When the 4.2 nm Pt nanocrystal (100 mL oleic acid and

100 mL oleylamine) was decorated with second metal across

the periodic table, it is interesting to note from Fig. 5 that a

volcano response was observed for the selectivity towards

cinnamyl alcohol (and also the yield) of this hydrogenation

reaction. Only with the exceptionally lower selectivity of Ni

(nickel is well known to be hydrogenation catalyst on its own)

and higher selectivity of Sn (tin can form PtSn alloy) the volcano

response across the transition elements in the periodic table

clearly reflects the changes in electronic properties of Pt

modified by the surface transition metal doper. This will be

discussed in later section. It is worth noting that the selectivity

with respect to the second metal doper reached the maximum

value of >99% when cobalt element was doped on the

Pt nanocrystal. As a result, Co doped Pt was selected

for detailed investigation of experimental parameters on the

reaction selectivity.

Effect of substrate concentration on selectivity and conversion

Fig. 6 shows clearly that the selectivity of the CoPt

nanocatalyst towards the production of cinnamyl alcohol is

linearly dependent on the concentration of cinnamaldehyde

used. The preferential hydrogenation of cinnamaldehyde to

cinnamyl alcohol increased gradually as the concentration

of the substrate increased reaching 100% selectivity at a

concentration of 0.983 mol L�1 (12% v/v cinnamaldehyde in

isopropanol). The notable improvement in selectivity from

94% to virtually 100% as substrate concentration increasing

from 0.04 mol L�1 to 0.983 mol L�1 can be attributed to stericFig. 4 Plots of XRD patterns of the preformed Pt sample and samples

obtained after doping with a second metal by polyol process.


effects as a result of the coverage of substrate molecules on

catalyst surface.44,45 The high coverage would enable the

cinnamaldehyde molecules to be aligned linearly for closer

surface packing of the adsorbed molecules hence facilitating the

hydrogenation of the terminal aldehyde. Thus, at low coverage,

cinnamaldehyde would likely adsorb with the CQC and CQO

bonds co-planar with the metal surface (i.e. a flat-lying geometry

that facilitates hydrogenation of unsaturated bonds), while

at high coverage, a tilted geometry is likely adopted with the

preferential CQO interaction with the surface as observed those

for alkoxides on Pt. As the concentration of cinnamaldehyde

increased, the conversion was also found to decrease (from ca.

85% to 60%), as seen from Fig. 6.

Effect of hydrogen pressure on selectivity and conversion

Under this liquid phase batch process, reaction parameters

such as 3.0 mg of unsupported CoPt catalyst, 5.0 mL of

3% vol/vol cinnamaldehyde at 100 1C over a reaction time of

2 h were kept constant, which enabled the effect of hydrogen

pressure on conversion and selectivity to be studied (Fig. 7). It

was found that the increasing the hydrogen pressure from

Table 3 Size effect of Pt nanocrystals on reaction selectivity towards cinnmyl alcohol at complete cinnamaldehyde conversion, the main sideproduct is hydrocinnamaldehyde with only trace of phenylpropanol detected4

Stabilizers

Particle size (TEM) Cinnamyl alcohol selectivity (%)Oleic acid (mL) Oleylamine (mL)

40 40 14.4 (XRD) 85.680 80 6.0 80.8120 120 4.8 49.1200 200 3.3 44.6300 300 3.1 30.2400 400 2.8 24.3

Fig. 5 A volcano trend in catalytic performance (selectivity and yield) for 4.2 nm unsupported Pt decorated with second metal. Conditions:

1.0 mg of unsupported Pt nanocrystals, 5.0 mL of 3% vol/vol cinnamaldehyde in isopropanol, 20 bar hydrogen, 100 1C and 2 h. Reactor volume

is 25 mL.

Fig. 6 Effect of cinnamaldehyde concentration on selectivity and

conversion. Conditions: 1.0 mg of unsupported CoPt nanocrystals,

5.0 mL of cinnamaldehyde at various concentrations in isopropanol,

100 1C, 20 bar of hydrogen, 2 h of reaction time. Reactor volume

is 25 mL.


1.5 bar to 10 bar resulted in an increase in the cinnamyl alcohol

selectivity sharply from 70% to 96%. The selectivity was then

levelled between 10 bar and 55 bar. Similarly, the conversion

increased steadily from 21.6% at 1.5 bar to 83% at 10 bar

reaching a plateau at about 20 bar, as shown in Fig. 7. This

trend is similar to the generally proposed trend of activity

trend in many hydrogenation reactions in liquid phase with

respect to hydrogen pressure in the literature.4 The Langmuir

adsorption model is generally adopted for catalytic

hydrogenation, which requires a mass transfer of hydrogen

gas to adsorbed hydrogen over active metal surfaces like Pt,

Pd and Ru.46 Since hydrogen gas solubility is poor at low

applied hydrogen gas pressure in the reactor, an increase in

pressure would therefore favour the dissolution of hydrogen

gas in isopropanol according to the Henry’s law from 1–10 bar

of hydrogen until it reached saturation.47 The important

information from this fit is to derive the saturation hydrogen

pressure, more data point with vigorous statistical treatment is

yet required in order to deduce the Langmuir equilibrium

constant.

Effect of temperature on selectivity and conversion

Fig. 8 shows that the extent of hydrogenation is dependent on

applied temperature while all other reaction conditions are

kept constant. The conversion increases gradually from 13% at

313 K to 19% at 373 K. A similar trend is observed for the

selective towards the unsaturated alcohol (from 88% to 98%

over the same temperature regime).

Effect of reaction time on cinnamyl alcohol formation

The concentration of cinnamyl alcohol increases almost

linearly as the reaction progresses with initial experimental

time from 0 to 4 h, as shown in Fig. 9. At the beginning of the

reaction, there was simultaneous formation of both

hydrocinnamaldehyde (i.e. the hydrogenation of the CQC)

and cinnamyl alcohol so the initial selectivity at the first

120 min was as low as 90% (not shown). However, further

experimental time for this batch reaction led to higher

selectivity to cinnamyl alcohol at the expense of the

hydrocinnamaldehyde until the cinnamyl alcohol was almost

exclusively formed after 90 min, a similar observation that was

also made by Breen et al.49 It was believed that the initial

catalyst was comprised of partially oxidised Pt nanoparticles

which was subsequently reduced to active metallic phase over

the 2 h under the reaction conditions.

However, by pre-reducing the catalyst material at 100 1Cwith

5 bar of hydrogen (99.99% BOC) for 1 h before injecting

the substrate concentration (3% vol/vol cinnamaldehyde in

isopropanol), the initial formation of hydrocinnamaldehyde

can indeed be greatly suppressed, as shown in Fig. 10,

giving high selectivity towards cinnamyl alcohol. It is therefore

evident that Pt based catalysts were partially oxidised in air as

previously shown.50–53 The pre-reduction step is important to

restore the metallic surface (it took time for the reduction under

the reaction conditions), which is essential for the production

of cinnamyl alcohol through the terminal adsorption of the

substrate. Partial oxidised Pt surface appears to give a greater

extent of the initial CQC hydrogenation.

Reaction kinetics

Considering the hydrogenation of cinnamaldehyde (CAL) to

its products, the reaction is represented by the equation;

CALþH2 �!k

products

The rate equation can be represented as

R = k[CAL]m[H2]n (1)

From Fig. 11, the concentration of cinnamaldehyde in the

hydrogenation reaction decreases with the progress of the

reaction where the rate constants, k, at different temperatures

can be deduced.

For a first order reaction, plotting ln[CAL] against time (t)

should give a straight line.48,53 Fig. 12 depicts that the

experimental data apparently fit to the first order reaction

with respect to cinnamaldehyde concentration.

Arrhenius equation below was used to estimate the reaction

activation energy. Fig. 13 shows the activation energy

Fig. 7 Langmuirian fit to experimentally determined data illustrating

the effect of hydrogen pressure on selectivity and conversion.

Conditions: 3.0 mg unsupported CoPt catalyst, 5.0 mL of 3% vol/vol

cinnamaldehyde in isopropanol, 100 1C and reaction time 2 h. Reactor

volume is 25 mL.

Fig. 8 A temperature effect on conversion and selectivity in the

hydrogenation of cinnamaldehyde over CoPt; conditions: 3.0 mg

unsupported CoPt, 5.0 mL of 3% vol/vol cinnamaldehyde in

isopropanol, 20 bar of hydrogen and reaction time 2 h. Reactor

volume is 25 mL.


Fig. 9 The effect of reaction time on the formation of cinnamyl alcohol catalysed by CoPt on carbon. Conditions: 30 mg supported catalyst, 60 mL

of 3% vol/vol cinnamaldehyde in isopropanol, 100 1C and 20 bar hydrogen. Reactor volume is 300 mL.

Fig. 10 Effect of time on cinnamyl alcohol selectivity and

cinnamaldehyde conversion over the pre-reduced CoPt nanocrystals.

Conditions: 30 mg supported catalyst, 60 mL of 3% vol/vol

cinnamaldehyde in isopropanol, 100 1C and 20 bar hydrogen.

Reactor volume is 300 mL.

Fig. 11 A profile of the concentration of cinnamaldehyde dependence on time at various temperatures.

Fig. 12 A fit of the experimental data assuming a first order reaction

with respect to substrate concentration at 373.15 K and 20 bar

hydrogen.


calculated from the slope of the graph.

ln k ¼ lnA� Ea

RTð2Þ

The apparent activation energy of 17.26 kJ mol�1 was

derived in our case. This value agreed with the finding of Li

et al., who obtained their activation energy of 18 kJ mol�1 over

cobalt boron amorphous alloy catalyst.52 Breen et al. estimated

the activation energy to be 37 KJ/mol over iridium-carbon

catalyst.49 All these low value of activation energies for

this catalysed reaction suggest the rates could be somehow

subjected from mass transport or diffusion limited.48

In contrast, in our testing conditions, parameters were

deliberately chosen so no mass or diffusion limitation could

be encountered (used 20 atmosphere of H2, see Fig. 7; high

substrate concentration, see Fig. 6, over 1 mg catalyst

under optimum stirring) only except that in the kinetic rate

law determinations where the considerably lower hydrogen

pressures and lower substrate concentrations were used. They

could affect the accurate measurements in activation energy. It

would be interesting to apply a classical mathematically

modelling to the system using Weisz equation54 in order to

determine the valid experimental parameters that are not prone

to mass limitations in combination with the use of a Rushton

turbine with a hollow agitator for these fast reactions in the

future.

Geometric and electronic effects

Fig. 14 (top) shows a collection of typical FTIR spectra after

CO chemisorption on Pt nanocrystals with different sizes as

compared with a spectrum collected from the cobalt decorated

Pt (4.2 nm) sample. It demonstrates clearly that there is a sharp

increase in multicarbonyl adsorption due to the increase in the

number of highly uncoordinated Pt surface site upon

decreasing the particle size. It is apparent that the cobalt

deposition occurred principally on the highly uncoordinated

Pt sites that blocked the formation of multicarbonyl band

(M peak). As a consequence, the disappearance of the

characteristic band at u(CO) of 1925 cm�1 was evidenced.

Thus, the beneficial geometric modification of Pt nanocrystal

by cobalt atoms is concluded. In addition, there was a red shift

in the linear mode indicative of electronic promotion of Pt

(back donation) by the surface cobalt atoms. Fig. 14 (bottom)

shows a collection of FTIR spectra collected after CO

chemisorption on surface modified 4.2 nm Pt nanocrystals

where a range of transition metals were chosen as the dopant.

As seen from the typical M–Pt spectra (NiPt and FePt), apart

from linear CO adsorption mode of the Pt nanocrystals, they

also show absorbance bands corresponding to linear modes of

M–CO around 2050–1950 cm�1.27,32

The trend observed is that the selective hydrogenation of

CQO is favoured for catalysts in which the multicarbonyl

adsorption band at around 1925 cm�1 is suppressed by the

second metal deposition. This phenomenon also occurred in all

the bimetallic systems but to varying degree of efficiency with

Co metal the most exemplary. The exception to this trend is the

PbPt system (Fig. 14 bottom) in which the linear mode of CO

adsorption almost disappeared. It was observed that there was

a sharp decrease in the intensity for the linearly adsorbed CO

on Pt mode (u(CO) between 2090–2040 cm�1) which suggested

the poison effect of Pb on Pt. Pt decorated with Mn and Co

metals showed similar u(CO) patterns and they both gave a

high selectivity to cinnamyl alcohol above 90%.

It is very interesting to find that second metal atoms

preferentially segregate at sites of low coordination at the

surface of the Pt crystal, which can suppress the formation of

the multi-carbonyl peak (1925 cm�1) despite their low surface

coverage. Cyclic voltammetry can be used to analyse the

surface of platinum crystal structures; the technique can

provide structural information of the materials once they are

impregnated on to an electrode. Using this technique it should

be possible to see if the surface of platinum had been

modification in terms of second metal decoration on the

surface. Cyclic voltammograms for Pt single crystal surfaces

are well known, and Pt/graphite catalysts show four features

that correspond (by analogy with the single crystal data) to

Pt{1 1 1} � {1 1 1} steps, Pt{1 0 0} � {1 1 1} steps, Pt{1 0 0}

terraces and Pt{1 1 1} terraces. Changes in the populations

of these surface features with catalyst treatment can be

investigated and the effects on catalytic activity and

selectivity determined. Wells and co-workers have shown

that it is possible to work out the presence of different sites,

and surface structures of different size crystals.55 Cyclic

voltammetry was therefore conducted on our three samples

CoPt (1 : 1), Commercial JM 5%Pt/C and Pt/C, which was

made by the polyol process. The CoPt/C and Pt/polyol/C

contained 2.4 and 6.1% m/m Pt respectively from ICP

analysis. The JM 5%Pt/C was used as a reference sample in

order to determine the crystal faces. Pt/polyol/C was also used

as a reference sample as the platinum particles were synthesised

by the same method but without the presence of the typical

cobalt, this eliminated any effects of unwashed stabilisers on

the surface that might have been present on the sample, which

could have blocked certain crystal planes thus giving a

false interpretation of the cobalt decoration on the surface

of platinum.

The cyclic voltammograms (Fig. 15) indicated the

disappearance of Pt {111} � {111} and Pt {100} � {111}

steps (distinctive features in CVs of Pt/graphite that

correspond with single crystal data upon the decoration of

Co atoms.55 It is clear that no stabiliser appeared to have

blocked the crystals planes as all the distinctive features of

platinum were visible from the spectra on the Pt/polyol/C

Fig. 13 Arrhenius plot for cinnamaldehyde hydrogenation over CoPt

nanocrystals.


sample which was synthesised using the same stabilisers as the

CoPt particles. Pulse CO chemisorption also confirmed site

blocking (coverage) on platinum nanoparticle by Co atoms

(Table 4). The CO: metal stoichiometry of 1 : 1 was assumed in

these measurements. The addition of cobalt clearly resulted in

a decrease of the platinum surface area. Thus, the preferential

site-blocking by Co atoms reinforces the earlier postulation of

catalytic deposition on low coordination, but highly reactive

sites upon the subsequent reduction of Co2+ in the presence of

Pt nanocrystals.

Thus, the geometric (site) modification by doper is clearly

demonstrated to play an important role in cinnamyl alcohol

selectivity but the electronic modification on Pt nanocrystals

exerted by metal dopant is not yet clear. On the other hand, the

strength of surface bonding (chemisorption) with adsorbed

atoms and molecules is generally known to depend strongly on

electronic properties of underlying metal or bimetallic surface.

In addition, it has been noted from the CO chemisorption

spectra (Fig. 14) that there is a clear red shift in the linear mode

of uCQO particularly upon the addition of cobalt doper.

Bearing in mind that the Co atoms are specifically decorated

on the low coordination sites at low surface coverage, the

results therefore clearly imply that the exposed Pt atom bearing

a CO ligand must have experienced an increasing degree of

back-bonding of its d-electrons (increasing d- of Pt) to the p* ofthe CO when it is placed closer to the Co atoms on a smaller

size of Pt crystals. X-ray Photoelectron Spectroscopy and

Auger Spectroscopy (surface sensitive techniques) which can

give information about the surface composition and electronic

interactions of the metals, were conducted. Analyses of the

surface compositions of the CoPt with different recipes’ (the

number in () was the recipe Co/Pt atomic ratio) using XPS as

shown in Table 5 confirm that the second metal did not

completely encapsulate but merely decorate the platinum

nanocrystals. Their surface percentages were in line with the

results from the EDX analyses in Table 2. It is however noted

from the XPS spectra that there was no detectable chemical shift

in the binding energies (Co 2p3/2 and 2p1/2 and Pt 4f7/2 and 4f5/2)

of both Pt and Co, presumably these innermost core electrons

might be less sensitive to any change in nearby chemical

environment. Interestingly, there were more significant and

progressive shifts in Pt Auger peaks NNN and Co LMM peaks

than their core peaks (Table 5). Co exerts a greater electronic

influence to Pt on the outermost electrons and vice versa as

reported in PtCo alloy in the literature,56 indicating the local

electronic influence of Pt crystal by increasing Co atoms content.

d-band Center analysis of cinnamaldehyde hydrogenation

With the appreciation of the electronic influence of the

Pt exerted by transition metal doper from the above

characterisations one key question is how important this

electronic effect to catalytic performance. The Density

Functional Theory, DFT summarized by the d-band center

model which has been successfully used in the literature to

model various chemisorption systems over metals, in which the

closer the d-band center is to the Femi level (towards zero), the

higher adsorption energy for adsorbate obtained.57 Here,

d-band center plot is used to examine the correlation

Fig. 14 (top) Plots of ATR-FTIR spectra from CO chemisorption

studies on Pt nanocrystals colloids with different particle sizes showing

different characteristic peaks of CO linear mode u(2060 cm�1, (L)),

multicarbonyl mode u(1925 cm�1, (M)) and bridging mode u(1830cm�1, (B)). The absence of multicarbonyl peak over 4.8 nm Pt

nanocrystal decorated with cobalt is demonstrated;4 (bottom) Plots

of ATR-FTIR spectra from CO chemisorption studies on Pt samples

with a second metal decoration.

Fig. 15 Cyclic voltammograms of 5% Pt (synthesised of polyol

reduction) on graphite; JM commercial 5%Pt on graphite and 5%

Co decorated Pt (synthesised of polyol reduction, 1 : 1 CoPt) on

graphite using 0.5 M sulfuric acid as the electrolyte, voltage was

swept between 0 and 0.8 V at a rate of 10 mV s�1.


between electronic effect and the catalytic performance of this

reaction. As seen from Fig. 16, a volcano relationship between

selectivity to cinnamyl alcohol with respect to the position of

d-band center of the particular 1st row transition doper metal

referenced to its Fermi level is obtained. The selectivity reaches

the highest point at the CoPt, but starts to decrease at FePt and

drops sharply at the NiPt catalyst on both side of the apex.

This variation in selectivity is believed due to the difference in

their d-band centre position from the Fermi level when the Pt

nanocrystal surface was electronically modified by the

transition metal. Amazingly, the same volcano response with

the apex on cobalt was presented by Stamenkovic et al. who

investigated the specific activity of bimetallic Pt electro-

catalysts for oxygen reduction reaction (ORR) and the band

center position of the added transition metal.58 They attributed

their best CoPt activity to the proper balance of adsorption

energy of O2 or reaction intermediates (O2�, O2

2�, H2O2) for

optimised surface electro-catalysis. They believed that for

metal surfaces that binded oxygen or intermediates too

strongly, as in the case of Pt, the d-band center was too close

to the Fermi level and the rate of the ORR was limited by the

availability of free Pt sites (poisoning effect). On the other

hand, when the d-band center was too far from the Fermi level,

as in their case of TiPt, the surface binded O2 and intermediates

too weakly to significantly promote the ORR activity.

Although our catalytic hydrogenation reaction was very

different from the ORR in the cathode, the adsorption

strengths of substrate or intermediates in the function of

electronic properties of the underlying bimetallic surfaces

must be closely related to this, giving the same volcano

response. Since cinnamaldehyde adsorption on catalyst

surface can take place with either adsorption of alkene or

terminal carbonyl moieties followed by the formation di-smetal C–C or di-s metal C–O bonds, respectively. As d-band

center moves close to Fermi level, a stronger and more

thermodynamic favourable di-s metal C–C bonds is

preferred over di-s metal C–O on NiPt and Pt surfaces,

which reduce the selectivity in the hydrogenation of the

CQO bond. As the d-band centre moves away from the

Fermi level, it is likely to expect the di-s metal C–C binding

energy to decrease, thus leading to a more selective CQO

hydrogenation pathway. Thus, the selectivity is maximized at

around FePt and CoPt, which indicate that the value of the

d-band centre on these surfaces correspond to the optimal

binding energy of CQO in cinnamaldehyde. However, the shift

in the d-band centre will also reduce the di-s C–O binding

energy and would likely make it too weak for hydrogenation to

occur when the d-band centre is too far away from the Fermi

level, such as that observed with the TiPt catalyst. As evident

from the poor conversion shown in Fig. 5, the drop in cinnamyl

alcohol selectivity at low conversion was due to rising

significance of other side products. Therefore, d-band center

indicates the need for a delicate balance between the strengths

of these two bonding configurations which would lead to a

volcano-type relationship for the production of cinnamyl

alcohol, as illustrated in Fig. 16.

Conclusion

Bimetallic nanostructures are currently used as industrial

catalysts for many important transformations. It has been

postulated that the second metal can play several pivotal

roles within the catalyst, including site blocking and

electronic promotion on the primary metal. However, a poor

understanding of these mechanisms hampers further

development within this area. It is often difficult to disentangle

and optimise these effects, especially using catalysts that have

been prepared using traditional techniques, as invariably side

reactions are promoted. In this paper, we have presented some

new ways to approach these problems. By using simple nano-

chemistry skills, unsupported Pt nanocrystals with tailored sizes

can be decorated with another metal atom in a controlled

Table 4 Pulse CO chemisorption analysis

Catalyst (on graphite) Surface area/g metal (m2 g�1) Atom dispersion (%)

Commercial Pt (JM) 4.8 1.8Pt (Polyol) 13.7 5.0CoPt (Polyol) 9.2 3.4

Table 5 X-ray photoelectron spectroscopy and Auger emission data of Co decorated Pt nanocrystals

Catalyst% surfaceatom ratio (Co/Pt)

B.E Co0

2P3/2B.E Co2+

2P1/2B.E Pt4f7/2

B.E Pt4f5/2

K.E AugerPtNNN

K.E AugerCoLMM2

CoPt (0.24) 0.14 778.15 793.04 70.7 74.0 183.3 775.2CoPt (1.47) 0.22 778.21 793.26 70.8 74.1 184.5 774.7CoPt (5.33) — 778.21 796.74 70.8 74.7 185.2 772.2Co (Lit.) — 778.2 � 0.3 796.4 � 0.05 71.2 � 0.61 74.3 � 0.09 770.70

Fig. 16 Selectivity towards cinnamyl alcohol on the hydrogenation of

cinnamaldehyde as a function of the d-band center of transitionmetal doper.


manner. Thus, we showed that the blockage of unselective low

coordination metal sites and the optimisation in electronic

influence of the decorated Pt surface, can be independently

studied in the selective hydrogenation of a,b-unsaturatedaldehydes. In this paper, we also show that the terminal CQO

hydrogenation can be achieved in high activity, whilst the

undesirable hydrogenation of the CQC group can be greatly

suppressed by some second metal dopers. In particular, the Co

decorated Pt nanocrystals display the best activity and

selectivity for the formation of cinnamyl alcohol. Our work

clearly demonstrates the advantage in engineering preformed

nanoparticles via a bottom-up construction and illustrates that

this route of catalyst design may lead to improved and greener

manufacturing processes.

Acknowledgements

We thank Dr Richard Smith of Johnson Matthey technology

centre, Sonning Common, Reading, UK for the XPS/AES

studies.

References

1 P. Gallezot and D. Richard, Catal. Rev., 1998, 40, 81.2 P. Claus, Top. Catal., 1998, 5, 51.3 S. Bhattacharjee, D. M. Dotzauer and M. L. Bruening, J. Am.Chem. Soc., 2009, 131, 3601.

4 S. C. Tsang, N. Cailuo, W. Oduro, A. T. S. Kong, L. Clifton,K. M. K. Yu, B. Thiebaut, J. Cookson and P. Bishop, ACS Nano,2008, 2, 2547.

5 M. Ash and I. Ash, in Specialty Chemical Source Book, SynapseInformation Ressources Inc., ed., New York, 1997.

6 B. Coq, F. Figueras, P. Geneste, C. Moreau, P. Moreau andM. G. Warawdekar, J. Mol. Catal., 1991, 78, 211.

7 D. Richard, J. Ockleford, A. Girior-Fendler and P. Gallezot,Catal.Lett., 1989, 3, 53.

8 L. Sordelli, R. Psaro, G. Vlaic, A. Cepparo, S. Recchia, C. Dossi,A. Fusi and R. Zanoni, J. Catal., 1999, 182, 186.

9 B. Coq, P. S. Kumbhar, C. Moreau, P. Moreau andM. G. Warawdekar, J. Mol. Catal., 1993, 85, 215.

10 R. Hubaut, J. P. Bonnelle and M. Daage, J. Mol. Catal., 1989, 55,170.

11 P. Gallezot, A. Girior-Fendler and D. Richard, Catal. Lett., 1990,5, 169.

12 J. Aumo, J. Lijla, P. Maki-Arvela, T. Salmi, M. Sundell, H. Vainioand D. Y. Murzin, Catal. Lett., 2002, 83, 219.

13 R. Touroude, J. Catal., 1980, 65, 110.14 B. Sen and M. A. Vannice, J. Catal., 1989, 115, 65.15 T. B. L. W. Marinelli, S. Nabuurs and V. Ponec, J. Catal., 1995,

151, 431.16 L. Mercadante, G. Neri, C. Milone, A. Donato and S. Galvagno,

J. Mol. Catal. A: Chem., 1996, 105, 93.17 G. Neri, L. Mercadante, C. Milone, R. Pietropaolo and

S. Galvagno, J. Mol. Catal. A: Chem., 1996, 108, 41.18 S. Galvagno, C. Milone, A. Donato, G. Neri and R. Pietropaolo,

Catal. Lett., 1993, 18, 349.19 U. K. Singh and M. A. Vannice, J. Catal., 2001, 199, 73.20 R. Malathi and R. P. Viswanath, Appl. Catal., A, 2001, 208, 323.21 M. Consonni, D. Jokic, D. Y. Murzin and R. Touroude, J. Catal.,

1999, 188, 165.

22 Z. Poltarzewski, S. Galvagno, R. Pietropaolo and P. Staiti,J. Catal., 1986, 102, 190.

23 M. A. Aramendia, V. Borau, C. Jimenez, J. M. Marinas, A. Porrasand F. J. Urbano, J. Catal., 1997, 172, 46.

24 R. Narayanan and M. A. El-Sayed, Nano Lett., 2004, 4, 1343.25 P. Fouilloux, in Heterogeneous Catalysis and Fine Chemicals, ed.

M. Guisnet, et al., Amsterdam, Elsevier, 2000, pp. 123–129.26 D. de Caro and J. S. Bradley, Langmuir, 1998, 14, 245.27 L. H. Little, in Infrared Spectra of Adsorbed Species, Academic

Press, London, 1966.28 L. C. deMenorval, A. Chaqroune, B. Coq and F. Figueras,

J. Chem. Soc., Faraday Trans., 1997, 93, 3715.29 O. Dulaurent, M. Nawdali, A. Bourane and D. Bianchi, Appl.

Catal., A, 2000, 196, 271.30 P. Kubanek, H. W. Schmidt, B. Spliethoff and F. Scuth,

Microporous Mesoporous Mater., 2005, 77, 89.31 T. I. Koranyi, J. Mihaly, E. Pfeifer, C. Nemeth, T. Yuzhakova and

J. Mink, J. Phys. Chem. A, 2006, 110, 1817.32 M. Mihaylov, K. Hadjiivanov and H. Knozinger, Catal. Lett.,

2001, 76, 59.33 K. Hadjiivanov, T. Venkov and H. Knozinger, Catal. Lett., 2001,

75, 55.34 K. Tedsree, A. T. S. Kong and S. C. Tsang,Angew. Chem., Int. Ed.,

2009, 121, 1471.35 S. H. Sun, C. B. Murray, D. Weller, L. Folks and A. Moser,

Science, 2000, 287, 1989.36 M. Chen, J. P. Liu and S. H. Sun, J. Am. Chem. Soc., 2004, 126, 8394.37 N. S. Sobal and M. Giersig, Aust. J. Chem., 2005, 58, 307.38 O. M. Boronin and V. S. Poltorak, Int. Chem. Eng., 1967, 7, 452.39 R. van Hardeveld and F. Hartog, Surf. Sci., 1969, 15, 189.40 S. Schimpf, M. Lucas, C. Mohr, U. Rodemerck, A. Bruckner,

J. Radnik, H. Hofmeister and P. Claus, Catal. Today, 2002, 72, 63.41 H. L. Hu and K. Sugawara, Mater. Lett., 2009, 63, 940.42 J. L. Cheng, H. H. Hng, H. Y. Ng, P. C. Soon and Y. W. Lee,

Mater. Res. Bull., 2009, 44, 95.43 M. Galeotti, A. Atrei, U. Bardi, B. Cortigiani, G. Rovida and

M. Torrini, Surf. Sci., 1993, 297, 202.44 M. Bron, D. Teschner, A. Knop-Gericke, B. Steinhauer,

A. Scheybal, M. Havecker, D. Wang, R. Fodisch, D. Honicke,A. Wootsch, R. Schlogl and P. Claus, J. Catal., 2005, 234, 37.

45 S. Fujii, N. Osaka, M. Akita and K. Itoh, J. Phys. Chem., 1995, 99,6994.

46 P. G. J. Koopman, A. P. G. Kieboom and H. Vanbekku, ColloidsSurf., 1981, 3, 1.

47 V. Schettino and R. Bini, Chem. Soc. Rev., 2007, 36, 869.48 G. C. Bond, inHeterogeneous Catalysis Principles and Applications,

Oxford Chemistry Series, Oxford University Press, Oxford, 1974.49 J. P. Breen, R. Burch andGomez-Lopez,Appl. Catal., A, 2004, 268,

267.50 H. T. Stokes, C. D. Mokowka, P. K. Wang, S. L. Rudaz and

L. P. Slichter, J. Mol. Catal., 1983, 20, 321.51 P. Ratnasam, A. J. Leonard, L. Rodrique and J. J. Fripiat,

J. Catal., 1973, 29, 357.52 H. X. Li, X. F. Chen, M. H. Wang and Y. P. Xu, Appl. Catal., A,

2002, 225, 117.53 J. E. House, in Principles of Chemical Kinetics, London, Academic

Press, 2nd edn, 2007.54 P. B. Weisz, Ber. Bunsen Ges. Phys. Chem., 1975, 79, 798.55 G. A. Attard, K. G. Griffin, D. J. Jenkins, P. Johnston and

P. B. Wells, Catal. Today, 2006, 114, 346.56 H. X. Li, H. Li and M. H. Wang, Appl. Catal., A, 2001, 207, 129.57 A. Ruban, B. Hammer, P. Stoltze, H. L. Skriver and J. K. Norskov,

J. Mol. Catal. A: Chem., 1997, 115, 421.58 V. R. Stamenkovic, B. S. Mun, M. Arenz, K. J. J. Mayrhofer,

C. A. Lucas, G. Wang, P. N. Ross and N. M. Aovic, Nat. Mater.,2007, 6, 241.

Citethis:Phys. Chem. Chem. Phys.,2011,13 ,25902602 PAPERfulltext.calis.edu.cn/rsc/physical chemistry chemical physics/c0cp01832e.pdf · 2590 Phys. Chem. Chem. Phys., 2011, 13 , 25902602

Documents