Size- and structure-controlled mono- and bimetallic Ir-Pd nanoparticles in selective ring opening of indan

Article abstract of DOI:10.1016/j.jcat.2013.01.004

Nearly monodispersed 1.6 nm Ir, 2.3 nm Pd nanoparticles, 2.7 nm Pd(core)-Ir(shell) and 2.2 nm Pd-Ir alloys with mixed surface atoms were synthesised in the presence of polyvinylpyrrolidone (PVP) and studied in the atmospheric ring opening of indan. The na

Full text of DOI:10.1016/j.jcat.2013.01.004

Journal of Catalysis 300 (2013) 113–124
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Size- and structure-controlled mono- and bimetallic Ir–Pd nanoparticles
in selective ring opening of indan
a
a
a
b
c
a,
⇑
Hessam Ziaei-azad , Cindy-Xing Yin , Jing Shen , Yongfeng Hu , Dimitre Karpuzov , Natalia Semagina
a
Department of Chemical and Materials Engineering, University of Alberta, 9107-116 St., Edmonton, AB, Canada T6G 2V4
Canadian Light Source Inc., 101 Perimeter Road, Saskatoon, SK, Canada S7N 0X4
Alberta Centre for Surface Engineering and Science, University of Alberta, Edmonton, Canada
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
Nearly monodispersed 1.6 nm Ir, 2.3 nm Pd nanoparticles, 2.7 nm Pd(core)–Ir(shell) and 2.2 nm Pd–Ir
alloys with mixed surface atoms were synthesised in the presence of polyvinylpyrrolidone (PVP) and
studied in the atmospheric ring opening of indan. The nanoparticles and supported catalysts were
characterised by UV–vis spectroscopy, TGA, TEM, EDX, TPR, XPS, ISS and XANES. The nanoparticle size
and structure control allowed insights into the bimetallic catalyst functioning. As soon as two contiguous
surface Ir atoms exist on the nanoparticle surface, they display the same catalytic properties, most likely
through the dicarbene ring-opening mechanism. Palladium serves only as a dispersing agent in
Pd(core)–Ir(shell) structures providing 100% Ir dispersion, or as an inert surface diluting metal in Pd–Ir
formulations with mixed nanoparticle surface. The Pd–Ir core–shell structure allows maximum selective
ring-opening yield, which is 19% higher than that for the industrial Pt–Ir catalyst.
Received 29 November 2012
Accepted 8 January 2013
Available online 13 February 2013
Keywords:
Bimetallic
Nanoparticles
PVP
Ring opening
Indan
Palladium
Iridium
Ó 2013 Elsevier Inc. All rights reserved.
Core–shell
Size control
synergism
1
. Introduction
Bimetallic catalysts play a significant role in the chemical indus-
thermal treatments in various atmospheres. This technique usually
results in a poor control over the particle size, structure and the
close vicinity between the two metals, giving rise to a wide family
of monometallic and bimetallic particles with different composi-
tions that coexist at the surface [5]. As the surface composition
in most cases is not controlled and/or known, the changes in cata-
lytic behaviour are typically assigned to synergism effects between
the two metals, and the role of each metal in the bimetallic formu-
lations is difficult to explain. On the contrary, if the surface bime-
tallic composition is well known, the metal functions can be
quantitatively described and either confirm or reject intrinsic syn-
ergism hypothesis.
In the recent years, much attention has been paid to the size-
and shape-controlled synthesis of metallic nanoparticles for cata-
lytic applications [6]. This interest mainly originates from the
quantum size effects created in nanoparticles as a result of a signif-
icant reduction in the particle size (below 5 nm), which can alter
the chemical activity and selectivity of the metallic nanoparticles
[6]. Colloid chemistry techniques have been applied for the con-
trolled synthesis of bimetallic nanoparticles, such as so-called ran-
dom alloy and the ‘core/shell’ structures. Steric stabilisers like
polyvinylpyrrolidone (PVP) are usually used during the synthesis
to protect nanoparticles from excessive growth [7]. Numerous
examples of such bimetallic catalytic nanoparticles have been
try as the addition of a second metal may promote the perfor-
mance of monometallic materials. The alloying of Pd with Au, Ag,
Cu or Pb for the selective hydrogenations or the promotion of
Pt-based catalysts with Ir and Re in the reforming units is just a
few of the most prominent examples in this area [1–4]. Bimetallic-
ity can change the activity, selectivity and stability of monometals
via electronic and geometric effects and the occurrence of mixed
sites [5]. Electronic effects can be explained as the redistribution
of charge in a metal, or charge transfer between two metals in a
bimetallic composition, that can modify the catalytic properties
through the mode and adsorption strength of the reaction partici-
pants. Geometric effects mostly concern the positioning of surface
atoms in bimetallic catalysts and sometimes can result in the dilu-
tion of surface active sites with those of an inactive metal. Finally, a
mixed site may be interpreted as an active site where both metals
participate in the catalytic action [5].
Traditionally, heterogeneous catalysts are prepared by the
deposition of molecular precursors on supports followed by
⇑
Corresponding author. Fax: +1 780 492 2881.
E-mail address: semagina@ualberta.ca (N. Semagina).
0
021-9517/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2013.01.004

1
14
H. Ziaei-azad et al. / Journal of Catalysis 300 (2013) 113–124
documented in the literature, such as the application of alloy-like
structures NiPd, CuPd, PtPd, PdM (M = Au, Pt, Zn) and PdRh for
hydrogenation [8–15], NiIr and PtRu for hydrogen generation
attacked by the carbon layer, which could facilitate the production
of 2MP and 3MP. The resistance to self-poisoning by carbon depos-
its is reportedly higher in smaller particles, which contain more
edges and corner atoms than face atoms [56].
Thus, in order to correlate the catalytic behaviour with nano-
particle metal nature of some definite size, nanoparticles’ mono-
dispersity is of crucial importance. The issue becomes even more
important when it comes to the comparison of mono- and bimetal-
lic nanoparticles, as the size effect should not influence a research-
er’s hypothesis on the role of the second metal in bimetallic
formulations. The colloidal techniques of nanoparticle size and
structure control become particularly advantageous in elucidating
mechanisms over bimetallic catalysts.
The present study is devoted to the development of size- and
structure-controlled mono- and bimetallic Ir–Pd nanoparticles
with high catalytic activity and selectivity in SRO. A Pt-free catalyst
was targeted due to the relatively high cost of Pt and its low sul-
phur tolerance. The work stands out of the numerous investiga-
tions on Ir-based SRO catalysts due to the enhanced control over
the nanoparticle size and structure, which makes it possible to cor-
relate the bimetallic nanoparticle composition with its catalytic
properties. We believe that such approach should become routine
in bimetallic catalysis to elucidate intrinsic or observed synergism
between the two metals.
Iridium is selected due to its known high ring-opening activity
and palladium due to its excellent hydrogenation activity as a
probable candidate for platinum replacement. Obtaining different
bimetallic structures allowed us getting insight into the bimetallic
catalyst functioning; moreover, it is advantageous to be able to
place the atoms of Ir, which is one of rarest elements on Earth, only
at the outer shell of nanoparticles for an optimum catalyst formu-
lation. We report both random alloys and Pd(core)–Ir(shell) nano-
particles as efficient catalysts in indan ring opening (sulphur-free,
as a model reaction). The comparison of the catalytic behaviour of
the developed catalysts and that one of the industrial Pt–Ir coun-
terpart allows suggesting them as prominent advanced materials
for SRO.
[
16,17], AuPt and NiPt for electrooxidation [18,19], AgAu, AuPd
and NiPt for reduction [20–22], AuPd and AgAu for CO oxidation
23,24], FeRu and FePt for the formation of single-walled carbon
[
nanotubes (SWNTs) [25], along with those of the core/shell struc-
tures Au/Pd, Pd/Pt and Ag/Pd for hydrogenation [26–29], M/Pt
(
[
M = Ru, Rh, Ir, Pd, or Au) for preferential CO oxidation in hydrogen
30], Pt/Sn for aromatisation [31], Au/M (M = Pd, Pt, Rh) for hydro-
gen generation [32,33], Pd/Pt for methane combustion [34] and Pd/
Pt for electrooxidation [35].
In the plethora of catalytic processes, the refining industry is
one of the most important areas in which bimetallic nanoparticles
play a key role. One of the pioneering examples in this category
dates back to the 1970s, when Exxon introduced the Pt–Ir bimetal-
lic catalysts to the reforming units with improved activity and sul-
phur resistance compared to the monometallic Pt previously used
[
4,36]. On the other hand, internal combustion engines are still the
main source of road transportation, and the role of catalysis in
increasing the yield of premium quality fuels produced in the refin-
eries is vital. The catalytic selective ring opening (SRO) of polyaro-
matic and naphthenic compounds found in the heavy oil fractions
is one of the key processes in increasing the quality of fuels, as dur-
ing the process, a C–C bond on the ring is cleaved only once, retain-
ing the same number of the carbon atoms in the product [37,38]. In
terms of the cetane number improvement, it is important that the
cleavage occurs at the substituted C–C bonds [39]. Most of the lit-
erature studies on SRO have been focused on relatively small mol-
ecules like methylcyclopentane (MCP), which has a boiling point
(
ꢀ72 °C) far from the diesel cut [40–43]. Another model com-
pound, indan (benzocyclopentane, C 10), with one aromatic and
9
H
naphthenic ring also has been applied as a test substance for the
SRO [37].
Since the 1970s, the ring opening of cyclic compounds has been
studied extensively using Pt-group-supported catalysts [37,41,42,
4
4–54]. It is reported that the RO of naphthenic rings relies on a
metal-driven chemistry and that the Ir has always stood out as
the most active metal for these reactions, although its high crack-
ing propensity may reduce the yield of the single cleavage products
2
. Experimental
[
44,45,48,50,55]. Not only the metal nature but also the size of the
2
.1. Materials
catalytic nanoparticles can alter the way they behave in RO reac-
tions due to the reaction structure sensitivity. RO of MCP is well
known for its structure sensitivity: Pt nanoparticles larger than
Hydrogen hexachloroiridate (IV) hydrate (H
2
IrCl
6
ꢁxH
2
O,
, 5%
Sigma–Aldrich), dihydrogen tetrachloropalladate (II) (H
w/v solution, Acros Organics), polyvinylpyrrolidone (PVP, average
molecular weight 40,000, Sigma–Aldrich), gamma alumina
2
PdCl
4
2
nm lead to cleavage only on substituted carbons, while below
that, cleavage on unsubstituted carbons is also possible. Such
behaviour is related to the metal crystal surface statistics, as the
presence of at least two contiguous edge atoms in bigger particles
facilitate the formation of complexes that yield 2-methylpentane
2
(c
-Al
2
O
3
, average pore size 58 Å, SAA 155 m /g, Sigma–Aldrich),
ethanol (95 vol.%, Fischer Scientific) and acetone (>99.7%, Fischer
Scientific) were used as received. Nitrogen, argon and hydrogen
of ultra-high purity (99.999%) were purchased from Praxair.
(
2MP) and 3-methylpentane (3MP) [41,46]. Size effect can also
change the activity of a specific metal for RO reactions. For exam-
ple, the RO of MCP over Rh, which operates on the unsubstituted
carbons, showed higher specific activity on higher loading Rh
Milli-Q water (18.2 M
X cm) was used throughout the work.
Benzocyclopentane (indan, 95 vol.%, Sigma–Aldrich) was distilled
once. A batch of Pt–Ir industrial catalyst was used to compare
the catalytic behaviour; its composition and other properties are
not reported due to the confidentiality issues.
particles (10% Rh/Al
Rh/Al ), which was ascribed to the corrugate flat morphology
higher proportion of face atoms) in larger Rh particles [42].
2 3
O ) than the lower loading samples (0.3%
2 3
O
(
Similarly, larger Ru (3 nm) turned out to be more active than
the smaller ones (1.2 nm) in the SRO of methylcyclohexane, where
2.2. Catalyst preparation
2
the strength of H chemisorption played the key role [47]. The
blocking and/or poisoning of the active sites can also change the
behaviour in RO reactions. For instance, the RO of MCP over Ir,
which happens only on the unsubstituted carbons, is reported
not to be size sensitive; however, the presence of the carbonaceous
layer on the Ir particles can result in the production of n-hexane as
well as 2MP and 3MP. This behaviour reportedly happens due to
the elimination of a large number of contiguous sites (face atoms)
Pd and Ir mono- and bimetallic nanoparticles were preformed
in a colloidal dispersion before deposition on
2 3
c-Al O . Alcohol
reduction is known as a simple and effective way to produce
well-dispersed nanoparticles [32]. Pd and Ir monometallic nano-
particles were synthesised by the one-step alcohol reduction
method, which has been described elsewhere for Pd [57]. Briefly,
2 4 2 6
50 ml of the 2 mM solution of either H PdCl or H IrCl in Milli-Q

H. Ziaei-azad et al. / Journal of Catalysis 300 (2013) 113–124
115
water together with 0.22 g of PVP (PVP/metal = 20:1 mol/mol) was
added to a 250-ml three-neck flask and mixed with 120 ml of
water and ethanol solution (3:2 v/v). The resulting mixture was
stirred and heated until it came to boil and then refluxed in air
under vigorous stirring for 3 h to complete the synthesis of PVP-
protected nanoparticles. A dark but transparent and homogeneous
colloidal dispersion of nanoparticles was obtained at the end of
synthesis without any precipitate.
Bimetallic nanoparticles were synthesised with two different
methods of co-reduction and hydrogen-sacrificial reduction to pro-
duce random alloy and core–shell particles, respectively. For the
co-reduction synthesis of Pd and Ir nanoparticles, 50 ml of the
under N
2
flow, held for 10 min; then, N
2
was switched to air, and
the temperature was further increased to 600 °C with a heating
rate of 5 °C/min. Samples studied by TGA were also analysed by
an Elementar Vario Micro elemental analyser for CHN percentage.
TEM images were recorded using a JEOL JEM2100 TEM device
operating at 200 kV. Samples for TEM analysis were prepared by
placing a drop of the colloidal dispersion of nanoparticles onto a
carbon-coated copper grid, followed by the evaporation of the sol-
vent at room temperature. The average diameter and the standard
deviation were calculated by counting over 200 particles using the
ImageJ software.
Scanning transmission electron microscopy (STEM) coupled
with energy dispersive X-ray (EDX) spectroscopy was performed
using a Hitachi HD-2000 operating at 200 kV on selected fresh cat-
alysts calcined in air at 200 °C for 1 h.
2
mM solution of metal precursors with two different molar ratios
of Pd:Ir = 1:1 and Pd:Ir = 1:3 was prepared and added to a 250-ml
three-neck flask. As the final Pd:Ir molar ratios in the resulting cat-
alysts were found as 2:1 and 1:1, respectively, the catalysts hereaf-
ter are referred as Pd2Ir1 and Pd1Ir1. A certain amount of PVP
2
Hydrogen temperature programmed reduction (H -TPR) was
performed using an AutoChem 2950 HP device (Micromeritics) un-
der atmospheric pressure. About 500 mg of the catalysts calcined
in air at 200 °C for 1 h was used for the analysis. Before the TPR
spectra were collected, samples were initially reduced in 10 ml/
min mixture of 10% H
in 10 ml/min of 10% O
tive gases were switched to He and Ar after the reduction and oxi-
dation steps, respectively; the samples were allowed to cool down
to ambient temperature. The TPR spectra were collected in 10 ml/
(
PVP/metals = 10:1 mol/mol) was added as well, and the rest of
the process was exactly the same as that of monometallic nanopar-
ticles. The mixture of metal precursors was simultaneously re-
duced in a one-step process in the alcohol–water solution using
the co-reduction technique. The resulting colloidal solution ob-
tained was macroscopically homogeneous and transparent with-
out any precipitate.
2
in Ar up to 375 °C for 1 h and then oxidised
in He mixture up to 400 °C for 2 h. The ac-
2
In addition to the Pd2Ir1 and Pd1Ir1 nanoparticles, another
bimetallic structure Pd(core)–Ir(shell) with a Pd:Ir molar ratio of
2
min mixture of 10% H in Ar from ambient temperature to 300 °C
1
:1 (hereafter referred as Pd(c)Ir(s)) was synthesised using the
with a heating rate of 5 °C/min.
hydrogen-sacrificial reduction method. Pd monometallic core
nanoparticles were synthesised as vide supra. After the refluxing
time, the solution was allowed to cool down to room temperature
followed by purging with hydrogen for 1.5 h. This procedure cre-
ates Pd hydride, which further serves to reduce Ir and form an Ir
shell around the preformed Pd nanoparticle. To create an Ir shell,
X-ray photoelectron spectroscopy (XPS) was performed using a
Kratos Axis 165 X-ray photoelectron spectrometer with a Mono Al
a source operating at 15 mA and 14 kV. The XPS analysis was
K
done on the pelletised samples which were calcined in air at
200 °C for 1 h followed by the reduction in hydrogen at 359 °C
for 1 h. Background subtraction (Shirley-type), smoothing and
peak fitting were performed using the CasaXPS software package.
All the core-level spectra were corrected with C 1s at 284.6 eV.
Ion scattering spectra of selected samples were obtained using a
Kratos Axis 165 X-ray photoelectron spectrometer equipped with
an ion scattering spectrometer, operated using noble gas ion beams
5
0 ml of 2 mM H
dropping funnel; the precursor solution was constantly purged
with N , while the Pd–Ir dispersion with H . The final dispersion
2
IrCl
6
ꢁxH
2
O solution in water was added via a
2
2
was left under hydrogen overnight. No precipitation was observed
at the end of synthesis, and the resulting colloidal dispersion was
macroscopically homogeneous and transparent.
+
(He ) at 1 keV and scattering angle of 135°. ISS of metallic silver foil
All the synthesised nanoparticles were supported on predried
-Al O . The nanoparticles were precipitated using acetone in the
2 3
presence of alumina [58]. The resulting powder was dried in the
fume hood overnight to form the so-called fresh catalysts, which
were then calcined in air at 200 °C for 1 h.
was used to reference the energy scale. Similar catalysts but with
higher metal loading (about 1 wt.%) was used for the analysis since
an ISS detector requires relatively high metal loading for powdered
samples. Prior to the analysis, samples underwent the calcination
and reduction pretreatment as described above for the XPS studies.
c
3
The Pd L -edge XANES of the various samples were recorded at
2.3. Catalyst characterisation
the Soft X-ray Microanalysis Beamline (SXRMB) 06B1-1 of the
Canadian Light Source (CLS; Saskatoon, Canada) using a Si (111)
The actual Pd loadings on
c
-Al
2
O
3
were determined by a Varian
double crystal monochromator (energy range, 1.7–10 keV; resolu-
ꢃ4
SpectrAA 220FS atomic absorption (AA) spectrometer after Pd dis-
solution in hot nitric acid. The loading of Ir-containing nanoparti-
cles was determined by neutron activation analysis (NAA).
Samples were irradiated for 110 s in the Cd shielded, epithermal
site of the reactor core. They were counted for 30 min each on an
Aptec CS11-A31C gamma detector, approximately 12 h after
irradiation.
UV-visible spectroscopy was performed on nanoparticles in the
colloidal solution with a Varian Cary 50 Scan UV-visible spectrom-
eter, 1 cm quartz cell.
tion, 1 ꢂ 10
DE/E). The same calcination and reduction pretreat-
ment as for the XPS analyses were done on the samples prior to
this analysis. Samples in powder form were mounted on the dou-
ble-sided conducting carbon tape. Measurements were performed
in total electron yield (TEY) mode by recording the drain sample
current and in fluorescence yield (FY) using a Si(Li) drift detector,
all at room temperature. Both TEY and FY spectra were similar,
but only TEY results will be presented here. Standard reference
Pd foil was also measured for the respective edge and used as
the reference for comparison. The resulting XANES spectra were
normalised using the Athena software.
Thermogravimetric analysis (TGA) was done using a TA instru-
ment Q500 TGA device. Ca. 10–20 mg of sample was used for TGA
measurements. TGA analyses performed on a selected catalyst (Ir/
2.4. Catalytic behaviour in indan ring opening
2 3
c-Al O ) under different conditions, that is, (i) fresh, (ii) calcined in
air at 200 °C for 1 h and (iii) calcined in air at 200 °C for 1 h fol-
lowed by the reduction in hydrogen at 359 °C for 1 h. Some pre-
Catalytic studies on the indan (benzocyclopentane) ring open-
ing were performed at 336 °C (internal temperature) and atmo-
spheric pressure in a flow stainless steel reactor (length of 16 in.,
i.d. ½ in.). Catalysts, calcined in air at 200 °C for 1 h, were diluted
dried
2 3
c-Al O was also tested for comparison. During the
analysis, the temperature was increased from ambient to 110 °C

1
16
H. Ziaei-azad et al. / Journal of Catalysis 300 (2013) 113–124
with 2 g of #150 mesh SiC. 1.2 mg of the active metal(s) was used
for each run (unless indicated otherwise). Gas flows were regulated
with calibrated mass flow controllers (Sierra Instruments). The sys-
tem was purged with Ar before and after each test. The catalysts
were initially pretreated with hydrogen in situ, 80 ml/min with
temperature ramp from ambient to 359 °C at 5 °C/min, held for
3. Results and discussion
3.1. Catalyst characterisation
The formation of nanoparticles in the colloidal dispersion was
confirmed by UV-visible spectroscopy (Fig. 1). The Pd precursor
solution presents a peak at about 206 nm, which has been reported
1
h. At the end of pretreatment, the reactor temperature and the
hydrogen flow rate were lowered to 350 °C and 50 ml/min, respec-
tively. The catalytic reaction started by bubbling hydrogen through
the bubbler with indan that was kept at the constant temperature
water bath of 10 °C. An indan flow rate of 2.8 ꢂ 10 g/min was
as the characteristic peak of PdCl [59]. During the synthesis of Pd
nanoparticles, this peak disappeared, and the colour of the medium
2
changed from yellow (before the reflux) to dark brown (after the
reaction). In the case of iridium, the starting solution shows two
distinct peaks at 433 nm and 488 nm, which are in agreement with
the literature data and can be ascribed to the ligand-to-metal
ꢃ5
confirmed using GC, which was calibrated with a Gilson HPLC
pump with a known indan flow rate. The reactor up- and down-
streamlines were heated to 220 °C in order to preheat the reactants
and avoid product condensation. The reaction products were ana-
lysed online every 30 min with a 430 Varian Gas chromatograph
equipped with an FID detector. The GC capillary column was a
WCOT fused silica column, 50 m length ꢂ 0.32 mm inside diame-
ter ꢂ 1.2 mm thickness. Initially, the oven temperature was stabi-
lised at 40 °C for 2.5 min; then, it started to increase at a rate of
2
ꢃ
charge-transfer (MLCT) absorption of [IrCl6] ions [16]. Similarly,
during the synthesis of Ir nanoparticles, the above-mentioned
peaks disappeared, the absorption in the visible region increased,
and the solution turned from pale yellow to dark brown after the
reflux. This can be ascribed to the nanoparticles’ formation as
metal nanoparticles have the ability of absorbing photons in the
UV-visible region due to the coherent oscillation of their conduc-
tion band electrons [60]. As an example of bimetallic nanoparticles,
the spectrum of the Pd(c)Ir(s) colloidal dispersion also confirms the
nanoparticle formation, both for Pd and for Ir counterparts.
3
0 °C/min until the temperature reached 110 °C, and then, it was
maintained at 110 °C for 20 min. FID and injector temperature
were 280 °C. The split ratio was 1. Helium flow rate was constant
at 25 ml/min. Each catalyst was tested for 240 min on stream.
For the online product analysis, after a steady state was achieved
One of the important issues about the use of PVP-stabilised
nanoparticles in a catalytic reaction is the removal of the protective
polymer before assessing the catalyst performance. It is reported
that part of the main chain of PVP can be adsorbed on the metal
surface as a result of hydrophobic interactions [57]. According to
Somorjai et al., an oxidation–reduction cycle is known to remove
PVP from metal nanoparticles at lower temperature than free
PVP decomposition. Maximum activity for ethylene hydrogenation
with PVP-containing catalysts was shown after an oxidation–
reduction pretreatment at 200 °C [61]. Fig. 2 presents the results
(
at 150 min time-on-stream), no more than 5% deviation in the
mass balance was observed (typically, within 2%) as compared to
the mass flow of incoming indan. The absence of external and
internal mass transfer limitations was verified experimentally by
varying the flow rate and estimating the Weisz–Prater criterion.
The reaction products, as well as indan impurities, were identi-
fied via GC–MS that was performed with an Agilent Technologies
7
890 GC coupled with a 5975C MSD. The GC column used is a
ZB-50 (Phenomenex) column, 30 m length ꢂ 25 mm i.d. ꢂ 25-
l
m
2 3
of thermogravimetric studies on the Ir/c-Al O catalyst sampled
thickness. Oven temperature was stabilised at 40 °C for 0.5 min,
under different conditions, that is, (i) fresh, (ii) calcined in air at
200 °C for 1 h and (iii) calcined in air at 200 °C for 1 h followed
by the reduction in hydrogen at 359 °C for 1 h. The third sample
is the one which was further used for the catalytic reaction. The
observed weight losses can be divided into three different regions:
initially, below 200 °C, the mass loss is mostly connected to the
desorption of water physically adsorbed to the surface in the wet
chemistry of synthesis. Between 200 °C and 400 °C, designated as
the ‘slope change region’, the largest mass loss occurs due to the
thermal decomposition of PVP which has been reported to occur
between 200 °C and 375 °C under air in the presence of Pt
then increased to 110 °C at 30 °C/min, and then increased to
2
80 °C at 50 °C/min. The desired single cleavage products of ring
opening are 2-ethyltoluene and n-propylbenzene. Main dealkyla-
tion products are o-xylene, ethylbenzene, toluene and benzene.
Compounds with GC retention times less than that of benzene
were named as ‘lights’, and products other than the major products
and lights were called ‘others’. The same main products were re-
ported in the indan ring opening over Pt–Ir catalysts literature
[
37]. Raw GC results were corrected for indan impurities, and all
calculations for indan conversion and product selectivities were
based on the corrected GC results. Indan purified by distillation
contains 0.09% benzene, 0.15% n-propylbenzene, 0.02% 2-
ethyltoluene, 0.02% lights and 2.91% other.
Selectivities are reported on a mass basis as molar selectivity
can give a distorted picture of indan utilisation because up to
9
mol of methane may be produced per mole of indan. The FID
detector is a mass-sensitive analyser that responds to the number
of carbon atoms entering per unit of time; thus, for each RO open-
ing product, the selectivity was determined as the GC area of each
product dividing by the difference between total GC area and the
GC area of indan. Single cleavage selectivity is defined as the sum
of the 2-ethyltoluene and n-propylbenzene formed divided by
the amount of indan that has reacted. Selective RO yield is defined
as the sum of 2-ethyltoluene and n-propylbenzene formed per the
total amount of indan fed. For each catalyst, 10 data points were
obtained at 90, 120, 150, 180 and 210 min of time-on-stream with
a duplicate experiment. The error never exceeded 10% and was
typically within 3%. Within the indicated times on stream the cat-
alysts did not show significant deactivation as will be seen from
the conversion and selectivity versus time-on-stream profiles vide
infra.
Fig. 1. UV–vis spectra of the selected PVP-stabilised nanoparticles and the starting
solutions of metal precursors.

H. Ziaei-azad et al. / Journal of Catalysis 300 (2013) 113–124
117
The analyses below were performed to confirm the nanoparticle
bimetallicity and structure, instead of physical mixture of the
monometallic counterparts. STEM coupled with EDX of the
Pd(c)Ir(s)/
the nanoparticles on
c
-Al
2
O
3
(Fig. 4) revealed a uniform distribution of
-Al due to a proper choice of the support,
c
2 3
O
that is, the used alumina has an average pore size of 5.8 nm, so that
the nanoparticles can be easily accommodated within the support
pores. The presence of Al, O, Pd and Ir was confirmed on the sur-
face; however, due to the limited resolution, the bimetallic nano-
particles could not be differentiated either in a layered structure
of Pd and Ir atoms (core–shell) or randomly mixed together (alloy).
Fig. 5 shows the H
monometallic Pd, a peak at 80 °C can be ascribed to the reduction
in PdO species [62,63]. The TPR profile of monometallic Ir catalyst
shows a negative peak at about 190 °C, similar to the reported
observations for Ir/Al , when the apparent TPR profile includes
the early (room temperature) reduction, H adsorption and desorp-
tion [64]. When a physical mixture of the monometallic Ir and Pd
catalysts was inserted to the TPR cell, the TPR profile, expectedly,
shows the summation of the less intense signals for the monoPd
and monoIr. However, the TPR profiles of the Pd1Ir1 and Pd(c)Ir(s)
catalysts show different trend compared to their monometallic
counterparts, with no distinct hydrogen consumption for the Pd
component and a broad negative peak over the low temperature
region (75–300 °C) for the Pd(c)Ir(s). The change of the reducibility
characteristics of a specific metal upon alloying is a typical phe-
nomenon in bimetallic catalysts [55,64,65] and indicative of the
formation of new bimetallic clusters, not the physically mixed
monometallic particles.
2
-TPR profiles of the selected catalysts. For the
x
2 3
O
2 3
Fig. 2. TGA profiles of Ir/c-Al O pretreated at different conditions, i.e. (i) fresh, (ii)
calcined in air at 200 °C for 1 h, and (iii) calcined in air at 200 °C for 1 h followed by
the reduction in hydrogen at 359 °C for 1 h, as used for the catalytic tests. Some
2
2 3
predried c-Al O was also tested for comparison. TGA programme: ramp from room
2
temperature to 110 °C under N , heating from 110 °C to 600 °C under air with the
heating rate of 5 °C/min.
nanoparticles [61]. Catalyst calcination at 200 °C before the TGA re-
duces the mass loss only by 0.3 wt.%; when the calcined catalyst is
further pretreated in hydrogen, the mass loss during the TGA in-
creases to about 2.1 wt.%. Since the amount of PVP used for the
2 3
synthesis of Ir/c-Al O nanoparticles was about 3.5 wt.%, the oxi-
dation–reduction pretreatment performed in this study had a great
impact on reducing the PVP residues from the samples. It should be
mentioned that the weight increase due to the oxidation during
the calcination at 200 °C is negligible (<0.01 wt.%). Besides, as the
metal loadings in catalysts are very low (ꢀ0.1–0.3 wt.%), the metal
oxidation weight increase is negligible as compared to the total ob-
served ꢀ2–3 wt.% mass loss. The CHN analysis was also done to
XPS analysis was also performed on selected monometallic and
bimetallic catalysts. No chlorine residuals from the precursor were
found in XPS survey scan. The XPS spectra of the Pd 3d and Ir 4f are
presented in Table 1 and Figs. 6 and 7. The binding energy (BE) val-
ues of 334.6 and 335.7 eV for Pd 3d5/2 in monoPd/
assigned to the pure and oxide Pd species, respectively [66,67].
The BEs of 60.7 and 62.2 eV for Ir 4f7/2 in monoIr/ -Al O3 corre-
2
c-Al O3 can be
2 3
strengthen the TGA results. The C wt.% for c-Al O , fresh, calcined
and the calcined–reduced catalysts were found to be 0.04, 2.0, 1.9
and 1.1 wt.%, respectively, which again shows the role of oxida-
tion–reduction pretreatment to reduce the amount of PVP. The fi-
nal range in the TGA profile is above 400 °C, where similar mass
losses for all the samples (including alumina) occurred and most
c
2
spond to the reported numbers for pure and oxidised Ir [68,69],
which forms during the catalyst handling in air even at room tem-
perature [70]. The bimetallic samples exhibited shifts in the BE val-
ues as compared to their monometallic counterparts, which can
imply the metal alloying. Starting with Pd1Ir1, the BE values of
Pd 3d5/2 shifted to higher values of 335.4 and 336.2 eV for the
metallic and oxidised Pd, respectively. Ir 4f7/2 BE values shifted to
2 3
likely originated from c-Al O than from PVP. The TGA and CHN
analyses performed in this study suggest that the aforementioned
oxidation–reduction treatment can be successfully applied to the
fresh catalysts before the catalytic reaction to remove the protect-
ing polymer.
Fig. 3 presents the TEM images of the synthesised nanoparticles.
No large agglomerates (above 5 nm) were observed, and the nano-
particles are near-spherical with narrow size distribution. Mono-
metallic Ir nanoparticles are the smallest in size (1.6 ± 0.2 nm).
The bimetallic Pd2Ir1 and Pd1Ir1 particles (2.2 nm diameter) are
close in size to monometallic Pd (2.3 ± 0.4 nm) but with wider size
distribution (up to 1.0 nm) suggesting the bimetallic nature of the
nanoparticles instead of a physical mixture of monometallic Pd and
Ir particles. Pd(c)Ir(s) sample shows the largest average size of
lower values (60.6 and 61.4 eV for the metallic and oxidised Ir).
ꢃ1
These shifts are in line with electron affinities of Pd (54 kJ mol
)
ꢃ1
and Ir (112 kJ mol ), which are indicative of the electron-donating
properties of Pd in this bimetallic combination. Similar shift trends
exist for the Pd(c)Ir(s) catalyst, but they are less significant, most
likely due to the fact that the there are fewer Pd–Ir neighbours in
the core–shell combination as compared to the uniformly mixed
Pd1Ir1 nanoparticles. The area percentage of the metallic Pd peaks
are 47%, 58% and 67% of the total Pd peak areas for mono Pd, Pd1Ir1
and Pd(c)Ir(s), respectively, which imply the higher percentage of
metallic Pd in Pd(c)Ir(s) as it is protected by the shell Ir layer from
oxidation.
2
.7 ± 0.5 nm with narrow size distribution, indicating that the pre-
formed core Pd nanoparticles of 2.3 ± 0.4 nm diameter are coated
with Ir atoms. The applied synthetic procedure for the Pd(c)Ir(s)
nanoparticles ensures the desired structure, as the Pd nanoparti-
cles were synthesised first followed by Pd–H formation in
hydrogen atmosphere. The hydride serves as a catalyst for Ir reduc-
tion on the Pd nanoparticle surface. This procedure, known as a
hydrogen-sacrificial technique, is shown in Scheme 1. The
Supporting information contains detailed explanations on how
the stoichiometric amount of Ir precursor to build a monolayer
shell was determined.
Ion scattering spectroscopy (ISS) was also performed on mono-
and bimetallic catalysts (Fig. 8) to provide information on the outer
surface layer of the bimetallic formulations. Although quantitative
analysis is limited by the uncertainty of the inelastic losses and the
neutralisation rate depending on ion trajectories [71], the spectra
indicate roughly similar amounts of Pd and Ir in Pd1Ir1 sample
and significant abundance of Ir (ꢀfivefold) in the outer shell of
Pd(c)Ir(s) catalyst. This result confirms that Ir was successfully
deposited on Pd nanoparticles following the hydrogen-sacrificial
technique (Scheme 1).

1
18
H. Ziaei-azad et al. / Journal of Catalysis 300 (2013) 113–124
Fig. 3. TEM images with average size and one standard deviation for the various Pd and Ir-containing nanoparticles: (a) Pd, (b) Ir, (c) Pd2Ir1, and (d) Pd(c)Ir(s). The image for
Pd1Ir1 is not shown; the size is 2.3 ± 1.0 nm. The scale bars correspond to 20 nm.
compared both to Pd foil and to Pd nanoparticles is significant
(
0.4–0.5 eV for Pd nanoparticles), and it is larger than the instru-
mental error (the beamline resolution is 0.3 eV, and 0.2 eV step size
was used in the scan). This observation may again confirm the
bimetallic structure of the Pd–Ir catalysts, instead of physical
mixtures of monometallic Pd and Ir nanoparticles.
From Fig. 9, one can compare the intensities by calculating the
peak areas, which vary in the order of Pd1Ir1 > Pd2Ir1 >
Pd(c)Ir(s) > Pd nanoparticles. This can be interpreted as the Pd in
Pd1Ir1 having relatively more positive charge, while that of the
Pd(c)Ir(s) seems to be more metal-like. The electron affinity of Ir
(
112 kJ/mol) is higher than that of Pd (54 kJ/mol), indicating elec-
tron-accepting properties of Ir, which result in more positive Pd
in bimetallic structures. This phenomenon is more pronounced in
Pd1Ir1 probably due to the higher Ir-to-Pd ratio in this sample as
compared to Pd2Ir1. Among the bimetallic structures, the least in-
tense edge jump is observed for Pd(c)Ir(s), even though its Ir-to-Pd
ratio is in between those for Pd1Ir1 and Pd2Ir1 samples. It is likely
that the Pd–Ir contact in Pd1Ir1 and Pd2Ir1 is more profound than
that of the core–shell structure, and the formation of alloy struc-
ture is more likely for the Pd1Ir1 and Pd2Ir1 samples (herein, un-
der the alloy structure, a nanoparticle composed of mixed Pd–Ir
atoms in all nanoparticle shells is assumed). One the other hand,
the lower white-line intensity of the Pd(c)Ir(s) compared to Pd2Ir1
(about 10% less) may be similarly attributed to the lower contact of
Pd–Ir in the former. Indeed, for the core–shell structure, only the
surface Pd atoms of the Pd core are in contact with Ir shell atoms,
resulting in higher ratio of Pd–Pd bonds compared with the Pd–Ir
ones. As a result, the XANES profile of this structure looks more like
that of the Pd monometallic nanoparticles, but the peak is still
shifted towards higher energy values (the same as for alloy PdIr
catalysts), indicating the nanoparticle bimetallicity. These results
may suggest that the hydrogen-sacrificial strategy has been effec-
tive in producing the targeted Pd(c)Ir(s) nanoparticle structure and
that the simultaneous reduction resulted in bimetallic structures
Scheme 1. Formation of Pd(c)Ir(s) nanoparticles by hydrogen-sacrificial reduction
method.
3
Pd L -edge XANES spectra were also recorded; the nanoparticle
monodispersity made them excellent samples for this analysis. All
the spectra have been normalised using the Athena software, so
that the edge jump intensity can be compared directly. The peak
on the rising edge of the XANES profiles in Fig. 9 (also refer as
‘
white line’) is due to excitations of Pd 2p1/2 electrons to the unoc-
cupied Pd 4d orbitals, and its intensity is proportional to the degree
of valence electrons’ occupancy of the absorber [72]. The change in
the white-line intensity and the peak position is reported as a
possible indication of the change in the oxidation state of the
absorbing metal [73]. As seen in Fig. 9, all three presumably bime-
tallic formulations exhibit peaks at higher energies as compared to
monometallic Pd foil and monometallic Pd nanoparticles, as well as
larger peak intensities, indicating higher positive charge on Pd in
bimetallic catalysts (it takes more energy to remove Pd 2p
electrons from oxidised Pd). The shift to higher energy values as
instead of
a physical mixture of monometallic Pd and Ir
nanoparticles.
Thus, monometallic Ir and Pd as well as bimetallic PdIr
nanoparticles were synthesised with narrow size distribution and

H. Ziaei-azad et al. / Journal of Catalysis 300 (2013) 113–124
119
2 3
Fig. 4. The STEM image with Pd and Ir EDX profiles of the Pd(c)Ir(s) nanoparticles supported on c-Al O .
the Pd monometallic nanoparticles are not active for the RO of in-
dan. In contrast, the Ir monometallic catalyst shows the highest
conversion (99%) among the synthetic nanoparticles. The presence
of Ir in the bimetallic catalysts ensures their higher activity as com-
pared to pure Pd. Notably, as seen from Table 2, similar Ir loadings
in the reactor (in the range of 0.6–0.84 mg) lead to a range of con-
versions from 9% to 91% depending on the bimetallic structure,
which makes the possibility of controlling the structure being of
crucial importance. The highest conversion for a bimetallic catalyst
is achieved with Pd(core) and Ir(shell) structure. Our working
hypothesis was that the Ir atoms forming one outer layer around
a Pd core display similar catalytic properties as the surface atoms
on the monometallic Ir nanoparticles. The atmospheric ring open-
ing of naphthenic rings is a metal-driven reaction, which is essen-
tially controlled by the accessibility of the surface metal [44]. The
f.c.c. cuboctahedral crystal surface statistics was applied for the Pd
and Ir monometallic particles with 2.3 nm and 1.6 nm diameters to
find the metal dispersion as 46% and 61%, respectively [74], which
allowed calculating the activity per surface Ir atoms of the mono-
metallic particles (Table 2). For the Pd(core) Ir(shell) particles,
the statistics were applied for the core Pd particles covered with
one extra layer of Ir atoms. The calculation details are presented
in the Supporting information; the amount of Ir in the sample is
enough (or nearly enough) to build one full Ir shell around the
Pd core, which results in 100% Ir dispersion. Based on the found
2 2
Fig. 5. H -TPR profiles of the selected catalysts in 10 ml/min mixture of 10% H in
Ar (heating rate: 5 °C/min).
Table 1
XPS binding energy values of Pd and Ir in selected samples after the
calcination–reduction pretreatment.
Catalyst
2 3
support: c-Al O )
Binding energy (eV)
Zero-oxidation state
(%)
(
Pd 3d5/2
Ir 4f7/2
Pd
Ir
334.6a
335.4
334.7
335.7b
336.2
335.9
dispersions of monometallic Ir and Pd(c)Ir(s), the activities per sur-
Pd
47
60.6 61.4 58
60.0 61.3 67
60.7 62.2
ꢃ2
face Ir atoms were found as 3.6 ꢂ 10
gindan/gsurface Ir/min irre-
Pd1Ir1
Pd(c)Ir(s)
Ir
55
59
49
spectively of the catalyst mono- or bimetallicity, indicating the
absence of either antagonism or synergism between the two met-
als arranged in the core–shell structure with one layer of Ir atoms.
External Pd core atoms, located in the close proximity to the cov-
ering Ir atoms, which would allow for the electron transfer be-
tween the metals, do not affect the Ir atoms electronic properties
enough to change their activity in indan ring opening. In this bime-
tallic structure, the role of a second metal (Pd) is only in being an
inert support allowing 100% dispersion of the active Ir metal.
Other bimetallic ‘alloy’ structures Pd1Ir1 and Pd2Ir1 showed
intermediate activities between monometallic Pd and Ir catalysts.
Due to similar sizes (2.3 nm and 2.2 nm vs. 2.3 nm for monometal-
lic Pd), the nanoparticle dispersion in all the catalysts is 46% (see
Supporting information for the calculation details). As a working
a
Values correspond to 0-valent and oxidised metal, as seen in Figs. 6 and 7.
Values correspond to 0-valent and oxidised metal, as seen in Figs. 6 and 7.
b
controlled structure, which makes them invaluable tool in finding
correlations between the bimetallic nanoparticle compositions
with its catalytic properties without possible size effects.
3
.2. Catalytic behaviour in indan ring opening
The indan conversion, single cleavage selectivity and product
distribution are presented in Fig. 10a, b and Tables 2 and 3. As seen,

1
20
H. Ziaei-azad et al. / Journal of Catalysis 300 (2013) 113–124
2 3
Fig. 6. Pd 3d XPS spectra of the selected c-Al O -supported catalysts: (a) Pd, (b) Pd1Ir1 and (c) Pd(c)Ir(s).
Fig. 7. Ir 4f XPS spectra of the selected
c
-Al
2
O
3
-supported catalysts: (a) Ir, (b) Pd1Ir1 and (c) Pd(c)Ir(s).
hypothesis, a uniform mixing of Pd and Ir atoms in the nanoparti-
cles was assumed, implying that there is no Ir abundance either on
the surface or in the core of the nanoparticles. With these assump-
tions, the activities per surface Ir atoms were found as being
deficiency in the particle shell, that is, Pd and Ir non-uniform dis-
tribution in the bimetallic structure; (ii) Pd inhibiting effect via
electronic transfer, when Ir atoms have relatively high coordina-
tion number with respect to Pd atoms, which amount is the highest
among the all catalysts and (iii) surface dilution of Ir atom ensem-
bles with Pd atoms which hinders indan chemisorption. The first
hypothesis is line with the literature for Pt–Ir reforming catalysts,
which as shown by X-ray diffraction and EXAFS, have a platinum-
rich surface and an iridium-enriched core [75]. Palladium, similar
to platinum, with lower heat of sublimation than iridium (about
ꢃ2
ꢃ2
4
.4 ꢂ 10
and 0.9 ꢂ 10
gindan/gsurface Ir/min for Pd1Ir1 and
Pd2Ir1 catalysts, respectively. The first value may be considered
as being in line with the surface Ir activity in the monometallic
or core–shell nanoparticles and confirms the hypothesis of the ab-
sence of Pd promoting effect on Ir. However, the second value can
reflect any of the following scenarios (or their combinations): (i) Ir

H. Ziaei-azad et al. / Journal of Catalysis 300 (2013) 113–124
121
2 3
Fig. 8. Ion scattering spectra of the selected c-Al O -supported catalysts: (a) Pd, (b) Ir, (c) Pd1Ir1 and (d) Pd(c)Ir(s).
The third hypothesis on the Ir surface dilution is viable if we
consider possible reaction mechanisms. Three mechanisms have
been proposed for the RO over metal-supported catalysts [76].
First, the dicarbene mechanism, which results in the cleavage of
unsubstituted secondary-secondary carbon atoms, necessitates
the metal–carbon bonding of two carbon atoms standing perpen-
dicular on the metal surface. The second mechanism, known as
p-adsorbed olefin mode, requires the flat adsorption of three
neighbouring carbon atoms, which can achieve cleavage at substi-
tuted C–C positions. And, a third mechanism competes with the
dicarbene mechanism and operates via a metallocyclobutane inter-
mediate containing a metal atom and four carbon atoms, which al-
lows the opening of substituted C–C bonds if an external methyl
group is involved in the intermediate [76]. This mechanism, as re-
ported, has higher activation energy than the dicarbene and can
operate only when the dicarbene path is blocked [76]. The RO
mechanism of alkyl and/or aromatic branched naphthenic rings
greatly depends on the nature of the metal as well as its particle
size and structure [45]. In general, RO on metals like Ir, Ru, Rh
and Pt–Ir mostly happen only on the unsubstituted carbons, while
on others such as Pt, Pd, Pt–Pd and Pt–Ni, substituted ones are usu-
ally involved as well [42,45]. A very comprehensive study on the
RO of several naphthenic molecules using different noble metals
can be found in the literature [44]. Among the different noble met-
als investigated in the literature, iridium reportedly displays the
strongest tendency to break unsubstituted C–C bonds via the dicar-
bene mechanism [76].
Fig. 9. Pd L
Al O . Pd foil was used as the reference.
2 3
3
-edge XANES spectra of the selected nanoparticles supported on c-
1
60 kcal/mol for Ir, 135 for Pt and 89 for Pd) has the lowest surface
energy, and so, it is more preferable to concentrate on the bimetal-
lic particle surface. However, in this case, the Pd1Ir1 catalyst would
not exhibit its high activity, which under the assumption of uni-
form Pd and Ir mixing in Pd1Ir1 catalyst corresponds well to the
activity of surface Ir atoms in monoIr and Pd(c)Ir(s) catalysts. In
addition, the single cleavage selectivity decreases with the increase
of surface iridium in the samples (Table 2). These results are in
good agreement with the well-documented propensity of iridium
for hydrogenolysis and cracking [4,37,44]. So, the first hypothesis
may be declined for the presented PdIr catalysts; it is not excluded
that after ageing and/or prolong exposure to the high-temperature
reaction conditions, the core enrichment with Ir may occur leading
to the loss of activity. Similarly, the Pd inhibiting effect via electron
transfer is unlikely due to the similarity of Ir atom properties lo-
cated on monoIr, Pd1Ir1 and Pd(c)Ir(s) nanoparticles.
From Table
3
presenting the product distribution, 2-
ethyltoluene is the major product, while the portion of
n-propylbenzene is much lower for all the samples, which is in line
with the dicarbene mechanism [44,45,76]. The alternative olefin
mode mechanism cannot be used to explain the presented results
due to low selectivity to n-propylbenzene. The dicarbene mode
requires two contiguous Ir atoms on the surface, and assuming
one active site consisting of two neighbouring Ir atoms, a Pd1Ir1
catalyst with 1:1 surface ratio of Pd-to-Ir atoms will display similar

1
22
H. Ziaei-azad et al. / Journal of Catalysis 300 (2013) 113–124
1
-to-1 which is 0.8-to-1 according to the NAA results, so more Ir
ensembles exist on the catalyst surface resulting in higher specific
activity of 8.8 vs. 7.2 for the monoIr. However, the Pd2Ir1 catalyst
with 2-to-1 Pd-to-Ir surface atom ratio possesses the lower propor-
tion of two-Ir-atom ensembles, which correlates with much lower
activity of the Pd2Ir1 catalyst. The single cleavage selectivity (S) for
all catalysts is also governed by Ir, irrespectively to its surround-
ings, and follows the conversion (X) trend: 73%
Pd2Ir1) ꢀ 75% S (48% X, monoIr) > 63% S (61% X, Pd1Ir1) > 52% S
91% X, Pd(c)Ir(s)). Palladium’s presence does not contribute to
the cetane improvement number either: the n-propylbenzene-to-
-ethyltoluene ratio follows the same conversion trend being
0.18 up to 50% conversion and ꢀ0.13 up to 90% conversion
Table 2). Similarly, no significant improvement in the yield of sin-
S (9% X,
(
2
ꢀ
(
gle cleavage products obtained when Pt was added to Ir [37,77],
which may point out the role of Pt only as the diluting material.
This again confirms the Pd role only as a Ir surface diluting (for al-
loy catalysts) or dispersing (for core–shell) agent, with the catalytic
properties governed by Ir active sites.
The existence of synergism in bimetallic formulations was
widely reported for MCP ring opening. For instance, the MCP ring
opening using Pt–Ru catalysts revealed a synergetic effect for
Pt–Ru catalysts prepared by the amine precursors, which was
explained as the formation of Pt–Ru dual surface sites [78]. In a
recent study, a synergetic behaviour was reported for the RO of
MCP over Pt–Ir/
the conversion sixfold higher than Pt/
than Ir/ -Al . Such behaviour was attributed to the formation
c
-Al
2
O
3
, where the bimetallic catalyst presented
c
-Al but twofold lower
2 3
O
c
2 3
O
of new active sites between Pt and Ir [43]. However, the synergetic
effects were not completely understood in both cases due to the
incomplete knowledge over the structure of particles obtained
via co-impregnation. Finally, in another recent study, Pt–Rh/
2 3
Al O catalysts were applied for the RO of MCP. The study showed
Fig. 10. Indan conversion (a) and single cleavage selectivity (b) vs. time-on-stream
over different -Al -supported nanoparticles at 336 °C and 0.1 MPa: Pd nano-
that only the catalysts synthesised by refilling method (surface re-
dox reaction), that possess large particles of greater than 1.7 nm,
presented the improved activity and modified selectivity com-
c
2 3
O
particles (d), Ir nanoparticles (ꢀ), Ir nanoparticles (half amount) (}), Pd2Ir1 (4),
Pd1Ir1 (N), Pd(c)Ir(s) (h), and industrial Pt–Ir catalyst (ꢂ).
2 3
pared to Pt/Al O [79]. Surface characterisation and cyclohexane
dehydrogenation probe reactions confirmed the presence of alloy
structures for the particles of larger than 1.7 nm, and it was shown
that under a hydrogen atmosphere, the surface of the catalyst is
enriched with Pt [80]. Bimetallicity also alters the selectivity of
RO reactions. For instance, the incorporation of Pt into Ir moderates
the cracking behaviour of iridium in the SRO of benzocyclopentane
(indan) and improves the selectivity towards the single cleavage
ꢃ2
activities per each surface Ir atom (Table 2 and 44 ꢂ 10
g
indan
/
-
ꢃ2
g
surface Ir/min) or per ensemble of two surface atoms (8.8 ꢂ 10
g
indan/gIr active site/min) as monometallic Ir or Pd(c)Ir(s) nanoparti-
ꢃ2
cles (3.6 ꢂ 10
indan
g /
ꢃ2
gsurface Ir/min or 7.2 ꢂ 10
gindan/gIr active site/min). The small differ-
ence may be attributed to the rounding up the Pd-to-Ir ratio to
Table 2
Catalytic performance of the synthesised and industrial catalysts in atmospheric indan ring opening at 336 °C.
Catalyst
support:
Al
Nano-
particle size
(nm)
Pd/Ir
loadings
(wt.%)
Conversionc
(%)
Single cleavage
selectivity/yield
(%)
Ir loading in
the reactor
(mg)
Ir
Surface Ir in
the reactor
(mg)
Activity
PB/ET
b
c
dispersiond
(%)
ꢃ2
(
c
-
(10
g
indan
/
ratio/PB
yielde
2
O
3
)
gsurface Ir/min)
Pd
Ir
2.3 ± 0.4
1.6 ± 0.2
1.6 ± 0.2
0.160/0
0/0.178
0/0.178
1
99
48
N/A
10/10
75/36
N/A
1.20
0.60
N/A
61
61
N/A
0.73
0.37
–
3.8
3.6
–
<0.09/<1
0.19/6
Ir (half
amounta)
Pd2Ir1
2.2 ± 0.8
0.091/
9
73/7
0.62
46
0.29
0.9
0.17/1
0
.098
0.047/0.11 61
Pd1Ir1
Pd(c)Ir(s)
2.3 ± 1.0
2.7 ± 0.5
63/39
52/48
0.84
0.71
46
100
0.39
0.71
4.4
3.6
0.11/4
0.15/6
0.269/
.385
N/A
91
0
Industrial Pt–
Ir
N/A
62
47/29
N/A
N/A
N/A
N/A
0.15/4
a
b
c
0
.6 mg of Ir in the reactor vs. 1.2 mg of the active metal(s) in all other reactions including for the industrial catalyst.
The TEM diameter of the as-prepared nanoparticles.
Average values between 120 and 210 min on stream.
The calculation details can be found in the Supporting information.
PB: n-propylbenzene, ET: 2-ethyltoluene.
d
e

H. Ziaei-azad et al. / Journal of Catalysis 300 (2013) 113–124
123
Table 3
Product distribution at 120 min on stream.
Catalyst (support:
c
-Al
2
O
3
)
Product distribution (wt.%)
2
-Ethyltoluene
n-Propylbenzene
o-Xylene
Ethylbenzene
Toluene
Benzene
Lights
Ir
11
31
6
36
41
26
<1
6
1
4
6
45
7
1
16
20
12
4
1
<1
2
4
2
22
3
<1
3
10
8
2
<1
<1
<1
1
17
2
<1
4
8
6
Ir (half amount)a
Pd2Ir1
Pd1Ir1
Pd(c)Ir(s)
Industrial Pt–Ir
4
1
a
0
.6 mg of Ir in the reactor vs. 1.2 mg of the active metal(s) in other runs; Pd is not included in the comparison due to low conversion.
2 3 2 3
Fig. 11. TEM photographs of (a) Ir/c-Al O after the reaction, 210 min on stream; (b) Pd(c)Ir(s)/c-Al O after the reaction, 210 min on stream.
products [55]. Or, the addition of Ni to Ir/Al
selectivity to higher octane number products in the RO of
,3-dimethylcyclohexane (DMCH) [48]. However, all these findings
2
O
3
may improve the
n-propylbenzene (Table 2), which is beneficial in terms of cetane
number improvement.
1
were reported for the catalysts prepared via traditional impregna-
tion techniques, which result in the active metal polydispersity and
hinders the ‘one-particle’ size effect studies [6].
4. Conclusions
The important significance of the current work is that by con-
trolling the size and structure, including the surface structure, of
mono- and bimetallic nanoparticles, we were able to distinguish
two metal roles and establish the absence of any other Pd function-
ing apart from being an inert diluting or dispersing agent for active
Ir. Our findings do not contradict the observed synergism hypoth-
eses but allow verifying whether the intrinsic synergism exists. In-
deed, according to the Table 2, similar Ir loadings in the reactor (in
the range of 0.6–0.8 mg) lead to a range of conversions from 9% to
The engineered synthesis of Pd, Ir and Pd–Ir nanosised catalysts
of 1.6–2.7 nm range enabled us to get insights into the Pd and Ir
roles in the sulphur-free atmospheric ring opening of indan. Irre-
spective of the catalyst structure and composition, as soon as
two contiguous surface Ir atoms exist on the nanoparticle surface,
they display the same catalytic properties, most likely through the
dicarbene ring-opening mechanism. Palladium serves only either
as a dispersing agent in Pd(core)–Ir(shell) structures providing
100% Ir dispersion or as a surface diluting inert metal in Pd–Ir for-
mulations with mixed nanoparticle surface. The surface composi-
tion should be carefully controlled so that the Pd-to-Ir atom ratio
on the surface does not exceed 1:1 to allow the existence of the
two contiguous Ir atoms. No intrinsic synergism or antagonism be-
tween the two metals was found, while the conversion improve-
ment (observed synergism) for some bimetallic formulations was
seen in the catalytic runs. The ability to control and know the
bimetallic nanoparticle surface structure is an indispensible tool
in verifying whether intrinsic synergism exists between the two
metals.
9
1% for bimetallic structure, which could be assigned to antago-
nism and/or synergism between Pd and Ir. The possibility to con-
trol the nanoparticle structure, thus, to know its surface
composition allowed us to reject the intrinsic synergism hypothe-
sis and assign the observed synergism to the variations in active
site architecture.
From the catalyst application point of view, for the same metal
loading in the reactor, that is, 1.2 mg of active metals(s), the
synthesised Pd(c)Ir(s) nanoparticles display the highest selective
ring-opening yield (48%) among the various samples in this study,
with a yield of about 19% higher than the commercial Pt–Ir catalyst
(
Table 2). As mentioned, a series of in-house Pt–Ir catalysts has
The hydrogen-sacrificial technique of liquid-phase colloid syn-
thesis may allow the production of a very active and Pt-free cata-
lyst formulation, i.e., Pd(c)Ir(s), which can be further examined in
a sulphur feed under industrial conditions. This work is currently
under way. With Ir being twice as expensive as Pd and one of
the rarest elements on Earth, its 100% dispersion makes the
Pd(c)Ir(s) catalyst preferable to the monometallic Ir. An increase
of 19% in the yield of desired products was observed with the
Pd(c)Ir(s) compared to a commercial Pt–Ir bimetallic catalyst,
been tested recently for the RO of indan at low pressure [55,77];
however, no significant improvement in the yield of single
cleavage products was observed after the addition of Pt to Ir. Our
Ir and Pd(c)Ir(s) nanoparticles also showed the absence of agglom-
eration under the reaction conditions (Fig. 11). Although the ratio
of n-propylbenzene-to-2-ethyltoluene for the developed catalysts
is the same as for the industrial catalyst, due to higher single
cleavage selectivity, the Pd–Ir catalysts allow a higher yield of

1
24
H. Ziaei-azad et al. / Journal of Catalysis 300 (2013) 113–124
which may suggest Pd(c)Ir(s) as a potential candidate to reduce the
consumption of platinum in refineries.
[32] N. Toshima, Pure Appl. Chem. 72 (2000) 317.
[
[
[
33] N. Toshima, K. Hirakawa, Polym. J. 31 (1999) 1127.
34] S.J. Cho, S.K. Kang, Catal. Today 93 (2004) 561.
35] L. DSouza, S. Sampath, Langmuir 16 (2000) 8510.
Acknowledgments
[36] J.H. Sinfelt, Catal. Today 53 (1999) 305.
[
37] U. Nylen, L. Sassu, S. Melis, S. Jaras, M. Boutonnet, Appl. Catal.
2006) 1.
38] M. Kangas, D. Kubicka, T. Salmi, D.Y. Murzin, Top. Catal. 53 (2010) 1172.
A 299
(
Financial support from the Imperial Oil – Alberta Ingenuity Cen-
ter for Oil Sands Innovation at the University of Alberta is grate-
fully acknowledged. STEM–EDX analysis was performed at the
Centre for Nanostructure Imaging, Department of Chemistry, Uni-
versity of Toronto; NAA at the Becquerel Laboratories Inc. in Miss-
issauga, Ontario. Dr. Y. Maham and Dr. S. Merali are acknowledged
for the help with TGA and AAS.
[
[39] R.C. Santana, P.T. Do, M. Santikunaporn, W.E. Alvarez, J.D. Taylor, E.L. Sughrue,
D.E. Resasco, Fuel 85 (2006) 643.
[
[
40] F. Garin, P. Girard, F. Weisang, G. Maire, J. Catal. 70 (1981) 215.
41] W.E. Alvarez, D.E. Resasco, J. Catal. 164 (1996) 467.
[42] D. Teschner, K. Matusek, Z. Paal, J. Catal. 192 (2000) 335.
[
[
43] A. Djeddi, I. Fechete, F. Garin, Appl. Catal. A 413–414 (2012) 340.
44] G.B. McVicker, M. Daage, M.S. Touvelle, C.W. Hudson, D.P. Klein, W.C. Baird Jr.,
B.R. Cook, J.G. Chen, S. Hantzer, D.E.W. Vaughan, E.S. Ellis, O.C. Feeley, J. Catal.
210 (2002) 137.
[
[
[
45] H. Du, C. Fairbridge, H. Yang, Z. Ring, Appl. Catal. A 294 (2005) 1.
46] F. Luck, J.L. Schmitt, G. Maire, React. Kinet. Catal. Lett. 21 (1982) 219.
47] C.G. Walter, B. Coq, F. Figueras, M. Boulet, Appl. Catal. A 133 (1995) 95.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2013.01.004.
[48] S. Dokjampa, T. Rirksomboon, D.T.M. Phuong, D.E. Resasco, J. Mol. Catal. A:
Chem. 274 (2007) 231.
[
49] S. Lecarpentier, J.V. Gestel, K. Thomas, J.-P. Gilson, M. Houalla, J. Catal. 254
2008) 49.
[50] R. Moraes, K. Thomas, S. Thomas, S.V. Donk, G. Grasso, J.-P.M. Houalla, J. Catal.
86 (2012) 62.
(
References
2
[
[
[
51] D. Kubicka, N. Kumar, P. Maki-Arvela, M. Tiitta, V. Niemi, H. Karhu, T. Salmi,
D.Yu. Murzin, J. Catal. 227 (2004) 313.
52] D. Kubicka, M. Kangas, N. Kumar, M. Tiitta, M. Lindblad, D.Yu. Murzin, Top.
Catal. 53 (2010) 1438.
[
[
1] G.C. Bond, A.F. Rawle, J. Mol. Catal. A: Chem. 109 (1996) 261.
2] L. Guczi, Z. Schay, Gy. Stefler, L.F. Liotta, G. Deganello, A.M. Venezia, J. Catal. 182
(
1999) 456.
[
[
3] H.R. Aduriz, P. Bodnariuk, B. Coq, F. Figueras, J. Catal. 129 (1991) 47.
4] J.L. Carter, G.B. McVicker, W. Weissman, W.S. Kmak, J.H. Sinfelt, Appl. Catal. 3
53] J.-W. Park, K. Thomas, J.V. Gestel, J.-P. Gilson, C. Collet, Appl. Catal. A 388
(
2010) 37.
(
1982) 321.
[
[
[
54] D. Kubicka, T. Salmi, M. Tiitta, D.Yu. Murzin, Fuel 88 (2009) 366.
55] U. Nylen, J.F. Delgado, S. Jaras, M. Boutonnet, Appl. Catal. A 262 (2004) 189.
56] J.G. Van Senden, E.H. Van Broekhoven, C.T.J. Wreesman, V. Ponec, J. Catal. 87
[
[
[
5] B. Coq, F. Figueras, J. Mol. Catal. A: Chem. 173 (2001) 117.
6] N. Semagina, L. Kiwi-Minsker, Catal. Rev. 51 (2009) 147.
7] N. Toshima, H. Yan, Y. Shiraishi, in: G. Schmid, N. Toshima, B. Corain (Eds.),
Metal Nanoclusters in Catalysis and Material Science. The Issue of Size Control,
Elsevier Science, Amesterdam, 2008, p. 49.
(
1984) 468.
[
[
[
[
[
57] T. Miyake, M. Teranishi, Chem. Mater. 10 (1998) 594.
58] X. Yan, H. Liu, K.Y. Liew, J. Mater. Chem. 11 (2001) 3387.
59] X. Huang, Y. Wang, X. Liao, B. Shi, Chem. Commun. (2009) 4687.
60] S. Link, M.A. El-Sayed, J. Phys. Chem. B 103 (1999) 8410.
61] R.M. Rioux, H. Songa, M. Grass, S. Habas, K. Niesz, J.D. Hoefelmeyer, P. Yang,
G.A. Somorjai, Top. Catal. 39 (2006) 167.
62] S.K. Matam, E.H. Otal, M.H. Aguirre, A. Winkler, A. Ulrich, D. Rentsch, A.
Weidenkaff, D. Ferri, Catal. Today 184 (2012) 237.
63] N.S. Babu, N. Lingaiah, R. Gopinath, P.S.S. Reddy, P.S.S. Prasad, J. Phys. Chem. C
[
[
8] P. Lu, T. Teranishi, K. Asakura, M. Miyake, N. Toshima, J. Phys. Chem. B 103
(
1999) 9673.
9] P. Lu, N. Toshima, Bull. Chem. Soc. Jpn. 73 (2000) 751.
[
[
10] N. Toshima, Y. Wang, Langmuir 10 (1994) 4574.
11] C.-R. Bian, S. Suzuki, K. Asakura, L. Ping, N. Toshima, J. Phys. Chem. B 106
[
[
(
2002) 8587.
12] N. Toshima, T. Yonezawa, K. Kushihash, J. Chem. Soc., Faraday Trans. 89 (1993)
537.
13] N. Toshima, Y. Shiraishi, T. Teranishi, M. Miyake, T. Tominaga, H. Watanabe, W.
Brijoux, H. Bonnemann, G. Schmid, Appl. Organomet. Chem. 15 (2001) 178.
14] L.M. Bronstein, D.M. Chernyshov, I.O. Volkov, M.G. Ezernitskaya, P.M. Valetsky,
V.G. Matveeva, E.M. Sulman, J. Catal. 196 (2000) 302.
[
[
[
2
111 (2007) 6447.
[
[
64] Y.-J. Huang, J. Xue, J.A. Schwarz, J. Catal. 111 (1988) 59.
65] J. Batista, A. Pintar, D. Mandrino, M. Jenko, V. Martin, Appl. Catal. A 206 (2001)
113.
[
[
[
66] K.S. Kim, A.F. Gossmann, N. Winograd, Anal. Chem. 46 (1974) 197.
67] A.B. Gaspar, L.C. Dieguez, Appl. Catal. A 201 (2000) 241.
68] R. Nyholm, A. Berndtsson, N. Martensson, J. Phys. C: Solid State Phys. 13 (1980)
L1091.
[
[
[
[
[
[
[
[
[
15] B. Zhao, N. Toshima, Chem. Express 5 (1990) 721.
16] S.K. Singh, Q. Xu, Chem. Commun. 46 (2010) 6545.
17] N. Toshima, K. Hirakawa, Appl. Surf. Sci. 121 (122) (1997) 534.
18] Y. Lou, M.M. Maye, L. Han, J. Luo, C.-J. Zhong, Chem. Commun. (2001) 473.
19] T.C. Deivaraja, J.Y. Lee, J. Electrochem. Soc. 151 (2004) A1832.
20] T. Endo, T. Yoshimura, K. Esumi, J. Colloid Interface Sci. 286 (2005) 602.
21] T. Endo, T. Kuno, T. Yoshimura, K. Esumi, J. Nanosci. Nanotechol. 5 (2005) 1875.
22] S.K. Ghosh, M. Mandal, S. Kundu, S. Nath, T. Pal, Appl. Catal. A 268 (2004) 61.
23] L. Guczi, A. Beck, A. Horvath, Z. Koppany, G. Stefler, K. Frey, I. Sajo, O. Geszti, D.
Bazin, J. Lynch, J. Mol. Catal. A: Chem. 204 (2003) 545.
[
[
[
69] F. Bernardi, J.D. Scholten, G.H. Fecher, J. Dupont, J. Morais, Chem. Phys. Lett.
4
79 (2009) 113.
70] F. Locatelli, B. Didillon, D. Uzio, G. Niccolai, J.P. Candy, J.M. Basset, J. Catal. 193
2000) 154.
(
71] J. Shen, X. Yin, D. Karpuzov, N. Semagina, Catal. Sci. Technol. doi: http://
dx.doi.org/10.1039/c2cy20443f.
72] J.A. Horsley, J. Chem. Phys. 76 (1982) 1451.
73] B.J. Hwang, C.-H. Chen, L.S. Sarma, J.-M. Chen, G.-R. Wang, M.-T. Tang, D.-G. Liu,
J.-F. Lee, J. Phys. Chem. B 110 (2006) 6475.
74] R.V. Hardeveld, F. Hartog, Surf. Sci. 15 (1969) 189.
75] Sinfelt, Bimetallic Catalysts: Discoveries, Concepts and Applications, John
Wiely & Sons, Inc., New York, 1983, p. 94.
76] P.T. Do, W.E. Alvarez, D.E. Resasco, J. Catal. 238 (2006) 477.
77] U. Nylen, B. Pawelec, M. Boutonnet, J.L.G. Fierro, Appl. Catal. A 299 (2006) 14.
78] G. Diaz, F. Garin, G. Maire, S. Alerasool, R.D. Gonzalez, Appl. Catal. A 124 (1995)
[
[
[
[
24] A.Q. Wang, C.M. Chang, C.Y. Mou, J. Phys. Chem. B 109 (2005) 18860.
25] X. Wang, W.B. Yue, M.S. He, M.H. Liu, J. Zhang, Z.F. Liu, Chem. Mater. 16 (2004)
7
99.
26] G. Schmid, H. West, J.O. Halm, J.O. Bovin, C. Grenthe, Chem. Eur. J. 2 (1996)
099.
27] Y. Mizukoshi, T. Fujimoto, Y. Nagata, R. Oshima, Y. Maeda, J. Phys. Chem. B 104
2000) 6028.
28] N. Toshima, Y. Shiraishi, A. Shiotsuki, D. Ikenaga, Y. Wang, Eur. Phys. J. D 16
2001) 209.
29] J. He, I. Ichinose, T. Kunitake, A. Nakao, Y. Shiraishi, N. Toshima, J. Am. Chem.
Soc. 125 (2003) 11034.
30] A.U. Nilekar, S. Alayoglu, B. Eichhorn, M. Mavrikakis, J. Am. Chem. Soc. 132
[
[
[
[
[
[
[
[
1
[
[
[
(
(
3
3.
79] P. Samoila, M. Boutzeloit, C. Especel, F. Epron, P. Marecot, Appl. Catal. A 369
2009) 104.
80] P. Samoila, M. Boutzeloit, C. Especel, F. Epron, P. Marecot, J. Catal. 276 (2010)
37.
[
[
(
(
2010) 7418.
2
31] S.J. Cho, R. Ryoo, Catal. Lett. 97 (2004) 71.