Comparative study of hydrotalcite-derived supported Pd2Ga and PdZn intermetallic nanoparticles as methanol synthesis and methanol steam reforming catalysts

Article abstract of DOI:10.1016/j.jcat.2012.05.020

An effective and versatile synthetic approach to produce well-dispersed supported intermetallic nanoparticles is presented that allows a comparative study of the catalytic properties of different intermetallic phases while minimizing the influence of differences in preparation history. Supported PdZn, Pd₂Ga, and Pd catalysts were synthesized by reductive decomposition of ternary Hydrotalcite-like compounds obtained by co-precipitation from aqueous solutions. The precursors and resulting catalysts were characterized by HRTEM, XRD, XAS, and CO-IR spectroscopy. The Pd²⁺ cations were found to be at least partially incorporated into the cationic slabs of the precursor. Full incorporation was confirmed for the PdZnAl-Hydrotalcite-like precursor. After reduction of Ga- and Zn-containing precursors, the intermetallic compounds Pd₂Ga and PdZn were present in the form of nanoparticles with an average diameter of 6 nm or less. Tests of catalytic performance in methanol steam reforming and methanol synthesis from CO₂ have shown that the presence of Zn and Ga improves the selectivity to CO₂ and methanol, respectively. The catalysts containing intermetallic compounds were 100 and 200 times, respectively, more active for methanol synthesis than the monometallic Pd catalyst. The beneficial effect of Ga in the active phase was found to be more pronounced in methanol synthesis compared with steam reforming of methanol, which is likely related to insufficient stability of the reduced Ga species in the more oxidizing feed of the latter reaction. Although the intermetallic catalysts were in general less active than a Cu-/ZnO-based material prepared by a similar procedure, the marked changes in Pd reactivity upon formation of intermetallic compounds and to study the tunability of Pd-based catalysts for different reactions.

Full text of DOI:10.1016/j.jcat.2012.05.020

Journal of Catalysis 293 (2012) 27–38
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Comparative study of hydrotalcite-derived supported Pd2Ga and PdZn
intermetallic nanoparticles as methanol synthesis and methanol steam
reforming catalysts
Antje Ota a, Edward L. Kunkes a, Igor Kasatkin a, Elena Groppo b, Davide Ferri c, Beatriz Poceiro d,
Rufino M. Navarro Yerga d, Malte Behrens a,
⇑
a Fritz-Haber-Institute of the Max-Planck-Society, Department of Inorganic Chemistry, Faradayweg 4-6, D-14195 Berlin, Germany
b University of Torino, Department of Inorganic, Physical and Material Chemistry, NIS Centre of Excellence and INSTM, Via P. Giuria 7, I-10125 Torino, Italy
c Empa, Swiss Federal Laboratories for Materials Science and Technology, Laboratory for Solid State Chemistry and Catalysis, Ueberlandstrasse 129, CH-8600 Dübendorf, Switzerland
d Instituto de Catálisis y Petroleoquímica (CSIC), Grupo de Energía y Química sostenibles (EQS), C/Marie Curie, 2, 28049, Madrid, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
An effective and versatile synthetic approach to produce well-dispersed supported intermetallic nano-
particles is presented that allows a comparative study of the catalytic properties of different intermetallic
phases while minimizing the influence of differences in preparation history. Supported PdZn, Pd2Ga, and
Pd catalysts were synthesized by reductive decomposition of ternary Hydrotalcite-like compounds
obtained by co-precipitation from aqueous solutions. The precursors and resulting catalysts were charac-
terized by HRTEM, XRD, XAS, and CO-IR spectroscopy. The Pd2+ cations were found to be at least partially
incorporated into the cationic slabs of the precursor. Full incorporation was confirmed for the PdZnAl-
Hydrotalcite-like precursor. After reduction of Ga- and Zn-containing precursors, the intermetallic com-
pounds Pd2Ga and PdZn were present in the form of nanoparticles with an average diameter of 6 nm or
less. Tests of catalytic performance in methanol steam reforming and methanol synthesis from CO2 have
shown that the presence of Zn and Ga improves the selectivity to CO2 and methanol, respectively. The
catalysts containing intermetallic compounds were 100 and 200 times, respectively, more active for
methanol synthesis than the monometallic Pd catalyst. The beneficial effect of Ga in the active phase
was found to be more pronounced in methanol synthesis compared with steam reforming of methanol,
which is likely related to insufficient stability of the reduced Ga species in the more oxidizing feed of the
latter reaction. Although the intermetallic catalysts were in general less active than a Cu-/ZnO-based
material prepared by a similar procedure, the marked changes in Pd reactivity upon formation of inter-
metallic compounds and to study the tunability of Pd-based catalysts for different reactions.
Ó 2012 Elsevier Inc. All rights reserved.
Received 16 March 2012
Revised 22 May 2012
Accepted 29 May 2012
Available online 17 July 2012
Keywords:
Methanol synthesis
Methanol steam reforming
Intermetallic compounds
Hydrotalcites
1. Introduction
nature of the active phase and with it the easy relation to real
powder catalysts has to be sacrificed. Recently, aerosol-derived
Comparative studies of different (bi- or inter)metallic phases in
nanostructured catalysts often suffer from limited comparability of
the catalytic materials themselves as a result of differences in their
preparation history. Application of the individually synthesis reci-
pes that have been optimized for the best performance of a given
catalyst system requires special and often unique conditions like
presence of certain ligands, solvents, support materials, promoters,
or thermal post-treatments. As a result, differences in composition,
dispersion, homogeneity, or metal-support contacts may compli-
cate the comparison of such nanocatalysts by convolution of extrin-
sic and intrinsic effects. Unsupported model catalysts are an
alternative [1], but with the support often also the nanostructured
alloy powders were presented as an elegant compromise and useful
platform for studying intermetallic metal phase in MSR and CO
oxidation on PdZn particles with a surface area of 6 m2 gꢀ1 [2].
While this approach successfully bridges the gap between mod-
el and real catalysts starting from the model side, we herein, pres-
ent a synthetic approach to intermetallic catalysts that allows
conserving the full complexity of supported nanocatalysts. At the
same time, this method assures high comparability of different
intermetallic compounds (IMCs) in a nanostructured form by
application of a common and flexible synthesis protocol to mini-
mize the role of different preparation history. This is achieved
using a facile aqueous co-precipitation technique to prepare
Hydrotalcite-like compounds (HTlc) with different elemental com-
binations as well-defined platform precursor materials for inter-
metallic compounds.
⇑
Corresponding author. Fax: +49 30 8413 4405.
E-mail address: behrens@fhi-berlin.mpg.de (M. Behrens).
0021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2012.05.020

28
A. Ota et al. / Journal of Catalysis 293 (2012) 27–38
The advantage of this precursor material is that divalent and tri-
catalysts unsuitable for portable applications. To that end, Pd sup-
ported on reducible oxides such as ZnO, In2O3, and Ga2O3 has
shown favorable reactivity similar to that of Cu-based catalysts
for both methanol synthesis and MSR [11], whereas Pd supported
on non-reducible supports has been shown to be selective to rWGS
and methanol decomposition. Additionally, Pd-based catalysts
have also shown good long-term stability and resistance to sinter-
ing. The alteration of the properties of Pd was attributed to the for-
mation of the IMCs Pd2Ga [12] and PdZn [11a,13] upon partial
reduction in the support components rendering these systems
ideal test cases for studying the potential of the HTlc precursor ap-
proach for IMC catalysts synthesis.
valent cations are uniformly distributed in slabs of edge-sharing
MO6 octahedra that allow a close interaction of all metal cations,
which is thought to be prerequisite to the uniform formation of
ꢀ
ꢁ
IMCs [3]. HTlcs exhibit the general composition M1II; M2II
1ꢀx
M3IIIðOHÞ ðCO3Þ ꢁ mH2O (0.25 6 x 6 0.33) and a huge flexibility
x
2
x=2
of reducible and non-reducible metal cations that can be incorpo-
rated into the structure, for example, MII = Mg2+, Mn2+, Co2+, Ni2+
,
Cu2+, Zn2+ and MIII = Al3+, Ga3+, Fe3+, Cr3+ [4]. This flexibility makes
a large number of element combinations accessible rendering HTlc
a platform precursor for many IMC systems. In this work, we use
the HTlc precursor approach to compare the role of different Pd-
based IMCs. We used Pd2+ as M1II and added reducible species,
Zn2+ and Ga3+ at the M2II and MIII sites, respectively, to obtain
the IMCs PdZn and Pd2Ga. It is noted that due to the charge distri-
bution the formed IMCs are sometimes denoted as ZnPd and
GaPd2. Here, we use in the field of catalysis more common form
PdZn and Pd2Ga for consistency with previous work.
2. Experimental
2.1. Synthesis conditions
Ternary palladium containing and binary Pd-free PdMgAl,
PdMgGa, PdZnAl, MgGa, ZnAl, and MgAl HTlc with M2+/M3+ molar ra-
tios of 70:30 were synthesized by co-precipitation. The nominal
composition of all Pd-M2+–M3+ samples was set to 1:69:30. A
mixed aqueous metal nitrate ([Pd2+] + [M2+] + [M3+] = 0.2 M) solu-
tion and 0.345 M basic precipitating agent solution were co-fed
at pH = 8.5. For MgGa and MgAl, HTlc precursors pure sodium car-
bonate solution and a precipitation temperature of 328 K were
used, whereas in case of PdZnAl HTlc a mixture of sodium carbon-
ate (0.3 M) and sodium hydroxide (0.045 M) and a temperature of
298 K has be applied in order to obtain an homogeneous precursor
sample. During precipitation, both solutions were added simulta-
neously dropwise into a 2-L precipitation reactor (Mettler-Toledo
LabMax). The nitrate solution was automatically pumped with a
constant dosing rate, and the basic solution was added to maintain
a constant pH of 8.5. After completion of addition, the mixture was
aged for 1 h at the same temperatures applied during synthesis.
The precipitate was filtered and washed twice with warm deion-
ized water in order to remove the nitrate and sodium ions and ob-
tain a conductivity of the filtrate lower than 0.2 mS/cm. The solid
was dried for 12 h at 353 K in air. After drying, a one-step decom-
position–reduction in 5 vol% H2/Ar (2 K/min) was performed at the
temperatures extracted from H2-TPR experiments that yielded in
the intermetallic phases and in metallic Pd in case of the PdMgAl
HTlc precursor. A reduction temperature of 523 K was applied for
the PdMgAl and PdZnAl system, whereas a reduction of 773 K is
needed to obtain Pd2Ga intermetallic particles.
We have shown recently [5] that Pd2Ga nanoparticles sup-
ported on an oxide matrix of MgO and MgGa2O4 can be obtained
by this synthetic approach and reported on their hydrogenation
properties. This method presents an efficient alternative to previ-
ously used top-down methods, such as etching or milling of mate-
rials obtained from high temperature melt synthesis, used
previously to synthesize this material [6]. In addition to catalyst
synthesis and characterization data from a variety of complemen-
tary techniques (XRD, BET, TPR, SEM, TEM, FTIR, and XAFS), we also
present catalytic performance data in two reactions involving
methanol, namely its synthesis from CO2 and methanol steam
reforming (MSR) to correlate structural with catalytic properties.
Methanol has been proposed as a promising energy storage
molecule for portable applications such as direct methanol fuel
cells, as well as fuel for automobile internal combustion engines
[7]. Additionally, methanol has been proposed as hydrogen storage
medium [8], whose gravimetric hydrogen density exceeds that of
compressed and even that of liquid hydrogen. Furthermore, the
use of CO2 as a carbon source for methanol synthesis enables the
simultaneous reduction in emissions of this greenhouse gas [7].
Methanol synthesis from CO2 (Reaction (1)) and its reverse
reaction, methanol steam reforming (MSR) comprise a class of
reactions critical to the application of methanol as an efficient en-
ergy carrier.
CO2 þ 3H2ꢀCH3OH þ H2O
CO2 þ H2ꢀCO þ H2O H0 ¼ þ41:2 kJ=mol
CH2OHꢀCO þ 2H2 H0 ¼ þ91:0 kJ=mol
D
H0 ¼ ꢀ49:8 kJ=mol
ð1Þ
ð2Þ
ð3Þ
A HTlc-based CuZnAl catalyst was prepared as described previ-
ously [14] and was used as reference material for the catalytic test
in methanol synthesis. Additional details about the synthesis con-
ditions of the CuZnAl HTlc precursor and the synthesis protocols of
the Pd-based HTlc are given in the supplementary information
(Fig. S1).
D
D
Both reactions are currently carried out at similar temperatures
(523–603 K) on Cu/ZnO-based catalysts. While MSR is not thermo-
dynamically limited and is carried out at atmospheric pressure,
methanol synthesis requires high pressures (30–100 bar) and low-
er operating temperatures due to unfavorable thermodynamics [9].
Reverse water gas shift (rWGS, Reaction (2)) is a side reaction to
methanol synthesis and thus diminishes selectivity. In MSR, meth-
anol decomposition (Reaction (3)) as well as rWGS are undesired
side reactions and yield CO-a poison for fuel cell electrodes. The
rates of both undesired reactions can be thermodynamically hin-
dered by operating at lower temperatures. Lower temperature
operation and low selectivity to rWGS or decomposition thus re-
main the driving forces behind the development and optimization
of methanol catalysts [9].
2.2. Characterization
X-ray diffraction. XRD patterns of the HTlc precursor and its
decomposition products were recorded on a STOE Stadi P diffrac-
tometer in transmission geometry using Cu Ka1 radiation, a pri-
mary Ge monochromater, and a 3° linear position sensitive
detector.
Specific surface area determination. Specific surface areas (SSA) of
the precursors and reduced compounds were determined by N2
adsorption–desorption measurements at 77 K by employing the
BET method (Autosorb-1C, Quantachrome). Prior to N2 adsorption,
the sample was outgassed at 353 K/423 K (precursor sample/re-
duced catalyst, respectively) to desorb moisture from the surface
and pores.
The current state of the industrial Cu/ZnO catalysts are well
suited to stationary operation, however suffer from pyrophoricity,
sintering with long reaction times, and instability to changes in
reaction conditions [10]. These shortcomings make Cu-based

A. Ota et al. / Journal of Catalysis 293 (2012) 27–38
29
Chemical analysis. About 5 mg was exactly weighted in and dis-
solved in 2 ml of aqua regia. The solutions were transferred and
filled up in 50-ml volumetric flasks. The content of the metals were
determined with ICP-OES (Vista RL, Varian) after matrix-matched
calibration.
Temperature-programmed reduction. The TPR experiments were
performed in a fixed-bed reactor using 400 mg of precursor
(100–250 lm sieve fraction). PdMgGa and PdMgAl precursors were
reduced in 5 vol% H2 in Argon (100 ml/min) with a heating rate of
2 K/min up to a sample temperature of 773 K. The temperature was
kept constant for 4 h. The PdZnAl precursor was heated to 1073 K
(2 K/min). The hydrogen consumption was monitored with a ther-
mal conductivity detector.
X-ray absorption fine structure. XAFS data were collected in the
XANES (X-ray absorption near edge fine structure) and EXAFS (Ex-
tended X-ray absorption fine structure) regions at the Pd K edge
(24.350 keV) were carried out at the Super-XAS beamline of the
Swiss Light Source (Villigen, Switzerland) in the transmission
mode, using ionization chambers as detectors and a Si(3 1 1) Quick
EXAFS monochromator. A Pd foil placed between the second and
third ionization chamber was measured simultaneously as an
internal reference. In situ XANES measurements during hydrogen
TPR were performed in transmission mode, using a 3-mm capillary
operated at 300 kV. With computer-assisted correction and align-
ment, the value of the spherical aberration constant Cs was kept
below 100 nm in the present experiments. Selected areas of the
high-resolution images have been Fourier transformed to obtain
Power spectra, and the lattice distances and angles were measured
for phase identification (accuracy 1% and 0.5°, respectively). Me-
tal dispersions for Pd-based catalysts were obtained by D = 1.1/d,
where d is the average particle size determined by TEM neglecting
the partial embedment of nanoparticles that reduces the availabil-
ity of active sites. An estimate of the number of surface metal sites
was calculated from this dispersion and the palladium loading,
assuming complete accessibility of the surface to the gas phase.
For bimetallic nanoparticles, incorporation of second metal (Zn or
Ga) was assumed stoichiometric according to PdZn and Pd2Ga.
2.3. Catalytic performance
2.3.1. Methanol synthesis from CO2
Methanol synthesis experiments were carried out in a stainless
steel fixed-bed flow reactor. 400 mg (100–200 lm) of HTlc precur-
sor was mixed with 2 g of crushed SiO2 chips and loaded into a
10 mm I.D. stainless steel reactor tube. Gas flows were controlled
and monitored with analog mass flow controllers (Brooks 5850A).
The PdZnAl and PdMgAl HTlc precursors were reduced in situ at
523 K (2 K/min) for 2 h in 100 ml/min (STP) of 20 vol% H2 in He.
The PdMgGa HTlc precursor was reduced at 773 K (2 K/min) for 5 h
in the same H2/He mixture. Upon completion of the reduction, the
reactor was cooled to 523 K, a 3:1 H2/CO2 mixture (100 ml/min)
containing 4 vol% Ar (as internal standard) was introduced into
the reactor, and the pressure was raised to 30 bar by means of a
back-pressure regulator (Tescom). The back-pressure regulator
and all high-pressure line were heated to 423 K to avoid condensa-
tion of water and methanol, and all low-pressure lines downstream
of the back-pressure regulator were heated to 393 K. Online analysis
of products was performed with a GC (Agilent 6890) equipped with
a J&W scientific Haysep Q Column and a TCD for analysis of non-
condensable gases and a Agilent DB1 column interfaced to an FID
for analysis of methanol. Methanol and CO were the main products
observed, along with traces (less than 0.1% selectivity) of methane.
After the start of the reaction, the catalysts were allowed to stabilize
for 20 h time-on-stream at 523 K. After this period, the activation
energies of methanol synthesis and rWGS were measured in the
temperature range of 463–548 K, with 4 h allowed for each temper-
ature point. Conversions of CO2 were below 3% insuring the absence
of thermodynamic artifacts.
reactor cell (Hilgenberg). The precursor (63–100 lm, 30 mg) was
filled into the capillary between quartz wool plugs. The cell was
connected to a gas manifold. The cell was convectively heated by
means of an air blower. Spectra were collected during the reduc-
tion in 5 vol% H2/N2 (100 ml/min) in the temperature range of
303–773 K. For in situ MSR experiments, an HPLC pump was used
to introduce 0.02 ml/min of a 1:1 M ratio methanol water solution
into an SiC filled evaporator at 120 °C, where the vapor was diluted
with 100 ml/min (STP) of He. The vapor sent to the reaction cell via
heated lines, and the reactor outlet gas composition was monitored
by MS. Data reduction was performed using the Athena 0.8.056
software package.
Fourier-transformed infrared absorption spectroscopy of CO. In
situ experiments were carried out in a custom built quartz cell
equipped with KBr windows allowing sample activation and suc-
cessive measurements in the 292–823 K temperature range, at
pressures from 10ꢀ4 to 760 Torr. The catalysts were pressed into
self-supporting pellets and activated in the same cell used for the
measurement. The thermal treatments were performed either in
dynamic vacuum or under static conditions (no flux), according
to procedures discussed below. In situ FTIR spectra on the investi-
gated catalysts were recorded at a resolution of 2 cmꢀ1 on a Nicolet
6700 instrument, in transmission mode. PdZnAl and PdMgAl HTlc
precursors were outgassed in dynamic vacuum up to 523 K for sev-
eral hours (until the pressure reached values below 10ꢀ4 Torr), fol-
lowed by reduction in H2 gas (2 cycles of 30 min, PH2 = 100 Torr).
H2 was removed from the cell at the same temperature, and the
sample was allowed to reach room temperature in dynamic vac-
uum. PdGaMg HTlc precursor was outgassed in dynamic vacuum
up to 823 K for several hours (until the pressure reached values be-
low 10–4 mbar), followed by reduction in H2 gas (4 cycles of 1 h,
PH2 = 100 Torr). H2 was removed from the cell at the same temper-
ature, and the sample was allowed to reach room temperature in
dynamic vacuum.
2.3.2. Methanol steam reforming
Methanol steam reforming (MSR) reaction was conducted using
a fixed-bed flow reactor at 523 K and atmospheric pressure. Activ-
ity tests were performed using 0.125 g of HTlc precursor diluted
with SiC at volume ratio of 3:1 to avoid adverse thermal effects.
Catalyst bed was placed in a 6 mm I.D. quartz tubular reactor with
a coaxially centered thermocouple. Prior to reaction, the PdZnAl
and PdMgAl HTlc precursors were reduced in situ at 523 K (2 K/
min) for 4 h in 60 cm3 (STP) of 5 vol% H2 in N2. The PdMgGa HTlc
precursor was reduced at 823 K (2 K/min) for 4 h in the same
H2/N2 mixture. The pretreating gases were flushed from the reactor
with N2 before admission of methanol–water–nitrogen reaction
mixtures. Methanol and water mixture was supplied into a pre-
heater by means of a liquid pump (Becton–Dickinson) before their
mixture with nitrogen carrier supplied by Brooks model 5850E
mass flow controller. For MSR reaction, the reactants were intro-
duced into the reactor in the molar ratio H2O/CH3OH = 1.0. In the
MSR tests, the total flow rate was kept at 26 ml/min (STP), and
the methanol concentration in the feed gas mixture was fixed at
Scanning electron microscopy. SEM images were acquired with a
Hitachi S4800 FEG microscope equipped with an EDS system
(EDAX) for elemental analysis. The samples were loosely dispersed
on a conductive carbon tape to preserve the as-prepared morphol-
ogy as much as possible. SEM images were acquired at low accel-
erating voltage, that is, 1.5 kV, while EDX spectra were acquired
at an accelerating voltage of 15 kV.
Electron microscopy. HRTEM images were acquired using a FEI
TITAN microscope, equipped with a field emission gun (FEG) and

30
A. Ota et al. / Journal of Catalysis 293 (2012) 27–38
28.4 vol% to avoid thermal effects derived from the high endo-
thermicity of the reaction.
composition of Pd containing and Pd-free HTlc precursors is listed
in Table 1. There are small deviations between the nominal and
measured composition. In particular the M3+ content of PdMgAl
and PdMgGa HTlc is increased to 35 mol% and the Pd content of
PdMgAl HTlc is slightly higher than 1 mol% (1.3 mol%).
The reaction products were analyzed on line by GC with TCD
(HP 6890 GC) equipped with CP-Porabond Q (CO2, water, metha-
nol, formaldehyde, methyl formate, and dimethyl ether) and
molecular 5A (H2, O2, N2, and CO) connected in series, using He as
carrier gas. Each value of total conversion is the average of two dif-
The specific surface areas of the HTlc precursor, as obtained by
nitrogen physisorption, varied between 34 and 127 m2/g (Table 1).
The substitution of M2+ by Pd2+ resulted in an increase in the sur-
face area in all samples probably due to hindered growth of the Pd-
distorted HTlc lattice. In general, the surface area increases from
PdMgGa < PdZnAl < PdMgAl which is in agreement with the broader
XRD line profiles suggesting smaller crystallite size. Differences in
surface area and XRD peak widths are compatible with the obser-
vation of SEM measurements. As shown in Fig. 2 the HTlc precur-
sors show similar platelet-like morphology, but the particle sizes
differed considerably. Small platelets were obtained for PdMgAl
(Fig. 2C) and PdZnAl (Fig. 2A). The thickness of the platelets is in
the low nanometer range, while the lateral dimension is approxi-
mately around 250 nm or below. Although for PdMgAl and PdMgGa
samples the same synthesis conditions were applied the lateral
ferent analyses taken after 3 h time-on-stream at
temperature.
a given
The hydrogen selectivity of the reaction was determined by
comparing the molar hydrogen production rate to the maximum
stoichiometrically allowable hydrogen production rate at a given
methanol conversion (4).
rH2
3rMeOH
S ¼
ð4Þ
If the reaction is dominated by MSR, H2, and CO2 selectivities
should lie at around 100%, whereas if methanol decomposition is
the primary reaction, the H2 selectivity lies at 67% and the CO2
selectivity near zero.
size of the PdMgGa platelets is increased to 0.4–1 lm. The PdMgGa
and PdMgAl platelets exhibit sandrose morphology, while the Pd
ZnAl platelets are randomly oriented. However, no large Pd
agglomerates were detected outside the platelets by SEM and also
EDX mapping suggests a homogenous metal distribution for all
samples (not shown).
3. Results and discussion
3.1. Properties of the HTlc precursor
PdZnAl, PdMgGa and PdMgAl HTlc precursors have been prepared
with 1 mol% of the cations being Pd2+ and a M2+:M3+ ratio of 70:30
by co-precipitation. The XRD patterns of all as-synthesized precur-
sor samples are typical of Hydrotalcite-like structure, as shown in
Fig. 1. No Pd containing or other crystalline phases were observed.
The crystallinity of the precursors decreased in the order of
PdMgGa > PdZnAl > PdMgAl as indicated by the decreasing intensity
as well as the increased peak broadening. The measured chemical
3.2. Reducibility of the HTlc precursors and IMC formation
3.2.1. TPR and MS measurements
TPR experiments were performed in order to study the IMC for-
mation and monitor the reduction process. All TPR profiles can be
separated into three regions: a room temperature reduction (RTR),
a low temperature reduction (LTR, <573 K) and high temperature
reduction region (HTR, >573 K). RTR contains any initial hydrogen
consumption detected upon switching to reducing atmosphere at
room temperature before the heating ramp was started. Fig. 3 in-
cludes the LTR and HTR processes, only. The reduction from Pd2+
to metallic Pd0 normally occurs in the RTR or LTR regime, at tem-
peratures lower than ca. 423 K. Quantification of the degree of
reduction from the hydrogen consumption turned out to be com-
plex for these samples. Metallic Pd particles form hydrides and
transfer spillover hydrogen to the support material; both phenom-
ena lead to higher hydrogen consumption than expected for stoi-
chiometric Pd2+ reduction [15]. The PdHx phase is not stable
during heating and decomposes at higher temperatures producing
a negative peak in the TPR profile. In the HTR regime hydrogen
consumption due to partial reduction of reducible metal oxides
overlaps with further H2 consumption due to conversion of CO2,
which is released as a decomposition product of the interlayer car-
bonate ions. As a matter of fact, CO and CH4 have been detected by
MS during LTR and HTR (Fig. 3, bottom). Thus, evolved CO2 is not
completely emitted from the catalyst bed, but varying fractions un-
dergo rWGS and methanation at the surface of the metallic nano-
particles. This resulted in considerably higher hydrogen
consumption than expected for stoichiometric reduction of the
metal cations. Apparently, hydride formation, spillover hydrogen,
and CO2 conversion are dominating the consumption, and it is only
considered in a semi-quantitative manner here, that is, by compar-
ison of the TCD peak integrals of the different samples in the differ-
ent temperature regimes.
Fig. 1. XRD patterns of (A) PdMgGa HTlc and simulated MgGa HTlc precursor; (B)
ex-PdMgGa HTlc after reduction at 773 K, ICDD 1-1235 MgO (green), ICDD 10-113
MgGa2O4 (orange); (C) PdZnAl HTlc precursor and ICDD 38-486 Zn6Al2(OH)16-
CO3ꢁ4H2O; (D) ex- PdZnAl HTlc after reduction at 523 K, ICDD 36-1451 ZnO (violet);
(E) PdMgAl HTlc precursor and ICDD 14-191 Mg6Al2(OH)16CO3ꢁ4H2O; (F) ex-PdMgAl
HTlc after reduction at 523 K. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)
As expected, no hydrogen consumption was observed in the
RTR and LTR regions for the Pd-free samples. For the Pd containing
precursors, the profiles in this region are different. RT and LT
hydrogen consumption were observed for PdMgGa and PdMgAl
samples, while for PdZnAl sample H2 uptake was not detected at

A. Ota et al. / Journal of Catalysis 293 (2012) 27–38
31
Table 1
Chemical composition and textural properties of the HTlc precursor materials.
Sample
Measured compositiona Pd:M2+:M3+ BET SA precursors (m2/g) BET SA reduced (m2/g) Pd content after reduction (wt.%) Particle size by TEM (nm)
ZnAl
PdZnAl
MgGa
PdMgGa 1.0:64.5:34.5
MgAl
0:70.3:29.7
1.0:69.5:29.5
0:63.2:36.8
53
85
34
48
111
127
n.d.b
–
1.46
–
1.81
–
3.13
–
86c
1.8c
–
130
116d
n.d.b
123c
6.1d
–
0:65.6:34.4
1.3:69.1:29.6
PdMgAl
2.2d
a
b
c
Determined by ICP-OES.
n.d. = not determined.
Reduced at 523 K.
d
Reduced at 773 K.
Fig. 2. SEM images of (A) PdZnAl, (B) PdMgGa, and (C) PdMgAl HTl precursor. (Bottom parts show magnification of top parts.)
RT. This suggests that in the former samples two different Pd2+ spe-
cies were present, one easily and one harder to reduce. The main
LTR peaks are found at 359, 382, and ca. 343 K for PdZnAl, PdMgGa
and PdMgAl, respectively, which are relatively high values for the
reduction of Pd2+. From this observation, the RTR peak can be as-
signed reduction to PdO-like species not incorporated into the
HTl lattice, while the LTR peaks are due to reduction of HTl-lattice
Pd2+ cations. This assumption is consistent with XANES measure-
ments discussed below. Furthermore, Pd hydride is observed only
for the PdMgAl (small negative peak around 423 K). Additional LT
hydrogen uptake was observed for PdZnAl at 520 K. It is known that
ZnO can be reduced in the presence of Pd at such mild tempera-
tures to form the IMC PdZn [13b,15]. Since the above-mentioned
LTR peak is absent for the Pd-free precursor, it is assigned to reduc-
tion of ZnO with consequent IMC formation (although small
amounts of CH4 were detected at this temperature). No further
peaks are detected in the LTR region for PdMgGa and PdMgAl sam-
ples. In PdMgAl, there are no other reducible species. In PdMgGa,
this observation indicates that the IMC formation takes place at
higher temperature. Therefore, the process cannot be accurately
studied by TPR, because of the overlap with CO2 conversion in
the higher temperature regime, see Fig. 3 bottom.
ZnO-component of the support. For MgGa and PdMgGa, the pres-
ence of Pd leads to an increase in the total HTR hydrogen consump-
tion and to a decrease in temperature from 736 to 695 K. In both
samples, CO was detected by MS. Thus, at sufficiently high temper-
ature CO2 from decarboxylation of the interlayer carbonate anions
undergoes rWGS on the reducible oxides [16]. In case of PdMgGa
the hydrogen uptake due to rWGS and support reduction overlaps
with IMC formation (Fig. 3). Only the non-reducible MgAl sample
does not show any significant hydrogen consumption over the
whole temperature range. Addition of Pd to this sample induced
an uptake of hydrogen at 679 K; correspondingly, CH4 was de-
tected by MS (Fig. 3). Therefore, it is concluded that CO2 deriving
from decomposition of the interlayer carbonates is converted to
methane over metallic Pd. TPR and MS data provide first evidence
that the catalytic properties of Pd have been markedly modified by
formation of IMCs. While PdMgAl (where no IMC is formed) was
found to be active in methanation, PdMgGa converts CO2 in the
rWGS to CO at comparable conditions. The HTR hydrogen conver-
sion of PdZnAl can be completely ascribed to the reducibility of the
support material. From the TPR results, the reduction temperatures
for IMC formation were estimated to 523 K for PdZnAl and 773 K
for PdMgGa. The PdMgAl reference sample was reduced at 523 K.
In the HTR regime, hydrogen consumption was detected for all
precursors, also for the Pd-free MgGa and ZnAl, except for MgAl. At
these temperatures, partial reduction of the reducible oxides is ex-
pected also in the absence of Pd. Starting from PdZnAl and ZnAl,
HTR consumption peaks having a similar integral area were ob-
served at approximately same temperature (808 and 823 K). No
MS-peaks of CO, or CH4 were noticed at that temperature (Fig. 3).
As a consequence, the peaks are assigned to reduction of the
3.2.2. XANES measurements
XANES measurements were carried out to gain more insight
into the Pd speciation in the precursor and in the RTR and LTR
events. Since TPR measurements suggested the presence of two
Pd-species in PdMgAl and PdMgGa, different reference materials
were used to evaluate the XANES data. Commercial PdO was
chosen as reference material for segregated Pd2+ species in the

32
A. Ota et al. / Journal of Catalysis 293 (2012) 27–38
structure, as shown in Fig. 4B and C. This analysis revealed that
68% of Pd2+ is present in octahedral coordination for PdMgGa, while
in case of PdMgAl only 61% of Pd2+ is incorporated. Both samples
showed a certain degree of RT reduction in the TPR (see Sec-
tion 3.2.1). Accordingly, also the XANES spectra changed upon H2
reduction at 323 K (Fig. 4). In particular for PdMgAl the near-edge
features approach those of metallic Pd0. The PdMgGa sample is
apparently more stable and its XANES changes to a lesser extent.
Still a significant shift of the second XANES feature to higher en-
ergy was observed also for this sample. This higher stability is in
agreement with the higher LT reduction peak temperature com-
pared with PdMgAl as observed in the TPR experiments (Fig. 3).
3.3. Properties of the ex-HTlc samples after reduction
The XRD patterns of the reduced samples presented in Fig. 1
contain only broad modulations of the background suggesting poor
crystallinity and small crystallize sizes. After reduction at 523 K,
PdMgAl is amorphous (Fig. 1F) and only heating to 773 K leads to
significant growth of XRD peaks due to poorly crystalline MgO
and MgAl2O4 (not shown). Formation of periclase and spinel
phases was also observed for PdMgGa after reduction at 773 K
(Fig. 1B). The PdZnAl sample showed very weak and broad peaks
due to ZnO after reduction at 523 K (Fig. 1D). The XRD data are
an agreement with HRTEM analysis of the phases of the support.
Because the metallic component was not able to be identified
by means of XRD, probably due to the low total Pd content and
the small crystallite size, XAFS and HRTEM were applied for phase
identification. Fig. 5 shows the XANES spectra of the reduced pre-
cursors at the reduction temperatures deduced from the TPR
experiment. Palladium foil, bulk Pd2Ga and PdZn [21] are shown
as reference materials. In general, lower amplitudes of the oscilla-
tions were observed in the XANES (Fig. 5) and EXAFS (Fig. 6) re-
gions for the catalyst samples as a result of lower coordination
number and/or higher disorder as expected for nanoparticles, in
particular for PdZnAl, whose EXAFS spectra were too noisy to give
reliable data. Nevertheless, the EXAFS spectra of the other catalysts
and the XANES of all samples show resembled those of the associ-
ated bulk references regarding the features of the fine structure on
the energy scale. This shows that the local environment around the
Pd atoms and the bond distances to the neighboring atoms are sim-
ilar in the nanocatalysts and the bulk references. In particular, good
agreement was observed for the PdMgAl catalyst and the pure Pd
reference, whereas there is a clear difference between Pd and the
IMC catalysts confirming the modification of Pd by alloying and
IMC formation.
Fig. 3. TPR profiles of Pd substituted and unsubstituted Hydrotalcite-like precur-
sors (top) and MS profiles (obtained by TGA-MS) in 5 vol% H2/Ar (bottom).
precursor, which is not incorporated into the HTl lattice. Pd2+ in
PdO prefers square planar coordination as well as Pd2+ in aqueous
solution and most other solids. On the contrary, incorporation into
the Hydrotalcite lattice requires octahedral coordination of Pd. The
low-spin d8 electronic configuration of Pd2+ makes regular octahe-
dral coordination unstable with respect to large tetragonal elonga-
tions. Octahedral coordination of Pd2+ with oxygen was obtained in
a high-pressure phase of PdO at 12 GPa [17]. Moreover, octahedral
environment was also observed for Pd cations having higher oxida-
tion states. For instance, Zn2PdO4 is a compound where Pd4+ has
regular octahedral symmetry in the cubic spinel structure [18],
while in LaPdO3 Pd3+ has a distorted perovskite structure that sta-
bilizes the trivalent ion in octahedral coordination [19]. The XANES
spectra of the HTl precursors are shown in Fig. 4; both in the unre-
duced state and after treatment at 323 K in hydrogen, that is, at a
temperature where RTR has already occurred, but low enough not
to trigger LTR. XANES spectra of PdO as reference for segregated
Pd2+ species before RTR and of metallic Pd for the state after RTR
were included.
The intensity ratio of the white line and the first feature of the
XANES spectrum of PdZnAl HTlc are significantly different from
that of PdO. It is similar to that reported for LaFe0.95Pd0.05O3 where
Pd exhibits a distorted octahedral coordination and occupies Fe
sites [20]. Changing to reducing atmosphere did not induce a
change of the XANES of PdZnAl, which is in agreement with the
TPR experiment that did not show any RT reduction. Therefore,
we assume that in this sample Pd2+ is fully incorporated into the
HTl structure and not available for reduction to metallic Pd below
323 K. Therefore, the PdZnAl spectrum recorded at RT in inert
atmosphere was taken as a second reference for Pd2+ that is incor-
porated into the HT structure and exhibits octahedral coordination.
Linear combination of PdO and PdZnAl spectra were used to obtain
the fraction of Pd present as PdO and as incorporated within HT
TEM revealed that the platelet-like morphology of the HTlc pre-
cursor was retained after reduction and all catalysts showed spher-
ical particles distributed on the platelets as presented in Figs. 7A
and 8A. HRTEM images were used to determine the net plane dis-
tances and plane angles by FFT. Reduction of PdMgGa yielded well
dispersed, supported nanoparticles as presented in Fig. 7A. The cor-
responding power spectrum of the HRTEM image, shown in Fig. 7B,
allowed identification of the intermetallic Pd2Ga phase (ICDD 65-
1511). In agreement with XRD, partially crystalline MgO and
MgGa2O4 were present as oxidic support [5]. However, for PdZnAl
two different Pd-Zn IMCs were identified after reduction at
ꢀ
523 K. The crystal planes ð1 1 1Þ and (1 1 1) of PdZn (ICDD 6-620)
were identified with the characteristic acute angle of 65° along
ꢀ
the ½1 0 1ꢂ zone axis (Fig. 8B and C). Differentiation of the closely
related crystal structures of Pd and PdZn is often difficult by
HRTEM as Pd particles produce similar patterns along the [1 1 0]
zone axis, but the angle between the corresponding (octahedral)
planes in the fcc Pd structure should be 70.5°. In several particles
the lattice spacings d111 were systematically smaller than
0.22 nm, and d200 were larger than 0.20 nm, while in pure Pd (ICDD

A. Ota et al. / Journal of Catalysis 293 (2012) 27–38
33
Fig. 4. (A) XANES spectra PdZnAl at RT/He (a) and 323 K/H2 (b), PdMgGa at RT/He (c) and 323 K/H2 (d), PdMgAl at RT/He (e) and 323 K/H2 (f), PdO (g) and Pd foil (h) reference.
(B and C) Linear combination fit of the XANES spectra of PdMgGa and PdMgAl precursor, by using PdO and PdZnAl as references for Pd2+ in square planar and octahedral
coordination, respectively.
46-1043) d111 = 0.225 nm and d200 = 0.195 nm. Based on these dif-
ferences the particles were identified as PdZn rather than Pd.
Additionally, nanoparticles with a core–shell structure were
present. In one particle the phase in the core was identified as
Pd2Zn (circled area, ICDD 6-629) surrounded by an ordered shell
of ZnO (Fig. 8E and F). This arrangement suggests that Pd2Zn might
be an intermediate of the reduction of PdZnAl. ZnO seems to have
segregated from the oxide support to encapsulate the metal core
by strong metal support interaction and supply further Zn atoms
for IMC formation through the reactive Pd-ZnO and Pd2Zn–ZnO
interfaces, finally resulting in formation of PdZn. Fig. 9 shows the
PdMgAl sample after reduction at 773 K. In STEM mode homoge-
neous and small-sized Pd nanoparticles on MgAlOx support were
observed (Fig. 9A) and HRTEM investigations (Fig. 9B) confirming
the presence of metallic Pd (ICDD 46-1043). The average particle
sizes were obtained from projected areas of the metal particles
and were 1.8, 2.2 and 6.1 nm for PdZnAl (based on 50 particles),
Fig. 5. In situ XANES spectra of (a) PdZnAl reduced at 523 K and (b) bulk PdZn, (c)
PdMgAl (400 particles) and PdMgGa (1500 particles), respectively.
PdMgGa reduced at 773 K, and (d) bulk Pd2Ga, (e) PdMgAl reduced at 523 K and (f)
Pd foil.
The considerably larger size of the Pd2Ga particles can be explained
by higher reduction temperature that was necessary to form the
IMC. The corresponding size distributions are shown in Fig. S2.
samples are shown in Fig. 10 as function of CO coverage. Several IR
absorption bands were observed in the range of 2100–2000 cmꢀ1
and 1970–1800 cmꢀ1. These bands are characteristic of linearly
and multiply bonded CO on Pd0 atoms, respectively [22]. As
FT-IR spectroscopy of adsorbed CO was adopted to investigate
the surface properties of the metallic Pd, PdZn and Pd2Ga compo-
nents in PdMgAl, PdZnAl and PdMgGa after reduction, respectively.
The FT-IR spectra of CO adsorbed at room temperature on the three

34
A. Ota et al. / Journal of Catalysis 293 (2012) 27–38
2080 cmꢀ1 gains intensity and slightly shifts to higher wavenum-
bers due to increased dipole–dipole coupling of the neighboring
CO groups.
In summary, small metal particles have been obtained from
HTlc precursors with at least partial Pd2+ incorporation on octahe-
dral sites in the cationic layers. Upon reduction in hydrogen, ele-
mental Pd segregated from the precursors at low temperature
and as the reduction temperature increases, strong modification
of the Pd phase are observed, which are attributed to the formation
of IMC particles. Although the absence of remainders of elemental
Pd in the IMC samples is hard to finally proof with each character-
ization method applied, the combination of imaging, bulk- and sur-
face-sensitive techniques all provide evidence that Pd has reacted
with the reducible oxide species in both samples. These results
strongly suggest that the state of the freshly reduced samples is
intermetallic. For PdZnAl the IMC PdZn is formed at 523 K probably
via a Pd2Zn intermediate stage. The average particle size is about
1.8 nm. Pd2Ga formed from PdMgGa at higher temperature
(773 K) and the average particle size is larger (6.1 nm). Compared
with the monometallic Pd particles obtained by PdMgAl, the
adsorptive properties of the IMC particles are strongly modified
suggesting differences in catalytic properties.
3.4. Catalytic properties of the IMCs
3.4.1. Methanol synthesis from CO2
Activity versus time-on-stream data for methanol synthesis and
rWGS at 523 K is shown in Fig. 11 for all Pd-based catalysts. After a
5 h stabilization period, all catalysts show only very slow loss of
activity with time-on-stream. PdMgGa and PdZnAl showed similar
steady-state activities for methanol synthesis of 18.7 and
17.6
nearly 30 times the methanol synthesis activity of PdMgAl
(0.65 mol/(gcat min)). In contrast to this difference, the rWGS of
all catalysts were within similar range, with PdMgGa
(27.8 mol/(gcat min)) being twice as active as PdMgAl and PdZnAl
with 14.3 and 13.5 mol/(gcat min), respectively. These results
lmol/(gcat min), respectively. These activities amounted to
Fig. 6. Pd K edge EXAFS spectra (k2 weighted) of PdGaMg (A) and PdMgAl (B) after
reduction at 773 K (a), during MSR at 523 K (b), after 90 min MSR(c). Reference
spectra are shown of bulk Pd2Ga (d in A), and Pd foil (e in A and d in B).
l
a
l
l
are consistent with those of Iwasa et al. [24] and Fujitani et al.
[25], who showed that Pd supported on ZnO and Ga2O3 possessed
significant methanol synthesis activity, whereas unsupported Pd
and Pd on irreducible SiO2 were almost inactive for methanol syn-
thesis. Additionally, the results of Iwasa showed far less significant
differences in rWGS rates on these catalysts. A similar observation
about rWGS activities can be made for the catalysts studied in this
work. From this observation it can be concluded that free Pd may
be responsible for the rWGS activity, whereas the intermetallic
sites lead to methanol formation. However, Iwasa et al. [24] had
shown, that the rWGS activity of Pd/ZnO decreases by only 40%
upon the complete conversion of Pd into a PdZn alloy. It is there-
fore possible that the intermetallic surfaces behave similarly to
Cu, in that they possess sites for both methanol synthesis and
rWGS.
reported by Kovnir et al. [1] and Conant et al. [13c] the absence or
strong reduction of multiple bonded carbonyl species should be
observed for Ga–Pd and Pd–Zn intermetallic compounds. Indeed,
mainly on-top coordinated CO species, characterized by IR absorp-
tion bands at 2080 cmꢀ1 and 2050 cmꢀ1 were observed for PdMgGa
and PdZnAl, as shown in Fig. 10A and B. In contrast, CO absorption
on PdMgAl sample revealed a much larger population of multiply
bonded sites compared with linearly coordinated CO (Fig. 10C).
At low CO coverage the band of linearly coordinated CO was ob-
served at 2020 cmꢀ1, which is relatively low, but still in the char-
acteristic range. According to Prinetto et al. [23] the frequency
red shift originates from CO absorption on Pd sites, whose elec-
tronic properties are modified by interaction with basic support
species. With increasing CO coverage, an additional band at
Fig. 7. (A) overview TEM images PdMgGa after reduction at 773 K in 5 vol% H2/Ar and (B) HRTEM image and corresponding FFT pattern of Pd2Ga.

A. Ota et al. / Journal of Catalysis 293 (2012) 27–38
35
Fig. 8. (A and D) Overview TEM images, and (B and E) HRTEM and corresponding FFT pattern (C and F) of PdZnAl after reduction at 523 K in 5 vol% H2/Ar.
Fig. 9. (A) STEM and (B) HRTEM images of PdMgAl after reduction at 773 K in 5 vol% H2/Ar.
Fig. 10. FT-IR spectra of CO adsorbed at room temperature on (A) PdZnAl (reduced at 523 K), (B) PdMgGa (reduced at 823 K) and (C) PdMgAl (reduced at 523 K) as a function of
CO coverage (hmax: blue, hmin: red). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
In order to compare the intrinsic activities and apparent activa-
tion energies for the Pd catalysts with those of a more active
CuZnAl catalyst, which was also prepared from a single phase HTlc
precursor, the space velocity had to be varied to reach similar con-
version levels in the low-% range for all samples. PdZnAl shows a
methanol selectivity of 60% and is more selective that PdMgGa
(47%), whereas the CuZnAl sample exhibits a higher selectivity of
82%. The methanol TOF of CuZnAl is 4–12 times higher than that
of the IMCs (Table 2). The TOFs of the IMC catalysts in this work
were estimated on basis of the average particle size as detected
by TEM without consideration of possible variations in surface
composition, segregation or particle embedment and thus are

36
A. Ota et al. / Journal of Catalysis 293 (2012) 27–38
likely afflicted with systematic errors. It should be noted, however,
that the intrinsic activities of the IMCs are still two orders of mag-
nitude higher than of PdMgAl.
The comparison of apparent activation energies between Cu-
and Pd-based catalysts results in an intuitive understanding to
the effect of IMC formation on reactivity. The Arrhenius plots for
methanol synthesis and rWGS are shown in Fig. 11B for Pd-based
catalysts tested at a space velocity of 10 mmol/(gcat min) and
CuZnAl at 93 mmol/(gcat min). It is interesting to note that the acti-
vation energy for methanol synthesis decreases from 72 kJ/mol in
PdMgAl to 68 kJ/mol in PdZnAl and 59 kJ/mol in PdMgGa upon the
substitution of Al with Ga and Mg with Zn, respectively, and ap-
proaches 54 kJ/mol as measured for CuZnAl. Similarly for rWGS,
substitution causes an increase in the apparent activation energy
from 69 kJ/mol in PdMgAl to 94 kJ/mol and 82 kJ/mol for PdZnAl,
PdMgGa, respectively, in the direction of CuZnAl (122 kJ/mol). The
results suggest that the reactivity of Pd in methanol synthesis
and rWGS approaches that of Cu upon substitution of irreducible
species with Zn and Ga. In particular, these extraordinary differ-
ences in reactivity correspond to the formation of PdZn and Pd2Ga
IMCs upon reduction of PdZnAl and PdMgGa, respectively and the
presence of only pure Pd upon reduction of PdMgAl, as evidenced
by the aforementioned characterization results. The similarity of
the catalytic properties of PdZn and Cu has been reported in liter-
ature and was attributed to the similar electronic structure. Both
exhibit similar valence electron densities of states (DOS) as con-
firmed by DFT, UPS and XPS [26].
The fact that the catalytic properties do not evolve with the
amount of ‘‘inert’’ species diluting the Pd metal atoms when going
from pure Pd to the 2:1 IMC Pd2Ga and further to the 1:1 IMC PdZn
suggests that rather this modification of the electronic structure is
responsible for the changes in activity and selectivity compared
with a purely geometric site-isolation concept. The electronic
structures depend on the crystal structures of the IMCs that differ
from that of elemental Pd and are not subjected to continuous
trends as expected for statistical fcc alloys, but need to be consid-
ered individually. Indeed, in case of Pd-Zn the Zn-poorer fcc alloy
Fig. 11. (A) Rate of CO2 hydrogenation over PdZnAl, PdMgGa and PdMgAl at 523 K
(30 bar, 400 mg catalyst, 100 ml/min CO2/H2). (B) Arrhenius plot for methanol
synthesis from CO2 (closed symbols: MeOH, open symbol: CO).
(a-phase) has recently been shown to exhibit catalytic properties
greatly different from those of PdZn and to resemble rather those
of elemental Pd [27].
CO2 selectivities observed by Iwasa et al. were near 95% and there-
by similar to those observed on CuZn based catalysts. Good activity
and high CO2 selectivity up to 99.4% and 98% were also reported for
unsupported intermetallic bulk compounds [28] and aerosol-de-
rived powders [2]. It should be noted that the particle size of the
PdZn compounds observed by Iwasa et al. [11b], Friedrich et al.
[28] and Halevi et al. [2] were far larger than the particles obtained
in this work. It may therefore be possible that under reaction con-
ditions the smaller particles decompose partially to yield metallic
Pd, or that the methanol decomposition reaction is structure sensi-
tive, and occurs faster on smaller particles. In accordance with the
latter statement, Karim et al. [29], found a positive correlation be-
tween the number of small particles (<1.5 nm) and decreasing
selectivity toward CO2 on Pd/ZnO catalysts using TEM analysis.
Iwasa et al. [11b] and more recently, Haghofer et al. [12a] have
shown that Pd supported on Ga2O3 is also very selective to MSR
with CO2 selectivities above 80%, and have attributed this high
selectivity to the formation of Ga-Pd IMCs. However, for the
PdMgGa sample investigated herein the CO2 selectivity (16%) is
only somewhat higher than that of the PdMgAl (6%) catalyst, in
which no IMCs are formed. Haghofer et al. [14] have shown by
in situ IR spectroscopy on Pd/Ga2O3 that the intermetallic Pd2Ga
surface is not entirely stable during MSR at 480 K, and that metallic
Pd patches forms on the surface, as evidenced by the presence of
bridge-bonded CO. At the higher temperatures (523 K) used in this
study, such surface decomposition may occur to a greater extent
and leave a higher fraction of exposed Pd, thus favoring methanol
decomposition and leading to lower CO2 selectivity.
3.4.2. Steam reforming of methanol
The catalytic activities of the Pd-based catalysts derived from
HTlc precursors in MSR tests are summarized in Table 3. For all
tested catalysts the main products were H2, H2O, CO and CO2.
Other reaction products such as formaldehyde or dimethylether
were not observed. The PdZnAl and PdMgGa catalysts showed sim-
ilar overall activity (9.5% methanol conversion), while the PdMgAl
catalyst was somewhat less active (7% methanol conversion). The
PdZnAl catalyst exhibited the highest hydrogen yield (660 lmol/
(gcat min)) of all catalysts tested, and this hydrogen yield was
23% and 75% higher than those of PdMgGa and PdMgAl, respectively
(Fig. 12). In terms of intrinsic activity for hydrogen production, the
PdZnAl and PdMgGa were 2 and 6 times, respectively, more active
than PdMgAl. The hydrogen selectivities of PdMgGa and PdMgAl
were significantly lower (70%) than that of PdZnAl (87%). Although
the monometallic Pd catalysts PdMgAl showed the lowest perfor-
mance also in this reaction, the differences to the IMC catalysts
in product yield and TOF are interestingly much smaller compared
with the methanol synthesis tests. Marked differences are however
seen in the selectivity toward CO2.
While the reactivity of methanol on PdMgAl and PdMgGa is al-
most entirely dominated by methanol decomposition, PdZnAl
shows considerable MSR activity with S(CO)2 = 61%. Iwasa et al.
[11a] have attributed this CO2 selectivity to the formation PdZn
intermetallic phases during reduction of Pd/ZnO. However, the

A. Ota et al. / Journal of Catalysis 293 (2012) 27–38
37
Table 2
Activities for methanol synthesis from 1:3 CO2:H2 ratio at 30 bar on Pd-based and Cu-based catalysts at 523 K.
Catalyst
Metal site (
lmol/gcat
)
XCO2 (%)
Rate (
l
mol/min tgcat
CO
6.2
)
Space velocity
(mmol/gcat min)
SMeOH (%)
TOF (minꢀ1
MeOH
)
MeOH
CO
PdZnAl
PdMgGa
PdMgAl
CuZnAl
73.4a
20.4a
0.6
1.0
0.3
1.3
9.1
10.5
0.3
10
10
10
93
60
47
4
0.12
0.51
0.003
1.30
0.08
0.57
0.07
0.28
11.7
6.8
47.6
90.9a
170.9b
222.4
82
a
Estimated from the metal particle size determined by TEM.
Measured with N2O reactive frontal chromatography.
b
Table 3
Activity results for the Pd-based IMC catalysts for the steam reforming of methanol at 523 K.
Catalyst
Metal sitesa
(l
mol/gcat
)
XMeOH (%)
SCO2 (%)
Rate [
l
mol/min g]
SH2 (%)
TOF (minꢀ1
MeOH
)
Methanol converted
H2 released
H2
PdZnAl
PdMgGa
PdMgAl
73
20
91
14.0
15.3
12.6
61.1
16.4
6.1
369
404
331
964
874
681
87
72
69
5.1
20.2
3.6
13.2
43.7
7.5
a Estimated from the metal particle size determined by TEM.
Another aspect that deserves consideration is the difference in
the interface of metallic and oxide phase due to the preparation
technique applied in this study compared with most literature re-
ports, where, for instance, well crystalline b-Ga2O3 has been used
as support [11a,12a,12c]. In our samples the IMCs are supported
on poorly crystalline MgO/MgGa2O4. The presence of the oxide
component can influence the reactivity pattern in MSR [28,30]
and also its nature should matter. However, in a study of Pd2Ga
particles of a similar size below 10 nm supported on crystalline
b-Ga2O3 we have observed similar MSR properties as in this work
[31]. This observation supports the hypothesis that the effect of
particle size plays a more decisive role than the nature of the gal-
lia-containing oxide support.
It should be noted that the promotional effect of Pd2Ga forma-
tion on the selectivity is far more evident during methanol synthe-
sis (as described previously) as compared to steam reforming. To
that end, the higher hydrogen pressures and virtual absence of
water during methanol synthesis may ensure the stability of the
Pd2Ga intermetallic surface responsible for the enhanced selectiv-
ity. In contrast, the presence of large concentrations of water dur-
ing MSR may lead to a surface decomposition. Such a surface
reaction can be expected to occur faster on small particles as stud-
ied in this work. However, the in situ EXAFS data (Fig. 6) do not
show clear indications of decomposition and Pd segregation under
MSR conditions. It is thus concluded that such decomposition takes
place only at the outer surface of the Pd2Ga particles while the bulk
remains in the IMC state. The product of the surface decomposition
is not easily characterized and might be monometallic Pd or a Pd-
rich alloy in addition to segregated Ga2O3. Corresponding to its
lower formation temperature and the lower oxophilicity of Zn,
the PdZn IMC seems to possess a greater stability in less reducing
atmospheres than Pd2Ga and the beneficial intrinsic catalytic prop-
erties observed in methanol synthesis prevail to a greater extent in
the PdZnAl catalyst under wet MSR conditions. Still the same con-
siderations about water-induced surface decomposition of the
PdZn phase to unselective Pd or the a-Pd-Zn alloy [27] that might
be accelerated by small and very reactive PdZn particles also hold
in this system and are a likely explanation for the differences in
performance observed between this and literature studies.
4. Conclusion
A flexible synthetic approach for the synthesis of nanoparticles
of mono- and intermetallic compounds has been presented. Pd
(2.2 nm), PdZn (1.8 nm), and Pd2Ga (6.1 nm) nanoparticles were
obtained from HTlc precursors with a homogeneous distribution
of Pd and intimate interaction with the other reducible metal spe-
cies. At a degree of substitution of 1 mol%, full incorporation of Pd2+
in the HTlc lattice has been confirmed for PdZnAl, while in PdMgAl
and PdMgGa the majority of Pd2+ was incorporated. Upon reduction
in hydrogen, metallic Pd segregated from the precursors at low
temperature and for PdZnAl the IMC PdZn is formed at 523 K prob-
ably via a Pd2Zn intermediate stage. Formation of Pd2Ga IMC in
PdMgGa requires 773 K. In comparison with the analogous PdMgAl
sample, which does not contain a second reducible species and
yields monometallic Pd nanoparticles, the CO absorption proper-
ties as well as the catalytic properties have markedly changed
through formation of the IMCs. Improved activities and selectivi-
ties in MSR, and methanol synthesis from CO2 have been observed.
This effect is larger for CO2 hydrogenation than for MSR. In the for-
mer reaction the Pd2Ga-based catalyst showed the best perfor-
mance, while the latter reaction is was the PdZn-based sample.
These differences are probably related to the different sensitivity
of the IMC surfaces against decomposition into Pd metal or a Pd-
enriched bimetallic phase and oxide in the presence of water.
The more reducing conditions of methanol synthesis seem to
Fig. 12. Rate of MSR of PdZnAl, PdMgGa and PdMgAl catalysts in MSR conducted at
523 K.

38
A. Ota et al. / Journal of Catalysis 293 (2012) 27–38
[8] a R. Schlögl, Chemsuschem 3 (2010) 209;
stabilize the IMC better than the more oxidizing conditions of MSR.
Under steady-state conditions, our Pd-based IMC catalysts are infe-
rior to a Cu/ZnO reference system. Nevertheless, the potential of
these materials for MSR operation at higher temperature and un-
der changing reaction conditions like on–off operations remains
to be investigated. The HTlc precursor approach has been shown
as a valuable tool to investigate IMC catalysts in a nanostructured
form and, in our future work, will be exploited for a deeper inves-
tigation of Pd-based IMCs and to explore different elemental com-
binations in comparative studies.
b P.J. de Wild, M.J.F.M. Verhaak, Catal. Today 60 (2000) 3;
c F. Joensen, J.R. Rostrup-Nielsen, J. Power Sources 105 (2002) 195.
[9] R. Schögl, Energy Storage Materials, Walter de Gruyter, Berlin, 2012.
[10] L. Guczi, A. Erdôhelyi, Catalysis for Alternative Energy Generation, first ed.,
Springer, 2012.
[11] a N. Iwasa, S. Masuda, N. Ogawa, N. Takezawa, Appl. Catal. A 125 (1995) 145;
b N. Iwasa, T. Mayanagi, N. Ogawa, K. Sakata, N. Takezawa, Catal. Lett. 54
(1998) 119.
[12] a A. Haghofer, K. Föttinger, F. Girgsdies, D. Teschner, A. Knop-Gericke, R.
Schlögl, G. Rupprechter, J. Catal. 286 (2012) 13;
b S. Penner, H. Lorenz, W. Jochum, M. Stöger-Pollach, D. Wang, C. Rameshan, B.
Klötzer, Appl. Catal. A 358 (2009) 193;
c H. Lorenz, S. Penner, W. Jochum, C. Rameshan, B. Klötzer, Appl. Catal. A 358
(2009) 203.
[13] a C. Rameshan, W. Stadlmayr, C. Weilach, S. Penner, H. Lorenz, M. Hävecker, R.
Blume, T. Rocha, D. Teschner, A. Knop-Gericke, R. Schlögl, N. Memmel, D.
Zemlyanov, G. Rupprechter, B. Klötzer, Angew. Chem. Int. Ed. 49 (2010) 3224;
b K. Föttinger, J.A. van Bokhoven, M. Nachtegaal, G.n. Rupprechter, J. Phys.
Chem. Lett. 2 (2011) 428;
c T. Conant, A.M. Karim, V. Lebarbier, Y. Wang, F. Girgsdies, R. Schlögl, A.
Datye, J. Catal. 257 (2008) 64.
[14] M. Behrens, I. Kasatkin, S. Kühl, G. Weinberg, Chem. Mater. 22 (2009) 386.
[15] N. Iwasa, T. Mayanagi, W. Nomura, M. Arai, N. Takezawa, Appl. Catal. A 248
(2003) 153.
[16] W. Jochum, S. Penner, R. Kramer, K. Föttinger, G. Rupprechter, B. Klötzer, J.
Catal. 256 (2008) 278.
[17] A.G. Christy, S.M. Clark, Phys. Rev. B 52 (1995) 9259.
[18] G. Demazeau, I. Omeran, M. Pouchard, P. Hagenmuller, Mater. Res. Bull. 11
(1976) 1449.
[19] S.-J. Kim, S. Lemaux, G. Demazeau, J.-Y. Kim, J.-H. Choy, J. Mater. Chem. 12
(2002) 995.
[20] A. Eyssler, P. Mandaliev, A. Winkler, P. Hug, O. Safonova, R. Figi, A. Weidenkaff,
D. Ferri, J. Phys. Chem. C 114 (2010) 4584.
Acknowledgments
The authors thank Edith Kitzelmann (XRD), Gudrun Auffermann
(ICP-OES), Gisela Lorenz (BET), and Stafania Sabatino (FTIR) for
their help with various characterizations. Gregor Wowsnick (bulk
Pd2Ga), Matthias Friedrich and Marc Armbrüster (bulk PdZn) and
Stefanie Kühl (CuZnAl) are acknowledged for providing the used
reference samples. Paul Scherrer Institute (SLS, SuperXAS) is
acknowledged for allocation of beamtime. Olga Safonova, Maarten
Nachtegaal, Ye Lu, Andrey Tarasov, and Patrick Kast are acknowl-
edged for their assistance during beamtime. Many thanks are given
to the COST Action CM0904 ‘‘Network for Intermetallic Compounds
as Catalysts in the Steam Reforming of Methanol’’ for providing this
multi-collaboration.
[21] M. Friedrich, A. Ormeci, Y. Grin, M. Armbruster, Z. Anorg. Allg. Chem. 636
(2010) 1735.
[22] N. Sheppard, C. De La Cruz, Catal. Today 70 (2001) 3.
[23] F. Prinetto, M. Manzoli, G. Ghiotti, M.d.J. Martinez Ortiz, D. Tichit, B. Coq, J.
Catal. 222 (2004) 238.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2012.05.020.
[24] N. Iwasa, H. Suzuki, M. Terashita, M. Arai, N. Takezawa, Catal. Lett. 96 (2004)
75.
[25] T. Fujitani, M. Saito, Y. Kanai, T. Watanabe, J. Nakamura, T. Uchijima, Appl.
Catal. A 125 (1995) L199.
[26] a A.P. Tsai, S. Kameoka, Y. Ishii, J. Phys. Soc. Jpn. 73 (2004) 3270;
b A. Bayer, K. Flechtner, R. Denecke, H.-P. Steinrück, K.M. Neyman, N. Rösch,
Surf. Sci. 600 (2006) 78;
References
[1] K. Kovnir, M. Armbrüster, D. Teschner, T.V. Venkov, F.C. Jentoft, A. Knop-
Gericke, Y. Grin, R. Schlögl, Sci. Technol. Adv. Mater. 8 (2007) 420.
[2] B. Halevi, E.J. Peterson, A. DeLaRiva, E. Jeroro, V.M. Lebarbier, Y. Wang, J.M.
Vohs, B. Kiefer, E. Kunkes, M. Havecker, M. Behrens, R. Schlögl, A.K. Datye, J.
Phys. Chem. C 114 (2010) 17181.
[3] a X. Duan, D.G. Evans, Layered Double Hydroxides, Springer-Verlag, Berlin,
Heidelberg, 2006;
b C. Forano, T. Hibino, F. Leroux, C. Taviot-Gueho, in: F. Bergaya, B.K.G. Theng,
G. Lagaly (Eds.), Handbook of Clay Science, Elsevier, 2006, p. 1021.
[4] F. Cavani, F. Trifirò, A. Vaccari, Catal. Today 11 (1991) 173.
[5] A. Ota, M. Armbrüster, M. Behrens, D. Rosenthal, M. Friedrich, I. Kasatkin, F.
Girgsdies, W. Zhang, R. Wagner, R. Schlögl, J. Phys. Chem. C 115 (2010) 1368.
[6] K. Kovnir, J. Osswald, M. Armbruster, D. Teschner, G. Weinberg, U. Wild, A.
Knop-Gericke, T. Ressler, Y. Grin, R. Schlogl, J. Catal. 264 (2009) 93.
[7] G.A. Olah, A. Goeppert, G.K.S. Pakash, Beyond Oil and Gas: The Methanol
Economy, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 2006.
c K.M. Neyman, K.H. Lim, Z.-X. Chen, L.V. Moskaleva, A. Bayer, A. Reindl, D.
Borgmann, R. Denecke, H.-P. Steinrück, N. Rösch, Phys. Chem. Chem. Phys. 9
(2007) 3470;
d Z.-X. Chen, K.M. Neyman, A.B. Gordienko, N. Rösch, Phys. Rev. B 68 (2003)
075417.
[27] B. Halevi, E.J. Peterson, A. Roy, A. DeLariva, E. Jeroro, F. Gao, Y. Wang, J.M. Vohs,
B. Kiefer, E. Kunkes, M. Hävecker, M. Behrens, R. Schlögl, A.K. Datye, J. Catal.
[28] M. Friedrich, D. Teschner, A. Knop-Gericke, M. Armbrüster, J. Catal. 285 (2012)
41.
[29] A. Karim, T. Conant, A. Datye, J. Catal. 243 (2006) 420.
[30] C. Rameshan, W. Stadlmayr, S. Penner, H. Lorenz, N. Memmel, M. Hävecker, R.
Blume, D. Teschner, T. Rocha, D. Zemlyanov, A. Knop-Gericke, R. Schlögl, B.
Klötzer, Angew. Chem. Int. Ed. 51 (2012) 3002.
[31] L. Li, B. Zhang, E. Kunkes, K. Föttinger, M. Armbrüster, D. Sheng Su, W. Wei, R.
Schlögl, M. Behrens, ChemCatChem, htttp://dx.doi.org10.1002/cctc.201200268.