Characterisation and microstructure of Pd and bimetallic Pd-Pt catalysts during methane oxidation

Article abstract of DOI:10.1016/j.jcat.2006.10.029

The catalytic oxidation of methane was studied over Pd/Al₂O₃ and Pd-Pt/Al₂O₃. It was found that the activity of Pd/Al₂O₃ gradually decreases with time at temperatures well below that of PdO decomposition. The opposite was observed for Pd-Pt/Al₂O₃, of which the activity decreases slightly with time. Morphological studies of the two catalysts showed major changes during operation. The palladium particles in Pd/Al₂O₃ are initially composed of smaller, randomly oriented crystals of both PdO and Pd. In oxidising atmospheres, the crystals become more oxidised and form larger crystals. The activity increase of Pd-Pt/Al₂O₃ is probably related to more PdO being formed during operation. The particles in Pd-Pt/Al₂O₃ are split into two different domains: one with PdO and the other likely consisting of an alloy between Pd and Pt. The alloy is initially rich in palladium, but the composition changes to a more equalmolar Pd-Pt structure during operation. The ejected Pd is oxidised into PdO, which is more active than its metallic phase. The amount of PdO formed depends on the oxidation time and temperature.

Full text of DOI:10.1016/j.jcat.2006.10.029

Journal of Catalysis 245 (2007) 401–414
www.elsevier.com/locate/jcat
Characterisation and microstructure of Pd and bimetallic Pd–Pt catalysts
during methane oxidation
Katarina Persson a,∗, Kjell Jansson b, Sven G. Järås a
a
Chemical Technology, Department of Chemical Engineering and Technology, KTH, Teknikringen 42, SE-100 44, Stockholm, Sweden
b
Arrhenius Laboratory, Department of Inorganic Chemistry, University of Stockholm, S. Arrhenius väg 12, SE-106 91 Stockholm, Sweden
Received 15 September 2006; revised 25 October 2006; accepted 28 October 2006
Available online 30 November 2006
Abstract
The catalytic oxidation of methane was studied over Pd/Al O and Pd–Pt/Al O . It was found that the activity of Pd/Al O gradually decreases
2
3
2
3
2 3
with time at temperatures well below that of PdO decomposition. The opposite was observed for Pd–Pt/Al O , of which the activity decreases
2
3
slightly with time. Morphological studies of the two catalysts showed major changes during operation. The palladium particles in Pd/Al O are
2
3
initially composed of smaller, randomly oriented crystals of both PdO and Pd. In oxidising atmospheres, the crystals become more oxidised and
form larger crystals. The activity increase of Pd–Pt/Al O is probably related to more PdO being formed during operation. The particles in Pd–Pt/
2
3
Al O are split into two different domains: one with PdO and the other likely consisting of an alloy between Pd and Pt. The alloy is initially rich
2
3
in palladium, but the composition changes to a more equalmolar Pd–Pt structure during operation. The ejected Pd is oxidised into PdO, which is
more active than its metallic phase. The amount of PdO formed depends on the oxidation time and temperature.
© 2006 Elsevier Inc. All rights reserved.
Keywords: Palladium; Platinum; Bimetal; Methane; TEM; PXRD; TPO; XPS; Stability; Catalytic combustion
1. Introduction
is considerably reduced, and the formation of thermal NOx
is decreased. Natural gas is a frequently used fuel in gas tur-
bines, and the major combustible component in natural gas is
methane.
Catalytic oxidation of methane under oxygen-rich condi-
tions has received considerable attention. One of the reasons
is the increasing interest in lean-burn natural gas vehicles
(NGVs). Compared with diesel engines, NGV engines produce
less NOx and particulates [1–3]. However, a problem with NGV
engines is the high level of unburned methane in the exhaust.
Because methane is a potent greenhouse gas, it is important to
reduce the amount emitted into the atmosphere. One approach
to abating methane emission is the use of catalytic exhaust con-
verters, in which the abatement is achieved by catalytic oxida-
tion.
A catalyst that often has been used for these applications is
palladium supported on various washcoats [6,7]. Many studies
have shown that the activity over fresh palladium catalysts is ex-
cellent for methane oxidation under lean conditions. However,
it has recently been reported that palladium-supported catalysts
have poor stability for methane conversion when the tempera-
ture is kept constant [8–13]. The initially high activity when the
catalysts are fresh drops significantly during operation, result-
ing in increasing difficulty in igniting the methane at desirable
temperatures.
Different additives may be used to improve the palladium
catalyst. One of the more promising additives is Pt, which con-
siderably improves the stability during reaction [8,9,11,14]. Re-
cent results from our lab have shown even a slight increase in
the activity for the Pd–Pt catalysts with time [11,15]. A con-
siderable number of papers have also shown that the activity
of the bimetallic Pd–Pt catalysts is improved in comparison to
Another application for catalytic oxidation of methane is
in catalytic gas turbine combustors [4,5]. Instead of combust-
ing the fuel homogeneously using a flame, the fuel is com-
busted over catalysts. In this way, the combustion temperature
*
Corresponding author. Fax: +46 8 108579.
E-mail address: katarina.persson@ket.kth.se (K. Persson).
0021-9517/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.10.029

402
K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
the monometallic palladium catalysts [8,16–18]. These proper-
ties make the bimetallic Pd–Pt catalysts more attractive than the
monometallic palladium catalysts.
(GC Varian 3800), equipped with a thermal conductivity detec-
tor.
Two different activity tests were performed for the as-
prepared catalysts. In the first test, the catalyst was studied for
12 h with the temperature maintained at 500 ◦C. In the second
test, the temperature was varied among 500, 550, and 600 ◦C,
with each temperature maintained for 30 min. A new sample
was used for each experiment.
In this work, the combustion behaviour during methane ox-
idation is correlated to morphologic changes of the catalyst on
various treatments. To investigate the reason for the decrease or
increase in activity for the monometallic palladium catalyst and
the bimetallic Pd–Pt catalyst, respectively, various temperature-
programmed oxidation analyses were carried out. The samples
were also studied by TEM, EDS, PXRD, and XPS. Further-
more, the thermal aging properties were evaluated for both the
monometallic and bimetallic catalysts.
2.3. Characterisation
Inductively coupled plasma–atomic emission spectroscopy
(ICP-AES) was performed on the catalysts to verify the amounts
of palladium and platinum. The specific surface areas of the
catalyst powders were measured by nitrogen adsorption at liq-
uid N2 temperature in a Micromeritics ASAP 2010 instrument.
The surface area was determined according to the Brunauer–
Emmett–Teller (BET) theory. Before analysis, the samples were
degassed for at least 2 h at 250 ◦C.
The crystal phases in the catalysts were monitored by pow-
der X-ray diffraction (PXRD), using both a Guinier–Hägg cam-
era with a radius of 40 mm and diffractometer (Siemens Dif-
fraktometer D5000). In the former case, Si was added as an
internal standard, and the powder X-ray photographs were eval-
uated by the Scanner System program [19] to obtain the d val-
ues. The phases were identified using the JCPD database. The
mean crystal sizes were estimated from the diffractometer data
using the Fundamental Parameter Approach application of the
TOPAS v2.0 software.
The redox properties of the catalysts were analysed by
temperature-programmed oxidation (TPO). The experiments
were carried out using a Micromeritics AutoChem 2910
equipped with a thermal conductivity detector. As-prepared cat-
alyst powder (100 mg) was placed on top of glass wool in a
quartz reactor, and 5 vol% O2/He was continuously flowed over
the sample at 20 ml/min. The temperature was cycled between
300 and 900 ◦C at a rate of 10 ◦C/min. The temperature cycle
was repeated twice; the results from the second cycle are pre-
sented in this paper, because water desorption affected the first
cycle. For most experiments, the sample was preoxidised at var-
ious temperatures and for varying durations before the second
temperature cycle. A new sample was used for each analysis.
The morphologies and sizes of the noble metal particles on
the alumina were studied by transmission electron microscopy
(TEM), using a JEOL 2000 FX microscope equipped with an
energy-dispersive X-ray spectroscopy (EDXS) detector (LINK
AN 10000). Surface analyses of the samples were conducted
2. Experimental
2.1. Catalyst preparation
A palladium catalyst and a palladium–platinum catalyst,
both supported on γ -Al2O3 (PURALOX HP-14/150, Sasol
Germany GmbH), were investigated in this study. The two cata-
lysts were prepared so as to have equal loadings of noble metals
(470 µmol metal/g catalyst powder). The molar ratio between
Pd and Pt in the bimetallic catalyst was 2:1. More details on the
catalysts are given in Table 1.
The alumina powder was impregnated with the metal/metals
by the incipient wetness technique. Solutions of palladium
(Johnson Matthey) and platinum nitrate (Kaabs Materials AB)
were mixed with deionised water before impregnation. The sup-
port was impregnated twice, with a drying step at 300 ◦C for 4 h
between steps, and calcined at 1000 ◦C for 1 h.
The catalyst powders were mixed with ethanol before
ball milling. Cordierite monoliths (400 cpsi, Corning), with
∅14 mm and length 10 mm, were then dip-coated in the slurry,
followed by a drying step at 100 ◦C. This procedure was re-
peated until 20 wt% of catalyst material was fastened onto
the monolith. Finally, the coated monoliths were calcined at
1000 ◦C for 2 h.
2.2. Activity tests
Catalytic activity measurements were carried out at at-
mospheric pressure using a tubular reactor placed inside a
programmable furnace. A gas mixture of 1.5 vol% CH4 in air
was fed to the reactor at a space velocity of 250 000 h−1 for all
types of activity tests. The temperature of the gas was recorded
by a thermocouple placed upstream of the monolith. The re-
action products were analysed on-line by gas chromatography
Table 1
A summary of the catalysts and their textural properties
a
a
Sample
Composition
5 wt% Pd/Al O
2 3
Pd loading
(wt%)
Pt loading
(wt%)
BET surface
area (m /g)
Crystal size
PdO (nm)
Crystal size
Pd–Pt (nm)
2
Pd/Al O
5.1
4.8
3.0
3.0
0
0
3.3
2.8
102
84
91
26
27
19
20
–
–
34
33
2
3
Pd/Al O -aged
5 wt% Pd/Al O -aged
2
3
2 3
PdPt/Al O
PdPt/Al O -aged
2:1 Pd:Pt/Al O
2
3
2 3
2:1 Pd:Pt/Al O -aged
71
2
3
2 3
a
Estimated by TOPAS—Fundamental Parameter Approach.

K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
403
with a Shimadzu ESCA 3400 using MgKα radiation. The sam-
ples were fastened onto one side of double-adhesive tape, the
other side fastened onto the instrument sample holder. The ex-
periments were performed at a pressure of 10−6 Pa using a
radiation power of 10 kV and 20 mA. After a scan in the
range of 0–600 eV, the energy ranges 62–80 eV (Al2p), 280–
300 eV (C1s), 330–350 eV (Pd3d), and 523–544 eV (O1s)
were scanned for 12 h. Al 1s was used to correct spectra for
charge-up effects, where spectra were corrected to an energy of
74.5 eV. The spectrometer resolution was taken from the C1s
spectra, and convolution was performed using all three energy
states, 335.2 eV (Pd0), 336.7 eV (Pd2+), and 337.5 eV (Pd4+),
for each sample.
methane conversion, but the activity dropped rapidly with time.
After 12 h, the conversion was only 11%.
In the same figure, the gray dotted curve represents the ac-
tivity of Pd/Al2O3 with the temperature alternated among 500,
550, and 600 ◦C. It is clear that the deactivation was enhanced
at higher temperatures, because the activity already at the third
temperature step was lower than the activity measured when
the temperature was maintained at 500 ◦C. The conversion at
the end of the test was only 4%, whereas that after the same
time for the sample maintained at 500 ◦C was 14%.
Fig. 2 shows the results of the two activity tests over
PdPt/Al2O3, but in a different range than displayed in Fig. 1.
For the 12-h test, the initial activity was lower than its mono-
metallic counterpart, and the activity showed a small drop dur-
ing the first 50 min. This initially lower activity of the bimetallic
catalyst is expected because the amount of palladium in the
catalyst was only two-thirds of the amount in Pd/Al2O3. In
addition, palladium is generally known to have higher activ-
ity for methane combustion than platinum [20]. However, the
conversion actually increased with time over the remainder of
the activity test. The increase levelled out at the end of the
test.
The gray dotted curve in Fig. 2 represents the activity over
PdPt/Al2O3 at 500, 550, and 600 ◦C. The higher-temperature
steps appear to enhance the activity slightly at 500 ◦C, in-
dicating that the temperature is important to the activation
process. It is interesting to note that the activity of PdPt/Al2O3
was actually higher than the activity of Pd/Al2O3 after only
6.6 h for the samples maintained at 500 ◦C; when the temper-
ature was cycled, PdPt/Al2O3 achieved superior activity more
rapidly.
2.4. Thermal aging
Both the catalytic powders and the monoliths were aged for
10 h in a tubular furnace at 1000 ◦C with a flow of air of 15 vol%
of steam. The activity tests were carried out under the same
conditions as described in Section 2.2, except for the temper-
ature being increased stepwise from 470 to 770 ◦C, with each
temperature step maintained for 1 h. Both as-prepared and aged
monoliths were tested. Before the activity tests, the temperature
was increased to 900 ◦C to stabilise the samples.
3. Results and discussion
3.1. Catalytic activity
The results from the activity measurements over Pd/Al2O3
and PdPt/Al2O3 are shown in Figs. 1 and 2, respectively. The
black curves display the activity over 12 h with the inlet temper-
ature to the catalyst maintained at 500 ◦C. As shown in Fig. 1,
the Pd/Al2O3 catalyst initially had excellent activity with 53%
The poor stability of methane conversion over monometal-
lic palladium catalysts has been explained by Narui et al. [8] as
a size effect of the Pd particles. Adding platinum to this cata-
Fig. 1. Methane conversion versus time on stream over Pd/Al O . The black
Fig. 2. Methane conversion versus time on stream over PdPt/Al O . The black
2 3
2
3
◦
◦
curve represents the activity of the sample kept at 500 C and the gray dotted
curve represents the activity of the sample kept at 500 C and the gray dotted
◦
◦
curve represents the activity of the sample at 500, 550 and 600 C, respectively.
curve represents the activity of the sample at 500, 550 and 600 C, respectively.

404
K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
lyst resulted in considerably more stable activity. Their Pd–Pt
catalyst showed lower particle growth than its monometal-
lic counterpart. Araya et al. [21] reached a different conclu-
sion, that the loss of dispersion of their Pd-based catalyst was
too small to be essential. Their fresh catalyst had a disper-
sion of 31.7%, and after the catalyst was used for 96 h, the
dispersion was 28.2%. Nonetheless, these authors observed a
significant drop in activity with time for their Pd-based cata-
lyst.
The poor stability of monometallic Pd catalysts also has been
attributed to poisoning by the water produced during oxida-
tion of methane [13,22,23]. The poisoning process has been
suggested to be due to formation of Pd(OH)2, which is less ac-
tive for methane oxidation, resulting in decreased activity. Even
though most authors have reported decreased activity when ex-
tra water is added to the feed stream, this does not prove that
Pd(OH)2 has been formed, especially not at this high temper-
ature, where hydrothermal processes of alumina may occur.
Thus, there may be other explanations for deactivation besides
Pd(OH)2 formation, for instance, hydrothermal reaction with
the support.
cooling can be attributed to oxidation of metallic Pd back to
PdO. The temperature hysteresis between decomposition and
reoxidation is a well-known phenomenon in supported palla-
dium catalysts [26,27]. The cause of this hysteresis has been
suggested to be an effect of a nucleation mechanism when Pd
is reoxidised [7]; however, the literature on this hysteresis phe-
nomenon is diverse.
The decomposition peaks during heating changed with
preoxidation. The total area of the decomposition peaks in-
creased slightly with preoxidation, indicating that more PdO
was formed during preoxidation. If this is also true under the
reaction conditions, then deactivation of Pd/Al2O3 may not be
due to loss of the PdO phase.
The ratio between the areas of the two decomposition peaks
also changes with preoxidation. The first peak became smaller
with increasing preoxidation time, whereas the second peak
became larger. These results are similar to those of Groppi
et al. [25], who explained this behaviour with two different Pd
oxide species. The first peak represents one type of oxide, trans-
ferred during heating into a second type of PdO, shown as the
second peak. However, the origin of these species has not been
established. During cooling, the reoxidation peak was not af-
fected by the preoxidation time.
3.2. Oxidation for various lengths of time
For the as-prepared Pd/Al2O3 sample, a broad decrease of
the oxygen signal between 450 and 630 ◦C was observed during
heating, indicating that the sample was taking up oxygen from
the feed gas. This is probably because the reoxidation of Pd
was not completed during the first cooling ramp (not shown in
figure), but continues during the second heating ramp. When the
sample was preoxidised, the oxygen uptake during the heating
ramp disappeared. Thus, the oxidation of Pd proceeded further
during the preoxidation step.
The changes of the catalysts with time while temperature
was held constant at 500 ◦C have been studied by TPO; the re-
sults are shown in Figs. 3 and 4. The catalyst powders were pre-
oxidised at 500 ◦C before the second TPO cycle. For Pd/Al2O3,
presented in Fig. 3, two positive peaks occurred during heating.
The first, small peak appeared at around 730 ◦C; the second,
larger peak appeared at 790 ◦C and persisted up to 900 ◦C. Dur-
ing cooling, a negative peak occurred, starting at 630 ◦C and
ending at 460 ◦C. The peaks appearing during heating are gen-
erally known to be due to oxygen release when PdO decom-
poses into its metallic state [24,25]. The negative peak during
The incomplete reoxidation on cooling in the TPO exper-
iment was also suggested by the reoxidation coefficient δ, de-
Fig. 3. TPO profiles of Pd/Al O . The sample was either as-prepared or pre-
Fig. 4. TPO profiles of PdPt/Al O . The sample was either as-prepared or pre-
2 3
2
3
oxidised for various lengths of time.
oxidised for various lengths of time.

K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
405
Table 2
◦
The reoxidation coefficient δ of the samples that were preoxidised at 500 C
during various lengths of time
Sample
Pd/Al O
As-prepared
1 h
5 h
10 h
1.33
2.43
1.37
3.07
–
4.03
1.41
3.88
2
3
PdPt/Al O
2
3
fined here as the area of the decomposition peaks divided by the
area of the reoxidation peak. If the catalyst were to be as oxi-
dised after the cooling ramp as before the decomposition peaks,
then δ would be equal to 1; that is, the areas of the decompo-
sition peaks and of the reoxidation peak would be equal. As
shown in Table 2, this is not the case for Pd/Al2O3, but the
value was >1, reflecting incomplete oxidation during cooling.
The δ value increased to some extent with increasing preoxi-
dation time. This result was associated more with variations in
the area of the decomposition peaks than with variations in the
reoxidation peak area.
The TPO profile of the PdPt/Al2O3 catalyst is shown in
Fig. 4 for various preoxidation periods at 500 ◦C. The reduc-
tion peak during heating appears at a lower temperature than
for Pd/Al2O3, indicating that Pt promotes the reduction of PdO.
This is in line with previous reports [9,28]. The temperature
at the maximum of the reduction peak was 664 ◦C for the as-
prepared PdPt/Al2O3 catalyst. By preoxidising the sample, the
peak maximum moved to slightly higher temperatures.
The preoxidation time strongly affected the intensity of the
reduction peak, considerably more than for the Pd/Al2O3 cata-
lyst. The peak intensity of the as-prepared PdPt/Al2O3 catalyst
was significantly lower than that for the sample preoxidised for
1 h. When the sample was preoxidised for 5 h, the peak intensity
was even larger. However, the change was smaller between 5
and 10 h of preoxidation. Our previous high-temperature XRD
studies [15] showed that the TPO peak during heating was most
likely attributable to PdO decomposition, as in the case of the
monometallic Pd catalyst. Therefore, the increase in the posi-
tive peak during heating presumably was due to increased PdO
formation during preoxidation. If this was also the case under
reaction conditions, then it was not surprising that the activity
increased with time. Previous studies at our laboratory showed
that the Pd–Pt alloy itself has poor activity for methane com-
bustion [29]. Consequently, if the PdO content in the catalyst
increases, then the activity also increases.
Fig. 5. TPO profiles of Pd/Al O . The sample was either as-prepared or pre-
oxidised at various temperatures.
2
3
3.3. Oxidation at various temperatures
3.3.1. The Pd/Al2O3 catalyst
To investigate the influence of temperature on the catalysts,
the samples were preoxidised at various temperatures for 1 h
before the second TPO cycle. The TPO profiles of Pd/Al2O3 are
presented in Fig. 5. Preoxidation at 500 and 600 ◦C minimised
oxygen uptake between 450 and 630 ◦C for the as-prepared
sample during heating.
Small variations of the PdO decomposition peaks were ob-
served for the various temperatures. The sample preoxidised at
600 ◦C showed a slightly lower intensity of the first peak, but
the second peak was somewhat higher than that for the sample
preoxidised at 500 ◦C. This indicates that both temperature and
duration of preoxidation affect the decomposition peaks.
The total area of the PdO decomposition peaks increased
slightly with preoxidation temperature, indicating that more
PdO was formed at 600 ◦C than at 500 ◦C. Despite the higher
PdO concentration, a greater drop in activity occurred at
the higher temperatures (see Fig. 1). Therefore, a decreasing
amount of PdO most likely was not the reason for the loss of
activity.
During cooling, the reoxidation peak appeared at the same
temperature, independent of the duration of preoxidation. How-
ever, the sample that was preoxidised for 10 h had a slightly
larger reoxidation peak, whereas the as-prepared sample had the
smallest peak. Nonetheless, the reoxidation peak during cooling
appeared at a lower temperature than for Pd/Al2O3.
The reoxidation coefficients were also determined for this
catalyst and are given in Table 2. All δ values for PdPt/Al2O3
were clearly >1 and in most cases increased with the duration
of preoxidation. Again, this is related mostly to the increasing
reduction peak and less to the changes in the reoxidation peak.
This result reflects a slow reoxidation process that did not reach
completion before the decomposition peak.
The reoxidation peaks during cooling were similar for the
various preoxidation temperatures. This result indicates that
once the catalyst is reduced, the material obtains the same mor-
phology independent of the pretreatment.
The reoxidation coefficients of Pd/Al2O3 are compiled in
Table 3 for the various temperatures; all values are >1. The δ
value of the sample preoxidised at 600 ◦C is greater than that
of the sample preoxidised at 500 ◦C, likely associated with the
higher decomposition peaks.
To gain insight into the behaviour of Pd/Al2O3 in the TPO
experiments, samples were cooled in helium to ambient tem-
perature at different stages during experiments. Thereafter, the
samples were studied ex situ by PXRD and TEM/EDS. Four

406
K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
Table 3
darker spots, whereas the alumina support is shown as the light-
gray areas.
The reoxidation coefficient δ of the samples that were preoxidised for 1 h at
different temperatures
Fig. 7a shows a representative palladium particle of the as-
prepared sample. As shown in the figure, the centre of the
particle is darker than the external parts. From the PXRD re-
sults, it can be concluded that both PdO and Pd are present
in the as-prepared material. Thus, the darker parts of the par-
ticle shown in the TEM image are most likely metallic palla-
dium (as an effect of denser material), whereas the lighter parts
presumably consist of PdO. Hence, the palladium-containing
particles appear to be not fully oxidised, but rather to con-
sist of a mixture of both Pd and PdO domains. Even though
PdO is the thermally favoured phase at this temperature range,
Ciuparu et al. [7] reported that metallic Pd and PdO may co-
exist in a catalyst for extended periods due to a slow oxidation
process.
◦
◦
◦
◦
◦
◦
Sample
As-pre- 400 C 500 C 600 C 655 C 665 C 800 C
pared
Pd/Al O
1.33
–
1.37
3.07
1.39
3.68
–
–
–
2
3
PdPt/Al O 2.43
2.32
3.06
2.89
2.56
2
3
The lack of facetted crystals indicates that the particles in
the as-prepared material were composed of smaller (5–10 nm),
randomly oriented crystals. At the outer surface of the parti-
cle, patches of smaller (3–5 nm) crystals are seen. According
to the elemental analyses, these particles consist of palladium,
but the oxidation state of the palladium cannot be determined
using EDS. In addition, the particle size likely is too small to
be detected by PXRD. Due to the polycrystalline PdO particles
around the areas of Pd metal, the description of this structure
is inconsistent with the core–shell model, which is supposed to
have uniform layers of PdO around the Pd metal.
Fig. 6. PXRD patterns of Pd/Al O for (A) as-prepared, (B) preoxidation at
2
3
◦
◦
◦
500 C, (C) preoxidation at 775 C, (D) preoxidation at 900 C. Reflex from
δ-Al O is labelled as S.
2
3
The oxidation mechanism of Pd is not very clear, and various
ideas on how it occurs have been proposed. Lyubovsky et al.
[30] observed by TEM rough polycrystalline PdO particles on
reoxidation by cooling the sample after thermal decomposition.
Based on this result, they excluded the possibility of a process
involving growth of a surface oxide on a shrinking metal core.
This idea appears to be consistent with the findings presented
in this paper. Because the as-prepared catalyst was calcined at
1000 ◦C (a temperature above the thermal stability of PdO), the
observed morphology is most likely due to this phenomenon of
reoxidation, resulting in polycrystalline particles with mixtures
of small PdO and Pd crystals.
The morphology changed after the catalyst was preoxidised
at 500 ◦C. A considerable amount of the particles became fully
oxidised, and some appeared to have a more crystalline struc-
ture. Fig. 7b shows an image of a representative palladium
particle; the internal structure is composed of single crystals
that overlap each other and form moire fringes with a rougher
material on the outside surface. The inner part is most likely
due to the small crystals, shown in the as-prepared material,
having agglomerated into a larger crystal. In other cases, the
particles appear to be built up of small PdO domains, as shown
in Fig. 7c.
different samples were selected. The first sample was the as-
prepared catalyst. The second sample was preoxidised for 10 h
at 500 ◦C, the same temperature as in the activity test. The third
sample was treated in the same way as the second sample before
heating to 775 ◦C, a temperature at which the PdO decomposi-
tion started but did not reach completion. The last sample was
taken after 5 min at 900 ◦C, when decomposition was com-
pleted.
The results of the PXRD analyses are presented in Fig. 6.
All samples were supported on δ-Al2O3. The as-prepared sam-
ple consisted mostly of PdO (2θ = 33.8◦), but small amounts of
metallic Pd (2θ = 40.1◦) were also detected. When the material
was preoxidised at 500 ◦C, the peak indicating metallic palla-
dium vanished along with an increase in the PdO reflection,
demonstrating that the sample was somewhat more oxidised.
When the temperature was raised to 775 ◦C, PdO decomposi-
tion started but was not completed, as demonstrated by reflec-
tions from both PdO and Pd. In the sample obtained at 900 ◦C,
only metallic palladium was detected. The PXRD outcome is
well in agreement with the TPO results. Furthermore, a sample
taken from the monolith after the 12 h-activity test had a similar
PXRD pattern (not shown in figure) as the sample preoxidised
at 500 ◦C, confirming that the preoxidation procedure is repre-
sentative of the used material.
According to Ho et al. [31], the uptake of oxygen for Pd
is very slow at ambient temperature and it remains adsorbed
on the surface of the Pd crystals without further penetration
into the bulk. Bulk oxidation into PdO becomes pronounced
only above 500 ◦C. Hence, when storing the incompletely oxi-
dised as-prepared sample at ambient temperature, insignificant
amounts of PdO will be formed. When the temperature is in-
Images from the morphological study by TEM/EDS of the
four samples are presented in Figs. 7a–7e. The analyses reveal
that all samples have well-distributed palladium-containing par-
ticles on the alumina surface in the range of 30–100 nm. Ac-
cording to the element analyses, the palladium particles are the

K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
407
(a)
(b)
(c)
(d)
(e)
◦
◦
Fig. 7. TEM images of the Pd/Al O catalyst. (a) As-prepared, (b, c) preoxidised at 500 C for 10 h, (d) preoxidised at 500 C for 10 h followed by further heating
2
3
◦
◦
up to 775 C where PdO decomposition has started but is incomplete and (e) heated up to 900 C where PdO decomposition is completed.
creased to 500 ◦C, as was the case for the preoxidised samples,
the oxidation will proceed further. The small crystals observed
in the as-prepared material are likely to reach full oxidation
and agglomerate into larger particles, resulting in fewer grain
boundaries. This may perhaps be one of the explanations of the
gradually decreasing activity.

408
K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
Lyubovsky et al. [32] have reported an increase in the activ-
3.3.2. The PdPt/Al2O3 catalyst
Preoxidation of PdPt/Al2O3 was also studied by TPO at var-
ious temperatures, as shown in Fig. 8. The intensity of the
decomposition peak during heating was clearly affected by
the preoxidation temperature. The preoxidation at 400 ◦C gave
only a small increase in the peak intensity compared to the
as-prepared sample. When the temperature was increased to
500 ◦C, the decomposition peak became considerably larger,
but the largest peak after preoxidation for 1 h was obtained for
the sample preoxidised at 600 ◦C. The sample preoxidised at
800 ◦C showed any increase in the peak intensity, similar to the
as-prepared sample. Hence, PdO formation with time is possi-
ble only below the temperature of decomposition.
The significant variation of the reduction peak with pre-
oxidation temperature suggests that more PdO was formed at
600 ◦C than at 500 ◦C, which would explain the larger increase
of methane conversion with time at the higher temperatures (see
Fig. 2). However, preoxidation at temperatures above the de-
composition of PdO are not favourable for PdO formation and
thus also not for activity.
The reoxidation coefficients of PdPt/Al2O3 are shown in
Table 3 for the various temperatures. A larger variation is seen
for PdPt/Al2O3 than for Pd/Al2O3, due mostly to the size of the
reduction peak and less to the reoxidation peak. During cooling,
the reoxidation peaks were almost equal for the various temper-
atures. Thus, the bimetallic sample formed more PdO during
heating compared with the monometallic sample.
By preoxidising the catalyst at a temperature on the decom-
position peak, a double peak was obtained, as shown in Fig. 9.
When preoxidation was carried out at 655 ◦C, the first peak was
smaller and appeared at 659 ◦C, whereas the second peak was
considerably larger and appeared at 676 ◦C. The reverse be-
haviour occurred for the sample preoxidised at 665 ◦C, where
the first peak was larger than the second peak. For both sam-
ity after treating the catalyst above the thermal decomposition
temperature. They suggest that this phenomenon results from
the rougher surface formed on reoxidation being more active
than the fully oxidised catalyst. This is consistent with the re-
sults presented in this paper, because the as-prepared catalyst
has polycrystalline morphology. During reaction, however, the
oxidation of the catalyst is most likely proceeding further, as all
characterisation techniques used in this study indicate, losing
active surface and thereby activity.
According to Carstens et al. [33], the activity for methane
combustion may be enhanced by producing small amounts of
metallic Pd on the surface of PdO. Pd is more effective in dis-
sociatively adsorbing methane, which normally is a difficult
molecule to split. The fragments of the methane molecules are
then rapidly diffused to the Pd–PdO interface, where reduc-
tion of the palladium oxide occurs. Increasing the degree of
oxidation of the sample, which properly takes place during re-
action, decreases the number of Pd–PdO interfaces, possibly
causing a decline in activity. However, the extent of the effects
of oxidation and crystal growth on catalytic activity is difficult
to quantify; other deactivation processes may occur simultane-
ously.
Another feature of the small crystals of metallic palladium
is that they can act as nucleation spots for the decomposition
of PdO, when Pd crystals are in boundary contact with PdO
crystals. This may be the cause of the first decomposition peaks
in the TPO experiments. If this is the case, then the amount of
metallic Pd in a sample reflects the size of the first peak in a
TPO experiment. This is in line with the first peak in the TPO
profile of the preoxidised samples being smaller, because the
sample contains less metallic Pd.
After a TPO experiment has reached 775 ◦C, the decompo-
sition of PdO into Pd is halfway completed. From TEM investi-
gations the material exhibits individual palladium particles that
are completely or partly reduced, but also some particles that
appear to be unaffected. The partly reduced particles reveal that
the transformation starts with a nucleation of Pd on the PdO
crystals (see Fig. 7d). As the reduction progresses, the metallic
palladium nucleates and grows on each crystal, forming a num-
ber of crystals. By heat treatment at 900 ◦C, these may form
a single crystal with time. This structure with partly reduced
particles during the thermal reduction of PdO is similar to that
observed by Datye et al. [34], who found only metallic Pd on
the surface of the PdO crystal. In contrast, Voogt et al. [35] sug-
gested a structure consisting of a metallic core with an oxide
skin. The former appears to be more in line with the findings of
the present paper.
A typical TEM image of the sample treated at 900 ◦C is pre-
sented in Fig. 7e. The particles in this sample are well facetted,
indicating that they consist of larger crystals. According to the
PXRD result of the sample heated to 900 ◦C, the particles pre-
sumably consist of metallic palladium. Chen et al. [36] have
observed a similar morphology of a thermally reduced palla-
dium catalyst.
Fig. 8. TPO profiles of PdPt/Al O . The sample was either as-prepared or pre-
oxidised at various temperatures.
2
3

K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
409
(a)
Fig. 9. Temperature-programmed oxidation plots of PdPt/Al O . The sample
2
3
◦
was preoxidised at temperatures on the reduction peak, i.e. 655 and 665 C.
ples, the total area of the peaks was somewhat larger than the
peak area of the sample that was not preoxidised, indicating
that some PdO formed during the treatment. This is probably
related to how far the decomposition has proceeded during the
heat treatment and how rapidly reoxidation is occurring during
cooling. However, further investigation is required to determine
the cause of the double peak.
To investigate the nature of the PdPt/Al2O3 at various tem-
peratures in the TPO experiments, the samples were cooled in
helium to ambient temperature. Preoxidation was carried out
for 10 h for all temperatures before further ex situ analyses. In
the PXRD diffraction patterns of these samples, δ-Al2O3, PdO
and Pd–Pt alloys were the only crystal phases observed. The as-
prepared sample had a PdO peak at 2θ = 33.8◦ and a metallic
phase between the 2θ values of metallic Pd and Pt at 81.3◦ and
82.2◦, respectively. The PdO(101) peaks are shown in Fig. 10a.
The sample preoxidised at 500 ◦C shows a diffraction pattern
similar to that of the as-prepared sample, but the PdO reflection
is considerably larger. At 655 ◦C, the temperature reached the
reduction peak in the TPO experiment, and the amount of PdO
in the sample was smaller, indicating that the decomposition of
PdO had started but had not reached completion. For the sample
obtained at 800 ◦C, the PdO peak vanished. A sample obtained
from the monolith after the 12 h-activity test was evaluated by
PXRD and exhibited a diffraction pattern similar to, but slightly
more oxidised than, that of the sample taken at 500 ◦C.
The changes in PXRD with the various temperatures for the
metallic peaks are shown in Fig. 10b. Calculating the solid solu-
tion composition, PtxPd1−x, revealed that the x value changed
with temperature. The solid solution of the as-prepared sam-
ple had a value of x = 0.43, whereas the solid solution of
the sample obtained at 500 ◦C had a value of x = 0.48. Con-
sequently, by preoxidising the sample below the temperature
of PdO decomposition, the material obtained a solid solution
(b)
Fig. 10. PXRD patterns of the PdPt/Al O catalyst for the (a) (101) reflex of
2
3
PdO (b) (311) reflex of Pt Pd
500 C (B), preoxidised at 655 C (C) and preoxidised at 800 C (D). Reflex
alloy, in as-prepared (A), preoxidised at
x
1−x
◦
◦
◦
from δ-Al O is labelled as S.
2
3
closer to x = 0.5. This is the most likely reason for the ability
of PdPt/Al2O3 to form PdO during reaction. Pd is dissolved out
from the solid solution, and because PdO is the stable phase,
under these conditions, the metallic Pd will oxidise into PdO.
Note, however, that PdO formation appears to be a finite process
and probably does not exceed x = 0.5. This is in line with the
conversion levelling out after 12 h in the activity test, as well as
with the increase of the reduction peak shown in the TPO pro-
files becoming smaller with time. Accordingly, the palladium
and platinum in the catalyst will not end up in two separate par-
ticles.
The solid solution of the sample taken at 800 ◦C had a com-
position close to the nominal value of the catalyst, that is,
x = 0.33. This indicates that once PdO decomposed, the metal-

410
K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
lic palladium was incorporated into the solid solution. When
the catalyst is reoxidising, Pd must diffuse out from the solid
solution to become oxidised, which may explain the slow reox-
idation for PdPt/Al2O3. However, this activation process most
likely will occur only after the catalyst is treated under con-
ditions favourable for PdO decomposition. If the catalyst is
operating below the decomposition point of PdO, then the activ-
ity will most likely stagnate after a while and remain constant.
To further investigate the catalyst, the preoxidised samples
were studied by TEM and EDS. The as-prepared sample con-
tains 30–50 nm noble metal particles, with almost constant
composition. The particles are well distributed and separated in
the alumina support. Each particle was divided into two parts;
EDS measurements confirmed that one part was rich in Pd and
the other part contained both Pd and Pt, thus corresponding to
PdO and PtxPd1−x alloy, respectively.
(a)
No evidence of metallic Pd was found in the Pd-rich part,
signifying that this domain consists of only PdO with no contri-
bution of Pd. The mixed palladium phase (PdO and Pd), which
is claimed to be the more active structure for methane conver-
sion [32,33], was present in the as-prepared Pd/Al2O3 catalyst.
As discussed previously, during the course of the experiment,
the structure in the monometallic catalyst changed as the metal-
lic Pd became oxidised, likely causing a decline of activity. In
the case of the bimetallic catalyst, the metallic Pd appeared
to be absent in the PdO part. Thus, less reconstruction of the
PdO part occurred during operation, possibly resulting in less
changes in activity and higher stability of methane conversion.
The process of ejecting Pd out from the alloy will result only in
increased PdO. Furthermore, due to the close contact with the
alloy, the PdO part in PdPt/Al2O3 will always have a metallic
source, which possibly will aid in splitting of the methane mole-
cules. This may be one reason for the greater activity after 6.6 h,
despite the lower palladium concentration in the PdPt/Al2O3
compared with Pd/Al2O3.
(b)
Fig. 11. TEM images of Pt Pd
/PdO particles in the PdPt/Al O catalyst
after (a) pre-oxidation at 500 C for 10 h, (b) pre-oxidation at 800 C for 10 h.
x
1−x
2 3
◦
◦
Fig. 11a shows a representative image of the material after
preoxidation at 500 ◦C. The particle has a morphology similar
to that of the as-prepared sample, but with a larger Pd-rich part.
The alloy exhibited facetted surfaces, whereas corrosion was
found on the interface with the oxide. The image also shows
that the oxide surfaces contained 2–3 nm particles, similar to
the monometallic palladium material.
was divided into two different components: one at 336.0 eV at-
tributable to Pd2+ and one at 337.3 eV attributable to Pd4+. In
this case, the energy of the Pd2+ component was slightly too
low compared with the values reported in the literature, indicat-
ing the possibility that Pd0 may also be on the surface of the
as-prepared bimetallic Pd–Pt catalyst, which was not assumed
to be the case at the time of this study. Platinum was also ob-
served in two oxidation states (Pt0 and Pt4+) with band maxima
at 314.4 and 317.0 eV, respectively.
At 800 ◦C, only well facetted PtxPd1−x particles were found
(see Fig. 11b). According to the EDS analysis, Pd and Pt were
evenly spread in the noble metal particles. Palladium was never
found without platinum.
This study has investigated how the XPS spectra of the Pd3d
band vary with preoxidation of the PdPt/Al2O3 catalyst. By as-
3.3.3. Surface chemistry
suming that all three oxidation states (Pd0, Pd2+, and Pd4+
)
The surface chemistry of as-prepared Pd and Pd–Pt catalysts
was previously investigated by XPS [29]. The monometallic
palladium catalyst showed a binding energy of 337.1 eV for
Pd3d5/2. The literature reports that this value is slightly too high
for pure PdO (Pd2+, 336.7 eV [37,38]), indicating the possibil-
ity of a mixture with Pd4+ (337.5 eV [39,40]). Pd4+ (PdO2)
is generally known to be highly unstable, but according to
Moroseac et al. [41], metal–support interactions may stabilise
this species. For the bimetallic Pd–Pt catalyst, the Pd3d5/2 band
may be present on the catalyst surfaces, the amount of each state
was determined by convolution using binding energies from the
literature (335.2 eV [42,43], 336.7 eV [37,38], and 337.5 eV
[39,40], respectively). The spectra are shown in Fig. 12, and the
binding energies and the oxidation states of Pd are compiled in
Table 4.
In the as-prepared sample, all oxidation states were ob-
served, dominated by Pd0 and Pd2+. Pd4+ is most likely present

K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
411
By oxidising the sample for 10 h at 500 ◦C, the intensity
of the Pd0 peak decreased along with an increase in the Pd2+
peak. This demonstrates that more PdO was present at the sur-
face of the sample compared to the as-prepared sample, similar
to the bulk composition observed by PXRD. The low fraction
of metallic palladium suggests that even the Pd–Pt alloy was
oxidised on the surface to some extent.
No Pd4+ was detected in the sample treated at 500 ◦C; thus,
it likely was not present during the combustion reaction. This
finding is not surprising, because Pd4+ is well known to have
low stability at higher temperatures. After heating at 800 ◦C,
only the metallic form of palladium was present on the surfaces
of the alloyed particles.
3.4. Thermal aging
The catalysts were aged at 1000 ◦C for 10 h in an air stream
with 15 vol% of steam. This is a very harsh aging, because the
temperature is above the normal operation temperature of the
catalysts and above the temperature window of PdO. However,
these conditions will speed up the aging and give an indication
of the long-term stability of the catalysts. The monoliths and the
catalyst powders were aged at the same time so as to facilitate
evaluation of the activity and characterization of the samples.
Fig. 13 shows the activities of as-prepared and aged Pd/
Al2O3 catalysts. The aged catalyst had considerably lower ac-
tivity than the as-prepared catalyst, where the largest losses
occurred at the third and forth temperature steps. The activity
decreased with time for both samples, although that of the aged
sample decreased less.
Fig. 12. Pd 3d core level spectra recorded by XPS on PdPt/Al O catalyst of
2
3
◦
as-prepared (A), Ar-etched (B), preoxidised at 500 C (C) and preoxidised at
◦
800 C (D).
Table 4
Binding energies of core electrons determined from XPS spectra of the different
catalyst materials
Fig. 14 shows the activities of the as-prepared and aged PdPt/
Al2O3. The loss of activity due to thermal aging was signifi-
cantly less than that for the monometallic Pd catalyst. No dif-
ference is observed between the combustion profiles of the aged
and the as-prepared catalysts for the two final temperature steps
at 720 and 770 ◦C. For the other temperature steps, small vari-
ations are seen. Nevertheless, the ability to maintain a constant
activity is sustained on aging.
Catalyst
Pd3d
(eV)
Oxidation
state
Relative
intensity (%)
5/2
a
5 wt% Pd/Al O
337.1
336.0
337.3
335.2
336.7
337.5
335.2
336.7
337.5
335.2
336.7
337.5
335.2
336.7
337.5
100
22
78
38
47
15
71
29
0
17
83
0
100
0
2
3
a
2:1 PdPt/Al O
2
3
0
PdPt/Al O as-prepared
Pd
2
3
2+
4+
0
Pd
Pd
Pd
Pd
Pd
Pd
Pd
Pd
Pd
Pd
Pd
The loss of activity at temperatures above 670 ◦C is most
likely related to the PdO decomposition, because the TPO
experiments showed PdO decomposition in this temperature
range. However, the TPO decomposition started already at
PdPt/Al O Ar-etched
2
3
2+
4+
0
◦
PdPt/Al O at 500 C
2
3
2+
4+
0
625
◦C. The lower decomposition temperature is probably re-
lated to the fourfold lower partial pressure of oxygen used in
the TPO experiments compared with the activity test. The de-
composition temperature of PdO was strongly influenced by the
partial pressure of oxygen, and the decomposition temperature
increased with increasing pressure [44,45].
◦
PdPt/Al O at 800 C
2
3
2+
4+
0
a
Previous investigation, see paper by Persson et al. [29].
The values of BET surface areas, crystal sizes, and noble
metal loadings of the samples are compiled in Table 1. The
drop of surface area with aging may decrease the activity of
both catalysts to some extent. But because the drop is in the
same range for the two catalysts, the differences in activity loss
are most likely not associated with the different surface areas.
TEM studies of the aged materials (not shown in any figure)
revealed that the particle size of the alumina was larger than
that in the as-prepared samples. According to the literature,
only on the surface of the catalyst, because no Pd4+ was de-
tected in the bulk phase by PXRD. To investigate whether this
in fact is the case, depth profiling by argon etching was carried
out. As expected, after the etching of the material, Pd4+ disap-
peared, and the oxidation states of the bulk materials consisted
of Pd2+ and Pd0. Because the amount of Pd0 is fairly high in
this sample, it is possible that some of the PdO phase also was
reduced during etching.

412
K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
◦
Fig. 15. TEM image of Pd/Al O sample, aged at 1000 C, 15 vol% steam for
2
3
10 h.
Fig. 13. Steady-state activity tests of Pd/Al O ; methane conversion as a func-
tion of time on stream. The black line represents the activity of the as-prepared
sample and the gray dotted line represents the activity of the aged sample.
2
3
with aging may not be attributed to increasing crystal size, but
rather may be an effect of Pd being encapsulated into a surface
coating layer of PdO over the metal, preventing good contact
with both Pd and PdO surfaces. Furthermore, the increase in
crystal size was very small with aging and may be within the
experimental error.
The catalyst in the present study was calcined at consider-
ably higher temperature than that used in most aging studies,
resulting in larger initial sizes of the palladium-containing par-
ticles. Due to the larger palladium particles, and thus the larger
distance between them, the ability to move in the alumina ma-
trix to obtain growing and sintering was reduced. Nevertheless,
the smaller crystals might migrate through the alumina mater-
ial.
Gélin et al. [1] have shown that alumina-supported Pt cata-
lysts are less resistant to thermal aging in steam at 600 ◦C than
Pd catalysts. The dispersion of Pt on the alumina decreased sig-
nificantly, resulting in lower activity. In contrast to monometal-
lic Pt catalysts, the alloyed Pd–Pt system showed significant
stabilisation against sintering [47]. This appears to also be the
case for the catalysts presented in this paper, for which only
small changes in the alloy crystal size were observed.
Vaporisation of the noble metals also may decrease the activ-
ity. According to the literature, Pt is more easily vaporised than
Pd, but the combination of the two make the platinum more re-
sistant against vaporisation [48]. Table 1 shows that the noble
metal loading was slightly decreased for both catalysts after ag-
ing. For the Pd/Al2O3 catalyst, the Pd loading was decreased
from 5.1 to 4.8 wt%. In PdPt/Al2O3, the Pd content remained
constant, and only the platinum content decreased. Because pal-
ladium in the form of PdO appears to be the active phase in the
bimetallic catalysts, the loss in Pt probably will not decrease
the activity severely. It is interesting to note that the alloy ap-
peared to stabilise the Pd phase, probably related to the solid
solution formation at higher temperatures having the nominal
composition value.
Fig. 14. Steady-state activity tests of PdPt/Al O . Methane conversion
2
3
as a function of time on stream. The black line represents the activity of
the as-prepared sample and the gray dotted line represents the activity of the
aged sample.
the transformation into θ-Al2O3 occurred at 1050 ◦C [46] and
thus is likely to occur during natural aging. This transforma-
tion will slightly decrease the surface area and thereby decrease
the activity. The PXRD diffraction patterns confirm the higher
content of θ-Al2O3 in the aged samples compared with the as-
prepared samples. Nonetheless, high amounts of the δ-Al2O3
phase were present in the aged sample; α-Al2O3 was not de-
tected even in the aged samples.
As shown in Table 1, the increase in the average crystal size
of PdO determined from the PXRD data was similar for the two
catalysts as well. Thus, the greater activity loss of Pd/Al2O3
A representative TEM image of the aged Pd/Al2O3 mate-
rials is presented in Fig. 15. The particles have more facetted

K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
413
structures, and almost no metallic Pd is found in the sample.
Similar morphology was found in the preoxidised sample, as
discussed previously. Thus, the reconstruction into a less ac-
tive morphology may be one explanation for the poor stability
against thermal aging.
According to TEM analyses of the aged PdPt/Al2O3 sam-
ple, the alloy part is unaffected by aging, although the PdO
part is larger. That the oxidation proceeded during aging is also
indicated by the x-values determined from PXRD; for the as-
prepared sample, x = 0.43, but this increased to x = 0.46 after
aging. However, the increase in the x-value also may be due to
the lower amount of Pt in the sample after aging.
In summary, thermal aging achieved a structure similar to
that after preoxidation at 500 ◦C; that is, the metallic Pd in
Pd/Al2O3 oxidised into PdO at the same time as the crystals
grew, whereas PdPt/Al2O3 maintained the two-domain struc-
ture in which the PdO part became larger and the alloy obtained
a structure closer to x = 0.5. Thus, Pd/Al2O3 acquired a less ac-
tive structure over time, whereas PdPt/Al2O3 achieved a more
active structure. However, thermal aging also causes other phe-
nomena that may result in a drop in activity for both catalysts,
such as decreased surface area and vaporisation of noble met-
als. This feature will probably be less pronounced at the normal
operation temperature of the catalysts, because they were not
observed for the preoxidised sample.
eration. By increasing the oxidation temperature, more PdO
was formed and stabilisation occurred more rapidly. How-
ever, the oxidation temperature must be below the PdO
decomposition temperature for this to occur.
• The morphology of the noble metal particles in PdPt/Al2O3
consisted of two parts, one rich in PdO and the other con-
sisting of an alloy between Pd and Pt. These two parts
were always found in close contact. The alloy initially had
a palladium-rich composition. During operation below the
PdO decomposition temperature, the composition of the al-
loy changed toward a structure of equal molarity of Pd and
Pt. The metallic palladium ejected from the solid solution
was oxidised into PdO, which is more active for methane
combustion, resulting in increased activity.
• On exposing PdPt/Al2O3 to higher temperature than al-
lowed for PdO stability, the PdO decomposes and becomes
incorporated into the solid solution between Pd and Pt.
Hence, on cooling, the extra Pd in the solid solution must
diffuse out from the solid solution to be oxidised back to
PdO. This is probably the reason for the slow reoxidation
process observed for the bimetallic catalyst.
• For the oxidation reaction of methane, stable conversion
was observed only for the PdPt/Al2O3 catalyst, where both
metallic Pd and PdO coexist on the surface, whereas for the
Pd/Al2O3 catalyst, all Pd transformed into PdO over time.
• The thermal aging was worse for Pd/Al2O3 than for
PdPt/Al2O3. Thus, Pt prevents the aging of Pd cata-
lysts. The difference in aging may be attributed both to
changes in the morphology and to the loss of palladium of
the monometallic catalyst, which is not observed for the
bimetallic catalyst.
4. Conclusions
Our study of the stability of methane conversion over
Pd/Al2O3 and PdPt/Al2O3, as well as the corresponding mor-
phologies, has led to the following conclusions:
• The activity over PdPt/Al2O3 was greater than that over
Pd/Al2O3 after 6.6 h for the conditions used in this study.
• The activity over Pd/Al2O3 dropped significantly with
time, and the loss was larger at higher temperatures. Be-
cause slightly more PdO was formed during operation in
oxidising atmosphere, this activity drop most likely cannot
be attributed to loss of Pd.
Acknowledgments
This work was supported by the Swedish Energy Agency.
The authors thank Sasol Germany GmbH for providing the
alumina support. They also thank Professor Y. Teraoka and
Dr. H. Kusaba at Kyushu University for helping with the XPS
measurements.
• The as-prepared Pd/Al2O3 catalyst was incompletely oxi-
dised initially, but formed more PdO during operation. The
palladium-containing particles were composed of poly-
crystalline domains of both PdO and Pd randomly ori-
ented inside the particles. Preoxidising the sample at 500 or
600 ◦C changed the morphology of the particles into larger
crystals and caused the sample to become fully oxidised.
The decline in activity may be due to this reconstruction.
By treating the catalyst at a temperature above that of ther-
mal PdO decomposition, the structure of the as-prepared
sample likely can be recovered.
• The thermal decomposition of PdO was initiated by cre-
ating one or two nuclei of Pd on each PdO crystal. The
completely reduced sample consisted of large facetted crys-
tals.
• The activity of PdPt/Al2O3 increased with time. Results
from TPO, XRD, XPS, and TEM indicate that the increase
was due to significantly greater PdO production during op-
References
[1] P. Gélin, L. Urfels, M. Primet, E. Tena, Catal. Today 83 (2003) 45.
[2] D.L. Mowery, M.S. Graboski, T.R. Ohno, R.L. McCormick, Appl. Catal.
B 21 (1999) 157.
[3] J.K. Lampert, M.S. Kazi, R.J. Farrauto, Appl. Catal. B 14 (1997) 211.
[4] W.C. Pfefferle, J. Energy 2 (1978) 142.
[5] M.F.M. Zwinkels, S.G. Järås, P.G. Menon, T.A. Griffin, Catal. Rev. Sci.
Eng. 35 (1993) 319.
[6] E.M. Johansson, D. Papadias, P.O. Thevenin, A.G. Ersson, R. Gabriels-
son, P.G. Menon, P.H. Björnbom, S.G. Järås, in: J.J. Spivey (Ed.), in:
Catalysis—Specialists Periodical Reports, vol. 14, Royal Soc. Chem.,
Cambridge, 1999, p. 183.
[7] D. Ciuparu, M.R. Lyubovsky, E. Altman, L.D. Pfefferle, A. Datye, Catal.
Rev. 44 (2002) 593.
[8] K. Narui, H. Yata, K. Furuta, A. Nishida, Y. Kohtoku, T. Matsuzaki, Appl.
Catal. A 179 (1999) 165.
[9] A. Ersson, H. Kušar, R. Carroni, T. Griffin, S. Järås, Catal. Today 83
(2003) 265.

414
K. Persson et al. / Journal of Catalysis 245 (2007) 401–414
[10] K. Persson, A. Ersson, A. Manrique Carrera, J. Jayasuriya, R. Fakhrai,
T. Fransson, S. Järås, Catal. Today 100 (2005) 479.
[11] K. Persson, A. Ersson, K. Jansson, N. Iverlund, S. Järås, J. Catal. 231
(2005) 139.
[12] Y. Ozawa, Y. Tochihara, M. Nagai, S. Omi, Chem. Eng. Sci. 58 (2003)
671.
[13] D. Roth, P. Gélin, M. Primet, E. Tena, Appl. Catal. A 203 (2000) 37.
[14] Y. Ozawa, Y. Tochihara, A. Watanabe, M. Nagai, S. Omi, Appl. Catal.
A 259 (2004) 1.
[27] J.G. McCarty, Catal. Today 26 (1995) 283.
[28] R. Strobel, J. Grunwaldt, A. Camenzind, S.E. Pratsinis, A. Baiker, Catal.
Lett. 104 (2005) 9.
[29] K. Persson, A. Ersson, K. Jansson, J.L.G. Fierro, S.G. Järås, J. Catal. 234
(2006) 14.
[30] M. Lyubovsky, L. Pfefferle, A. Datye, J. Bravo, T. Nelson, J. Catal. 187
(1999) 275.
[31] Y.S. Ho, C.B. Wang, C.T. Yeh, J. Mol. Catal. A Chem. 112 (1996)
287.
[15] K. Persson, A. Ersson, S. Colussi, A. Trovarelli, S.G. Järås, Appl. Catal.
B 66 (2006) 175.
[16] H. Yamamoto, H. Uchida, Catal. Today 45 (1998) 147.
[17] K. Nomura, K. Noro, Y. Nakamura, Y. Yazawa, H. Yoshida, A. Satsuma,
T. Hattori, Catal. Lett. 53 (1998) 167.
[18] C.L. Pieck, C.R. Vera, E.M. Peirotti, J.C. Yori, Appl. Catal. A 226 (2002)
281.
[19] K.E. Johansson, T. Palm, P.E. Werner, J. Phys. Sci. Instrum. 13 (1980)
1289.
[20] R.B. Anderson, K.C. Stein, J.J. Feenan, L.J.E. Hofer, Ind. Eng. Chem. 53
(1961) 809.
[21] P. Araya, S. Guerrero, J. Robertson, F.J. Garcia, Appl. Catal. A 283 (2005)
225.
[22] C.F. Cullis, T.G. Nevell, D.L. Trimm, J. Chem. Soc. Faraday Trans. 68
(1972) 1406.
[23] R. Burch, P.K. Loader, F.J. Urbano, Catal. Today 27 (1996) 243.
[24] G. Groppi, C. Cristiani, L. Lietti, P. Forzatti, in: A. Corma, F.V. Melo,
S. Mendioroz, J.L.G. Fierro (Eds.), in: Stud. Surf. Sci. Catal., vol. 130,
Elsevier Science, Amsterdam, 2000, p. 3801.
[32] M. Lyubovsky, L. Pfefferle, Appl. Catal. A 173 (1998) 107.
[33] J.N. Carstens, S.C. Su, A.T. Bell, J. Catal. 176 (1998) 136.
[34] A.K. Datye, J. Bravo, T.R. Nelson, P. Atanasova, M. Lyubovsky, L. Pfef-
ferle, Appl. Catal. A 198 (2000) 179.
[35] E.H. Voogt, A.J.M. Mens, O.L.J. Gijzeman, J.W. Geus, Surf. Sci. 350
(1996) 21.
[36] J.J. Chen, E. Ruckenstein, J. Phys. Chem. 85 (1981) 1606.
[37] D.H. Kim, S.I. Woo, J.M. Lee, O. Yang, Catal. Lett. 70 (2000) 35.
[38] K. Otto, L.P. Haack, J.E. de Vries, Appl. Catal. B 1 (1992) 1.
[39] Y. Bi, G. Lu, Appl. Catal. B 41 (2003) 279.
[40] A.L. Guimarães, L.C. Dieguez, M. Schmal, J. Phys. Chem. B 107 (2003)
4311.
[41] M. Moroseac, T. Skála, K. Veltruská, V. Matolín, I. Matolínová, Surf.
Sci. 566–568 (2004) 1118.
[42] R.J. Bird, P. Swift, J. Electron Spectrosc. Relat. Phenom. 21 (1980) 227.
[43] G. Johansson, J. Hedman, A. Berndtsson, M. Klasson, R. Nilsson, J. Elec-
tron Spectrosc. Relat. Phenom. 2 (1973) 295.
[44] W.E. Bell, R.E. Inyard, M. Tagami, J. Phys. Chem. 70 (1966) 3735.
[45] H. Zhang, J. Gromek, G.W. Fernando, S. Boorse, H.L. Marcus, J. Phase
Equilib. 23 (2002) 246.
[46] C. Morterra, G. Magnacca, Catal. Today 27 (1996) 497.
[47] A. Morlang, U. Neuhausen, K.V. Klementiev, F.W. Schütze, G. Miehe,
H. Fuess, E.S. Lox, Appl. Catal. B 60 (2005) 191.
[48] M. Chen, L.D. Schmidt, J. Catal. 56 (1979) 198.
[25] G. Groppi, G. Artioli, C. Cristiani, L. Lietti, P. Forzatti, in: E. Inglesia, J.J.
Spivey, T.H. Fleish (Eds.), in: Stud. Surf. Sci. Catal., vol. 136, Elsevier
Science, Amsterdam, 2001, p. 345.
[26] R.J. Farrauto, M.C. Hobson, T. Kennelly, E.M. Waterman, Appl. Catal.
A 81 (1992) 227.