Applied Catalysis A: General 495 (2015) 54–62

Contents lists available at ScienceDirect

Applied Catalysis A: General

journal homepage: www.elsevier.com/locate/apcata

Pd catalysts for total oxidation of methane: Support effects

James B. Miller ∗ , Manaswita Malatpure
Department of Chemical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213, United States

a r t i c l e i n f o a b s t r a c t

Article history: We prepared a series of Pd catalysts for total oxidation of methane on different supports: Al2 O3 ,
Received 18 December 2014 ZrO2 –CeO2 and CeO2 . We characterized the catalysts’ structures by XRD and their surface chemistries
Received in revised form 27 January 2015 by temperature programmed reduction (TPR) and oxidation (TPO); and measured their activities for
Accepted 31 January 2015
methane oxidation in temperature programmed reaction experiments (TPRx). By fitting a kinetic model
Available online 7 February 2015
to the TPRx results, we extracted apparent activation barriers (Eapp ) and pre-exponents (Aapp ) for methane
oxidation over each of the catalysts. Activity correlates strongly with Aapp , which we interpret as a relative
Keywords:
measure of the density of PdO–Pd* (* is an O-vacancy) site pairs. TPR results support this view: PdO in low
Supported Pd catalysts
Total methane oxidation
activity (low Aapp ) catalysts, such as Pd/CeO2 , is relatively resistant to reduction, meaning that formation
Temperature programmed reaction of the Pd* partner of the site pair is difficult—and the number of site pairs that can form is low.
Temperature programmed reduction © 2015 Elsevier B.V. All rights reserved.

1. Introduction apparent reaction pre-exponent, which we interpret as a measure

Supported palladium catalysts for total hydrocarbon oxida-
tion have important applications in power generation, where they
enable methane combustion at relatively low temperature for
avoiding formation of thermal NOx ; in emissions control, includ-
ing automotive catalysts; and as the active elements of sensors
that detect combustible gasses. A number of reviews have been
published on supported Pd catalysts for methane oxidation [1–5].
At a mechanistic level, the catalytic chemistry of methane oxida-
tion on Pd catalysts is complex. Oxidation is usually described as
a Mars–van Krevelen cycle [6], in which a pair of sites—PdO and
Pd* (where * is an ‘O-vacancy’)—act in concert to activate the sta-
ble C H bond. The activity of the catalyst depends on a number
of factors, but the density and activity of site pairs are among the
most important [6,7]. These characteristics are functions of the
size, morphology and surface chemistry of PdO crystallites that
reside on the support surface, which in turn depend on the details
of catalyst preparation [6,8–11], operating conditions [10] and,
significantly, the interactions of PdO with its support [7,12–15].
Different supports can stabilize PdO crystallites in different sizes
and morphologies [16]. The support can even be an active partic-
ipant in the catalytic chemistry through exchange of its O atoms
with PdO [17,18]. In this work, we compare Pd catalysts pre-
pared on three different supports, Al2 O3 , ZrO2 –CeO2 , and CeO2 .
Microkinetic analyses of the temperature-programmed methane
oxidation data show that activity correlates strongly with Aapp , the
of the relative density of site pairs on the catalyst surface. Results
of temperature-programmed reduction experiments support this
view.
2. Methods
2.1. Catalyst preparation
Table 1 lists the catalysts studied in this work. They were pre-
pared by impregnating Al2 O3 (Strem 13-2525, ‘Al’), ZrO2 –CeO2
(Alfa Aesar 39216, ‘ZrCe’) or CeO2 (Alfa Aesar 12925, ‘Ce’) oxides
with solutions of Pd(NO3 )2 (10 wt% in 10% HNO3 , Aldrich 380040,
‘NO3 ’) or Pd(NH3 )4 (NO3 )2 (tetraamine palladium nitrate, 10 wt%,
Aldrich 377384, ‘TAPN’) precursors. For each preparation, approx-
imately 1 g of oxide was first dried at 400 K for 30 min in a vacuum
oven. Then, precursor solution was stirred into the dried powder in
20 ␮L increments until the powder started to become sticky. The
precursor was decomposed by calcination in a box furnace at 773 K
for 1 h; the heating rate of the furnace is approximately 40 K/min.
Table 1 summarizes the Pd contents of the catalysts calculated
from the amount of precursor solution added to their supports. The
1.0 Pd(NO3 )/Ce sample was prepared by two impregnations, with
an intermediate calcination at 773 K. We also evaluated a com-
mercially available 1% Pd/Al2 O3 (Alfa Aesar, ‘1.0 Pd/Al(AA)’) as a
reference material.
2.2. X-ray diffraction

∗ Corresponding author. Tel.: +1 920 397 5855; fax: +1 412 268 7139. X-ray diffraction patterns were acquired for the supports and
E-mail address: jbmiller@andrew.cmu.edu (J.B. Miller). the calcined catalysts using a Pananalytical X’pert Pro X-ray

http://dx.doi.org/10.1016/j.apcata.2015.01.044
0926-860X/© 2015 Elsevier B.V. All rights reserved.
 J.B. Miller, M. Malatpure / Applied Catalysis A: General 495 (2015) 54–62
Table 1
Supported Pd catalysts studied in this work. Except for 1.0 Pd/Al(AA), Pd contents are
calculated from preparation. Pd dispersion and crystallite size are estimated from
the results of pulse CO chemisorption experiments.
Calculated Dispersion
Pd (wt%)
1.0 Pd/Al(AA) 1.02* 0.23
1.4 Pd(NO3 )/Al 1.35 0.21
1.4 Pd(TAPN)/Al 1.44 0.32
1.1 Pd(NO3 )/ZrCe 1.13 0.20
0.5 Pd(TAPN)/ZrCe 0.54 0.41
0.5 Pd(NO3 )/Ce 0.52 0.10
1.0 Pd(NO3 )/Ce 0.95 – –
*
From manufacturer’s certification.
diffractometer in theta–2theta (–2) configuration. X-rays with
1.54 Å wavelength were generated from a Cu source operated at
35 kV and 10 mA. Diffraction patterns were recorded over a 2 range
from 10◦ to 90◦ with 0.0525◦ step size.
2.3. CO pulse chemisorption
Pd dispersion was characterized by pulse CO chemisorption
using a Micromeretics 2920 instrument. The instrument is config-
ured as a flow-through, fixed-bed quartz U-tube reactor enclosed
in a high-temperature furnace. Samples are held in place within the
reactor by a small quartz wool plug. For the chemisorption exper-
iment, 0.10–0.15 g of catalyst was first pre-treated in 50 ccm 10%
H2 /Ar (Airgas) for 2 h at 523 K and then cooled to 323 K in 50 ccm
He (Airgas). A series CO pulses (9.991% CO/He, Airgas) were injected
into the He flow; a thermal conductivity detector (TCD) monitored
relative CO content of the reactor effluent during each pulse. Injec-
tions were terminated when the size of the detected pulses no
longer changed. Dispersion (fraction of Pd atoms at the surfaces
of crystallites) and crystallite size were estimated from CO uptake
assuming adsorption stoichiometry of 1:1 CO:Pd.
2.4. Temperature-programmed oxidation, reduction and reaction
Supports and catalysts were characterized by temperature-
programmed reaction (TPRx), oxidation (TPO) and reduction (TPR)
using a Micromeretics 2950HP instrument. This instrument is also
configured as a flow-through, fixed-bed quartz U-tube reactor
enclosed in a high-temperature furnace. Samples are held in place
within the reactor by a small quartz wool plug. A thermocouple is
located at the top (inlet) of the catalyst bed. Mass flow controllers
regulate the flow of reaction and treatment gases. An integrated
TCD reports a signal proportional to the difference between the
thermal conductivities of the reactor inlet and outlet streams.
Before entering the TCD, TPRx product gas flows through a cold
trap to remove water produced during the oxidation reaction.
For the TPRx experiment, a ∼0.1–0.2 g sample was first pre-
treated in 50 ccm air (zero grade, Valley National) at 873 K for
15 min. After cooling to 323 K, the flow was changed to 50 ccm
2% CH4 in air (Matheson). The catalyst temperature was increased
at 10 K/min to 873 K, where it was held for 15 min; we refer to
this as the 1up portion of the experiment. To complete the first
reaction cycle, the catalyst was cooled at 10 K/min to 323 K in the
CH4 /air mixture (1dn). A second cycle was started immediately
without changing flow from the CH4 /air mixture (2up and 2dn).
For supported Pd samples, a third cycle, preceded by a new air
pre-treatment, was also performed (3up and 3dn).
For TPO characterization, a ∼0.1 g sample was first cooled to
278 K flowing Ar. At 278 K, the flow was changed to 50 ccm of
10.0% O2 /He (Valley National). The sample was heated to 873 K
55
at 10 K/min and then held at 873 K for 15 min before cooling at
20 K/min to 300 K in O2 /He flow. The reactor was purged with
He and the sample was cooled to 233 K before changing the flow
to 50ccm 10.8%H2 /Ar (Valley National) for TPR. The sample was
Crystallite
heated to 873 K at 10 K/min and then held at 873 for 15 min. Next,
size (nm)
the sample was cooled in 10.8% H2 /Ar at 20 K/min to 423 K and
4.8 10 K/min to 288 K. Flow was changed to Ar and a second TPO was
5.3
conducted. We report results of the TPR and the second TPO only;
3.6
5.7 the first TPO served as a pre-treatment to deliver a baseline, oxi-
2.7 dized state for the TPR experiment.
11.1
3. Results
3.1. X-ray diffraction
Figures A1, A2 and A3 display diffraction patterns for Al2 O3 -
supported catalysts, ZrO2 –CeO2 supported catalysts and CeO2
supported catalysts. No diffraction features that could be indexed
to either Pd or PdO were identified in any of the Pd-catalysts. This
result reflects small primary crystallite size, or perhaps simply rela-
tively low Pd loadings [19]. Furthermore, Pd addition did not change
the structure of any of the of the parent oxides. Al2 O3 and Pd/Al cat-
alysts all displayed diffraction patterns characteristic of ␥-Al2 O3 .
ZrO2 –CeO2 and Pd/ZrCe catalysts’ diffraction patterns reflect a mix-
ture of monoclinic and tetragonal ZrO2 . No features that could be
indexed to CeO2 were identified in the patterns of ZrO2 –CeO2 or the
Pd/ZrCe catalysts. CeO2 and Pd/Ce catalysts all displayed patterns
characteristic of a highly crystalline CeO2 .
3.2. CO pulse chemisorption
Pd dispersions and crystallite sizes of the catalysts are reported
in Table 1. Dispersions ranged from 0.10 and 0.41, correspond-
ing to crystallite sizes between 2.7 and 11 nm. Dispersions in
this range are typical of supported Pd catalysts made by solu-
tion impregnation of Pd salts [9,19–22]. For both the Al2 O3 and
ZrO2 –CeO2 supports, catalysts prepared from the TAPN precur-
sor display higher dispersion then those prepared from the NO3
precursor. Others have reported dependencies of dispersion on
precursor choice in incipient wetness preparation of supported
Pd catalysts [9,22]; while beyond the scope of this work, we note
that the differences likely reflect differences in the decomposition
kinetics of the precursors. We measured the dispersion of only
one of our two ceria-supported catalysts; it displayed the lowest
dispersion within the sample set. Others have observed a depend-
ence of Pd dispersion on support composition [12,20]. Even within
cerias, Pd dispersion has been observed to depend on the details
of support morphology and chemistry. For example, Satsuma et al.
report that high surface area cerias interact strongly with Pd to
keep Pd crystallites small [15]. On the other hand, the interaction
between Pd and highly crystalline (like ours), low surface area,
deeply reduced cerias is weaker, allowing larger Pd structures to
form.
3.3. TPRx
Figure 1 shows the TPRx results for 1.0 Pd/Al(AA) as a plot of TCD
signal vs. reaction temperature. The 0.0 to ∼0.05 TCD signal scale
corresponds to fractional CH4 conversions, x, of 0.0–1.0. Results for
the first reaction cycle (1up and 1dn) are plotted in blue, with a
solid line for the increasing temperature portion of the cycle (1up)
and a dotted line for the decreasing temperature portion (1dn).
The second cycle is plotted in red (2up and 2dn) and the third
cycle in green (3up and 3dn). For this catalyst, complete conver-
sion is achieved across all three cycles. The catalyst exhibits highest
activity—conversion at any reactor temperature is highest—during
56 J.B. Miller, M. Malatpure / Applied Catalysis A: General 495 (2015) 54–62

Fig. 1. TPRx of 1.0 Pd/Al(AA), 0.1002 g sample size, displayed as TCD signal (relative
methane conversion) vs. reactor temperature. Blue lines are for the first reac-
tion cycle (1), red for the second cycle (2) and green for the third (3). Increasing
temperature portions of each cycle (up) are plotted as solid lines; decreasing tem- Fig. 2. TPR of 1.0 Pd/Al(AA). Solid blue line is the signal from the increasing tem-
perature portions (dn) as dotted lines. The catalyst is most active during 1up. A perature portion of the experiment. Features are assigned to PdO reduction (273 K),
‘low-temperature feature’ is observed during 1up and 3up. (Color version available Pd H decomposition (347 K) and interactions of H2 with the apparatus (640 K). Dot-
online.) ted red line is the signal from the decreasing temperature portion of the experiment.
The feature at 335 K corresponds to re-formation of Pd H. (Color version available
online.)
the 1up portion of the experiment. Conversion is significantly lower
in 1dn than in 1up, but activity changes from 1dn through the
end of the experiment are small. The need for supported Pd to
be “conditioned” or “stabilized” by cyclic exposure to different
reaction environments has been widely reported [1,6,10,19,21,23];
during conditioning, PdO crystallites achieve their steady state
morphology and surface chemistry. The 1up and 3up portions of
the experiment—those that were immediately preceded by an air
treatment—display a low-temperature feature at around 480 K. We
will consider the significance of this feature in the Discussion sec-
tion; here, we simply note that it likely reflects a transition in PdO
surface chemistry.
Results for 1.4 Pd(NO3 )/Al appear in Fig. A4. The TPRx response
of this sample is similar to that of 1.0 Pd/Al(AA): complete conver-
sion in the temperature range of the experiment, highest activity
in the 1up portion of the experiment, and appearance of the low-
temperature feature in 1up and 3up. The TPRx response of 1.4
Pd(TAPN)/Al, shown in Fig. A5, is similar to those of the other Pd/Al
catalysts, with the exception that it does not display a ‘condition-
ing period’—CH4 conversions are similar across all three cycles. The
state of the as-prepared catalyst must not be very different from the
steady state at reaction conditions. Complete consideration of sta-
bilization differences between 1.4 Pd(NO3 )/Al and 1.4 Pd(TAPN)/Al
is beyond the scope of this report, but we note that they are likely Fig. 3. TPR (increasing temperature portion of the experiment) of all catalysts stud-
linked to differences in the decomposition/thermal evolution of the ied in this work. Temperatures of features that correspond to the PdO→Pd transition
TAPN and Pd(NO3 )2 precursors. are indicated. Dotted lines are results for the bare oxides.
Figures A6 and A7 display TPRx results for 0.5 Pd(TAPN)/ZrCe
and 1.1 Pd(NO3 )/ZrCe. Like the Pd/Al catalysts, both Pd/ZrCe cat- 3.4. TPO/R
alysts achieve full conversion well before 873 K. Unlike Pd/Al, the
activity of Pd/ZrCe increases with the number of reaction cycles. Figure 2 displays the TPR of 1.0 Pd/Al(AA). Three primary fea-
Neither Pd/ZrCe sample exhibits the low-temperature feature that tures appear. The large positive feature at 273 K is evidence of
was characteristic of the Pd/Al catalysts. hydrogen consumption; it has been linked to reduction of sur-
TPRx results for 0.5 Pd(NO3 )/Ce and 1.0 Pd(NO3 )/Ce appear face PdO [6,12,16]. The negative feature at 347 K is attributable
in Figs. A8 and A9. The Ce-supported catalysts are clearly less to hydrogen evolution upon decomposition of Pd hydride [16].
active than the others—neither delivers complete CH4 conversion The broad feature near 640 K appears in blank runs and most
in the temperature range of the experiment. Both Pd/Ce samples likely results from interaction of H2 with the thermocouple’s metal
exhibit activity loss as the number of reaction cycles increases. Like sheath. During the decreasing temperature portion of the experi-
Pd/ZrCe, the low-temperature feature is not present in TPRx over ment, a hydrogen consumption feature, related to re-formation of
Pd/Ce. the hydride, appears at 335 K.
We performed 2 cycle TPRx experiments on the bare oxides Others have successfully used the low-temperature reduction
(not shown in this report). All three exhibited measurable activ- feature to interpret oxidation activity of supported Pd catalysts
ity at high temperatures. At 873 K, CH4 conversion is ∼30% over [6,12,16]; therefore, in Fig. 3 we compare TPR results for all catalysts
both Al2 O3 and ZrO2 –CeO2 ; and ∼12% over CeO2 . during the increasing temperature portion of the experiment at
 J.B. Miller, M. Malatpure / Applied Catalysis A: General 495 (2015) 54–62 57

4. Discussion

4.1. Mirokinetic model of methane oxidation

To formally compare the activities of the catalysts, we fit a

kinetic model to the reaction data. We use the model of Fujimoto
et al. [6], which describes methane oxidation as a sequence of the
following elementary steps:
K1
O2 + ∗ ←→ O2 ∗
k2
O2 + ∗ −→ 2O∗
K3
CH4 + ∗ ←→ CH4 ∗
k4
CH4 ∗ + O∗ −→ CH3 ∗ +HO∗

Fig. 4. TPO of 1 Pd/Al(AA). Solid green line is the signal from the increasing temper-
ature portion of the experiment. The feature at 559 K is assigned to consumption of
O2 during oxidation of surface Pd. Dotted red line is the signal from the decreasing
temperature portion of the experiment. (Color version available online.)
temperatures below 500 K. The location of the maximum tempera-
ture of the PdO reduction feature depends on the catalysts’ support;
reduction maxima increase in the order Pd/Al < Pd/ZrCe < Pd/Ce.
Yue, Zhou et al. [12,16] reported a reduction maximum for Pd/Al2 O3
that was ∼24 K below that of Pd/ZrO2 –CeO2 /Al2 O3 ; differences
between our Pd/Al and Pd/ZrCe catalysts are similar.
Figure 4 is the TPO of 1.0 Pd/Al(AA). The broad negative feature
at around 650 K is characteristic of the support itself. Of greater rel-
evance to activity is the feature at 559 K, which has been attributed
to oxidation of Pd to PdO [16,24]. No notable features appear in the
decreasing temperature portion of the experiment. Figure 5 com-
pares the increasing temperature portions of the TPO experiments
for all catalysts. Trends are less clear than in the TPR experiments.
1.0 Pd/Al(AA) and the two Pd/ZrCe catalysts display significant oxi-
dation features between 500 and 650 K; features below ∼500 K in
the Pd/ZrCe catalysts are related to their support. Prepared Pd/Al
and Pd/Ce catalysts have weak or no oxidation features in the
500–650 K interval.
→
→
→ CO2 ∗, CO3 ∗ . . .
K5
2HO∗ ←→ H2 O + O ∗ +∗
K6
CO2 ∗ ←→ CO2 + ∗
K7
CO3 ∗ ←→ CO2 + O∗
In this sequence, O* represents surface PdO and * represents Pd*,
a surface Pd atom or ‘O-vacancy’. Methane adsorbs at a vacancy
(step 3) and is oxidized in rate-limiting steps 4 as H atoms are
removed from the C atom by successive reaction with O*. Adsorbed
species produced during the oxidation steps (HO*, CO2 * and CO3 *)
are in equilibrium with their gas phase counterparts (steps 5, 6 and
7). Note that the sequence includes cyclic consumption (step 4)
and regeneration (steps 1–2, 5 and 7) of O*—a Mars–van Krevelen
cycle. The requirement that both types of sites, Pd* for adsorption
of methane and PdO for oxidation of adsorbed methane, be present
for the oxidation reaction to proceed is central to understanding
the activity of the catalyst.
Applying the steady state approximation in the case where HO*
is the most abundant surface species to the reaction sequence, Fuji-
moto derived a rate law for methane oxidation [6]:

k4 K3 K5 [CH4 ]1.0 [O2 ]0.0 kapp [CH4 ]

r= =
[H2 O]1.0 [H2 O]

This is an interesting result; it underscores the well-known inhib-

iting effect of product water on the reaction kinetics [21].
To fit the rate law to our data, it must be integrated for a plug
flow reactor. Au-Yeung et al. report the result of the integration as
[25]:

2FCH4
 1

kapp = ln −x
ns 1−x

where FCH4 is the methane molar feed rate, x is methane conver-

sion and nS is the number of ‘accessible Pd sites’ in the reactor. We
calculate nS from the catalyst’s Pd content and its Pd dispersion
(Table 1). It is important to recognize that nS contains no informa-
tion on the fraction of surface Pd atoms that participate in active
site pairs, PdO–Pd*; we will return to this point shortly.
We assume that kapp has Arrhenius form:

kapp = Aapp exp(−Eapp /RT )

Fig. 5. TPO (increasing temperature portion of the experiment) of all catalysts stud-
ied in this work. Temperatures of features that correspond to the Pd→PdO transition
are indicated.
where Aapp and Eapp are the apparent pre-exponent and activa-
tion barrier associated with the rate constant. Our strategy for
interpreting the reaction data is to calculate kapp as a function
58 J.B. Miller, M. Malatpure / Applied Catalysis A: General 495 (2015) 54–62
Fig. 6. Arrhenius analysis of methane oxidation TPRx results for 1.0 Pd/Al(AA). Blue
lines are for the first reaction cycle (1) and red for the second cycle (2). Increasing
temperature portions of each cycle (up) are plotted as solid lines; decreasing tem-
perature portions (dn) as dotted lines. The catalyst is most active during 1up. The
‘low-temperature feature’ observed in the raw reaction data (Fig. 1) appears as a
discontinuity in the plot of 1up. (Color version available online.)
of temperature and extract values of Eapp and Aapp for each cat-
alyst from (Arrhenius) plots of lnAapp vs. 1/T. Differences in Eapp
reflect changes in the temperature dependence of rate constants
(step 4) and equilibrium constants (steps 3 and 5)—a measure of
the inherent activity of the catalyst. With kapp normalized to the
total number of surface Pd sites in the reactor, differences in Aapp
can reflect differences in fraction of surface Pd atoms that partici-
pate in PdO–Pd* pairs: larger Aapp means that a higher fraction of
surface Pd atoms participate in PdO–Pd* pairs. Conversely, lower
fractional participation of surface Pd atoms in site pairs means that
nS over counts the number of active sites, thereby suppressing Aapp .
4.2. Fitting the microkinetic model to the TPRx data
Figure 6 presents the results of the Arrhenius analysis of the first
two TPRx cycles of 1.0 Pd/Al(AA). As before, the first cycle is colour-
coded blue (1up and 1dn) and the second cycle red (2up and 2dn);
solid lines represent the increasing temperature portion of the cycle
(1up and 2up) and dotted lines represent the decreasing tempera-
ture portion of the cycle (2dn and 1dn). As expected from the raw
results (Fig. 1), the plot for 1up is located at lower temperatures
(higher 1/T) than the others.
The low temperature feature observed in the raw reaction
data (Fig. 1) is easily identified as a discontinuity in the Arrhe-
nius plot. This discontinuity has been observed by others [10,26];
Carstens et al. [10] provide perhaps the most complete context for
it. They suggest that during rapid heating, a transient, reduced Pd
forms on the surface of PdO crystallites and accelerates the rate of
low-temperature combustion. Then, as temperature increases, the
reduced Pd is reoxidized and activity returns to ‘normal’ levels. They
make the related observation that PdO formed in an oxygen atmo-
sphere (as in the air pre-treatment that precedes 1up and 3up) is a
mixture of amorphous and crystalline forms, while PdO formed in a
reaction atmosphere (as in the other portions of our experiment) is
largely crystalline. Our results link these observations, suggesting
that only the mixed form of PdO, the one that is formed in oxygen,
Table 2
Kinetic parameters, Eapp and lnAapp , and kapp (at 723 K) for 1.0 Pd/Al(AA). Uncer-
tainties reflect the range of kinetic parameters estimated from replicate TPRx of the
same catalyst.
Eapp (kcal/mol) ln(Aapp (s−1 )) kapp @723 K (s−1 )
Cycle 1, up, low 47 ± 1 45 ± 1 210,000 ± 10,000
temperature
Cycle 1, up, high 45 ± 1 39 ± 1 1500 ± 100
temperature
Cycle 1, down 47 ± 1 38 ± 1 220 ± 10
Cycle 2, up 44 ± 1 36 ± 1 150 ± 10
Cycle 2, down 49 ± 1 39 ± 1 160 ± 10
can display this transient. And furthermore, within this sample set,
the mixed form appears to develop only on samples supported by
Al2 O3 . Pd supported on ZrO2 –CeO2 and CeO2 does not display the
low-temperature feature/discontinuity.
Broad linear regions exist in the Arrhenius plots for all four por-
tions of the 1.0 Pd/Al(AA) TPRx experiment (Fig. 6). For all but the
1up portion of the experiment, we fit the data corresponding to
5–50% conversion. Below 5% the data are noisy; above 50% devi-
ations from linearity are observed as model assumptions break
down and supports become active at higher temperatures. The 1up
portion of the experiment has two linear regions, one below and
one above the discontinuity. Table 2 summarizes values of Eapp and
lnAapp extracted from the Arrhenius plots of Fig. 6; for a simple
activity comparison, we calculated kapp (at 723 K) from lnAapp and
Eapp . Reported uncertainties reflect the range that we typically see
in repeat runs starting with a new charge of the same catalyst.
In the 1up portion of the experiment, kapp is significantly higher
at temperatures below the discontinuity. Because Eapp is the same
on both sides of the discontinuity, higher kapp is attributable exclu-
sively to higher Aapp . As noted earlier, higher Aapp suggests a higher
density of active site pairs. Thus, this observation is consistent with
Carstens’ view of transient formation of new reduced Pd sites on
the PdO surface at temperatures below the discontinuity, which, in
turn, participate in new—and transient—PdO–Pd* active site pairs.
From the first through the second cycle of the experiment, kapp
approaches a constant value as the catalyst achieves its steady state
morphology and surface chemistry. Activity is stable in the sec-
ond cycle, kapp (2up) ∼ kapp (2dn), but there are differences between
the lnAapp and Eapp parameters that contribute to kapp : both are
lower during the increasing temperature portion of the experiment.
This observation is likely a manifestation of reversible changes in
PdO–Pd’s surface chemistry/morphology that have been linked to a
widely reported temperature hysteresis in the methane oxidation
activity of supported Pd and Pt catalysts [3,27,28].
We use the 2up portions of the TPRx experiments to com-
pare the catalysts. This is a convenient choice because, having
not been preceded by an air treatment, the low-temperature fea-
ture/discontinuity is not a factor in the analysis of Pd/Al reaction
data. For Pd/ZrCe catalysts, we pick a linear region between 5 and
50% conversion, as we did for Pd/Al. We fit the reaction data from
the less-active Pd/Ce catalysts from ∼5 to 30% conversion. Depar-
tures from linearity occur at lower conversions for Pd/Ce than in the
other samples because higher temperatures are required to achieve
them. Table 3 compares lnAapp , Eapp and kapp (723 K) of all catalysts
estimated from kinetic analysis of the 2up portions of the TPRx
experiments. The rate constants span several orders of magnitude.
The samples fall into three groups; in order of decreasing activ-
ity, they are 1.0 Pd/Al(AA) > ‘prepared’ Pd/Al ∼ Pd/ZrCe > PdCe. This
activity rank-order is consistent with results previously reported
by others [16,22,29].
Our samples display a range of Eapp , from 34 kcal/mol (Pd/Ce) to
44 kcal/mol (1.0 Pd/Al(AA)). Eapp of ∼40–44 kcal/mol (after correc-
tion for inhibition by H2 O) for methane oxidation over Pd/ZrO2 and
 J.B. Miller, M. Malatpure / Applied Catalysis A: General 495 (2015) 54–62 59

Table 3 when supported on ZrO2 –CeO2 /Al2 O3 and CeO2 /Al2 O3 than when
Kinetic parameters, Eapp and lnAapp , and kapp (at 723 K) from the 2up portion of the
supported on Al2 O3 alone, confirming the role of ZrO2 –CeO2 and,
TPRx experiments for all catalysts.
in particular, CeO2 in stabilization of PdO.
Eapp (kcal/mol) ln(Aapp (s−1 )) kapp @723 K (s−1 ) Fujimoto et al. [6] also made a link between PdO stability and
1.0 Pd/Al(AA) 44 36 150 morphology, noting that small crystallites, which are more inti-
1.4 Pd(NO3 )/Al 43 33 19 mate contact with their ZrO2 support, can have more stable surface
1.4 Pd(TAPN)/Al 40 30 5.9 Pd O bonds (TPR maxima at higher temperature)—and display
1.1 Pd(NO3 )/ZrCe 38 28 5.3
lower TOF for methane oxidation. Roth et al. [9] made the same
0.5 Pd(TAPN)/ZrCe 38 28 4.8
0.5 Pd(NO3 )/Ce 35 22 0.092 connection between crystallite size and TOF for Pd/Al2 O3 cata-
1.0 Pd(NO3 )/Ce 34 20–21* 0.022–0.088* lysts. We observe a related trend—kapp (Table 3) decreases with
*
Pd dispersion was not measured for this sample. Ranges of A and k were cal- crystallite size (Table 1)—within both our prepared Pd/Al catalysts
culated using the minimum and maximum dispersions measured for the other and within our Pd/ZrCe catalysts, but not across supports. Pd/Ce
samples. presents a unique case; among our catalysts, it displayed the high-
est PdO stability and lowest activity—but the largest Pd crystallite
size. Satsuma et al. provide partial context for this result [15]. They
Pd/Al2 O3 catalysts have been reported by others [6,10,21,23,25,30];
observe that Pd supported by high surface area cerias can be eas-
our estimates for Pd/Al and Pd/ZrCe are consistent with those
ily and completely oxidized to highly dispersed PdO crystallites
reports. Within our sample set, activity differences are driven pri-
or other relatively inactive forms, such as Pd O Ce. On the other
marily by differences in lnAapp —kapp decreases with decreasing
hand, highly crystalline (like ours), low surface area, deeply reduced
lnAapp despite smaller Eapp . Because Aapp provides a relative mea-
cerias interact less strongly with Pd, allowing for growth of larger,
sure of the number of active site pairs, this pattern suggests that the
(partially) metallic structures that can be more active for hydro-
density of site pairs on the catalyst surface is a key determinant of
carbon oxidation. Understanding the relationships between ceria’s
activity. Furthermore, it indicates that 1.0 Pd/Al(AA), with highest
morphology and surface chemistry, and the activities of Pd/CeO2
Aapp , presents the most site pairs, followed by ‘prepared’ Pd/Al and
catalysts for hydrocarbon oxidation is an interesting topic that
Pd/ZrCe. Pd/Ce presents the fewest.
deserves additional investigation.
Our results underscore the role of the support in determin-
4.3. Correlation of PdO stability and catalyst activity ing PdO surface chemistry, and ultimately its activity for methane
oxidation. As we noted earlier, among our catalysts, only Pd/Al
TPR results support the view of site pair density as the primary displayed transient low-temperature activity when pre-treated in
determinant of activity. Figure 7 shows the relationship between air; Al2 O3 demonstrates a unique ability to stabilize the amor-
the location of the TPR maximum and lnAapp : as the reduction tem- phous/crystalline form of PdO that may be responsible for the
perature increases, lnAapp decreases. Pd/Al catalysts display the transient. At higher reaction temperatures, oxidation activity
lowest temperature TPR features, suggesting that creation of sur- depends on the stability of PdO on the catalyst surface. PdO is least
face Pd* from PdO to form the site pairs required for oxidation is stable on Al2 O3 , meaning that conversion of PdO to Pd* for creation
relatively facile. On the other hand, TPR features of the Pd/Ce cat- of PdO–Pd* site pairs is relatively facile—and that Pd/Al catalysts
alysts appear at temperatures more than 40 K higher than Pd/Al, are the most active. PdO is most stable on low-activity Pd/Ce.
consistent with a more stable surface PdO—and more difficult for-
mation of the Pd* partner of the site pair. Pd/ZrCe catalysts display
5. Conclusion
TPR maxima at intermediate temperatures. These results likely
reflect the contributions of lattice oxygen from easily reducible
Supported Pd catalysts display activities for total methane oxi-
supports, like CeO2 and ZrO2 –CeO2 , to PdO stability [31,32].
dation that depend on the composition of the support: among
Fujimoto et al. arrived at a similar conclusion about the relation-
our samples, the observed activity order is Pd/Al > Pd/ZrCe > Pd/Ce.
ship between high PdO stability and low density of PdO–Pd* pairs
Activity is related to the relative density of PdO–Pd* active site
in their study of Pd/ZrO2 catalysts [6]. Zhou et al. [12] and Colussi
pairs on the catalyst surface, which correlates with Aapp . The den-
et al. [14] reported that fraction of Pd present as an oxide is greater
sity of site pairs, in turn, depends on the stability of surface PdO:
if PdO is too stable, than it cannot be partially converted to Pd*
to form the site pairs required for reaction. Among the supports
examined, CeO2 delivers an especially stable PdO, which accounts
for the relatively low activity of the Pd/Ce catalyst.

Acknowledgement

This work was supported by Carnegie Mellon University through

Fig. 7. The pre-exponent of the oxidation rate constant decreases (the catalyst
becomes less active) as the temperature of PdO reduction in the TPR experiment
increases (PdO becomes more stable). Green triangles are results for 1.0 Pd(NO3 )/Ce
calculated using the minimum and maximum Pd dispersions of the other catalysts.
(Color version available online.)
a Berkman Faculty Development grant; by the College of Engineer-
ing through a Dean’s Infrastructure Grant; and by the Department
of Chemical Engineering.
The authors thank George Huber and members of the Huber
group at the University of Wisconsin-Madison for assistance with
CO chemisorption characterization; and Meghan Swanson and
Mark Zanella at Mine Safety Appliances Company for many helpful
discussions.
Appendix A.
See Figs. A1–A9.
60 J.B. Miller, M. Malatpure / Applied Catalysis A: General 495 (2015) 54–62

Fig. A1. XRD patterns of Al2 O3 and Pd/Al catalysts. All spectra are typical of ␥-Al2 O3 , with no evidence of either Pd or PdO.

Fig. A2. XRD patterns of ZrO2 –CeO2 and Pd/ZrCe catalysts. All patterns reflect a mixture of tetragonal and monoclinic phases of ZrO2 , with no evidence of CeO2 , PdO or Pd.

Fig. A3. XRD patterns of CeO2 and Pd/Ce catalysts. All patterns display the features of CeO2 , with no evidence of either Pd or PdO.
 J.B. Miller, M. Malatpure / Applied Catalysis A: General 495 (2015) 54–62
Fig. A4. TPRx of 1.4 Pd (NO3 )/Al, 0.1004 g sample size. The catalyst is most active
during 1up. A ‘low-temperature feature’ is observed during the 1up and 3up portions
of the experiment. (Color version available online.)
Fig. A5. TPRx of 1.4 Pd (TAPN)/Al, 0.0982 g sample size. Catalyst activity is
approximately constant throughout the experiment. A ‘low-temperature feature’
is observed during 1up and 3up portions of the experiment. (Color version available
online.)
Fig. A6. TPRx of 0.5 Pd (TAPN)/ZrCe, 0.1949 g sample size. Catalyst activity increases
throughout the experiment. No ‘low-temperature features’ are observed. (Color
version available online.)
61
Fig. A7. TPRx of 1.1 Pd (NO3 )/ZrCe, 0.1966 g sample size. Catalyst activity increases
throughout the experiment. No ‘low-temperature features’ are observed. (Color
version available online.)
Fig. A8. TPRx of 0.5 Pd (NO3 )/Ce, 0.2023 g sample size. The catalyst never achieves
full conversion; activity declines throughout the experiment. No ‘low-temperature
features’ are observed. (Color version available online.)
Fig. A9. TPRx of 1.0 Pd (NO3 )/Ce, 0.0996 g sample size. The catalyst never achieves
full conversion; activity declines throughout the experiment. No ‘low-temperature
features’ are observed. (Color version available online.)
References
[1] Y.-H. Chin, D.E. Resasco, in: J.J. Spivey (Ed.), Catalysis, Royal Scoiety of Chem-
istry, Cambridge, 1999, pp. 1–39.
[2] G. Centi, J. Mol. Catal. A: Chem. 173 (2001) 287–312.
[3] D. Ciuparu, M.R. Lyubovsky, E. Altman, L.D. Pfefferle, A.K. Datye, Catal. Rev. 44
(2002) 593–649.
[4] T.V. Choudhary, S. Banerjee, V.R. Choudhary, Appl. Catal. A: Gen. 234 (2002)
1–23.
62 J.B. Miller, M. Malatpure / Applied Catalysis A: General 495 (2015) 54–62
[5] P. Gelin, M. Primet, Appl. Catal. B 39 (2002) 1–37.
[6] K. Fujimoto, F.H. Ribeiro, M. Avalos-Borja, E. Iglesia, J. Catal. 179 (1998)
431–442.
[7] S.K. Matam, G.L. Chiarello, Y. Lu, A. Weidenkaff, D. Ferri, Top. Catal. 56 (2013)
239–242.
[8] L.M.T. Simplicio, S.T. Brandao, E.A. Sales, L. Lietti, F. Bozon-Verduraz, Appl. Catal.
B: Environ. 63 (2006) 9–14.
[9] D. Roth, P. Gelin, A. Kaddouri, E. Garbowski, M. Primet, E. Tena, Catal. Today 112
(2006) 134–138.
[10] J.N. Carstens, S.C. Su, A.T. Bell, J. Catal. 176 (1998) 136–142.
[11] N. van Vegten, M. Maciejewski, F. Krumeich, A. Baiker, Appl. Catal. B 93 (2009)
38–49.
[12] R. Zhou, B. Zhao, B. Yue, Appl. Surf. Sci. 254 (2008) 4701–4707.
[13] S. Colussi, A. Trovarelli, C. Cristiani, L. Lietti, G. Groppi, Catal. Today 180 (2012)
124–130.
[14] S. Colussi, A. Trovarelli, E. Vesselli, A. Baraldi, G. Comelli, G. Groppi, et al., Appl.
Catal. A 390 (2010) 1–10.
[15] A. Satsuma, R. Sato, K. Osaki, K. Shimizu, Catal. Today (2012) 61–65.
[16] B. Yue, R. Zhou, Y. Wang, X. Zheng, Appl. Catal. A: Gen. 295 (2005) 31–39.
[17] W.R. Schwartz, L.D. Pfefferle, J. Phys. Chem. C 116 (2012) 8571–8578.
[18] D. Ciuparu, F. Bozon-Verduraz, L.D. Pfefferle, J. Phys. Chem. B 106 (2002)
3434–3442.
[19] P. Castellazzi, G. Groppi, P. Forzatti, A. Baylet, P. Marecot, D. Duprez, Catal. Today
155 (2010) 18–26.
[20] G. Groppi, C. Cristiani, L. Lietti, C. Ramella, M. Valentini, P. Forzatti, Catal. Today
50 (1999) 399–412.
[21] F.H. Ribeiro, M. Chow, R.A. Dalla Betta, J. Catal. 146 (1994) 537–544.
[22] M. Schmal, M.V.M. Souza, V.A. Alegre, M.A.P. da Silva, D.V. Cesar, C.A.C. Perez,
Catal. Today 118 (2006) 392–401.
[23] C.F. Cullis, B.M. Willatt, J. Catal. 83 (1983) 267–285.
[24] S.C. Su, J.N. Carstens, A.T. Bell, J. Catal. 176 (1998) 125–135.
[25] J. Au-Yeung, K. Chen, A.T. Bell, E. Iglesia, J. Catal. 188 (1999) 132–139.
[26] J.G. McCarty, Catal. Today 26 (1995) 283–293.
[27] R.J. Farrauto, M.C. Hobson, T. Kennelly, E.M. Waterman, Appl. Catal. A 81 (1992)
227–237.
[28] A. Amin, A. Abdeli, R. Hayes, M. Votsmeier, W. Epling, Appl. Catal. A 478 (2014)
91–97.
[29] P.O. Thevenin, A. Alcalde, L.J. Pettersson, S.J. Jaras, J.L.G. Fierro, J. Catal. 215
(2003) 78–86.
[30] P. Araya, S. Guerrero, F. Robertson, F.J. Gracia, Appl. Catal. A 283 (2005)
225–233.
[31] C. Bozo, N. Guilhaume, J.-M. Herrmann, J. Catal. 203 (2001) 393–406.
[32] L. Kai, W. Xuezhong, Z. Zexing, W. Xiaodong, W. Duan, J. Rare Earth 25 (2007)
6–10.