Increase of Pd surface area by treatment in dioxygen

Article abstract of DOI:10.1016/j.jcat.2004.03.035

The surface area of Pd(111) and Pd(110) single crystals increased after oxidation in O₂ transformed them to PdO. The surface area of the oxide decreased after complete reduction in H₂. The techniques of STM, TPD, XPS, AES, and LEED were employed to study the Pd(111) and Pd(110) single-crystal surfaces after they were subjected to O₂ oxidation, methane combustion, and H₂ reduction. The surface area of the treated palladium single crystals was measured by ¹⁸O isotope exchange and by direct measurements using the STM image. These two methods showed agreement within 20%. After oxidation in O₂ (100 and 150 Torr) at 600 K, the surface area for both Pd(111) and Pd(110) single crystals increased by a factor of approximately two. The effect was more pronounced on the Pd(111) surface. The oxidized surfaces were covered with 3- to 4-nm semispherical oxide agglomerates that formed a cauliflower-like structure 10-20 nm in size. Similar surface structures were observed after exposure of the Pd single crystals to a lean O₂ and CH₄ reaction mixture (O₂:CH ₄=10:1). Thus, the oxidized single crystal becomes amorphous. Reduction in H₂ decreased the surface area of the preoxidized Pd(111) and Pd(110) crystals. An amorphous metallic surface was produced after H ₂ reduction at 373 K, whereas a smooth surface with characteristic single-crystal features was observed after reduction at 673 K. These experiments suggest that oxidizing a Pd metal catalyst or reducing an oxidized Pd catalyst, for example, before palladium metal surface area measurement, will affect the surface area of the sample. It also shows that the increase in surface area on Pd catalysts after oxidation treatment is caused by surface roughening.

Full text of DOI:10.1016/j.jcat.2004.03.035

Journal of Catalysis 225 (2004) 7–15
www.elsevier.com/locate/jcat
Increase of Pd surface area by treatment in dioxygen
a
a
b
a,∗
Jinyi Han, Guanghui Zhu, Dmitri Y. Zemlyanov, and F.H. Ribeiro
a
Worcester Polytechnic Institute, Department of Chemical Engineering, Worcester, MA 01609-2280, USA
b
Materials and Surface Science Institute, University of Limerick, Limerick, Ireland
Received 13 November 2003; revised 22 March 2004; accepted 24 March 2004
Available online 24 April 2004
Abstract
The surface area of Pd(111) and Pd(110) single crystals increased after oxidation in O transformed them to PdO. The surface area of the
2
oxide decreased after complete reduction in H . The techniques of STM, TPD, XPS, AES, and LEED were employed to study the Pd(111)
2
and Pd(110) single-crystal surfaces after they were subjected to O oxidation, methane combustion, and H reduction. The surface area
2
2
1
8
of the treated palladium single crystals was measured by O isotope exchange and by direct measurements using the STM image. These
two methods showed agreement within 20%. After oxidation in O (100 and 150 Torr) at 600 K, the surface area for both Pd(111) and
2
Pd(110) single crystals increased by a factor of approximately two. The effect was more pronounced on the Pd(111) surface. The oxidized
surfaces were covered with 3- to 4-nm semispherical oxide agglomerates that formed a “cauliflower-like” structure 10–20 nm in size. Similar
surface structures were observed after exposure of the Pd single crystals to a lean O and CH reaction mixture (O :CH = 10:1). Thus, the
2
4
2
4
oxidized single crystal becomes amorphous. Reduction in H decreased the surface area of the preoxidized Pd(111) and Pd(110) crystals. An
2
amorphous metallic surface was produced after H reduction at 373 K, whereas a smooth surface with characteristic single-crystal features
2
was observed after reduction at 673 K. These experiments suggest that oxidizing a Pd metal catalyst or reducing an oxidized Pd catalyst,
for example, before palladium metal surface area measurement, will affect the surface area of the sample. It also shows that the increase in
surface area on Pd catalysts after oxidation treatment is caused by surface roughening.

2004 Elsevier Inc. All rights reserved.
Keywords: Palladium oxidation; PdO morphology; PdO surface area measurement
1
. Introduction
lyst for methane combustion [5], is an example. Depending
on the oxygen partial pressure and the temperature, either
Pd metal or Pd oxide can be the thermodynamically sta-
ble phase. In air at atmospheric pressure, PdO is the active
phase up to 1073 K [6–8]. Therefore, the oxidized sam-
ples will have to be reduced to Pd metal (usually in H2,
see, for example [9–11]) before the measurement of sur-
face area by chemisorption. It is implicitly assumed then
that the surface area does not change after PdO reduction or
Pd oxidation. In general, the surface area may change dur-
ing reduction. Indeed, H2 reduction has been suggested to
modify the surface morphology and the effect varies with
the treatment temperature [12,13]. The possible changes in
surface area after reduction and oxidation prompted us to
study the effect of reduction and oxidation on surface area.
A direct consequence of accurate measurements on PdO sur-
face area is to allow for the proper measurement of turnover
rates. The surface morphology of palladium oxide formed on
Pd(111) and Pd(110) single crystals and the influence of H2
reduction on the oxidized surfaces were studied by scanning
The turnover rate (TOR) is defined as the number of
molecules reacted on an active site per unit of time [1]. It
is one of the most important parameters used in quantify-
ing the properties of a catalyst [2,3]. To calculate the TOR,
it is necessary to know the number of active sites, a rarely
known parameter. In practice, the total surface area is mea-
sured and the total number of sites is calculated from it by
assuming a certain site density. For example, the adsorption
of H2, O2, CO, and H2–O2 titration is used to measure the
surface area of supported metal catalysts [1,4], but the cat-
alyst must be in the reduced state for these techniques to
work; in many cases the active catalyst is not in metallic
form. Palladium, which is referred as the most active cata-
*
Corresponding author. Current address: School of Chemical Engineer-
ing, Purdue University, 480 Stadium Mall Drive, West Lafayette, IN 47907-
2100. Fax: (765) 494-0805.
E-mail address: fabio@purdue.edu (F.H. Ribeiro).
0021-9517/$ – see front matter  2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2004.03.035

8
J. Han et al. / Journal of Catalysis 225 (2004) 7–15
tunnelling microscopy (STM) and temperature-programmed
desorption (TPD). The surface area measurements were car-
ried out by 18O isotope exchange [14,15] and by surface area
integration of STM images. We will show that after oxida-
tion in O2 (100 and 150 Torr) at 600 K, the surface area for
both Pd(111) and Pd(110) single crystals increases by a fac-
tor of approximately two and that reduction in H2 decreases
the surface area. Thus, one should realize when working
with Pd catalysts that there will be variation of surface area
after oxidation and reduction treatments.
30 min in 150 Torr of O2 at 600 K and then cooled to
room temperature (RT) before oxygen was pumped out. The
oxidized sample was heated to the desired reduction tem-
perature at a pressure in the reaction chamber better than
−7
10 Torr and then 1 Torr of H2 was introduced for 1 min.
−6
The H2 was evacuated at a pressure better than 10 Torr
before the reduced sample was cooled to RT.
The crystal structure of the oxide formed on the sur-
face of Pd single crystals by oxidation in 150 Torr O2 at
600 K for 30 min was checked by X-ray diffraction (XRD,
Rigaku Geigerflex) with Cu-Kα radiation source and a Ni
filter. The sample was dismounted from the standard STM
sample holder before analysis by XRD.
2
. Experimental methods
The combustion of methane on Pd single crystals was
3
The experiments were carried out in a specially de-
performed in the 615 cm batch reactor with reaction gases
signed system, which consisted of three chambers: ultra-
high vacuum (UHV) analysis chamber, UHV STM cham-
ber, and high-pressure reactor. The analysis chamber was
equipped with X-ray photoelectron spectroscopy (XPS),
Auger-electron spectroscopy (AES), low-energy electron
diffraction (LEED), and TPD. The STM chamber accommo-
dated an ambient-temperature UHV STM (RHK, Inc.). The
being introduced from a gas manifold in the following or-
der: N2 (624 Torr), O2 (160 Torr), and finally CH4 (16 Torr).
The reactants were mixed for 30 min before reaction by a
circulation pump Model MB-21 (Metal Bellows, Inc.) at a
3
−1
nominal rate of 1000 cm min . Reaction (600 K) was nor-
mally carried out for 60 min with about 2.5% of the CH4
being consumed. The reaction mixture was analyzed with an
Agilent 6890 Series gas chromatograph using a thermal con-
ductivity detector, and a 15-ft Carboxen 1000, 60/80 mesh
column.
−10
base pressure in the UHV chambers were 5 × 10
1 Torr = 133.3 Pa). The high-temperature/high-pressure
treatments were carried out in the high-pressure reactor
Torr
(
−
8
The 18O isotope exchange experiment was performed in
the high-pressure reactor following the procedure described
before [14]. Briefly an oxidized palladium sample was ex-
(
base pressure 2 × 10 Torr). The sample could be trans-
ferred between the chambers, without exposure to the at-
mosphere, by means of a 142 cm long transfer arm.
posed to 5 Torr 18O at 600 K for 12 s, and the uptake of 18O
The 0.8 mm thick, 7.2 mm diameter Pd(111) and 1 mm
thick, 8.5 mm diameter Pd(110) single crystals (both Prince-
2
exchanged was then measured by TPD analyzing all gases
containing labeled oxygen.
◦
ton Scientific Corp., misalignment < 0.5 ) were used as
planar model catalysts. The temperature was measured by
a chromel–alumel thermocouple spot-welded onto the side
of the sample. In the analysis chamber, the sample was
heated by electron-bombardment from the rear. In the reac-
tion chamber, the sample was heated by an infrared lamp
3. Results
3.1. Surface area measurements: STM image analysis and
1
8
(
Research, Inc.) from the front.
O isotope exchange
The temperature-programmed desorption spectra were
−
1
collected at a constant heating rate of 5 K s . The cov-
erage was calculated by integrating the area under the
TPD peak and measured in monolayers (1 monolayer =
The surface areas were measured by STM image analy-
sis and 18O isotope exchange. The method using STM im-
ages relies on the generation of the surface topographic
image by STM and therefore can be used for samples
that are conductors and that can be outlined by the mi-
croscope tip. The surface area is then calculated by taking
discrete image pixels and joining three contiguous points
to make one triangle. Integration of the area of individ-
ual triangle gives the total area. In detail, each STM im-
age is collected as a grid of 512 × 512 points. Each point
(i, j) of a STM image can be described in Cartesian coordi-
nates by xi,j , yi,j , zi,j (x, y, z projections). Three adjacent
1
5
−2
1
1
ML = 1.53 × 10 atoms cm on Pd(111) and 0.94 ×
15
−2
0
atoms cm on Pd(110)) by comparing to standard cal-
ibration values [16,17]. STM images were obtained using
Pt–Ir tips electrochemically etched in NaCl/NaNO3 melt.
The areas presented here are an average of at least 10 images;
the variation in the area among the images is about 20%.
The crystal cleaning procedures consisted of cycles of
+
Ar sputtering at room and elevated temperatures, flashing
in UHV, and exposure to O2 and NO2 followed by anneal-
ing at 1100 K for 60 s in UHV. The sample cleanness was
checked by TPD, AES, and LEED. The NO2 and O2 gases
were introduced into the analysis chamber through a capil-
lary dozer.
points, for instance, (xi,j yi,j zi,j ), (x
i+1,j i+1,j i+1,j
y
z
)
and (xi,j+1 yi,j+1 zi,j+1) are used to construct a 3D ori-
ented triangle. The total surface area is obtained by summing
all triangle areas for each i and j from 1 to 512. Since
other triangles can be constructed on other sets of the adja-
cent points, for instance (xi,j yi,j zi,j ), (xi,j+1 yi,j+1 zi,j+1)
The following experimental protocol was used for H2 re-
duction experiments. The single crystal was oxidized for

J. Han et al. / Journal of Catalysis 225 (2004) 7–15
9
and (xi+1,j+1 yi+1,j+1 zi+1,j+1), the surface area is aver-
aged through the possible choice of triangles. No filtering or
smoothing process was applied to the STM images prior to
surface area calculations.
lic Pd-supported catalysts were prepared by calcination at
1123 K in air for 24 h followed by quenching to LN2 temper-
ature; metal is the stable phase under these conditions. The
oxide phase was then prepared by treating the metallic phase
at 973 K in air for 24 h. Before the surface area measure-
ment by chemisorption, the oxide-supported catalysts were
reduced in 1 atm H2 at 373 K for 1. After oxidation, the
surface area for the Pd-supported catalysts increased approx-
imately by a factor of 2.5 (Table 2).
1
8
The O isotope exchange method is based on the detailed
1
8
characterization of the O exchange kinetics by Au-Yeung
et al. [18]. The conditions described in the experimental
methods section were designed to ensure that the exchange
between 16O in PdO and O2 isotope in the gas phase would
18
1
8
happen only on the surface, without appreciable O diffu-
sion to the bulk. In fact, the rate for recombination of oxygen
3
.2. Oxidation of Pd(111) and Pd(110) single crystals
−1
atoms at the surface of PdO was about 4 s at 600 K [18]
and this amounts to about 50 turnovers per site in the 12 s the
exchange lasts. The number of equivalent layers that oxygen
atoms diffuse under exchange conditions was about 0.05 ML
on both Pd(111) and Pd(110) [19]. Thus, only surface oxy-
gen should be exchanged.
Exposure of the Pd(111) and Pd(110) single crystals to
50 Torr of O2 at 600 K for 30 min resulted in palladium ox-
1
ide formation, which was monitored by AES and XPS (spec-
tra not shown). The Auger peaks of oxygen were located at
4
90 and 510 eV, and the O/Pd atomic ratio was approxi-
mately 0.7. This value was lower than the expected value for
PdO due to electron beam decomposition of the oxide, with
the oxygen desorbing as O2 [14,20]. The core-level Pd 3d5/2
peak was located at 336.8 eV, shifted 1.8 eV from the metal-
lic peak position at 335.0 eV due to oxidation. On Pd(111)
after oxidation, no X-ray diffraction pattern corresponding
to PdO could be observed; only the peaks corresponding to
the metal were observed. The oxygen uptakes on the Pd(111)
and Pd(110) surfaces were 22 ML and 27 ML, respectively.
The STM images obtained after exposure of Pd(111) and
Pd(110) single crystals to 150 Torr of O2 at 600 K for 30 min
are shown in Fig. 1. Both oxidized Pd(111) and Pd(110) sin-
gle crystals showed a similar amorphous PdO surface mor-
phology. The surface was composed of semispherical oxide
agglomerates that tended to aggregate into a “cauliflower-
like” structure. In general, the agglomerate size varied de-
pending on the oxygen pressure and the exposure time. Un-
der the reaction conditions in Fig. 1, the agglomerates grown
on the Pd(111) and Pd(110) surfaces were 4.3 ± 0.7 and
The results of the surface area measurement on palla-
dium after oxidation in O2 and catalytic combustion of a
lean CH4 mixture (O2:CH4 = 10:1) at 600 K are summa-
rized in Table 1. It was found out that the two proposed
methods: surface area integration of STM images and 18
O
exchange agreed within ± 20%. In addition, the two meth-
ods provided consistent results with the measurement of the
surface area of a polycrystalline Pd foil following CH4 com-
bustion in lean condition [14] which is listed in the last row
in Table 1.
Table 2 provides comparisons on the surface area increase
after O2 oxidation of Pd(110), Pd(111), 8.5% Pd/Al2O3, and
1
0% Pd/ZrO2 [10]. For supported catalysts, the surface areas
were determined by H2–O2 titration, after sample reduction.
The procedure was described in detail in [10]. Briefly, metal-
Table 1
Comparison of surface area determined by STM image analysis (STM) and
18
18
O isotope exchange ( O)
3
.4 ± 1.1 nm in diameter, respectively. Although the surface
a
Sample
Pd(111)
Treatment
condition
Surface area increase
roughness did not allow us to draw any conclusion on pref-
erential orientations that the oxide agglomerates grew, prior
work has shown that for the Pd(111) surface, the oxide ag-
glomerates appeared first in the vicinity of steps, whereas
18
STM
O
Oxidation
Reaction
Reaction
Reaction
1.8
3.2
1.6
–
2.2
3.0
1.8
2.2
Pd(110)
Pd foil
¯
the ꢀ110ꢁ direction was preferential during the early stages
b
of Pd(110) oxidation [19].
Oxidation carried out at 100 Torr O , 600 K, 10 min, and reaction at 600 K,
The surface morphologies of Pd(111) and Pd(110) after
methane combustion in lean condition at 600 K are presented
in Fig. 2. A clearer “cauliflower-like” structure 20 ± 5 nm
in size is visible on both Pd surfaces. The surface area was
2
1
6 Torr CH , 160 Torr O , 624 Torr N , 60 min.
4
2
2
a
Oxide surface area
Metal surface area
Surface area increase =
.
b
Monteiro et al. [14].
Table 2
Surface area increase after oxidation of Pd catalysts
a
Sample
Surface area increase
Oxidation condition
Method of measurement
H –O titration
Reference
8.5% Pd/Al O
2.7
2.5
1.9
2.3
973 K, air, 24 h
973 K, air, 24 h
[10]
[10]
Present work
Present work
2
3
2
2
10% Pd/ZrO
H –O titration
2
2
2
Pd(110)
Pd(111)
600 K, 150 Torr O , 30 min
STM image analysis
STM image analysis
2
600 K, 150 Torr O , 30 min
2
a
Oxide surface area
Metal surface area
Surface area increase =
.

10
J. Han et al. / Journal of Catalysis 225 (2004) 7–15
Fig. 1. STM images obtained after exposure of Pd(111) (a and b) and Pd(110) (c and d) single crystals to 150 Torr of O , at 600 K for 30 min. The sample bias
2
was 1 V; the tunneling current was 0.5 nA.
higher as compared to the oxidized surface in pure O2, par-
ticularly for the Pd(111) surface (Table 1).
the formation of an amorphous Pd metal overlayer. As
shown by STM in Figs. 3a and 3c, the surface was rough
with metal agglomerates visible. The agglomerates were not
of spherical shape as before reduction (see Fig. 1), but elon-
gated with an aspect ratio around 2–2.5:1 (4.5 ± 0.5 nm vs
2.1 ± 0.4 nm on Pd(110) and 3.5 ± 0.7 nm vs 1.6 ± 0.4 nm
on Pd(111)). On the Pd(111) surface, the agglomerates were
slightly angled with respect to the monoatomic steps, which
were barely distinguishable. On the Pd(110) surface, the
3
.3. H2 reduction
To investigate the effect of H2 reduction on surface mor-
phology, the Pd(111) and Pd(110) surfaces preoxidized in
50 Torr of O2 at 600 K for 30 min were studied after
1
reduction in 1 Torr H2 at 373 and 673 K. The lower tem-
perature of 373 K is used for H2–O2 titration on supported
Pd catalysts, see, for example [9,10]. The higher tempera-
ture of 673 K is used under vacuum for removing adsorbed
hydrogen present on Pd after reduction and before hydro-
gen chemisorption [21]. According to the PdHn phase dia-
gram [22], no β-phase palladium hydride is expected under
these reduction conditions. No oxygen was observed by TPD
and by AES after the reduction at both 373 and 673 K, show-
ing that the samples were completely reduced.
¯
“rod-shaped” agglomerates were aligned along ꢀ110ꢁ.
Temperature-programmed desorption did not reveal any
hydrogen desorption after reduction at 673 K. A diffuse
(1 × 1) pattern referred to the bare unreconstructed surface
was observed by LEED for both Pd(111) and Pd(110) sur-
faces. The surface reduced at 673 K was smoother than that
after 373 K reduction as shown in Figs. 3b and 3d with
monoatomic steps visible on the Pd(111) surface. One no-
ticeable difference was that the step was curved and not
straight as on a clean Pd(111) surface. Several holes with
mean size of 3 nm and monoatomic depth did not heal on the
terraces. A mesoscopic ordered island structure composed of
The desorption of residual hydrogen was detected by
TPD from the Pd(110) surface reduced at 373 K. No LEED
pattern was observed after 373 K reduction, which revealed

J. Han et al. / Journal of Catalysis 225 (2004) 7–15
11
Fig. 2. STM images obtained from Pd(111) (a and b) and Pd(110) (c and d) surfaces exposed to 16 Torr CH , 160 Torr O , and 624 Torr N at 600 K for
4
2
2
60 min. The sample bias was 1 V; the tunneling current was 0.5 nA.
alternating bright and dark islands appeared on the Pd(110)
surface after H2 reduction at 673 K. The mean size of the
island was approximately 14 nm in the ꢀ100ꢁ direction and
Table 3
Surface area decrease after reduction of preoxidized Pd(111) and Pd(110)
evaluated by integration of STM images
¯
Oxidized
sample
H reduction
2
temperature
Surface area
increase factor
2
0 nm in the ꢀ110ꢁ direction. The height difference between
a
the islands was 0.14 nm, which is equal to the height of the
monoatomic step on the Pd(110) surface. Each mesoscopic
island was composed of elongated agglomerates with aspect
ratio of 3–4:1 (6.9 ± 1.1 nm vs 2.2 ± 0.6 nm).
Pd(110)
No reduction
1.9
1.5
1.1
3
6
73 K
73 K
Pd(111)
No reduction
2.3
1.3
1.1
The surface area decrease after H2 reduction is summa-
rized in Table 3. It is shown that the surface area of the
preoxidized Pd(111) and Pd(110) single crystals decreased
even after reduction at 373 K, and the effect was more pro-
nounced after 673 K reduction.
3
6
73 K
73 K
Oxidation conditions were 150 Torr O at 600 K for 30 min. Reduction
conditions were 1 Torr H at 373 or 673 K for 1 min.
2
2
a
Oxide surface area
Surface area increase factor = Metal surface area
.
combustion, and H2 reduction. In the method of 18O isotope
4
. Discussion
1
8
exchange, the exchange conditions between O in the gas
4
.1. STM image analysis and 18O isotope exchange
phase and O in PdO should be chosen to ensure enough
16
exchange turnovers and to limit bulk diffusion.
We report the surface area integration of STM images and
O isotope exchange as methods to measure the surface
area on palladium single crystals after O2 oxidation, CH4
In certain circumstances, experimental errors also exist
in the STM image analysis method. At low bias voltages
STM probes the oxide–substrate interface, whereas STM be-
1
8

12
J. Han et al. / Journal of Catalysis 225 (2004) 7–15
Fig. 3. STM images obtained after reduction in 1 Torr of H of the preoxidized Pd(111) (a and b) and Pd(110) (c and d) surfaces at 373 K (a, c) and at 673 K
2
¯
(b, d). Arrows designate the step orientation in (a and b), and ꢀ110ꢁ in (c and d). The sample bias was 1 V; the tunneling current was 0.5 nA.
comes more sensitive to the vacuum–oxide interface at high
ers is estimated to be in the range of 5–20 ML; and the band
gap for bulk PdO is reported to be 4.0–5.0 V [25]. The ab-
sence of the “bias-voltage” dependence could be explained
by the high surface conductivity of oxygen vacancies gener-
ated under UHV conditions.
The tip effect can also introduce experimental errors. In
fact, the image of an object on the surface might be just tip
shape deconvolution. The agglomerates might appear larger
in the STM image. No changes of the surface topology and
of the surface area were observed when the tip was changed,
thus implying that this effect was not significant. On the
other hand, if the tip cannot follow the contour of the surface,
the measured surface area by STM will be in error. The good
agreement between the STM data and the 18O exchange data
(Table 1) allows us to rule out this effect as well.
bias voltage. This effect was demonstrated for instance by
Bertrams et al. [23] for a thin Al2O3 film on NiAl(110). No
change of the tip height was detected when moving between
an Al2O3 island and the NiAl substrate at bias voltage be-
low 4 V (the band gap of Al2O3 is approximately 7–8 V).
On the other hand, an apparent height of 3.5 Å with re-
spect to the metal support was observed after the bias voltage
was increased above 4 V. Hagendorf et al. [24] showed that
CoO(001) islands on the Ag(001) surface appeared to be
depressed at the bias voltage between −1.5 and +2.2 V,
whereas above and below these voltages the islands were
protruding from the substrate. The maximum height varia-
tion was 6 Å. In our experiments this effect was not signifi-
cant. The variation of the size and height of the oxide clusters
was in the range of 0.1–0.3 Å when the bias voltage was
changed from 0.1 to 5.0 V. The thickness of the oxide lay-
Another important issue is the dependence of the surface
area calculated from a STM image as a function of the grid

J. Han et al. / Journal of Catalysis 225 (2004) 7–15
13
step as discussed by Brown et al. [26]. The x–y step used in
agglomerates is different on the Pd(111) and Pd(110) sur-
faces. The most dense Pd(111) surface expanded more than
the more open Pd(110) surface. The concentration of dis-
solved oxygen also might affect the oxide agglomerate size.
Oxidation of Pd single crystals including Pd(111), Pd(100)
and Pd(110) suggested that the oxygen diffusion from sur-
face to the bulk depends on the surface crystallography with
the highest rate of diffusion determined on the open Pd(110)
surface. Indeed, the oxygen uptakes were 13 ML, 17 ML,
and 25 ML on Pd(111), Pd(100), and Pd(110) respectively
after exposure to 25 Torr O2 at 600 K for 10 min [19]. There-
fore the Pd → PdO transformation might proceed at a higher
bulk oxygen concentration for a given O2 gas pressure for
the Pd(110) single crystal, resulting in a greater number
of nucleation sites. This should lead to smaller agglomer-
ates.
It is interesting to note that the surface area increase
after oxidation of a well-annealed metallic supported cata-
lyst agrees with the corresponding result on single crystals
(Table 2). This is to be expected as metal crystallites on
a supported catalyst are composed of low index planes, as
shown for example by TEM [40]. Thus, the increase in sur-
face area observed on supported Pd catalysts after oxidation
is caused by surface roughening.
our study was approximately 1 Å (500 × 500 Å with image
resolution 512×512 lines), two times smaller than the Pd–O
bond length (2.02 Å). The proper choice of the grid step is
supported by the comparison with the 18O isotope exchange
method as shown in Table 1.
4
.2. Oxidation of Pd single crystals
Oxidation of Pd single crystals in low pressure (< 1 Torr,
−6
typically < 10
Torr) O2 has been studied previously,
see, for example [16,17,20,27–37]. Chemisorption of oxy-
gen atoms on Pd single crystals results in a p(2 × 2) over-
layer with 0.25 ML coverage on Pd(111) [16,38], p(2 × 2)
and c(2 × 2) overlayers on Pd(100) with 0.25 and 0.5 ML
coverages, and the reconstructed c(2 × 4) superstructure on
Pd(110) with 0.5 ML coverage [17]. The oxygen uptake be-
yond the saturation coverage could be reached (2.3 ML on
Pd(111), 0.8 ML on Pd(100), and 1.8 ML on Pd(110)) by
dosing stronger oxidants such as NO2 [16,27,30]or exposing
to O2 at high pressures and temperatures [17,20,29]. A sur-
face oxide appears on Pd(111) [16,20] and Pd(110) [32,39],
which presents a complex crystal structure distinct from that
of PdO or the original metallic facets. On Pd(100), the sur-
face oxide appears to have a structure close to PdO(001).
It is suggested that PdO forms through a nucleation mecha-
nism [16]. Once the bulk oxide forms, the surface roughens
4.3. H2 reduction
[
16,27].
It is possible that bulk palladium oxide could be formed
The measurement of surface area for supported Pd cat-
alysts requires reduction of the oxidized sample, usually in
H2, before chemisorption. Reduction in H2, however, results
in changes of the surface area as demonstrated in Table 3.
Note that the reduction temperature is an important parame-
ter. Scanning tunnelling microscopy images of the preoxi-
dized Pd(111) and Pd(110) surfaces reduced at a temperature
of 673 K demonstrated a fairly smooth surface with clearly
visible monoatomic steps (Fig. 3). The surface area after
reduction was only about 1.1 times higher than the metal
surface area compared to the twofold increase after preoxi-
dation. On a supported catalyst, reduction at high tempera-
tures might also cause Pd sintering. Suh and Park [41] using
transmission electron microscopy (TEM) observed signif-
icant sintering of finely dispersed Pd/C catalyst following
treatment at 673 K. Logan et al. [13] detected only Rh metal
agglomerates with well-defined low index facets after reduc-
tion in H2 at 773 K. At 373 K however, the rate of reduc-
tion, i.e., the rate of generation of free Pd atoms exceeds
the rate of their diffusion and recrystallization and therefore
the surface remains amorphous. TEM studies by Datye et
al. [12] reported the formation of an amorphous Rh metal-
lic layer on crystalline RhO agglomerates during reduction
at 323 K in hydrogen. The surface area decreased even af-
ter 373 K reduction but it could be due to the collapse of
the Pd agglomerates after the PdO → Pd transformation. It
is reasonable to assume that the large agglomerates should
experience greater shrinking due to the bigger contribution
of bulk atoms compared to surface atoms. Thus, the big-
only when a critical concentration of dissolved oxygen was
reached in the near-surface region. The formation of this
stoichiometric PdO was characterized by the drastically
dropped oxygen uptake rate, the complete fading of the
metallic Pd(110) LEED pattern, the 0.2–0.7 O/Pd atomic
ratio detected by AES, and the roughened cauliflower-like
surface structure imaged by STM [19]. The roughened sur-
face structure with an increase in surface area could be ra-
tionalized as a major lattice expansion during the chemical
transformation ‘Pd metal’ → ‘Pd oxide’. The Pd atom den-
sity for the oxide is only 60% of the density for the metal
structure. In addition, the Pd–Pd nearest neighbor distance
is 3.04 Å for PdO, about 10% larger than the 2.75 Å for Pd
metal.
The fact that we did not observe any XRD pattern of PdO,
which should be detectable for the thick PdO layer (2–7 nm
estimated from STM images), is an indication that the ox-
ide layer is either amorphous or crystalline with crystallites
too small to diffract coherently. The STM images do not re-
veal preferential growth and thus we conclude that the oxide
layer is amorphous. A number of semispherical oxide ag-
glomerates were observed in this study (Fig. 1) possibly due
to many nucleation sites of oxide growth. Unfortunately we
do not know the nature of these sites. It might be structural
defects but most probably this is a fluctuation of the oxygen
concentration in the near-surface region, which causes a lo-
cal transformation Pd → PdO. The mean size of the oxide

14
J. Han et al. / Journal of Catalysis 225 (2004) 7–15
ger agglomerates collapse more. This might explain why the
oxidized Pd(110) surface showed only 24% surface area de-
crease after H2 reduction at 373 K, whereas the surface area
decreased 42% for oxidized Pd(111).
before Pd surface area measurement can modify the Pd sur-
face area on supported catalysts.
The main conclusion from the H2 reduction experiments
is that the area of the reduced sample is different from the
area of the actual oxidized catalyst.
Acknowledgment
We gratefully acknowledge support from the Office of
Basic Energy Sciences, Chemical Sciences, U.S. Depart-
ment of Energy, Grant DE-FG02-00ER15408.
5
. Summary
References
Planar model catalysts Pd(111) and Pd(110) were used
to study palladium oxidation and methane combustion. The
surfaces were studied by STM, TPD, XPS, AES, and LEED.
The advantages of using a model catalyst approach were
demonstrated by accurate surface area measurements, which
were performed by the surface area integration of STM im-
ages and by 18O isotope exchange. The surface area inte-
gration of STM images was performed by triangulation of
the STM image based on the x/y grid with a 1 Å step. Sur-
[
1] M. Boudart, G. Djega-Mariadasson, in: Kinetics of Heterogeneous
Catalytic Reactions, Princeton University Press, Princeton, 1984, p. 7.
2] M. Boudart, Chem. Rev. 95 (1995) 661.
[
[
3] F.H. Ribeiro, A.E.S. Von Wittenau, C.H. Bartholomew, G.A. Somor-
jai, Catal. Rev. 39 (1997) 49.
[
[
4] J.E. Benson, H.S. Hwang, M. Boudart, J. Catal. 30 (1973) 146.
5] R.B. Anderson, K.C. Stein, J.J. Feenan, L.J.E. Hofer, Ind. Eng.
Chem. 53 (1961) 809.
[6] C. Mallika, O.M. Sreedharan, J.B. Gnanamoorthy, J. Less-Common
face 1 O atoms were exchanged by O atoms in gas phase
6
18
Met. 95 (1983) 213.
[
[
7] W.E. Bell, R.E. Inyard, M. Tagami, J. Phys. Chem. 70 (1966) 3735.
8] G. Bayer, H.G. Wiedemann, Thermochim. Acta 11 (1975) 79.
during 18O exchange experiments, and the uptake was then
1
8
measured by TPD. The area measurement by O isotope
exchange and by STM image analysis showed similar val-
ues.
[9] K.-I. Fujimoto, F.H. Ribeiro, M. Avalos-Borja, E. Iglesia, J. Catal. 179
1998) 431.
10] F.H. Ribeiro, M. Chow, R.A.D. Betta, J. Catal. 146 (1994) 537.
11] G. Prelazzi, M. Cerboni, G. Leofanti, J. Catal. 181 (1999) 73.
12] A.K. Datye, Top. Catal. 13 (2000) 131.
(
[
[
[
[
Exposure of the Pd(111) and Pd(110) single crystals to
1
00–150 Torr O2 at 600 K led to an increase in surface
13] A.D. Logan, K. Sharoudi, A.K. Datye, J. Phys. Chem. 95 (1991) 5568.
area by a factor of approximately two. The effect was more
pronounced for the Pd(111) surface. This might be because
the close packed Pd(111) surface experienced a greater lat-
tice expansion during the Pd → PdO transformation than the
open Pd(110) surface. The other explanation is that because
of the higher oxidation rate on the (110) surface, smaller ox-
ide agglomerates were formed due to a higher density of
oxide nuclei. The smaller oxide agglomerates on the (110)
surface created a lower surface area than the bigger agglom-
erates in the case of the Pd(111) surface. On both surfaces,
oxide agglomerates formed an amorphous structure. A sur-
face with more pronounced cauliflower-like structure was
observed after treatment of the Pd single crystals in a lean
O2 + CH4 reaction mixture.
[14] R.S. Monteiro, D. Zemlyanov, J.M. Storey, F.H. Ribeiro, J. Catal. 199
2001) 291.
15] R.S. Monteiro, D. Zemlyanov, J.M. Storey, F.H. Ribeiro, J. Catal. 201
2001) 37.
16] G. Zheng, E.I. Altman, Surf. Sci. 462 (2000) 151.
(
[
[
(
[17] J.W. He, P.R. Norton, Surf. Sci. 204 (1988) 26.
[18] J. Au-Yeung, A.T. Bell, E. Iglesia, J. Catal. 185 (1999) 213.
[19] J. Han, D. Zemlyanov, G. Zhu, F.H. Ribeiro, in preparation.
[20] E.H. Voogt, A.J.M. Mens, O.L.J. Gijzeman, J.W. Geus, Surf. Sci. 373
(1997) 210.
[21] P.C. Aben, J. Catal. 10 (1968) 224.
[22] J. Shu, B.P.A. Grandjean, A. Van Neste, S. Kaliaguine, Can. J. Chem.
Eng. 69 (1991) 1036.
[
23] T. Bertrams, A. Brodde, H. Neddermeyer, J. Vac. Sci. Technol. B 12
1994) 2122.
(
[24] C. Hagendorf, R. Shantyr, K. Meinel, K.M. Schindler, H. Nedder-
meyer, Surf. Sci. 532–535 (2003) 346.
The H2 reduction of preoxidized Pd(111) and Pd(110)
crystals resulted in the decrease of surface area of the oxi-
dized Pd(111) and Pd(110) crystals. An amorphous metallic
surface was found after reduction at 373 K, whereas an al-
most smooth surface with area of only 1.1 times that of clean
metallic surface was observed after 673 K hydrogen reduc-
tion. Characteristic single crystal features were observed by
LEED after 673 K reduction in H2.
[25] C.Q. Sun, Surf. Rev. Lett. 7 (2000) 347.
[26] C.A. Brown, P.D. Charles, W.A. Johnsen, S. Chesters, Wear 161
(1993) 61.
[27] G. Zheng, E.I. Altman, Surf. Sci. 504 (2002) 253.
[28] E. Lundgren, G. Kresse, C. Klein, M. Borg, J.N. Andersen, M. De San-
tis, Y. Gauthier, C. Konvicka, M. Schmid, P. Varga, Phys. Rev. Lett. 88
(2002) 246103/1.
[29] B. Klotzer, K. Hayek, C. Konvicka, E. Lundgren, P. Varga, Surf.
Sci. 482–485 (2001) 237.
[30] B.A. Banse, B.E. Koel, Surf. Sci. 232 (1990) 275.
In conclusion, the surface area of Pd(111) and Pd(110)
single crystals changes after they are subjected to O2 oxi-
dation, methane combustion, or H2 reduction. The surface
area change after oxidation on single crystals agrees with
the ones on supported catalysts and it us due to roughening
of the PdO surface. These results imply that pretreatments
[31] H. Niehhus, C. Achete, Surf. Sci. 369 (1996) 9.
[32] M. Nishijima, M. Jo, Y. Kuwahara, M. Onchi, Solid State Commun. 60
(1986) 257.
[
33] F.P. Leisenberger, G. Koller, M. Sock, S. Surnev, M.G. Ramsey, F.P.
Netzer, B. Klotzer, K. Hayek, Surf. Sci. 445 (2000) 380.
34] D.T. Vu, K.A.R. Mitchell, O.L. Warren, P.A. Thiel, Surf. Sci. 318
(1994) 129.
[

J. Han et al. / Journal of Catalysis 225 (2004) 7–15
15
[
[
[
35] T.W. Orent, S.D. Bader, Surf. Sci. 115 (1982) 323.
[38] H. Conrad, G. Ertl, J. Kueppers, E.E. Latta, Surf. Sci. 65 (1977) 245.
36] V.A. Bondzie, P. Kleban, D.J. Dwyer, Surf. Sci. 347 (1996) 319.
37] J. Goschnick, M. Wolf, M. Grunze, W.N. Unertl, J.H. Block, J. Lobo-
da-Cackovic, Surf. Sci. 178 (1986) 831.
[39] G. Ertl, P. Rau, Surf. Sci. 15 (1969) 443.
[40] A.K. Datye, J. Catal. 216 (2003) 144.
[41] D.J. Suh, T.-J. Park, J. Catal. 149 (1994) 486.