CO, NO, and H2 adsorption on ceria-supported Pd

Article abstract of DOI:10.1006/jcat.1996.0070

We have examined the adsorption of CO, NO, and H₂ on model Pd/ceria catalysts, formed by vapor deposition of Pd metal onto a flat, polycrystalline ceria support, using temperature programmed desorption. For CO adsorption on freshly deposited Pd films, a substantial fraction of the CO reacts to CO₂ in TPD, between 450 and 550 K, indicating that oxygen from ceria can react with CO on Pd. Prereduction of the catalyst by heating in CO or H₂ results in TPD curves which are virtually identical to that observed for Pd/α-Al₂O₃(0001), suggesting that electronic interactions between ceria and Pd have a negligible effect on the adsorption properties. For NO, there are differences between Pd/ceria and Pd/α-Al₂O₃(0001). A significantly higher fraction of the NO on Pd/ceria reacts to N₂ and N₂O, and prereduction of the Pd/ceria increases the amount of NO that can adsorb. Both of these results are interpreted as indicating that NO can adsorb on reduced ceria sites which are formed when lattice oxygen at the Pd-ceria interface is transferred to the Pd. For H₂, the desorption curves are similar to that observed on Pd/α-Al₂O₃(0001), except that the low-temperature feature assigned to the bulk hydride is less well resolved. If water is formed upon H₂ adsorption, it apparently readsorbs on the ceria. The implications of these results for reactions involving NO and CO on Pd/ceria are discussed.

Full text of DOI:10.1006/jcat.1996.0070

JOURNAL OF CATALYSIS 159, 112–118 (1996)
ARTICLE NO. 0070
CO, NO, and H2 Adsorption on Ceria-Supported Pd
H. Cordatos and R. J. Gorte
Department of Chemical Engineering, University of Pennsylvania, Philadelphia, Pennsylvania 19104
Received August 18, 1995; revised October 30, 1995; accepted October 31, 1995
In order to address the issue of possible differences be-
tween ceria-supported Pt, Rh, and Pd, we compare here
the adsorption properties for CO, NO, and H2 on ceria-
supported Pd, Rh, and Pt using a series of model catalysts
prepared under UHV conditions. Previous studies have
demonstratedthatadsorptionpropertiesonmodelcatalysts
of this type can be related to catalytic rates measured under
steady-state conditions. For example, changes with particle
size in the desorption kinetics of NO from Rh on model cat-
alysts (6) can be related to changes in the steady-state NO
reduction rates observed on high-surface-area catalysts (7).
Similarly, changes in desorption kinetics of CO from Pt with
particle size could be related to observations of steady-state
COoxidationrates(8, 9). Finally, TPDstudiesofmodelRh–
ceria catalysts, prepared using similar conditions to those in
the present study, exhibited a ceria-mediated CO oxidation
process (3, 10) which is also observable on high-surface-
area catalysts (11).
Our present results demonstrate that CO adsorbed on
Pd can react with oxygen from the ceria framework to form
CO2 in the same manner reported for Rh on ceria. With this
exception, the desorption features for CO are otherwise
very similar to that found for Pd on ꢀ-Al2O3(0001) (12),
suggesting that the Pd itself is not strongly affected. For
NO, we observe significantly more dissociation than was
found for Pd/ꢀ-Al2O3(0001). Also, for reduced ceria, some
of the NO adsorbs and dissociates on the ceria itself, as
demonstrated by the significant increase in the adsorption
uptakes. The implications of these results in automotive
catalysts will be discussed.
We have examined the adsorption of CO, NO, and H2 on model
Pd/ceria catalysts, formed by vapor deposition of Pd metal onto a
flat, polycrystalline ceria support, using temperature programmed
desorption. For CO adsorption on freshly deposited Pd films, a sub-
stantial fraction of the CO reacts to CO2 in TPD, between 450 and
550 K, indicating that oxygen from ceria can react with CO on
Pd. Prereduction of the catalyst by heating in CO or H2 results in
TPD curves which are virtually identical to that observed for Pd/ꢀ-
Al2O3(0001), suggesting that electronic interactions between ceria
and Pd have a negligible effect on the adsorption properties. For
NO, there are differences between Pd/ceria and Pd/ꢀ-Al2O3(0001).
A significantly higher fraction of the NO on Pd/ceria reacts to N2
and N2O, and prereduction of the Pd/ceria increases the amount of
NO that can adsorb. Both of these results are interpreted as indi-
cating that NO can adsorb on reduced ceria sites which are formed
when lattice oxygen at the Pd-ceria interface is transferred to the Pd.
For H2, the desorption curves are similar to that observed on Pd/ꢀ-
Al2O3(0001), except that the low-temperature feature assigned to
the bulk hydride is less well resolved. If water is formed upon H2
adsorption, it apparently readsorbs on the ceria. The implications
of these results for reactions involving NO and CO on Pd/ceria are
c
discussed. ꢀ 1996 Academic Press, Inc.
INTRODUCTION
Ceria-supported group VIII metals have received con-
siderable attention in recent years, mainly due to their ap-
plications in automotive emissions-control catalysis. Ceria
is added to automotive catalysts primarily for its oxygen-
storage properties, which result from this oxide’s ability to
cycle between CeO2 and CeO2ꢀx. However, many questions
remain about the nature of interactions between ceria and
catalytic metals and there are indications that Pt, Rh, and
Pd do not all interact in the same manner with ceria. For
example, in at least one catalyst formulation reported in the
patent literature (1), it appears that an attempt was made
to limit contact between Rh and ceria and maintain con-
tact between Pt and ceria. Furthermore, Rh forms stable
perovskites, while Pt and Pd do not (2). Differences were
also reported from our laboratory on the oxygen transfer
between ceria and supported Pt and Rh (3, 4), although
more recent results have shown that this difference could
have been due to the ceria structure (5).
EXPERIMENTAL PROCEDURE
Since the pretreatment and preparation conditions used
for ceria can strongly affect the final properties (5), all of
the experiments were performed on the same substrate.
This substrate was prepared by spray pyrolysis of a 0.3-
N aqueous solution of Ce(NO3)3 (99.5% purity, Johnson
Matthey) on an ꢀ-Al2O3(0001) single crystal, using an N2
carrier gas, while holding the sapphire crystal at ꢁ650 K.
After ceria deposition, the sample was placed in an oven
and held at ꢁ900 K, in air for 2 hr. This procedure was
repeated three times to produce a final ceria film ꢁ20 ꢁm
112
0021-9517/96 $18.00
c
Copyright ꢂ 1996 by Academic Press, Inc.
All rights of reproduction in any form reserved.

CO, NO, AND H2 ADSORPTION ON CERIA-SUPPORTED Pd
113
thick. Finally, the sample was heated overnight at 970 K, a
For CO and 15NO, adsorption was performed at room
temperature at least 250 K higher than the decomposition temperature, while H2 was adsorbed at 120 K. Typical ex-
temperature of Ce(NO3)3 (13).
posures were approximately 10 L (1 Langmuir = 10ꢀ6 Torr
A complete description of the films is given elsewhere ꢅ sec = 3.7 ꢄ 1014 molecules/cm2 for NO at 300 K), which
(5). Briefly, X-ray diffraction patterns of films prepared us- were sufficient to achieve saturation. For all TPD measure-
ing the same procedure demonstrated that the sample had ments, the heating rate was 10 K/sec. After each NO TPD,
the CeO2 fluorite structure. Using Scherrer’s formula for oxygen resulting from dissociation was removed by expos-
the line-broadening of the peak at 28.4ꢃ, we estimated the ing the sample to CO and heating to 700 K. The desorption
ceria crystallite size to be ꢁ9–12 nm. Scanning electron mi- rates, rd, in each figure are in arbitrary units, but all curves
crographs showed that the films were relatively flat, with are drawn to the same scale.
cracks or fissures separating irregularly shaped islands of
ꢁ1ꢁm in size. The surface area was found from BET mea-
RESULTS
surements to be 12 m2/g. If one assumes that the ceria con-
sists of dense, spherical particles, this surface area would
be obtained for particles 0.1 ꢁm in diameter, which implies
there must be some porosity in the islands.
CO Adsorption on Pd/CeO2
The TPD curve following a saturation exposure of the
as-deposited, fresh, 5 ꢄ 1015Pd/cm2 film to CO is shown in
Fig. 1a. It is very similar to curves obtained from Pd parti-
cles on ꢀ-Al2O3(0001) (12), with the exception that a sig-
nificant amount, ꢁ20%, of the CO reacts to CO2 between
450 and 550 K. The CO peak exhibits a single maximum at
490 K and a low-temperature shoulder at ꢁ370 K, similar
to what is observed for CO from Pd single crystals (14).
These desorption features have previously been assigned
to CO which is bridge bonded and terminally bonded, re-
spectively (15). The source of the oxygen needed to form
CO2 appears to be the ceria. The Pd was not exposed to
The UHV chamber used for the TPD measurements
has a base pressure of ꢁ4 ꢄ 10ꢀ10 Torr. It is equipped
with a cylindrical mirror analyzer for Auger electron spec-
troscopy (AES) and a quadrupole mass spectrometer, in-
terfaced with a computer for data acquisition and multi-
plexing. To improve sensitivity and baselines of the TPD,
the quadrupole was placed inside a stainless-steel cone with
an aperture, in front of which the sample could be placed.
The oxide substrate was mounted on a Ta foil that could be
cooled to 120 K or heated resistively to 900 K. A chromel-
alumel thermocouple, attached to the back side of the ꢀ-
Al2O3 crystal via a ceramic adhesive, was used to moni-
tor the temperature. The adsorbates were introduced to
the chamber through two beam dosers in order to main-
tain a low base pressure. The pressure in front of the doser
has been estimated to be approximately 20 times higher
than the background pressure, and exposures quoted in this
manuscript were determined using this factor.
The ceria substrate was cleaned by ion bombardment at
500 K to remove all impurities observable by AES, followed
by annealing in 1 ꢄ 10ꢀ7 Torr O2 at 900 K for 30 min. The
final oxidation-step involved cooling the sample to 400 K
in 1 ꢄ 10ꢀ7 Torr O2. Following evacuation, the noble met-
als were deposited at room temperature. The model cata-
lysts at this stage will be referred to as “fresh.” The AES
peak intensities for Pd as a function of metal coverage were
very similar to those reported for Pd/ꢀ-Al2O3(0001), sug-
gesting that the growth of the metal film was nearly two-
dimensional (12). Simply heating the samples above 650 K
resulted in an irreversible, decreased peak intensity for Pd
in AES, along with an increased intensity for oxygen, in-
dicating that the metal film coalesced into particles at this
temperature. Heating the samples in 1ꢄ10ꢀ7 Torr CO or H2
for 10 min at 650 K, followed by briefly heating to 800 K in
vacuum, induced changes in the desorption curves. Samples
pretreated in this manner will be referred to as “reduced”.
For most of the experiments in this study, CO was used as
the reductanct.
FIG. 1. Saturation TPD curves following CO adsorption at room tem-
perature on a film of 5 ꢄ 1015 Pd/cm2 (a) as deposited on ceria (fresh) and
(b) after mild reduction in CO (reduced).

114
CORDATOS AND GORTE
oxygen prior to the adsorption of CO; furthermore, if the the TPD curves for NO on Pd/ꢀ-Al2O3(0001) were virtu-
CO2 was the product of coadsorbed oxygen, it would ap- ally identical to those obtained for Pd on amorphous silica
pear in a peak centered at a lower temperature (16). We (19), we suggest that the sapphire support is noninteract-
found no evidence for carbon deposition from either AES ing and that the results on this surface simply represent the
or subsequent TPD curves; therefore, the oxygen is not the desorption of NO on small Pd particles. Figures 2 and 3
result of CO dissociation. Rather, the results for Pd/ceria show the desorption results for NO from the fresh and re-
are very similar to those reported for Rh/ceria (3), where duced Pd films of 5 and 1 ꢄ 1015 Pd/cm2, respectively, with
reduction of the ceria is the source of the oxygen.
the analogous results for the same metal coverages on the
After heating the film to 700 K during this first TPD sapphire crystal in the upper right corner. The results on ꢀ-
measurement, the Pd film coalesced into larger particles, as Al2O3(0001) were obtained after heating the films to 700 K
shown by AES. Subsequent TPD curves for CO were very to form particles. To remove oxygen formed by NO disso-
similar to that shown in Fig. 1a. Except that the amount ciation, CO was adsorbed and the sample flashed to 700 K
of CO2 decreased monotonically, similar to what has been between NO TPD measurements. 15NO was used in all of
reported for Rh/ceria (3). After the film had coalesced into the experiments to eliminate any ambiguity in determining
particles, the sum of the CO and CO2 peaks remained ap- the products.
proximately constant, indicating that the size of the parti-
For 5 ꢄ 1015 Pd/cm2, most of the NO desorbs unreacted
cles did not change substantially following the first anneal- for the sapphire substrate, as shown in Fig. 2. For Pd/ceria,
ing cycle. Using the sum of he CO and CO2 areas in TPD however, only a small fraction of the NO desorbs intact
to determine the dispersion and to calculate a particle size, on both the fresh and reduced surfaces. The TPD curves
as described elsewhere (17), we estimate the average Pd
particle diameter to be ꢁ9 nm. The CO TPD curve on the
reduced sample, obtained after heating it in CO at 650 K,
is shown in Fig. 1b. There is no significant change in the
desorption features for CO, but the amount of CO2 formed
is now negligible. Annealing to 950 K in vacuum restored
some of the ability of the surface to oxidize CO, as reported
previously for Rh/ceria (3), indicating that oxygen could be
brought out from the bulk of the ceria.
In order to investigate the effect of Pd particle size, the
same adsorption measurements were performed on a film
with1ꢄ1015 Pd/cm2. TheTPDcurvesarenotshownbecause
they are so similar to those obtained for 5 ꢄ 1015 Pd/cm2.
The minor differences were as follows: First, the fraction of
the CO which reacted to CO2 in TPD was higher and CO2
formation appeared to start at slightly lower temperatures.
For example, the first TPD curve on the fresh film gave a
CO2 peak area which was 30% of the CO peak area. The
increased fraction of CO reacting to CO2 is consistent with
oxygen originating at the edge of the metal crystallites, as
suggested in recent reaction measurements (18). Second,
the particle size determined after annealing this sample to
700 K, again determined by CO adsorption capacity, was
found to be 5.2 nm, which is smaller than that obtained for
larger metal coverages and consistent with that found for
similar metal coverages on ꢀ-Al2O3(0001) (12). There is no
evidence for CO adsorption on either oxidized or reduced
ceria. Finally, a higher fraction of the CO on the reduced
sample desorbed in the 370 K peak for the lower Pd cover-
age, consistent with the smaller Pd particle size.
NO Adsorption on Pd/CeO2
FIG. 2. TPD curves following NO adsorption at room temperature on
In order to examine the role of ceria in the adsorption
and dissociation of NO on Pd, we compare the TPD curves
(a) a fresh and (b) a reduced film of 5 ꢄ 1015 Pd/cm2 on ceria. The NO TPD
curves obtained from a Pd/ꢀ-Al2O3 film with the same Pd loading, taken
for Pd/ceria with those for Pd/ꢀ-Al2O3(0001) (12). Since from reference (12), are shown in the insert at the top right of the figure.

CO, NO, AND H2 ADSORPTION ON CERIA-SUPPORTED Pd
115
ported for lanthana (20). First, we observed negligible ad-
sorption on the ceria without Pd, demonstrating that the ox-
idized ceria cannot adsorb NO. The evidence that NO can
adsorb on the reduced ceria comes from the total amounts
of NO which adsorb. A reasonable estimate of the satu-
ration coverages for NO can be obtained by summing the
areas under the NO, N2, and N2O features and comparing
thesetothesaturationCOcoverage. ForPd/ꢀ-Al2O3(0001),
this ratio was 0.7 (12). On the fresh and reduced Pd/ceria
catalysts, these ratios were 0.8 and 3.5, respectively. Ob-
viously, the large amount of NO on the reduced sample
suggests that adsorption occurs on the ceria. NO reoxidizes
CeO2ꢀx and desorbs as N2. Adsorption on reduced ceria
may also explain why more NO decomposes in TPD on
the fresh sample. If Pd is oxidized and ceria reduced, either
at room temperature or during the temperature ramp, NO
could adsorb on the ceria. Coadsorption of NO and oxygen
on Pd should certainly decrease the NO decomposition rate
(21) and the CO results suggest that the Pd properties are
not significantly changed, so that reaction on reduced ce-
ria is the most reasonable explanation of the increased NO
decomposition. The similarities in the desorption temper-
atures for both fresh and reduced Pd/ceria further agrees
with this interpretation.
The TPD curves for 1 ꢄ 1015 Pd/cm2 are qualitatively
similar, as shown in Fig. 3. We again find that most of the
NO dissociates to desorb as N2 at the same temperatures
observed for the higher Pd coverage, only the amounts are
much less. Reducing the catalyst again increases the ad-
sorption uptake. The sum of the peak areas for NO, N2,
and N2O, ratioed to the CO and CO2 areas, were 1.0 for
Pd/ꢀ-Al2O3(0001), 0.9 for fresh Pd/ceria, and 2.0 for re-
duced Pd/ceria. Since the N2 desorption features overlap
with those observed on Pd/ꢀ-Al2O3(0001) at this Pd cover-
age, desorption from Pd could contribute to the features.
We also observe a trace of N2O on the Pd/ceria, which was
not observed for Pd/sapphire, suggesting that the N2O is
formed on the oxide.
FIG. 3. TPD curves following NO adsorption at room temperature on
(a) a fresh and (b) a reduced 1 ꢄ 1015 Pd/cm2 film on ceria. The NO TPD
curves obtained from a Pd/ꢀ-Al2O3 film of the same film loading, taken
from reference (12), are shown in the insert at the top right of the figure.
on Pd/ceria are qualitatively similar, independent of pre-
treatment, showing unreacted NO in a peak centered at
510 K, with a leading shoulder extending to 350 K. The
main desorption product is N2, which desorbs in a narrow
feature at 510 K and a second, overlapping, broad feature
H2 Adsorption on Pd/CeO2
The TPD curves obtained following a saturation expo-
at ꢁ590 K. A small amount of N2O is also observed in peaks sure of H2 at 120 K to fresh Pd/ceria catalysts with Pd
at 390 and 510 K. The effect of reducing the Pd/ceria was coverages of 5 ꢄ 1015 and 1 ꢄ 1015 Pd/cm2 are given in
to dramatically increase the amount of N2 formed and to Fig. 4, along with the corresponding TPD curve for 5 ꢄ 1015
increase slightly the N2O, while not changing the molecu- Pd/cm2 on ꢀ-Al2O3(0001). For Pd/ꢀ-Al2O3(0001), one ob-
lar NO peak. The simultaneous desorption of NO, N2 and serves a well-defined peak below 200 K for the larger parti-
N2O at 510 to 550 K on both Pd/ceria and Pd/alumina sug- cles, which is attributable to subsurface H2 (12). This feature
gests that some surface processes remain unchanged, but is not observed to any great extent on smaller Pd particles
the relative amounts of each product are quite different.
on ꢀ-Al2O3(0001). Surface hydrogen on Pd/ꢀ-Al2O3(0001)
The two important questions are (i) how does reducing desorbs in a broad feature between 250 and 400 K, essen-
the catalyst change the adsorption properties and (ii) how tially independent of particle size. For the Pd/ceria samples,
does ceria affect the decomposition of NO. We believe the onlymolecularH2, nowater, desorbedintwobroadpeaksat
answer to both these questions is that NO adsorbs on the 200and320K. ExceptfortheamountofH2 whichdesorbed,
reduced ceria, which is similar to what was previously re- both TPD profiles were essentially the same. While H2 is

116
CORDATOS AND GORTE
reported that the properties of Pt and Rh particles on ce-
ria were different in that CO on Rh/ceria resulted in CO2
formation, while very little CO2 was observed for Pt/ceria
(3, 4). More recently, it was shown that Rh particles on
CeO2(111) and ceria films annealed at high temperatures
do not exhibit CO2 formation in TPD (22). The pretreat-
ment conditions used for the ceria, therefore, appear to
strongly affect the final properties of the catalyst. To do
a fair comparison of Pt, Pd, and Rh on ceria, therefore,
we examined the desorption of CO from each of the three
metals deposited on the same ceria substrate, sputtering the
sample clean between the deposition of each metal. The re-
sults for fresh Rh/ceria and Pt/ceria, with metal coverages
of 1.5 ꢄ 1015/cm2, are shown in Figs. 5a and 5b.
The results for Rh/ceria are virtually identical to TPD
curves reported earlier (3). CO desorbs at 490 K, with a
shoulder to lower temperatures, similar to that observed
on Rh single crystals. Substantial amounts of CO2 are also
formed, consistent with what would be expected for coad-
sorbed CO and oxygen. Contrary to previous reports, the
results for Pt, Fig. 5b, also show significant CO2 formation
and could be consistent with coadsorbed CO and oxygen.
For both Pt and Rh, reduction of the surfaces by heating
in either CO or H2 gave samples which exhibited CO TPD
curves similar to that for Pt or Rh on alumina (23), with no
evidence for CO2 formation.
FIG. 4. TPD curves following H2 adsorption at 120 K on pure ceria as
The previous results for Pt/ceria appear to be the result
of high-temperature annealing of the ceria, which has
well as for two different loadings of Pd on ceria (1 and 5 ꢄ 1015 Pd/cm2).
The TPD curve obtained from a 5 ꢄ 1015 Pd/cm2 loading on ꢀ-Al2O3,
taken from reference (12), is shown in the insert at the top right of the
figure.
seen below 250 K, which might indicate subsurface species,
the desorption feature is broad, unlike the alumina case.
The most interesting questionregardingH2 fromPd/ceria
is why is water not observed in TPD. If the explanations for
CO and NO adsorption on these catalysts are correct, that
the Pd is oxidized by contact with the ceria, we should form
water from the coadsorbed hydrogen and oxygen. As we
discussed earlier, heating the samples in 1ꢄ10ꢀ7 Torr of H2
for 10 min at 500 K was sufficient to reduce the sample so
that adsorption of CO showed only CO in TPD. Therefore,
wesuggestthatwatersimplyreadsorbsontheceriaandthen
either desorbs in a such a broad feature that it is lost in the
baseline or is absorbed into the ceria bulk. (The presence of
hydrogen in ceria has been discussed by others (28).) The
conclusion that water is simply not observed in our TPD
measurement was varified in coadsorption measurements
with O2 and H2, in which the TPD curves were similar to
those reported here and no water was seen.
Comparisons with Rh/CeO2 and Pt/CeO2
It is interesting to consider how specific the interactions
are between ceria and the catalytic metal. Previously, it was
FIG. 5. TPD curves following CO adsorption at room temperature of
a fresh film of (a) 1.5 ꢄ 1015 Rh/cm2 and (b) 1.5 ꢄ 1015 Pt/cm2 on ceria.

CO, NO, AND H2 ADSORPTION ON CERIA-SUPPORTED Pd
117
already been shown to cause deactivation (5). The NO tion temperatures are a sensitive measure of the adsorption
desorption results reported in this earlier study of Pt/ceria properties of the catalyst, changes in the metal properties
were identical to results for Pt/alumina (24). If it were should be reflected in the TPD curves. There is also no evi-
simply a question of the ceria having been reduced, that dence for ceria migration onto the metals, such as has been
study should have observed enhanced N2 formation and reported for titania-supported metals (27).
new desorption features, similar to what was observed in
The picture of ceria support effects that emerges, then, is
the present study on Pd/ceria. This further points out the one in which oxygen and possibly NO move relatively freely
importance of ceria structure in the adsorption properties between the ceria and the catalytic metal. Lattice oxygen
of ceria-supported metals.
from CeO2 at the metal–ceria boundary provides oxygen to
the metal for oxidation reactions, while reduced sites at the
boundary provide adsorption sites for NO. These reduced
sites can be oxidized by NO, giving up N2 and completing
the oxidation–reduction cycle.
DISCUSSION
The central questions addressed in this paper are as fol-
lows: (i) What is the nature of support interactions between
ceriaandgroupVIIImetals?(ii)Howdotheadsorptionand
reaction properties of group VIII metals supported on ceria
differ from catalysts formed on alumina supports? Each of
these issues will be addressed in turn.
Regarding the adsorption and reaction properties of
ceria-supported Pd, the TPD results show that there are
significant changes due to the presence of ceria. For CO, ob-
viously, the reaction with lattice oxygen from the ceria pro-
vides a second pathway for CO oxidation. Previous work
with Rh/ceria has shown that this second pathway exhibits
different reaction orders for both CO and O2 from that ob-
served for Rh/alumina under reducing conditions and also
exhibits a lower activation energy (18). Therefore, the oxi-
dation process observed in TPD influences the steady-state
reaction rates, at least under some conditions. The results
reported here for NO suggest that reactions of this molecule
will also be strongly affected by the presence of ceria un-
der net reducing conditions. The relatively free movement
of molecules between sites on the catalytic metal and on
the ceria provides a means for enhancing NO dissociation.
This could be very important for metals like Pt and Pd which
are not active for NO dissociation. Therefore, based on our
TPD results, one would predict that the presence of ceria
could significantly enhance NO reduction. Comparisons of
rates for NO reduction by CO on Pd/alumina and Pd/ceria
are presently being performed in our laboratory to test this
prediction.
First, the ability of CO on Pd, Pt, and Rh to react with lat-
tice oxygen from ceria suggests that all three catalytic met-
als exhibit similar interactions with ceria. Since each of the
metals have different stable oxidation states and form dif-
ferent compounds, interfacial compound formation would
not appear to be necessary for the reaction to occur. It is
possible that oxidation of CO occurs through a complex
at the boundary between the metals and ceria, and evi-
dence for these boundary states has been presented from
IR spectra on lanthana-promoted Rh/SiO2 (25). Since H2 is
equally effective at reducing ceria in our studies, a similar
state would have to be invoked for hydrogen. However, it is
our opinion that the best explanation for our results is that
the metals are oxidized by the reduction of ceria since the
TPD curves for CO are so similar to those observed when
CO and oxygen are coadsorbed. While the redox process,
M + CeO2 → MOx + Ce2O3, is not thermodynamically
favored for these catalytic metals based on bulk thermody-
namicproperties(3), thereareanumberofreasonsthether-
modynamic data for bulk materials may not be meaningful.
First, it is known that CeO2 can be reduced by as much
as 17% without changing from the fluorite structure to the
hexagonal Ce2O3 structure (29). Second, the observed sen-
sitivity for CO2 formation from Rh/ceria on ceria structure
suggests that the properties of ceria in these nonannealed
films are not the same as for bulk structures (5). Indeed,
there is precedence for changes in the free energies of ox-
ides like ceria with particle size in that high-pressure forms
of ZrO2 and Y2O3 can be stabilized at low pressures by
forming small crystallites (26).
SUMMARY
The desorption measurements in this study indicate that
the effect of ceria on Pd is very similar to that found for
Rh and Pt. There is no evidence for electronic interactions
which modify the adsorption properties of the metal, but
it appears that lattice oxygen from the ceria can be trans-
ferred to the metal surface. This results in CO2 formation at
relatively low temperatures upon the adsorption of CO. The
reduction of ceria by oxygen transfer also frees sites for NO
adsorption, and NO decomposes readily on reduced ceria.
This may provide an additional reaction pathway for NO
decomposition.
Electronic effects are often invoked to explain support
effects, but these do not appear to be important for ceria.
Following reduction of the ceria, which should maximize in-
teractions based on most theories of support interactions,
the desorption curves for CO from Pd/ceria are identical
to those found for Pd on an inert substrate, ꢀ-Al2O3(0001)
(12). Similar results were found for Rh and Pt. Since desorp-
ACKNOWLEDGMENTS
This work was supported by the NSF, MRL Program, Grant DMR 88-
19885. Supplies and supplimentary support were obtained from the DOE,
Basic Energy Sciences, Grant DE-FG03-85-13350.

118
CORDATOS AND GORTE
REFERENCES
15. Van Hardeveld, R., and Hartog, F., Adv. Catal. 22, 75 (1972).
16. Stuve, E. M., Madix, R. J., and Brundle, C. R., Surf. Sci. 146, 155 (1984).
1. Joyner, R. W., “Zirconium in Catalysis: a Review of Current Scientific 17. Roberts, S., and Gorte, R. J., J. Chem. Phys. 93, 5337 (1990).
Literature” prepared for Magnesium Electron Ltd. 18. Bunluesin, T., Cordatos, H., and Gorte, R. J., J. Catal. 157, 222 (1995).
2. Tejuca, L. G., Fierro, J. L. G., and Tascon, J. M. D., Adv. Catal. 36, 237 19. Xu, X., and Goodman, W. D., Catal. Lett. 24., 31 (1994).
(1989).
20. Cordatos, H., and Gorte, R. J., Appl. Catal. in press.
21. Schmick, H. D., and Wassmuth, H. W., Surf. Sci. 123, 471 (1982).
22. Stubenrauch, J., and Vohs, J. M., J. Catal. in press.
3. Zafiris, G. S., and Gorte, R. J., J. Catal. 139, 561 (1993).
4. Zafiris, G. S., and Gorte, R. J., Surf. Sci. 276, 86 (1992).
5. Cordatos, H., Bunluesin, T., Stubenrauch, J., Vohs, J. M., and Gorte, 23. Altman, E. I., and Gorte, R. J., Surf. Sci. 195, 392 (1988).
R. J., J. Phys. Chem. in press.
24. Altman, E. I., and Gorte, R. J., J. Phys. Chem. 93, 1993 (1989).
25. Underwood, R. P., and Bell, A. T., J. Catal. 109, 61 (1988).
26. Scandan, G., Foster, C. M., Frase, H., Ali, M. N., Parker, J. C., and
Hahn, H., Nanostructured Mater. 1, 313 (1992).
6. Altman, E. I., and Gorte, R. J., J. Catal. 113, 185 (1988).
7. Oh, S. H., and Eickel, C. C., J. Catal. 128, 526 (1991).
8. Altman, E. I., and Gorte, R. J., Surf. Sci. 172, 71 (1986).
9. Zafiris, G. S., and Gorte, R. J., J. Catal. 140, 418 (1993).
10. Zafiris, G. S., and Gorte, R. J., J. Catal. 143, 86 (1993).
11. Oh, S. H., and Eickel, C. C., J. Catal. 112, 543 (1988).
12. Cordatos, H., Bunluesin, T., and Gorte, R. J., Surf. Sci. 323, 219 (1995).
13. Wendlandt, W. W., Anal. Chim. Acta 15, 435 (1956).
14. Engel, T., and Ertl, G., Adv. Catal. 28, 1 (1979).
27. Ko, C. S., and Gorte, R. J., J. Catal. 90, 59 (1984).
28. Laachir, A., Perrichon, V., Badri, A., Lamotte, J., Catherine, E., Laval-
ley, J. C., El Fallah, J., Hilaire, L., le Normand, F., Quemere, E.,
Sauvion, G. N., and Touret, O., J. Chem. Soc. Faraday Trans. 87, 1601
(1991).
29. Sanchez, M. G., and Gazquez, J. L., J. Catal. 104, 120 (1987).