Effects of Pd particle size and ceria loading on NO reduction with CO

Article abstract of DOI:10.1006/jcat.2000.2985

There has been significant interest in using Pd-only catalysts for automobile three-way exhaust gas conversion. Three-way converters are designed to reduce emissions of CO, NO_x, and uncombusted hydrocarbons. These catalysts often contain noble metals, e.g., Pd, Pt, and Rh, with a ceria promoter supported on alumina. In view of this, the kinetics of the NO + CO reaction at 433-573 K were studied over supported palladium with and without ceria promotion. For the ceria promoted samples, the interface between ceria and palladium was changed by varying particle size and the quantity of ceria postdeposited on the Pd particles. The Pd particle size affected activity and apparent reaction orders. For the unpromoted catalyst, the Pd particle size of ~ 4-5 nm average dia had a lower turnover frequency than 1-nm size particles, indicating structure sensitivity of the NO + CO reaction. However, the ceria-promoted samples displayed a maximum in catalytic activity for 2-nm palladium. This maximum in reaction rate indicated a balance between the Pd/ceria interface that promotes NO dissociation and the Pd metal surface that catalyzes product formation. TPD results on adsorbed NO were consistent with this interpretation.

Full text of DOI:10.1006/jcat.2000.2985

Journal of Catalysis 195, 193–206 (2000)
doi:10.1006/jcat.2000.2985, available online at http://www.idealibrary.com on
Effects of Pd Particle Size and Ceria Loading on NO Reduction with CO
�
�
,1
Joseph H. Holles, Robert J. Davis, Thomas M. Murray,† and James M. Howe†
�
Department of Chemical Engineering and †Department of Materials Science and Engineering,
University of Virginia, Charlottesville, Virginia 22904-4741
Received April 21, 2000; revised June 28, 2000; accepted June 29, 2000
its use as a promoter for Pd has not been as widely studied
The kinetics of the NO + CO reaction from 433 to 573 K were (6, 8).
studied over supported Pd with and without ceria promotion.
In addition, temperature-programmed desorption (TPD) of prod-
ucts formed from preadsorbed NO was performed. For the ceria-
promoted samples, the interface between Pd and ceria was changed
by varying particle size and the amount of ceria postdeposited on
the Pd particles. The Pd particle size affected both activity and
apparent reaction orders. For the unpromoted catalyst, the Pd par-
ticle size of about 4.5-nm average diameter, determined by dihydro-
gen chemisorption, had a lower turnover frequency than 1-nm size
particles, indicating structure sensitivity of the NO + CO reaction.
The NO + CO reaction was found to be structure sen-
sitive over Pd single crystals with the Pd(111) facet being
more active than either Pd(110) or Pd(100) (2, 9, 10). In
addition, large particles (120 nm) of Pd supported on alu-
mina were much more active than small particles (<3 nm)
(
2, 3). The Pd surface structure also affects dinitrogen
selectivity for this reaction. The less active crystal faces
tend to produce more N2 and less N2O (2). Furthermore,
temperature-programmed desorption of adsorbed NO in-
However, the ceria-promoted samples showed a maximum in cata- dicated that only N2 desorbed from the small particles
lytic activity for 2-nm Pd. This maximum in reaction rate suggests a whereas both N2 and N2O desorbed from large particles
balance between the Pd/ceria interface which promotes NO dissoci- (11). Interestingly, no appreciable particle size effect on
ation and the Pd metal surface which catalyzes product formation.
Results from TPD of adsorbed NO are consistent with this interpre-
tation. High-resolution transmission electron microscopy was used
to confirm the proximity of small ceria crystallites to the supported
rd ei na ic tt ri oo gn e (n2 )s.electivity was observed during the NO + CO
In our previous investigation of supported Pd for the
NO + CO reaction, we showed that substantial improve-
Pd particles.
�c 2000 Academic Press
ments in reaction rates are realized when ceria is added as
a promoter (6). This enhanced activity due to ceria addi-
tion is similar to that previously reported for Rh, which
was attributed to new ceria–metal interfacial sites that
promote NO dissociation (7). These results suggest that
increasing the interfacial contact between Pd and ceria
rm e aa yc t if ou nr t. her enhance catalytic activity for the NO + CO
Key Words: NO; NOx; CO; Pd (palladium); ceria; kinetics;
reduction; NO + CO reaction; TPD; TEM; chemisorption.
INTRODUCTION
Recently, there has been considerable interest in using
palladium-only catalysts for automobile three-way exhaust
gas conversion (1). Three-way converters (TWC) are de-
signed to reduce emissions of carbon monoxide, nitrogen
oxides, and uncombusted hydrocarbons. These catalysts
typically contain noble metals such as Pt, Pd, and Rh with
a ceria promoter supported on alumina. Traditionally, the
principal function of the Rh is to control emissions of ni-
trogen oxides (NOx) (1) by reaction with carbon monoxide.
However, use of Pd to control NOx emissions has increased.
As a result, the NO + CO reaction has been studied on high
surface area Pd/Al2O3 powder catalysts (2–6). Although ce-
ria effectively promotes Rh for the NO + CO reaction (7),
Therefore, the present study examines the influence
of Pd particle size and ceria loading on catalytic activ-
ity for the NO + CO reaction. Dihydrogen chemisorption,
temperature-programmed desorption (TPD) of NO, and
transmission electron microscopy were used to character-
ize the catalyst samples.
Typical three-way catalysts operate at temperatures
around 900 K and do not exhibit appreciable NOx reduc-
tion activity except in a narrow temperature window in
the vicinity of catalyst light-off (1). This is a concern dur-
ing cold start since the catalyst has not yet warmed up to
operating temperature and a significant amount of total
emissions are released during this period. Thus, the reac-
tion temperatures used in this work are less than 573 K
to simulate the state of the catalyst relevant to cold-start
conditions.
1
To whom correspondence should be addressed. Department ofChem-
ical Engineering, Chemical Engineering Bldg. Rm. #117A, University of
Virginia, 102 Engineers Way, Charlottesville, VA 22904-4741. Fax: (804)
9
82-2658. E-mail: rjd4f@virginia.edu.
193
0021-9517/00 $35.00
Copyright �c 2000 by Academic Press
All rights of reproduction in any form reserved.

1
94
HOLLES ET AL.
EXPERIMENTAL
The Pd/Al2O3 samples were prepared by slurrying � -
tions of N2 (mass 28), NO (mass 30), O2 (mass 32), and
N2O (mass44) were monitored throughout the temperature
ramp with the mass spectrometer. The amount of each gas
desorbed was quantified by comparing the area under the
TPD curve to peak areas of N2, NO, and O2 (99.993% BOC
Gases) and N2O (99.97% , BOC gases) calibration pulses
injected after each experiment. The calibration pulses were
also used to determine the contribution of N2O cracking to
mass 30 (NO) and mass 28 (N2).
Al2O3 (Alfa Aesar, 99.97% ) and palladium(II) acetylace-
tonate (acac) (Aldrich, 99% ) in toluene for 2 h at 353 K
and drying under vacuum in a rotary evaporator followed
by 24 h in air at 473 K. Samples were then calcined in flow-
�
1
ing air (BOC gases) by heating to 673 K at 0.5 K min
and then remaining at 673 K for 4 h. Subsequent reduc-
tion of the catalyst occurred at 673 K for 2 h in flowing
dihydrogen (99.999% from BOC gases, passed through a
Matheson Model 8371V purifier). For catalysts supported
on CeOx/Al2O3, the promoter was first deposited on the
alumina by slurrying cerium(III) acetylacetonate hydrate
The percent of the Pd metal atoms exposed was de-
termined using H2 chemisorption (99.999% from BOC
Gases passed through a Supelco OMI purifier) in a Coulter
Omnisorp 100CX system. Sample pretreatment consisted
of reduction in flowing H2 at 663 K for 30 min followed
(
Aldrich) and alumina in toluene for 2 h at 353 K and then
� 5
by evacuation (10 mbar) at 673 K for 45 min to re-
drying and calcining as described above. The palladium was
then deposited on the CeOx/Al2O3 using the same proce-
dure used for the alumina-supported samples. Samples with
CeOx on Pd/Al2O3 had 1 wt% palladium deposited and re-
duced prior to addition of ceria. After incorporation of the
promoter, the catalyst was calcined and reduced again.
The reaction experiments were performed in a quartz,
single-pass, fixed bed reactor containing approximately
move sorbed hydrogen (surface and bulk). Adsorption
isothermswere obtained at 303K. The hydrogen uptake was
calculated from the difference between total chemisorp-
tion and reversible chemisorption. Elemental analysis was
performed by Galbraith Laboratories Inc., Knoxville, TN
and by Southern Testing and Research Laboratories Inc.,
Wilson, NC.
Samples were prepared for transmission electron mi-
croscopy (TEM) by placing the catalyst in ethanol and son-
icating for 0.5 to 1 h to disperse the sample. One drop of
solution was then placed on lacey carbon supported on a
5
0 mg of catalyst diluted in chromatographic silica gel
(
Fisher). Feed gases included 5.00% NO/5.07% CO/He and
He (99.999% ) from BOC Gases, CO (UHP) from Air Prod-
ucts, and NO (99% ) from MG Industries. Gas flow rates
were controlled using mass flow controllers (Brooks Series
3
00-mesh copper grid (Ted Pella, Inc.). The samples were
then dried in air and stored in a desiccator. The micrograph
images of the catalyst samples were acquired using a JEOL
�
1
5
850C) and varied from 5 to 250 ml min . Total pressure
in the reactor was atmospheric with partial pressures of the
reactants varied by changing the individual flow rates while
simultaneously adjusting the flow rate of pure He. Reac-
tant and product analyses involved a combination of gas
chromatography (HP 5890 Series II with an Alltech CTR
I column) and mass spectrometry (Dycor MA 100 model).
The standard catalyst pretreatment consisted of heating at
4
000 EX model transmission electron microscope at 400 kV,
achieving a maximum magnification of 600,000� .
RESULTS
Results from elemental analysis are presented in Table 1.
5
23 K for 1 h under flowing He. Product analysis was per- The Pd/CeOx/Al2O3 samples had palladium deposited after
formed after a period of 45 min on stream at each set of the ceria whereas the CeOx/Pd/Al2O3 samples had ceria
reaction conditions to allow the reaction to reach steady deposited after the palladium. For the samples with ceria
state. Arrhenius-type plots were determined using 5.07 kPa deposited first, the loading of cerium was on the order of
of each NO and CO for the NO + CO reaction. For the de- that needed for a theoretical monolayer (� 9 wt% Ce, based
2
termination of reaction orders, the partial pressure of NO on the Ce surface density of 4.83 Ce/nm derived from the
or CO was held constant at 4.05 kPa while the pressure of CeO2(110) plane). Very small and broad X-ray diffraction
the other gas was varied from 1.01 to 16.2 kPa.
peaks associated with CeO2 indicated that the promoter
TPD experiments were performed in an atmospheric was well dispersed on the alumina.
pressure flow system with a quadrupole mass spectrome-
Results from H2 chemisorption on the supported metals
ter (Dycor MA 100 model). A fresh catalyst sample of 1 g are included in Table 1. Representative dihydrogen adsorp-
was heated in He (99.999% from BOC Gases) at 773 K tion isotherms for 5 wt% Pd/Al2O3, 1 wt% Pd/Al2O3, and
for 3 h. The sample was then cooled to 303 K and then ex- 0.5 wt% Pd/Al2O3 are presented in Figs. 1a–1c. Adsorption
�
1
posed to 10 ml min NO (99% from MG Industries) in isotherms for the ceria-promoted samples are similar. The
1
�
1
0 ml min He for 0.25 h. The system was then purged isotherms for two of the samples exhibited a large uptake
with He at 303 K for 2 h to remove weakly adsorbed NO. of dihydrogen at about 3 kPa due to the formation of a �-
The temperature of the sample was increased linearly by phase hydride which is stable at 303 K at pressures greater
�
1
5
K min to promote desorption of products into a He than 2.3 kPa (12). The isotherms for the 0.5 wt% Pd/Al2O3
�
1
carrier stream (20 ml min ). The gas phase concentra- did not show the large uptake associated with �-hydride

PALLADIUM–CERIA CATALYSTS FOR NO + CO
195
TABLE 1
formation in Pd particles. This result indicates that the Pd
particles in this sample were too small to have significant
internal volume to form the �-phase hydride. Average par-
ticle sizes for the samples, determined using d = 0.9/fraction
Elemental Analyses and Chemisorption Results
wt%
Metala
wt%
Particle
Catalyst
Ceriuma
H/Mb
size (nm) metal exposed (13), are also shown in Table 1. Dihydrogen
spillover from the Pd to the ceria resulted in the measured
dispersion of the postdeposited ceria samples being unre-
0
1
5
.5 wt% Pd/Al2O3
wt% Pd/Al2O3
wt% Pd/Al2O3
0.51
0.96
4.92
—
0.87
4.12
10.2
0.70
0.76
0.60
—
—
—
10.83
8.66
10.66
7.03
4.89
10.2
28.0
0.79
0.59
0.18
—
1.1
1.5
4.5
—
<1
2.0
9.0
3.1
3.1
3.1
alistically high, and therefore, the dispersion was assumed
equal to the 5 wt% CeO /Pd/Al O sample.
CeOx/Al2O3
x
2
3
1
5
1
5
1
3
wt% Pd/CeOx/Al2O3
wt% Pd/CeOx/Al2O3
0 wt% Pd/CeOx/Al2O3
wt% CeOx/Pd/Al2O3
0 wt% CeOx/Pd/Al2O3
0 wt% CeOx/Pd/Al2O3
1
Transmission electron micrographs are shown in Fig. 2 for
0.46
0.10
0.29
�
-Al2O3, 10 wt% CeOx/Al2O3, 5 wt% CeOx/Pd/Al2O3, and
3
0 wt% CeOx/Pd/Al2O3. A comparison of Fig. 2b to Fig. 2a
c
0.29
reveals that small ceria crystallites of about 3-nm average
size are dispersed over the alumina support. Figure 2cshows
a large Pd particle (� 20 nm) with a small ceria crystallite
about 5 nm deposited on it. Finally, Fig. 2d shows a slightly
larger size Pd particle with numerous ceria crystallites of 3
to 7 nm deposited around it.
0.29c
a
Quantitative analysis by Galbraith Laboratories Inc. and Southern
Testing and Research Laboratories Inc.
b
Total chemisorption-reversible chemisorption at 303 K.
Based on value for 5 wt% CeOx/Pd/Al2O3 sample.
c
FIG. 1. Dihydrogen chemisorption isotherms for (a) 5 wt% Pd/Al2O3, (b) 1.0 wt% Pd/Al2O3, and (c) 0.5 wt% Pd/Al2O3. Lines indicate extrapolated
values of total and reversible dihydrogen uptake.

1
96
HOLLES ET AL.
FIG. 2. TEM images of (a) Al2O3, (b) CeOx/Al2O3, (c) 5 wt% CeOx/Pd/Al2O3, and (d) 30 wt% CeOx/Pd/Al2O3.

PALLADIUM–CERIA CATALYSTS FOR NO + CO
197
FIG. 2—Continued

1
98
HOLLES ET AL.
FIG. 2—Continued

PALLADIUM–CERIA CATALYSTS FOR NO + CO
199
FIG. 2—Continued

2
00
HOLLES ET AL.
FIG. 3. Arrhenius-type plots for the NO + CO reaction over 5 wt% Pd/Al2O3, 1.0 wt% Pd/Al2O3, 0.5 wt% Pd/Al2O3, 10 wt% Pd/CeOx/Al2O3,
wt% Pd/CeOx/Al2O3, and 1.0 wt% Pd/CeOx/Al2O3 for reactant mixtures of 5.07 kPa of each NO and CO.
5
For the NO + CO reaction, the only products observed catalysts as a function of temperature. Dinitrogen selectiv-
were N2, N2O, and CO2. No NO2 was detected in any reac- ity is defined as the rate of N2 production divided by the
tion run. Evaluation of the Weisz–Prater criterion indicated rate of N2 and N2O production. For all Pd particle sizes, the
that no pore diffusion limitations existed. Conversions for presence of ceria decreased the dinitrogen selectivity.
these experiments ranged from 3 to 6% .
The influence of reactant partial pressures on the rate of
Figure 3 shows Arrhenius-type plots for the NO + CO the NO + CO reaction for Pd/Al2O3 catalysts as a function
reaction on the Pd/Al2O3 and Pd/CeOx/Al2O3 catalysts as a
function of particle size. Apparent activation energies de-
rived from these plots are presented in Table 2. Clearly,
the ceria-promoted Pd catalysts are much more active than
any of the unpromoted catalysts. Figure 4 compares the
dinitrogen selectivity for the Pd/Al2O3 and Pd/CeOx/Al2O3
TABLE 2
Apparent Activation Energies and Reaction Orders
Apparent
Reaction order
activation
�
Catalyst
energy (kJ mol 1)
NO
CO
0
1
5
1
5
1
5
1
3
.5 wt% Pd/Al2O3
wt% Pd/Al2O3
wt% Pd/Al2O3
wt% Pd/CeOx/Al2O3
wt% Pd/CeOx/Al2O3
0 wt% Pd/CeOx/Al2O3
wt% CeOx/Pd/Al2O3
0 wt% CeOx/Pd/Al2O3
0 wt% CeOx/Pd/Al2O3
51.0
42.6
59.8
68.9
76.6
56.2
31.3
43.0
37.1
0.51
0.72
� 0.49
� 0.51
� 1.11
1.02
a
� 0.50
a
a
a
a
a
a
1.3
0.67
� 0.97
� 0.32
FIG. 4. N2 selectivity vs temperature for the NO + CO reaction
over 5 wt% Pd/Al2O3, 1.0 wt% Pd/Al2O3, 0.5 wt% Pd/Al2O3, 10 wt%
Pd/CeOx/Al2O3, 5 wt% Pd/CeOx/Al2O3, and 1.0 wt% Pd/CeOx/Al2O3.
a
Indicates a nonconstant reaction order. See figures for details.

PALLADIUM–CERIA CATALYSTS FOR NO + CO
201
and CO were positive at low partial pressures and negative
at high partial pressures, over the range of conditions in-
vestigated. The maximum in rate occurred near equimolar
concentrations of NO and CO for these samples.
The Arrhenius-type plots for the catalysts with ceria de-
posited after palladium loading are shown in Fig. 7. Addi-
tion of 5 wt% ceria to the palladium results in an activity
increase of approximately 1 order of magnitude. The sub-
sequent addition of up to 30 wt% ceria results in only a
marginal increase in activity. For comparison with the sam-
ples with palladium added on top of ceria, the turnover
frequency for the 10 wt% CeOx/Pd/Al2O3 is almost identi-
cal to that of the 1 wt% Pd/CeOx/Al2O3 sample. Apparent
activation energy and reaction orders for the CeOx/
Pd/Al2O3 samples are also given in Table 2. Apparent acti-
�
1
vation energies were all similar at about 37 kJ mol and,
compared to the unpromoted 1 wt% Pd/Al2O3 sample, the
addition of ceria had very little effect on the apparent
FIG. 5. Reaction order determination for the NO + CO reaction for
wt% Pd/Al2O3 and 0.5 wt% Pd/Al2O3. Figure 5a is for a fixed PNO =
.05 kPa with a varying PCO. Figure 5b is for a fixed PCO = 4.05 kPa with a
5
4
varying PNO.
of particle size is summarized in Figs. 5a and 5b for CO and
NO, respectively. The lines represent a linear regression
of the rate data which yield orders for NO of 1.02 � 0.15
for 5 wt% Pd/Al2O3 and 0.51 � 0.15 for 0.5 wt% Pd/Al2O3
(
based on NO consumption). The plot indicates the reac-
tion order for CO becomes less negative and the reaction
order for NO becomes less positive as the particle size de-
creases. Results for the other catalysts are shown in Table 2.
Promotion of the Pd by ceria significantly altered the reac-
tion orders. For instance, the NO reaction order was 1.02
on 5 wt% Pd/Al2O3 while for similar-sized Pd particles, the
NO reaction order on 10 wt% Pd/CeOx/Al2O3 was positive
at low partial pressures of NO and became negative for
high partial pressures of NO. Effects of CO and NO partial
pressures on reaction rates as a function of particle size on
the promoted catalysts are shown in Figs. 6a and 6b, respec-
FIG. 6. Reaction order determination for the NO + CO reaction for
5
wt% Pd/CeOx/Al2O3 and 1.0 wt% Pd/CeOx/Al2O3. Figure 6a is for a fixed
PNO = 4.05 kPa with a varying PCO. Figure 6b is for a fixed PCO = 4.05 kPa
tively. For some catalysts, the reaction orders in both NO with a varying PNO.

2
02
HOLLES ET AL.
FIG. 7. Arrhenius-type plots for the NO + CO reaction over 1.0 wt% Pd/Al2O3, 5 wt% CeOx/Pd/Al2O3, 10 wt% CeOx/Pd/Al2O3, and 30 wt%
CeOx/Pd/Al2O3 for reactant mixtures of 5.07 kPa of each NO and CO.
activation energy. Ceria did affect the NO and CO reac- 25 wt% ceria resulted in the reappearance of an O2 desorp-
tion orders, however. The activity of a catalyst with 5 wt% tion feature and a decrease in the magnitude of the N2O
ceria exhibited a maximum with respect to reactant par- peak, which shifted to a slightly higher temperature. For
tial pressure, which is similar to the behavior of several 10 wt% CeOx/Al2O3 with no Pd present, there was a large
Pd/CeOx/Al2O3 samples. Catalysts with higher ceria load- O2 desorption peak at 710 K with N2 and N2O desorption
ings exhibited constant reaction orders for NO and CO over similar to that of the previous sample. Apparently, the ef-
the range of conditions tested.
fect of ceria addition after loading of Pd particles results
The effect of Pd particle size on the adsorption of NO in suppression of the O2 peak and the increase of the N2O
and subsequent TPD is shown in Fig. 8. The signal for NO peak as the ceria loading increases. High loadings of ce-
is not shown since it was very large and similar for all cata- ria result in a new O2 desorption feature due to the ceria
lysts. For the smallest Pd particles (1.1 nm), the major des- crystallites.
orption feature is attributed to O2 which desorbed with a
Figure 10 shows the TPD plots for various sizes of Pd par-
peak temperature of 720 K. Dinitrogen desorbed in a very ticlesdeposited onto CeOx/Al2O3. The major feature for the
broad peak at � 550 K and a smaller peak at 700 K. A small smallest particles (<1 nm) was a large O2 peak at 710 K.
amount of N2O also desorbed with a peak at 700 K. For Other features include two broad N2 desorption peaks at
slightly larger particles (1.5 nm), the main feature was also 450 and 700 K and an N2O desorption peak at 700 K. In-
an O2 desorption peak at 710 K. For the largest Pd particles creasing the Pd particle size to 2.0 nm suppressed the O2
(
4.5 nm), the O2 desorption feature was no longer present. peak while increasing the magnitude of the N2 and N2O
Instead, the major desorption product was N2O with two peaks, which were also shifted to lower temperatures of
broad desorption peaks at 430 and 610 K. Therefore, the ef- 670 and 620 K, respectively. For the largest ceria-promoted
fect of increasing the Pd particle size from 1 to 5 nm appears Pd particles (9 nm), two well-defined N2 and N2O des-
to be the suppression of O2 desorption and enhancement orption peaks were observed at temperatures of 620 and
of N2O desorption.
590 K, respectively. The overall effect of Pd particle size for
The TPD plots in Fig. 9 show the effect of depositing the Pd/CeOx/Al2O3 samples was to suppress O2 formation
ceria on top of supported palladium. For 1 wt% Pd/Al2O3 while enhancing N2 and N2O formation.
(
1.5-nm Pd particles), the major desorption feature at 710 K
was again due to O2. However, addition of 5 wt% ceria to
this sample suppressed the O2 desorption peak. A broad
N2O desorption peak at 675 K and a second smaller peak
DISCUSSION
The TEM images clearly show that ceria typically forms
at 440 K were observed instead. Incorporation of additional small, well-dispersed crystallites on the alumina support.

PALLADIUM–CERIA CATALYSTS FOR NO + CO
203
stable atomic nitrogen species (Na) is thought to be more
important than NO dissociation in determining the kinetics
of this reaction under some conditions. This conclusion is
also indicated by previous NO TPD studies that have shown
the less active Pd(100) surface is 80% covered by N atoms
while the more active Pd(111) surface is only 20% covered.
This concept can be used to explain the particle size effect
for unpromoted and ceria-promoted Pd.
To understand the role of Pd particle size on ceria-
promoted Pd for the NO + CO reaction, it is desirable to
first study the effect of Pd particle size on unpromoted Pd.
Previous results by Rainer et al. have demonstrated that ac-
tivity for the NO + CO reaction decreases as the Pd particle
size decreases from 120 to 6.0 nm (2). The specific activity of
our 5 wt% Pd/Al2O3 (4.5-nm Pd particles) was only slightly
higher than the activity reported by Rainer et al. for a cata-
lyst with 6.0-nm Pd particles. However, as we decreased
Pd particle size to 1.5 nm or smaller, we observed a three-
fold increase in turnover frequency. Rainer et al. reported
an increase of activity of 50% as the Pd particle size was
decreased from 6 to <3 nm, but this observation was not
discussed.
FIG. 8. TPD curves for N2, O2, and N2O desorption following a sat-
uration exposure to NO for 0.5 wt% Pd/Al2O3, 1.0 wt% Pd/Al2O3, and
5
wt% Pd/Al2O3.
For the samples with ceria added after Pd incorporation, ce-
ria was seen to deposit as small crystallites on top of and/or
near the Pd particles. It does not appear that ceria formed
a very thin film that covers the particle. As the amount of
postdeposited ceria increased, the number of ceria crys-
tallites in contact with the Pd particles increased. Although
the Pd particles shown in these TEM images are larger than
the average particle size determined by chemisorption, they
serve to illustrate that ceria crystallites were observed near
the metal particles. For the samples with Pd deposited after
ceria, the Pd was still found to be in contact with small ceria
crystallites. However, we did not find any evidence for ceria
migration on top of the Pd particles.
The NO + CO reaction over single-crystal Pd has been
shown to be structure-sensitive (2, 9). The Pd(111) sur-
face was observed to be more active than the Pd(100) sur-
face which is in turn more active than the Pd(110) surface.
The activity results contrast with the observation that the
Pd(100) surface is more active than the Pd(111) surface for
FIG. 9. TPD curves for N2, O2, and N2O desorption following a sat-
uration exposure to NO for 1.0 wt% Pd/Al2O3, 5 wt% CeOx/Pd/Al2O3,
NO dissociation (2, 9). Therefore, removal of a thermally 30 wt% CeOx/Pd/Al2O3, and 10 wt% CeOx/Al2O3.

2
04
HOLLES ET AL.
carbon and CO2 whereas larger particles cannot (14). This
disproportionation ability was attributed to small particles
preferring icosahedral packing instead of the standard face-
centered cubic structure (15). We suggest that the observed
increase in NO + CO activity for small particlesresults from
the enhanced ability to dissociate NO, similar to the case
described above for CO dissociation.
It is also instructive to discuss our results in light of those
for Rh, a typical NOx reduction catalyst. The activity for
the NO + CO reaction over supported Rh reportedly de-
creases as Rh particle size decreases from 70 to <1 nm (16).
An increase in activity was not observed for very small Rh
particles, which is in contrast to our observation with sup-
ported Pd. This is not surprising in light of the ability of Rh
to readily dissociate NO (17, 18). In addition, under iden-
tical experimental conditions for NO + CO, the observed
reaction orders on Rh/Al2O3 imply a high N atom coverage
whereas the orders on Pd/Al2O3 (4.5 nm) suggest a low N
atom coverage (6). Finally, very small Rh particles are not
required to dissociate CO (19).
In addition to Pd particle size, ceria plays a crucial role
in the NO + CO reaction. The turnover frequency for the
ceria-promoted 5 wt% Pd was approximately 2 orders of
magnitude greater than that for the unpromoted 5 wt% Pd.
Although alumina-supported ceria catalyzed the NO + CO
reaction with no metal present, the reaction rate, on a per
gram basis, was a factor of 4 lower than any other cata-
lyst studied in this work. Therefore, the enhancement in
FIG. 10. TPD curves for N2, O2, and N2O desorption following a sat-
uration exposure to NO for 1.0 wt% Pd/CeOx/Al2O3, 5 wt% Pd/CeOx/ observed rates is either due to a modification of the metal
Al2O3, and 10 wt% Pd/CeOx/Al2O3.
particles or creation of new sites at the metal/promoter in-
terface. To explain the enhanced activityfor ceria-promoted
The measured NO and CO reaction orders as a function Rh, Oh proposed that NO dissociates at interface sites (7).
of Pd particle size help to provide an understanding of this Based on our observations of the kinetics of unpromoted
increase in activity for the smallest particles. In an earlier and ceria-promoted Pd, we have extended this explana-
paper, a reaction rate expression was developed for the tion to include Pd (6). The enhancement of NO dissoci-
NO + CO reaction over Rh and Pd in which the NO reac- ation at the metal–promoter interface can also be used
tion order can be positive, zero, or negative, depending on to explain the effect of Pd particle size on catalysis by
the relative surface coverages of NO, CO, and N atoms (6). ceria-promoted Pd.
High coverage of NO results in a negative NO reaction or-
For the sample with the largest Pd particles (10 wt%
der, high coverage of CO results in a positive NO reaction Pd/CeOx/Al2O3), which should have the smallest interfa-
order, and high coverage of N atoms results in zero reac- cial contact relative to metal particle surface area, the ob-
tion orders for both NO and CO. For the 5 wt% Pd/Al2O3 served promotional effect was the smallest. In this case,
sample, a NO reaction order of 1.02 indicates high CO cov- the enhanced NO dissociation provided by the promoter
erage (and correspondingly low coverages of NO and N). interface did not affect surface coverages of adsorbed in-
However, as the particle size decreased, the NO reaction termediates until intermediate NO pressures were attained,
order decreased. This implies that the surface coverage of as indicated by the NO reaction order. At intermediate NO
N atoms increases as the particle size decreases. Enhanced pressures, the coverage of N atom was high and the reac-
NO dissociation and/or N atom stabilization can account for tion appeared to be zero order in NO. The promotional
this change in reaction order. Since the catalytic activity in- role of the Pd/ceria interface is evidently diminished for
creased while the NO reaction order decreased (indicating very large particles. For the 5 wt% Pd/CeOx/Al2O3 sam-
higher N atom coverage), we believe that enhanced NO ple with the highest catalytic activity, the proportion of
dissociation is more likely. The ability of small Pd parti- promoter/metal interface to Pd surface area was greater,
cles to enhance dissociation of adsorbed molecules has also thus providing more dissociated N atoms to the Pd metal
been observed in other systems. Ichikawa et al. showed that surface. However, as the Pd particles were made smaller,
small particles of Pd (<2 nm) can disproportionate CO to the amount of open surface facets increased, which resulted

PALLADIUM–CERIA CATALYSTS FOR NO + CO
205
in a partial stabilization and higher fractional surface cov-
The large increase in rate accomplished by adding 5 wt%
erage of N atoms. The high observed promotional effect of ceria indicated that submonolayer coverages of ceria can
ceria on this sample indicates a balance between the interfa- affect activity. The fact that the rate did not increase signif-
cial contact with promoter and the stabilization of N atoms icantly as additional ceria was added may indicate prefer-
at the surface. This balance is between the ability of the ential formation of ceria crystallites near the Pd particles
interface to provide N atoms through increased NO disso- during the calcination process. In addition to activity, the
ciation at the Pd–ceria interface and the ability of the metal effect of increasing the Pd/ceria interfacial contact can be
particle to react the N atoms off the surface. For the small- seen in the reaction orders. Incorporation of 5 wt% ceria
est particles, the ability of open facets to stabilize adsorbed changed the NO reaction order from the constant 1.02 for
N atoms overcomes the ability of the interface to enhance 1 wt% Pd/Al2O3 to a variable reaction order with a maxi-
dissociation. The high fractional coverage of N atoms then mum similar to the 5 and 10 wt% Pd/CeOx/Al2O3 samples.
results in decreased activity, as observed for single crystals. Again, this type of reaction order behavior indicates that
A high coverage of N atoms results in a NO reaction order NO dissociation is enhanced at intermediate pressures. For
of zero, even for low PNO on the smallest particles.
the 10 wt% ceria sample the constant reaction order in-
Apparent activation energies for the unpromoted cata- dicates that even at intermediate to high PNO, the addi-
lysts were similar at around 50 kJ/mol and there was no ob- tional Pd/ceria interface is dissociating NO rapidly enough
vious influence of particle size. For the ceria-promoted sam- to prevent inhibition by NO. Finally, at 30 wt% ceria, the
ples, the apparent activation energies were slightly higher, increased Pd/ceria interface dissociates NO rapidly enough
ranging from 56.2 to 76.6 kJ/mol. The apparent activation to build up N atoms on the Pd metal and alter the NO and
energy for the least promoted sample was similar to that for CO reaction orders appropriately.
the unpromoted samples. Dinitrogen selectivity was typi-
The results from TPD of preadsorbed NO substantiate
cally lower for the ceria-promoted than for the unpromoted the conclusions derived from kinetic analysis. Indeed, the
samples. The temperature appears to have a larger effect desorption of products was significantly affected by Pd par-
on selectivity than presence of the promoter. At a tempera- ticle size. For unpromoted Pd, the smallest Pd particles
ture of 520 K, the selectivity was highest for the smallest un- (<1.5 nm) were able to desorb dioxygen whereas desorp-
promoted particles and lowest for the largest unpromoted tion peaks for N2 and N2O were very broad and small. How-
particles. This result is consistent with those of Rainer et al. ever, for the largest Pd particles (4.5 nm), no O2 desorption
who reported that the more open single-crystal faces had was observed and instead the oxygen was desorbed in the
higher N2 selectivities than the close-packed surfaces (2). form of N2O product. Since enhanced NO dissociation can
For the ceria-promoted samples, the most active sample lead to O2 desorption, and less NO dissociation can lead
had the highest selectivity while the less active samples to N2O desorption, the TPD results provide additional ev-
had lower selectivity. The lower selectivity for the ceria- idence for the enhanced dissociation of NO on small Pd
promoted samples is not easily interpreted in light of the ex- particles. Desorption of O2 from Pd particles during NO
pected increase in NO dissociation for these samples which TPD was not previously observed by Xu and Goodman or
should result in greater N2 production due to the presence Cordatos et al.; however, they did not have particles smaller
of additional N atoms. This inconsistency was also observed than 1.6 nm (11, 20). The O2 desorption peak for very small
by Rainer et al. for Pd/Al2O3 catalysts (2) and Oh and Eickel particles may suggest the predominant facet that is exposed.
for Rh/Al2O3 catalysts (16). Perhaps ceria also affects prod- For example, some authors did not observe O2 desorption
uct formation. It is important to note that although selec- during TPD from single-crystal Pd(111) previously exposed
tivity decreased slightly, the activity for both N2 and N2O to NO (18, 21). Others report that chemisorbed O2 desorbs
formation increased significantly with ceria promotion.
from Pd(111) at temperatures greater than about 800 K
In an attempt to separate the effects of the Pd/ceria in- (22, 23), which exceeds the maximum temperature in our
terface and Pd particle size, we prepared a series of samples work. Dioxygen desorbing from the more open Pd(110)
with constant particle size and increased the Pd/ceria inter- surface gives two desorption peaks at about 700 and 800 K,
face by depositing ceria on top of the Pd. The rate mea- with the low-temperature peak originating from subsurface
surements showed that the addition of only 5 wt% ceria to oxygen (24–26). Desorption of subsurface oxygen prepared
1
wt% Pd/Al2O3 provided an order of magnitude increase by decomposition of NO2 on Pd(111) generates lower tem-
in activity. The rate on the 10 wt% ceria sample was essen- perature O2 TPD features (27), which are consistent with
tially the same. The activities of 1 wt% Pd/CeOx/Al2O3 and O2 desorption from the more open Pd(110) crystal face pre-
1
0 wt% CeOx/Pd/Al2O3 were similar, indicating that the viously exposed to O2. Apparently, open-crystal planes of
order of ceria addition is not a major factor in the catalyst Pd facilitate incorporation of subsurface oxygen. Thus, the
preparation method. The presence of additional ceria up observation of O2 desorption from small Pd particles re-
to 30 wt% total resulted in only a slight increase in activ- ported here indicates that they likely expose open-faceted
ity. Dinitrogen selectivities for all ceria-containing samples surfaces which may lead to subsurface oxide formation. Our
were similar.
TPD results for N2 and N2O as a function of particle size are

2
06
HOLLES ET AL.
qualitatively similar to those of Cordatos et al. and Xu and mation and/or desorption. The variations in apparent re-
Goodman since the smallest particles evolved mostly N2 action orders and results from TPD were consistent with
and very little N2O and the larger particles produced more the idea that NO dissociation is promoted on very small
N2O and less N2 (11, 20). Enhanced NO dissociation on the particles (1 nm) and by addition of ceria.
smaller particles is also consistent with N2 being the pre-
Characterization of the catalysts by TEM indicated that
dominant N-containing product evolved during TPD from ceria was typically in the form of small crystallites from 3
small particles whereas N2O is preferred on large particles. to 7 nm in diameter. For the samples with ceria deposited
Addition of ceria on top of supported Pd particles sup- after Pd, the ceria crystallites were typically deposited near
pressed O2 desorption at low ceria loading. In addition, a or on top of the Pd particle. No evidence of a ceria thin film
low-temperature N2 peak appeared, indicating enhanced on the Pd was detected.
NO dissociation. Higher loadings of ceria resulted in the
reappearance of a large O2 desorption feature at � 700 K.
ACKNOWLEDGMENT
Since O2 evolution was detected from supported ceria
(
without Pd present) during TPD of adsorbed NO, we at-
This work was supported by the Division of Chemical Sciences, Office
of Basic Energy Sciences, Office of Energy Research, U.S. Department of
Energy.
tribute the reappearance of an O2 feature on high ceria-
loaded Pd/Al2O3 to the ceria component.
For the samples with ceria deposited prior to the Pd, O2
desorption was only observed from the sample with the
smallest Pd particles. Enhanced NO dissociation on this
sample was indicated by the preferential desorption of N2
instead of N2O. As the Pd particle size increased, the O2
desorption was completely suppressed. For the largest par-
ticles, both N2O and N2 desorption features became well
defined, similar to features seen in single-crystal studies.
Indeed, the N2O and N2 peak temperatures are close to the
single-crystal values of � 510 K for N2O and 650 K for N2
reported by Schmick and Wassmuth and 480 and 670 K by
Conrad et al. for a Pd(111) surface (18, 28). The N2O and
N2 peak temperatures for the sample with the largest Pd
particles are also similar to the results obtained by Rainer
et al. for 8 nm, unpromoted Pd/Al2O3 particles (525 and
REFERENCES
1
2
. Taylor, K. C., Catal. Rev.-Sci. Eng. 35, 457 (1993).
. Rainer, D. R., Vesecky, S. M., Koranne, M., Oh, W. S., and Goodman,
D. W., J. Catal. 167, 234 (1997).
. Rainer, D. R., Koranne, M., Vesecky, S. M., and Goodman, D. W.,
J. Phys. Chem. B 101, 10769 (1997).
. Almusaiteer, K., and Chuang, S. S. C., J. Catal. 180, 161 (1998).
. Almusaiteer, K., and Chuang, S. S. C., J. Catal. 184, 189 (1999).
. Holles, J. H., Switzer, M. S., and Davis, R. J., J. Catal. 190, 247 (2000).
7. Oh, S. H., J. Catal. 124, 477 (1990).
. Le Normand, F., Hilaire, L., Kili, K., Krill, G., and Maire, G., J. Phys.
Chem. 92, 2561 (1988).
3
4
5
6
8
9
. Vesecky, S. M., Chen, P., Xu, X., and Goodman, D. W., J. Vac. Sci.
Technol. A 13, 1539 (1995).
0. Vesecky, S. M., Rainer, D. R., and Goodman, D. W., J. Vac. Sci. Technol.
1
6
00 K, respectively) (2). The agreement between the TPD
A 14, 1457 (1996).
1
1
1
1. Xu, X., and Goodman, D. W., Catal. Lett. 24, 31 (1994).
2. Benson, J. E., Hwang, H. S., and Boudart, M., J. Catal. 30, 146 (1973).
3. Boudart, M., and Dj e´ ga-Mariadassou, G., “Kinetics of Heterogeneous
Catalytic Reactions.” Princeton Univ. Press, Princeton, NJ, 1984.
4. Ichikawa, S., Poppa, H., and Boudart, M., J. Catal. 91, 1 (1985).
of NO from promoted, large Pd particles and the Pd single-
crystal studies illustrates that the Pd/ceria interface does
not significantly affect the results over the largest Pd parti-
cles in this study. When comparing the promotional effect
1
of ceria on TPD, the enhanced NO dissociation provided 15. Poppa, H., Moorhead, R. D., and Avalos-Borja, M., J. Vac. Sci. Tech-
nol. A 7, 2882 (1989).
6. Oh, S. E., and Eickel, C. C., J. Catal. 128, 526 (1991).
7. Belton, D. N., DiMaggio, C. L., Schmieg, S. J., and Ng, K. Y. S., J. Catal.
by the Pd/ceria interface served to increase the amount of
1
1
N2 desorbed at lower temperatures. Similar increased N2
desorption was observed by Cordatos and Gorte for Pd
particles on flat polycrystalline ceria supports (29).
157, 559 (1995).
1
1
2
8. Schmick, H.-D., and Wassmuth, H.-W., Surf. Sci. 123, 471 (1982).
9. Solymosi, F., and Erd o¨ helyi, A., Surf. Sci. 110, L630 (1981).
0. Cordatos, H., Bunluesin, T., and Gorte, R. J., Surf. Sci. 323, 219 (1995).
CONCLUSIONS
21. Davies, P. W., and Lambert, R. M., Surf. Sci. 110, 227 (1981).
2
2
2
2. Conrad, H., Ertl, G., Kuppers, J., and Latta, E. E., Surf. Sci. 65, 245
1977).
3. Guo, X., Hoffman, A., and Yates, J. T., Jr., J. Chem. Phys. 90, 5787
1989).
4. He, J.-W., Memmert, U., and Norton, P. R., J. Chem. Phys. 90, 5088
1989).
(
Palladium particles of 1-nm average diameter were more
active than 4.5-nm Pd particles for the NO + CO reaction.
Results from kinetic studies and TPD of NO suggest that
NO dissociation is enhanced on the smallest particles, but
(
(
that N atoms are also held more strongly on the Pd. When 25. Milun, M., Pervan, P., Vajic, M., and Wandelt, K., Surf. Sci. 211/212,
8
87 (1989).
ceria is used to promote the Pd, the activity for NO + CO
was a maximum for 2-nm-sized Pd particles. A maximum
in activity results from the balance between the Pd/ceria in-
terface that enhances NO dissociation and the close-packed
2
2
2
6. Bondzie, V. A., Kleban, P., and Dwyer, D. J., Surf. Sci. 347, 319 (1996).
7. Banse, B. A., and Koel, B. E., Surf. Sci. 232, 275 (232).
8. Conrad, H., Ertl, G., Kuppers, J., and Latta, E. E., Surf. Sci. 65, 235
(
1977).
planes of the Pd particle surface that facilitate product for- 29. Cordatos, H., and Gorte, R. J., J. Catal. 159, 112 (1996).