Investigation of Pt/γ-Al2O3 catalysts with locally high Pt concentrations for oxidation of CO at low temperatures

Article abstract of DOI:10.1016/j.jcat.2003.08.017

The development of low-temperature active catalysts is important for applications such as catalytic combustion of VOC and abatement of vehicle exhausts. A new method for preparing supported catalysts, in which Pt is distributed in locally high concentrations on the γ-Al₂O₃ support, was studied. These catalysts were compared with a conventionally prepared Pt/γ-Al₂O₃ catalyst in which Pt was deposited evenly on the support. A significant improvement of the activity was observed for the catalysts prepared with locally high Pt concentrations when CO (1%, 1000 and 100 ppm) was oxidized at a constant O₂ concentration (10%). The Pt/1% Al₂O₃ catalyst with largest Pt crystallites (4.5 nm) showed the highest activity while the Pt/100% Al₂O₃ catalyst with smallest Pt crystallites (2.3 nm) showed the lowest activity.

Full text of DOI:10.1016/j.jcat.2003.08.017

Journal of Catalysis 221 (2004) 252–261
www.elsevier.com/locate/jcat
Investigation of Pt/γ -Al2O3 catalysts with locally high Pt concentrations
for oxidation of CO at low temperatures
Karl Arnby,a,b,∗ Anders Törncrona,c,1 Bengt Andersson,a,d and Magnus Skoglundh a,b
a
Competence Centre for Catalysis, Chalmers University of Technology, SE-412 96 Gothenburg, Sweden
b
Applied Surface Chemistry, Chalmers University of Technology, SE-412 96 Gothenburg, Sweden
c
Perstorp Formox, SE-284 80 Perstorp, Sweden
d
Chemical Reaction Engineering, Chalmers University of Technology, SE-412 96 Gothenburg, Sweden
Received 22 May 2003; revised 18 August 2003; accepted 22 August 2003
Abstract
A new method for preparing supported catalysts, in which Pt was distributed in locally high concentrations on the γ -Al O support,
2
3
was studied. These catalysts were compared with a conventionally prepared Pt/γ -Al O catalyst in which Pt was deposited evenly on the
2
3
support. The object was to ascertain whether it is possible to prepare catalysts that retain heat released from exothermic reactions to a higher
extent and thereby become more low-temperature active than a conventionally prepared catalyst. A significant improvement of the activity
was observed for the catalysts prepared with locally high Pt concentrations when CO (1%, 1000 and 100 ppm) was oxidized at a constant
O concentration (10%). The improved activity is discussed in terms of heat transfer, mass transfer, and structure sensitivity. Differences in
2
heat transfer appear to be the least probable reason for the enhanced activity for the catalysts with locally higher Pt concentrations, whereas
structural effects also seem to be an unlikely explanation. Differences in mass transfer seem, however, to be a more likely reason for the
improved activity.
2003 Elsevier Inc. All rights reserved.
Keywords: Catalytic activity; CO oxidation; Low temperature; Platinum; Alumina; Mass transport; Heat transport; Structure sensitivity
1. Introduction
tion, and radiation. Even if the heat is rapidly removed from
the active sites, large temperature gradients between the ac-
tive sites and the support material can arise which affect the
catalytic activity. The topic of such temperature gradients
was first discussed by Damköhler [1], and since that sev-
eral theoretical [2–7] and experimental [8–16] studies have
been contradictory. Early theoretical studies [3–6] report in-
stantaneous temperature rises in crystallites in the order of
several hundred degrees Celsius, lasting up to 10−10 s. In-
stead of considering the transient gradients, Holstein and
Boudart [7] calculated the constant temperature gradients
between the active metal and the support at steady state,
(Tmetal −Tsupport)/Tsupport, during exothermic reactions to be
less than 0.03%.
Large temperature gradients between active sites and sup-
port material may result in hot active sites, but to experi-
mentally verify the existence of such hot sites is a difficult
task. In 1973, Mark and Low [9] used IR radiometry to mea-
sure temperature changes when prereduced Ni/SiO2 was ox-
idized. The authors did not find any emission gradients and
concluded that the Ni crystallites and the silica support had
The development of low-temperature active catalysts is
important for applications such as catalytic combustion of
VOC (volatile organic compounds) and abatement of vehicle
exhausts. The usage of such catalysts will in these applica-
tions reduce both emissions of harmful compounds and costs
for external heating of the feed gas (generally required for
VOC combustion). To obtain high activity at low temper-
atures, it is essential to optimize the effects from heat and
mass transfer in the catalyst. For reactions which are struc-
ture sensitive it is also important to optimize the size of the
active sites.
Heat released from chemical reactions on the surface of a
catalyst is removed from the active sites to the surrounding
support material and the gas phase by convection, conduc-
*
Corresponding author.
E-mail address: arnby@surfchem.chalmers.se (K. Arnby).
Present address: Eka Chemicals AB, SE-445 80 Bohus, Sweden.
1
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2003.08.017
2003 Elsevier Inc. All rights reserved.

K. Arnby et al. / Journal of Catalysis 221 (2004) 252–261
253
the same temperature during the oxidation. However, Kem-
ber and Sheppard [10] reported a temperature difference of
190 ◦C between the metal and the support material when
CO was oxidized over Pd/SiO2. They measured the Pd crys-
tallite temperature using IR spectroscopy and the support
temperature with a thermocouple. Sharma et al. [14] argued,
however, that the support temperature reported by Kember
and Sheppard was incorrectly measured, which resulted in
too high temperature gradients between the Pd crystallites
and the silica support. For CO oxidation over Pt/SiO2, stud-
ied in situ by FTIR absorption spectroscopy, Sharma et al.
[14] reported quite small temperature gradients, e.g., at a
gas-phase temperature of 200 ◦C the support and crystal-
lite temperature was 210 and 235 ◦C, respectively. Matyi et
al. [12] reported temperature gradients in the same range
for the reaction between H2 and CO (Fischer–Tropsch syn-
thesis) over Fe/SiO2. Using Mössbauer spectroscopy they
found temperature gradients of 13 and 19 ◦C between the
metal and the support at gas-phase temperatures of 275 and
300 ◦C, respectively. Further, Frost et al. [15] used neutron
resonance radiography to study possible microscopic tem-
perature inhomogeneities within a Pt/Sm2O3 catalyst used
for CO hydrogenation. The authors did, however, not find
any significant temperature gradients between Pt and Sm2O3
under the experimental conditions used. Clearly it is an ex-
perimental challenge to measure whether the heat released
from exothermic reactions can give rise to sufficiently large
temperature gradients between the metal and the support to
significantly increase the reaction rate over a catalyst.
For a given catalyst material at fixed bulk concentrations
of the reactants, the reaction rate can vary due to local varia-
tions in concentrations at the catalyst surface. Mass transfer
limitations locally decrease the concentration of the reac-
tant resulting in concentration gradients which affect the
correspondingreaction rate. This is most interesting for reac-
tions with negative reaction orders since the reaction rate in-
creases with decreasing concentration. The oxidation of CO
on Pt is especially interesting since it is highly exothermic
and has negative reaction order with respect to CO concen-
tration, due to self-poisoning of the active sites by CO at low
temperatures.
In this investigation we have studied a new method for
preparing supported catalysts. The object was to ascertain
whether it is possible to prepare catalysts that locally, around
the active sites, retain heat released from exothermic reac-
tions to a higher extent than a conventionally prepared cat-
alyst. Using the new preparation method, the active phase,
platinum, was deposited on a small part (1, respectively
10%) of the available γ -Al2O3 support while the major part
of the support material was left unimpregnated.The catalysts
prepared using this method were compared with a conven-
tionally prepared Pt/γ -Al2O3 catalyst in which platinum was
deposited on the entire part (100%) of the alumina support. If
the heat released from the exothermic reaction between CO
and O2 was isolated in the close vicinity of the active sites,
the reaction rate would self-accelerate at lower temperatures
over the catalysts with local Pt concentration compared to
the conventional catalyst with evenly distributed platinum.
2. Experimental
2.1. Catalyst preparation
The catalysts were prepared by depositing three differ-
ent wash coats on separate monolithic substrates. Each wash
coat contained boehmite, γ -Al2O3, and Pt.
2.1.1. Boehmite sol (A)
The boehmite sol was prepared by adding 30 g of
boehmite powder (Dispersal S, Condea Chemie) to a solu-
tion containing 16.2 g 1 M HNO3 and 154 g distilled water
under moderate stirring. The dispersed powder was stirred
for at least 30 min in order to obtain a stable boehmite sol
without aggregated particles.
2.1.2. Wash coat with uniform Pt density, conventional
catalyst—Pt/100% Al2O3 (B)
A 64.0 g γ -Al2O3 powder (Puralox S Ba 70, Condea
Chemie) was added to 100 g distilled water under contin-
uous stirring. The pH was adjusted to 1.9 by adding nitric
acid before impregnation with 1.035 g 0.8 M platinum(II) ni-
trate solution (Hereaus) diluted with distilled water to 10 g.
The impregnated powder was dried at 100 ◦C for 15 h and
calcined at 500 ◦C in air for 1 h. The resulting surface con-
centration of Pt on this powder was about 0.0754 µmol/m2
γ -Al2O3. The Pt/γ -Al2O3 powder and boehmite sol, pre-
pared according to Section 2.1.1. were added to a mixture
of 199 g ethanol and 30 g 1 M nitric acid. The slurry (Pt/γ -
Al2O3, ethanol, nitric acid, and boehmite sol) was finally
ball-milled at a constant rate of 50–70 rpm for 10–24 h.
Since the reactivity for structure-sensitive reactions de-
pends on the shape and size of active sites, the catalytic
performance can be improved by optimizing these forms.
Even for a thoroughly studied reaction such as CO oxidation
over supported noble metal, there is no consensus whether it
is structure insensitive or the activity increases with increas-
ing size of the active sites [17,18]. A study by McCarthy
et al. [19] showed that CO oxidation over Pt/α-Al2O3 can
be structure sensitive depending on the CO concentration.
More recently, Zafiris and Gorte studied the same system
as above [20] but observed that the activity increased with
increasing Pt crystallite size irrespective of the CO concen-
tration and suggested that this is because CO desorbs more
easily from large Pt crystallites.
2.1.3. Wash coat with medium local Pt density—Pt/10%
Al2O3 (C)
A portion, 10 wt% (6.4 g), of the total amount of γ -Al2O3
powder intended for use in the preparation of the alumina
slurry was added to 50 g distilled water under continuous

254
K. Arnby et al. / Journal of Catalysis 221 (2004) 252–261
Table 1
Sample data
Catalyst
Pt/100% Al O
2
Pt/10% Al O
2
Pt/1% Al O
2 3
3
3
Amount applied wash coat (g)
Pt content (mg)
0.15
0.30
100
0.20
0.39
10
0.30
0.59
1
Amount Pt-coated Al O (%)
2
3
a
b
a
b
a
b
Amount adsorbed CO (µmol)
0.52 (0.54)
0.58 (0.58)
0.55 (0.51)
c
a
b
a
b
a
b
Dispersion (%)
Pt surface area (dm )
48 (50)
41 (41)
26 (24)
d
2
a
b
a
b
a
b
3.6 (3.7)
2.3
4.0 (4.0)
2.7
3.8 (3.5)
4.5
e
Mean Pt crystallite diameter (nm)
f
Mean support particle radius (µm)
1.25
151
1.25
152
1.25
162
2
S
(m /g wash coat)
BET
a
Before activity study.
b
c
d
e
f
After activity study.
Assuming Pt:CO stoichiometry of 1:0.7 [23].
2
Assuming 0.08 nm /Pt atom [24].
Assuming spherical particles [24].
Estimation from SEM studies.
stirring. The pH was adjusted to 1.9 by adding nitric acid
before impregnation with an aqueous platinum(II) nitrate so-
lution, dried, and calcined, as described in Section 2.1.2.
The resulting surface concentration of Pt on this powder
was about 0.754 µmol/m2 γ -Al2O3. The Pt/γ -Al2O3 pow-
der, the remaining γ -Al2O3 powder (90 wt% of the total
amount of γ -Al2O3), and the boehmite sol (A) were mixed
in the same proportions and with the same compounds as de-
scribed in Section 2.1.2. and the resulting slurry was finally
ball-milled according to the procedure in Section 2.1.2.
and weighed. This stepwise procedure was repeated until the
desired amount of wash coat was applied on each monolith
sample (Table 1). The wash-coated monoliths were finally
calcined in air at 500 ◦C for 1 h in order to fixate the wash
coat on the monolith framework. During the calcination, the
boehmite binder particles were dehydrated and transformed
into γ -Al2O3.
The catalysts slurries (B)–(D) had the same Pt content
(0.2 wt% based on the dry wash coat). The difference be-
tween the slurries was that Pt was deposited on either 100%
(B), 10% (C), or 1% (D) of the total alumina content, and
these proportions did not change during the ball milling,
since both Pt/Al2O3 and Al2O3 particles were affected to
same extent by the grinding. The amount of applied wash
coat differed in an inverse proportion to the Pt dispersions,
resulting in constant Pt surface area (measured with CO
chemisorption) for all catalysts prepared. Basic character-
istics of the catalysts are given in Table 1. Moreover, three
additional catalysts were prepared in similar procedures as
above but with constant Pt loading instead of constant Pt sur-
face area.
2.1.4. Wash coat with high local Pt density—Pt/1%
Al2O3 (D)
A small portion, 1 wt% (0.64 g), of the total amount of
γ -Al2O3 powder intended for use in the preparation of the
alumina slurry was added to 40 g distilled water under con-
tinuous stirring. The slurry was impregnated with an aque-
ous platinum(II) nitrate solution, dried, and calcined using
the same method as described in Section 2.1.2. The result-
ing surface concentration of Pt on this powder was about
7.54 µmol/m2 γ -Al2O3. The Pt/γ -Al2O3 powder, the re-
maining γ -Al2O3 powder (99 wt% of the total amount of
γ -Al2O3), and the boehmite sol (A) were mixed and finally
ball-milled according to the procedure described in Sec-
tion 2.1.3.
2.2. Catalyst characterization
The dispersion and surface area of Pt for the catalysts
were measured with CO chemisorption [21]. The chemisorp-
tion measurements were performed in a continuous flow
reactor system described elsewhere [22]. The Pt dispersion
was measured for each catalyst twice (first as fresh samples
and secondly after the activity tests). The experiments were
performed by first prereducing the catalyst in 10% H2 at
400 ◦C for 30 min. The catalyst was then cooled to 0 ◦C in
N2 and was kept at this temperature during the entire experi-
ment. After about 10 min, the catalyst was instantly exposed
to 50 ppm CO in N2, while measuring the outgoing con-
centration of CO. During this step, CO was chemisorbed on
the surface Pt atoms. When the outlet concentration reached
2.1.5. Preparation of monolith catalysts—Pt/1% Al2O3,
Pt/10% Al2O3, and Pt/100% Al2O3 (E)
Small samples (23 mm long and 13 mm in diameter) of
washed and dried monolithic cordierite with a cell density
of 400 CPSI (cells per square inch) were used as framework
for the wash coats. The wash coat was uniformly applied
onto the monolith sample by immersing the monolith in one
of the three catalyst slurries prepared according to (B), (C),
or (D). After each immersion, the slurry in the monolith
channels was removed by gently blowing with pressurized
air. The wet monolith sample was dried in hot air (300 ◦C)

K. Arnby et al. / Journal of Catalysis 221 (2004) 252–261
255
Table 2
Summary of experiments
a
b
Figure
1
Catalyst
Gas composition
CO (vol ppm)
Temperature
Flow
Space velocity
◦
−1
O
(vol %)
( C)
(ml/min)
(h
)
2
Pt/100% Al O
10,000
10,000
10,000
10
10
10
50–280
50–280
50–270
1000
1000
1000
17,000
17,000
17,000
2
3
3
3
Pt/10% Al O
2
3
Pt/1% Al O
2
3
2
3
Pt/100% Al O
1000
1000
1000
10
10
10
50–220
70–220
60–210
1000
1000
1000
17,000
17,000
17,000
2
Pt/10% Al O
2
3
Pt/1% Al O
2
3
Pt/100% Al O
100
100
100
10
10
10
40–170
40–190
50–170
1000
1000
1000
17,000
17,000
17,000
2
Pt/10% Al O
2
3
Pt/1% Al O
2
3
a
N
as balance.
2
b
◦
Heating and cooling ramp at 5 C/min.
the inlet CO concentration, the catalyst was saturated with
CO and the reactor was flushed with pure N2 whereby
nonchemisorbed CO desorbed from the sample. Then once
again, the catalyst was exposed to 50 ppm CO in N2. Sub-
tracting the CO responses from the two experiments gives
an area correlating with the amount of chemisorbed CO
on Pt. The Pt dispersion and the corresponding Pt surface
area of the catalysts were calculated from the amount of
chemisorbed CO divided by the total Pt content, using a sto-
ichiometric factor of 0.7 adsorbed CO molecules per Pt sur-
face atom for dispersion [23] and a surface area of 0.08 nm2
per Pt atom [24].
with two thermocouples, which were located 11 mm in front
of the catalyst and in one of the monolith channels close
to the catalyst front. The product gases were continuously
analyzed with respect to CO and CO2 with IR instruments
(UNOR 6N Maihak).
The light-off and extinction processes for CO oxidation,
using constant O2 concentration (10%) and varying the CO
concentration (1%, 1000 and 100 ppm), were studied by first
increasing the reactor inlet temperature at a constant rate of
5 ◦C/min and then quenching the reaction by constant cool-
ing at 5 ◦C/min. The three catalysts were initially reduced
in 10% hydrogen at 400 ◦C for 15 min, followed by oxida-
tion in 10% oxygen at 400 ◦C for 15 min. All experiments
are summarized in Table 2.
The mean platinum crystallite diameter was calculated
assuming spherical crystallites according to Anderson and
Pratt [24],
dPt = 6VPt/APt,
(1)
3. Results
where VPt is the total platinum volume, obtained by dividing
the platinum mass with its density, while APt is the platinum
surface area determined by CO chemisorption.
3.1. Catalyst characterization
The platinum dispersion, the platinum surface area, the
mean platinum crystallite size, and the BET area of the cata-
lysts are given in Table 1. The dispersions for the fresh
and used catalysts do not differ significantly, indicating that
the catalysts did not sinter during the activity studies. The
Pt/100% Al2O3 catalyst shows the highest Pt dispersion
(49%), while the dispersion is 41% for Pt/10% Al2O3 and
25% for Pt/1% Al2O3. The corresponding platinum sur-
face areas are rather constant for the catalysts (3.6 dm2 for
Pt/100% Al2O3, 4.0 dm2 for Pt/10% Al2O3, and 3.6 dm2
for Pt/1% Al2O3, with an error range of 0.2 dm2). CO
chemisorption studies for both empty reactor and for unim-
pregnated alumina support showed no significant CO uptake,
which strongly indicates that CO only chemisorbs on plat-
inum during the experimental conditions used.
The specific surface area and the mean pore diameter of
the wash coats were determined (with an accuracy of 1%)
by nitrogen adsorption after the activity experiments accord-
ing to the BET method using an ASAP 2010 instrument
(Micromeritics). The BET surface area determinations were
based on six measurements at relative pressures of N2 in the
range of 0.03–0.20. The used cross-sectional area of the ni-
trogen adsorbate was 0.162 nm2.
2.3. Activity studies
The influence of the different platinum distributions on
the oxidation of CO was studied using constant gas com-
positions under temperature ramps at atmospheric pressure.
The experiments were performed in the reactor described
above. For the activity tests the catalytic performance of each
catalyst was compared per the same unit of surface area of
platinum metal (3.6–4.0 dm2). The gases (CO, O2, and N2 as
balance) were introduced into the reactor via mass flow con-
trollers (Bronkorst Hi-Tec). Temperatures were measured
The average diameter of the Pt crystallites calculated
from the CO chemisorption data is 2.3, 2.7, and 4.5 nm
for Pt/100% Al2O3, Pt/10% Al2O3, and Pt/1% Al2O3, re-
spectively. While preparing the catalysts, the platinum was
impregnated on the wash coat before the wash coat was de-

256
K. Arnby et al. / Journal of Catalysis 221 (2004) 252–261
Fig. 1. Conversion of CO versus inlet gas temperature during oxidation of 1% CO in 10% O over Pt/γ -Al O . Diamonds (100%); Pt/100% Al O , triangles
2
2
3
2 3
◦
◦
(10%); Pt/10% Al O , and squares (1%); Pt/1% Al O . Open symbols, heating ramps (5 C/min); and filled symbols, cooling ramps (5 C/min).
2
3
2 3
posited on the monolith. This preparation technique may trap
some of the platinum in pores inaccessible to the reactants.
Platinum trapping could be avoided by first depositing the
wash coat on the monolith, and then impregnating the stabi-
lized wash coat with the Pt precursor. When considering this
trapping effect the actual Pt crystallite size could be some-
what smaller than calculated for the catalysts.
The BET surface areas of the three wash coats are sim-
ilar (SBET, 150–162 m2/g stabilized wash coat; mean pore
diameter, 75 Å).
Pt/1% Al2O3, respectively. For all catalysts a pronounced
hysteresis between the light-off and extinction temperature
regions is observed. For the cooling ramp (extinction) the ac-
tivity of the different catalysts differs markedly. The Pt/1%
Al2O3 catalyst has highest low-temperature activity, while
the conventional catalyst, Pt/100% Al2O3, has the lowest
activity. The T50’s for extinction are 177 ◦C for Pt/100%
Al2O3, 165 ◦C for Pt/10% Al2O3, and 144 ◦C for Pt/1%
Al2O3. The values for T50 are summarized in Table 3.
3.3. Oxidation of 1000 ppm CO
3.2. Oxidation of 1% CO
The oxidation of 1000 ppm CO with 10% O2 over the
three catalysts is shown in Fig. 2, and the corresponding
light-off and extinction temperatures are summarized in Ta-
ble 3. The conventional catalyst, Pt/100% Al2O3, reaches
T50 at the highest temperature, both for light-off (183 ◦C)
and extinction (166 ◦C). On the contrary, the lowest T50’s are
observed for the Pt/1% Al2O3 catalyst (T50,light-off 164 ◦C,
T50,extinction 127 ◦C). The Pt/10% Al2O3 sample shows an
intermediate activity in comparison to the other two cata-
lysts, with T50,light-off at 177 ◦C and T50,extinction at 151 ◦C.
The appearances of the CO conversion profiles are similar to
those for 1% CO in 10% O2 as described above.
Fig. 1 shows oxidation of 1% CO over Pt/100% Al2O3,
Pt/10% Al2O3, and Pt/1% Al2O3 with 10% O2 during heat-
ing and cooling ramps. The graph displays the effect of the
inlet temperature on the CO conversion. The conversion fol-
lows a typical light-off process for CO oxidation over plat-
inum which can be divided in three different activity regions.
At low temperatures, the reaction is self-inhibited by a high
CO coverage on the active sites [25] and the conversion is
thus very low. At higher temperatures, the CO conversion is
high and the surface coverage of CO is low. In this region the
reaction rate is limited by the transport of reactants to the ac-
tive sites of the catalyst. In the intermediate light-off region,
during the heating ramp, the reaction is autocatalyzed by the
evolved reaction heat which results in a rapid increase from
low to high conversion. While cooling the inlet gas, the re-
action proceeds from the region with high conversion to the
region with low conversion. In the intermediate extinction
interval, the reaction rate decreases fast. The extinction from
high to low conversion is somewhat less rapid than the light-
off process.
3.4. Oxidation of 100 ppm CO
Fig. 3 shows the oxidation of 100 ppm CO with 10% O2
over the three catalysts and the T50 values during the tem-
perature ramp experiments are presented in Table 3. The CO
conversion profiles resemble the corresponding profiles for
1000 ppm and 1% CO with sharp light-off and extinction
intervals; see above. Even with this rather low CO concentra-
tion, a marked hysteresis between the light-off and extinction
processes is seen. The hysteresis is most pronounced for the
For the heating ramps the light-off temperatures, T50
(temperature at 50% conversion), are rather constant, 232,
237, and 228 ◦C for Pt/100% Al2O3, Pt/10% Al2O3, and

K. Arnby et al. / Journal of Catalysis 221 (2004) 252–261
257
Table 3
Temperatures for 50% conversion (T ) for light-off and extinction experiments
50
Figure
Experiment
Catalyst
Pt/100% Al O
2
Pt/10% Al O
2
Pt/1% Al O
2 3
3
3
1
1% CO
◦
T
T
light-off ( C)
extinction ( C)
232
177
237
165
228
144
50
50
◦
2
3
1000 ppm CO
◦
T
light-off ( C)
extinction ( C)
183
166
177
151
164
127
50
50
◦
T
100 ppm CO
◦
T
light-off ( C)
extinction ( C)
126
117
113
94
102
72
50
50
◦
T
Fig. 2. Conversion of CO versus inlet gas temperature during oxidation of 1000 ppm CO in 10% O over Pt/γ -Al O . Diamonds (100%), Pt/100% Al O ;
2
2
3
2 3
◦
◦
triangles (10%), Pt/10% Al O ; and squares (1%), Pt/1% Al O . Open symbols, heating ramps (5 C/min); and filled symbols, cooling ramps (5 C/min).
2
3
2 3
Fig. 3. Conversion of CO versus inlet gas temperature during oxidation of 100 ppm CO in 10% O over Pt/γ -Al O . Diamonds (100%), Pt/100% Al O ;
2
2
3
2 3
◦
◦
triangles (10%), Pt/10% Al O ; and squares (1%), Pt/1% Al O . Open symbols, heating ramps (5 C/min); and filled symbols, cooling ramps (5 C/min).
2
3
2 3

258
K. Arnby et al. / Journal of Catalysis 221 (2004) 252–261
Pt/1% Al2O3 catalyst. The low-temperature activity is high-
est for Pt/1% Al2O3 and lowest for Pt/100% Al2O3. The
T50,light-off is 126 ◦C for Pt/100% Al2O3, 113 ◦C for Pt/10%
Al2O3, and 102 ◦C for Pt/1% Al2O3. The corresponding
T50,extinction is 117 ◦C for Pt/100% Al2O3, 94 ◦C for Pt/10%
Al2O3, and 72 ◦C for the Pt/1% Al2O3 sample.
mesoporous catalysts rapid reactions such as CO oxidation
can be diffusion limited in a single pore [27] due to slow
diffusion. Mass transfer limitations will reduce the reaction
rate for reactions with positive order while it will increase
the rate of reactions with negative reaction order. The kinet-
ics for CO oxidation over Pt can be divided in two regimes
since it is self-inhibited [28]: at high CO concentrations the
reaction order is negative (−0.62), while at low concentra-
tions it follows first-order reaction. The breakpoint between
these two regions is when the partial pressure of CO is about
13 Pa [28]. Other studies [19,29] have shown that this break-
point is not fixed, but varies with temperature. According to
Weisz and Prater [30], mass transfer will affect a first-order
reaction for Weisz-modulus values equal or higher than one.
Reactions with negative reaction orders are influenced by
mass transfer at Φ ꢀ |n|−1 [31].
4. Discussion
The results from the CO oxidation experiments for the
three catalysts with Pt deposited with different local concen-
trations on the available alumina support are presented in
Figs. 1–3. The graphs show clearly that there are significant
differences in activity between the catalysts. The catalyst
with Pt deposited on only 1% of the total amount of alumina
support, Pt/1% Al2O3, shows the highest low-temperature
activity for both heating and cooling ramps. The conven-
tional catalyst, with Pt deposited on the entire amount of alu-
mina support, Pt/100% Al2O3, has the lowest activity, while
the catalyst with Pt deposited on 10% of the alumina support,
Pt/10% Al2O3, shows intermediate activity for CO oxida-
tion. The hysteresis, the difference between the light-off and
extinction temperatures, is smallest for Pt/100% Al2O3, in-
creases for Pt/10% Al2O3, and is highest for Pt/1% Al2O3.
Complementary to the experiments in this study, three addi-
tional catalysts were prepared with constant Pt loading and
tested by oxidizing 0.2% CO in air. This test also showed
that the catalyst prepared with high local Pt density had the
highest activity, while the conventionally prepared catalyst
had the lowest activity for CO oxidation.
It is obvious that the sample with highest local Pt density,
Pt/1% Al2O3, shows the highest low-temperature activity. In
contrast, the lowest activity for CO oxidation is found for
the conventionally prepared sample, Pt/100% Al2O3. Since
the amount of platinum surface area between the catalysts is
constant, the differences in activity for CO oxidation most
be due to the deposition and distribution of the platinum in
the wash coat. To explain these results, we have considered
three effects. One possible explanation is that mass transfer
of reactants to the active Pt sites affects the activity. An-
other is that heat transfer at the active sites in the catalysts
differs. Even if it is uncertain whether CO oxidation is a
structure-sensitive reaction [17,18], this can also be a possi-
ble explanation for the differences in activity since the mean
size of the Pt crystallites in the three samples varies.
For spherical Pt/Al2O3 particles is the Weisz modulus
given by
rp2rv
Φ =
,
(2)
Deffcwc
where rp is the radius for the Pt/Al2O3 particles, Deff the ef-
fective diffusion of CO, cwc the concentration of CO in the
wash coat, and rv is the reaction rate per active catalyst vol-
ume. Expressing as usual Deff via the bulk [32] and Knudsen
diffusion coefficient [33] and using rp = 1.25 µm (estima-
tion using scanning electron microscopy), at 50% conver-
sion we have Φ = 0.12 (Pt/1% Al2O3), Φ = 17 × 10−3
(Pt/10% Al2O3), and Φ = 2.4 × 10−3 (Pt/100% Al2O3).
None of these values are close to the regime where mass
transport limitations affect the reaction rate (at Φ ∼ 0.6 or
higher). The figures used for the calculation were bulk tem-
perature, 500 K; pore diameter, 75 Å; and total flow rate,
1.67 × 10−5 m3/s. The active catalyst volume was obtained
by multiplying the wash-coat mass (see Table 1) with the
fraction which was Pt-impregnated, divided by the density
(1500 kg/m3).
When considering the entire wash coat, instead of one
Pt/Al2O3 particle, mass transfer limitations could affect the
reaction rate due to a long diffusion distance from the bulk
phase to the catalyst surface. The Weisz modulus for a planar
layer (wash coat) with the depth, δwc, is expressed as
δw2 crwc
Φwc
=
,
(3)
Deffcwc
where rwc is the reaction rate for all the wash-coat mate-
rial, irrespective if Pt is locally distributed or not. Using
the same figures as above results in Φwc = 0.60 (catalyst
4.1. Difference in mass transfer
Pt/1% Al2O3), Φwc = 0.38 (Pt/10% Al2O3), and Φwc
=
A typical criterion if the catalytic activity is affected by
mass transfer of reactants to the active sites of the catalyst
is that the net transport effect should alter the true chemical
rate by more that 5% [26]. This criterion can be determined
calculating the Weisz modulus, Φ = ηφ2 (η is the effec-
tiveness factor and φ the Thiele modulus), which compares
the reaction rate versus the diffusion of the reactants. For
0.31 (Pt/100% Al2O3). These results, where the entire wash
coat is regarded, indicate that the experiments performed in
this study are close to being mass transfer limited at 50%
conversion. As the Weisz modulus increases with increas-
ing conversion, mass transfer most probably does affect the
activity for CO oxidation at higher conversions, especially

K. Arnby et al. / Journal of Catalysis 221 (2004) 252–261
259
during the beginning of the extinction region. In correspon-
dence with Zhdanov and Kasemo [27] diffusion limitation
will provide a higher reaction rate for CO oxidation because
the self-inhibition regime will be postponed. On the contrary,
at low conversions, such as during the start of the light-off
region it seems unlikely that mass transfer influences the
reaction. This can explain why the differences in activity be-
tween the catalysts are larger for the extinctions compared
to the light-off processes.
4.2. Difference in heat transfer
The differences in activity between the catalysts could
be due to differences in heat transfer. First we will ana-
lyze whether transient temperature gradients between the
active sites and the support material could affect the low-
temperature activity for the catalysts. Both Luss [3], and
Steinbrüchel and Schmidt [6] report that heat evolved from
exothermic catalytic reactions can raise temporarily the tem-
perature of the active sites several hundred degrees Celsius
above the temperature of the support material. Luss finds the
Fig. 4. Model of heat transfer for a spherical Pt/Al O particle adjacent to
a stagnant gas phase.
2
3
For a stagnant gas phase the Nusselt number is 2 (Nu =
2rph/λ = 2, where h is the heat transfer coefficient and λ
the thermal conductivity). The generated heat flux, qg, and
the heat loss from the Pt/Al2O3 particle by conduction, qc,
is
temporary temperature raise only to last in order of 10−13
s
qg = r ꢀH,
(4)
(5)
before the active site is cooled down. However, Steinbrüchel
and Schmidt report longer periods before the temperature
declines, up to 10−10 s. In our study the number of CO mole-
cules reacted per second and per surface atom of Pt (turnover
frequency) is between 0.01 and 10 at complete conversion of
CO (depending on the CO concentration in the feed), which
is somewhat higher than under typical laboratory condi-
tions [7]. The average time period between two consecutive
reactions at one active site is then at maximum 0.1 s, which
is a tremendously longer time period than the period for
a transient temperature rise according to Steinbrüchel and
Schmidt. We conclude that the transient temperature rises
is too instantaneous, compared to the turnover frequency, to
have a possibility of affecting the low-temperature activity
of the catalysts.
qc = hSp ꢀT,
where r is the reaction rate per Pt/Al2O3 particle, ꢀH the
reaction heat, Sp the area of the particle, and ꢀT the temper-
ature gradient between Pt/Al2O3 particles and the surround-
ings. The reaction rate is estimated by dividing the amount
of CO reacted per second with the number of Pt/Al2O3 par-
ticles in the catalyst, np (the number of Pt/Al2O3 particles
is trivial to calculate knowing the particle radius as well as
the mass and density of the wash coat). At steady state, the
generated heat is equal to the heat loss, resulting in
FCO convꢀH
ꢀT =
,
(6)
4πrpnpλ
where FCO is the flow rate of CO in the feed and conv
the conversion. The highest temperature gradient generated
in the experimental study was while oxidizing 1% CO at
100% conversion. Using FCO = 6.83 × 10−6 mol/s, ꢀH =
283 kJ/mol, and λ = 3.86 × 10−2 W/mK [34] (dry air at
200 ◦C) results in ꢀT = 13 × 10−3 ◦C (for catalyst Pt/1%
Al2O3), ꢀT = 2.0 × 10−3 ◦C (Pt/10% Al2O3) and ꢀT =
0.26 × 10−3 ◦C (Pt/100% Al2O3). For the experiments with
lower CO concentrations (1000 and 100 ppm) the corre-
sponding ꢀT for each catalyst will be one and two mag-
nitudes lower, respectively. Instead of estimating the tem-
perature rise for a Pt/Al2O3 particle, the same model can
be used to calculate the temperature rise for a Pt crystallite.
Elementary estimates with the parameters corresponding to
our system indicate that in this case ꢀT is negligibly small
(< 10−5 ◦C). These calculations clearly show that the reac-
tion heat evolved in our experiments cannot heat Pt/Al2O3
particles or Pt crystallites sufficiently enough to produce sig-
nificant differences in activity between the catalysts.
Instead of discussing the transient heat effects at one spe-
cific active site, heat from exothermic reactions could accu-
mulate in Pt/Al2O3 particles causing constant temperature
gradients between Pt/Al2O3 particles and the surroundings.
The generated heat is transferred by conduction to the gas
phase and to the surrounding particles in the wash coat. Hot
areas may thus occur on the level of an array of alumina
particles. However by using a heat transfer model similar to
Holstein’s and Boudart’s [7], we will illustrate an extreme
case and calculate the maximum temperature gradient be-
tween spherical Pt/Al2O3 particles and the unimpregnated
alumina support. Even though heat transfer between parti-
cles in the wash coat is expected to dominate, the Pt/Al2O3
particle in this model is assumed to be isolated from the
support material and only adjacent to a stagnant gas phase,
see Fig. 4. This condition minimizes heat transfer from the
Pt/Al2O3 particle and thus overestimates the temperature
gradient. The gas bulk and the unimpregnated alumina sup-
port are assumed to have equal temperature.

260
K. Arnby et al. / Journal of Catalysis 221 (2004) 252–261
Interesting to note from the experiments is the hys-
cle size at high CO concentrations. Though the results from
the activity tests in this investigation can be explained in
terms of structure sensitivity, it is still very doubtful. Even
if McCarthy and Zafiris insist, there is still a disagreement
whether or not CO oxidation over Pt/Al2O3 is a structure-
sensitive reaction. Furthermore, in the experiments reported
in the literature the range of crystallite sizes is much wider
(from some nm up to 100 nm) compared to the narrow range
(2.3–4.5 nm) in our study. It seems unlikely that the rela-
tively small difference in particle size really causes the large
difference in activity observed in this investigation. Espe-
cially the difference in activity between the Pt/100% Al2O3
and the Pt/10% Al2O3 catalyst is difficult to describe in
terms of structure sensitivity since the estimated crystallite
sizes are so similar (2.3 and 2.7 nm, respectively).
teresis in temperature between the ignition and extinction
processes. For exothermic reactions this phenomenon usu-
ally takes place due to the evolved reaction heat. In our
experiments the hysteresis is ranging between 55 and 84 ◦C
when oxidizing 1% CO. Surprisingly, for 100 ppm CO the
hysteresis is still marked (9–17◦C) and this behavior cannot
be explained as a heat effect since the amount of generated
heat is insignificant. This hysteresis is probably an effect
from the self-inhibition of CO. From the start of the igni-
tion process, the surface is covered by CO preventing the
reaction to take off, causing a shift in the light-off to higher
temperatures. Since there is no self-inhibition during the ex-
tinction, the hysteresis appears as the reaction can proceed
at lower temperatures before being quenched.
4.3. Difference in specific activity and structure sensitivity
5. Concluding remarks
In this study CO has been oxidized over Pt/γ -Al2O3 cata-
lysts. One aspect worth noting for this system is that there is
no general agreement whether CO oxidation over noble met-
als is structure insensitive or the reaction rate increases with
increasing noble metal crystallite size [17,18].
Pt/γ -Al2O3 catalysts prepared with locally higher Pt con-
centrations in the wash coat showed a considerably higher
activity for CO oxidation at low temperatures compared to a
conventionally prepared Pt/γ -Al2O3 catalyst with homoge-
neous Pt concentration in the alumina support.
Using Pt/α-Al2O3 catalysts with crystallite sizes ranging
from 2.8 to 100 nm, McCarthy et al. [19] showed that CO
oxidation is structure sensitive over Pt/Al2O3 at low CO con-
centrations (< 2000 ppm), while it is structure insensitive at
high concentrations (> 1% CO). The authors speculate that
PtO forms and the structure sensitivity is because the oxygen
in PtO may be more easily extracted from larger crystallites
than from smaller ones. These speculations are supported by
several studies, reporting that platinum oxide can be formed
at adequate temperatures [35–39], that oxygen is more eas-
ily reduced from larger PtO crystallites [35,40], and also that
reduced Pt is more active for CO oxidation than oxidized
platinum [35,41].
Zafiris and Gorte [20] studied CO oxidation over Pt/α-
Al2O3 for Pt crystallites with average diameters of 14 and
1.7 nm. The authors concluded that CO oxidation is structure
sensitive even at high CO concentrations. The result was ex-
plained by CO desorption occurring more easily from large
Pt crystallites, due to less curvature of the Pt particles, and
this desorption controls the oxidation rate since it opens up
the surface for O2 adsorption.
Three reasons for the differences in activity between the
catalysts are discussed: differences in mass transfer, differ-
ences in heat transfer, and structural differences between the
catalysts. Mass transfer limitations were considered both for
Pt/Al2O3 particles and for the entire wash coat of the cata-
lysts. When considering heat transfer both temporary effects
and heat accumulation in the active phase were taken into
account. Differences in heat transfer appear to be the least
probable reason for the enhanced activity for the catalysts
with locally higher Pt concentration, whereas structural ef-
fects also seem to be an unlikely explanation. Differences in
mass transfer seem, however, to be a more likely reason for
the enhanced activity for the catalysts with platinum distrib-
uted locally in the wash coat.
To distribute Pt locally in the wash coat may also give
positive catalytic consequences in other research areas than
low-temperature oxidation of CO. By using a broader ap-
proach it could be possible to find the explanation to the
improved activity.
In our study, the Pt/1% Al2O3 catalyst with largest
Pt crystallites (4.5 nm) showed the highest activity while
the Pt/100% Al2O3 catalyst with smallest Pt crystallites
(2.3 nm) showed the lowest activity. The results from all
light-off and quenching experiments, except for the light-
off test with 1% CO, seem to be in accordance with the
structure sensitivity reasoning in which larger Pt crystallites
have higher low-temperature activity for CO oxidation than
smaller ones. In the light-off experiment with 1% CO the
activity is very similar for all three catalysts. This specific
experiment, however, coincides with the results of McCarthy
et al. [19] in which the activity is independent of the parti-
Acknowledgments
Prof. Bengt Kasemo, Dr. Erik Fridell, and Dr. Ann Grant
are gratefully acknowledged for valuable discussions. This
work has been performed within the Competence Centre for
Catalysis, which is financially supported by The Swedish
Energy Agency and the member companies: AB Volvo,
Johnson Matthey-CSD, Perstorp AB, Saab Automobile AB,
AVL-MTC AB, Akzo Catalyst, and the Swedish Space Ad-
ministration.

K. Arnby et al. / Journal of Catalysis 221 (2004) 252–261
261
References
[23] P. Lööf, B. Kasemo, S. Andersson, A. Frestad, J. Catal. 130 (1991)
181.
[24] J.R. Anderson, K.C. Pratt, Introduction to Characterization and Testing
of Catalysts, Academic Press, North Ryde, Australia, 1985.
[1] G. Damköhler, Z. Phys. Chem. 193 (1943) 16.
[2] J. Wei, Chem. Eng. Sci. 21 (1966) 1171.
[3] D. Luss, Chem. Eng. J. 1 (1970) 311.
[25] R.J.H. Grisel, J.J. Slyconish, B.E. Niewenhuys, Top. Catal. 16/17
(2001) 425.
[4] S.W. Chan, M.J.D. Low, W.K. Mueller, AIChE J. 17 (1971) 1499.
[5] E. Ruckenstein, C.A. Petty, Chem. Eng. Sci. 27 (1972) 937.
[6] C. Steinbruchel, L.D. Schmidt, Surf. Sci. 40 (1973) 693.
[7] W.L. Holstein, M. Boudart, Lat. Am. J. Chem. Eng. Appl. Chem. 13
(1983) 107.
[8] J.A. Cusumano, M.J.D. Low, J. Catal. 17 (1970) 98.
[9] H. Mark, M.J.D. Low, J. Catal. 30 (1973) 40.
[10] D.R. Kember, N. Sheppard, J. Chem. Soc., Faraday Trans. 2 77 (1981)
1321.
[26] J.B. Butt, Reaction Kinetics and Reactor Design, 2nd ed., Dekker, New
York, 2000.
[27] V.P. Zhdanov, B. Kasemo, Catal. Lett. 72 (2001) 7.
[28] J. Wei, Adv. Catal. 24 (1975) 57.
[29] A. Drewsen, A. Ljungqvist, M. Skoglundh, B. Andersson, Chem. Eng.
Sci. 55 (2000) 4939.
[30] P.B. Weisz, C.D. Prater, Adv. Catal. 6 (1954) 143.
[31] D.E. Mears, Ind. Eng. Chem., Proc. Des. Dev. 10 (1971) 541.
[11] T.S. Cale, J. Catal. 90 (1984) 40.
[32] C.N. Satterfield, Heterogeneous Catalysis in Industrial Practice, 2nd
ed., McGraw–Hill, New York, 1991.
[12] R.J. Matyi, J.B. Butt, L.H. Schwartz, J. Catal. 91 (1985) 185.
[13] G. Georgiades, V.A. Self, P.A. Sermon, Angew. Chem. Int. Ed. 26
(1987) 1042.
[14] S. Sharma, D. Boecker, J. Maclay, R.D. Gonzalez, J. Catal. 110 (1988)
103.
[33] B.E. Poling, J.M. Prausnitz, J.P. O’Connell, The Properties of Gases
and Liquids, 5th ed., McGraw–Hill, New York, 2001.
[34] G. Hellsten, in: A. Bernhardsson (Ed.), Tabeller och Diagram, Energi-
och Kemiteknik, Almqvist & Wiksell, Falköping, 1992, p. 33.
[15] J.C. Frost, P. Meehan, S.R. Morris, R.C. Ward, Catal. Lett. 2 (1989)
97.
[35] R.W. McCabe, C. Wong, H.S. Woo, J. Catal. 114 (1988) 354.
[16] J. Kapicka, M. Marek, J. Catal. 119 (1989) 508.
[17] M. Che, C.O. Bennet, Adv. Catal. 36 (1989) 55.
[18] V.P. Zhdanov, B. Kasemo, Surf. Sci. Rep. 39 (2000) 25.
[19] E. McCarthy, J. Zahradnik, C. Kuczynski, J.J. Carberry, J. Catal. 39
(1975) 29.
[20] G.S. Zafiris, R.J. Gorte, J. Catal. 140 (1993) 418.
[21] A. Holmgren, B. Andersson, J. Catal. 178 (1998) 14.
[22] M. Skoglundh, H. Johansson, L. Löwendahl, K. Jansson, L. Dahl, B.
Hirschauer, Appl. Catal. B 7 (1996) 299.
[36] A. Borgna, F. Le Normand, T. Garetto, C.R. Apesteguia, B. Moraweck,
Catal. Lett. 13 (1992) 175.
[37] N. Hartmann, R. Imbihl, W. Vogel, Catal. Lett. 28 (1994) 373.
[38] B.L.M. Hendriksen, J.W.M. Frenken, Phys. Rev. Lett. 89 (2002)
046101.
[39] V.P. Zhdanov, B. Kasemo, Surf. Sci. Rep. 45 (2002) 231.
[40] E.S. Putna, J.M. Vohs, R.J. Gorte, Surf. Sci. 391 (1997) 1178.
[41] Y. Barshad, X. Zhou, E. Gulari, J. Catal. 94 (1985) 128.