Exploiting the synergy of titania and alumina in lean NOx reduction: In situ ammonia generation during the Pd/TiO2/Al 2O3-catalysed H2/CO/NO/O2 reaction

Article abstract of DOI:10.1016/j.jcat.2003.07.005

In situ DRIFTS, XPS, HREM, XRD, and reactor studies have been used to examine and elucidate the catalytic performance of Pd particles supported on TiO₂/Al₂O₃. These materials deliver 100% NO_x conversion under demanding and technically relevant conditions (low temperature, high oxygen excess) in the presence of mixed H₂+CO reductant feeds. In particular, they are far superior to the corresponding Pd/TiO₂ and Pd/Al₂O₃ catalysts. It is shown that the synergy that operates between the titania and the alumina components involves the intrinsic surface chemistry of these oxides rather than formation of mixed oxide phases. Specifically, titania is critical for the formation of NCO on Pd while alumina promotes subsequent hydrolysis of NCO to ammonia, which then reduces NO_x. At high temperature a second NO_x reduction channel operates with Pd/TiO₂/Al₂O₃ which greatly extends the useful temperature range. This also involves in situ formation of ammonia - in this case directly from reaction between H ₂and NO.

Full text of DOI:10.1016/j.jcat.2003.07.005

Journal of Catalysis 221 (2004) 20–31
www.elsevier.com/locate/jcat
Exploiting the synergy of titania and alumina in lean NOx reduction:
in situ ammonia generation during the Pd/TiO2/Al2O3–catalysed
H2/CO/NO/O2 reaction
Norman Macleod, Rachael Cropley, James M. Keel, and Richard M. Lambert ∗
Department of Chemistry, University of Cambridge, Lensfield Rd, Cambridge CB2 1EW, UK
Received 17 April 2003; revised 11 July 2003; accepted 21 July 2003
Abstract
In situ DRIFTS, XPS, HREM, XRD, and reactor studies have been used to examine and elucidate the catalytic performance of Pd par-
ticles supported on TiO /Al O . These materials deliver 100% NO conversion under demanding and technically relevant conditions (low
x
2
2 3
temperature, high oxygen excess) in the presence of mixed H + CO reductant feeds. In particular, they are far superior to the corresponding
2
Pd/TiO and Pd/Al O catalysts. It is shown that the synergy that operates between the titania and the alumina components involves the in-
2
2 3
trinsic surface chemistry of these oxides rather than formation of mixed oxide phases. Specifically, titania is critical for the formation of NCO
on Pd while alumina promotes subsequent hydrolysis of NCO to ammonia, which then reduces NO . At high temperature a second NO
x
x
reduction channel operates with Pd/TiO /Al O which greatly extends the useful temperature range. This also involves in situ formation of
2
2 3
ammonia—in this case directly from reaction between H and NO.
2
2003 Elsevier Inc. All rights reserved.
Keywords: NO; Palladium; CO; H ; Lean burn; Al O ; TiO ; NH ; DRIFTS
2
2
3
2
3
1. Introduction
conditions. Although this process is very effective, the re-
quirement for a NH3 reservoir and injection system makes
use of this technology problematic for transport applications.
Safer alternatives such as urea [5–7], which can be used to
generate NH3 in situ, add further complexity and cost.
Recently we identified a number of palladium-based cata-
lysts that generate NH3 in situ when fed with H2/CO/NO
mixtures in the presence of very high oxygen concentra-
tions. Hydrogen and CO are both frequently encountered in
the exhaust streams produced by hydrocarbon combustion
processes; therefore, this strategy could potentially remove
the requirement for injection of additional reductant into the
exhaust. Two separate systems capable of in situ NH3 gen-
eration have been identified.
Thus Pd/Al2O3 catalysts fed with H2 + CO + NO + O2
reactant streams at relevant partial pressures exhibited much
improved performancecompared to the results obtained with
either H2 or CO used alone [8–10]. In situ DRIFTS stud-
ies revealed that this improvement in the NOx conversion
was attributable to the generation of ammonia formed via the
production and subsequent hydrolysis of isocyanate (NCO)
species [9]. This increase in NOx conversion obtained in
the presence of mixed H2 + CO reductant feeds is signifi-
The exhaust streams generated by diesel and lean burn
gasoline engines and stationary power plants contain very
high background oxygen levels, making effective utilisa-
tion of any reductant species present for catalytic reduction
of the nitrogen oxides produced extremely difficult [1–4].
The environmental consequences arising due to release of
these NOx (NO + NO2) species have been well documented
(acid rain, photochemical smog, depletion of stratospheric
ozone). The lack of an effective catalyst/reductant combina-
tion for this process is one of the main obstacles preventing
widespread introduction of fuel-efficient lean burn engines,
which have the potential to significantly reduce the global
CO2 burden.
For stationary power sources and chemical plants the
NH3 SCR process [3,4] is currently the preferred option
for NOx abatement. Ammonia is utilised due to its unique
high activity toward NOx reduction under these oxygen-rich
*
Corresponding author.
E-mail address: rml1@cam.ac.uk (R.M. Lambert).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2003.07.005
2003 Elsevier Inc. All rights reserved.

N. Macleod et al. / Journal of Catalysis 221 (2004) 20–31
21
cant as the H2 and CO concentrations in exhaust streams are
closely linked due to equilibration via the water gas-shift re-
action.
formed on a VSW ARIES system equipped with a HA-100
hemispherical analyser. Catalyst samples were mounted by
pressing a small amount of the material between two disks
of pure aluminum. Separating the disks produced two speci-
mens consisting of a thin film of catalyst powder adhering to
the aluminum. XP spectra were acquired with Mg-Kα radi-
ation and quoted binding energies (BEs) are referred to the
C 1s emission at 284 eV. The different contributions to the
Ti 2p3/2,1/2 emission were extracted by curve fitting.
Pd/TiO2 catalysts, on the other hand, display very high
activity for the reduction of lean NOx in the absence of CO,
i.e., with H2 + NO + O2 mixtures [11–13]. These catalysts
are active in two distinct temperature regimes, indicating
operation of two different mechanisms for NOx reduction.
By means of in situ DRIFTS we were able to attribute the
lower temperature (∼110 ◦C) NOx reduction channel to a
mechanism involving dissociation of NO on reduced metal
(Pd0) sites [13]. The second, higher temperature (∼250 ◦C),
NOx reduction channel operated via formation and subse-
quent reaction of NH3 [13], in this case generated directly
by reaction of H2 with NO. While the low temperature
channel operated over a rather narrow temperature region
(∼100–130◦C), the NH3-driven channel maintained NOx
conversions >50% over a much wider temperature range
(200–280◦C).
Here we are concerned with the effects of CO. Specif-
ically, we report on the H2 + CO + NO + O2 reaction
over Pd/TiO2 and Pd/TiO2/Al2O3 catalysts. Our aim was
to identify the reason for the improved performance of
the mixed support catalyst Pd/TiO2/Al2O3 [14] compared
to Pd/TiO2 and Pd/Al2O3 under the same conditions. De-
tailed characterisation of the TiO2/Al2O3 support has been
performed which, combined with in situ DRIFTS results,
has enabled us to elucidate aspects of the reaction mecha-
nism.
Catalyst testing was performed in a quartz microreactor
system described previously [15]. The feed composition em-
ployed contained 5% O2 and 500 ppm NO in all cases with
a total reductant concentration of 4000 ppm, delivered to the
reactor with a total flow of 200 ml min−1. One hundred mil-
ligrams of sample was employed, corresponding to a recip-
rocal weight time velocity of w/f = 0.03 g s ml−1. Prior
to testing the samples were calcined in air (60 ml min−1
)
for 6 h at 500 ◦C. The reactor outflow was analysed using a
chemiluminescence NOx (NO+NO2) detector (Signal 4000
series) and two dual-channel NDIR detectors (Siemens Ul-
tramat 6) calibrated for CO/CO2 and NO/N2O. Hydrogen
consumption was monitored via a quadrupole mass spec-
trometer (Hiden RGA 301). Nitrogen production was cal-
culated by subtracting the N2O contribution from the to-
tal NOx conversion (GC analysis produced no evidence for
NH3 in the reactor outlet during these experiments). The cat-
alyst temperature and all analyser outputs were continuously
monitored and recorded by PC-based software. Conversion
profiles, typically containing 1000 data points per channel,
were obtained as the catalyst temperature was raised from 50
to 450 ◦C with a linear ramp of 2 ◦C min−1. These runs were
repeated several times in order to ensure stable and repro-
ducible performance. Temperature-programmed desorption
(TPD) experiments were also performed in this system, em-
ploying a He flow of 150 ml min−1 with a heating rate of
10 K min−1. The sample (100 mg) was initially heated to
500 ◦C in He, cooled to 50 ◦C, and subsequently dosed with
a feed containing 500 ppm NO + 5% O2. The TPD ramp
was then performed between 50 and 550 ◦C. Desorption of
NO/N2O was monitored by NDIR, NO2 by chemilumines-
cence, and O2/N2 by mass spectrometry.
2. Experimental
Mixed TiO2/Al2O3 supports, with varying TiO2 loadings,
were prepared by hydrolysis of titanium isopropoxide in the
presence of a γ -Al2O3 support (Aldrich, 120 m2/g). The
resulting solid materials were dried and calcined in flow-
ing air for 10 h at 500 ◦C. Palladium was subsequently
impregnated by incipient wetness using aqueous solutions
of palladium(II)nitrate (Aldrich) to yield 0.5 wt% metal
loading. Following impregnation the catalysts were dried in
air overnight at 110 ◦C, calcined in air at 500 ◦C for 6 h,
and subsequently crushed/sieved to yield grain sizes in the
range 255–350 µm. The 0.5 wt% Pd/γ -Al2O3 and 0.5 wt%
Pd/TiO2 (Degussa P25) catalysts were prepared in a simi-
lar manner. Metal dispersion was determined using the CO
methanation technique [15]. Prior to this measurement the
samples were calcined in air (60 ml min−1) for 6 h at 500 ◦C
followed by reduction in H2 (60 ml min−1) for 1 h at 150 ◦C.
Metal dispersions were calculated assuming a 1:1 CO to sur-
face metal atom ratio.
DRIFTS experiments were performed with a Perkin-
Elmer GX2000 spectrometer equipped with an MCT de-
tector and a high-pressure, high-temperature DRIFTS cell
(Thermo Spectra-Tech) fitted with ZnSe windows. Spectra
were acquired at a resolution of 4 cm−1 typically averaging
64 scans. Background spectra were obtained from samples at
the relevant temperature in a flow of helium, following iden-
tical pretreatment to that utilised in the reactor experiments.
All spectra were taken after 1 h in the relevant gas mix unless
otherwise stated. The flow rate through the DRIFT cell was
varied so as to maintain conversions in the range 10–20%.
Therefore the spectra displayed correspond to the state of
the catalyst toward the front end of the sample bed in the
tubular reactor.
Powder XRD patterns were obtained on a Philips PW1710
instrument employing Cu-Kα radiation. Transmission elec-
tron microscopy was performed on a JOEL 3011 microscope
operating at 300 kV and equipped with a Princeton Gamma-
Tech energy dispersive X-ray spectrometer. XPS was per-

22
N. Macleod et al. / Journal of Catalysis 221 (2004) 20–31
Table 1
Palladium dispersion
a
Sample
CO uptake
Dispersion
(%)
(µmol/g
)
cat
0.5 wt% Pd/Al O
22.0
14.1
19.2
18.3
12.7
47
30
41
39
27
2
3
0.5 wt% Pd/TiO
2
0.5 wt% Pd/2.5 wt% TiO /Al O
0.5 wt% Pd/10 wt% TiO /Al O
0.5 wt% Pd/25 wt% TiO /Al O
2
2 3
2
2 3
2
2 3
a
Assumes 1:1 CO to surface metal atom ratio.
3. Results
3.1. Catalyst characterisation
Metal dispersion data for the Al2O3, TiO2, and various
TiO2/Al2O3-supported palladium samples are listed in Ta-
ble 1. In general the metal dispersion varied in line with the
total surface area, with the alumina-supportedsample having
a higher dispersion (47%) than that of TiO2 (30%). For the
mixed support catalysts the palladium dispersion decreased
as the TiO2 loading increased.
Fig. 1. Powder XRD patterns obtained from Pd/TiO /Al O catalysts with
varying TiO loadings following calcination in air at 500 C for 6 h.
2
2
2 3
◦
X-ray diffraction patterns obtained from Pd/TiO2/Al2O3
catalysts with varying TiO2 loadings are shown in Fig. 1.
These data were obtained after calcination of the samples
in air at 500 ◦C for 6 h. The sample containing 2.5 wt%
TiO2 displayed a pattern identical to that of the γ -Al2O3
support, indicating that the TiO2 component was highly dis-
persed over the surface of the alumina. The same is true of
the 10 wt% TiO2/Al2O3 sample, although in this case weak
reflections due to the anatase form of TiO2 were also visible.
On increasing the loading further to 25 wt% the intensity of
the anatase reflections increased considerably, indicating an
increase in the number and size of crystalline TiO2 particles
(line broadening analysis gave an average TiO2 particle size
of 10 nm for this sample). The 10 wt% TiO2/Al2O3 sample
was found to yield the best catalytic performance, and was
therefore characterised in more detail by HREM/EDX, XPS,
and TPD.
HREM images obtained from this sample are shown in
Fig. 2. A number of TiO2 particles in the size range 2–5 nm
were imaged, examples of which are shown. However, EDX
analysis indicated that a significant proportion of the TiO2
was present in a highly dispersed form, as spectra obtained
from areas containing no visible TiO2 particles (utilising a
relatively large electron beam probe diameter, ∼50 nm) fre-
quently displayed significant Ti peak intensity.
XPS spectra (Ti 2p) obtained from the 0.5 wt% Pd/TiO2
and 0.5 wt% Pd/10 wt% TiO2/Al2O3 samples are com-
pared in Figs. 3a and 3b, respectively. In the case of the
Pd/TiO2 sample, the peak found at 459.2 eV is assigned
to Ti4+. The corresponding oxygen peaks (not shown) were
resolved into two species, one at 530.8 eV and a smaller
peak a 531.6 eV, corresponding to oxygen in the bulk tita-
nia and oxygen from surface hydroxyl groups, respectively.
Fig. 2. HREM images obtained from the Pd/10 wt% TiO /Al O catalyst showing presence of TiO particles in the 2–5 nm size range.
2
2
3
2

N. Macleod et al. / Journal of Catalysis 221 (2004) 20–31
23
Fig. 3. Ti 2p XP spectra obtained from (a) Pd/TiO and (b) Pd/10 wt%
2
TiO /Al O samples.
2
2 3
The Ti(2p) peak obtained from the 10 wt% TiO2/Al2O3 sam-
ple, shown in Fig. 3b, was significantly broader than that
observed with Pd/TiO2, indicating the presence of multiple
oxidation states. Curve fitting indicates the presence of Ti2+
and Ti3+ in addition to the dominant (90% of the total sig-
nal) Ti4+ peak. The Ti:Al surface atomic ratio determined by
XPS was estimated as ∼1:3 for this sample. This is further
evidence for the presence of a highly dispersed TiO2 com-
ponent, as the number density of 3D TiO2 particles imaged
by HREM was insufficient to give such a high Ti:Al ratio.
Temperature-programmed desorption spectra obtained
from the 0.5 wt% Pd/Al2O3, 0.5 wt% Pd/TiO2 (P25) and
0.5 wt% Pd/10 wt% TiO2/Al2O3 samples following expo-
sure to NO + O2 (500 ppm NO + 5% O2) at 50 ◦C are
shown in Figs. 4a, 4b, and 4c, respectively. (The Pd/Al2O3
and Pd/TiO2 profiles have been published previously [13]
and are included here for the purpose of comparison with
Pd/TiO2/Al2O3.) The desorption spectra obtained from the
Al2O3 supported catalyst are dominated by two regimes, one
centred at 150 ◦C and one at 390 ◦C. The low-temperature
peaks are due to NO and NO2, while the high-temperature
peaks are due to NO, NO2, and O2. A further small NO2
desorption peak is visible at 290 ◦C. On the TiO2 supported
sample, Fig. 4b, two NO2 desorption peaks were again ob-
served (220 and 340 ◦C) along with an O2 peak at 340 ◦C.
A small NO desorption peak appeared at 150 ◦C but no
Fig. 4. Temperature-programmed desorption spectra obtained from
(a) Pd/Al O , (b) Pd/TiO , and (c) Pd/10 wt% TiO /Al O catalysts fol-
2
3
2
2
2 3
◦
lowing exposure to NO + O at 50 C. [NO] = 500 ppm, [O ] = 5%.
2
2
high-temperature NO desorption was observed in this case.
However, the most important point to note from this TPD
data is that in the case of the mixed TiO2/Al2O3 support,
Fig. 4c, the desorption profiles obtained contain a combi-
nation of the features observed on the individual oxides.
For example, the high-temperature NO peak at 420 ◦C is
characteristic of alumina while the NO2 desorption feature
at 230 ◦C is associated with the TiO2 component.
Corresponding DRIFT spectra were obtained at 50 K
intervals. Fig. 5 displays an extract from this data set,
showing spectra obtained from the Pd/TiO2, Pd/10 wt%
TiO2/Al2O3, and Pd/Al2O3 samples at 250 ◦C. These spec-
tra are simplified due to desorption of the majority of the
nitrite species present at lower temperatures. Although sig-
nificant overlap of various nitrate bands is still present, it is
clear that the mixed support displays features characteristic
of both constituent oxides. For example the intense peak at
∼1618 cm−1 observed on both titania-containing samples
is assigned to the asymmetric stretching mode of a bridging
nitrate species adsorbed on TiO2 sites [16–18]. The corre-

24
N. Macleod et al. / Journal of Catalysis 221 (2004) 20–31
Fig. 5. DRIFT spectra obtained with various supported Pd catalyst follow-
◦
◦
ing exposure to NO + O flow at 50 C and subsequent heating to 250 C
2
in flowing He. [NO] = 500 ppm, [O ] = 5%.
2
sponding symmetric stretch occurs at ∼1230 cm−1. On the
other hand the peak at 1303 cm−1 observed on alumina-
containing samples is associated with the symmetric stretch-
ing mode of a bidentate nitrate adsorbed on alumina [19,20].
In this case the asymmetric stretch occurs at ∼1563 cm−1
.
This observation that the surface of the mixed support con-
tains adsorption sites characteristic of both alumina and ti-
tania components is important in relation to the catalytic
properties discussed below.
Fig. 6. Conversion profiles for (a) hydrogen and (b) NO as a function of
x
temperature over various palladium-containing catalysts. [H ] = 4000 ppm,
2
−1
[NO] = 500 ppm, [O ] = 5%, w/f = 0.03 g s ml
.
2
3.2. NO reduction by H2
Conversion versus temperature profiles obtained during
the reduction of NO by hydrogen (4000 ppm H2 + 500 ppm
NO + 5% O2) over the various supported palladium cata-
lysts are shown in Fig. 6. As found previously [10,11],
the Pd/Al2O3 catalyst displayed very low activity under
these conditions, with the NOx conversion remaining below
10% over the entire temperature range studied. In contrast
with the Pd/TiO2 catalyst, NOx conversions of ∼70% were
achieved in two distinct temperature regimes. One centred at
low temperature (110 ◦C), in the region were H2 conversion
approaches 100%, and one at higher temperature (240 ◦C).
The presence of two reduction maxima indicates operation
of two distinct mechanisms for NOx reduction over this cata-
lyst as noted previously by Ueda et al. [11,12]. Although the
Pd/Al2O3 catalyst was relatively inactive, when modified by
the addition of TiO2 the sample displayed much improved
performance, achieving NOx conversions of ∼50% in both
temperature regimes.
Fig. 7 shows the corresponding DRIFT spectra obtained
under identical conditions (4000 ppm H2 + 500 ppm NO +
5% O2) at 120 ◦C: this temperature corresponds to the low-
temperature NOx reduction pathway shown in Fig. 6. At
this temperature, Pd/Al2O3 showed bands due to adsorbed
Fig. 7. DRIFT spectra obtained with various supported palladium catalysts
◦
in flowing H + NO + O at 120 C. [H ] = 4000 ppm, [NO] = 500 ppm,
2
2
2
[O ] = 5%.
2

N. Macleod et al. / Journal of Catalysis 221 (2004) 20–31
25
3.3. The influence of carbon monoxide: NO reduction by
H2 + CO
Fig. 9 shows the influence of CO on the H2/NO/O2 reac-
tion over Pd/Al2O3, Pd/TiO2 and the Pd/10 wt% TiO2/Al2O3
catalysts, maintaining the total reductant concentration at
4000 ppm (3500 ppm H2 + 500 ppm CO + 500 ppm
NO + 5% O2). It is important to note that CO alone is a
very poor reductant for NOx under these conditions [8]. In
the case of Pd/Al2O3, the presence of a small quantity of
CO increased significantly the maximum NOx conversion
compared to the results obtained with H2 alone (cf. Fig. 6)
from approximately of 6 to ∼25%. This increase in NOx
conversion occurred in the temperature region where reduc-
Fig. 8. DRIFT spectra obtained with various supported palladium catalysts
◦
in flowing H + NO + O at 240 C. [H ] = 4000 ppm, [NO] = 500 ppm,
2
2
2
[O ] = 5%.
2
water (1644 cm−1) and hydrated nitrate species (1403 and
1327 cm−1 [21]). No evidence for the presence of Pd ni-
trosyl species was found with this catalyst. In contrast,
under these conditions the Pd/TiO2 catalyst gave bands
that may be assigned to Pd2+–NO (1842 cm−1), Pd+–NO
(1787 cm−1), Pd0–NO (1722 cm−1), and a bent Pd nitro-
syl species (1693 cm−1) [13,22–30]. The remaining bands
in this spectrum are assigned to variously TiO2 coordinated
nitrates. Over the TiO2/Al2O3 sample a broad peak was ob-
served in the nitrosyl region centred at 1795 cm−1. The
broad and asymmetric nature of this peak indicates adsorp-
tion of NO on a range of different palladium sites (Pd+
and Pd0).
The corresponding spectra obtained at 240 ◦C are shown
in Fig. 8. This temperature corresponds to the high-tempera-
ture NOx reduction pathway apparent in Fig. 6. The Pd/
Al2O3 spectrum shows bands due to anhydrous nitrates due
to desorption of the water present at the lower tempera-
ture. On the Pd/TiO2 sample, in addition to nitrates, evi-
dence for the presence of adsorbed NH3 was also clearly
observed. The asymmetric and symmetric N–H-stretching
modes (3400–3150 cm−1) observed are split due to adsorp-
tion on two different Lewis acid sites on the TiO2 surface
[16,31]. Neither of the NH3-bending modes was clearly re-
solved due to overlap with strong nitrate bands in these re-
gions. On the mixed support evidence for the presence of
NH3 was also observed, although the intensity of these bands
was somewhat reduced compared to Pd/TiO2. In contrast,
no evidence for the presence of NH3 was observed with the
Pd/Al2O3 catalyst under these conditions.
Fig. 9. Influence of CO on the H /NO/O reaction. Conversion profiles for
2
2
(a) hydrogen, (b) CO, and (c) NO as a function of temperature over var-
x
ious palladium-containing catalysts. [H ] = 3500 ppm, [CO] = 500 ppm,
2
−1
[NO] = 500 ppm, [O ] = 5%, w/f = 0.03 g s ml
.
2

26
N. Macleod et al. / Journal of Catalysis 221 (2004) 20–31
found to display the best performance, presumably due to its
containing the optimum balance between TiO2 and Al2O3
sites, as is discussed in more detail below.
The influence of the H2:CO ratio on NOx conversion
at four different temperatures over Pd/10 wt% TiO2/Al2O3
is shown in Fig. 10. In these experiments the catalyst was
maintained at the specified temperature while the ratio of
H2:CO was varied. The total reductant concentration in
all cases was maintained at 4000 ppm. The correspond-
ing nitrogen selectivity, defined as %SN = [N2]/([N2] +
2
[N2O]) × 100, is also shown. Very high NOx conversions
were achieved over a relatively wide range of H2:CO ra-
tios. In general, increasing the CO concentration results in
an increase in the temperatures at which the highest NOx
conversions were observed. This reflects the shift in re-
ductant light-off temperatures that occurs. NOx conversions
>80% were achieved with [H2]/([H2] + [CO]) values in
the range 0.5–0.9 and temperatures of 150–180◦C. The ni-
trogen selectivity increased with increasing temperature and
displayed a slight tendency to decrease with increasing CO
concentration. Under conditions delivering complete NOx
conversion, nitrogen selectivities in the range 65–85% were
observed.
The corresponding DRIFT spectra obtained at 150 ◦C
from Pd/Al2O3, Pd/TiO2, and Pd/TiO2/Al2O3, in the pres-
ence of H2 + CO + NO + O2 (3000 ppm H2 + 1000 ppm
CO + 500 ppm NO + 5% O2) are shown in Fig. 11. Most
of the bands observed below 1700 cm−1 are assigned to
various nitrate/nitrite, carbonate, and bicarbonate species as
discussed in detail elsewhere [9]. Evidence for the presence
of NH3 species was obtained on both Al2O3-containing cat-
alysts (i.e., those which show improved performance in the
Fig. 10. Influence of [H ]/([H ] + [CO]) ratio on total NO reduction and
2
2
◦
corresponding nitrogen selectivity in the temperature range 130–170 C
for Pd/10 wt% TiO /Al O . Total reductant concentration maintained at
2
2 3
−1
4000 ppm. [NO] = 500 ppm, [O ] = 5%, w/f = 0.03 g s ml
.
2
tant conversion approaches 100% (the oxidation of CO and
H2 track each other closely with all three catalysts). In the
case of the Pd/TiO2 catalyst this improvement in catalytic
performance on addition of CO did not occur. In fact the
presence of CO actually decreased the NOx conversion com-
pared to the values obtained in its absence, as shown by
comparison with Fig. 6. The low-temperature reduction fea-
ture was shifted by 20 ◦C toward higher temperature and the
maximum NOx conversion decreased from ∼70% to less
than 40%. However, the high-temperature process was not
directly influenced by the presence of CO (additional exper-
iments showed that the level of NOx conversion obtained
via this mechanism is influenced only by the absolute H2
concentration and not the H2:CO ratio). For the mixed sup-
port Pd/TiO2/Al2O3 sample, as with Pd/Al2O3, a beneficial
effect was again observed on addition of CO. However, in
this case the promotional influence of CO was even greater
than that observed with Pd/Al2O3, with 100% NOx conver-
sion being obtained over the temperature range 135–155◦C
under these conditions. Of the various TiO2 loadings tested
(2.5, 10, and 25 wt% TiO2), the 10 wt% TiO2 sample was
Fig. 11. In situ DRIFT spectra obtained with various supported palladium
◦
catalysts in flowing H + CO + NO + O at 150 C. [H ] = 3000 ppm,
2
2
2
[CO] = 1000 ppm, [NO] = 500 ppm, [O ] = 5%.
2

N. Macleod et al. / Journal of Catalysis 221 (2004) 20–31
27
presence of CO). On Pd/Al2O3 the band at 1609 cm−1 is
assigned to a bending mode of adsorbed NH3. This assign-
ment is in agreement with the appearance of a broad feature
at ∼3250 cm−1 characteristic of N–H-stretching vibrations.
On Pd/TiO2/Al2O3 the 1609 cm−1 mode was obscured
due to the presence of strongly absorbing nitrate bound to
TiO2 sites. However the N–H-stretching modes (3246 and
3193 cm−1) were much more prominent with this catalyst,
indicating a higher NH3 concentration and accounting for
the much-improved activity of Pd/TiO2/Al2O3. No evidence
was obtained for the presence of NHx species on the Pd/TiO2
sample at this temperature. Bands in the 2100–1900 cm−1
region are assigned to various adsorbed CO species as dis-
cussed below. Formate was also observed on the Pd/Al2O3
catalyst (3005, 2910, and 1592 cm−1), although this species
has been shown to be a spectator under the conditions em-
ployed [9].
It should be noted that in the absence of CO no evi-
dence for the presence of NH3 species was obtained on either
Pd/Al2O3 or Pd/TiO2/Al2O3 at this temperature (150 ◦C).
This is illustrated in Fig. 12 for the Pd/10 wt% TiO2/Al2O3
catalyst. The N–H- and O–H-stretching frequency regions
of spectra obtained in the presence of two different gas-
phase compositions are shown. In the presence of 4000 ppm
H2 + 500 ppm NO + 5% O2 (spectrum A) a broad fea-
ture was observed with a maximum ∼3400 cm−1, which is
characteristic of hydrogen bonded O–H groups. On replac-
ing part of the H2 feed by CO (3000 ppm H2 + 1000 ppm
CO + 500 ppm NO + 5% O2, spectrum B), weak new fea-
tures characteristic of N–H-stretching modes developed in
the 3300–3100 cm−1 region. The insert shows the corre-
sponding difference spectrum, which accentuates the N–H-
stretching modes at 3251 and 3180 cm−1. The negative band
at 3708 cm−1 is attributable to loss of “free” O–H groups
due to their hydrogen bonding with adsorbed NH3 [16]. The
shoulder at 2821 cm−1 is consistent with the presence of
NH4+ species also [32].
The most likely route for NH3 formation requiring partic-
ipation of CO is via the generation and subsequent hydrol-
ysis of NCO species. However, NCO was never observed
in DRIFT spectra when H2 was present in the gas feed, most
likely due to its very rapid hydrolysis under these conditions.
We have previously shown that while NCO preformed on a
Pd/Al2O3 catalyst was relatively stable in flowing NO + O2,
it was rapidly consumed in a flow containing H2 + NO + O2
yielding NH3 and CO2 [9]. That NCO is indeed produced on
our Pd/TiO2/Al2O3 catalyst in the absence of H2 is shown
by the results presented in Fig. 13. Here the spectra obtained
from Pd/TiO2, Pd/10 wt% TiO2/Al2O3, and Pd/Al2O3 cata-
lysts after 1 h in a flow containing 4000 ppm CO + 500 ppm
NO + 5% O2 at 150 ◦C are compared. The correspond-
ing time-resolved spectra in the region 2300–1700 cm−1 are
shown in Figs. 14, 15, and 16 for Pd/Al2O3, Pd/TiO2/Al2O3,
and Pd/TiO2, respectively.
On Pd/Al2O3 bands were observed at 2152 and 2080 cm−1
assigned to linear Pd+–CO and Pd0–CO, respectively
[33–35]. Bridging and triply bonded carbonyl species were
also observed, as shown by the bands at 1975 and 1911 cm−1
[33,35]. The bands at 2255 and 2233 cm−1 may be assigned
with confidence to isocyanate (NCO) species adsorbed on
the alumina support [36–38]. The appearance of two bands
could be due to either adsorption of isocyanate on two
different surface sites [37,39], or to the presence of cova-
lently bonded NCO and anionic NCO− [40]. All the bands
Fig. 12. In situ DRIFT spectra showing influence of CO on the intensity
◦
of N–H-stretching modes at 150 C over Pd/10 wt% TiO /Al O . Spec-
2
2 3
trum A: [H ] = 4000 ppm, [NO] = 500 ppm, [O ] = 5%. Spectrum B:
Fig. 13. In situ DRIFT spectra obtained with various supported palla-
2
2
◦
[H ] = 3000 ppm, [CO] = 1000 ppm, [NO] = 500 ppm, [O ] = 5%. Inset
dium catalysts in flowing CO + NO + O at 150 C. [CO] = 4000 ppm,
2
2
2
shows corresponding difference spectrum.
[NO] = 500 ppm, [O ] = 5%.
2

28
N. Macleod et al. / Journal of Catalysis 221 (2004) 20–31
Fig. 14. Time-resolved DRIFT spectra showing accumulation of sur-
face species following introduction of CO + NO + O flow to the
2
◦
DRIFTS cell containing a Pd/Al O sample at 150 C. [CO] = 4000 ppm,
2
3
[NO] = 500 ppm, [O ] = 5%.
2
Fig. 16. Time-resolved DRIFT spectra showing accumulation of sur-
face species following introduction of CO + NO + O flow to the
2
◦
DRIFTS cell containing a Pd/TiO sample at 150 C. [CO] = 4000 ppm,
2
[NO] = 500 ppm, [O ] = 5%.
2
tively constant throughout the remainder of the experiment.
Within the first few minutes an additional band developed
at 2166 cm−1 on the Pd/TiO2/Al2O3 catalyst: this was not
observed with Pd/Al2O3. As discussed below, this is as-
signed to a Pd–NCO species. Although the features in this
region were not well resolved on the Pd/TiO2/Al2O3 cata-
lyst, at least one further species was present, as shown by
the gradual increase in intensity in the 2122 cm−1 region
(Fig. 15). This may be due to the generation of cyanide (CN)
species [44].
The corresponding time-resolved DRIFT spectra ob-
tained in a flow containing 4000 ppm CO + 500 ppm NO +
5% O2 over the Pd/TiO2 sample at 150 ◦C are shown in
Fig. 16. A sharp and intense doublet at 2168/2159cm−1 was
seen to develop rapidly following introduction of the CO +
NO + O2 reactant mix. This peak is characteristic of NCO
adsorbed on the palladium component [41–43]. Interestingly
no support-bound NCO was observed on this catalyst in the
presence of oxygen, indicating that Ti–NCO species were
not stable under these conditions. In separate experiments
employing NO + CO only, Ti–NCO species were observed.
This is in agreement with the work of Kondarides et al.,
who investigated the NO + CO + O2 reaction on Rh/TiO2-
based catalysts and also only observed Ti–NCO species in
the absence of gas-phase oxygen [44]. It has also been shown
that NCO is considerably less stable on TiO2 than other
supports such as SiO2 [39]. The intensity of the Pd–NCO
feature passed through an initial maximum and then contin-
ued to increase gradually throughout the remainder of the
experiment. The other prominent feature in this figure is as-
sociated with Pd nitrosyl species, Pd+–NO at 1974 cm−1
and Pd0–NO at 1735 cm−1. The intensity of these features
increased rapidly to reach a maximum after ∼10 min and
then decreased gradually thereafter. Similar features were
Fig. 15. Time-resolved DRIFT spectra showing accumulation of surface
species following introduction of CO + NO + O flow to the DRIFTS cell
2
◦
containing a Pd/10 wt% TiO /Al O sample at 150 C. [CO] = 4000 ppm,
2
2 3
[NO] = 500 ppm, [O ] = 5%.
2
below 1650 cm−1 can once again be assigned to various ni-
trate/nitrite and carbonate/bicarbonate species [9,13].
On the Pd/TiO2/Al2O3 sample an intense peak was
observed at 1801 cm−1 assigned to a Pd+–NO species
[22,23], with a shoulder at 1730 cm−1 indicating the pres-
ence of Pd0–NO species [22–24]. Alumina-bound NCO
(2255/2233 cm−1) was also observed on this catalyst.
In the 2200–2000 cm−1 range, in addition to the linear
Pd+–CO (2150 cm−1) and Pd0–CO (2084 cm−1) ob-
served with Pd/Al2O3, further bands were observed at 2166
and 2122 cm−1. The development of these bands is more
clearly seen by referring to the time resolved data in Figs. 14
and 15. On both Pd/Al2O3 and Pd/TiO2/Al2O3 catalysts
the Pd+–CO and Pd0–CO (∼2152 and 2080 cm−1, respec-
tively) species appeared first, their intensities remaining rela-

N. Macleod et al. / Journal of Catalysis 221 (2004) 20–31
29
observed on the Pd/TiO2/Al2O3 catalyst, Fig. 15, and seem
to be associated with the Pd/TiO2 component. In contrast
to the Pd/Al2O3 and Pd/10 wt% TiO2/Al2O3 samples, vari-
ous carbonyl species (2150–1900 cm−1) were only observed
for a very brief initial period following introduction of the
CO + NO + O2 reactant mix over Pd/TiO2.
In the case of the Pd/10 wt% TiO2/Al2O3 mixed sup-
port catalyst, the titania caused a significant increase in NOx
conversion during the H2 + NO + O2 reaction (Fig. 6), re-
sulting in performance similar to Pd/TiO2. DRIFTS data
clearly revealed that this improvement in performance, in
both low- and high-temperature regimes, is attributable to
the same mechanisms that occur with Pd/TiO2, i.e., forma-
tion of reduced Pd and NH3 species at the relevant temper-
atures (Figs. 7 and 8). This is good evidence for the close
association of Pd and TiO2 in the mixed support catalyst.
Given the high activity of Pd/TiO2 for NOx reduction by
hydrogen, it is of considerable interest to elucidate the re-
sponse of this catalyst to addition of CO to the H2 +NO+O2
gas feed (in real engine exhausts the H2 and CO concentra-
tions are closely linked due to the water gas-shift reaction).
As shown by comparing the results in Figs. 6 and 9, chang-
ing the reductant feed from 4000 ppm H2 to 3500 ppm H2 +
500 ppm CO resulted in a considerable decrease in the NOx
conversion achieved via the low-temperature mechanism, in-
volving dissociation of NO on reduced metal sites, over this
catalyst. Carbon monoxide induced a similar poisoning in-
fluence on the H2 + NO + O2 reaction over platinum/Al2O3
catalysts [8,48]. This poisoning influence of CO on Pd/TiO2
and Pt/Al2O3 is related to the fact that NH3 was not formed
on these catalysts under reaction conditions, in contrast to
the significantly more active Pd/Al2O3 sample. In the ab-
sence of NH3 generation the up-shift in light-off temperature
that occurs on addition of CO favours direct combustion of
the reductants by O2 rather than reaction with NO.
4. Discussion
The key issue is this—why is the mixed support cat-
alyst (Pd/TiO2/Al2O3) far superior to both Pd/TiO2 and
Pd/Al2O3? The answer does not reside in formation of mixed
or doped oxide phases because the spectroscopic, micro-
scopic, and diffraction data indicate that no such phases
are formed. Therefore before discussing the catalytic re-
sults it is useful to review briefly structural aspects of the
TiO2/Al2O3 mixed support. A similar method to that em-
ployed here was used by Schmal and co-workers to prepare
a series of Pt/TiO2/Al2O3 catalysts [45,46]. These materials
were characterised by a combination of XRD, TPR, DRS-
UV–vis, and TEM and it was concluded that they consisted
of a two-dimensional layer of titanium oxide covering the
γ -alumina support. This interpretation is consistent with our
XRD, XPS, and EDX spectroscopy results for the 10 wt%
TiO2 sample. These suggest that a significant proportion of
the titania was present in a very highly dispersed form on
the alumina support. XPS gave a Ti:Al surface atomic ratio
of 1:3 for this sample. Given the surface area of our Al2O3
starting material (120 m2 g−1), a straightforward calcula-
tion suggests that monolayer coverage would correspond
to approximately 7.0 wt% TiO2 loading. This is in accord
with the HREM results which show that small (2–5 nm)
3D TiO2 crystallites are also present at 10 wt% TiO2 load-
ing. However, note that complete coverage of the alumina
was not achieved with this titania loading, as shown by
TPD/DRIFTS of adsorbed NOx species: these data clearly
reveal that surface sites associated with the Al2O3 compo-
nent were still available for adsorption.
The vastly superior performance of Pd/TiO2 compared to
Pd/Al2O3 for NOx reduction by H2 (Fig. 6) under oxygen-
rich conditions has been reported previously [11–13]. We
showed that the low-temperature reduction channel found
with Pd/TiO2 (occurring in the temperature range close to
H2 light off) was associated with the formation of reduced
palladium (Pd0) sites on the catalyst surface [13]. The high-
temperature (240◦C) NOx reduction channel observed with
Pd/TiO2 was shown to occur via the formation and subse-
quent reaction of ammonia [13]. The strikingly low activity
of Pd/Al2O3 under these conditions was due to palladium re-
maining predominantly in an oxidised state on this support.
Why palladium is more easily reduced on TiO2 as compared
to Al2O3 is not entirely clear. A possible explanation is that
PdO is stabilised on alumina by a strong support interac-
tion, perhaps involving formation of a palladium aluminate
phase [47].
In striking contrast to Pd/TiO2, the Pd/TiO2/Al2O3 mixed
support catalyst exhibited very strong promotion by CO for
NOx reduction—even more pronounced than in the case of
Pd/Al2O3. The DRIFT spectra in Fig. 11 indicate that this
higher activity of Pd/TiO2/Al2O3 compared to Pd/Al2O3 is
due to the higher concentration of NH3 generated on the sur-
face of the former.
Since NH3 is not formed on Pd/TiO2 in the presence of
H2 + CO + NO + O2, how do we account for the fact that
the presence of TiO2 promotes the formation of this species
on the mixed support catalyst? To answer this it is neces-
sary to consider the intermediate NCO species. NCO is not
observed in the presence of H2 + O2 due to its rapid hydrol-
ysis [9]. Therefore to investigate formation of this species it
was necessary to remove H2 from the gas feed. As shown
in Fig. 13, in the presence of CO + NO + O2, Pd–NCO
species are observed in the DRIFT spectra obtained from
both Pd/TiO2 and Pd/TiO2/Al2O3, although the concentra-
tion is much greater on the former. Although Pd–NCO is not
observed in the DRIFT spectra obtained from Pd/Al2O3, its
presence may be inferred from the observed accumulation
of Al–NCO. That is, the NCO species originate on the metal
component followed by spillover onto the alumina surface.
What is very clear is that the TiO2 component of Pd/TiO2/
Al2O3 catalysts promotes formation of NCO on the surface
of metal particles with which it is in intimate contact. It is
likely that this translates into enhanced rates of NH3 forma-

30
N. Macleod et al. / Journal of Catalysis 221 (2004) 20–31
tion when H2 is added to the gas mix. However, the fact that
NH3 was only observed on Pd/Al2O3 and Pd/TiO2/Al2O3
on addition of H2, even though Pd/TiO2 had the highest
concentration of Pd–NCO species in the absence of hydro-
gen, suggests that the Al2O3 component is necessary for
rapid hydrolysis of NCO to ammonia under these condi-
tions. Although both Al2O3 and TiO2 have been shown
previously to be active for NCO hydrolysis [49,50], no di-
rect comparison of the activity of these two oxides for this
reaction has been performed. Furthermore, in separate ex-
periments we have observed that NCO species preadsorbed
on the support component of Pd/TiO2 catalysts during the
CO + NO reaction were subsequently rapidly consumed
when oxygen was added to the gas mix. This accounts for
the fact that Ti–NCO species were never observed under the
oxygen-rich conditions discussed in this work. In contrast,
on Al2O3 preadsorbed NCO was relatively stable in the pres-
ence of oxygen but was consumed rapidly in the presence of
H2 + O2 [9]. Therefore in the presence of the full reaction
mix (H2 + CO + NO + O2), NCO hydrolysis to ammonia
dominates on the alumina surface while on TiO2 rapid NCO
combustion may compete with hydrolysis.
How are we to rationalise the observation that TiO2 pro-
motes the formation of Pd–NCO species? The most likely
explanation is in terms of the relative surface coverages
of CO and NO on palladium particles supported on the
TiO2 and Al2O3 components, under conditions of com-
petitive adsorption. (Other things being equal, one expects
CO to adsorb more strongly than NO.) One of the ma-
jor differences between the spectra observed with Pd/TiO2,
Pd/TiO2/Al2O3, and Pd/Al2O3 catalysts in the presence of
CO + NO + O2 is the very different concentrations of
Pd carbonyl and Pd nitrosyl species (Figs. 13–16). With
Pd/TiO2 and Pd/TiO2/Al2O3 catalysts the surface of the
Pd particles is dominated by Pd–NO species (Pd+–NO and
Pd0–NO at 1801/1974 and 1935 cm−1, respectively), while
on Pd/Al2O3 only carbonyl species (Pd+–CO, Pd0–CO, and
bridged CO at 2152, 2080, and 1975 cm−1, respectively) are
observed. This difference in relative coverage must influence
the rate at which NCO is formed, with higher NO concentra-
tions favouring isocyanate production.
experiments in which H2 was replaced by H2O, where it
was found that the promoting effect of directly added wa-
ter was substantially less than that of hydrogen. Factors
contributing to this clear difference between the effects of
hydrogen and water could be as follows. NCO formation
on the metal surface occurs via reaction between adsorbed
CO and N. Formation of the latter, by NO dissociation, is
likely to be the rate-determining step [52]. It is significant,
therefore, that enhanced rates of NO dissociation have been
observed in the presence of hydrogen [53–55]. Note also
that the intense Pd–NO bands at 1800 and 1735 cm−1 we
observed in the presence of CO + NO + O2 on Pd/TiO2
and Pd/TiO2/Al2O3 (Fig. 13) were not seen in the pres-
ence of H2 + CO + NO + O2 (Fig. 11), consistent with
extensive NO dissociation in the later case. Moreover, the
bands due to the linear and bridged Pd–CO species ob-
served on Pd/TiO2/Al2O3 in the presence of CO + NO + O2
(Fig. 13, 2084 and 1974 cm−1, respectively) were signifi-
cantly red-shifted in the presence of hydrogen (to 2064 and
1956 cm−1, respectively; Fig. 11). This is attributed to coad-
sorbed hydrogen weakening the C–O bond by enhancing
back-donation of electron density into the CO antibonding
π orbital [56,57]. The same would be expected with NO,
weakening the N–O bond and thus enhancing the rates of
NO dissociation in the presence of H2 [53–55]. This in turn
would increase the rate of NCO production and subsequent
NH3 formation.
An alternative explanation for the role of hydrogen in this
system relates to the mobility of isocyanate species. Lorimer
and Bell [58] found while investigating the CO + NO reac-
tion over a Pt/SiO2 disk that NCO species were deposited
on a second SiO2 disk located centimeters away in the same
in situ infrared cell. This was explainable only via the gen-
eration of gas-phase HNCO, with the source of hydrogen
considered to be reverse spillover of hydroxyl groups from
the support to the metal surface. In the present case it is pos-
sible therefore that the role of hydrogen is to enhance the
rate of migration of isocyanate species from the metal to the
support via the generation of gas-phase HNCO.
Although the Pd/TiO2/Al2O3 catalyst discussed here de-
livers extremely high NOx conversions under demanding
conditions, which includes the presence of additional wa-
ter vapour [14], there remain aspects of this system that
could be significantly improved. It is clear from Fig. 10
that the highest NOx conversions are achieved with rela-
tively H2-rich mixtures. Although >80% NOx conversion
can be achieved at the 1:3 H2:CO ratio encountered in most
engine exhausts, improving the performance of the system
under more CO-rich conditions would clearly be desirable.
Widening the temperature range over which the catalyst
maintains very high NOx conversions would also be bene-
ficial, although the fact that NH3 is also formed via a sec-
ond, high-temperature mechanism ensures that the signif-
icant NOx conversion (>50%) can be maintained over a
very wide temperature range (150 ◦C in the data shown in
Fig. 9). Identifying suitable promoters could be one method
The mechanism whereby TiO2 affects Pd particles so as
to strongly favour NO adsorption may be purely electronic
in origin. However, it also possible that TiO2 is directly in-
volved in the catalytic reaction. Reducible metal oxides such
as CeO2 are known to participate in CO oxidation by re-
acting with CO molecules adsorbed near the metal-support
interface, generating CO2 and oxygen vacancies [51]. In the
present case this process could result in a reduction of the
steady-state CO coverage on the Pd surface, thus enhancing
the NO coverage.
Although our results strongly suggest that NCO species
are intermediates in the enhanced conversion of NO in the
presence of H2 + CO over Pd/TiO2/Al2O3, the role of H2
cannot solely be that of generating H2O for subsequent NCO
hydrolysis. This was demonstrated by the result of control

N. Macleod et al. / Journal of Catalysis 221 (2004) 20–31
31
of achieving these aims. Overall, we believe this system rep-
resents a promising alternative strategy for the control of
NOx emissions generated under oxygen-rich conditions.
[16] T.J. Dines, C.H. Rochester, A.M. Ward, J. Chem. Soc., Faraday
Trans. 87 (1991) 643.
[17] K. Hadjiivanov, H. Knözinger, Phys. Chem. Chem. Phys. 2 (2000)
2803.
[18] T. Chafik, A.M. Efstathiou, X.E. Verykios, J. Phys. Chem. B 101
(1997) 7968.
[19] W.S. Kijlstra, D.S. Brand, E.K. Poels, A. Bliek, J. Catal. 171 (1997)
208.
5. Conclusions
[20] F.C. Meunier, J.P. Breen, V. Zuzaniuk, M. Olsson, J.H.R. Ross, J. Ca-
tal. 187 (1999) 493.
[21] A.L. Goodman, E.T. Bernard, V.H. Gassian, J. Phys. Chem. A 105
(2001) 6443.
[22] T. Maillet, J. Barbier, P. Gelin, H. Praliaud, D. Duprez, J. Catal. 202
(2001) 367.
[23] K. Almusaiteer, S.S.C. Chuang, J. Catal. 184 (1999) 189.
[24] N. Sheppard, C. De La Cruz, Catal. Today 70 (2001) 3.
[25] A.J. Renouprez, J.F. Trillat, G. Bergeret, P. Delichere, J.L. Rousset, J.
Massardier, D. Loffreda, D. Simon, F. Delbecq, P. Sautet, J. Catal. 198
(2001) 243.
[26] A. El Hamdaoui, G. Bergeret, J. Massardier, M. Primet, A. Renouprez,
J. Catal. 148 (1994) 47.
[27] F.B. Noronha, M.A.S. Baldanza, M. Schmal, J. Catal. 188 (1999) 270.
[28] A.M. Venezia, L.F. Liotta, G. Deganello, P. Terreros, M.A. Pena,
J.L.G. Fierro, Langmuir 15 (1999) 1176.
[29] K. Almusaiteer, S.S.C. Chuang, J. Catal. 180 (1998) 161.
[30] T.E. Hoost, K. Otto, Laframboise, J. Catal. 155 (1995) 303.
[31] G. Ramis, L. Yi, G. Busca, M. Turco, E. Kotur, R.J. Willey, J. Ca-
tal. 157 (1995) 523.
[32] N.Y. Topsøe, H. Topsøe, J.A. Dumesic, J. Catal. 151 (1995) 226.
[33] M. Valden, R.L. Keiski, N. Xiang, J. Pere, J. Aaltonen, M. Pessa, T.
Maunula, A. Savimäki, A. Lahti, M. Härkönen, J. Catal. 161 (1996)
614.
1. Pd/TiO2/Al2O3 catalysts provide very high (100%) NOx
conversions under demanding conditions (low temper-
ature, high oxygen excess) in the presence of mixed
H2 + CO reductant feeds. A second high-temperature
NOx reduction channel also operates which greatly ex-
tends the useful temperature range.
2. Both NOx reduction channels generate NH3 in situ. The
low-temperature channel operates via formation and
subsequent hydrolysis of NCO species while the high-
temperature channel forms NH3 directly from reaction
between H2 and NO.
3. The observed synergy between titania and alumina de-
rives from the intrinsic surface chemistry of the pure
oxides and not from mixed oxide formation. Thus the
Pd/TiO2 component of the Pd/TiO2/Al2O3 catalyst is
active toward NCO production while the Al2O3 com-
ponent promotes subsequent hydrolysis of NCO to am-
monia.
[34] V. Pitchon, M. Primet, H. Praliaud, Appl. Catal. 62 (1990) 317.
[35] K.I. Choi, M.A. Vannice, J. Catal. 127 (1991) 465.
[36] T. Bánsági, F. Solymosi, J. Catal. 202 (2001) 205.
[37] R. Dictor, J. Catal. 109 (1988) 89.
[38] A. Acke, B. Westerberg, M. Skoglundh, J. Catal. 179 (1998) 528.
[39] F. Solymosi, L. Völgyesi, J. Raskó, Z. Phys. Chem. 120 (1980) 79.
[40] M.L. Unland, J. Catal. 31 (1976) 459.
[41] J. Raskó, F. Solymosi, J. Catal. 71 (1981) 219.
[42] R.J. Gorte, L.D. Schmidt, B.A. Sexton, J. Catal. 67 (1981) 387.
[43] J. Kiss, F. Solymosi, J. Catal. 179 (1998) 277.
[44] D.I. Kondarides, T. Chafik, X.E. Verykios, J. Catal. 191 (2000) 147.
[45] N.S. De Resende, M. Schmal, J.-G. Eon, in: G. Poncelet, et al. (Eds.),
Preparation of Catalysts, Vol. 6, Elsevier Science, Amsterdam, 1995,
p. 1059.
[46] N.S. De Resende, J.-G. Eon, M. Schmal, J. Catal. 183 (1999) 6.
[47] T.E. Hoost, K. Otto, Appl. Catal. A 92 (1992) 39.
[48] K. Yokota, M. Fukui, T. Tanaka, Appl. Surf. Sci. 121/122 (1997) 273.
[49] R. Dümpelmann, N.W. Cant, D.L. Trimm, J. Catal. 162 (1996) 96.
[50] M. Kleeman, M. Elsener, M. Koebel, A. Wokaun, Ind. Eng. Chem.
Res. 39 (2000) 4120.
[51] M. Fernández-García, A. Martínez-Arias, L.N. Salamanca, J.M. Coro-
nado, J.A. Anderson, J.C. Conesa, J. Soria, J. Catal. 187 (1999) 474.
[52] J.H. Miners, A.M. Bradshaw, P. Gardner, Phys. Chem. Chem. Phys. 1
(1999) 4909.
[53] W.C. Hecker, A.T. Bell, J. Catal. 92 (1985) 247.
[54] E. Shustorovich, A.T. Bell, Surf. Sci. 289 (1993) 127.
[55] R. Burch, T.C. Watling, Catal. Lett. 37 (1996) 51.
[56] A. Erdöhelyi, M. Pásztor, F. Solymosi, J. Catal. 98 (1986) 186.
[57] F. Solymosi, A. Erdöhelyi, M. Lancz, J. Catal. 95 (1985) 567.
[58] D. Lorimer, A.T. Bell, J. Catal. 59 (1979) 223.
Acknowledgments
R.C. acknowledges support from the Cambridge Univer-
sity Oppenheimer Fund and from Johnson Matthey plc. We
are grateful to Dr. D.A. Jefferson for access to the HREM
facilities.
References
[1] V.I. Pârvulescu, P. Grange, B. Delmon, Catal. Today 46 (1998) 233.
[2] A. Fritz, V. Pitchon, Appl. Catal. B 13 (1997) 1.
[3] G.T. Went, L.J. Leu, R.R. Rosin, A.T. Bell, J. Catal. 134 (1992) 492.
[4] G. Busca, L. Lietti, G. Ramis, F. Berti, Appl. Catal. B 18 (1998) 1.
[5] M. Koebel, M. Elsener, M. Kleemann, Catal. Today 59 (2000) 335.
[6] L.F. Xu, R.W. McCabe, R.H. Hammerle, Appl. Catal. B 39 (2002) 51.
[7] E. Seker, N. Yasyerli, E. Gulari, C. Lamert, R.H. Hammerle, Appl.
Catal. B 37 (2002) 27.
[8] N. Macleod, R.M. Lambert, Appl. Catal. B 35 (2002) 269.
[9] N. Macleod, R.M. Lambert, Appl. Catal. B, in press.
[10] R. Burch, M.D. Coleman, Appl. Catal. B 23 (1999) 115.
[11] A. Ueda, A. Takayuki, A. Masashi, T. Kobayashi, Catal. Today 45
(1998) 135.
[12] A. Ueda, T. Nakao, M. Azuma, T. Kobayashi, Chem. Lett. (1998) 595.
[13] N. Macleod, R. Cropley, R.M. Lambert, Catal. Lett. 86 (2003) 69.
[14] N. Macleod, R.M. Lambert, Catal. Commun. 3 (2002) 61.
[15] N. Macleod, J. Isaac, R.M. Lambert, J. Catal. 193 (2000) 115.