Ammonia formation over Pd/Al2O3 modified with cerium and barium

Article abstract of DOI:10.1016/j.cattod.2016.01.012

We report experimental results for ammonia formation from nitric oxide and either a direct source of hydrogen or from a mixture of carbon monoxide and water over palladium based catalysts. Specifically, the addition of barium or cerium into an alumina supported palladium sample was studied. Static and transient flow reactor experiments were performed in order to identify the effects of temperature and the presence of oxygen on the activity for ammonia formation. Modification of Pd/Al₂O₃ with cerium proved to be beneficial for the activity due mainly to its enhancement of the water-gas-shift reaction, thus providing a higher availability of hydrogen for ammonia formation, but also because it remains active in the presence of slightly oxidizing global conditions when hydrogen is provided directly to the feed. Although the modification of Pd/Al₂O₃ with barium did not affect the ammonia formation during static conditions, the activity during lean/rich cycling increased. This is important for applications of passive selective catalytic reduction.

Full text of DOI:10.1016/j.cattod.2016.01.012

Catalysis Today 267 (2016) 210–216
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Ammonia formation over Pd/Al O modified with cerium and barium
2
3
a,∗
a
b
c
Emma Catherine Adams , Magnus Skoglundh , Pär Gabrielsson , Mats Laurell ,
Per-Anders Carlssona
a
Competence Centre for Catalysis, Chalmers University of Technology, SE-412 96 Göteborg, Sweden
Haldor Topsøe A/S, P.O. Box 213, DK-2800 Lyngby, Denmark
Volvo Car Corporation, 97624, Torslanda HDIN, SE-405 31 Gothenburg, Sweden
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
We report experimental results for ammonia formation from nitric oxide and either a direct source of
hydrogen or from a mixture of carbon monoxide and water over palladium based catalysts. Specifically,
the addition of barium or cerium into an alumina supported palladium sample was studied. Static and
transient flow reactor experiments were performed in order to identify the effects of temperature and
the presence of oxygen on the activity for ammonia formation. Modification of Pd/Al2O3 with cerium
proved to be beneficial for the activity due mainly to its enhancement of the water-gas-shift reaction,
thus providing a higher availability of hydrogen for ammonia formation, but also because it remains active
in the presence of slightly oxidizing global conditions when hydrogen is provided directly to the feed.
Although the modification of Pd/Al2O3 with barium did not affect the ammonia formation during static
conditions, the activity during lean/rich cycling increased. This is important for applications of passive
selective catalytic reduction.
Received 22 September 2015
Received in revised form 8 December 2015
Accepted 13 January 2016
Available online 10 February 2016
Keywords:
Catalytic exhaust aftertreatment
Passive-SCR
NOx reduction
NH3 formation
©
2016 Elsevier B.V. All rights reserved.
1
. Introduction
stored in the form of urea-in-water solution on-board the vehicle.
Although this technique is acceptable for heavy-duty vehicles, dif-
ficulties arise when applied to smaller passenger cars. Some of the
problems encountered are due to extra weight associated with the
need for an additional urea storage tank as well as the urea injection
system, which is complex, costly and increases the risk of creating
Combustion of gasoline and diesel in vehicles results in the for-
mation of harmful products, including nitrogen oxides (NOx), which
are known to be responsible for various environmental issues such
as photochemical smog and acid rain [1,2]. At present, the fuel
economy of gasoline fuelled vehicles can be improved by ensuring
that the combustion takes place in the presence of excess oxy-
gen, so-called lean operation [3]. Practically, lean operation makes
an NH slip [2,5,10]. Ammonia emissions are undesirable since NH3
3
is a toxic air pollutant and is known to contribute to the production
of secondary particulate matter [11,12].
it challenging to reduce NOx to N , driving the development for
Passive SCR is a newly emerging technique for NOx abatement in
both diesel and lean-burn spark ignition gasoline passenger vehi-
cles. The concept of this technique is to generate an onboard supply
2
new NOx abatement concepts such as NOx storage reduction (NSR)
and selective catalytic reduction (SCR) whereby the fuel economy
can be improved whilst the tailpipe emissions are kept sufficiently
low to fulfill future, and increasingly stringent, regulations [4–7].
of NH in the vehicle by utilizing the NOx that is readily available in
3
the exhaust stream. Although the current concept of this technique
is to produce NH3 whilst the engine undergoes rich operation (low
air-to-fuel ratio), a long-term goal is to suggest methods and/or cat-
alyst formulations which would allow the formation of NH3 under
stoichiometric or even slightly lean conditions, thereby minimiz-
ing the fuel penalty incurred during the periods required for NH3
formation. The formed NH3 can then be stored in an SCR catalyst
placed downstream of the ammonia formation catalyst and used to
reduce slipped NOx when the engine is set back to operate under
lean conditions. If sufficient amounts of NH3 can be produced dur-
ing the rich periods and stored for complete reaction with NOx
to form N2 during the subsequent lean periods, an external urea
Selective catalytic reduction of NOx with ammonia (NH -SCR) is
3
currently the preferred technology for NOx abatement from sta-
tionary sources and larger vehicles including trucks and buses [8].
This technique relies on selective reduction of NOx with NH3 over
a suitable catalyst to form N2 in the presence of elevated levels
of O2 [2,9]. However, due to concerns with the safety and toxicity
associated with ammonia transportation and storage, the NH3 is
∗ Corresponding author.
E-mail address: emma.adams@chalmers.se (E.C. Adams).
http://dx.doi.org/10.1016/j.cattod.2016.01.012
0
920-5861/© 2016 Elsevier B.V. All rights reserved.

E.C. Adams et al. / Catalysis Today 267 (2016) 210–216
211
injection system, as required by current NH -SCR technology, may
not be needed [13–15].
further impregnation in order to keep the noble-metal particle
distribution and size constant in all samples being investigated.
Barium and cerium containing samples were prepared to a nomi-
nal loading of 10 wt.% (calculation based on pure Ba and Ce) using
aqueous solutions of Ba(NO3)2 and Ce(NO3)2 respectively. In order
3
Another technique, which also depends on periodic cycling
between lean and rich operation, is the afore-mentioned NSR
system. A well-studied material used for this application is
Pt/Ba/Al O , over which NH formation has previously been
to stabilize Ba in the form of BaCO , a successive impregnation
2
3
3
3
observed during operational switches. Castoldi et al. [16] report
that, as the Ba content of the formulation is increased, the amount
of NH3 formed over the catalyst by the reaction of stored NOx with
H2 also increases but is only observed after the formation of N2
has diminished. As reported by Pihl et al. [17] and Cumaranatunge
et al. [18] this may be related to the fact that NH3 can be generated
in significant quantities at the catalyst inlet but as it progresses
axially through the catalyst it acts as a hydrogen carrier, reducing
step was carried out on this sample using an aqueous solution of
ammonium carbamate [24]. All samples were then freeze-dried and
calcined in air at 550 C for 1 h after each impregnation step.
◦
The powder catalysts were then coated onto cordierite mono-
lith substrates. This was done by preparing solutions consisting
of the powder catalyst, a suitable binder material (Disperal P2),
water and ethanol. The monolith was then carefully immersed into
the slurry, allowing the catalyst-containing liquid to be dispersed
evenly throughout the channels by capillary forces. The monolith
stored NOx to N . This was confirmed by Clayton et al. [19] who
2
◦
also determined that the reactivity of NH3 and H2 with stored NOx
was then dried in air at 90 C for 2 min to remove the water and
◦
is comparable at temperatures above 230 C.
ethanol, before being briefly calcined in air for a further 2 min at
◦
In automotive applications however, deviations in driving pat-
terns result in conditions where the concentration of H2 present
in the exhaust stream does not meet the requirement for ammonia
formation for significant periods. Despite this, passive SCR may play
an important role in NOx abatement provided that a direct supply
500 C and weighed. If necessary, the process was repeated until
the desired mass of catalyst and binder was deposited (200 ± 3 mg).
◦
Once coated, the monoliths were calcined in air at 600 C for 2 h to
ensure complete removal of water and phase transformation of the
Disperal P2 to alumina.
of H is not necessary. Carbon monoxide and water are also present
The specific surface area of the prepared powder samples was
2
◦
in the exhausts during rich operation [20] and H can be formed by
measured by nitrogen physisorption at −196 C using an ASAP 2010
2
the water-gas-shift (WGS) reaction;
(Micrometrics) sorptometer. Prior to adsorption, all samples were
◦
dried at 200 C for 2 h under vacuum in order to remove any residual
CO + H O ↔ CO + H
(1)
2
2
2
water. Respective surface areas were then determined according
to the standard Brunauer-Emmet-Teller (BET) method [25] using
Hence, water could be a viable hydrogen source for the further for-
P/P = 0.05–0.20. The specific surface area of the pure support mate-
0
mation of NH . Ceria, particularly when metal-modified, has been
3
◦
rials; both fresh and after calcination in air at 600 C for 1 h, was
extensively studied due to its excellent properties that catalyze the
WGS reaction. When metal-modified, the oxygen storage capac-
ity (OSC) and reducibility of the ceria support increase [21,22].
The reducibility of the ceria support is of importance since it is
able to utilise lattice oxygen to facilitate the exchange process
measured to determine the thermal stability of the supports. The
prepared powder catalysts were characterized directly after calci-
nation.
The distribution of Pd, Ba and Ce particles throughout the
alumina carrier phase was determined by scanning transmission
electron microscopy coupled with energy-dispersive X-ray spec-
troscopy (STEM-EDX). This allowed the confirmation of a successful
synthesis procedure. A small amount of the prepared sample was
crushed together with ethanol to form a suspension which was sub-
sequently ultrasonicated for 1 min. One drop of the suspension was
then applied to a Cu-grid coated with continuous carbon. STEM-
EDX maps were then acquired at a spot size of 5, corresponding to
a beam current of 0.7 nA, using an FEI-Talos electron microscope.
Aperture sizes of 2000 and 70 ␮m were selected with a camera
length of 125 cm. The minimum time for elemental mapping of the
sample was 10 min.
2
CeO ↔ Ce O + 1/2O , thereby resulting in an oxygen vacancy.
2
2
3
2
Thus for the water-gas-shift reaction, CO is able to adsorb onto
metal sites whilst H O dissociation takes place on the oxide
2
vacancy, resulting in high activity [22].
As we reported previously, Pd/Al O3 is a promising material for
2
passive SCR, over which high selectivity to NH3 can be achieved
over a broad temperature interval [23]. The objective of the present
study is to investigate whether the modification of Pd/Al O3 with
2
cerium or barium can further enhance the temperature interval for
NH3 formation from NO + H2 and NO + H O + CO feed gas compo-
2
sitions. Since passive SCR is a technique that depends on periodic
switching between rich and lean operating conditions, the influ-
ence of the presence of O2 on the ammonia formation is also of
interest and is studied by use of both steady-state (oxygen-step)
and transient (oxygen-pulse) experiments.
2
.2. Kinetic measurements in gas flow reactor
Continuous gas flow reactor studies were carried out to deter-
mine the activity and selectivity of the prepared catalysts. The
reactor setup employed in this investigation has previously been
described by Kannisto et al. [26]. The system consists of a quartz
tube surrounded with a heating coil and insulation. It contains two
thermocouples to measure the temperature 10 mm before and in
the centre of the monolith sample. Two uncoated (blank) cordierite
monoliths were placed on either side of the sample monolith in the
quartz tube in order to reduce otherwise significant heat losses and
achieve a nearly isothermal sample. The inlet gas composition was
controlled using mass flow controllers (Bronkhorst Hi-Tech LOW-
ꢁP-FLOW) and the outlet gas composition was analyzed using an
FTIR gas-analyzer (MKS 2030 HS). All experiments were carried out
with an Ar balance in order to keep the total gas flow constant at
2000 ml/min, corresponding to a gas hourly space velocity (GHSV)
2
. Experimental
2.1. Catalyst preparation and characterization
A 30 g batch of 0.5 wt.% Pd/Al O3 was prepared via incipi-
2
ent wetness impregnation of an aqueous Pd precursor solution
(NH ) Pd(NO ) ) into a commercially-available ꢀ-Al O support
(
3
4
3
2
2
3
material (Puralox SBa200). The prepared sample was subsequently
frozen in liquid N2 and freeze-dried for approximately 12 h to
ensure complete sublimation of water with preserved dispersion
of Pd and minimal damage to the support structure. The obtained
◦
powder catalyst was then calcined in air at 550 C for 1 h.
The prepared Pd/Al O3 catalyst was then divided into three
2
−
1
aliquots, two of which underwent subsequent modification with
either cerium or barium. The same Pd/Al O3 sample was used for
of 40,000 h . Prior to kinetic analysis, all samples were pretreated
◦
in a flow of 5 vol.% O2 for 30 min at 500 C in order to remove any
2

2
12
E.C. Adams et al. / Catalysis Today 267 (2016) 210–216
Table 1
carbonaceous contaminants from the monolith surface and allow
fair comparison.
To investigate the temperature region for which the catalyst is
active for ammonia formation, temperature programmed reaction
experiments were conducted. All samples were exposed to feed
Surface area characterization of prepared samples.
2
Surface area (m /g)
Nominal composition (%)
Pd
Ba
Ce
Pd/Al2O3
Ba/Pd/Al2O3
Ce/Pd/Al O
185
160
152
0.5
0.5
0.5
–
10
–
–
–
10
compositions containing 500 ppm NO and either 1500 ppm H (feed
2
gas A) or 1500 ppm CO accompanied by 2% H O (feed gas B). During
2
2
3
exposure to both gas mixtures, the inlet temperature was varied
◦
◦
from 500 to 120 C at a rate of 7 C/min. The system was then kept
at this temperature for 10 min before increasing the temperature to
determined as 0.58 ± 0.26 wt.%. The variation in metal content cal-
culated during the imaging conducted at different positions is
explained by variations in the distribution of Pd on the microscopic
scale. Barium was also found to be homogeneously distributed into
the parent Pd/Al2O3 sample, which can be seen in Fig. 2. In the top
panels, the presence of small, well-distributed regimes with bar-
ium, most likely in the form of baria, that are evenly distributed
throughout the Al2O3 support is observed. The lower panels how-
ever, show that there are also larger needle-shaped regions of
barium and oxygen likely in the form of Ba(OH)2·H2O, BaO crys-
tallites present in the sample that are entirely separate from the
alumina carrier phase. Fig. 3 shows an obtained STEM-EDX image
of the prepared Ce/Pd/Al O sample. It can be observed that cerium
◦
5
00 C at the same rate. The low ramp rate was chosen as to ensure
nearly stationary conditions.
To evaluate the influence of the stoichiometric number, S, of
the feed on the formation of NH over the catalysts, steady-state
3
◦
experiments at 250, 350 and 450 C were subsequently carried out.
The S-value characterizes the net oxidizing-reducing character of
the inlet feed and is defined as [27];
S = 2
[O ] + [NO]
(2)
2
[
CO] + [H ]
2
Again, the two feed compositions were fed to the reactor though
this time various levels of oxygen were also added. The O2 concen-
tration was varied from 0.105 to 0.015% (corresponding to S-values
between 1.73 and 0.33) in steps of 150 ppm, each of which lasted
for 20 min in order to allow steady state to be reached.
2
3
is very well distributed throughout the Al O support material with
2
3
a particle size of approximately 5–10 nm in diameter.
To determine the transient effect of switching between oxygen
excess and deficient conditions, feed gas A or B was fed constantly
to the reactor. A supply of 2 vol.% O2 was then pulsed into the
flow at periodic intervals of 4 min whereafter the O2 pulse was
extinguished to emulate rich conditions. A total of 10 lean/rich
pulses were introduced in succession in order for a consistent,
repeatable behaviour of the catalytic material to be obtained. It
was typically seen that the initial 3–4 pulses in the sequence gave
responses which were higher in concentration with time. After this,
the concentration of NH3 formed during each pulse was constant
and reproducible.
3.2. Kinetic evaluation
Fig. 4 shows the formation of NH3 in the presence of NO and
either a direct source of H2 (panel a) or a combination of CO
and H O (panel b) as a function of the inlet gas temperature. The
2
inlet concentration of NO was 500 ppm and the chosen limiting
reagent for this reaction, thus 500 ppm is the maximum concentra-
tion of NH3 that can be generated under steady-state conditions.
It can therefore be seen that complete conversion of NO to NH3
is achieved over Pd/Al O over the broad temperature range of
2
3
◦
250–500 C when H2 is available for direct reaction in the feed,
with the Ce-modified sample showing a similar (although slightly
3
. Results
lower) conversion trend. Negligible amounts of N O were detected
2
throughout the course of this reaction. Upon comparison of the
WGS-assisted reaction, the temperature window for ammonia for-
mation becomes more narrow where the Ce-modified sample is
the most active sample with a light-off temperature for ammo-
3
.1. Catalyst characterization
The determined specific surface areas of the synthesized cata-
◦
lysts are displayed in Table 1. Here it can be seen that there is a
small decrease in the specific surface area after further impregna-
tion of the Pd/Al O reference sample, with the Ce-modified sample
nia formation of approximately 220, compared to 250 C as seen
for the non-modified and Ba-containing sample. The Ba-modified
sample shows a maximum of around 50% conversion of NO to NH3
when exposed to both feed compositions although the tempera-
2
3
exhibiting the lowest value.
◦
The results from the STEM-EDX analysis of the Pd/Al O ,
ture of this maximum shifts from 250 to 350 C upon switching
2
3
Ce/Pd/Al O and Ba/Pd/Al O3 samples are displayed in Figs. 1–3
to H O as the source of hydrogen for NH3 formation. Unlike the
2
3
2
2
respectively. As can be seen in Fig. 1, the synthesis of Pd/Al O3
reaction when H2 was directly available, N O is now detected in
2
2
resulted in a highly dispersed noble metal-phase throughout the
support material. The Pd content throughout the sample was
significant quantities over all samples. A maximum of ∼50 ppm
is formed over the Pd/Al O sample, centred at a temperature of
2
3
Fig. 1. STEM-EDX maps for Pd, O and Al for the 0.5 wt.% Pd/Al2O3 reference sample.

E.C. Adams et al. / Catalysis Today 267 (2016) 210–216
213
Fig. 2. STEM-EDX maps for Ba, O and Al for 10 wt.% Ba/Pd/Al2O3.
◦
approximately 270 C. The addition of Ce and Ba to this formula-
activity of this reaction is significantly enhanced, from almost no
tion significantly enhances the amount of N O formed but shifts
production of NH3 over the reference Pd/Al O3 sample to concen-
2
2
the formation towards lower temperatures, i.e. ∼100 ppm N O is
trations in slight excess of 200 ppm for the sample modified with Ce.
Interestingly, the activity for NH3 formation over the Ba-modified
sample is higher during the experiments conducted in the presence
of low concentrations of oxygen compared to the prior tempera-
2
◦
formed over the Ba-modified sample at 220 C and ∼140 ppm is
◦
formed over the Ce-modified sample at 200 C.
Fig. 5 shows the steady state formation of NH3 from feed gas
A (left-hand column) and B (right-hand column) over Pd/Al O3 (a
ture programmed experiment performed in the absence of O . This
2
2
and d), Ba/Pd/Al O (b and e) and Ce/Pd/Al O3 (c and f) as a func-
observation is particularly noteable in the case where H is directly
2
3
2
2
tion of the stoichiometric number of the feed gas at 450, 350 and
available in the feed gas.
◦
2
50 C. The dashed red line indicates the point where S = 1, indi-
Fig. 6 shows the transient behaviour of each sample when
switching from rich to lean conditions for different temperatures
in the presence of either feed gas A (left-hand column) or B (right-
hand column). Formation of ammonia is observed for all samples
immediately after O2 is removed from the feed. In the presence of
large excess of oxygen, however, no formation of NH3 is observed
over any sample. The barium-modified sample shows enhanced
cating a stoichiometrically balanced feed. To the right of this line
(
(
S > 1) the feed becomes net-oxidizing, whereas the left-hand side
S < 1) represents net-reducing feeds. Higher amounts of NH3 are
formed over all catalysts when H2 is directly available for reac-
tion, i.e. no prerequisite reaction is required, as was seen by the
temperature-programmed experiments previously discussed. For
the majority of the experiments, the formation of NH decreases as
activity for NH formation under transient conditions as compared
3
3
the stoichiometric number of the gas feed increases and is entirely
suppressed when the feed becomes net-oxidizing. However, upon
observation of the NH formation over Ce/Pd/Al O in the presence
to stationary conditions. This can be said of all temperature and
gas feed compositions with the exception of the WGS-assisted
reaction conducted at 250 C (Fig. 6f). Again, the benefit of impreg-
◦
3
2
3
◦
of NO and H , it can be seen that at low temperature (250 C) it is
nation of cerium into the catalyst formulation during WGS-assisted
conditions can be seen and is especially pronounced during the
2
possible to produce low amounts of NH even when S > 1. A further
3
◦
trend regarding the Ce-modified sample is the increased activ-
ity for NH3 formation in the presence of WGS reaction conditions
compared to the other samples. In particular the low-temperature
rich phase of the transient experiment conducted at 250 C where
the ammonia formation is approximately 200 ppm higher over the
Ce/Pd/Al O3 sample compared to Pd/Al O3 and Ba/Pd/Al O . With
2
2
2
3
Fig. 3. STEM-EDX maps for Ce, O and Al for 10 wt.% Ce/Pd/Al2O3.

2
14
E.C. Adams et al. / Catalysis Today 267 (2016) 210–216
Fig. 4. Ammonia formation over reference, Ce-modified and Ba-modified 0.5 wt.%
Pd/Al2O3 during temperature programmed reaction in the presence of (a) 500 ppm
NO and 1500 ppm H2 (feed gas A) and (b) 500 ppm NO, 1500 ppm CO and 2 vol.% H2O
−
1
(
feed gas B). Balance Ar is employed to maintain the space velocity at 40,000 h and
◦
a temperature ramp rate of 7 C/min was used.
Fig. 6. Transient formation of NH3 over reference, Ce-modified and Ba-modified
.5 wt.% Pd/Al2O3 formed during lean/rich cycling versus the time of reaction. Each
0
reaction protocol was conducted in the presence of gas feed containing 500 ppm NO
and either 1500 ppm H2 (left-hand column) or 1500 ppm CO accompanied by 2 vol.%
H2O (right-hand column) at an inlet temperature of either 450 (a and d) 350 (b and
◦
−1
e) or 250 C (c and f). Ar balance was employed to maintain GHSV = 40,000 h . Grey
panels indicate the switch from lean to rich conditions by removal of oxygen from
the feed gas.
◦
regard to N O formation during the low-temperature (250 C) tran-
2
sient experiments, small amounts were observed during only the
lean pulses when H2 was available directly in the feed. Formation
of approximately 20 ppm N O was observed over all catalyst for-
2
mulations. Conversely, in the WGS-assisted experiments, N O was
2
only detected during the rich phases. Furthermore, the formation of
N O over the Ba- and Ce-modified samples was higher (∼65 ppm)
2
compared to the Pd/Al O sample (∼15 ppm).
2
3
4. Discussion
In line with the previous study [23], we have shown that alu-
mina supported palladium catalysts are suitable for the selective
reaction of NO to NH3 using both a direct source of hydrogen but
also hydrogen formed from H O and CO via the WGS reaction. How-
2
ever, in the present study an extra layer of complexity has been
added by including a third component, i.e. barium or cerium, to the
Fig. 5. Steady-state formation of NH3 as a function of the stoichiometric number, S,
of the feed gas at 250, 350 and 450 C over Pd/Al2O3 (a and d), Ba/Pd/Al2O3 (b and
◦
Pd/Al O3 reference sample. To allow fair comparison of the inves-
e) and Ce/Pd/Al2O3 (c and f) in the presence of 500 ppm NO with either 1500 ppm
H2 (left-hand columns) or 1500 ppm CO accompanied by 2 vol.% H2O (right-hand
columns). The O2 concentration was varied between 0 and 1050 ppm (S = 0.33–1.73)
2
tigated formulations, a first-stage synthesis of 0.5 wt.% Pd/Al2O3
was performed in order to serve as a reference catalyst. This ref-
erence material was subsequently split into three aliquots, two of
which underwent subsequent modification by impregnation with
aqueous solutions of either barium or cerium precursor salts. This
−
1
in steps of 150 ppm and Ar was used as balance to maintain GHSV = 40,000 h
.
Highlighted data points represent responses which have not yet reached steady-
state.

E.C. Adams et al. / Catalysis Today 267 (2016) 210–216
215
method assumed that the dispersion of Pd throughout the parent
sample was homogeneous so that the noble metal particle size dis-
tribution in each impregnated sample could be considered equal
and therefore comparable, resulting in the additionally impreg-
nated compound (i.e., Ce or Ba) being the only factor responsible for
observed differences in activity. Márquez et al. [28] report on the
modelling of the preferred adsorption sites of Pd atoms onto the
surface of a stable (100) ꢀ-alumina support. They state that strong
reaction of NO + H2 was not observed to the same extent during
the transient lean/rich switching experiments. For most cases (with
the exception of low-temperature conditions) these samples now
exhibit behaviour similar to that of the Pd/Al O3 reference sample.
2
The short, excessively lean pulses employed during this experi-
ment means that conditions are not as strongly reducing as those
present in the temperature programmed experiment. Hence, it can
be assumed that kinetic effects due to SMSI, as discussed above,
are less pronounced resulting in higher NH3 formation. The pos-
itive effect of lean/rich cycling over these samples is interesting
and, since the concept of passive-SCR depends on a delicate type
of operational switching between different gas stoichiometries, it
may be of interest to further investigate the effect of transients over
this sample. Since relatively long (4 min) pulses were chosen in the
present investigation, shortening the lengths of pulses or having a
longer lean period compared to the rich one may further increase
the activity for NH3 formation over this formulation.
(
−3.81 to −3.27 eV) and localized bonds are formed between Pd
and tetrahedral sites of the alumina. As the synthesized Pd/Al O
2
3
catalyst was stabilized through calcination, it is highly unlikely that
the active phase would undergo any significant change as a result of
the secondary impregnation step conducted at room temperature.
Thus, the Pd structure is considered to be similar in all as-prepared
samples. However, due to strong metal-support interaction (SMSI)
during reaction, ammonia formation may be suppressed. This will
be discussed more below.
The formation of NH from NO and H was studied in the late 70s
The NOx storage properties of the BaO component were
observed when H2 was used directly in the feed gas during the
transient lean/rich switching experiments. Evidence for the stor-
age of NO during lean pulses is the delay in time before 500 ppm
NO was detected by the FTIR (not displayed) which is not seen for
the reference and cerium-doped samples. The time delay observed
is approximately 200 s and can be considered as the period required
for total surface saturation to take place; NO is oxidized on the palla-
dium sites prior to storage in the form of nitrates on the BaO phase.
Subsequently, when O2 is removed from the feed and conditions
are rich again, the stored nitrates are reduced by the supplied H2
[35,36]. This explains the slight increase in NH3 formation activity
observed when a switch from lean to rich conditions is initiated.
After a short time, the stored NOx supply is depleted and the NH3
formation activity over the Ba-modified sample reaches steady
state. No NOx storage was observed during the WGS-assisted tran-
sient reactions, explaining why there is no initial peak production
of NH3 when switches to rich conditions are employed.
With regard to the Ce-modified sample, an interesting obser-
vation of this sample in the presence of NO + H2 is the ability to
temporarily form low amounts of NH3 in the presence of slightly
oxidizing global conditions at low temperature. Although this effect
is not yet understood, the experiment has been repeated and the
results confirmed. Thus, this phenomenon is an area of interest for
future research and is most likely to be related to the oxygen stor-
age capacity of the ceria. In the experimental protocol conducted,
the oxygen feed was decreased stepwise in terms of concentra-
tion, starting from a large excess of that required for stoichiometric
conditions to be met. Thus, when exposed to the first stoichiomet-
ric value at which the formation of low amounts of ammonia is
observed (S = 1.53), this may indicate the concentration at which
the ceria is able to buffer the oxygen slightly, extending the period
required for total oxidation of the sample to take place. This is sug-
gested because a near but not fully steady-state response regarding
the formed NH3 is reached over these globally lean points, which
may indicate that some active sites of the sample are exposed to
a locally reducing environment due to the unique oxygen storage
properties of ceria. If these steps however, were allowed to reach
steady-state as a result of extended periods of time under oxidiz-
ing conditions, the oxygen storage capacity of the ceria will likely
diminish and the palladium sites subsequently become oxidized,
which may result in a loss of activity for ammonia formation. As
the purpose of the present investigation is to investigate potential
catalyst formulations suitable for passive-SCR automotive appli-
cations, better fuel economy can be achieved if NH3 can indeed
be formed under stoichiometric or, even better, slightly oxidizing
global conditions rather than more rich conditions.
3
2
when the TWC was being developed for automotive applications.
It has been reported that noble metal particles contain the active
sites required for this reaction to take place efficiently [29,30].
Thus it could be expected that, for the reaction of NO with H , the
2
formation of NH3 would be similar over all three samples if the
number and properties of available active sites in the samples are
the same. Although the NH3 formation over the cerium-modified
sample is similar though slightly lower compared to the reference
sample when H2 is directly available, the barium-modified sample
forms significantly less NH . This is most obvious in the temper-
3
ature programmed experiments in Fig. 4a. This may indicate that
the availability of palladium sites, active for NH3 formation, has
decreased as a result of the second impregnation step. It has been
reported that when Pt/Al O3 catalysts are impregnated with Ba,
2
the Pt sites can become partially covered with BaO, decreasing the
number of available Pt sites [16,31]. Wang et al. have also reported
on this effect and concluded it is due to the SMSI effect between
barium and the noble metal [32]. Furthermore, the authors found
that the coverage of Pt by barium is reversible, dependent on the
environment the sample is exposed to. Under reducing conditions,
barium migrates towards Pt and covers a portion of the Pt sites,
whereas under oxidizing conditions barium migrates from the Pt
sites. As the conditions of the temperature programmed experi-
ments are strongly reducing, it is possible that partial encapsulation
of Pd by both barium and cerium contribute to the loss in NH3
formation. Effectively, this would lead to a lower conversion of
NO due to lower number of available sites active for NO dissoci-
ation. Although this may explain the slightly lower NH3 formation
observed over the Ce-modified sample, such site blockage can-
not be solely responsible for the significantly lower formation of
NH3 observed for the Ba-modified sample. Furthermore, storage of
nitrogen oxides is ruled out, not only because rich conditions are
employed, but also because the length of experiment is sufficiently
◦
long to saturate the sample with NOx. Above 300 C during both
direct H2 and WGS-assisted feed gas exposure, NO is completely
converted. The complete conversion of NO shows that a sufficiently
high number of sites active for NO dissociation are accessible. Since
no NO2 or N O formation is detected at these temperatures, a
2
change in selectivity must occur. The selectivity towards NH3 is
likely suppressed by slow reaction between adsorbed N or NHx
species and adsorbed H atoms or by too low hydrogen coverage.
The change in selectivity can be due to charge transfer from barium
to palladium, affecting the electronic properties of Pd as a result of
barium addition. This has been observed for Pd catalysts both under
reducing [33] and oxidizing conditions [34] and may also be valid
for the Ce-modified sample.
Interestingly, the lower NH3 formation observed over the Ba-
and Ce-modified samples during the temperature programmed
A substantial benefit from the impregnation of cerium is seen
during all WGS-assisted NH3 formation experiments. This was

2
16
E.C. Adams et al. / Catalysis Today 267 (2016) 210–216
expected as the superior WGS reaction properties of cerium are
well reported in literature, though the majority of investigations
have been conducted by using ceria as the support material. Here
we show that the addition of merely 10 wt.% cerium into the cata-
lyst formulation results in a significant increase in activity for NH3
formation. As briefly mentioned in the introduction, the formation
of oxygen vacancies within ceria allows for the rapid dissociation of
water to take place prior to reaction with carbon monoxide that has
been adsorbed on the Pd sites. In the case where Al O is used as the
support material and cerium is absent in the catalyst formulation,
water activation is not as favourable. Kugai et al. [37] have com-
pared the WGS activity of catalysts supported by ceria and alumina
and suggest that the combination of noble metal and ceria support
Centre for Catalysis, which is financially supported by Chalmers
University of Technology, the Swedish Energy Agency and
the member companies: AB Volvo, ECAPS AB, Haldor Topsøe
A/S, Volvo Car Corporation, Scania CV AB, and Wärtsilä
Finland Oy.
References
[
[
1] M.A. Gómez-García, V. Pitchon, A. Kiennemann, Environ. Int. 31 (2005)
45–467.
2] N. Macleod, R. Cropley, J.M. Keel, R.M. Lambert, J. Catal. 221 (2004) 20–31.
2
3
4
[3] Y. Liu, M.P. Harold, D. Luss, Appl. Catal. B: Environ. 121 (2012) 239–251.
[
[
4] E. Fridell, M. Skoglundh, B. Westerberg, S. Johansson, G. Smedler, J. Catal. 183
1999) 196–209.
5] S.M. Park, M.-Y. Kim, E.S. Kim, H.-S. Han, G. Seo, Appl. Catal. A: Gen. 395
(2011) 120–128.
(
creates new active sites with improved H O activation behaviour
2
than the reportedly monofunctional sites present when alumina is
used as the carrier phase.
[6] A. Lindholm, N. Currier, J. Li, A. Yezerets, L. Olsson, J. Catal. 258 (2008)
73–288.
2
[
[
7] N. Macleod, R.M. Lambert, Appl. Catal. B: Environ. 46 (2003) 483–495.
8] C. Ciardelli, I. Nova, E. Tronconi, D. Chatterdee, T. Burkhardt, M. Weibel, Chem.
Eng. Sci. 62 (2007) 5001–5006.
5
. Conclusions
[
9] P. Forzatti, L. Lietti, I. Nova, E. Tronconi, Catal. Today 151 (2010) 202–211.
[
10] P.R. Dasari, R. Muncrief, M.P. Harold, Catal. Today 184 (2012) 43–53.
2
This study shows that, although the modification of Pd/Al O3
[11] N.V. Heeb, A.-M. Forss, S. Brühlmann, R. Lüscher, C.J. Saxer, P. Hug, Atmos.
Environ. 40 (2006) 5986–5997.
with cerium and barium has no beneficial effect on the formation
of NH3 via NO and directly available H2 during the temperature
programmed experiments, the addition of cerium proves advan-
[
12] T.D. Durbin, J.T. Pisano, T. Younglove, C.G. Sauer, S.H. Rhee, T. Huai, J.W. Miller,
G.I. MacKay, A.M. Hochhauser, M.C. Ingham, R.A. Gorse, L.K. Beard, D. DiCicco,
N. Thompson, R.J. Stradling, J.A. Rutherford, J.P. Uihlein, Atmos. Environ. 38
(2004) 2699–2708.
tageous when H O is used as the hydrogen source. This can be
2
[
[
13] C.D. DiGulio, J.A. Pihl, J.E.I.I. Parks, M.D. Amiridis, T.J. Toops, Catal. Today 231
justified by the enhanced water-gas-shift properties of ceria, thus
providing higher concentrations of hydrogen available for ammo-
nia formation compared to those samples without cerium. The
cerium-modified sample also shows positive behaviour during the
oxygen step experiments when H2 is directly available in the feed
gas. Namely, the formation of NH3 is still substantial at stoichio-
metric and even slightly oxidizing global environments of the feed
gas. This is suspected to be due to the oxygen storage capacity
of the cerium component but requires further studies to be con-
firmed. It is important however, to keep in mind that this is most
likely a temporary effect which is only observed in the presence
of conditions which are still very close to stoichiometric opera-
tion. In automotive applications, lean operation would occur in
much longer pulses, with a much greater oxygen excess in order to
conserve fuel economy. Thus, the absence of NH3 formation over
the Ce-modified sample during transient pulsing is a better rep-
resentation of how this sample would behave in a real application.
Although the impregnation with barium to the catalyst formulation
(
2014) 33–45.
14] V.Y. Prikhodko, J.E. Parks, J.A. Pihl, T.J. Toops, SAE Int. J. Engines 7 (2014)
1235–1243.
[
[
[
15] S.H. Oh, T. Triplett, Catal. Today 231 (2014) 22–32.
16] L. Castoldi, I. Nova, L. Lietti, P. Forzatti, Catal. Today 96 (2004) 43–52.
17] J.A. Pihl, J.E. Parks, C.S. Daw, T.W. Root, SAE Tech. Paper 2006-01-3441,
2006.
[
[
18] L. Cumaranatunge, S.S. Mulla, A. Yezerets, N.W. Currier, W.N. Delglass, F.H.
Ribeiro, J. Catal. 246 (2007) 29–34.
19] R.D. Clayton, M.P. Harold, V. Balakotaiah, Appl. Catal. B: Environ. 84 (2008)
616–630.
[20] N.W. Cant, D.C. Chambers, I.O.Y. Liu, Appl. Catal. B: Environ. 46 (2003)
51–559.
5
[
21] B. Perada-Ayo, U. De La Torre, J.R. González-Marcos, M.P. amd
González-Velasco, Catal. Today 241 (2015) 133–142.
[22] V.M. Shinde, G. Madras, Appl. Catal. B: Environ. 123–124 (2012) 367–378.
[
23] E.C. Adams, M. Skoglundh, M. Folic, E.C. Bendixen, P. Garbrielsson, P.-A.
Carlsson, Appl. Catal. B: Environ. 165 (2015) 10–19.
[24] J. Dawody, M. Skoglundh, S. Wall, E. Fridell, J. Mol. Catal. A 225 (2005)
259–269.
[
[
25] S. Brunauer, P.H. Emmett, E. Teller, J. Am. Chem. Soc. 60 (1938) 309–319.
26] H. Kannisto, X. Karatzas, J. Edvardsson, L.J. Pettersson, H.H. Ingelston, Appl.
Catal. B: Environ. 104 (2011) 74–83.
did not affect the activity for NH formation during static conditions
[27] J.C. Summers, K. Baron, J. Catal. 57 (1979) 380–389.
3
[28] A.M. Márquez, J.F. Sanz, Appl. Surf. Sci. 238 (2004) 82–85.
[29] H.G. Stenger Jr., J.S. Hepburn, Energy Fuels 1 5 (1987) 412–416.
[30] T.P. Kobylinski, B.W. Taylor, J. Catal. 33 (1974) 376–384.
in the presence of either feed gas mixture, it was interesting that
cycling of the feed gas between lean and rich conditions resulted
in improved activity for NH3 formation. As reasonably long pulses
were conducted in this paper, it may be of interest in the future
to determine whether shortening the length of the pulses or even
altering the balance between lean and rich operation would result
in higher activity over this sample.
[31] V. Labalme, N. Benhamou, N. Guilhaume, E. Garbowski, M. Primet, Appl. Catal.
A: Gen. 133 (1995) 351–366.
[
32] X. Wang, D.H. Kim, J.H. Kwak, C. Wang, J. Szanji, C.H.F. Peden, Catal. Today 175
2011) 78–82.
(
[33] D.T. Wickham, B.W. Logsdon, S.W. Cowley, J. Catal. 128 (1991) 198–209.
[34] F. Klingstedt, H. Karhu, A. Kalantar Neyestanaki, L.-E. Lindors, T. Salmi, J.
Vayrynen, J. Catal. 206 (2002) 248–262.
[
35] T. Kobayashi, T. Yamada, K. Kayano, SAE Tech. Paper 970745, 1997.
[36] S. Matsumoto, Cattech 4 (2000) 102–109.
37] J. Kugai, E.B. Fox, C. Song, Appl. Catal. A: Gen. 456 (2013) 204–214.
Acknowledgements
[
This work was financially supported by the Swedish Energy
Administration through the FFI program and the Competence