Ammonia is a hydrogen carrier in the regeneration of Pt/BaO/Al2O3 NOx traps with H2

Article abstract of DOI:10.1016/j.jcat.2006.11.008

We report that the regeneration of a Pt/BaO/Al₂O₃ NOx trap in H₂ occurs through the formation of NH₃ as an intermediate. When NH₃ is used instead of H₂, the reduction process is equivalent and equally effective to using H₂, thus confirming our hypothesis. The regeneration is localized, occurring at the plug flow front and is limited by mass transfer of the gas phase reductant to the catalyst. The process is consistent with the release of NOx from the trapping material, followed by its reduction over Pt. The high selectivity of Pt/BaO/Al₂O₃ in forming mostly N₂ during regeneration is achieved by the release of NOx only in the presence of H₂ which guarantees the formation of N₂ and NH₃ and only small amounts of N₂O. An oxygen source on the catalyst support, usually the stored NOx, is necessary for the oxidation of NH₃ to N₂.

Full text of DOI:10.1016/j.jcat.2006.11.008

Journal of Catalysis 246 (2007) 29–34
www.elsevier.com/locate/jcat
Ammonia is a hydrogen carrier in the regeneration of Pt/BaO/Al2O3 NOx
traps with H2
L. Cumaranatunge a, S.S. Mulla a, A. Yezerets b, N.W. Currier b, W.N. Delgass a, F.H. Ribeiro a,∗
a
School of Chemical Engineering, Purdue University, Forney Hall of Chemical Engineering, 480 Stadium Mall Drive, West Lafayette, IN 47907-2100, USA
b
Cummins Inc., 1900 McKinley Ave, Columbus, IN 47201, USA
Received 15 September 2006; revised 7 November 2006; accepted 7 November 2006
Available online 14 December 2006
Abstract
We report that the regeneration of a Pt/BaO/Al O NOx trap in H occurs through the formation of NH as an intermediate. When NH is
2
3
2
3
3
used instead of H , the reduction process is equivalent and equally effective to using H , thus confirming our hypothesis. The regeneration is
2
2
localized, occurring at the plug flow front and is limited by mass transfer of the gas phase reductant to the catalyst. The process is consistent with
the release of NOx from the trapping material, followed by its reduction over Pt. The high selectivity of Pt/BaO/Al O in forming mostly N
2
3
2
during regeneration is achieved by the release of NOx only in the presence of H which guarantees the formation of N and NH and only small
2
2
3
amounts of N O. An oxygen source on the catalyst support, usually the stored NOx, is necessary for the oxidation of NH to N .
2
3
2
© 2006 Elsevier Inc. All rights reserved.
Keywords: Mechanism of reduction of NOx traps; Pt/BaO/Al O NOx traps; Ammonia as a NOx reduction intermediate
2
3
1. Introduction
A few studies have been devoted to the analysis of the re-
generation or the reduction phase. Burch and Millington [5]
proposed a mechanism in which NO is decomposed on reduced
Pt sites, and the role of the reductant is to reduce the oxidized
Pt to Pt0, which is capable of reducing NO to N2. On the other
hand, direct reaction between released NOx species and the
reductant molecules on the precious metal has also been pro-
posed [6]. Here the debate on the possible mechanism govern-
ing NOx release remains open. Kabin et al. [7] proposed that the
NOx is released as a result of the heat generated by the exother-
mic reactions upon switching to the regenerating gases (thermal
release), while others have proposed a decrease in the equilib-
rium stability of the stored nitrates due to either the decrease
in the partial pressure of oxygen [8,9] or the establishment of
a net reducing environment [9,10]. In addition, Liu and Ander-
son [9] have proposed a different mechanism of NOx release in
which the reductant molecule (or its activated form spilled over
from Pt) may interact directly with the stored NOx on the stor-
age component where the nitrates are reduced to nitrites which
then release NO and gaseous oxygen. Recently, Nova et al. [11]
reported a similar mechanism that involves the activation of H2
on Pt sites, followed by spillover on the alumina support toward
The NOx storage/reduction (NSR) process is one of the tech-
nologies in development to abate the NOx (NO + NO2) in the
exhaust emitted from combustion engines. This system works
by the Pt catalyst transforming the NO in the exhaust to NO2,
which then reacts with a barium or potassium component to
form a stable compound, which is periodically reduced to re-
lease most of the nitrogen as N2 [1]. This cycle of oxidation,
trapping, release and reduction is performed, for example, with
a 60 s capture (oxidation and trapping) followed by a four sec-
ond regeneration (release and reduction). On a commercial sys-
tem, the efficiency in transforming the emitted NOx to N2 is
over 95%. This is remarkable, as the Pt itself is selective for
the formation of N2 from NO and H2 only in a narrow range of
NO to H2 ratio, as reported in the literature [2,3] and as we will
show below. A more detailed account of this technology can be
found in a recent review [4].
*
Corresponding author. Fax: +1 765 494 0805.
E-mail address: fabio@purdue.edu (F.H. Ribeiro).
0021-9517/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.11.008

30
L. Cumaranatunge et al. / Journal of Catalysis 246 (2007) 29–34
nitrate adspecies that decompose to gaseous NOx, which is then
reduced on Pt or is directly reduced by spilled over hydrogen.
They also included the possibility of a mechanism involving
surface diffusion of NOx adspecies toward reduced Pt sites,
which reduce the NOx to N2.
The objective of this contribution is to gain further insight
into the mechanism governing the reduction of stored NOx
species and provide a model of how the reduction process oc-
curs, including why it is so selective to N2. We will show that
the reduction process can be explained by the release of NO or
NO2 which is then optimally reduced to N2 or over-reduced to
NH3 on the Pt clusters. If N2O is formed, most of it continues to
react with hydrogen on the Pt surface to form N2 or NH3. Most
importantly, we will show that the NH3 formed is as effective
as H2 in the reduction/regeneration process and is eventually
transformed to N2.
NO, NO2, N2O, NH3 and H2O concentrations in the outlet
gas stream were detected with an FTIR gas analyzer (MKS
MultiGasTM Analyzer, Model 2030), while the N2 concentra-
tion was detected with a quadrupole mass spectrometer (SRS
RGA 200). The mass spectrometer was calibrated to measure
N2 concentrations in the 0–6500 ppm range either by the in-
jection of pulses of known volumes of N2 or by sampling cal-
ibrated N2/Ar mixtures. Argon was used as the carrier gas to
allow for the measurement of the released N2. Mass flow con-
trollers were used to control all the gas flows except for the
experiments where high concentrations (>1000 ppm) of NH3
and NO were utilized. These flows were controlled through nee-
dle valves. The system was automated to switch 3-way valves
between lean-rich cycles. Thermocouples were placed 6 mm
before and after the catalyst sample to verify inlet and outlet
gas temperatures.
2. Experimental methods
3. Results and discussion
The Pt/Al2O3 and Pt/BaO/Al2O3 catalysts used in this study
were supplied by EmeraChem in monolithic form. Both mono-
liths had a cell density of 200 channels per in2. The Pt loading
for both samples was ca. 50 g ft−3 of monolith. The Ba loading
in the Pt/BaO/Al2O3 sample was 20 wt%. The Pt/Al2O3 sample
was cut into a 1-inch long core weighing about 3 g and having
a cross-section of 60 channels. The total gas flow rate over this
sample was 6.6 L min−1 (space velocity of 80,500 h−1). The
percentage of metal exposed (PME) or metal dispersion of this
catalyst, defined as the ratio of the number of surface Pt atoms
to the total number of Pt atoms, was measured by H2–O2 titra-
tion [12] and was 42%. The Pt/BaO/Al2O3 catalyst was cut into
a 3-inch long core with a cross section of 60 cells. The total
flow rate over this sample was 7.0 L min−1 corresponding to
a gas hourly space velocity of 30,000 h−1. The PME of Pt in
this sample was 60%. This gives an exposed Pt content of 6.2
and 8.2 µmol g−1 for the Pt/Al2O3 and Pt/BaO/Al2O3 samples,
respectively.
Fig. 1 shows a comparison of the evolution of the outlet gas
concentrations from a Pt/BaO/Al2O3 catalyst after the switch
to regeneration gases containing either H2 or NH3. The solid
lines indicate regeneration with 0.75% H2/Ar and the dashed
lines indicate regeneration with 0.53% NH3/Ar. The number of
hydrogen atoms per unit of time flowing over the sample in
the regenerating gas mixture was kept nearly identical in both
cases to illustrate the effectiveness of H2 and NH3 for regener-
ation. The regeneration phase (ca. 3 min long) was preceded by
a 7 min long trapping (lean) phase that contained 350 ppm NO
and 10% O2, balance Ar, and a 5 s purge with Ar. To obtain re-
producible results, it was necessary to run 4–5 lean-rich cycles.
The nitrogen balance between capture and regeneration phases
was found to close within experimental error. After regenera-
tion for 3 min with H2 or NH3, the same amount of NOx was
stored (0.54 mmol) during the subsequent 7-min capture phase,
and the N2 selectivities for both reductants were similar.
As seen in Fig. 1A, the N2 and H2O traces for both regener-
ating mixtures increase rapidly to a constant value until the end
when they sharply decrease. This rectangular wave shape in-
dicates a “plug flow” type of mechanism implying a complete
The experimental apparatus used for this study is described
in detail elsewhere [13] and briefly here. All the experiments
reported here were run at 300 ◦C, except where specified. The
◦
Fig. 1. Evolution of the species after the switch to the regeneration phase following a 7 min trapping period with 350 ppm NO/10%O /Ar at 300 C over a
2
Pt/BaO/Al O monolith. The solid lines represent regeneration with 0.75% H /Ar and the dashed lines represent regeneration with 0.53% NH /Ar. The space
2
3
2
3
−1
velocity over the catalyst is 30,000 h . (A) N , H O and NH traces, (B) NO , NO and N O traces.
2
2
3
2
2

L. Cumaranatunge et al. / Journal of Catalysis 246 (2007) 29–34
31
reaction between the reductant and the NOx to produce N2 and
H2O. Since the reductant is the limiting reagent, this implies
that the time for regeneration should be inversely proportional
to the amount fed per unit time, as we verified experimentally.
Note, however, that the data do not provide a mechanism for
how the NOx is released from the trapping sites. Fig. 1A also
shows that the H2O and N2 traces in the case of NH3 as the
reductant continue at constant levels for a 15% longer dura-
tion before starting to decrease, when compared to the case
with H2 as the reductant. This difference is caused by the un-
certainty in the reproducibility of the total flow rate. The H2O
trace for NH3 as a reductant did not return to zero because the
NH3/Ar mixture contained H2O as an impurity. Accounting for
this H2O impurity, the area under the H2O traces of Fig. 1A
for the two reductants are the same within error (ca. 1.39 and
1.31 mmol). The N2 level is higher when NH3 is used as the
reductant (0.74 mmol vs 0.22 mmol) since in addition to the ni-
trogen species in the stored NOx, NH3 also contributes (through
oxidation of NH3 by the NOx) to the total N2 formed. This
makes the calculation of N2 selectivity less precise.
Since the NH3 and H2 (H2 data not shown since only non-
quantifiably small H2 amounts were observed in the mass spec-
trometer before breakthrough, and the H2 concentration mea-
surement was not precise) traces evolve close to the end of the
cycle, the reductants are proposed to be limiting in the regen-
eration phase. As seen in Fig. 1A, NH3 appears in the effluent
(slips) only after 30–35 s into the regeneration cycle. The NH3
(or H2) is consumed below detection level until this point and
starts to slip only when the stored NOx starts to deplete toward
the end of the catalyst bed. Thus, the shape of the NH3 evolu-
tion curves for the H2 and NH3 cases are consistent with our
plug flow model. In the case of regeneration by H2, the NH3
evolution curve is the result of the competition between the
generation (by NOx–H2 reaction) and consumption (discussed
below) of NH3 at the end of the catalyst bed. In the case of
regeneration by NH3, the NH3 evolution curve has the charac-
teristic “S” shape of strong gas adsorption seen, for example,
during adsorption of NO2 on BaO. Examining the NH3 curves,
we propose that the reason why NH3 is seen only toward the end
of the cycle must be due to the fact that the NH3 front moves
along the length of the catalyst bed in a plug flow manner while
getting consumed in reducing NOx to N2 (thereby regenerating
the catalyst). When the NH3 front reaches the end of the cata-
lyst bed, it begins to break through due to the absence of NOx
to oxidize the NH3 to N2 and H2O. This breakthrough is simi-
lar to the model presented by Epling et al. [14] for NOx storage,
where the NOx sorption zone propagates down the catalyst bed
in a plug flow manner and begins to breakthrough after reaching
the end of the catalyst bed.
tant was completely consumed during the initial stages of the
regeneration phase. The time delay in evolution of H2 decreased
proportionally with increasing H2 concentration as should be
the case if the reductant is the limiting species in this reac-
tion. We also observed that lowering the temperature for the
tests to 242 ◦C made no difference on the reduction profiles,
except that the amount of N2O produced was about twice as
high. This insensitivity to temperature suggests that the regen-
eration is limited by transport of reactants and not by kinet-
ics.
The sharp rise in the N2 trace, as seen in Fig. 1A, implies
that the H2O trace, the other product of the reduction reaction,
should follow a similar profile. Although the curve shapes are
similar, a delay is seen in the H2O trace, relative to N2, for
both reductants. We investigated this phenomenon by including
H2O (ca. 7.5%) in both the capture and regeneration phases.
Interestingly, the usual NOx spike that arises immediately after
the switch to the regeneration phase decreased by a factor of
two compared to the dry feed conditions, and the H2O trace
had a sharp rise similar to the N2 trace (i.e., no delay) when
H2O was included in the feed. In light of this data, we propose
the following. It is well known that H2O decreases the NOx
storage capacity of the NSR catalysts [4] and hence H2O could
be competing for some of the NOx sites. In the absence of H2O
in the trapping or lean-phase, the sites that generally favor the
adsorption of H2O over NOx are occupied by NOx. However,
when H2O is formed during the regeneration phase due to the
reductant coming into contact with either residual O2 or stored
NOx, it first adsorbs on those sites that favor H2O adsorption
over NOx, causing the delay in H2O evolution, while displacing
the NOx that were stored on those sites, resulting in a larger
NOx spike. When H2O is added to the feed during the trapping
phase, it is preferentially adsorbed on some of the sites during
the capture phase (decreasing NOx storage) and this prevents
the adsorption of the H2O that is formed during the regeneration
phase, resulting in the sharp rise (no delay) in the H2O trace and
a decreased NOx spike. However, a small NOx spike (about 5%
of total NOx stored) is still observed at the beginning of the
regeneration phase even in the presence of H2O in the trapping
phase. We hypothesize that this is due to the combination of
desorption (explained below) and a few highly reactive sites
that release NOx as a result of the low concentration of H2 or
NH3 that contacts these sites in the initial phase of regeneration.
The NO and N2O traces in Fig. 1B are similar for both the
reductants except for a higher initial spike with NH3. The con-
centration of NO and N2O (after the spike) is approximately
constant with time. We propose that their shape is a result of
depletion of H2 at the end of the moving reduction front, and
the reactions that take place in the non-reducing environment
encountered there. In particular, we have observed that the re-
action between NO and reduced Pt in the absence of adsorbed
hydrogen will produce N2O until the surface is titrated to Pt–O
and the reaction stops. Fig. 1B also shows that the NO2 de-
crease (after the initial rise) is linear with time in both cases.
We propose that this is simply due to the desorption of NO2
arising from the shift in equilibrium between the surface and
the gas phase that is accompanied with the switch to the re-
Experiments performed by varying the H2 concentration in
the regenerating phase over the range of 1.0–2.5% at the same
total flow on the Pt/BaO/Al2O3 catalyst (that has stored a sim-
ilar NOx amount during the preceding capture phase) have
shown that the time required for regeneration is inversely pro-
portional to H2 concentration. The selectivity to N2 was main-
tained at 80–85%. In all cases, the evolution of H2 (observed in
the mass spectrometer) was delayed, indicating that the reduc-

32
L. Cumaranatunge et al. / Journal of Catalysis 246 (2007) 29–34
generating gases [9]. As the catalyst bed is reduced in a plug
flow type mechanism, the amount of NO2 available for desorp-
tion ahead of the front decreases linearly with time, resulting
in a linear decrease in NO2 evolution with time. To reinforce
the hypothesis of NO2 desorption, we performed experiments
with an inert substitute (Ar only with no reductant) in the re-
generation phase spanning the same regeneration time period
(ca. 3 min) as with H2. After a 7 min trapping phase containing
350 ppm of NO, 10% of O2, balance Ar, we observed 12% of
the stored NOx desorbing as NO2 and 6% as NO under the inert
flow conditions during the 3 min regeneration period compared
to 7% as NO2, 3% as NO under H2 rich conditions.
To summarize the findings, NH3 or H2 are capable of re-
generating the trap in a similar way. We will investigate next
the reaction of NH3 or H2 with NO over a Pt/Al2O3 catalyst
(without the trapping component) as a way to simulate the re-
duction steps (Figs. 2A and 2B). Most of the experiments were
carried out under nearly isothermal conditions (300 ◦C). How-
ever, in experiments where a significant amount of NH3 was
formed, the outlet gas temperature was 30–40 ◦C higher than
that of the inlet. For this reason we carried out experiments at
200 ◦C to make sure the selectivity was not due to a hot spot.
We observed similar product selectivity to that at 300 ◦C. The
Pt/Al2O3 monolith catalyst was first exposed to NO/Ar mixture
for 2 min and then H2 (or NH3) was added to the NO flow at
varying NO/reductant ratios for 6 min (the reactions reached
steady state in less than 1 min). The data reported here was ob-
tained by averaging the concentrations after 2 min of reaction.
The NO/reductant (H2 or NH3) ratio was varied to simulate
the different NOx/reductant environments that occur along the
NSR trap.
an atomic N basis. Fig. 2A shows that as the NO/H2 ratio in-
creases from 0.4 (excess H2) to 4.6 (excess NO) the product
selectivity changes from mostly NH3 at low NO/H2 ratios to
N2O at high NO/H2 ratios. The highest N2 selectivity was ob-
served for a stoichiometric ratio (1:1) of NO/H2. The formation
of mostly NH3 under low NO/H2 ratios (reducing conditions)
has previously been observed by Shelef et al. [15]. The prod-
uct selectivities we have obtained for the NO/H2 reaction over
the Pt/Al2O3 sample are similar to that reported by Kosaki et
al. [2] for their Pt/Al2O3 powdered catalyst. Similarly, as the
NO/NH3 ratio is varied from 0.67 (excess NH3) to 5 (excess
NO), Fig. 2B, the product selectivity changes from mostly N2
at low NO/NH3 ratios to N2O at high NO/NH3 ratios. Otto et
al. [16] have previously shown that NH3 is an effective reduc-
tant in the removal of NO from waste effluents over supported
Pt. Although similar experiments with NO2 + H2 or NH3 were
not performed, we expect similar results with respect to the
product selectivity as with the experiments with NO described
above, with the possibility of NO2 being more reactive. We have
also carried out experiments involving N2O and H2 at 300 ◦C
over the Pt/Al2O3 catalyst. In this case, with ca. 50 ppm N2O
and ca. 1% H2 at 300 ◦C, the N2O was reduced to N2 and H2O
at a total conversion of 90%. These results for reduction with
H2 or NH3 show that selectivity to N2 is high only in specific
ranges of oxidant/reductant ratios. This information is impor-
tant to the formulation of our regeneration model.
Fig. 3 shows a schematic of the proposed regeneration mech-
anism for a single monolith channel of a Pt/BaO/Al2O3 NSR
catalyst. The figure illustrates the propagation of the reductant
front along the catalyst bed with complete regeneration of the
trapping sites. A zoomed-in version of the chemistry occurring
within the reductant front is also illustrated in Fig. 3. It shows
that as the NOx (NO + NO2) is released from the trapping
sites (the exact mechanism of release is not known), it reacts
with H2 over Pt to form NH3, N2 and N2O as the N-containing
species. The selectivity of the individual species will depend on
the local NOx/H2 concentration ratios as shown in Fig. 2A. In
regions where the H2 level is high compared to the NOx, the re-
action with the released NOx over Pt will form mostly NH3 and
some N2. The NH3 formed will further react with more NOx to
As seen in Fig. 2 when all the reduction chemistry occurs
on Pt, then the steady state selectivity to N2 is a strong func-
tion of the NO/reductant ratio. The H2 conversions noted in
Fig. 2A were calculated based on the amounts of N2, NH3 and
N2O formed and the appropriate reaction stoichiometry, since
the H2 concentration was not measured precisely. The selectiv-
ity to N2 is defined as the ratio of two times the total amount of
N2 formed to the total amount of nitrogen species in the prod-
uct (NO2, N2O, NH3, and N2) formed during the reaction on
◦
Fig. 2. Steady-state product selectivity and reactant conversion at 300 C as a function of NO: reductant ratio for the reaction of (A) NO + H and (B) NO + NH
2
3
over a Pt/Al O monolith.
2
3

L. Cumaranatunge et al. / Journal of Catalysis 246 (2007) 29–34
33
Fig. 3. Schematic of the reduction mechanism for a Pt/BaO/Al O monolith regenerated with H . Bottom panel illustrates a zoomed-in picture of the reaction front.
2
3
2
give either N2 or N2O. If the N2O is formed behind the front, it
will be reduced to N2 by H2/NH3. As the front approaches the
end of the catalyst, the supply of NOx starts to deplete and will
be insufficient to react with the NH3 formed upstream, leading
to the NH3 breakthrough seen in Fig. 1A. Before the reaction
front reaches the end of the bed, most of the non-selective prod-
ucts formed by the reaction of H2 with NOx over Pt, such as
N2O and NH3, will further react to form N2 and maintain the
high N2 selectivity of the overall NSR catalyst. Using this ra-
tionale, and by reference to Fig. 2A, one might expect that as
the concentration of H2 is depleted at the end of the front, N2O
should become a major product. This is not observed experi-
mentally. We believe the reason is that there is a self-adjusting
mechanism. As the concentration of H2 is lowered, the amount
of released NO is also lowered thus keeping the local overall ra-
tio unfavorable to the formation of N2O. The N2O is formed in
a stoichiometric reaction between Pt and NO to form N2O and
Pt–O and is confined to the leading edge of the reaction front.
The reduction model proposed indicates that the trapping
phase of the NSR cycle should not extend up to the full ca-
pacity of the trap since in that case the NOx released during
regeneration would not be captured downstream. Furthermore,
we propose that having an oxygen storage capacity (OSC) sub-
strate such as ceria would allow the oxidation of NH3 to N2
when there are no more nitrates to oxidize it, and, hence, the
OSC would decrease the NH3 slip in the NSR catalysts.
anism of surface diffusion of NO-containing species that are
reduced on Pt. We propose a localized reaction front which
travels through the bed with complete regeneration of the trap-
ping sites as it propagates down the catalyst bed. This chemistry
seems to be fast enough to make the process mass transfer lim-
ited and does not depend on whether the reactant is H2 or NH3.
Thus, the reaction kinetics and associated rate constants are not
necessary to model the quantitative behavior of this system. The
efficiency of the NSR catalyst in converting NOx to mostly N2
is achieved because the other products that form over Pt, such
as N2O and NH3, are either not favored (N2O) or can further re-
act to produce N2. The low observed amounts of NO and N2O
are explained by a self-limiting NO production rate as the H2
concentration is depleted at the leading edge of the front. The
hydrogen on the surface of Pt will react first to reduce NO and
N2O before it will spillover to release more NO. As the hydro-
gen on the surface is depleted, the NO in the gas phase will
adsorb and react on Pt to produce N2O until the surface is ox-
idized and the remaining NO will escape. The NO2 profile is
due to simple desorption. The NH3 profile measured at the end
of the cycle when using H2 as a reductant is a result of insuf-
ficient NOx left in the bed to consume the NH3 formed as the
front reaches breakthrough.
Acknowledgments
The authors thank the Indiana 21st Century Research and
Technology Fund and Cummins Inc. for financial support of
this work.
4. Conclusion
We have proposed a model for reduction by H2 where NH3
is the hydrogen carrier. In this model, derived from the results
presented here, NO and NO2 are first released into the gas phase
and then reduced over Pt. It is not necessary to invoke a mech-
References
[1] M. Takeuchi, S.i. Matsumoto, Top. Catal. 28 (2004) 151.
[2] Y. Kosaki, A. Miyamoto, Y. Murakami, Chem. Lett. 8 (1979) 935.

34
L. Cumaranatunge et al. / Journal of Catalysis 246 (2007) 29–34
[3] J. Siera, P. Cobden, K. Tanaka, B.E. Nieuwenhuys, Catal. Lett. 10 (1991)
[10] S. Poulston, R.R. Rajaram, Catal. Today 81 (2003) 603.
[11] I. Nova, L. Lietti, L. Castoldi, E. Tronconi, P. Forzatti, J. Catal. 239 (2006)
244.
[12] J.E. Benson, M. Boudart, J. Catal. 4 (1965) 704.
[13] S.S. Mulla, N. Chen, L. Cumaranatunge, W.N. Delgass, W.S. Epling, F.H.
Ribeiro, Catal. Today 114 (2006) 57.
335.
[4] W.S. Epling, L.E. Campbell, A. Yezerets, N.W. Currier, J.E. Parks, Catal.
Rev. 46 (2004) 163.
[5] R. Burch, P.J. Millington, Catal. Today 26 (1995) 185.
[6] T. Maunula, J. Ahola, H. Hamada, Appl. Catal. B 26 (2000) 173.
[7] K.S. Kabin, R.L. Muncrief, M.P. Harold, Catal. Today 96 (2004)
79.
[14] W.S. Epling, J.E. Parks, G.C. Campbell, A. Yezerets, N.W. Currier, L.E.
Campbell, Catal. Today 96 (2004) 21.
[15] M. Shelef, J.H. Jones, J.T. Kummer, K. Otto, E.E. Weaver, Environ. Sci.
Technol. 5 (1971) 790.
[8] A. Amberntsson, H. Persson, P. Engstrom, B. Kasemo, Appl. Catal. B 31
(2001) 27.
[16] K. Otto, M. Shelef, J.T. Kummer, J. Phys. Chem. 74 (1970) 2690.
[9] Z. Liu, J.A. Anderson, J. Catal. 224 (2004) 18.