Investigation of NO reduction by H2 on Pd monolith with transient and isotopic exchange techniques: II. H2/D2 exchange in the reduction of NO

Article abstract of DOI:10.1006/jcat.2002.3669

The kinetics and mechanism of catalytic reduction of nitrogen oxide (NO) by hydrogen on an alumina-based palladium monolith were studied under atmospheric pressure at 155°C. Transient kinetic experiments, as well as isotopic exchange techniques, were appl

Full text of DOI:10.1006/jcat.2002.3669

Journal of Catalysis 210, 30–38 (2002)
doi:10.1006/jcat.2002.3669
Investigation of NO Reduction by H2 on Pd Monolith with Transient
and Isotopic Exchange Techniques
II. H2/D2 Exchange in the Reduction of NO
∗
∗
∗
K. Rahkamaa-Tolonen,∗,1 T. Salmi, D. Yu. Murzin, L. Barreto Dillon, U. Lassi,† and R. L. Keiski†
❛
❛
∗
Laboratory of Industrial Chemistry, Process Chemistry Group, Faculty of Chemical Engineering, Abo Akademi, FIN-20500 Turku-Abo,
Finland; and †Department of Process and Environmental Engineering, University of Oulu, FIN-90014 Oulu, Finland
Received October 15, 2001; revised March 13, 2002; accepted May 6, 2002
catalysts, which leads to the decay in active NOx -adsorption
sites and causes a rapid decline in the catalyst activity (2, 3).
Therefore, selective catalytic reduction (SCR) has turned
out to be a more successful way to remove NO. The current
SCR technology uses NH3; however, there are drawbacks
withthisprocessandtheutilisationofotherreductants, such
as CH4, CO, and H2, is desirable (4). Hydrogen can act as a
reductant of NO on noble metal catalysts, provided the con-
centrationofO2 islowintheexhaustgas(5). Inthepresence
of excess of O2, hydrogen is easily consumed in the simple
combustion with O2 (6, 7), but not in the reactions with ni-
trogen oxides. As a result of the thermal cracking reactions
that take place in the flame, especially with incomplete com-
bustion, hydrogen is formed and emitted, as are hydrocar-
bons that are different from the ones present in the fuel (8).
It is, therefore, worth revealing the surface reaction mech-
anisms of NOx reduction by H2. One of the main practical
applications of transient methods in catalytic kinetics has
been undoubtedly the study of transient behaviour of three-
way automotive catalysts. It is well-known that these cata-
lysts normally operate under transient conditions (9, 10).
Fundamental knowledge related to elementary reaction
kinetics and mechanisms on single-metal components of
three-way catalysts will also be indispensable in the design
of improved converters. In the present work, the catalytic
reduction of NO by H2/D2 was investigated with transient
step-response experiments. Isotopic exchange experiments
provide information on the behaviour of the reactants.
Deuterium step changes were utilised also to trace the hy-
drogen reaction pathways during the water and ammonia
formation. The adsorption and dissociation of NO was ex-
amined by temperature-programmed desorption (TPD).
The kinetics and mechanism of catalytic reduction of nitrogen
oxide (NO) by hydrogen on an alumina-based palladium monolith
were studied under atmospheric pressure at 155◦C. Transient ki-
netic experiments, as well as isotopic exchange techniques, were
applied in order to improve understanding of the reactions occur-
ring on the surface of the noble metal catalyst. Nitrogen, nitrous
oxide, ammonia, and water were detected as reaction products in
the reduction of nitrogen oxide by hydrogen. The dissociation of
NO on the catalyst surface is the crucial step affecting the overall
reaction. Ammonia was formed by stepwise hydrogenation of ad-
sorbed nitrogen atoms, which was confirmed by isotope transient
techniques. Reaction pathways explaining the reduction of NO with
c
hydrogen and deuterium were proposed. ꢀ 2002 Elsevier Science (USA)
KeyWords:nitrogenoxides;hydrogen;deuterium;reduction;pal-
ladium; monolith; transient kinetics; isotopic exchange; TPD.
1. INTRODUCTION
Emissions from vehicles are suppressed by catalytic con-
version, i.e., total oxidation of carbon monoxide and hydro-
carbons and reduction of nitrogen oxides. In environmental
catalysis, the main issue is to optimise catalytic converters
to remove pollutants. Therefore, a detailed understanding
of the kinetic processes is needed. The reaction mechanisms
for the abatement of nitrogen oxides (NOx ) are of particu-
lar interest, since they are environmentally very unfriendly
compounds. Most of the modern vehicles have after-
treatment devices, typically monolithic catalytic converters,
to transform pollutants in the exhaust gas to less harmful
components. A principally straightforward way to remove
NO would be self-decomposition. The noncatalytic reac-
tion is retarded by the high activation energy (Ea > 364 kJ/
mol) (1). Kinetically, the reaction rate can be enhanced via
decreasing the activation energy with an effective catalyst.
The main obstacle of the self-decomposition of NO and
N2O is, however, the strong oxygen adsorption on most
2. EXPERIMENTAL
2.1. Catalysts
1 To whom correspondence should be addressed. Fax: +358-2-2154479.
E-mail: Katariina.Rahkamaa@abo.fi.
Pd-based metallic catalysts provided by Kemira Metalkat
Oy (Finland) were used in this study. The catalyst sample
30
0021-9517/02 $35.00
c
ꢀ 2002 Elsevier Science (USA)
All rights reserved.

NO REDUCTION BY H2
31
consisted of flat and corrugated metal foils coated with a after the steady state was attained during the NO + H2 re-
support, mainly containing γ -Al2O3. Noble metals were action. Calibration was carried out with gas mixtures con-
impregnated on the support in an aqueous solution. The taining known concentrations of the components before
supportmaterialcontentofthealumina-supportedPd(3%) displaying the NO + H2 data as mole fractions in tables
catalyst was 40 g/m2 and the metal amount in the washcoat and figures. For deuterium-containing products, the total
was3wt%. TheBETsurfaceareaofthecatalystdetermined signal for each m/z value divided by the signal of a carrier
by nitrogen physisorption was 106 m2/g. The pore volume gas is shown in Figs. 6–8. The experimental procedure is
of the catalyst was 0.32 cm3/g and the mean pore diame- described in detail in the previous part of the present arti-
ter was 12 nm. The palladium particle size characterised by cle (11).
carbon monoxide chemisorption was about 12 nm. A de-
tailed description of the 1% Pd on the alumina-supported
catalyst can be found in the preceding part of the present
3. RESULTS AND DISCUSSION
work (11); the most pertinent details are given below. The
support material content of the 1% Pd catalyst was 41 g/m2.
The BET surface area of the Pd catalyst was 112 m2/g. The
pore volume of the catalyst was 0.33 cm3/g and the mean
pore diameter was 12 nm. The palladium particle size was
about 3.2 nm.
3.1. Temperature-Programmed Desorption of NO
NO (at m/z = 30), N2 (at m/z = 28), O2 (at m/z = 32),
and N2O (at m/z = 44) were the main desorption prod-
ucts from these catalysts, whereas no formation of NO2 (at
m/z = 46)wasdetected. Thisisconsistentwiththeresultsof
Huang et al. (13) and Valden et al. (14), who have conducted
NO-TPD experiments on the same type of catalyst surfaces.
It is, however, difficult to draw conclusions between peak
maxima and positions, due to the different pretreatment
procedures and heating rates they used.
2.2. NO-TPD Experiments
NO-TPD (temperature-programmed desorption) mea-
surements were carried out to obtain information on the
surface adsorption states of the Pd-based catalysts sup-
ported on γ -alumina. Adsorption of NO on 1% Pd and 3%
Pd catalysts was studied in order to determine the effect of
the amount of palladium on the adsorption and desorption.
Analogous experiments were carried out over the alumina
support, too, mainly consisting of γ -alumina, to evaluate
the effect of the support material on the surface adsorption
states. The volume of the catalyst monolith was 1.4 cm3. In
the pretreatment stage, catalysts were evacuated for 2 h and
then reduced under hydrogen flow for 10 min at 500◦C. The
catalysts were exposed to 5% NO/Ar at room temperature
for 10 min. The gas flows were regulated with mass flow con-
trollers (Bronkhorst). The NO-TPD measurements were
carried out in a quartz chamber in vacuum at 30–800◦C at a
linear heating rate of 30◦C/min. A split of a product gas flow
was taken through a capillary into a quadrupole mass spec-
trometer (Carlo Erba Instruments Q.T.M.D.). The experi-
mental setup for the NO-TPD measurements is presented
in detail in our earlier publication (12).
Figures 1a–1d show the comparative TPD curves for the
individual desorption products from the alumina support,
1% Pd/alumina and 3% Pd/alumina catalysts, following ad-
sorption of 5% NO/Ar at room temperature for 10 min.
The adsorption of NO on catalysts occurred mainly molec-
ularly, as indicated by the large amounts of NO desorbed
and small amounts of N2O and N2 formation as a result of
partial dissociation of NO during heating. Two prominent
NO-TPD peaks for the support were observed, at around
310 and 480◦C (Fig. 1a). The lower temperature state of NO
was dominant and had a higher adsorption capacity. For
the 1% Pd catalyst, three adsorption states were observed,
at temperatures around 250, 310, and 440◦C. The lowest
temperature NO state was caused by NO–Pd interaction,
because it was not observed in the case of the support mate-
rial. The middle state was observed as the shoulder in a TPD
spectrum consisting mainly of the interaction between NO
and support material. The highest temperature state was
due to NO support as well as NO–Pd interaction. It can be
concluded that the interaction between NO and alumina
support was significant, which has also been observed by
other researchers (14, 15). Thus, the support itself is ac-
tive and a strong adsorption of NO onto the support was
2.3. Experimental Procedure of
Step-Response Experiments
Reduction of NO with hydrogen was studied with tran- observed.
sient step-response experiments. A single step of both re-
In the literature, there has been a lot of discussion on
actants was performed under different concentration con- the number and the origin of NO-TPD peaks on alumina-
ditions (stoichiometric, excess of NO, and excess of H2). based catalyst surfaces. Huang et al. (13) observed three
Pretreatments with NO and H2 were carried out to investi- peaks, at 100, 410, and 530◦C after NO adsorption; in con-
gate their effect on the final concentration of products and trast, Kijlstra et al. (16) observed only two peaks, at 180
on transient kinetics. Tracing the hydrogen pathway by D2 and 375◦C. The differences in temperature maxima might
contributes to the understanding of the role of hydrogen in be due to different heating rates. Peak maxima are shifted
the reduction of NO. Hydrogen was replaced by deuterium to higher temperatures when heating rate is increased (15).

32
RAHKAMAA-TOLONEN ET AL.
FIG. 1. TPD spectra of (a) NO, (b) N2, (c) N2O, and (d) O2 for the alumina support, 1% Pd/alumina and 3% Pd/alumina, following adsorption of
5% NO at room temperature (25◦C) for 10 min. Heating rate: 30◦C/min.
The origin of the lower temperature adsorption state of NO with the Pd catalyst. On the support nitrogen was formed
is not well understood. The highest temperature NO-TPD only during the first 15 s after the switch. In addition to
peak was accompanied by oxygen desorption, and this cor- TPD, these step-response experiments also confirmed the
responds to the decomposition of nitrate and nitrite species N2O formation on the support. The formation of N2 and
(14, 16). This explanation is supported also by our experi- N2O was observed in small amounts in the same tempera-
ments where oxygen desorption occurred in the same tem- ture range where NO desorption occurred, with maxima at
perature range as the highest temperature NO adsorption around 250 and 440◦C over 1% Pd/alumina. We conclude
state observed, around 440◦C (Figs. 1a and 1d).
thatpalladiummetalparticlesaremostlyresponsibleforthe
N2, N2O, and O2 were formed over the alumina support, decomposition of NO, because the formation of N2O was
1% Pd/alumina and 3% Pd/alumina catalysts, as can be seen smallerandtheformationofN2 washigherinthecaseof1%
from Figs. 1b–1d. Especially, the formation of N2O on the and 3% Pd catalysts than in the case of support (Figs. 1b and
support was remarkable. This might be due to Zr, La, and 1c). The TPD results of NO from the alumina-supported Pd
Ce oxides present in the support. In addition to TPD, ad- catalysts in this study are consistent with those of Valden
sorption of NO was studied by a step-response experiment et al. (14) and Ciuparu et al. (17), who also demonstrated
from Ar to a 1% NO/Ar mixture at 155◦C with the Pd cata- that palladium particles have a central role in the decom-
lyst and the support. After the switch (400 s) mole fractions position of NO.
of formed N2O and N2 were 1.0 × 10−4 on the Pd cata-
AccordingtoFig. 1d, oxygendesorptionoccurredasasin-
lyst and the mole fraction of formed N2O was 4.0 × 10−5 gle sharp peak around 470◦C for the support. In the case of
on the support alone. The transient response of N2O was 1% Pd/alumina and 3% Pd/alumina catalysts, oxygen des-
higher than that of N2 in the beginning of the experiment orbed in two states, having also a higher temperature state

NO REDUCTION BY H2
33
FIG. 2. Ar → (1% NO + 1% H2)/Ar step-response experiment on
the Pd catalyst at 155◦C.
at around 750◦C. The origin of this state is not fully known,
but it is somehow related to the oxidation states of palla-
dium because it was not observed in the case of support.
Palladium oxide supported on alumina is found to decom-
pose to metallic palladium at around 700–800◦C, depending
on the oxygen partial pressure (18). It can also be observed
that the support was almost inactive in oxygen formation.
3.2. Reduction of NO with H2
From Figs. 2–7 it can be seen that NO reduction with H2
generated H2O and N2 as the main products and N2O and
NH3 as the by-products. The responses of the reduction of
NO + H2 with 1% of each reactant in the reacting mixture
are shown in Fig. 2. During the first 350 s after the step
change from Ar to the reaction mixture, the temperature
increased from 155 to 180◦C. A slight maximum in the nitro-
gen response was detected at the beginning of the NO+H2
step. Water and ammonia started to liberate once N2, N2O,
and NO had almost reached the steady state.
FIG. 3. Ar → 1% NO/Ar → (1% NO + 1% H2)/Ar step-response ex-
periment on the Pd catalyst at 155◦C: (a) Ar → 1% NO/Ar step;
(b) 1% NO/Ar → (1% NO + 1% H2)/Ar step.
gen, which reacts with the OH groups present in the ox-
ide catalyst, releasing a small amount of water, as can be
seen from Fig. 4. The isotopic exchange experiment of H2
and D2 on the support material provided the supporting
Figure 3 shows the responses of reduction of NO with
H2 after a pretreatment with NO. Figure 3a presents a step
response from Ar to a 1% NO/Ar mixture with the reduced
Pd catalyst at 155◦C. Minimal amounts of N2O and N2 were
formed. The transient response of N2O was higher than that
of N2 in the beginning of the experiment. After introduc-
ing H2 to the reactor, the response of N2O appeared first
(Fig. 3b). Nitrogen-containing species reached their steady
states faster than water.
Figure 4 shows the transient responses of the reduction
of NO with H2 with a pretreatment of hydrogen. Once
H2 entered the monolith, the steady state was reached al-
most immediately, with minor formation of water. Hydro-
gen molecules easily adsorb and dissociate on palladium,
due to the low activation energy for adsorption (19). The
FIG. 4. 1% H2/Ar → (1% NO + 1% H2)/Ar step-response experi-
surface of the catalyst was then covered by atomic hydro- ment on the Pd catalyst at 155◦C.

34
RAHKAMAA-TOLONEN ET AL.
evidence for water formation (11). During this experiment
NO showed the highest relative product distribution with
asmallamountofH2Owasformedalsoonthemodifiedalu- respecttoNatomstowardsN2, asrevealedbyTable1. Itwas
mina support. When NO was introduced into the reactor, slightly higher for the experiments without any pretreat-
a N2 response was immediately observed but it decreased ment and values were close for pretreatment experiments.
quickly. The second product formed was N2O, which indi- N2O was the second most abundant nitrogen component
cated the abundance of NO∗ and N∗ species on the surface. into which NO converts in all three cases. The mole frac-
The formation of water started again after the N2 and N2O tions of N2O at the steady state were practically equal for
responses had almost reached their steady states. The NH3 all cases. The selectivity of NO with respect to ammonia
response was rather slow.
was small in all three cases.
In the case of no catalyst pretreatment and H2 pretreat-
The relative product distribution with respect to H atoms
ment, a maximum in the response of N2 was detected when during the steady state is also presented in Table 1. Water
NO was introduced into the reactor. The more intensive was clearly the most abundant product in the three cases.
one was obtained after the H2 pretreatment. The N2 re- Preferentially H2O was formed from hydrogen atoms. The
sponse decreased quickly, possibly because of the decrease ammonia formation was rather minor, reaching 5% in all
in activity of the catalyst due to the oxygen, water, and cases.
hydroxyl accumulation on the surface (20). The nitrogen
The relative amount of N, NO, and H adsorbed on the sur-
response was the fastest, which indicated rapid dissociation face are the key factors determining the selectivity. There-
of NO on the metallic surface. The explanation might be fore, during the transient period of the experiments there
the availability of the reductant.
was a small variation in selectivity towards N2, N2O, and
The response of N2O differed considerably in the NO NH3. However, it can be concluded that transient periods
pretreatment experiment compared to the experiments of the experiments were fractionally different due to differ-
without and with H2 pretreatment during the transient pe- ent surface coverages in the beginning of the experiments,
riod. In the case of NO pretreatment, a maximum for the but in steady state catalysts reached similar states.
N2O response was detected as H2 was introduced into the
Once NO and H2 were introduced into the reactor in
monolith. In the other two cases, the N2O response reached the case of no catalyst pretreatment, all the active sites of
the steady state rapidly without showing any maximum. For the catalyst were free and started to be initially covered by
the H2 pretreatment experiment, a small decline was de- NO∗ and H2∗. At this point there were still many vacant sites,
tected in the middle of the NO + H2 step, then the steady which enhanced the dissociation of NO∗ into N∗ and O∗, as
state was quickly attained.
explained by Hirano et al. (21). High coverage of N∗ on
The NH3 responses in all three cases reached the steady the surface promoted the formation of diatomic nitrogen,
state at the same time. The shape of the responses and the which was detected to be the primary reaction product of
mole fraction of NH3 at steady state were similar. The re- NO. N2O was also produced due to the presence of NO∗
sponses of H2O did not differ significantly; they showed the and N∗:
similar transient responses, reaching the steady state almost
at the same time.
NO∗ + N∗ → N2O + 2∗.
[1]
The conversion of NO and the relative product distribu- Therefore, in the case of no catalyst pretreatment the prob-
tion with respect to N and H atoms at steady state in the ability of forming N2 was higher than in the case of NO
NO and H2 experiments (1% NO/Ar and 1% H2/Ar pre- pretreatment, where NO pretreatment has an oxidising ef-
treatment as well as without any pretreatment) are shown fect. ItisknownthatNOdissociationisfavouredonmetallic
in the Table 1. In these three cases, high conversions, which Pd sites. In the case of pretreatment by NO, N2O was the
were in practice equal, were obtained. Selectivity towards first product seen and there was a maximum in the N2O re-
N2, N2O, and NH3 at steady state did not vary significantly. sponse after introduction of the reaction mixture because
TABLE 1
Conversion of NO and Relative Product Distribution with Respect to N Atoms from NO and H Atoms from H2
during the Steady State (1% NO + 1% H2 in Ar) over the Alumina-Based Pd Catalyst
Relative product distribution
with respect to N atoms
Relative product distribution
with respect to H atoms
Experiment
NO conversion
N2
N2O
NH3
H2O
NH3
Without pretreatment
Pretreatment with NO
Pretreatment with H2
0.98
0.96
0.97
0.69
0.64
0.66
0.22
0.24
0.25
0.09
0.12
0.09
0.95
0.95
0.95
0.05
0.05
0.05

NO REDUCTION BY H2
35
NO was preadsorbed on the surface and the dissociation of These experiments implied also that the desorption of H2O
only one NO molecule is needed for the formation of N2O is a slow step. According to Hirano et al. (21) NH∗ was easily
(Fig. 3b). The changes in relative amounts of NO∗ and N∗ formed, but then it was only slowly further hydrogenated,
coverages on the catalyst surface explained the maximum which might also be one reason for the slow response of
in the response of N2O. NO adsorption as well as oxida- NH3. According to the study of the NO–H2 reaction on
tion of metallic clusters blocked active sites for hydrogen Pd (111), hydrogen was adsorbed dissociatively on this sur-
dissociation and thus enhanced the incomplete reduction face, penetrating into the subsurface at very low tempera-
to N2O. Gas-phase oxygen was not observed during the ture, already at −183◦C (23). It has been found that in the
NO pretreatment experiment. This could be explained by formation of water the reaction between O∗ and Hbulk is
the strong adsorption of oxygen atoms on the metal sur- preferred over the reaction of adsorbed oxygen with H∗
face, which was confirmed by the NO-TPD experiments adsorbed on the surface. Therefore, the formation rate of
(Fig. 1d).
H2O was dominated by the transport of atomic hydrogen
This fact explains the peaks in N2O and N2 responses between the bulk and the surface. Formation of NH3 might
detected at the beginning of the NO step, when still some follow an analogous pathway. In the current study, this be-
active sites of the catalyst are vacant and allow the disso- haviour is supposed to be the least likely explanation for
ciation of NO∗ to N∗ and O∗ (Fig. 3a). The N2 signal was the delay of responses of hydrogen-containing species.
slightly lower than the N2O signal in the beginning of the
Step-response experiments from Ar to 1% NO/Ar and
step reponse. Once NO enters the monolith, it is adsorbed (1% NO + 1% H2)/Ar mixtures at 155◦C were carried out
on the metal surface. NO oxidises the surface of the catalyst. on the support to clarify the role of the support material.
Previous studies have demonstrated that NO dissociation The consumed amount of NO was 0.184 mmol/g washcoat
does not proceed on surfaces fully covered with NO at low during the NO adsorption experiment. A minor amount
temperatures (18). It is well established now that the disso- of N2O was formed on the support. A very small amount
ciation of NO requires the existence of vacant sites in the of nitrogen was formed only during 15 s at the beginning of
neighbourhood of the adsorbed molecule. For the selectiv- the switch. The result indicated that the adsorption of NO
ity of NO to N2 or N2O, the availability of a vacant site and formation of N2O happened on the support material
adjacent to NO∗ and N∗ plays a crucial role. In this case, to some extent. During the (1% NO + 1% H2)/Ar experi-
nitrogen was formed, as NO started to cover the surface of ment, a small amount of H2O was formed on the modified
the catalyst, but there existed still vacant sites for the disso- alumina support. Minor amounts of N2O and NH3 were
ciation of NO. However, once the active sites were covered also formed. The N2 response was visible for 15 s at the
by NO∗ and O∗, the formation of N2O became higher and beginning of the experiment. This implied that only a small
it then decreased with the coverage of NO, indicating that amount of NO dissociated on the support. After the switch
Pd was not an effective catalyst for the self-decomposition from H2 to D2, very small amounts of HD and HDO were
of NO. Our NO-TPD result showed that oxygen did not formed. Previous H2/D2 experiments (11) indicated that H2
desorb before 380◦C. Therefore, in the absence of any re- and D2 dissociated and the exchange reaction with hydro-
ducing agent the active sites were blocked by oxygen at gen and deuterium happened on the support material. The
temperatures below 380◦C.
experiments also implied that OH groups from alumina can
In the absence of H2 only very little N2 and N2O were take part to some extent in the water formation. In addition,
formed, as can be seen from Fig. 3a. Therefore, hydrogen is adsorption experiments (11) on the support material have
needed to react with O∗, keeping the metal atoms available shown that large amounts of NH3 and H2O are adsorbed
for reactions. Arena et al. (22) stated that hydrogen quickly on the support.
reduces the nitrogen oxides species present over the noble
According to Hecker and Bell (24) and Burch and
metal, thus cleaning the surface of the catalyst. The cata- Watling (25) the fast enhancement of the rate of NO de-
lyst became covered by H∗, which reacted with O∗ to form composition by H2 is probably due to H-assisted NO de-
O–H∗ and further H2O∗. A small amount of water was also composition. NO decomposition occurs by the following
formedthroughareactionofspilloverHandOH(11). Once reaction:
the cleaning of the surface was completed, new sites were
accessible for NO to adsorb and dissociate, forming N2, for
NO∗ + H∗ → N∗ + OH∗.
[2]
which a prominent signal was detected. Dissociation of NO However, the most recent evaluation of single-crystal data
is also needed for the formation of N2O and NH3. concludes that both reactions can be adequately explained
The delay of H2O and NH3 responses in reaching the by an initial direct dissociation of NO followed by removal
steady state is most probably due to the consecutive for- of the fragments by adsorbed reductant (26). Thus, based
mation mechanism of water and ammonia as well as to the on the transient responses, the following mechanism for the
adsorption of water and ammonia on the support, which is abatement of NO over the Pd monolith is proposed. The
confirmed by the adsorption–desorption experiments (11). nitrogen-containing reaction products N2, NH3, and N2O

36
RAHKAMAA-TOLONEN ET AL.
water:
+H∗
−H∗
+H∗
−H∗
∗
∗
O∗
OH
H O .
[8]
ꢀ
ꢁ
ꢀ
ꢁ
2
The isotopic transient experiments (Figs. 5–7) confirmed
that ammonia and water were formed by stepwise hydro-
genation, as is demonsrated in the following section.
3.3. Reduction of NO with H2/D2
Figures 5–7 show the transient states of the reduction
of NO with hydrogen after an isotopic exchange with deu-
terium. Figure 5 shows the responses of the experiment
carried out using equal amounts of reactants 1% NO + 1%
H2/D2. The following two figures correspond to the results
of experiments performed with different amounts of reac-
tants, in which 2% NO + 1% H2/D2 (Fig. 6) and 1% NO +
2% H2/D2 (Fig. 7) have been used.
The switch from H2 to D2 is presented in the graphs
shown in Figs. 5a, 6a, and 7a and the switch from D2 to H2
in the graphs shown in Figs. 5b, 6b, and 7b for the three dif-
ferent experiments carried out. During NO + H2 step, the
formation of water, nitrogen, ammonia, and nitrous oxide
were detected, as discussed earlier. H2O was the product
FIG. 5. Ar → (1% NO + 1% H2)/Ar → (1% NO + 1% D2)/Ar →
(1% NO + 1% H2)/Ar → Ar step-response experiment on the Pd cata-
lyst at 155◦C: (a) (1% NO + 1% H2)/Ar → (1% NO + 1% D2)/Ar step;
(b) (1% NO + 1% D2)/Ar → (1% NO + 1% H2)/Ar step.
are formed via dissociation of NO:
∗
ꢀ
ꢁ
NO + ∗
NO ,
[3]
[4]
∗
∗
∗
ꢀ
ꢁ
NO + ∗
N + O .
N2 is formed by combination of two adatoms of nitrogen:
2N∗
N + 2 .
2
[5]
[6]
∗
ꢀ
ꢁ
At lower temperatures N2 may also be formed via
NO∗ + N∗ → N2 + O∗ + ∗.
N2O is formed by combination of adsorbed NO and nitro-
gen (Eq. [1]).
From Fig. 3b it can be seen that the abundance of NO∗
promotes reaction [1] at the expense of reactions [5] and
[6]. In contrast H∗ contributes to reaction [5] (Fig. 4).
The most probable mechanism for ammonia formation
is stepwise hydrogenation of adsorbed nitrogen atoms:
+H∗
−H∗
+H∗
−H∗
+H∗
−H∗
∗
∗
∗
3
N∗
NH
NH
NH .
[7]
ꢀ
ꢁ
ꢀ
ꢁ
ꢀ
ꢁ
2
FIG. 6. Ar → (2% NO + 1% H2)/Ar → (2% NO + 1% D2)/Ar →
(2% NO + 1% H2)/Ar → Ar step-response experiment on the Pd cata-
lyst at 155◦C: (a) (2% NO + 1% H2)/Ar → (2% NO + 1% D2)/Ar step;
(b) (2% NO + 1% D2)/Ar → (2% NO + 1% H2)/Ar step.
The surface oxygen reacts with dissociatively adsorbed hy-
drogen, forming surface hydroxyls, which are converted to

NO REDUCTION BY H2
37
molecules in their neighbourhood to react with and to form
a higher amount of N2O. Nitrous oxide, on the other hand,
was not formed when an excess of hydrogen was present
(Fig. 7). As can be seen in the figure, nitric oxide response
was not clearly noticeable either, which indicated that the
total dissociation of NO on the surface prevents the reac-
tion of N∗ with chemisorbed NO∗.
In addition to N2, N2O, and NH3, the deuterium-
containing reaction products HDO, D2O, NH2D, NHD2,
and ND3 were observed. Formation of HD was not ob-
served because D∗ reacts much faster with OH∗ or NH2∗
than with H∗, as discussed in the preceding part of the
present work (11). Formation of products with m/z = 20,
i.e., D2O and ND3, was detected to be the highest in all
cases. Argon, which was used as a carrier gas, has a frag-
ment of m/z = 20. The response of m/z = 20 is mainly at-
tributed to deuterated water (D2O), based on the fact that
this compound behaves the same as H2O. Therefore, its
signal was similar to the response of water obtained in the
NO + H2 steps. Following the same behaviour as H2O and
NH3, the amount of produced D2O and ND3 was highest in
excess hydrogen (Fig. 7). Assuming that the ND3 response
followed more or less the same path as that of ammonia, it
was possible then to relate part of the m/z = 20 response to
ND3 and thus recognise the behaviour of H2O.
A peak in the response of m/z = 17 (HDO and NHD2)
was observed in the three experiments when the isotopic
step change was performed (H2 → D2 and D2 → H2). These
signals belonged mainly to HDO. The rest of this signal be-
longed to NHD2. During the switch from H2 to D2, HDO
response declined at the same time the D2O response in-
creased. This indicates that the formation of D2O is via iso-
topic exchange of HDO∗ and D∗, as was proposed in part I
(11). ThemaximuminHDOandNHD2 responseduringthe
isotopic step change from deuterium to hydrogen (Figs. 5b,
6b, and 7b) can be due to the reaction between the spillover
deuteriumonthesupportandtheOH∗ aswellasOHgroups
in the support. A small amount of it can be assigned to the
NHD2 formed via reaction between spill-over deuterium
and NH∗. The responses also showed that ND3 is formed
via stepwise deuteration. At the same time, the isotopic ex-
change in ammonia can take place, as was demonstrated
in part I (11). It is difficult to estimate the relative impor-
tance of these two pathways due to the overlapping mass
numbers.
FIG. 7. Ar → (1% NO + 2% H2)/Ar → (1% NO + 2% D2)/Ar →
(1% NO + 2% H2)/Ar → Ar step responses on the Pd catalyst at 155◦C:
(a) (1% NO + 2% H2)/Ar → (1% NO + 2% D2)/Ar step; (b) (1% NO +
2% D2)/Ar → (1% NO + 2% H2)/Ar step.
mainly formed, having the highest value in steady state in
the presence of excess H2 (Fig. 7).
NH3 formation was also detected in the three experi-
ments, having the highest value in steady state when hy-
drogen was the excess reactant (Fig. 7), which corresponds
with our earlier results (20). The signal of m/z = 16 (NH2)
was used to follow the formation of NH3. This preference
of forming NH3 and H2O when there was excess hydrogen
deals with the availability of hydrogen atoms. The excess of
hydrogen atoms allowed formation of hydrogen-containing
species instead of N2 or N2O, even when the responses of
these two last compounds were faster. In the excess of NO
(Fig. 6) the amount of ammonia was much smaller than with
stoichiometric and H2-rich gas mixtures. This was caused by
the lack of H∗ atoms to react with N∗, which instead had
many other N∗ and NO∗ in its surroundings to produce N2
and N2O.
Based on the transient results, the mechanism for NO
reduction with hydrogen and deuterium is proposed. For-
mation of N2, N2O, NH3, and H2O follows the mechanisms
already mentioned above (Eqs. [1]–[8]).
Of the nitrogen-containing reaction products N2 has the
highest amount, with stoichiometric and H2-rich gas mix-
tures. The amount of N2O formed was higher than N2 in ex-
cess NO (Fig. 6). This may be caused by the inhibition effect
of adsorbed oxygen. Once the molecules of NO adsorbed
on the surface they dissociated to form N∗ and O∗. Since
Deuterium adsorbes dissociatively;
∗
D2 + 2∗
2D .
[9]
ꢀ
ꢁ
there was an excess of NO and probably few vacant sites Formation of deuterated water can follow two different
on which NO∗ could dissociate, N∗ atoms had more NO∗ pathways. The first of them corresponds to the reaction of

38
RAHKAMAA-TOLONEN ET AL.
dissociated water with deuterium:
under H2-rich or stoichiometric conditions. Ammonia and
water were formed by stepwise hydrogenation of adsorbed
nitrogen and oxygen atoms, respectively, which was con-
firmed by isotopic transient experiments.
∗
∗
∗
ꢀ
ꢁ
H O + ∗
OH + H ,
[10]
[11]
[12]
2
OH∗ + D∗
HDO + ∗,
D O + H .
∗
ꢀ
ꢁ
HDO∗ + D∗
∗
∗
ꢀ
ꢁ
2
ACKNOWLEDGMENTS
❛
The second route is the stepwise one starting from atomi-
cally adsorbed oxygen:
This work is part of the activities at the Abo Akademi Process Chemistry
Group within the Finnish Centre of Excellence Programme (2000–2005)
of the Academy of Finland. Financial support from the Graduate School in
Chemical Engineering (GSCE) and the Academy of Finland is gratefully
acknowledged.
+D∗
+D∗
∗
∗
O∗
OD
D O .
[13]
ꢀ
ꢁ
ꢀ
ꢁ
2
−D∗
−D∗
Formation of ND3 is also thought to follow two routes. In
one hand, it can follow the H–D isotopic exchange in am-
monia when there is some NH∗3 or NH∗2 available (as shown
by separate experiments in part I):
REFERENCES
1. Glick, H. S., Klien, J. J., and Squire, W., Chem. Phys. 27, 850 (1957).
2. Iwamoto, M., and Hamada, H., Catal. Today 10, 57 (1991).
3. Iwamoto, M., and Yahiro, H., Catal. Today 22, 5 (1994).
4. Armor, J. N., Appl. Catal. B 1, 221 (1992).
5. Li, Y., and Armor, J. N., J. Catal. 150, 376 (1994).
6. Iwamoto, M., and Takeda, H., Catal. Today 27, 71 (1996).
7. Burch, R., Millington, P. J., and Walker, A. P., Appl. Catal. B 4, 64
(1994).
∗
∗
∗
ꢀ
ꢁ
NH + ∗
NH + H ,
[14]
[15]
[16]
[17]
[18]
[19]
[20]
3
2
NH∗2 + D∗
NH2D∗
NH D + ∗,
∗
ꢀ
ꢁ
2
ꢀ
ꢁ
NH D + ∗,
2
8. Lox, E. S. J., and Engler, B. H., in “Handbook of Heterogeneous Cata-
lysis” (G. Ertl, H. Kno¨zinger, and J. Weitkamp, Eds.), Vol. 4, p. 1559.
VCH, Weinheim, 1997.
∗
∗
NH2D∗ + D∗
NHD∗2
NHD + H ,
ꢀ
ꢁ
2
ꢀ
ꢁ
9. Shelef, M., and Graham, G. W., Catal. Rev.–Sci. Eng. 36, 433 (1994).
10. Taylor, K. C., Catal. Rev.–Sci. Eng. 35, 457 (1993).
11. Rahkamaa-Tolonen, K., Salmi, T., Murzin, D. Yu., Barreto-Dillon, L.,
Karhu, H., Keiski, R. L., and Va¨yrynen, J., J. Catal., in press.
12. Rahkamaa, K., Pulkkinen, U., Wa¨rna˚, J., Zhou, Y., Salmi, T., and
Keiski, R. L., in “Proceedings of the 14th International Congress
of Chemical and Process Engineering, Praha, August 2000,” Czech
Society of Chemical Engineering, Czech Republic, 2000.
13. Huang, S.-J., Walters, A. B., and Vannice, M. A., Appl. Catal. B 26, 101
(2000).
14. Valden, M., Keiski, R. L., Xiang, N., Pere, J., Aaltonen, J., Pessa, M.,
Maunula, T., Savima¨ki, A., Lahti, A., and Ha¨rko¨nen, M., J. Catal. 161,
614 (1996).
15. Falconer, J. L., and Schwarz, J. A., Catal. Rev.–Sci. Eng. 25, 141
(1983).
NHD + ∗,
2
NHD∗2 + D∗
ND∗3
ND + H ,
∗
∗
ꢀ
ꢁ
3
ꢀ
ꢁ
ND + ∗.
3
Another probable route is stepwise deuteration of ad-
sorbed nitrogen:
+D∗
−D∗
+D∗
−D∗
+D∗
−D∗
∗
∗
∗
3
N∗
ND
ND
ND .
[21]
ꢀ
ꢁ
ꢀ
ꢁ
ꢀ
ꢁ
2
Further kinetic studies are needed to quantify the relative
importance of these different routes.
16. Kijlstra, W. S., Brands, D. S., Poels, E. K., and Bliek, A., J. Catal. 171,
208 (1997).
17. Ciuparu, D., Bensalem, A., and Pfefferle, L., Appl. Catal. B 26, 241
(2000).
4. CONCLUSIONS
18. Farrauto, R. J., Hobson, M. C., Kennelly, T., and Waterman, E. M.,
Appl. Catal. A 81, 227 (1992).
19. Miyazaki, E., J. Catal. 65, 84 (1980).
20. Rahkamaa, K., Salmi, T., Keiski, R. L., Wa¨rna˚, J., and Zhou, Y., Chem.
Eng. Sci. 56, 1395 (2001).
21. Hirano, H., Yamada, T., Tanaka, K. I., Sierra, J., Cobden, P., and
Nieuwenhuys, B. E., Surf. Sci. 262, 97 (1992).
22. Arena, G. E., Bianchini, A., Centi, G., Vazzana, F., and Vitali, P., Stud.
Surf. Sci. Catal. 122, 109 (1999).
23. De Wolf, C. A., and Nieuwenhuys, B. E., Surf. Sci. 469, 196 (2000).
24. Hecker, W. C., and Bell, A. T., J. Catal. 92, 247 (1985).
25. Burch, R., and Watling, T. C., Catal. Lett. 37, 51 (1996).
26. Nieuwenhuys, B. E., Adv. Catal. 44, 249 (2000).
The reduction of NO with H2 and D2 was studied with
transient experiments on the Pd monolith. The experiments
showedthecomplexityofthedeuteratedproductformation
in the deuteration step and provided a possible means of
discriminating between different reaction mechanisms. Ni-
trogen, nitrous oxide, ammonia, and water were detected
as reaction products in NO reduction by hydrogen. The dis-
sociation of NO on the catalyst surface is the crucial step,
dominating the overall reaction behaviour. Ammonia for-
mation was favoured in excess H2. Under NO-rich con-
ditions the formation of N2O was essentially higher than