Simultaneous removal of NO and N2O over Pd-ZSM-5 catalysts and FT-IR observations of their decomposition routes to N2

Article abstract of DOI:10.1246/bcsj.74.1499

The simultaneous removal of NO and N₂O with propylene was studied over Pd-H-ZSM-5 catalysts and the results were compared with reports of the simultaneous removal of NO and N₂O in the presence of O₂ over Co-ZSM-5 and Fe-MFI. The N₂O conversion was suppressed in the presence of NO. The simultaneous removal rates were essentially the same as the NO conversion rates over Co-ZSM-5 and Fe-MFI. On the contrary, the simultaneous removal in the presence of O₂ was accelerated over Pd-H-ZSM-5 and conventional Pd/SiO₂ compared to the sum of the corresponding individual removal rates of N₂O and NO. An intense FTIR peak of NO^±0 ligated to the Pd cation inside the ZSM-5 cage was seen at 1868/cm in the promoted simultaneous removal reactions. Under the conditions for the individual removal of N₂O or NO, and the unpromoted simultaneous removal of NO + N₂O, only a weak peak at 1842/cm was observed. The oxygen atom dissociating from N₂O should be the key to promoted catalysis in the co-existence of NO and N₂O to maintain the Pd cation site clear. The oxygen controlled the concentration of surface carbonaceous species to effectively decompose NO, and so as not to inhibit NO activation.

Full text of DOI:10.1246/bcsj.74.1499

c. Jpn.
© 2001 The Chemical Society of Japan
Bull. Chem. Soc. Jpn., 74, 1499–1505 (2001) 1499
e Chemical
ciety of Ja-
n
e Chemical
ciety of Ja-
n
M
Simultaneous Removal of NO and N2O over Pd-ZSM-5 Catalysts and
FT-IR Observations of their Decomposition Routes to N2
Simultaneous Removal of NO + N2O
Y. Izumi et al.
Yasuo Izumi,* Tetsuro Kizaki, and Ken-ichi Aika
01
99
05
SJA8
09-2673
409
Department of Environmental Chemistry and Engineering, Interdisciplinary Graduate School of Science and
Engineering, Tokyo Institute of Technology, 4259 Nagatsuta, Midori-ku, Yokohama 226-8502
(Received December 1, 2000)
Simultaneous removal of nitrogen monoxide and dinitrogen oxide with propene was investigated over Pd-H-ZSM-
5 catalysts. Results were compared with reports of simultaneous removal of NO and N2O in the presence of O2 over Co-
ZSM-5 and Fe-MFI. The N2O conversion was suppressed in the presence of NO, and the simultaneous removal rates
were essentially the same as the NO conversion rates over Co-ZSM-5 and Fe-MFI. In contrast, the simultaneous removal
in the presence of O2 was found to be accelerated over Pd-H-ZSM-5 and conventional Pd/SiO2 compared to the sum of
the corresponding individual removal rates of NO and N2O. An intense FT-IR peak of NO 0 ligated to the Pd cation in-
side the ZSM-5 cage was observed at 1868 cm−1 in the promoted simultaneous removal reactions. Under the conditions
for the individual removal of NO or N2O and the unpromoted simultaneous removal of NO + N2O, only a weak peak at
1842 cm−1 was observed. The oxygen atom that dissociates from N2O should be the key to promoted catalysis in the co-
existence of NO and N2O in order to maintain the Pd cation site clear. The oxygen controls the concentration of surface
carbonaceous species to effectively decompose NO, and so as not to inhibit the NO activation.
00
01
.2
The catalytic removal of nitrogen oxides has been intensive-
ly investigated in relation to environmental problems.1 Elimi-
nating nitrogen monoxide usually entails the formation of dini-
trogen oxide as a by-product, which also affects the environ-
ment. In this context, simultaneous catalytic removal of NO
and N2O is a simplified and improved process for environmen-
tal catalysis. Simultaneous removal of NO and N2O was re-
ported on Co-ZSM-52 and on Fe-MFI catalysts.3 Atmospheric
oxygen and/or water often affects the rates of NO and/or N2O
catalytic removal.2,3 It is important to investigate catalysts ac-
tive in the presence of oxygen and/or water vapor. They are
present in industrial situations where there is a need for NO
and/or N2O removal.
From a scientific point of view, the reaction mechanism of
NO catalytic removal with reductants (e.g., hydrocarbons or
ammonia) is rather complicated. This is because the process
has the potential to form various intermediates and products;
N2, N2O, NO2, CO and CO2 are all possible products when a
hydrocarbon is utilized as a reductant. In contrast, if the simul-
taneous removal process allows us to selectively form N2, the
reaction mechanism should be relatively simple.
N2O, whereas the N2O conversions were inhibited with the
presence of NO.
In this paper, the simultaneous removal of NO and N2O over
Pd-H-ZSM-5 catalysts is reported and compared with results
for Pd-Na-ZSM-5 and conventional Pd/SiO2 catalysts. Palladi-
um is one of the components of the three-way catalyst (Rh, Pt,
Pd) for automobile exhaust gas. Because palladium generally
activates hydrocarbons,4–7 propene was chosen as the reductant
for NxO (x = 1, 2) rather than ammonia.8 ZSM-5 has been
widely investigated as an NO reduction catalyst by using hy-
drocarbons, especially as the metal-exchanged form.2,4,7,9–11
The promoted simultaneous reactions rather than the sum of
the individual removal rates of NO or N2O are reported for Pd-
H-ZSM-5 and Pd/SiO2. The simplified reaction mechanisms
(the routes from NO and N2O to N2) were monitored by in situ
FT-IR (Fourier-transformed infrared) spectroscopy. The key
features that affect the promotion of NO removal in the pres-
ence of N2O are proposed.
Experimental
Catalyst Preparation. The Pd-H-ZSM-5 catalyst was pre-
pared by continuously mixing 3.0 g of H-ZSM-5 and 0.10 g of
palladium(Ⅱ) nitrate in distilled water for 24 h. The mixture was
filtrated and washed seven times with distilled water. The powder
that was obtained was dried at 393 K for 12 h and heated in air at
673 K for 4 h. ICP (inductively coupled plasma) results demon-
strated that all the Pd ions were incorporated in/on the zeolite (Pd
1.5 wt%). H-ZSM-5 was prepared by the ion exchange of Na-
ZSM-5 (SiO2/Al2O3 = 23.3, primary crystalline size 0.1–0.5 µm,
Tosoh) with aqueous ammonium nitrate at 290 K for 24 h. ICP
measurements revealed that 98% of the sodium sites were substi-
tuted with protons. The pore area of ZSM-5 is 5.3 by 5.6 Å or 5.1
The Co-ZSM-5 catalyst is effective for both NO and N2O
removal. Simultaneous removal was tested on Co-ZSM-5 with
methane in the presence of O2 at 673–773 K, and the rates
were essentially the same as those for the removal of NO
alone.2 In the simultaneous removal on Fe-MFI with propane
in the presence of O2 at 523 − 773 K, the NO conversion rates
were similar to those for the individual conversion of NO,
whereas the N2O conversion rates were significantly sup-
pressed when compared to the individual conversion of N2O.3
A common feature of these two metal-exchanged zeolites is
that the NO conversions were not affected by the presence of

1500 Bull. Chem. Soc. Jpn., 74, No. 8 (2001)
Simultaneous Removal of NO + N2O
by 5.5 Å, and the pore volume is 0.29 cm3 g−1 based on the cyclo-
hexane adsorption measurements. A Pd-Na-ZSM-5 catalyst was
prepared from Na-ZSM-5 (Pd 1.6 wt% in dry sample) using a sim-
ilar procedure. The atomic ratio of incorporated palladium atoms
to Na+ ions in Na-ZSM-5 before mixing with Pd2+ was 11%.
A conventional Pd/SiO2 catalyst was prepared by the incipient
impregnation method12,13 by using silica (Davisson No. 952) and
palladium(Ⅱ) nitrate with distilled water. The powder that was ob-
tained was dried at 393 K for 12 h and subsequently heated in air
at 673 K for 4 h. The Pd loading was 0.5 wt%.
Kinetic Measurements. The catalytic removals of NO and/
or N2O were performed in a closed circulating system (117 cm3)
over either 0.12–0.16 g of Pd-H-ZSM-5 and Pd-Na-ZSM-5 or
conventional Pd/SiO2 catalysts. The flow rate of gas was 10–30
cm3 min−1. A closed circulating system connected to a vacuum
system was chosen rather than a conventional flow system in order
to correlate the reactivity to infrared spectroscopic data obtained
on a basis of vacuum cell + catalyst disk + reaction gas. Before
catalysis, the catalysts were treated in oxygen (27 kPa) for 1h, in
hydrogen (27 kPa) for 1 h, and finally under vacuum for 0.5 h at
653 K. The reaction gas pressures were 0.56, 0.56, 8.1, and 0.56
kPa for NO, N2O, O2, and propene, respectively. The remainder of
the gas necessary to obtain a pressure of 101.3 kPa was made up
with helium (balance). The combinations of reaction gases used
were as follows: NO; NO + O2; NO + C3H6; NO + O2 + C3H6
(individual removal of NO); N2O; N2O + O2; N2O + C3H6; N2O
+ O2 + C3H6 (individual removal of N2O); NO + N2O; NO +
N2O + O2; NO + N2O + C3H6; and NO + N2O + O2 + C3H6 (si-
multaneous removal). Most of the kinetic measurements were
performed with impurity moisture levels less than 0.01 kPa. A
control experiment was done in the presence of 0.56 kPa water gas
for Pd-H-ZSM-5 and Pd/SiO2.
Molecular Sieve 5A (N2, O2, and CO) and Unibeads C (NO,
N2O, and CO2) columns were utilized for GC analysis with a de-
tector system of TCD (Shimadzu GC-8A). The reaction rates be-
came constant within 1 h for all the reactions at 548 K. The reac-
tion temperature was fixed, based on preliminary reactions: the
formation rates of N2 and O2 reached the steady state and were
maintained for more than 24 h at 548 K. The relatively low tem-
perature was the compromise between high enough conversion
and better quality of FT-IR data. Increasing temperature caused
an increase in the conversion, but did decrease the S/N ratio and
peak intensity of FT-IR spectra due to the larger background and
desorption of surface species at > 600 K. Although a steady state
can be realized only for zero-order reactions in a closed circulat-
ing system, the observed constant rates are nearly steady-state
rates because catalysis occurred on Pd-H-ZSM-5 at relatively low
temperature (548 K) and relatively low conversion. The (pseudo)
steady-state rates were used in this work. The gas decrease in-
duced by sampling was taken into account in estimating the reac-
tion rate.
Fig. 1. In situ infrared sample cell.
ic measurements. The sample was cooled down from 653 K (pre-
treatment) to 548 K (reaction temperature), and the reaction gas
was introduced to the sample. The in situ spectra were measured
at 548 K. After the introduction of reaction gas, the infrared peak
intensity first increased and then saturated within 1–3 h. The
spectra for the catalyst sample before the introduction of reaction
gas were subtracted from the in situ spectra. Equivalent N2 forma-
tion rates were obtained for NO/N2O reactions with powder in a
U-shaped tube in the closed circulating system as compared to
pressed disk catalysts in an infrared cell connected to the closed
circulating system. This fact suggests that bulk mass transfer ef-
fects to obtain the catalyst active sites were not important on the
overall limiting kinetic rate.
Results
Kinetics. Twelve kinds of NxO (x = 1, 2) removal reac-
tions listed in the Experimental section were performed over
Pd-H-ZSM-5 and conventional Pd/SiO2 catalysts at 548 K.
The steady-state rates are summarized in Table 1. For all the
reactions (individual or simultaneous reactions and in the pres-
ence or absence of O2 and propene), the rates were faster on
Pd-H-ZSM-5 than on conventional Pd/SiO2, even taking into
account the difference of Pd loading of each catalyst (1.5 and
0.5 wt%, respectively). Namely, the rates on Pd-H-ZSM-5
were more than three times faster than those observed on Pd/
SiO2 in the corresponding reaction. H-ZSM-5 (Table 1) and
SiO2 were entirely inactive or showed smaller reactivity for
NO removal at 548 K. NO reduction with propane or propene
over H-ZSM-5 is reported at 673 K.5 This result demon-
strates that the Pd atom is the pertinent component of the ac-
tive site for NxO removal in these reaction conditions. The ef-
ficiency of Pd-H-ZSM-5 was reported for NO reduction with
methane.4,7,20 The acidity of zeolite was suggested to be an im-
portant factor in controlling the reaction rate.
FT-IR Measurements. The FT-IR spectra were measured by
a Jasco, Valor Ⅲ (single beam configuration, energy resolution 1
cm−1, MCD detector), in transmission mode with an in situ quartz
glass cell (Fig. 1). The in situ cell, capped by sodium chloride
windows,14–19 was combined with a closed circulating system.
Eighty milligrams of catalysts were pressed into a disk (φ = 20
mm). The temperature and ambient gas pressure for the sample
disk can be controlled in the in situ cell within the limits of 173–
823 K and 0–101.3 kPa, respectively. Gas phase spectra were
measured using the in situ cell without a catalyst sample.
For the reactions of NxO and NxO + O2 (Table 1-i and ii),
the rates of simultaneous reactions were almost the same as
those for the individual reaction rates of N2O. In the absence
The sample disk was treated in the same procedure as for kinet-

Y. Izumi et al.
Bull. Chem. Soc. Jpn., 74, No. 8 (2001) 1501
Table 1. Steady-State N2 Formation Rates (at 1 h Passed from the Beginning of Reaction; 10−6 mol h−1 g-cat−1) from NO
and/or N2O over Pd-H-ZSM-5 and Supported Pd/SiO2 Catalysts at 548 K
Reaction Gas
(i)
NxO
0.7
5.1
5.0
—
0.7
2.2
0
1.6
0.2
0.5
0.6
—
(ii)
NxO + O2
0.4
(iii)
NxO + C3H6
(iv)
NxO + O2 + C3H6
5.6( 0.1)
21( 1.2)
34( 0.9)a)
28( 0.9)
4.2
Catalyst
Pd-H-ZSM-5
NxO Type
NO
N2O
4.9
23
28
—
2.7
—
0
14
5.3
5.0
—
—
—
—
—
—
NO + N2O
NO + N2O + H2Ob)
Pd-Na-ZSM-5
H-ZSM-5
Pd/SiO2
NO
N2O
NO
N2O
NO
N2O
—
0
12
0.5( 0.1)
2.5( 0.1)
4.0( 0.1)
3.1( 0.3)
0.4( 0.05)
1.7( 0.1)
3.7( 0.2)
—
—
—
—
NO + N2O
NO + N2O + H2Oa)
The pressures of NO, N2O, O2, and propene were 0.56, 0.56, 8.1, and 0.56 kPa. The balance of the pressure to 101.3 kPa
was made up with helium.
a) Corresponding to 2.7% conversion (2[N2]/([NO] + 2[N2O]). b) 0.56 kPa of water was also present.
Fig. 2. The changes of concentration of reactants and products on the Pd-H-ZSM-5 catalyst for individual removal in NO + O2
+ propene (a) and for simultaneous removal in NO + N2O + O2 + propene (b). Reaction conditions are in footnote of Table 1.
of propene at 548 K, the reaction of N2O to N2 was so fast that
the reaction of NO was negligible or the surface reaction site
of NO was blocked by the presence of N2O.
for NO + N2O. The rate data deviations were estimated as the
standard deviations when several reaction runs were performed
in the same reaction conditions. Hence, in the presence of pro-
pene at 548 K, removal reactions of both NO and N2O pro-
ceeded. The reaction of NO was not suppressed by the coex-
istence of N2O for these Pd catalysts.
In the presence of O2 and propene (Table 1-iv), reactions be-
came rather faster than in the case in propene (Table 1-iii), ex-
cept for the case of N2O removal over Pd-H-ZSM-5. The reac-
tion enhancement in the presence of O2 was reported for the
NO reduction by methane/propane over the Co-ZSM-5 cata-
lysts.2 The rates of simultaneous reactions were larger than the
sum of the individual reactions for both Pd-H-ZSM-5 and Pd/
SiO2 catalysts (promotion ratio 28( 9) and 33( 12)%, re-
spectively). The promotions of simultaneous removal contrast
to the results of simultaneous removal on Co-ZSM-5 and Fe-
MFI with hydrocarbon in the presence of O2. The simulta-
neous removal rates were essentially the same as those ob-
For the reaction of NxO + propene (Table 1-iii), the reac-
tions were dramatically promoted. All the reactions were pro-
moted by the addition of propene, demonstrating that the N2O
reduction was also promoted by propene. The promotion of
NO reduction by propene is well known. The major reaction
route of propene should be the interaction with N2O, based on
the fact that the reaction in N2O + propene was about five
times faster than the reaction in NO + propene over both Pd-
H-ZSM-5 and Pd/SiO2 catalysts (Table 1). The rates of simul-
taneous reactions were the sum of the independent reactions of
NO or N2O for Pd-H-ZSM-5 and larger than the sum of the in-
dependent reactions for the conventional Pd/SiO2 catalyst
(76( 22)% enhancement). The error of enhancement ratio
was based on the data deviations (Table 1) of individual re-
moval rates for NO or N2O and of simultaneous removal rate

1502 Bull. Chem. Soc. Jpn., 74, No. 8 (2001)
Simultaneous Removal of NO + N2O
similar to the case of Fig. 2a. The CO formation rates reached
a maximum in the first 30 min and became negligible in 1–2 h
in Figs. 2a and b. The reason may be that the oxygen atom ad-
sorption on the Pd surface gradually increased and became
constant within 1–2 h of the reaction being initiated, which is
consistent with the decrease of N2O (Fig. 2b) within first 30
min.
FT-IR Observations. The in situ FT-IR measurements
were performed for Pd-H-ZSM-5 in NO, NO + propene, NO
+ O2 + propene, NO + N2O + propene, and NO + N2O + O2
+ propene at 548 K. Spectra were measured at this tempera-
ture. Peaks were observed in the region of 2000–1700 cm−1
(Fig. 3). The wavenumbers of the peaks are summarized in Ta-
ble 2, along with the reported values of NO adsorption.
An intense absorption peak was observed at 1880 cm−1 in
NO and at 1868 cm−1 in NO + N2O + O2 + propene (Figs. 3a
and e). A resolved shoulder peak also exists in NO at 1836
cm−1, but the region was not well resolved in NO + N2O + O2
+ propene. The intense peak in NO + N2O + O2 + propene
was broader than that in NO, and was not symmetric, suggest-
ing a weak feature on the lower-energy side.
The intense peak was not detected in NO + propene, in NO
+ O2 + propene, or in NO + N2O + propene (Figs. 3b– d). A
medium-intensity peak was observed at 1842 cm−1 in NO +
propene. The peak intensity became weaker or the peak disap-
peared by the addition of O2 or N2O. These peaks were not
symmetric, suggesting an additional weak feature around 1870
cm−1.
An intense, broad peak was observed at 2269–2251 cm−1 in
NO + O2 + propene and in NO + propene at 548 K for Pd-
Na-ZSM-5 (not shown). Under these conditions, no peak was
detected in this region for Pd-H-ZSM-5.
Fig. 3. In situ FT-IR spectra for Pd-H-ZSM-5 for 180 min at
548 K in various reaction conditions.
Two very broad peaks were also observed at 1730 and 1650
cm−1 in NO at 548 K for Pd-H-ZSM-5. The intensity was neg-
ligible, even when compared toweak peaks at1842–1836cm−1
(Fig. 3). The two broad peaks were also observed in NO at
548 K for conventional Pd/SiO2 catalyst. The FT-IR spectrum
for conventional Pd/SiO2 was essentially the same as those in
Refs. 12 and 13.
served for the individual removal rates of NO on Co-ZSM-5
and Fe-MFI.2,3
The effects of 0.56 kPa of water were tested for Pd-H-ZSM-
5 and Pd/SiO2 catalysts in NO + N2O + O2 + propene.
20% of activity loss was observed in the case of both cata-
lysts (Table 1). The N2 formation reactions over Pd-Na-ZSM-
5 were always slower than those over Pd-H-ZSM-5 (Table 1).
The time courses of gas conversion on the basis of N and C
atoms during the individual removal of NO and simultaneous
removal in the presence of O2 and propene on Pd-H-ZSM-5 are
illustrated in Figs. 2a and b, respectively. As the concentration
(reactants and products) was based on N or C atom, the con-
centration of N2O (0.56 kPa) was twice as large as that of NO
(0.56 kPa) at t (reaction time) = 0 in Fig. 2b. Figs. 2a and b
show the time courses during 300 and 120 min, respectively.
In Figure 2a (individual NO removal), 130% of CO2 was
formed compared to the degree of N2 formation (C and N atom
basis), but CO was produced in negligible quantities. Based
on the detected products ratio at the steady state (after 60 min
in Fig. 2), the reaction stoichiometry is simplified as 2NO +
C3H6 + 7/2O2 → N2 + 3CO2 + 3H2O (the CO2: N of N2 ratio
is 150%). The hydrocarbon combustion C3H6 + 9/2O2 →
3CO2 + 3H2O seems not to dominate the process. In Figure
2b (promoted simultaneous removal), the NO decrease and the
N2 increase were by about six times faster than in Fig. 2a. The
N2O decreased in the first 30 min, and remained (apparently)
constant later. The formation ratios of CO and CO2 to N2 were
Discussion
FT-IR Peak Assignments. The infrared absorption mea-
surements were performed at 548 K, and three kinds of peaks
(1880, 1868 and 1842–1836 cm−1) were observed (Table 2 and
Fig. 3). The NO adsorption peak on H-ZSM-5 was observed at
1877 cm−1, which is almost the same wavenumber as for free
NO: 1876 cm−1. However, the intensity was only a few per-
cent compared to the peak due to NO adsorption on Pd-H-
ZSM-5. The N2O adsorption was not observed on Pd-H-ZSM-
5 in the detection limit of FT-IR.16 Hence, the observed peaks
should be due to adsorbed NO. It is also possible for them to
be intermediate species of the reaction route starting from NO
and terminating with N2, because the FT-IR measurements
were performed at 548 K. However, the adsorption peaks due
18
to NCO,21 NO2,10,11 or N2 in the region of 1920–1800 cm−1
are not reported, or the adsorbed species have already desorbed
at 548 K. The intense, broad peak at 2269–2251 cm−1 ob-
served on Pd-Na-ZSM-5 can be assigned to NCO.21
The adsorption of NO on reduced metallic particles on a Pd/

Y. Izumi et al.
Bull. Chem. Soc. Jpn., 74, No. 8 (2001) 1503
Table 2. Observed Absorption Peak Wavenumbers (cm−1) in the FT-IR Spectra for Pd-H-ZSM-5- in Various Reactions at 548 K (a)
and Reported Values of NO Adsorption Wavenumbers (cm−1) (b)
(a)
Catalyst
NO
NO + propene NO + O2 + propene
NO + N2O + propene
NO + N2O + O2 + propene
Pd-H-ZSM-5
1880 (s),
1842 (m)
1842 (w, br)
1842 (m, br)
1868 (s)
1836 (m, sh)
2994 (w)
2937 (w)
2903 (w)
Blank
1876
The pressure of the gas was the same as in Table 1.
(b)
Catalyst
NO
Reference
12, 13
Pd/SiO2
1660–1645,
1750–1730
Pd-ZSM-5
Cu-ZSM-5
1881, 1836 (NO)
1827, 1734 (twin (NO)2δ−),
1813 (NOδ−),
10
22
1906 (NOδ+),
1964 (Cun–NOδ+
)
Fe-Ya)
1918, 1815 (twin NO)
1882, 1850, 1760 (NO)
23
a) Similar peak pattern was observed also for Fe-mordenite and Fe-ZSM-5.
SiO2 surface was reported at 1660–1645 and 1750–1730 cm−1
(Table 2).12,13 They were assigned as linear or bridged NO on
the Pd particles. Particle size effects on the NO stretching
wavenumber are known. The NO stretching wavenumber in-
creases as the coverage of NO increases, due to the dipole–di-
pole interactions.13 However, even when the particle size ef-
fects and the dipole–dipole coupling effects are taken into ac-
count, the wavenumbers of peaks in Fig. 3 are larger by more
than 100 cm−1 compared to reported peaks on metallic Pd.
This difference demonstrates that the observed NO peaks
(1880, 1868 and 1842–1836 cm−1) were ascribed to NO on
isolated Pd cations inside the zeolite cage and that broad, very
weak peaks at 1730 and 1650 cm−1 were due to NO on metal-
lic Pd particle under our reaction conditions.
compared to the free NO (1876 cm−1). Whether the 4d orbital
of the Pd cation back-donates charge to the NO π* orbital or
whether the filled NO σ orbital donates a charge to the unfilled
Pd 4d orbital, the νN-O stretching wavenumber will decrease or
increase, respectively, compared to the free value of 1876
cm−1.24
The peak at 1842 cm−1 appears in the presence of propene,
and can be assigned as bridging or multifold CO on the Pd in-
stead. As no corresponding peak of terminal CO on Pd was
detected within the range 2100–1900 cm−1 and CO formation
in 3 h of the reactions NO + propene + O2 or NO + N2O +
propene at 548 K was negligible (Fig. 2), the possibility of the
1842 cm−1-peak assignment being due to CO seems small.
The peaks observed at 1881 and 1836 cm−1 on oxidized 0.5
wt% Pd-ZSM-5 at 300 K were assigned to mononitrosyl on Pdꢀ
ligated by lattice oxygen atoms and to mononitrosyl on Pdꢀ,
where one of the O ligands was replaced by a water molecule,
respectively.10 In the cases shown in Figs. 3b–e, it is possible
that the water molecule of the reaction product ligated to the
Pd ion.
As examples of NO adsorption on metal cations inside zeo-
lites, the absorption peak values are listed in Table 2b for Pd-
ZSM-5,10 Cu-ZSM-522 and Fe-ZSM-5, Fe-Y, and Fe-morden-
ite.23 The peak energy splitting of symmetric and asymmetric
peaks of twin NO species was 93–103 cm−1. Looking at the
splitting between peaks in Fig. 3, which is 26–44 cm−1, we do
not assign them to twin NO species. In fact, their intensities
did not follow the same trend in the time course of measure-
ments of the catalytic reaction or of the gradual gas-phase
evacuation. For the mononitrosyl species, the adsorption peak
energy variedby 93–151cm−1 forCu-ZSM-5 and 32–122cm−1
for Fe-Y (Fe-ZSM-5) due to the difference of charge on the ni-
trosyl group. Therefore, the peaks at 1880, 1868, and 1842–
Inthepresence of O2, an additional peak at 2133–2125 cm−1
+
has been reported, and this was assigned as NO2 (Cu-MFI-
81),22 NO2 (Pd-H-ZSM-5),10 or NO+ (Pd-H-ZSM-5).11 No
peaks were observed in this wavenumber region for Pd-H-
ZSM-5 and Pd-Na-ZSM-5 in in situ reaction condition in this
study. This comparison is not contradictory because the peak
at 2133 cm−1 completely desorbed at 373 K,11 whereas our
measurement temperature of FT-IR was 548 K.
0
1836 cm−1 can be assigned as mononitrosyl NOδ+, NO , and
NOδ−, respectively, on the Pd cation. Rather than other species
(N2O, propene, or O2) directly affecting the wavenumber of ni-
trosyl on Pd (change from Figs. 3a to b–d), these other species
The NxO removal was enhanced by the coexistence of NO
and N2O in the case of reactions of NO + N2O + O2 + pro-
pene on Pd-H-ZSM-5 when compared to the sum of the indi-
vidual removal rates in the corresponding conditions (Table 1).
The reactions were monitored by in situ FT-IR. The intense
0
may preferentially block adsorption sites for NOδ+ or NO .
The charge on the nitrosyl group is based on each wavenumber

1504 Bull. Chem. Soc. Jpn., 74, No. 8 (2001)
Simultaneous Removal of NO + N2O
peak of mononitrosyl NO 0 was observed with similar peak in-
tensity to the NO adsorption (Figs. 3e and a), suggesting the
importance of mononitrosyl NO 0 in catalysis. In the absence
of N2O in the reaction mixture, only the very weak peak of
mononitrosyl NOδ− (or weak CO) was observed in NO + O2
+ propene (Fig. 3c).
In the absence of O2, the simultaneous removal rate of NO
+ N2O was the sum of the individual removal rates of NO and
N2O on Pd-H-ZSM-5 (Table 1). Only the peak due to mononi-
trosyl NOδ− (or weak CO) was observed both in NO + pro-
pene and in NO + N2O + propene (Figs. 3b and d).
Reaction Mechanism. One of the possible effects of pro-
pene is to offer carbon in the reaction route via NCO.5,21 The
activation of propene is believed to be related to the acidity of
the zeolite.4,20 The effects of acidity were implied in the com-
parison of the activity between Pd-H-ZSM-5 and Pd-Na-ZSM-
5 (Table 1). The N2 formation rates were always higher over
more acidic Pd-H-ZSM-5 than over less acidic Pd-Na-ZSM-5.
Compared to the gas-phase propene molecules (νCH = 2991.5,
2932.7, and 2869.8 cm−1),25 three weak peaks at 2994, 2937,
and 2903 cm−1) were observed for Pd-H-ZSM-5 in NO + N2O
+ O2 + propene at 548 K (Table 2). As these three peaks
should be typical vibration peaks of CH2 and CH3, surface car-
bonaceous species were proposed to exist during catalysis.
Surface carbonaceous species derived from propene react with
adsorbed NO (or NO2 in the presence of O2),3,6 or adsorbed
CO reacts with nitrogen species, and further reacts with anoth-
appropriate concentration of surface carbonaceous species to
below the level that leads to inhibition of NO adsorption and
dissociation (Scheme 1). The importance of carbonaceous
species produced from NO + hydrocarbon has been suggested
elsewhere.3,6,26,27 The effects of oxygen derived from N2O may
be similar to the effects of the oxygen derived from O2 or NO
(Scheme 1),2 but the dissociation of N2–O is thermodynamical-
ly far easier than that of O–O or N–O. N2O decomposition is
reported to be insensitive to the presence of O2, whereas NO
decomposition is often influenced by the presence of
O2.2,4,6,8,9,11,20,28 This general trend may suggest different active
sites for NO and N2O dissociation. An alternative interpreta-
tion is that the reactivity of NO and N2O is different with metal
ion in zeolite, and dissociated oxygen species may have differ-
ent chemical natures. More reactive oxygen species dissociat-
ed from N2O were reported on Cu-ZSM-5 (Cu2+-O−)28 and on
Fe-ZSM-5 (α-oxygen).29 Rate dependence on reactant pres-
sure should be monitored to make this effect of O of N2O more
certain. The N2O concentration was constant in Fig. 2b after
30 min of reaction, due to the steady-state balance of N2O de-
composition and the formation from NO. With the major ef-
fects of dissociated oxygen derived from N2O to remove the
carbonaceous species on surface Pd cation, the NO 0 on the Pd
cation (1868 cm−1) became the major surface species (Fig. 3e).
As the concentration of this species is the balance of the reac-
tion path rates in Scheme 1, the promotion effects were not ob-
served in NO + N2O + propene (in the absence of O2) on Pd-
H-ZSM-5.
er NO molecule (or NO2 in the presence of O2) to form N2 +
5,21
CO2 or N2O (Scheme 1-i).
The carbonaceous species can
The surface site of the 1868 cm−1-peak is suggested to be
responsible for the dissociation of NO (Scheme 1-i) in the pro-
moted simultaneous removal. Aparicio et al. reported the cor-
relation between the N2O decomposition reaction and the
mononitrosyl IR peak intensity, and suggested that the mono-
nitrosyl site is the active site for Fe-ZSM-5.23 The Pd ion in-
troduced to H-ZSM-5 (0.3 wt% of Pd) was reported to exist as
Pd2+ in the reaction condition of the NO + CH4 + O2 reaction
at 773 K, based on the X-ray absorption fine structure spec-
troscopy.7 The Pd ion introduced to H-ZSM-5 (0.44 wt% of
Pd) existed as Pd2+ below 610 K and is suggested to be the ac-
tive site for NO reduction by methane, based on the FT-IR
measurements.11 As the pertinent Pd sites for NO and/or N2O
decompositions were found to be dispersed Pd2+ inside zeolite
(Pd-H-ZSM-5) and small Pd particles (Pd/SiO2) in this paper,
the reaction mechanism of NO and/or N2O decompositions
should be different.
be consumed to form CO and CO2 in the route of by-reaction
(Scheme 1-iv) or can remain as surface deposits that act as an
inhibitor of catalysis. It is unclear why the NCO species was
detected on Pd-Na-ZSM-5 and not on Pd-H-ZSM-5 in NO +
propene or in NO + O2 + propene at 548 K. The N2 formation
rates on the former were slower (55–75%) than those of the
latter. The NCO on the Pd of the former catalyst may poison
the site, but NCO did not adsorb on the latter catalyst in the
time scale of the FT-IR measurements (i.e. stayed on the sur-
face during a very short time), and may convert quickly to N2.
We propose that the key to the promotion of simultaneous
removal on Pd-H-ZSM-5 is the dissociated oxygen atom of the
adsorbed N2O (Scheme 1-ii). The oxygen atom separated from
N2O on the catalyst surface may play a role in controlling the
The intensity of two very broad peaks (1730 and1650 cm−1)
in NO at 548 K for Pd-H-ZSM-5 that could be ascribed to
mononitrosyl on Pd particles was negligible. Either only a
small amount of Pd particles existed, or the surface of the Pd
metal particles was dominantly covered by oxygen (similar ef-
fects were reported in Ref. 11) such that it could not adsorb
NO in the reaction conditions used for NO and/or N2O remov-
al. Either possibility supports Pd2+ sites inside zeolite to be
active sites in the case of Pd-H-ZSM-5 catalyst.
The promoted simultaneous removal was observed for Pd/
SiO2 but not for Pd-H-ZSM-5 in the presence of propene and
absence of O2 (Table 1-iii). One of the possible reasons of this
difference is the different dissociation ability of N2–O bond ei-
ther in the case of Pd microparticle (Pd/SiO2) or Pd ion (Pd-H-
Scheme 1. Proposed reaction pathway of the promoted si-
multaneous catalytic removal of NO and N2O in the pres-
ence of O2 and propene.

Y. Izumi et al.
Bull. Chem. Soc. Jpn., 74, No. 8 (2001) 1505
References
ZSM-5). The steady-state formation rate of N2 + 1/2O2 from
N2O was as much as 10 times faster on Pd-H-ZSM-5 than on
Pd/SiO2 (Table 1-i), but the initial non-stoichiometric N2 for-
mation was faster on Pd/SiO2 rather than on Pd-H-ZSM-5.
Therefore, the dissociation of N2–O bond was faster on Pd mi-
croparticle surface than on Pd ion site, but the formation and
desorption of O2 was slow on Pd microparticles. In the pres-
ence of propene, formed surface O species can be converted by
carbonaceous species derived from propene. N2–O bond dis-
sociation (Scheme 1-ii) was fast enough on Pd/SiO2, and the
promoting effect of O(ad) (i.e. promoted simultaneous remov-
al) for the reaction step (iv) of Scheme 1 may be observed in
the presence of propene only for Pd/SiO2.
1
2
3
M. Shelef, Chem. Rev., 95, 209 (1994).
Y. Li and J. N. Armor, Appl. Catal., B, 3, 55 (1993).
M. Kogel, R. Monnig, W. Schwieger, A. Tissler, and T.
Turek, J. Catal., 182, 470 (1999).
4
(1995).
5
C. J. Loughran and D. E. Resasco, Appl. Catal., B, 7, 113
M. D. Amiridis, T. Zhang, and R. J. Farrauto, Appl. Catal.,
B, 10, 203 (1996).
6
(1995).
7
H. Matsumoto and S. Tanabe, J. Phys. Chem., 99, 6951
A. Ali, W. Alvarez, C. J. Loughran, and D. E. Resasco, Ap-
pl. Catal., B, 14, 13 (1997).
8
K. J. Zhen, M. M. Khan, C. H. Mak, K. B. Lewis, and G. A.
The biggest question still remains; why simultaneous re-
moval was promoted via the oxygen atom dissociated from
N2O only on the Pd ion and not on the Co and Fe ions.2,3 As
H-ZSM-5 was inactive for NO removal in the reaction condi-
tions in this paper, the difference may be ascribed to the differ-
ence of valence state among metals. These metal-zeolite cata-
lysts were prepared from Pd2+, Co2+, and Fe2+.2,3 As ex-
changed Co2+ and Fe2+ can be oxidized to Co3+ and Fe3+, dis-
sociated O may be bound to Co3+ or Fe3+ and stabilized. On
the other hand, the redox of palladium is between Pd0 and Pd2+
and dissociated O may be unstable when it is bound to Pd2+.
These difference of valence states among metals may be the
major reason that promoted simultaneous removal of NO +
N2O was observed only in the case of Pd catalysts. The oxy-
gen atom dissociated from N2O may be relatively unstable on
the Pd site, and further reaction with carbonaceous species
formed from propene can occur very quickly.
Somorjai, J. Catal., 94, 501 (1985).
9
R. Burch and S. Scire, Appl. Catal., B, 3, 295 (1994).
10 C. Descorme, P. Gelin, M. Primet, and C. Lecuyer, Catal.
Lett., 41, 133 (1996).
11 L. J. Lobree, A. W. Aylor, J. A. Reimer, and A. T. Bell, J.
Catal., 181, 189 (1999).
12 C. M. Grill and R. D. Gonzalez, J. Phys. Chem., 84, 878
(1980).
13 S. Moriki, Y. Inoue, E. Miyazaki, and I.Yasumori, J. Chem.
Soc., Faraday Trans. 1, 78, 171 (1982).
14 Y. Izumi, T. Chihara, H. Yamazaki, and Y. Iwasawa, J.
Phys. Chem., 98, 594 (1994).
15 Y. Izumi, K. Asakura, and Y. Iwasawa, J. Catal., 127, 631
(1991).
16 Y. Izumi, K. Oshihara, and K. Aika, Chem. Lett., 1998,
727.
17 Y. Izumi and Y. Iwasawa, J. Phys. Chem., 96, 10942
(1992).
Conclusions
18 Y. Izumi, M. Hoshikawa, and K. Aika, Bull. Chem. Soc.
Jpn., 67, 3191 (1994).
19 Y. Izumi, T. Chihara, H. Yamazaki, and Y. Iwasawa, J.
Chem. Soc., Dalton Trans., 3667 (1993).
1. Simultaneous removal of NO + N2O was promoted in O2
+ propene over Pd-H-ZSM-5 and conventional Pd/SiO2 cata-
lysts, and in propene on conventional Pd/SiO2, when compared
to the sum of the corresponding individual removal rates of
NO or N2O.
2. In the conditions that promoted simultaneous removal,
an intense peak ascribed to the NO 0 on Pd cation in a ZSM-5
cage was observed at 1868 cm−1. When NO was introduced to
Pd-ZSM-5, the mononitrosyl (NOδ+) peak was observed at
1880 cm−1. In the other conditions used for individual remov-
al or unpromoted simultaneous removal, a medium or weak
peak was observed at 1842 cm−1.
3. Based on peak intensity in FT-IR spectra, the pertinent
state of palladium for NO and/or N2O decomposition was
found to be Pd2+ for Pd-H-ZSM-5, in contrast to the two broad
peaks ascribed to surface metal particles observed for conven-
tional Pd/SiO2.
4. The promotion effects of the oxygen atom dissociated
from N2O were suggested as a mechanism for keeping the Pd
site clear from extra carbonaceous species. The relative stabil-
ity of the oxygen may be related to the difference between pro-
moted simultaneous removal on the Pd site and essentially the
same rates of simultaneous removal as NO removal (N2O re-
moval was inhibited) on the Co and Fe sites.
20 Y. Nishizaka and M. Misono, Chem. Lett., 1993, 1295.
21 G. R. Bamwenda, A. Ogata, A. Obuchi, J. Oi, K. Mizuno,
and J. Skrypek, Appl. Catal. B, 6, 311 (1995).
22 M. Iwamoto, H. Yahiro, N. Mizuno, W. Zhang, Y. Mine, H.
Furukawa, and S. Kagawa, J. Phys. Chem., 96, 9360 (1992).
23 L. M. Aparicio, W. K. Hall, S. Fang, M. A. Ulla, W. S.
Millman, and J. A. Dumesic, J. Catal., 108, 233 (1987).
24 G. W. Smith and E. Carter, J. Phys. Chem., 95, 2327
(1991).
25 G. Herzberg, “Molecular Spectra and Molecular Structure
Ⅲ. Electronic Spectra and Electronic Structure of Ployatomic
Molecules,” Van Nostrand Reinhold Co., New York (1966), p.
650.
26 H. Y. Chen, T. Voskoboinikov, and W. M. H. Sachtler,
“Proceedings of 2nd World Congress on Environmental Cataly-
sis,” J. N. Armor, Ed., Miami (1998), p. 105.
27 G. P. Ansell, A. F. Diwell, S. E. Golunski, J. W. Hayes, R.
R. Rajaram, T. J. Truex, and A. P. Walker, Appl. Catal., B, 2, 81
(1993).
28 M. V. Konduru and S. S. C. Chuang, J. Catal., 196, 271
(2000).
29 G. I. Panov, A. K. Uriarte, M. A. Rodkin, and V. I. Sobolev,
Catal. Today, 41, 365 (1998).
This work was partially supported by a grant from Toray
Science Foundation (No. 98-3901, Y.I.).