Professional Documents
Culture Documents
www.elsevier.com/locate/apcatb
Abstract
In this study, new Fe2O3 based materials are developed for the selective catalytic reduction (SCR) of NOx by NH3 in diesel exhaust. As a
result of the catalyst screening, performed in a synthetic model exhaust, ZrO2 is considered to be the most effective carrier for Fe2O3. The
modification of the Fe2O3/ZrO2 system with tungsten leads to drastic increase of SCR performance as well as pronounced thermal stability.
These results show that tungsten acts as bifunctional component. The highest catalytic activity is observed for ZrO2 that is coated with
1.4 mol% Fe2O3 and 7.0 mol% WO3 (1.4Fe/7.0W/Zr). By the use of this catalyst quantitative conversion of NOx is obtained between 285 and
430 8C with selective formation of N2. Here, the turnover frequency of NOx per Fe atom is found to be 35 105 s1 that indicates a high
catalytic performance. The SCR activity of the 1.4Fe/7.0W/Zr material is decreased in the presence of H2O and CO2, whereas it is increased
by NO2.
Temperature programmed reduction by H2 (HTPR) analyses show that the Fe sites of the 1.4Fe/7.0W/Zr catalyst are mainly in the form of
crystalline Fe2O3, whereby relatively small oxide entities are also present. The strongly aggregated Fe2O3 species are associated with the
presence of the promoter tungsten. Based upon stationary catalytic examinations as well as diffuse reflectance infrared fourier transform
spectroscopy (DRIFTS) studies we postulate an Eley Rideal type mechanism for SCR on 1.4Fe/7.0W/Zr catalyst. The mechanistic model
includes a redox cycle of the active Fe sites. As first reaction step, we assume dissociative adsorption of NH3 that leads to partial reduction of
the iron as well as to production of very reactive amide surface species. These amide intermediates are supposed to react with gaseous NO to
form N2 and H2O. In the final step, the reduced Fe sites be regenerated by oxidation with O2. As a side reaction of SCR, imide species,
originated from decomposition of amide, are oxidized by NO2 or O2 into NO.
# 2005 Elsevier B.V. All rights reserved.
Keywords: SCR; Diesel exhaust; NOx; NH3; Fe2O3; ZrO2; WO3; Reaction mechanism
0926-3373/$ – see front matter # 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcatb.2005.07.004
N. Apostolescu et al. / Applied Catalysis B: Environmental 62 (2006) 104–114 105
from Ströhlein Instruments employing N2 as adsorbate. The Siemens). After passing the Binos 5 analyser H2O and NH3
crystalline phase and BET surface area of the carriers as well are removed from the gas stream by a condenser unit (3 8C)
as pure Fe2O3 are summarized in Table 1. After modification and liquid H3PO4, respectively, to avoid cross sensitivity in
of these materials with dopants, i.e. Fe2O3, WO3, phosphate the NOx and N2O analysis.
or sulfate, both PXRD and BET give the same results as that The performance of the catalysts is in terms of percentual
demonstrated in Table 1. conversion of NOx (X(NOx)), Eq. (1); for the most active
material the percentual conversion of NH3 (X(NH3)), Eq. (1)
and the percentual selectivity of N2O (S(N2O)), Eq. (2) are
4. Catalytic studies also shown.
Fig. 1. Effect of the W load on the temperature with 90% NOx conversion in
SCR on fresh (*) and aged (&) 1.4Fe/W/Zr catalysts. Reaction conditions: Fig. 2. PXRD patterns of the fresh (a) and aged (b) 1.4Fe/7.0W/Zr catalyst.
m(catalyst) = 5.9 g, c(NO) = c(NH3) = 500 ppm, c(O2) = 5 vol.%, N2 bal- The reflexes indicated by arrows are referred to WO3, while all the other
ance, W/F = 0.18 g s scm3 and DT/Dt = 1.7 K min1. Aging conditions: peaks are assigned to tetragonal ZrO2. For simplicity the ZrO2 reflexes are
72 h at 600 8C in air. only labelled for pattern (b).
As the 1.4Fe/7.0W/Zr catalyst exhibits the highest SCR the TOF value is two orders of magnitude smaller than for
performance among all the materials tested (Table 2) as well Fe-ZSM-5 zeolite [26]. At higher temperature the 1.4Fe/
as good thermal stability this sample is employed in the 7.0W/Zr catalyst shows a slight decrease of NOx conversion.
following examinations. If not mentioned otherwise the The SCR reaction also results in the formation of N2O as by-
catalyst is thermally pretreated as described above (72 h at product, whereby the evolution of N2O starts at 250 8C; the
600 8C). maximum N2O selectivity of 6% is reached at 500 8C. The
The BET surface area of the 1.4Fe/7.0W/Zr catalyst SCR data measured under stationary conditions at 200, 250,
(60 m2 g1) is not affected by the thermal treatment. 350 and 450 8C are listed in Table 3 together with the
Furthermore, the PXRD pattern of fresh 1.4Fe/7.0W/Zr corresponding results of TPR. The comparison shows that
shows the typical reflexes of tetragonal ZrO2, while in the the difference of X(NOx) as well as S(N2O) is rather small,
diffractogram of the aged sample signals of WO3 are also whereas for X(NH3) there is a large distinction between
observed (Fig. 2). These additional reflexes are attributed to stationary and dynamic data. This proves that stationary
agglomeration of WO3 species that leads to the formation of conditions are not yet attained in TPR. The rather low
crystalline domains. conversion of NH3 observed in dynamic measurement is
associated with its thermal desorption from the catalyst.
8.3. Comparison of TPR with stationary data and mass Contrary, there is no significant release of NH3 from the
balance of nitrogen reactor and the stainless steel pipes of the bench. This is
proven by NH3 TPD performed in the absence of a catalyst.
The TPR data of the 1.4Fe/7.0W/Zr catalyst recorded The TPD spectrum of the 1.4Fe/7.0W/Zr material shows
under the standard reaction conditions show that the SCR NH3 desorption between 85 and 550 8C with two broad
reaction begins at 150 8C (Fig. 3) and leads to entire peaks at 155 and 295 8C and a shoulder at about 425 8C (not
conversion of NOx in the temperature range from 285 to depicted). The NH3 uptake of the 1.4Fe/7.0W/Zr catalyst is
430 8C. The complete conversion of NOx corresponds to a 108 mmol/gcat. The fresh and aged material as well as the
TOF value of 35 105 s1 (Table 2). This turnover pure ZrO2 support exhibit the same NH3 TPD profile.
frequency indicates that the 1.4Fe/7.0W/Zr sample is very Based upon the stationary data the N2 concentration is
active in the selective catalytic reduction of NOx. However, calculated from the balance of nitrogen (Eq. (3)) in order to
Table 3
Comparison of the SCR performance of the aged 1.4Fe/7.0W/Zr catalyst under dynamic (TPR) and stationary conditionsa
T (8C) X(NOx)TPR (%) X(NOx)stat (%) S(N2O)TPR (%) S(N2O)stat (%) X(NH3)TPR (%) X(NH3)stat (%)
200 36 44 0 0 21 43
250 90 96 0 0 74 96
350 100 100 2 0 93 100
450 98 94 5 2 99 97
a
Reaction conditions: m(catalyst) = 5.9 g, c(NO) = c(NH3) = 500 ppm, c(O2) = 5 vol.%, N2 balance, W/F = 0.18 g s scm3, in TPR: DT/Dt = 1.7 K min1.
N. Apostolescu et al. / Applied Catalysis B: Environmental 62 (2006) 104–114 109
determine the N2 selectivity (Eq. (4)). The result indicates a 50% at about 325 8C. Since the evolution of both NOx and
selective formation of N2 between 200 and 350 8C. At N2O is not observed, it is very likely that the NH3/O2 reaction
450 8C, where N2O forms, the selectivity of N2 is negligibly on 1.4Fe/7.0W/Zr results in formation of N2 (Eq. (6)).
reduced to 0.98.
4NH3 þ 3O2 ! 2N2 þ 6H2 O (6)
ðcðNOx Þin cðNOx Þout Þ þ ðcðNH3 Þin
cðNH3 Þout Þ 2cðN2 OÞ
cðN2 Þ ¼ (3) 8.5. Effect of O2 on the SCR activity of the 1.4Fe/7.0W/
2 Zr catalyst
2cðN2 Þ
SðN2 Þ ¼ (4) As ammonia is strongly oxidized by O2 on the 1.4Fe/
cðNOx Þin cðNOx Þout
7.0W/Zr catalyst (Section 8.4) the effect of oxygen on SCR
Furthermore, the stationary results show a molar NOx:NH3 activity is examined. For this purpose the concentration of
stoichiometry of 1. In agreement with literature [12] and the O2 is varied from 0 to 15 vol.%. Fig. 5 evidences that the
mass balance of nitrogen this ratio is referred to the overall SCR reaction is drastically suppressed at temperatures above
reaction given in Eq. (5). 300 8C if the O2 concentration increases to 10 or 15 vol.%.
NH3 is then completely oxidized above 340 8C, while N2O is
4NO þ 4NH3 þ O2 ! 4N2 þ 6H2 O (5)
only formed in negligible concentrations being smaller than
12 ppm (not depicted). This observation is in good agreement
8.4. Effect of O2 on oxidation of NH3 on the with the NH3 oxidation data that show an enhanced
1.4Fe/7.0W/Zr catalyst conversion of ammonia with increasing concentration of
O2. Hence, we conclude that at elevated temperatures SCR is
The catalytic oxidation of NH3 by O2 is examined as SCR restricted by the competitive consumption of ammonia by
can principally be limited by competitive NH3/O2 reaction. catalytic NH3/O2 reaction. However, Fig. 5 also shows that O2
The oxidation of NH3 is carried out in the absence of NOx, is an essential component for SCR because in the absence of
whereby different concentrations of O2, i.e. 5, 10 and O2 only a small amount of NOx is converted.
15 vol.%, are used. In these TPR experiments X(NH3) first
shows negative values due to thermal desorption and later on it 8.6. Effect of CO2, H2O and NO2 on the SCR activity of
starts increasing (Fig. 4). With 5 vol.% oxygen in the feed the 1.4Fe/7.0W/Zr catalyst
NH3 conversion starts increasing at 305 8C, whereby a
conversion of 50% is reached at 370 8C. Complete NH3 The effect of carbon dioxide is examined by increasing
oxidation is observed at 450 8C (Fig. 4). As compared to TPR the concentration to 5 and 10 vol.% in the feed. The TPR
with 5 vol.% O2 the oxidation of NH3 is accelerated if 10 and data (not depicted) show that the temperature at which NOx
15 vol.% O2 are dosed. However, the measurements, in which conversion reaches 90% does not change in comparison
10 and 15 vol.% O2 is added, show very similar results. The to the measurement performed in the absence of CO2.
conversion of NH3 increases already at 250 8C and reaches However, the temperature region exhibiting complete NOx
110 N. Apostolescu et al. / Applied Catalysis B: Environmental 62 (2006) 104–114
Fig. 5. Effect of O2 on SCR performance of aged 1.4Fe/7.0W/Zr catalyst. Fig. 6. Effect of H2O on SCR performance of aged 1.4Fe/7.0W/Zr catalyst.
Reaction conditions: m(catalyst) = 5.9 g, c(NO) = c(NH3) = 500 ppm, Reaction conditions: m(catalyst) = 5.9 g, c(NO) = c(NH3) = 500 ppm,
c(O2) = 0, 5, 10 or 15 (—) vol.%, N2 balance, W/F = 0.18 g s scm3 and c(O2) = 5 vol.%, c(H2O) = 5 vol.%, N2 balance, W/F = 0.18 g s scm3
DT/Dt = 1.7 K min1. and DT/Dt = 1.7 K min1.
reduction is diminished. Entire conversion of NOx is only The high selectivity of N2O increasing to 90% is
observed between 285 and 390 8C with 5 vol.% CO2, while attributed to the pre-treatment procedure. In TPR the
in the presence of 10 vol.% it occurs from 285 to 360 8C. catalyst is exposed to the feed gas for 30 min at 150 8C
From these results we deduce that CO2 leads to a slight before the measurement is started. In case of NO2 dosage
diminution of the SCR activity of 1.4Fe/7.0W/Zr catalyst. this treatment obviously leads to the formation of a
A stronger deactivation is observed when water is added substantial amount of NH4NO3. This interpretation is
to the feed (Fig. 6), whereby the same results are obtained supported by stationary SCR data that are recorded without
for 5 and 10 vol.% H2O. In both measurements, NOx corresponding pre-treatment. These results (not illustrated)
conversion of 90% is only attained at 275 8C, while total show a maximum N2O selectivity of 10% only that ranges
removal of NOx occurs between 320 and 365 8C. However, from 220 to 225 8C.
in the presence of H2O nitrous oxide is not formed. The Ammonium nitrate is not formed if the standard TPR
deactivation of the catalyst by CO2 and especially by H2O is conditions are used because no NO2 is dosed here.
very likely associated with the competitive adsorption on the Furthermore, the 1.4Fe/7.0W/Zr catalyst exhibits a rather
active Fe sites of 1.4Fe/7.0W/Zr. low NO oxidation activity as compared to the highly active
In contrast to H2O, the dosage of 250 ppm NO2 (c(NO)in/
c(NO2)in = 1) results in clear increase of catalytic perfor-
mance (Fig. 7) as a strong diminution of nitrogen oxides is
observed in the whole temperature range. After the start of
TPR the conversion of NOx first decreases and then it goes
through a minimum of 42% at 180 8C. A detailed analysis
shows that only the concentration of NO2 increases in this
region, while NO decreases continuously. Complete con-
version of NOx is found at 260 8C, whereas the conversion
restarts decreasing at 350 8C. Furthermore, the TPR data
indicate the formation of nitrous oxide at low temperature
showing peak selectivity at 190 8C.
The maximum of NOx conversion observed at 150 8C is
referred to the reaction of NO2 with NH3 to form NH4NO3. It
is well known in SCR that ammonium nitrate is produced
below 200 8C if a relatively high concentration of NO2 is
present [41]. With rising temperature the thermal decom-
position of NH4NO3 prevails that results in the production of
Fig. 7. Effect of NO2 on SCR performance of aged 1.4Fe/7.0W/Zr catalyst.
N2O and H2O [41]. Hence, the evolution of N2O observed in Reaction conditions: m(catalyst) = 5.9 g, c(NO) = c(NO2) = 250 ppm,
the low temperature region is associated with the thermal c(NH3) = 500 ppm, c(O2) = 5 vol.%, N2 balance, W/F = 0.18 g s scm3
decomposition of the formed ammonium nitrate. and DT/Dt = 1.7 K min1.
N. Apostolescu et al. / Applied Catalysis B: Environmental 62 (2006) 104–114 111
Pt systems [42]. The stationary data show a NO2 yield of 5% and W species and hence a careful conclusion should be
at 150 8C, while the maximum yield amounts to 10% drawn from this simplification. The modified HTPR curve
(340 8C). With 10 and 15 vol.% O2 the peak yield of NO2 is (not illustrated) shows a H2/Fe ratio of 1.3 that points to the
also found at 340 8C being 15 and 20%, respectively. predominant existence of crystalline Fe2O3; for a-Fe2O3 the
H2/Fe ratio is 1.5 as already mentioned above. Moreover, the
8.7. Fe sites of the 1.4Fe/7.0W/Zr catalyst HT/LT ratio is estimated to be 1 indicating that in addition to
the crystalline aggregates other Fe3+ containing species are
The nature of the iron sites of the 1.4Fe/7.0W/Zr catalyst also present. Based upon the estimated ratios of LT/HT and
is studied by the use of HTPR. For reference purposes H2/Fe as well as the respective values for a-Fe2O3 we
corresponding examinations are also performed with 7.0W/ carefully deduce that about one half of the Fe3+ sites exists as
Zr, 1.4Fe/Zr and a a-Fe2O3 sample. The a-Fe2O3 pattern is crystalline entities, while the second half is in the form of
prepared by PVA technique, whereby the final calcination is other structures. It has been shown that iron-exchanged ZSM-
carried out at 600 8C for 5 h; the crystalline phase is 5 zeolites reveal different Fe3+ oxo species, i.e. isolated,
confirmed by PXRD. oligomeric or strongly aggregated structures [24,47]. In
The pure iron oxide sample indicates a HTPR pattern HTPR, such sites exhibit a H2 consumption peak at ca. 420 8C
(Fig. 8(a)) that is typical for a-Fe2O3, i.e. a low temperature [20,48] that is in the range of the LT peak of 1.4Fe/7.0W/Zr.
(LT) and a high temperature (HT) signal is observed with a This feature at 420 8C is generally associated with the
HT/LT peak ratio very close to 8. The total consumption of reduction of Fe3+ to Fe2+ species, while the further conversion
H2 that is expressed as molar ratio of H2/Fe amounts to 1.5. to elemental Fe occurs at very high temperatures (>900 8C).
Moreover, the PXRD pattern of the iron oxide removed after These temperatures are not considered in our HTPR
the first HTPR peak shows exclusively reflexes that are experiments; thus, this potential reduction step is not possible
referred to Fe3O4. These results confirm literature data that to ascertain. Moreover, Chen and Sachtler [28] have observed
prove the reduction of Fe2O3 by H2 to be a two-step in their HTPR study of Fe-ZSM-5 a H2 consumption peak that
mechanism according to the sequence Fe2O3 ! Fe3O4 ! Fe appears at 430 8C. The authors associate this feature with the
[43–46]. existence of an oxygen bridged binuclear Fe complex.
The HTPR curve of the 1.4Fe/7.0W/Zr sample reveals a Additionally, for an overexchanged Fe-ZSM-5 sample
broad LT peak, that is shifted to higher temperature (540 8C) Grünert and co-workers have reported the coexistence of
as referred to a-Fe2O3, and two overlapped HT signals located crystalline Fe2O3 and Fe3+ oxo sites [48]. This material
at 730 and 840 8C (Fig. 8(b)). The latter HT feature also reveals a HT/LT ratio of 3.5 and a H2/Fe ratio of 1.4. Taking
appears in case of 7.0W/Zr (Fig. 8(c)) and is therefore into account the literature data mentioned above we interpret
assigned to the reduction of WO3 species. Furthermore, it is the HTPR profile of the 1.4Fe/7.0W/Zr catalyst with the pre-
obvious that the relative intensity of the LT peak is much more dominant existence of crystalline Fe2O3 (ca. 50%), whereby a
pronounced as compared to a-Fe2O3. To estimate the HT/LT significant amount of relatively small oxide entities is also
ratio as well as the H2 consumption of the 1.4Fe/7.0W/Zr present. However, the detailed structures of the latter species
catalyst we subtract the HTPR pattern of 7.0W/Zr from that of can hardly be explained by the use of HTPR only.
1.4Fe/7.0W/Zr. However, it must be noted here that this Contrary, the 1.4Fe/Zr catalyst shows a completely
procedure does not imply possible interactions between Fe different HTPR profile as compared to 1.4Fe/7.0W/Zr. Here,
a single HTPR peak is observed that appears at 480 8C. In
agreement with our HTPR analysis of a-Fe2O3 (Fig. 8(a)) as
well as reference data from Grünert and co-workers [48]
this result evidences the absence of a- and g-Fe2O3. As
discussed for the 1.4Fe/7.0W/Zr catalyst we relate this
single peak to the exclusive presence of small Fe3+
containing oxide structures. Furthermore, the H2/Fe ratio
found to be 0.85 leads to the conclusion that in HTPR some
Fe3+ species are reduced to a lower valency than 2+; for
comparison the reduction of Fe3+ to Fe2+ species corre-
sponds to a H2/Fe ratio of 0.5. A very similar interpretation
has been proposed by Chen and Sachtler for the HTPR of a
Fe-ZSM-5 type catalyst [28]. In PXRD analysis performed
after the HTPR analysis of 1.4Fe/Zr no reflexes attributable
to Fe or iron oxides are detected.
The different HTPR data obtained for 1.4Fe/Zr and
1.4Fe/7.0W/Zr provide evidence that the promoter markedly
Fig. 8. HTPR patterns of: (a) a-Fe2O3 prepared by PVA technique, causes the formation of larger oxide units, i.e. crystalline
(b) 1.4Fe/7.0W/Zr, (c) 7.0W/Zr and (d) 1.4Fe/Zr. Fe2O3. This effect might be related to the presence of
112 N. Apostolescu et al. / Applied Catalysis B: Environmental 62 (2006) 104–114
Fig. 9 shows the rate of NOx reduction versus the 1440 and 1610 cm1 (Fig. 10). Furthermore, two broad
concentration of NH3 at 300 8C. The rate is defined by features are observed at higher wavenumbers, i.e. at 3265
Eq. (7) and refers to the steady state. In these measurements and 3355 cm1. Contrary, the catalyst exposed to NO instead
the standard feed gas is taken, whereby ammonia of NH3 does not show any DRIFTS bands. From these results
concentration ranges from 90 to 1050 ppm. First, r(NOx) we deduce that ammonia is the species that adsorbs on the
increases linearly, while at NH3 concentrations above catalyst surface, while NO reacts from the gas phase.
500 ppm it changes no more. These data exclude a Langmuir The DRIFT spectrum of the Fe free 7.0W/Zr system,
Hinshelwood type mechanism that should result in decrease recorded after NH3 treatment, shows bands at 1200 and
of SCR rate if one reactant is dosed in excess. However, the 1610 cm1 only, whereas no feature is detected at
curve is typical for the Eley Rideal type mechanism, i.e. one 1440 cm1. Moreover, same peaks as for 1.4Fe/7.0W/Zr
educt adsorbs on active sites and the second one reacts from are found in the spectral range above 3000 cm1.
gas phase. This type of mechanism is also discussed by The peaks observed at ca. 1220 and 1610 cm1 are
Ramis et al. for SCR on Fe2O3 [4]. referred to the das(NH3) and ds(NH3) vibrations of ammonia
molecularly adsorbed on Lewis acid sites [49]. Furthermore,
FTr
rðNOx Þ ¼ ðcðNOx Þin cðNOx Þout Þ (7) the band located at 1440 cm1 is associated with the d(NH)
Ts nFe2 O3
vibration of imide species [4,50]. This feature is not
The DRIFT spectrum of the 1.4Fe/7.0W/Zr catalyst recorded interpreted as the ds(NH4) vibration of ammonium species
during NH3 exposure at 200 8C evidences three bands at 1245, that might be originated from the reaction of NH3 with
Brønsted acid OH groups [49,51]. Such species are formed
on 1.4Fe/7.0W/Zr catalyst by NH3 treatment at 50 8C that is
performed after heating at 500 8C in N2. The band of the oxidant. The measurements show the fastest rate of imide
ds(NH4) vibration, depicted in Fig. 11(a), is located at conversion if NO2 is present, whereas no reaction occurs in
1465 cm1 that is very close to the frequency of the d(NH) NO/N2 atmosphere. As imide does not react with NO it is
vibration. Subsequent treatment at 200 8C in N2 leads to deduced that this surface species plays no role in SCR, i.e.
complete disappearance of the 1465 cm1 band, while a new N2 formation. However, the DRIFTS data indicate complete
feature is detected at 1440 cm1 (Fig. 11(b)). For conversion of imide by O2 and NO2. Since N2O is not
comparison, the same experiments are carried out with formed in SCR at 200 8C (Table 3) we conclude that imide is
the 7.0W/Zr catalyst. The DRIFT spectrum recorded at oxidized here into NO according to Eqs. (8) and (9). In
50 8C also shows the 1465 cm1 band. This peak is totally contrast to our conclusion, Ramis et al. [4] speculate amide
removed at 200 8C, whereas no signal appears at 1440 cm1. to be involved in the production of N2O.
These results prove that the bands at 1440 and 1465 cm1
are associated with different surface species, i.e. imide and 4ðNHÞ þ 3O2 ! 4NO þ 2H2 O (8)
ammonium. Furthermore, the DRIFTS data indicate
4ðNHÞ þ 6NO2 ! 10NO þ 2H2 O (9)
formation of imide species by heating the catalyst to
200 8C after exposure to NH3 at 50 8C. We assume that the However, we agree with Ramis et al. [4] that imide is
production of imide is referred to the thermal decomposition originated from amide species. Following this route we
of NH3 adsorbed on catalytically active iron sites. A more assume that first NH3 is molecularly coordinated to the
detailed mechanism is presented below. Fe sites of the catalyst resulting in partial reduction of
The DRIFTS bands at 3265 and 3355 cm1, found after the iron. The amide species might then be formed by H+
NH3 exposure at 50 and 200 8C, are ascribed to the abstraction. However, our DRIFTS study does not evidence
symmetric and asymmetric stretching vibrations of the the existence of amide, probably due to its thermal instabil-
coordinated ammonia [49,52]. ity. We therefore postulate that amide rapidly undergoes
The DRIFTS results of 7.0W/Zr and 1.4Fe/7.0W/Zr decomposition into imide, whereby a H+ is abstracted. Since
provide evidence that the imide species are exclusively imide is obviously not involved in SCR, we suppose that the
formed on the Fe2O3 component of the 1.4Fe/7.0W/Zr amide intermediate is the key component for SCR, i.e. the
catalyst. On the other hand, it is obvious that the molecular reaction of amide with gaseous NO might lead to N2
NH3 is principally coordinated to all available Lewis acid formation. This supposition is consistent with the mechan-
sites, e.g. Zr4+, W6+ and Fe3+. As the NH3 TPD pattern of ism proposed by Ramis et al. [4].
1.4Fe/7.0W/Zr and pure ZrO2 is the same, it is very likely In the final reaction step, the reduced Fe sites have to be
that that the main part of NH3 adsorbs on the Zr4+ sites. regenerated. We assume that this regeneration occurs by
After the NH3 exposure at 200 8C and purging with N2, oxidation with O2 as, in addition to NO and NH3, the oxygen
the 1.4Fe/7.0W/Zr catalyst is treated with NO/N2, O2/N2 or is also essential for SCR reaction (Fig. 5). Summarizing the
NO2/O2/N2 as described in Section 6. Fig. 12 shows the single steps of our proposed catalytic cycle (Fig. 13) the
development of the DRIFTS band of the imide species stoichiometry of the overall reaction includes the consump-
(1440 cm1) with time and also in dependence of the type of tion of 2NH3, 2NO and 0.5O2, while 2N2 and 3H2O are
produced. This stoichiometry is in agreement with Eq. (5)
that is deduced from stationary SCR data (Table 3).
The reaction mechanism that we propose for 1.4Fe/7.0W/
Zr catalyst is quite different from that postulated in the
literature for Fe-ZSM-5 zeolite [53–57]. By the use of Fe-
ZSM-5 material NH3 is protonated by the Brønsted acid sites
of the zeolite to form NH4+. Then, two ammonium ions react
with one NO2 molecule that results in the production of
NO2(NH4+)2 species. This ionic intermediate is subse-
quently converted with another NO leading to the formation
Fig. 12. Development of the DRIFTS band of the d(NH) vibration of imide,
located at 1440 cm1, in the presence of 20 vol.% NO (^), 20 vol.% O2
(&) and 500 ppm NO2/20 vol.% O2 (~). In these gas mixtures, N2 is used Fig. 13. Postulated reaction mechanism for SCR on the 1.4Fe/7.0W/Zr
as balance and T = 200 8C. catalyst.
114 N. Apostolescu et al. / Applied Catalysis B: Environmental 62 (2006) 104–114
of N2 and H2O. In the Fe-ZSM-5 catalyzed SCR reaction the [16] R.Q. Long, R.T. Yang, Appl. Catal. B 33 (2001) 97.
role of the Fe sites lies in the acceleration of the NO [17] R.Q. Long, R.T. Yang, J. Catal. 207 (2002) 274.
[18] G. Qi, R.T. Yang, R. Chang, Catal. Lett. 87 (2003) 67.
oxidation to yield the needed NO2. For comparison, if a NO/ [19] A.V. Salker, W. Weisweiler, Appl. Catal. A 2 (2000) 221.
NO2 mixture is used, i.e. NO2 is already present in the feed, [20] K. Krishna, G.B.F. Seijger, C.M. van den Bleek, M. Makkee, G. Mul,
the Fe-ZSM-5 and H-ZSM-5 zeolites reveal the same H.P.A. Calis, Catal. Lett. 86 (2003) 121.
activity for the reaction of NH4+. The mechanism of SCR on [21] A.Z. Ma, W. Grünert, Chem. Commun. 1 (1999) 71.
[22] M. Schwidder, F. Heinrich, M. Kumar, A. Brückner, W. Grünert, Stud.
Fe-ZSM-5 shows that the reaction is profoundly effected by
Surf. Sci. Catal. 154 (2004) 2484.
the zeolitic Brønsted acid sites, whereas the mechanism of [23] K.M. Santhosh, M. Schwidder, W. Grünert, A. Brückner, J. Catal. 227
1.4Fe/7.0W/Zr is based upon the presence of the Lewis acid (2004) 384.
Fe3+ sites. [24] M. Schwidder, K.M. Santhosh, A. Brückner, W. Grünert, Chem.
Commun. 6 (2005) 805.
[25] M. Schwidder, K.M. Santhosh, K. Klementiev, M. Marga, A. Brück-
ner, W. Grünert, J. Catal. 231 (2005) 314.
9. Conclusion [26] R.Q. Long, R.T. Yang, J. Catal. 188 (1999) 332.
[27] R.Q. Long, R.T. Yang, Catal. Lett. 74 (2001) 201.
The catalyst screening proves that ZrO2 is the most [28] H.Y. Chen, W.M.H. Sachtler, Catal. Today 42 (1998) 73.
effective support for Fe2O3 among the tested materials. By [29] Q. Sun, Z.X. Gao, H.Y. Chen, W.M.H. Sachtler, J. Catal. 201 (2001)
modifying this system with tungsten the SCR activity as well 89.
[30] H.Y. Chen, Q. Sun, B. Wen, Y.H. Yeom, E. Weitz, W.M.H. Sachtler,
as thermal stability is clearly increased. The optimum Catal. Today 96 (2004) 1.
catalyst contains 1.4 mol% Fe2O3 and 7.0 mol% WO3 and [31] Q. Sun, Z.X. Gao, B. Wen, W.M.H. Sachtler, Catal. Lett. 78 (2002) 1.
exhibits a high SCR activity with selective production of [32] M. Pfeifer, B. van Setten, P. Spurk, Y. Demel, T. Kuhl, J. Gieshoff,
N2. Furthermore, the catalyst reveals pronounced thermal E. Lox, T. Kreuzer, DE 103 08 288 A1.
[33] A. Schäfer-Sindlinger, M. Pfeifer, P. Spurk, Y. Demel, T. Kreuzer,
stability. Detailed studies of the SCR reaction suggest an
E. Lox, DE 100 20 100 A1.
Eley Rideal type mechanism involving a redox cycle in [34] D.R. Lide (Ed.), Handbook of Chemistry and Physics, CRC Press,
which the Fe sites are periodically consumed and regene- London, 1994.
rated. We assume that amide species is the essential [35] K. Yamazaki, N. Takahashi, H. Shinjoh, M. Sugiura, Appl. Catal. B 53
component in the formation of N2. (2004) 1.
[36] S. Kureti, W. Weisweiler, K. Hizbullah, Appl. Catal. B 43 (2003) 281.
[37] D. Ciuparu, A. Ensuque, G. Shafeev, F. Bozon-Verduraz, J. Mater. Sci.
Lett. 19 (2000) 931.
References [38] M. Hino, K. Arata, Chem. Lett. 9 (1979) 477.
[39] A. Kayo, T. Yamaguchi, K. Tanabe, J. Catal. 83 (1983) 99.
[1] M. Guyon, P. Blanche, C. Bert, L. Philippe, I. Messaoudi, SAE Paper [40] N.W. Hurst, S.J. Gentry, B.D. McNicool, Catal. Rev. 24 (1982)
2000-01-2910. 233.
[2] G. Busca, L. Lietti, G. Ramis, F. Berti, Appl. Catal. B 18 (1998) 1. [41] W. Weisweiler, C. Walz, Chem. Ing. Tech. 1–2 (2002) 117.
[3] W. Weisweiler, B. Retzlaff, B. Hochstein, Staub-Reinhalt. Luft 3 [42] M. Crocoll, S. Kureti, W. Weisweiler, J. Catal. 229 (2005) 480.
(1988) 119. [43] E.E. Unmuth, L.H. Schwartz, J.B. Butt, J. Catal. 61 (1980) 242.
[4] G. Ramis, L. Yi, M. Turco, E. Kotur, R.J. Willey, J. Catal. 157 (1995) [44] O.J. Wimmers, P. Arnoldy, J.A. Moulijn, J. Phys. Chem. 90 (1986)
523. 1331.
[5] W. Weisweiler, E. Mallonn, O. Görke, Chem. Ing. Tech. 1–2 (2003) [45] U. Colombo, F. Gazzarrini, G. Lanzavecchia, Mater. Sci. Eng. 2 (1967)
72. 125.
[6] L. Chmielarz, P. Kustrowski, M. Zbroja, W. Lasocha, R. Dziembaj, [46] R. Brown, M.E. Cooper, D.A. Whan, Appl. Catal. 3 (1992) 177.
Catal. Today 90 (2004) 43. [47] A.V. Kucherov, C.N. Montreuil, T.N. Kucherova, M. Shelef, Catal.
[7] W. Weisweiler, B. Maurer, in: W. Essers (Ed.), Dieselmotorentechnik Lett. 56 (1998) 173.
1998, Expert-Verlag, Renningen, 1998, p. 92. [48] F. Heinrich, C. Schmidt, E. Löffler, M. Menzel, W. Grünert, J. Catal.
[8] H. Bosch, F. Janssen, Catal. Today 2 (1988) 369. 212 (2002) 157.
[9] A. Kato, S. Matsuda, F. Nakajima, M. Imanari, Y. Watanabe, J. Phys. [49] G. Busca, Langmuir 2 (1986) 577.
Chem. 85 (1981) 1710. [50] S. Kagami, T. Onishi, K. Tamaru, J. Chem. Soc. Faraday Trans. 1 (80)
[10] A. Palmqvist, E. Jobson, L. Andersson, R. Granbro, M. Wendin, L. (1984) 29.
Megas, P. Nisius, A. Wiartalla, G. Lepperhoff, P. Blakeman, T. [51] N.Y. Topsøe, H. Topsøe, J.A. Dumesic, J. Catal. 151 (1995) 226.
Ilkenhans, D. Webster, G. Haeffler, P. Voorde, P. Schmidt-Zhang, [52] J.B. Peri, J. Phys. Chem. 69 (1965) 231.
U. Guth, M. Wallin, SAE Paper 2004-01-1294. [53] R.Q. Long, R.T. Yang, J. Catal. 194 (2000) 80.
[11] S. Wu, K. Nobe, Ind. Eng. Chem. Prod. Res. Dev. 16 (1977) 136. [54] R.Q. Long, R.T. Yang, J. Catal. 198 (2001) 20.
[12] R.J. Willey, H. Lai, J.B. Peri, J. Catal. 130 (1988) 319. [55] R.Q. Long, R.T. Yang, J. Catal. 207 (2002) 224.
[13] R.J. Willey, E. Kotur, J. Kehoe, G. Busca, in: Y.A. Attia (Ed.), Proc. [56] H.Y. Huang, R.Q. Long, R.T. Yang, Appl. Catal. A 235 (2002)
Int. Symp. Adv. Sol-Gel Process. Appl., 1994, p. 351. 241.
[14] R.Q. Long, R.T. Yang, J. Catal. 186 (1999) 254. [57] V. Sanchez-Escribano, T. Montanari, G. Busca, Appl. Catal. B 58
[15] R.Q. Long, R.T. Yang, Appl. Catal. B 27 (2000) 87. (2005) 19.