Low-temperature NOx reduction with ethanol over Ag/Y: A comparison with Ag/γ-Al2O3 and BaNa/Y

Article abstract of DOI:10.1016/j.jcat.2006.12.013

A multistep mechanism has been elucidated for the reduction of NOx in the presence of ethanol over silver-exchanged zeolite Y (Ag/Y). Ethanol reacts with O₂ and/or NO₂ to form acetaldehyde at temperatures as low as 200 °C. Surface acetate ions, formed from the oxidation of acetaldehyde, react with NO₂ to yield nitromethane, a critical intermediate in subsequent deNOx chemistry. CN^-, NC^-, and NCO^- are intermediates likely bound to silver ions. Both CN^- and NC^- are stable toward reaction under experimental conditions. A significant difference exists between the catalytic activities of Ag/Y and Ag/γ-Al₂O₃; oxidation of ethanol to acetate at low temperature is significantly faster over Ag/Y than over Ag/γ-Al₂O₃, and both NO₂ and O₂ are effective oxidants over Ag/Y. With Ag/Y, pretreatment with either O₂ or H₂ does not affect the yield of N₂, which approaches 60% and remains constant for at least 5 h, making this catalyst promising for NOx reduction.

Full text of DOI:10.1016/j.jcat.2006.12.013

Journal of Catalysis 246 (2007) 413–427
www.elsevier.com/locate/jcat
Low-temperature NOx reduction with ethanol over Ag/Y:
A comparison with Ag/γ -Al O and BaNa/Y
2
3
∗
Young Hoon Yeom, Meijun Li, Wolfgang M.H. Sachtler, Eric Weitz
Department of Chemistry and Institute for Catalysis in Energy Processes, Northwestern University, Evanston, IL 60208, USA
Received 28 September 2006; revised 12 December 2006; accepted 17 December 2006
Abstract
A multistep mechanism has been elucidated for the reduction of NOx in the presence of ethanol over silver-exchanged zeolite Y (Ag/Y).
◦
Ethanol reacts with O and/or NO to form acetaldehyde at temperatures as low as 200 C. Surface acetate ions, formed from the oxidation
2
2
−
−
−
of acetaldehyde, react with NO to yield nitromethane, a critical intermediate in subsequent deNOx chemistry. CN , NC , and NCO are
intermediates likely bound to silver ions. Both CN and NC are stable toward reaction under experimental conditions. A significant difference
2
−
−
exists between the catalytic activities of Ag/Y and Ag/γ -Al O ; oxidation of ethanol to acetate at low temperature is significantly faster over
2
3
Ag/Y than over Ag/γ -Al O , and both NO and O are effective oxidants over Ag/Y. With Ag/Y, pretreatment with either O or H does not
2
3
2
2
2
2
affect the yield of N , which approaches 60% and remains constant for at least 5 h, making this catalyst promising for NOx reduction.
2
©
2007 Elsevier Inc. All rights reserved.
Keywords: deNOx; Silver; Zeolite Y; Gamma alumina; Kinetics; Ethanol; Acetate; Acetaldehyde; FTIR; SCR
1
. Introduction
emulsifying agent [6]. It then can be distilled from this mixture
via “on-board” heating. However, the reported working temper-
ature of the Ag/γ -Al2O3 catalyst is rather high compared with
the typical exhaust gas temperature of light-duty diesel engines
Although diesel engines are intrinsically more fuel-efficient
than gasoline-powered engines, their exhaust gases are more
heavily laden with nitrogen oxides and particulates than those
of a comparable gasoline engine whose exhaust has been treated
with a standard three-way catalyst [1]. Thus, a societal benefit
from the fuel efficiency of diesel engines requires efficient re-
moval of nitrogen oxides (NOx) from diesel exhaust. A goal in
◦
of 200–250 C [7].
◦
In previous work, we found that at 200 C, the yield of N2
from NOx reduction by ethanol was ∼55% on Ag/Y, com-
pared with only ∼20% on Ag/γ -Al2O3 [8]. More recent studies
have explored the reaction mechanism for the SCR of NOx
with ethanol over Ag/γ -Al2O3 [9–12]. Intermediates including
NOx removal is to reduce NOx to N with an added reductant
and a highly active and selective catalyst.
2
−
−
−
CH3COO [9], CH2=CHO [10], NO3 [11], and NCO [11,
1
2] have been identified by FTIR spectroscopy. However, re-
Ethanol has been successfully used in the selective cat-
markably little definitive mechanistic data on the SCR of NOx
on Ag/Y are available. It would be particularly interesting to de-
termine how the nature of the intermediates and differences in
the reaction mechanism and/or kinetics relate to the differences
in catalytic activity of the two supported silver catalysts of in-
terest in this work: Ag/γ -Al2O3 and Ag/Y. Understanding the
alytic reduction (SCR) of NOx on Ag/γ -Al O , an alumina-
supported silver catalyst [2,3]. This system has demonstrated
good resistance to both water and sulfur [4], and NOx con-
version can exceed 80% in the temperature range of 350–
2
3
◦
4
90 C [5]. Ethanol is a particularly attractive reductant for
NOx removal from diesel exhaust because it is environmentally
relatively benign and can be blended with diesel fuel using an
◦
cause of the superior activity of Ag/Y near 200 C is important,
because this low-temperature activity is highly desirable for its
removing NOx from diesel exhaust.
*
The present study used FTIR to identify reaction interme-
diates in deNOx chemistry occurring over Ag/Y with ethanol
Corresponding author.
E-mail address: weitz@northwestern.edu (E. Weitz).
0021-9517/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.12.013

414
Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
as an added reductant. These findings form the basis for our
discussion of the differences in catalytic activity between Ag/γ -
Al2O3 and Ag/Y. This comparison is based on both experiments
in the present work and previously published work on Ag/γ -
Catalytic tests were carried out in a fixed-bed microflow re-
actor. In brief, 0.2 g of the powdered Ag/Y catalyst was packed
into the quartz reactor. The feed gas composition was regu-
lated by mass-flow controllers (URS-100, UNIT Instruments).
Two separate experiments were performed with the flow reac-
tor. First, Ag/Y was exposed to a flowing mixture of He and H2
Al O [9]. We also have considered the effect of the oxidation
2
3
state of Ag on the catalytic activity of Ag/Y.
. Experimental
Alumina-supported silver catalysts (Ag/γ -Al O , contain-
−
1
◦
in a 1:1 ratio at a flow rate 40 ml min at 25 C. The temper-
◦
ature of the fresh Ag/Y sample was then increased to 300 C
2
◦
−1
◦
with a temperature ramp of 5 C min . After 3 h at 300 C,
the Ag/Y was cooled to 200 C and purged with He for 20 min.
◦
2
3
Ag/Y was then exposed to a flow of 336 ppm EtOH, 500 ppm
ing 4 wt% of Ag) were prepared by impregnating γ -Al2O3
−1
NO , 7% O , and 1% water at a flow rate of 200 ml min
,
2
2
(
2
Nanostructures and Amorphous Materials Inc., 99%, 11 nm,
50 m g ) with an aqueous solution of 0.05 M silver ni-
2
−1
with He making up the balance of the gas flow. Water va-
por was introduced into the reaction system by bubbling He
trate. The mixture of Ag/γ -Al2O3 and silver nitrate solution
was stirred for 1 h and kept at 298 K until the water evapo-
rated. The dried sample was then calcined at 773 K for 5 h in
through a H O saturator. The composition of the effluent was
2
analyzed on-line with a gas chromatograph (GC) equipped with
a thermal conductivity detector (5A column for N2). Before an-
alyzing the effluent, the mixture was allowed to flow through
the system for at least 30 min. The conversion of NO2 was de-
termined from the formation of N2 [2]. The Ag/Y catalyst was
an O flow. Ag/Y samples were obtained by threefold wet-ion
2
exchange of Na/Y (Si/Al = 2.5, Aldrich) at ambient tempera-
ture with a 0.1 M AgNO3 solution. Before each exchange, the
Ag/Y slurry was stirred for 3 h, followed by vacuum filtering,
thorough washing with doubly deionized H2O, and drying in
◦
calcined at 500 C for 5 h in a mixture of O2 and He (volumet-
−
1
◦
ric ratio = 1/3) at a flow rate of 40 ml min . The Ag/Y was
air. This material was then calcined for 5 h at 500 C in an O2
◦
+
then cooled to 200 C and exposed to a flow of the same com-
flow. ICP analysis revealed that 99.6% of the Na ions were
replaced by Ag ions.
+
position as described above.
Isocyanic acid (HNCO) was prepared by mixing stearic acid
In situ FTIR spectra were recorded in transmission mode us-
ing a homemade IR cell and a Bio-Rad Excalibur FTS-3000
infrared spectrometer equipped with a HgCdTe (MCT) detector.
The infrared cell, described in detail previously [13], consists
(
Aldrich, 98%) with sodium cyanate (Aldrich, 96%) and heat-
◦
ing to the melting point of the stearic acid (67–72 C) [14].
Authentic HCN was prepared by a similar method. Stearic
acid was mixed with potassium cyanide (Sigma-Aldrich, 97%).
Ethanol (Aldrich, 99.8%), acetic acid (Supelco, 99%), acetic
acid-2- C (Cambridge Isotope, 99% C), nitrogen dioxide
(
of a stainless-steel cube with two CaF windows that can be
2
differentially pumped. The tungsten grid that supports the cat-
alysts can be resistively heated. Its temperature is measured by
a Chromel–Alumel thermocouple spot-welded to the center of
the grid.
13
13
15
Matheson Tri-gas, 99%), nitric oxide- N (Cambridge Isotope,
15
9
8% N), and nitromethane (Aldrich, 99%) were used as re-
ceived. Caution: Care must be taken in handling and/or synthe-
sizing the toxic gases used in this study.
Typically, a small brush was used to “paint” ∼10 mg of sam-
ple, suspended in a water slurry, onto the 1.5 × 1.5 cm tungsten
grid, with the temperature held at 353 K. Before each exper-
−
6
3. Results
iment, catalysts were heated in vacuum (2 × 10 Torr) for
3
h at 703 K and then cooled to the desired temperature. After
this treatment, and before the catalyst was exposed to reactants,
a spectrum of the catalyst was recorded and used as the “back-
ground.”
Elucidation of the reaction mechanism for NOx reduction
with acetaldehyde over BaNa/Y and with ethanol over Ag/γ -
Al O has provided a very detailed picture of the chemistry
2
3
Unless stated otherwise, each reported spectrum is the result
of averaging 70 scans at 4-cm resolution under static condi-
tions. A DTGS detector and KBr windows were used in place
involved in this process. The temperature dependence of the
rates of reaction of the intermediates identified in these stud-
ies provides insight into why BaNa/Y affects NOx reduction at
−
1
of a MCT detector and CaF cell windows when it was neces-
a much lower temperature than Ag/γ -Al O [9].
2 3
2
−
1
sary to obtain spectra of absorptions below ∼700 cm . Time-
resolved spectra were obtained using the “rapid-scan” mode of
the FTIR.
Identifying the intermediates is a crucial step in determin-
ing an overall reaction mechanism. FTIR spectroscopy was the
principal tool used to determine the intermediates in deNOx
reactions occurring on Ag/γ -Al O and BaNa/Y [9,13] and
Unless otherwise stated, gas mixtures of NO + O (and,
2
2
3
in some cases, other gases) were prepared at room tempera-
ture and allowed to mix and react for a suitable time (typically
has been used in this study of Ag/Y. The use of isotopically
labeled compounds aids in assigning absorptions to specific in-
termediates. Experiments also have been performed in which
the Ag/Y sample was exposed to H2 before contact with the
NOx mixture. The objective of these experiments was to deter-
mine whether reduction of Ag, which is expected on exposure
of Ag/Y to H2, affects the performance of the Ag/Y catalyst.
overnight) such that the NO, NO , and O in the mixture were
2
2
in equilibrium. This means that in the mixtures used in this
study, all of the NO was converted to NO . For such mixtures, it
2
was verified that no other (or negligible other) reaction occurred
in the gas phase during mixing.

Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
415
◦
Fig. 1. Time resolved spectra recorded after (a) γ -Ag/Al O or (b) Ag/Y was exposed to 100 Torr of a mixture of 3.8 Torr of EtOH + 94.4 Torr of O2 at 200 C.
2
3
The exposure time is indicated for each spectrum.
3
.1. Oxidation of ethanol by O2 on Ag/γ -Al2O3 and Ag/Y
On exposing Ag/γ -Al2O3 or Ag/Y to a mixture of etha-
Comparing the rate of decrease of ethanol in Fig. 1, traces (a)
and (b), demonstrates a much more rapid loss in (b). Thus,
◦
ethanol is readily oxidized by O at 200 C on Ag/Y. Both the
2
2
−1
relatively higher surface area (∼900 vs ∼250 m g ) and the
nol + O2, oxygenated intermediates are observed. In Fig. 1,
greater quantity of silver (∼26 vs ∼4 wt%) for Ag/Y versus
trace (a), absorption bands at 1391 (CH3 bending), 2901 (C–H
−
1
2 3
Ag/γ -Al O could contribute to the high rate of oxidation. De-
stretch), 2978 (C–H stretch), and 3673 cm (O–H stretch) are
assigned to ethanol, based on a comparison with the spectrum
of authentic ethanol. The absorption at 1745 cm is due to
gas-phase acetaldehyde, and the band at 1580 cm is due to
surface acetate [9]. The absorption at 1451 cm results from
the overlap of the two bands: a bending mode ethanol and the
tails relevant to the rate of ethanol oxidation under different
−
1
conditions on Ag/γ -Al O have been provided in our previous
2
3
−
1
work on this system [9].
−
1
3.2. N-containing intermediates in the reaction of NO2 and
ethanol over Ag/γ -Al2O3
−
symmetric CO stretching vibration of surface acetate. The sur-
2
face acetate is a result of the oxidation of acetaldehyde. The IR
bands of acetaldehyde and surface acetate gradually increase in
intensity with reaction time.
In trace (b), the absorptions at 2901, 2978, and 3673 cm
are again due to ethanol. In the top trace, the band at 1699 cm
After exposure of Ag/γ -Al O to a mixture of C H OH +
2
3
2
5
−
1
NO + O , an intermediate is observed at 2258 cm (Fig. 2). In
2
−
−
1
1
−1
Fig. 2, trace (a), the gas-phase feature centered at 2348 cm is
due to CO , which is present as a result of incomplete purging
2
first forms and then gradually decreases in intensity, whereas
of the sample compartment of the spectrometer. The absorption
at 2258 cm , which does not disappear on evacuation of the
−
1
−1
the bands at 1393, 1446, 1474, 1619, and 1656 cm gradually
−
1
increase in intensity. The 1699 cm absorption is assigned to
acetaldehyde because the same band is observed when Ag/Y
is exposed to authentic acetaldehyde. A similar band is also
observed when BaNa/Y is exposed to acetaldehyde [13]. The
cell, is due to an adsorbed species. Data for this intermediate
◦
were typically acquired at 150 C, because the intensity of the
−
1
◦
2258 cm absorption is lower at 200 C.
13
15
C-labeled EtOH and NO are used to provide additional
2
−
1
absorption of gas-phase acetaldehyde, centered at 1745 cm
,
data for an assignment of this intermediate. In traces (b) and (c),
−
1
−1
13
seen in the bottom trace is weak compared with the correspond-
ing band in trace (a), because most of the acetaldehyde has
the absorption at 2258 cm is shifted to 2197 cm on
C
−
1
15
isotopic substitution and to 2244 cm on N isotopic substi-
tution. As indicated in Table 1, the asymmetric stretching vibra-
−
1
been adsorbed. The bands at 1393, 1446, 1474, and 1656 cm
correspond to absorptions observed when Ag/Y is exposed to
−
1
tion of isocyanate groups (N=C=O) is shifted by ∼8–18 cm
−
1
15
14
−1
authentic crotonaldehyde. The band at 1619 cm is best as-
signed to the overlap of an absorption of crotonaldehyde and
surface acetate. These data clearly indicate that some surface
acetate is formed by oxidation of acetaldehyde before all of the
acetaldehyde reacts to form crotonaldehyde.
due to substitution of N for N and by ∼56–62 cm as a re-
sult of 13C labeling. In addition, an isotopic shift of cyanate
−
1
15
(M–OCN, M = K, or Na) of 18 cm on N substitution
13
12
is seen. However, after substitution of C for C, adsorbed
cyanates usually absorb at significantly lower wavenumbers

416
Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
sorbing at 2258 cm 1. Therefore, the absorption in trace (a) can
−
(
see Table 1) and have different shifts than those seen here. As
−
shown in Table 1, fulminate (another isomer of NCO) absorbs
be confidently assigned to NCO , adsorbed on Ag/γ -Al2O3.
−1
14
15
at 2202 cm in HgONC with an isotopic shift ( N → N)
Because this IR band is virtually identical with respect to its
shape and position on bare γ -Al2O3, the absorbing species is
most likely Al–NCO.
−1
of 37 cm . Thus, neither fulminate nor cyanate is compatible
with either the position or the isotopic shift of the species ab-
As is evident from Fig. 2, trace (d), one band due to an ad-
−
1
sorbed species is observed at 2258 cm when Ag/γ −Al O is
2
3
exposed to HNCO. Gas-phase HCN is also observed. Judging
from the positions of the CN and NC absorptions given in Ta-
ble 1, neither surface NC nor CN is observed under our experi-
mental conditions; these species would be expected to absorb at
considerably lower wavenumbers. Interestingly, Bion et al. [11]
−1
observed four bands at 2127, 2155, 2255, and 2228 cm on ex-
posing Ag/Al2O3 to a mixture of EtOH + NO + O2. Based on
−
1
our data, the 2255 cm absorption observed by Bion is com-
−
patible with adsorbed NCO . Their observation of additional
absorptions is likely due to differences in their catalysts and ex-
perimental conditions. Although the positions and the isotopic
shift of the NCO absorption do not allow us to uniquely deter-
mine whether this species is an anion or neutral, the fact that it
is strongly bound to Ag/γ -Al O suggests that it is an anion,
2
3
and thus we consider it so in subsequent usage.
−
12
−
−1
13
−1
Both N CO at 2258 cm and N CO at 2197 cm are
13
observed when 2- C EtOH is used as the reductant [Fig. 2,
−
1
trace (b)]. Clearly, the band at 2197 cm in trace (b), due
−
13
−1
Fig. 2. (a) Spectra recorded after γ -Ag/Al O was exposed to 100 Torr of
2
3
to N CO , is much stronger than the band at 2258 cm
a mixture composed of 3.8 Torr EtOH + 3.8 Torr NO + 94.4 Torr O2 for
12
−
2
due to N CO . This intensity difference demonstrates that
most of the carbon atoms in the isocyanate moiety come from
the methyl carbon of EtOH. In previous work, a pathway was
discussed for the formation of isocyanate from acetaldehyde,
◦
1
min at 150 C. Before spectra were taken, gas-phase species were pumped
out. (b) EtOH was replaced by 2- C EtOH, and (c) NO was replaced by
NO . (d) Spectrum was taken after γ -Ag/Al O was exposed to 0.3 Torr of
13
2
15
2
2
3
◦
authentic HNCO at 200 C.
Table 1
15
14
13
12
Isotopic shift on substituting N for N and C for
C
14
12
15
13
14
N → 15
12
C → 13
N and
C
N
C
N
C
Ref.
ꢀν
ꢀν
HNCO
2266
2258
2315
2261
2258
2244
2305
2249
2112
2152
2165
2058
2188
2064
2206
2197
2255
2199
2074
8
60
61
60
62
56
Ag/γ -Al O + NCO
14
10
12
18
18
37
32
30
36
This work
[32]
[32]
2
3
SiNCO
Pt/Al O + NCO
2
3
2130
[32]
MOCN (M = Na, K)
HgONC
HCN
SiCN
SiNC
2170
2202
2090
2218
2100
2144
2051
2167
2061
2097
2011
51
39
47
40
[33]
[33]
[22]
[22]
AlCN
AlNC
CN (or NC) vibrations in silver compounds
−
1
CN or NC stretch (cm
)
Ref.
c
AgCN monomer
AgNC monomer
AgCN crystal
2171
[34]
[34]
[35]
[36]
[12]
c
2075
2164
a
Ag–CN in Ag/Y
2166
2127 (AgCN)
165 (AlCN)
155 (AlCN)
b
Ag/Al O + CN
2
3
2
2
a
b
c
+
Ag ion exchanged zeolite Y.
CN was detected from the reaction EtOH + NO + O on Ag/Al O .
2
2 3
Calculated.

Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
417
which is the product of ethanol oxidation [13]; a schematic rep-
resentation of that pathway is
3.3. N-containing products of the reaction of NO2 and ethanol
over Ag/Y
−
1
3
+
13
2
Ethyl nitrite, which is a precursor for acetaldehyde [9], is
formed from the reaction of EtOH and NO2:
CH3CHO + NO2 →→ H + N CO + CO .
It is well known that alumina catalyzes ethanol dehydration [12,
2
NO2 + EtOH → C2H5ONO + HNO3,
1
5–17]:
C2H5ONO → C2H5O + NO.
CH3CH2OH → CH2=CH2 + H2O.
The major difference between Ag/γ -Al O and Ag/Y is that
2
3
acetaldehyde is also formed from the direct oxidation of EtOH
To determine whether ethene is formed under our experimen-
◦
by O over Ag/Y, at 200 C, whereas on Ag/γ -Al O , most of
2
2
3
13
tal conditions, Ag/γ -Al2O3 was exposed to a mixture of 2- C
EtOH+NO2 +O2. The results, shown in Fig. 3, indicate a band
◦
the acetaldehyde is formed from ethyl nitrite at 200 C. Ac-
etaldehyde can then react with NO2 to produce nitromethane. It
can also be oxidized to surface acetate, which can subsequently
−1
at 949 cm that is assigned to the CH2 wagging vibration of
ethene. This assignment is based on literature data [18] and
a comparison with the authentic gas-phase spectrum of ethene
obtained in our IR cell. Ethene can then react with NO2 to pro-
duce N-containing intermediates.
react with NO to produce the aci-anion of nitromethane.
2
Fig. 4 displays changes in the spectra recorded after expos-
ing Ag/Y to a mixture of EtOH + NO + O . In trace (a), the
2
2
−1
absorptions at 1393, 2906, and 2980 cm are due to ethanol.
13
12
Gas-phase H CN and H CN are observed at 707 and
19 cm , respectively. Interestingly, there is a disparity in
intensity, which we address in Section 4. Broad features are
present at 680–720 cm and 930–950 cm in the spectrum
obtained at 21 s. These are due to ethyl nitrite formed from
the reaction of ethanol and NO2 [9]. This experiment was con-
ducted at 250 C, because the ethene concentration was too low
to be detected below this temperature. Even though gas-phase
hydrogen cyanide is observed, surface cyanide is not detected
−1
The absorption centered at 1616 cm is due to gas-phase
−1
7
−1
NO2, and the absorption at 1699 cm is due to adsorbed ac-
−1
etaldehyde. The absorption at 1208 cm is likely a zeolite
−
1
−1
framework vibration. Both the NO2 and the ethanol absorp-
tion rapidly decrease in intensity due to reaction. The 2000–
−
1
2
500 cm region is enlarged in the inset. In trace (a) in the
◦
−1
insert, absorptions are present at 2090 and 2184 cm , and
gas-phase CO2 is seen at 2348 cm . The intensity of the ab-
sorption band at 2184 cm gradually decreases with reaction
time and becomes a shoulder of a new absorption that grows in
at 2168 cm . Although the intensity of the band at 2090 cm
−
1
−
1
◦
on Ag/γ -Al2O3 under these conditions at 200 or 250 C. Pre-
sumably, the steady-state concentration of surface cyanide is
too low to allow detection, due to its rapid reaction with wa-
−
1
−1
increases slightly with time, its position does not shift. Neither
−
1
−1
ter (vida infra). Two bands at 668 and 649 cm are due to the
of the absorptions at 2168 and 2090 cm decreases signifi-
bending modes of 12CO2 and CO2, respectively [18]. There
13
cantly on evacuation, demonstrating strong adsorption of these
species on the Ag/Y surface [see trace (h)]. The rapid disap-
12
13
is also a disparity in the intensities of CO2 and CO2, which
was discussed previously [13].
−
1
pearance of the band at 2184 cm with time indicates that
−1
this species is reactive. The absorption at 2184 cm is due
to adsorbed NCO because its position [see trace (e) in Fig. 5]
is close to that of NCO observed after exposing Ag/Y to au-
thentic HNCO. Gas-phase HNCO, at 2268 cm , is first seen
in trace (b), and its intensity gradually decreases as the reac-
tion progresses. Trace (h) shows an additional weak feature at
236 cm ; this band is assigned in Figs. 5 and 6. We cannot
definitively determine whether the absorption at 2184 cm is
due to NCO or NCO ; however, because it is strongly adsorbed
on Ag/Y, it is likely NCO . The absorption band at 3632 cm
−
1
−
1
2
−1
−
−
−1
is assigned to the OH stretching vibration. The same absorp-
tion is observed whenever Ag/Y is exposed to authentic HCN.
Therefore, this OH absorption is probably the result of disso-
+
−
ciation of HCN: HCN → H + CN . The proton thus formed
can attach to a framework oxygen ion.
The experimental data displayed in Fig. 5 provide additional
information for the assignment of the three absorptions at 2168,
−
1
2
090, and 2236 cm . These absorptions are stable on evacu-
ation; they form when Ag/Y is exposed to a mixture of EtOH
−1
and NO . The intensity of the absorption band at 2236 cm
2
is dependent on both reaction time and the EtOH/NO2 ratio.
The band intensity increases with time when the EtOH/NO2
ratio is 1 but is hardly observable when the EtOH/NO2 ratio
Fig. 3. Gas-phase spectra were recorded at each time frame after Ag/Y was
13
exposed to 70 Torr of a mixture composed of 3.8 Torr 2- C EtOH + 3.8 Torr
◦
NO + 60 Torr of O at 250 C.
2
2

418
Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
Fig. 4. Time resolved spectra that was recorded after Ag/Y was exposed to 100 Torr of a mixture composed of 3.8 Torr EtOH + 3.8 Torr NO + 92.4 Torr of O at
2
2
◦
00 C, (a)–(g) each spectrum was taken at each indicated time, (h) spectrum was subsequently taken after gas-phase species were pumped off. The insert is a blow
up of the ∼2000 to ∼2450 cm region.
2
−
1
The absorptions at 2168 and 2090 cm 1 match well with
those that appear when Ag/Y is exposed to authentic HCN;
compare traces (a) and (d). These two absorptions, as well as
−
−
1
the absorption at 2236 cm , shift by the same amount [com-
15
13
pare traces (a), (b), and (c)] when either NO2 or 2- C EtOH
is a reactant. This implies that these three species contain a sim-
ilar moiety. Based on comparisons of band positions and the
isotopic shifts specified in Table 2, the band at 2168 cm is
assigned to CN (cyanide ion), and the band at 2090 cm is
−
1
−
−
−1
−1
assigned to NC (isocyanide ion). The band at 2236 cm does
not match any of the bands in traces (b) to (e).
To assign the band at 2236 cm 1, we performed another
series of experiments; the results are shown in Fig. 6. When
acetaldehyde [trace (b)] and crotonaldehyde [trace (c)] are used
as reductants, three bands are observed at positions matching
those of the bands in trace (a). Interestingly, when acetic acid
−
−1
is used as the reductant, the absorption at 2236 cm is not
observed even though the other two absorption bands are ob-
served.
Fig. 5. Spectra were taken after Ag/Y samples were exposed to a mixtures
◦
for 10 min at 200 C. Before spectra were taken, each sample pumped on to
remove gas-phase species. The mixtures were (a) 3.8 Torr EtOH + 3.8 Torr
We also exposed Ag/Y to acetonitrile, propanenitrile, pyru-
vonitrile, and acrylonitrile. The resulting spectra are displayed
in traces (e) to (h) of Fig. 6. The only absorption in traces (b)
15
NO + 92.4 Torr of O , (b) 3.8 Torr EtOH + 3.8 Torr NO + 92.4 Torr of
2
2
2
1
3
O , and (c) 3.8 Torr 2- C EtOH + 3.8 Torr NO + 92.4 Torr of O . Spectra
2
2
2
taken after Ag–Y was exposed to authentic (d) 0.15 Torr HCN and (e) 0.15 Torr
HNCO.
−
1
to (h) that matches the 2236 cm band is that in trace (g). On
exposing Ag/Y to pyruvonitrile [trace (g)], absorptions appear
−
1
is 1/3. Although we have not determined the EtOH/NO2 ratio
that optimized the intensity of this absorption, this band be-
comes weaker when the EtOH/NO2 ratio is <1.0.
at 2063, 2112, 2168, and 2236 cm . This spectrum was ob-
◦
tained after exposing Ag/Y to pyruvonitrile at 200 C and then
cooling to 100 C. The absorption at 2236 cm was not ob-
◦
−1

Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
419
Fig. 6. Spectra were taken after each Ag/Y was exposed to a mixture of
a) 3.8 Torr EtOH + 3.8 Torr NO + 92.4 Torr of O , (b) 3.8 Torr acetalde-
(
2
2
hyde + 3.8 Torr NO + 92.4 Torr of O , (c) 3.8 Torr crotonaldehyde + 3.8 Torr
2
2
NO + 92.4 Torr of O , and (d) 3.8 Torr AcOH + 3.8 Torr NO + 92.4 Torr of
2
2
2
◦
O for 15 min at 200 C. Before taking each spectrum, gas-phase species were
2
pumped off. Ag/Y was exposed to (e) 0.4 Torr of acetonitrile, (f) 0.4 Torr propa-
◦
nenitrile, and (h) 0.26 Torr of acrylonitrile at 200 C. (g) Ag/Y was first exposed
◦
◦
to 0.4 Torr of pyruvonitrile at 200 C and subsequently cooled to 100 C.
Table 2
Assignments of bands in Fig. 5
Fig. 7. Time resolved spectra after Ag/Y was exposed (a) to 1 Torr of ni-
−
1
14
N → 15
12
C → 13
Traces
cm
Assignments
N
C
tromethane and (b) to 10 Torr of a mixture containing a nitromethane/NO
ratio = 1/4. Temperature was 200 C.
2
ꢀν
ꢀν
◦
(a)
2090
AgNC, (NC)
stretch
stretch
stretch
2
168
236
AgCN, (CN)
a
−1
the 2236 cm absorption in trace (a) comes from the same ab-
sorbing moiety as in pyruvonitrile, namely –(C=O)C≡N. How-
ever, this moiety is part of a larger molecule than pyruvonitrile
that is strongly bound to the Ag/Y surface. (The possible route
for the production of this species is discussed in Section 4.1.)
Although s-triazine and melamine have been observed on Co-
ZSM-5 during nitromethane decomposition at 553 K [20], nei-
ther these molecules nor their derivatives were seen under our
reaction conditions.
–
CN , (CN)
2
2057
1
5
(
b)
c)
Ag NC, (NC)
31
35
35
stretch
stretch
stretch
1
5
2
2
133
201
AgC N, (CN)
1
5 a
–C N , (CN)
1
3
(
2048
AgN C, (NC)
40
46
41
stretch
stretch
stretch
1
3
2
2
122
195
Ag CN, (CN)
1
3
a
–
CN , (CN)
(
d)
e)
2090
168
2190
AgNC, (NC)
AgCN, (CN)
AgNCO, (NCO)
as stretch
stretch
2
stretch
(
3
.4. Reactions of nitromethane on Ag/Y
O
a
–
CN=R–C–C≡N.
−
−
In previous work, we reported that NCO , but not CN ,
was observed when Ag/γ -Al2O3 was exposed to nitromethane
◦
served above 100 C, because this absorption is not thermally
stable above 100 C. The three absorptions at 2168, 2112, and
(
2
NM) [9]. Absorptions are seen at 719 (not shown here), 2090,
178, and 2266 cm [Fig. 7, trace (a)] on exposing Ag/Y
at 200 C to NM. Gas-phase HCN is readily identified as the
◦
−
1
−1
◦
2
063 cm , shown in trace (g), are probably the result of a ther-
◦
−
1
−1
mal reaction of pyruvonitrile over Ag/Y at 200 C, because
they are not observed when Ag/Y is exposed to pyruvonitrile at
absorber at 719 cm . The band at 2178 cm can be at-
tributed to a combination of two different absorbing species.
This is verified by the observation that it splits into two bands
on evacuation (not shown here), with the absorption develop-
◦ ◦
0 C. When pyruvonitrile is introduced at 60 C, only a broad
6
−
1
absorption at 2228 cm is observed, with shoulders at 2220
and 2236 cm
It has been reported that in an Ar matrix, the CN stretching
−1
−1 −
ing at 2164 cm due to CN and absorption developing at
.
−
1
−
2190 cm due to NCO . (These assignments were discussed
in the previous section.) Gas-phase CO is also observed at
2348 cm . Interestingly, on exposure of Ag/Y to NM, ad-
sorbed isocyanide, with an absorption at 2090 cm , is formed
−
1
vibration of pyruvonitrile is at 2225 cm [19]. However, the
2
−
1
−1
band at 2236 cm in trace (a) cannot be due to pyruvonitrile
monomer, because both its thermal stability and the band posi-
tion differ from that of authentic pyruvonitrile. We believe that
−1
−
before CN is observed, and its intensity first increases, then

420
Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
decreases. On the other hand, both the gas-phase HNCO ab-
−
1
−
sorption at 2266 cm and the absorption due to both CN
−
and NCO increase in intensity with reaction time. On a longer
time scale, the intensity of HNCO decreases while that of CO2
increases.
Interestingly, very little CN and NC is observed when
Ag/Y is exposed to a mixture of NM and NO2 in a 1:4 ra-
−
−
−
tio [Fig. 7, trace (b)]. On the other hand, both NCO , at
190 cm , and gas-phase HNCO, at 2266 cm , are concur-
−1
−1
2
rently observed if NM is in excess relative to NO2; significant
−
−
CN and NC absorptions are also observed. A very weak
−
band due to NC is seen in trace (a), where the NM to NO2 ratio
is 1:4. This band is barely observable in subsequent traces be-
−
−
cause the absorption due to NC overlaps with a strong NCO
absorption. When the sample is evacuated after reaction, ab-
−
−
sorptions due to NC and CN become significantly smaller
−
than those shown in trace (a). This demonstrates that NC and
CN are formed under these conditions, but at much lower con-
−
centrations than when Ag/Y is exposed to only NM.
Comparing traces (a) and (b) of Fig. 7 demonstrates that the
intensity of CO2 absorption keeps increasing and that its rel-
ative intensity is larger than that of the band for HNCO. This
indicates that NM reacts more rapidly with NO2 than it does
when in contact only with Ag/Y.
3
.5. Thermal stability of surface acetate and nitrate on Ag/Y
Fig. 8. Temperature of Ag/Y sample was increased after Ag/Y was exposed
◦
(a) to 1.2 Torr of acetic acid at 200 C and (b) to 61.2 Torr of a mixture with an
NO /O ratio = 1/50.
2
2
On exposure of Ag/Y to acetic acid, strong and broad ab-
−
1
sorptions in the 1200–1900 cm
region are seen [Fig. 8,
+
observed on Na/Y and BaNa/Y, where NO is produced by an
trace (a)]. The absorptions at 1795, 1777, 1727, 1419, and
−1
analogous reaction [8,13]. With a temperature increase from
0 to 100 C, gas-phase NO2 gradually increases in intensity
due to desorption, and the integrated area of the surface nitrate
1
293 cm [labeled in the top trace in (a)] are due to gas-phase
◦
4
acetic acid. This assignment is based on a comparison with the
gas-phase spectrum of authentic acetic acid. Except for the ab-
◦
−
1
band decreases by ∼75%. With a further increase to 200 C,
sorptions at 1777 and 1795 cm , all acetic acid absorptions
overlap with the broad bands of adsorbed acetate. We know
of no evidence in the literature, or in our previous studies, to
surface nitrate almost completely disappears. In previous work,
we reported on the thermal stability of surface nitrate on Ag/γ -
Al O [8]. In contrast, with Ag/γ -Al O , the change in inten-
2 3 2 3
sity of surface nitrate is small when the temperature is increased
from 50 to 100 C; this indicates that surface nitrate is much
◦
indicate that acetate undergoes a chemical reaction at 50 C un-
der our experimental conditions. Except for the absorptions due
to gas-phase acetic acid, all other absorptions are assigned to
surface acetate. As expected, the intensity of absorptions due
to gas-phase acetic acid increase with increasing sample tem-
perature, due to the desorption of acetic acid. Even though it
is difficult to quantitatively integrate the surface acetate ab-
sorptions, due to their overlap with absorptions of gas-phase
molecules, the overall intensity change of surface acetate seems
◦
more thermally labile on Ag/Y than on Ag/γ -Al2O3.
3.6. Reactivity of N-containing surface species
−
−
3.6.1. CN and NCO on Ag/γ -Al2O3
Fig. 9, trace (a), shows time-resolved spectra taken after
Ag/γ -Al2O3 was exposed to authentic HCN. As shown, the in-
small, implying that surface acetate is thermally stable even at
◦
− −1
tensity of CN at 2085 cm decreases with exposure time,
2
2
00 C. Likewise, on Ag/γ -Al2O3, surface acetate is stable at
◦
−
−1
00 C [9].
whereas the intensity of the NCO absorption at 2255 cm in-
creases. As with NCO , by themselves the position and shift of
the CN absorption on Ag/γ -Al2O3 do not allow us to uniquely
identify this moiety as either an anion or a neutral. However,
the fact that it is strongly bound to Ag/γ -Al2O3 suggest that it
is the anion, and thus we refer to it as such.
We subsequently performed the same experiment in the
presence of added water [Fig. 9, trace (b)]. Both CN , at
2085 cm , and NCO , at 2258 cm , are seen before Ag/γ -
Al2O3 is exposed to water [top trace in Fig. 9b]. When water
−
On exposing Ag/Y to NO2, two broad absorptions are ob-
−
1
served in the 1200–1500 cm region (see the bottom trace
in (b) of Fig. 8). As seen with Na/Y and BaNa/Y, these features
are in the region where surface nitrates are known to absorb [8].
−
1
−1
A broad absorption at 2081 cm , with a shoulder at 2178 cm
◦
(
not shown in the figure), is also present at 40 C. This absorp-
tion is due to NO produced from the dissociative heterolytic
adsorption of N2O4 on the Ag/Y surface [8,13]. N2O4 is, of
+
−
−
1
−
−1
+
course, in equilibrium with NO2. NO absorptions have been

Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
421
−
is admitted, the CN band vanishes instantaneously. The inten-
Ag/γ -Al2O3, which had been pre-exposed to water, was ex-
posed to HCN. In that case, the intensity of the CN band
rapidly decreased because it reacted with water, and this band
−
−1
−
sity of the NCO absorption at 2258 cm also decreases on
exposure to water, but at a slower rate than for CN . Clearly,
−
−
−
CN is not simply displaced by water molecules, but is rapidly
was hardly seen after 10 s because CN was rapidly converted
−
−
−
converted to NCO by reacting with water on the Ag/γ -Al2O3
to NCO . The peak position of the NCO absorption moved
from 2258 to 2250 cm as its intensity decreased as a re-
sult of this reaction. An analogous shift in the position of the
−
1
surface, based on the following reasoning. The absorption due
to CN rapidly decreases after the introduction of water and
vanishes completely after 0.6 s, due to either rapid displace-
ment or rapid reaction. However, in a separate experiment (not
shown here), the CN absorption was still observed when
−
−
NCO absorption was observed on evacuation of the space over
−
a Ag/γ -Al2O3 surface on which NCO is adsorbed in the ab-
−
sence of added water. Thus, this shift was likely a consequence
−
of changes in the interactions between surface NCO species
as the surface concentration decreases and/or to interaction with
the added water. Therefore, we conclude that the introduction of
−
water leads to rapid loss of CN accompanied by a slower loss
−
−
of NCO . Because NCO also reacts with water, we interpret
this behavior to indicate that CN initially reacts with water to
form NCO on Ag/γ -Al2O3. As discussed previously, NCO
is in equilibrium with HNCO via a reaction with H . HNCO
can then react with water to form ammonia and CO2 [13].
−
−
−
+
−
To obtain data on the reactivity of NCO with water with-
out interference from CN , Ag/γ -Al2O3 was first exposed to
−
authentic HNCO, after which the cell was evacuated to remove
−
1
any gas-phase HNCO. A strong absorption band at 2255 cm
,
−
1
with a shoulder band at 2273 cm , is seen in the bottom trace
−
1
−
in (a) of Fig. 10. The 2255 cm absorption is due to NCO ,
−
1
but the absorption at 2273 cm could be due to physisorbed
HNCO because it is in the appropriate spectral region, and it
gradually disappears on evacuation.
After this treatment, water was introduced into the cell. As
−
−1
expected, the NCO band at 2255 cm and the associated
shoulder rapidly decreased in intensity. To demonstrate that
this is due to a reaction with water as opposed to displace-
ment by water, separate experiments were performed in which
Fig. 9. (a) Time resolved spectra that were registered after Ag/γ -Al O was ex-
2
3
◦
posed to 0.1 Torr HCN at 200 C, (b) Ag/γ -Al O was first exposed to 0.1 Torr
2
3
◦
of HCN for 2 min and subsequently exposed to 3 Torr of water.
only gas-phase species were monitored. Even at 90 C, gas-
◦
Fig. 10. (a) Ag/γ -Al O was first exposed to 0.04 Torr of HNCO at 200 C and the cell was subsequently evacuated to remove gas-phase species. After the
2
3
◦
evacuation, the sample was exposed to 10 Torr of water, (b) Ag/γ -Al O was first exposed to 0.04 Torr of HNCO at 90 C and subsequently evacuated. After
the evacuation, the sample was exposed to 10 Torr of water and temperature was increased to 200 C. Temperature ramping was 12 C min from 90 to 120 C,
2
3
◦
◦
−1
◦
◦ −1 ◦ ◦ −1 ◦
0 C min from 120 to 190 C, 4 C min from 190 to 200 C.
3

422
Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
−
1
phase molecules such as CO2 at 668 cm and NH3 at 931
HNCO) could form [20]. However, the bands that are observed
do not resemble absorptions obtained when Ag/γ -Al O is ex-
posed to authentic cyanuric acid. These absorption are also not
−1
and 966 cm are formed after water is introduced to a cat-
alyst covered with adsorbed NCO . The assignments of NH3
and CO2 are based on a comparison of gas-phase spectra of
authentic CO2 and NH3 obtained in our IR cell and literature
data [18]. The intensities of these gas-phase absorptions fur-
ther increase on increasing the temperature from 90 to 120 C.
However, after a temperature rise from 120 to 200 C, there is
no further significant change in the intensity of the absorption
of either CO2 or NH3. Consistent with this observation, only
negligible NCO is present on the surface at this point. This
clearly demonstrates that NCO is very reactive toward water
even below 120 C. To obtain more data on the thermal stability
of NCO on Ag/γ -Al2O3, the temperature of the sample was
increased to 300 C (not shown here). Accompanying this in-
crease in temperature is a gradual decrease in the intensity of
the NCO absorption, but it does not vanish completely.
2
3
−
◦
observed during the reaction of EtOH + NO + O at 200 C.
2
2
The reason for the absence of these absorptions may be due
to the rapid reaction of HCN with water, which is produced as
part of the reaction sequence. Though the assignment and full
interpretation of these absorptions is interesting, it is beyond the
scope of this work. Nonetheless, it is clear that the oligomers of
HNCO that have been observed by Cant on Co/Y do not form
under our experimental conditions on Ag/Y.
◦
◦
−
−
◦
−
3.6.2. N-containing intermediates on Ag/Y
◦
To assess their reactivity on Ag/Y, four N-containing species
−
1
with absorptions at 2090, 2168, 2184, and 2236 cm were gen-
−
erated by exposing Ag/Y to a mixture of EtOH + NO2 + O2 at
◦
−
2
00 C. After reaction, the cell was evacuated and the resulting
As judged by changes in the intensity of NCO absorption,
−
◦
spectrum is shown in trace (a) of Fig. 11. Distinct bands due to
NCO reacts with added NO2, but at 200 C, it is less reac-
tive with NO2 than with water (not shown here). Bion et al.
−
−
R–CN, CN , and NC are apparent. Under these conditions,
−
−
−1
also studied the reactions of NCO with water, NO2, and O2
at 300 C and reported that NCO is more reactive with water
than with NO2 or O2 [12].
On exposing Ag/γ -Al2O3 to authentic HCN at 200 C in
the absence of water, absorption bands at 1373, 1448, 1536,
and 1603 cm are observed (not shown here). The multiplic-
ity of absorptions may be due to HCN which has polymerized
on the Ag/γ -Al2O3 surface and the absorptions at 1536 and
an absorption due to NCO at 2184 cm is observed at early
times. The observed bands retain most of their original intensity
when the sample temperature is increased to 300 C. To probe
◦
−
◦
◦
the reactivity of these species, NO2 was reintroduced. Compar-
ing traces (a) and (c) clearly shows no significant change in
intensity of the three absorptions, indicating that these adsor-
bates do not rapidly react with NO2. Water was then introduced
into the cell, after which the change in intensity of these ab-
sorptions was again assessed. Comparing traces (c) and (e) also
shows no significant change in intensity of any of the three
bands, indicating low reactivity with water under these condi-
−1
−1
1
603 cm would come from the C=N moieties, because these
13
bands shift to lower frequency with H CN. Because CN can be
−
oxidized to NCO on Ag/γ -Al2O3, cyanuric acid (a trimer of
Fig. 11. Panel A: (a) Spectrum was taken after Ag/Y was exposed to a mixture of 3.8 Torr EtOH + 3.8 Torr NO + 92.4 Torr of O2 for 14 min and cell was
2
subsequently evacuated for 5 min, (b) spectra was taken immediately after subsequent exposure to 4 Torr NO , (c) subsequently evacuated after 7 min exposure to
2
4
Torr NO , (d) subsequently exposed to 8 Torr of H O, (e) subsequently evacuated after 6 min exposure to 8 Torr of H O. Panel B: Ag/Y was exposed to 0.04 Torr
2
2
2
of HNCO for 3 min and the IR cell is pumped on for 10 s. Spectra were taken at each time frame immediately after subsequent exposure to 10 Torr of water.

Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
423
tions. These adsorbates also do not react significantly with NO
or O2 (not shown).
A small shoulder at 2218 cm 1 in trace (b) is due to gas-
phase N2O possibly formed from the disproportionation of NO
and/or decomposition of ammonium nitrate on Ag/Y [13]. The
ammonium nitrate would form from the reaction of ammonia
with nitric acid, which is an intermediate in deNOx chem-
istry [21]:
−
1
2
NO → N O + 2O2
2
and/or
NH4NO3 → N2O + 2H2O.
It also could be formed from the decomposition of ethyl nitrite,
which is a product of the reaction of EtOH and NO2 [9].
−
1
Gas-phase features at 2000–2100 cm in trace (d) are due
−
to the added water. To estimate the concentration of NC and
CN in the top trace in (a), a fresh sample of Ag/Y was held
−
◦
◦
under vacuum at 430 C for 3 h and then cooled to 200 C.
The Ag/Y was subsequently exposed to authentic HCN, with
the amount of adsorbed HCN controlled to obtain the desired
peak heights. When the peak heights of NC and CN were
comparable to that observed when these molecules are formed
Fig. 12. Ag/Y was pre-exposed to 1.6 Torr of AcOH for 20 min and subse-
quently evacuated before the introduction of 5.9 Torr of NO (not shown). From
2
top to bottom, the elapsed time for the first three traces is: 1 min, 2 min, and
◦ ◦
.5 min (all at 150 C). The temperature is then increased from 150 to 320 C
using a ramp of 30 C min .
3
−
−
◦
−1
from the reaction of EtOH and NO , the amount of adsorbed
HCN was calculated. This calculation provides estimated con-
centrations of cyanide and isocyanide produced by the reaction
2
3
.6.3. Reactivity of surface acetate with NO2 on Ag/Y
As discussed previously [9], surface acetate does not re-
◦
act with NO2 on Ag/γ -Al2O3 at 200 C. Here we report data
on the reactivity of surface acetate with NO2 on Ag/Y. When
Ag/Y is pre-exposed to acetic acid and then exposed to NO2,
N-containing species at 2168 and 2090 cm and CO2 form
even at 150 C (Fig. 12), demonstrating a reaction. On increas-
ing temperature, the rate of the reaction of acetate with NO2
increases, and physisorbed acetic acid, at 1697 cm , disap-
pears. The intensities of surface acetate (at 1240–1500 cm
and gas-phase NO2 also decrease, and gas-phase CO2 forms.
However, the overall rate of reaction of surface acetate is slow
of NO + EtOH. Based on this calculation, the number of ad-
2
−
−
sorbed HCN molecules (equal to the sum of CN + NC ) in
the top trace in (a) is ∼1 HCN molecule per super cage.
−1
−
Even though NCO is not visible in the traces in (a), this
◦
ion is initially observable when Ag/Y is exposed to the indi-
cated mixture. To obtain additional data on the reactivity of
−1
−
NCO , Ag/Y was exposed to HNCO, after which the cell was
−1
)
evacuated. The resulting spectra are shown in (b) of Fig. 11.
On the introduction of water, the intensity of both gas-phase
−
HNCO and surface NCO first increase, then decrease. The
◦
at 150–200 C. Acetate does not react with NO2 on the Ag/γ -
Al2O3 surface at 200 C [9].
initial increase in intensity of these absorptions could be due
to displacement of HNCO by water adsorbed on the IR cell
wall. Note the 10 s evacuation time for the spectrum shown;
evacuating the sample for more than 10 s leads to an intensity
◦
To obtain data on the rate of reaction of NO2 to form N2
with acetic acid as a reductant, we exposed 200 mg of Ag/Y
to a mixture of 1000 ppm acetic acid + 500 ppm NO2 + 7%
−
of NCO that is too weak to observe. If the cell is evacuated
◦
−1
O2 + 1% H2O at 200 C at a flow rate of 200 ml min . The
N2 GC peak is too small to be integrated, demonstrating that
acetic acid is a poor reductant. In contrast, NO2 conversion
to N2 is ∼56% when ethanol is used as reductant under oth-
after exposure of the Ag/Y to HNCO and then resealed, no in-
−
crease in the intensity of NCO or gas-phase HNCO is evident.
−
−1
The NCO band at 2190 cm is very small ∼216 s after wa-
−
ter is introduced, indicating that NCO reacts with water on the
◦
erwise identical conditions (Fig. 13). In summary, at 200 C,
Ag/Y surface. Products of this reaction are NH3 and gas-phase
surface acetate reacts slowly with NO2 on Ag/Y but not on
Ag/γ -Al2O3. Thus, acetic acid is a very poor reductant for con-
version of NOx to N2.
−
1
CO2, which absorbs at 2348 cm
:
ꢀ
ꢁ
+
−
HNCO H + NCO + H2O → NH3 + CO2.
−
The formation of CO2 demonstrates that NCO reacts with
water. Ammonia is a co-product of this reaction [21]; however,
because ammonia is a much weaker IR absorber than CO2, it
is not observed under the conditions used to generate Fig. 11.
3.7. Effect of pretreatments on of Ag/Y
The effect of pretreating the Ag/Y catalyst on the rate of NO2
conversion to N , was investigated in two separate experiments
(Fig. 13). Data acquisition was initiated 30 min after Ag/Y was
exposed to the reactants, to ensure that the flow reactor sys-
tem had stabilized. The data on N2 yield versus temperature,
2
−
1
The gas-phase features at 2000–2150 cm are due to water.
The rate of reaction of NCO with NO2 (not shown here) is
comparable to that for NCO with water.
−
−

424
Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
proximately five times smaller than those seen when Ag/Y was
pretreated by O2. Even though the peak heights can vary from
experiment to experiment, this result indicates that the oxida-
tion state of silver depends on the pretreatment and affects the
−
−
concentrations of adsorbed CN and NC . However, there is
no measurable difference in the yield for NO2 conversion to N2.
4
. Discussion
The primary objective of this study is to understand the
differences in performance of two silver-containing catalysts,
Ag/Y and Ag/γ -Al2O3, with regard to the temperature depen-
dence and the yield of deNOx reactions with ethanol as a reduc-
tant. To do this, we focus on the mechanism and the differences
in the nature and reactivity of the intermediates. We start our
discussion by outlining the mechanism for deNOx chemistry
over Ag/γ -Al O , which has been reported previously [13].
2
3
◦
Fig. 13. NO conversion to N over a Ag/Y zeolite catalyst at 200 C with
2
2
ethanol as an added reductant. The Ag/Y catalyst was treated with H (!) and
2
4
4
.1. deNOx mechanism
with O ("). The experimental details are described in Section 2. (336 ppm
2
−
1
EtOH, 500 ppm NO , 7% O , and 1% H O at a flow rate of 200 ml min .)
2
2
2
.1.1. deNOx reaction mechanism on Ag/γ -Al2O3
Two reaction pathways have been identified as participat-
ing in deNOx chemistry over Ag/γ -Al2O3 with ethanol as the
reductant. Both of these pathways lead to the formation of ni-
tromethane as a crucial intermediate. The first pathway involves
oxidation of ethanol to form acetaldehyde. This mechanism is
temperature-dependent, with oxidation by NO dominating at
2
◦
◦
2
00 C and oxidation by O2 dominating at 300 C [9]. As dis-
cussed in Section 3.2, this reaction competes with the dehydra-
tion of ethanol. This dehydration reaction decreases in impor-
tance as the concentration of NO (or O ) increases. The dehy-
2
2
◦
dration reaction at 250 C involves only ∼20% of the ethanol in
a mixture of 3.8 Torr EtOH+3.8 Torr NO2 +94.4 Torr O2. Un-
der typical reaction conditions, for a mixture of EtOH + NO2 +
◦
O2 at 200 C, dehydration is of minor importance.
Once acetaldehyde is formed, it readily reacts to form ad-
sorbed acetate ions even in the absence of added O2. These
◦
acetate ions do not significantly react near 200 C but become
◦
reactive near 300 C. A band due to adsorbed acetaldehyde can
be seen at 1706 cm at 37 C (not shown) but is not observed
at 200 C, due to the low surface concentration of acetaldehyde
−
1
◦
◦
Fig. 14. (a) Before Ag/Y is exposed to 50 mTorr of HCN, it is pretreated with
◦
at elevated temperature. The reaction of acetate ions with NO2
leads to formation of the aci-anion of nitromethane via what has
been called the “ionic channel” [9]:
O
at 430 C for 3 h. (b) Before Ag/Y is exposed to 50 mTorr of HCN, Ag/Y
2
◦
was pretreated with H at 430 C. 10 mg of Ag/Y sample is used in both exper-
2
iments and gas-phase O or H is pumped way after the pretreatment.
2
2
−
−
CH3CO + NO2 → CH2CO + HNO2,
2
2
−
−
published previously, show that 85% NO2 is converted to N2
CH2CO + NO2 → CH2NO + CO2
2
2
◦
◦
at ∼300 C [8]. The rate of the deNOx reaction at 200 C was
also monitored after Ag/Y had been pre-exposed to acetic acid;
no difference was observed (not shown here). As can be seen in
Fig. 13, both catalysts have superior catalytic activity for NO2
−
−
2
CH3CO + 2NO2 → CH2NO + CO2 + HNO2.
2
The other channel for nitromethane formation involves the
formation of methyl radicals as a result of decomposition of
acetyl radicals formed from the reaction of acetaldehyde with
NO . Reaction of methyl radicals with NO then leads to ni-
◦
reduction (∼56% N2 yield) with ethanol at 200 C, and they do
not deactivate during this time period.
2
2
To determine whether any change in the intensity of cyanide
and isocyanide occurs due to pretreatment, we performed two
experiments; the findings are shown in Fig. 14. Interestingly,
tromethane. This channel “bypasses” surface acetate involved
in the ionic channel, which displays little reactivity below
300 C. CO is a co-product of the dissociation of acetyl radi-
◦
−
−
the peak areas of the absorption of both CN and NC are ap-
cals to form methyl radicals and thus serves as a marker for this

Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
425
“
radical channel”:
als) are between 2125 and 2175 cm 1, and those of MNC are
−
between 2045 and 2055 cm [22].
−
1
•
•
The mechanism for forming R–CN differs from that for
forming the other N-containing species on Ag/Y. When acetic
acid is the reductant, no R–CN intermediate is observed. This
molecule is seen when EtOH, AA, or crotonaldehyde is used
as the reductant. R–CN is thermally stable and relatively non-
CH3CO → CH + CO.
3
◦
However, no CO is observed over Ag/γ -Al2O3 at 200 C in-
dicating that acetyl radicals are not efficiently formed at that
temperature. Thus, neither pathway leads to significant NOx
◦
◦
reduction at temperatures substantially below 300 C: For the
reactive under experimental conditions, even at 300 C. When
ionic path, reaction of acetate is rate-limiting, whereas forma-
tion of acetyl radicals and/or the reaction of methyl radicals
with NO2 appears to be the rate-limiting step(s) for the radical
pathway.
ethanol is the reductant, it first must be oxidized to acetalde-
hyde on Ag/Y. Aldol condensation occurs with acetaldehyde
on Ag/Y in the presence of insufficient NO and O to stoi-
2
2
chiometrically react with acetaldehyde. In excess NO2 and O2,
acetaldehyde preferentially reacts with these molecules to give
surface acetate (and acetyl radicals), whereas the aldol conden-
sation becomes negligible. As expected from this description,
the intensity of the R–CN absorption depends on the EtOH/NO2
ratio. If there is excess NO2, then very little, if any, R–CN is
4
.1.2. deNOx mechanism on Ag/Y
Ethanol is readily oxidized to acetaldehyde (AA) by O2 over
◦
Ag/Y at 200 C. Once acetaldehyde (AA) is formed, it reacts
with both NO2 and O2 to form surface acetate. This reaction
sequence is schematically summarized as
present.
The stable R–CN molecule that can form on Ag/Y cannot
be pyruvonitrile, because the positions of IR absorption bands
of pyruvonitrile differ from those observed here, and pyru-
EtOH + NO2 (or O2) → AA,
AA + NO2 (or O2) → acetic acid.
◦
vonitrile reacts on Ag/Y at 200 C. Instead, the absorption at
−
1
Acetaldehyde can react to form crotonaldehyde on Ag/Y, but
this reaction is effectively suppressed in the presence of typical
concentrations of NOx; under these conditions, surface acetate
prevails.
2236 cm is likely due to a molecule that has a –(CO)–CN
functional group. This R–CN molecule on Ag/Y also appears
to have a longer chain length than pyruvonitrile. Scheme 1
shows a plausible schematic mechanism for the production of
this R–CN molecule.
Surface acetate will react with NO2 to form nitromethane.
◦
This reaction, although slow at 200 C, takes place on Ag/Y,
No R–CN is observed on Ag/γ -Al2O3. This can be ex-
plained by the fact that neither crotonaldehyde nor any other
product of aldol condensation is formed on the Ag/γ -Al2O3
surface. When Ag/γ -Al2O3 is exposed to AA or AA + NO2
acetaldehyde is efficiently oxidized to acetate [9].
◦
and its rate increases as the temperature approaches 300 C.
Various N-containing intermediates can form over Ag/Y, as
discussed in Section 3.3. An open question is whether these
intermediates accumulate during the reaction and poison the
catalyst. Three of the four N-containing intermediates observed
−
−
It is known that CN on Ag/γ -Al2O3 and both NC and
−
−
−
−
CN on Ag/Y are precursors for NH in NO reduction. How-
on Ag/Y—NCO , NC , and CN —are products of the reac-
tion of nitromethane. The fourth, R–CN, is formed from the
3 2
ever, we do not believe that these ions are major precursors for
−
−
NH3 in these systems. Although NC and CN are products of
NM decomposition [23], under the present experimental con-
ditions, the reaction of NM with NO2 is more rapid than NM
decomposition. Thus, NM will preferentially react with NO2
to produce dinitromethane, which is expected to be in equilib-
rium with its aci-anion, a precursor for the aci-anion of dini-
tromethane:
reaction of aldol condensation products with NO2. Isocyanate
−
(
NCO ) does not “poison” the Ag/Y surface because it reacts
readily with water:
–
Al–OH + OCN–Al– + H2O → –Al–O–Al– + NH3 + CO2.
◦
Isocyanate ions also react with NO2, but at 200 C, this reaction
−
is slower than that of NCO with water. The other intermedi-
−
−
NO + CH –NO → O N–CH –NO .
2
2
2
2
−
−
2
2
ates, CN , NC , and RCN, could “poison” the Ag/Y catalyst
because they do not react swiftly with H2O, NO2, or O2.
With the exception of R–CN, the mechanism of forma-
tion of these N-containing intermediates from the reaction of
ethanol and NO2 is virtually identical on Ag/Y and Ag/γ -
Al2O3. The immediate precursor for the N-containing species
is nitromethane (NM). Interestingly, when NM decomposes on
This reaction has been discussed in detail previously [13].
The dinitromethane anion subsequently decomposes to NCO ;
−
however, if the concentration of NO2 is low, then some of the
1
2
NM might decompose to HCN + H2O + O2.
−
Ag/Y, NC (the initial N-containing intermediate) converts to
−
CN .
The cyanide and isocyanide ions are likely in equilibrium,
because their energies are expected to be similar. Calculations
show an energy difference between MgCN and MgNC of 1.5
−1
and 5.5 kcal mol between AlCN and AlNC [22]. As expected,
the CN stretching frequencies of MCN (M = 13 different met-
Scheme 1.

426
Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
−
In summary, oxidation of ethanol to acetate is not the rate-
NC absorptions become very weak after exposure of Ag/Y
to H2 (Section 3.7) but remain strong after exposure of Ag/Y
to O2. Stable CN and NC moieties are not seen on Ag/γ -
determining step under the present experimental conditions; in
fact, surface acetate is readily formed in either the presence or
the absence of N-containing surface species. When the surface
is saturated with acetate, the reaction of acetate with NO2 is
−
−
Al2O3, because these species are very reactive with H2O. The
−
−
superior reactivity of CN and NC on Ag/γ -Al2O3 as com-
◦
very slow at 200 C. A slow reaction of surface acetate on an
pared to Ag/Y may be due to differing metal–support interac-
−
−
acetate-saturated surface was also observed on BaNa/Y [13].
No deactivation of the NO2 reduction process by EtOH is ob-
served. It provides an N2 yield of ∼56%.
tions of the silver ion on these supports. CN and NC are
+
expected to be bound to Ag ions located in site II of Ag/Y,
because such ions are probably unable to access cations in the
β-cage with a window diameter of 2.8 Å.
Surface nitrates do not poison the Ag/Y catalyst, because
they are only weakly bound to the surface. On the other hand,
if the surface is saturated with acetate, then the reaction with
NO2 is slow even at temperatures where the reaction is efficient
at lower concentrations of surface acetate.
5. Conclusions
The mechanism for deNOx chemistry over Ag/Y with
ethanol as an added reductant is qualitatively similar to that
over Ag/γ -Al2O3, but there are some important differences in
the reactivity of some intermediates. As has been observed
for Ag/γ -Al2O3, in the presence of a mixture of NO2 and
O2 ethanol is converted to acetaldehyde over Ag/Y. However,
unlike with Ag/γ -Al2O3, under experimental conditions, ac-
etaldehyde is formed mainly as a result of oxidation of ethanol
by O2. Over Ag/Y, oxidation with O2 successfully competes
with the oxidation of ethanol by NO2 at a lower temperature
than has been observed with Ag/γ -Al2O3.
Chemistry that occurs subsequent to formation of acetalde-
hyde parallels the major features of the deNOx chemistry re-
ported for acetaldehyde as an added reductant over BaNa/Y.
Acetaldehyde interacts with Ag/Y to form surface acetate. This
surface acetate reacts with NO2 to form the aci-anion of ni-
tromethane, which is in equilibrium with nitromethane. Ni-
tromethane can react with another NO2 molecule to form a dini-
tromethane molecule, which decomposes to HNCO. These
steps are depicted in Scheme 2.
4
.2. The effect of reducing Ag with H2
The oxidation state of silver and the form in which it is
present in a zeolite can change during the course of a reaction
and/or as a result of pretreatment with H2 or O2. Hattori’s group
reported that the addition of hydrogen increases the activity
of Ag-zeolites for the selective reduction of NO by hydrocar-
δ+
bons [24]. They attributed this effect to the formation of Ag
n
+
clusters due to partial reduction and agglomeration of Ag [25].
It also has been reported that the conversion of NO to N2 with
CH4 over a Ag, H/ZSM-5 increases with the amount of reduced
0
silver [26]. During the SCR reaction over this catalyst, Ag is
gradually oxidized by O2 in the feed, resulting in gradually de-
creasing NO conversion to N2 with time on stream.
In flow reactor experiments, we compared the N2 yield with
◦
or without a catalyst pretreatment in a flow of H2 at 300 C.
Within the limits of experimental error, the yields of N2 are
the same. Under our experimental conditions, both reductants
(
EtOH) and oxidants (O2 and NO2) are present, and the oxida-
The chemistry after HNCO formation is well established and
involves hydrolysis of HNCO to form NH3 and the reaction
of NH3 with nitrous acid (HONO) to form ammonium nitrite.
Ammonium nitrite is thermally unstable above 100 C and de-
composes to form H2O and N2.
tion state and morphology of the silver can change. For exam-
ple, some of the Ag ions in Ag/Y pretreated with O2 can be
reduced by EtOH and form Ag clusters, whereas some of the
metallic silver in Ag/Y pretreated with H2 may be oxidized to
Ag and/or cationic clusters [27].
As discussed previously, roughly one Ag ion per supercage
is accessible to CN and NC . Because our data indicate that
Ag is fully exchanged, the remaining silver cations are likely
located in sites inaccessible to HCN or form neutral silver clus-
ters and/or cationic clusters. Such clusters may be major con-
tributors to NO2 reduction. Even though the role of silver in
NOx reduction on Ag/γ -Al2O3 remains a matter of debate, it
is generally accepted that silver cations coexist with other sil-
ver species [28–31]. Hattori et al. proposed that dispersed silver
clusters are responsible for NOx reduction on Ag-zeolites [25].
+
◦
+
Although the mechanism for deNOx chemistry on Ag/Y is
qualitatively similar to what is observed on Ag/γ -Al2O3, the
yield of N2 is much higher with Ag/Y at temperatures below
+
−
−
◦
300 C. This is due essentially to the rate of reaction of surface
acetate, which is much faster on Ag/Y than on Ag/γ -Al2O3,
where virtually no reaction of surface acetate is observed below
◦
about 300 C. However, this reaction appears to be the rate-
+
Ag ions are probably not important active sites for NO2 reduc-
tion with EtOH, however, because CN and NC , are strongly
bound to these cations.
−
−
−
Finally, we discuss two observations that suggest that CN
−
+
−
−
and NC , are bound to Ag ions. First, NC and CN ions
are much more reactive on BaNa/Y and Na/Y zeolites than on
Ag/Y. Second, on Ag/Y the intensity of CN and NC absorp-
−
−
−
tion depends on the oxidation state of silver. Both CN and
Scheme 2.

Y.H. Yeom et al. / Journal of Catalysis 246 (2007) 413–427
427
Table 3
[3] T. Noto, T. Murayama, S. Tosaka, Y. Fujiwara, SAE 2001-01-1935.
Assignments of bands in Fig. 6
[4] S. Sumiya, M. Saito, M. Furuyama, N. Takezawa, K. Yoshida, React.
Kinet. Catal. Lett. 64 (1998) 239;
−
1
Traces
cm
Assignments
AgNC, (NC)
N. Aoyama, S. Sumiya, N. Kakuta, K. Yoshida, React. Kinet. Catal.
Lett. 65 (1998) 139;
A. Abe, N. Aoyama, S. Sumiya, N. Kakuta, K. Yoshida, Catal. Lett. 51
(1998) 5.
(a)–(d)
2090
stretch
stretch
stretch
2
2
168
236
AgCN, (CN)
a
–CN , (CN)
(
(
(
(
e)
f)
g)
h)
2277, 2306, 2330
2268, 2303
2063, 2112, 2168, 2236
2256
CH CN, (CN)
3 stretch
[5] J.F. Thomas, S.A. Lewis, B.G. Bunting, J.M. Storey, R.L. Graves,
P.W. Park, SAE SP-1942, 199.
[6] J.H. Lee, S.J. Schmieg, S.H. Oh, Ind. Eng. Chem. Res. 43 (2004) 6343.
[7] S. Schmieg, B.K. Cho, S.H. Oh, Appl. Catal. B 49 (2004) 113;
T. Miyadera, Appl. Catal. B 13 (1997) 157;
CH CH CN, (CN)
stretch
3
2
b
–CN , (CN)
stretch
CH =CH–CN, (CN)
2
stretch
O
T. Noto, T. Murayama, S. Tosaka, Y. Fujiwara, SAE SP-1626, 45;
S. Kameoka, Y. Ukisu, T. Miyadera, Phys. Chem. Chem. Phys. 2 (2000)
367;
a
–
CN=R–C–C≡N.
O
b
S. Kameoka, T. Chafik, Y. Ukisu, T. Miyadera, Catal. Lett. 51 (1998) 11.
8] M. Li, Y.H. Yeom, E. Weitz, W.M.H. Sachtler, J. Catal. 235 (2005) 201.
9] Y.H. Yeom, M. Li, M.H. Sachtler, E. Weitz, J. Catal. 238 (2006) 100.
–
CN=H C–C–C≡N.
3
[
[
[
10] Q. Wu, H. He, Y. Yu, App. Catal. B 61 (2005) 107.
limiting step in deNOx chemistry with ethanol or acetaldehyde
as a reductant over both catalysts.
[11] S. Kameoka, Y. Ukisu, T. Miyadera, Phys. Chem. Chem. Phys. 2 (2000)
367.
There is no significant effect on the yield of N2 as a result
of pretreating Ag/Y with either O2 or H2, despite the fact that
it is generally accepted that such pretreatments oxidize metallic
[12] N. Bion, J. Saussey, M. Haneda, M. Daturi, J. Catal. 217 (2003) 47.
[13] Y.H. Yeom, B. Wen, W.M.H. Sachtler, E. Weitz, J. Phys. Chem. B 108
(
2000) 5386.
14] G. Fischer, J. Geith, T.M. Klapotke, B. Krumm, Z. Naturforsch. B 57
2002) 19.
15] S. Golay, R. Doeper, A. Renken, Appl. Catal. A 172 (1998) 97.
[
+
+
0
silver to Ag and reduce Ag to Ag , respectively. Oxidation
in the flowing reaction stream may rapidly reverse any reduc-
tion occurring due to pretreatment with H2. Thus, in the steady
state, the reacting surface may be oxidized independent of the
(
[
[16] S. Golay, R. Doepper, A. Renken, Chem. Eng. Sci. 54 (1999) 4469.
[17] E.M. Cordi, J.L. Falconer, Catal. Lett. 38 (1996) 45.
[18] NIST Chemistry WebBook, http://webbook.nist.gov.
−
pretreatment used. Evidence also indicates that both NC and
CN are bound to Ag on Ag/Y.
[
19] M.P. Bernstein, S.A. Sandford, L.J. Allamandola, Astrophys. J. 476 (1997)
32.
−
+
9
Finally, the yield of N2 does not change significantly
over time when Ag/Y is exposed to a flowing mixture of
EtOH/NO2/O2/H2O (336 ppm EtOH, 500 ppm NO2, 7% O2,
and 1% water at a flow rate of 200 ml min ). Under these
conditions, as shown in Fig. 13, the yield of N2 can approach
[
20] A. Satsuma, A.D. Cowan, N.W. Cant, D.L. Trimm, J. Catal. 181 (1999)
165.
[21] Y.H. Yeom, J.D. Henao, M. Li, M.H. Sachtler, E. Weitz, J. Catal. 231
(2005) 181.
−
1
[22] D.V. Lanzisera, L. Andrews, J. Phys. Chem. A 101 (1997) 9660.
[23] W.-F. Hu, T.-J. He, D.-M. Chen, F.-C. Liu, J. Phys. Chem. A 106 (2002)
∼
60%, making this system promising for low-temperature
NOx removal with ethanol.
7294.
[24] J. Shibata, K.-I. Shimizu, Y. Takada, A. Shichi, H. Yoshida, S. Satokawa,
A. Satsuma, T. Hattori, J. Catal. 227 (2004) 367.
[
25] A. Satsuma, J. Shibata, K.-I. Shimizu, T. Hattori, Catal. Surv. Asia 9
Acknowledgments
(2005) 75.
[
26] C. Shi, M. Cheng, Z. Qu, X. Bao, J. Mol. Catal. A 235 (2005) 35.
27] M.D. Baker, G.A. Ozin, J. Godber, J. Phys. Chem. 89 (1985) 305.
This work was partially supported by the American Chemi-
cal Society Petroleum Research Fund (Grant 41855-AC5) and
by the Chemical Sciences, Geosciences and Biosciences Divi-
sion, Office of Basic Energy Sciences, Office of Science, US
Department of Energy (Grant DE-FG02-03ER15457).
[
[28] P. Sazama, L. Cˇ apek, H. Drobna, Z. Sobalik, J. D eˇ de cˇ ek, K. Arve,
B. Wichterlová, J. Catal. 232 (2005) 302.
ˇ
[29] K. Arve, L. Capek, F. Klingstedt, K. Eränen, E.-E. Lindfords, D.Yu.
Murzin, J. D eˇ de cˇ ek, Z. Sobalik, B. Wichterlová, Top. Catal. 30 (2004) 91.
30] K.A. Bethke, H.H. Kung, J. Catal. 172 (1997) 93.
31] X. She, M. Flytzani-Stephanopoulos, J. Catal. 237 (2006) 79.
[
[
References
[32] P.G. Harrison, E.W. Thorton, J. Chem. Soc. Faraday 1 72 (1976) 2484.
33] B.A. Morrow, I.A. Cody, J. Chem. Soc. Faraday 1 71 (1975) 1021.
[
[
[
1] T.V. Johnson, SAE 2002-01-0285.
2] M.D. Kass, J.F. Thomas, S.A. Lewis, J.M. Storey, N. Domingo, R.L. Gra-
ves, A. Panov, P. Park, SAE 2003-01-3244.
[34] O. Dietz, V.M. Rayon, G. Frenking, Inorg. Chem. 42 (2003) 4977.
[35] G.A. Bowmaker, B.J. Kennedy, J.C. Reid, Inorg. Chem. 37 (1998) 3968.
[36] C.J. Blower, T.D. Smith, J. Chem. Soc. Faraday Trans. 90 (1994) 931.