Ammoxidation of Ethane to Acetonitrile over Metal-Zeolite Catalysts

Article abstract of DOI:10.1006/jcat.1997.1947

Ammoxidation of ethane to acetonitrile was studied over a variety of metal ion exchanged zeolite catalysts. We discovered that ethane can be efficiently converted to acetonitrile over some Cozeolite catalysts. The type of zeolite is very important. In this regard, ZSM-5, beta, and NU-87 are superior to others. Among various transition metal cations, Co²⁺ is most active for acetonitrile formation. Kinetic studies on Co-ZSM-5 show that the nitrile formation rate is first order in NH₃, 0.5 order in C₂H₆, and 0.8 order in O₂. In the absence of O₂, no reaction occurs. A reaction scheme is proposed, whereby C₂H₄, a reactive intermediate, is thought to add to a strongly adsorbed NH₃ forming an adsorbed ethylamine, which is subsequently dehydrogenated to form C₂H₃N.

Full text of DOI:10.1006/jcat.1997.1947

JOURNAL OF CATALYSIS 173, 511–518 (1998)
ARTICLE NO. CA971947
Ammoxidation of Ethane to Acetonitrile over Metal–Zeolite Catalysts
�
1
Yuejin Li and John N. Armor
Air Products and Chemicals, Inc., 7201 Hamilton Boulevard, Allentown, Pennsylvania 18195-1501
Received July 21, 1997; revised October 17, 1997; accepted October 22, 1997
and variable selectivities for CO2 formation. HCN was also
Ammoxidation of ethane to acetonitrile was studied over a va- formed during this reaction with a relative constant selec-
riety of metal ion exchanged zeolite catalysts. We discovered that tivity of � 5% . Earlier, a USSR patent (4) disclosed that
ethane can be efficiently converted to acetonitrile over some Co-
ethane was converted to acetonitrile with a maximum yield
of 10% over a Cr–Nb–Mo oxide catalyst at 350–500 C with
a contact time of 19 s. By-products of this reaction were not
specified.
Recently, we reported (5) that Co-beta zeolite cata-
lyst was very efficient for C2H6 ammoxidation to C2H3N
zeolite catalysts. The type of zeolite is very important. In this
regard, ZSM-5, beta, and NU-87 are superior to others. Among
�
2
+
various transition metal cations, Co is most active for acetonitrile
formation. Kinetic studies on Co-ZSM-5 show that the nitrile for-
mation rate is first order in NH3, 0.5 order in C2H6, and 0.8 order in
O2. In the absence of O2, no reaction occurs. A reaction scheme is
proposed, whereby C2H4, a reactive intermediate, is thought to add
to a strongly adsorbed NH3 forming an adsorbed ethylamine, which
(
Eq. [1])
3
2
C2H6 + NH3 + O2 = C2H3N + 3H2O.
[1]
is subsequently dehydrogenated to form C2H3N.
�c 1998 Academic Press
The C2H3N formation rate over Co-beta catalyst was 1–2
orders of magnitude higher than many metal oxide cata-
lysts that are normally used for alkane oxidation or am-
INTRODUCTION
There has been a great deal of interest in convert- moxidation reactions, e.g., VPO, SbVO and those used in
4
ing the relatively inert but inexpensive alkanes to value- Refs. (3) and (4). In this paper, we present a more complete
added chemicals, such as olefins (via dehydrogenation), picture of using metal-exchanged zeolites to catalyze the
oxygenates (via oxidation), and higher hydrocarbons (via C H ammoxidation reaction. Our interest in using metal-
2
6
oxidative methane coupling). One commercially success- exchanged zeolites as catalysts to activate alkanes derives
ful example of alkane conversion is the production of from our earlier experience (6, 7) with the NO /CH /O re-
x
4
2
maleic anhydride by selective oxidation of butane over action over Co-zeolite catalysts, where an alkane, CH , is
4
2+
V–P–O-based catalyst (1). An alkane reacting with am- activated by the adsorbed NO species on the Co sites.
x
2+
monia in the presence of oxygen (ammoxidation) can also Here, we wish to show that metal cations, especially Co
,
form a carbon-nitrogen bond. In this regard, the recently when stabilized in a zeolite environment, have unusualcata-
announced acrylonitrile process by propane ammoxidation lytic properties that are not commonly seen with traditional
presents a breakthrough in alkane conversion. In this pro- metal oxide systems. The way the alkane is activated in this
cess, acrylonitrile, a unsaturated C3 nitrile, is produced over case is quite different from the NO /CH /O reaction, and a
x
4
2
V–Sb–Al based oxide catalyst (2). However, this type of reaction scheme is proposed for C H ammoxidation over
2
6
catalysis has not been successfully extended to ethane. Ace- Co-zeolite catalysts.
tonitrile, a saturated C2 nitrile, cannot be efficiently pro-
duced using the catalysts that are effective for propane
ammoxidation (3).
EXPERIMENTAL
The metal–zeolite catalysts were prepared by the cation
exchange method. Typically, an NH4-zeolite was exchanged
with a metal cation in an aqueous solution, e.g. cobalt ac-
etate solution, at an elevated temperature. As an example,
the preparation of Co-ZSM-5 is described in detailed here,
and other catalysts can be found in our earlier publications
A few attempts by others were reported to carry out the
ethane ammoxidation reaction using metal oxide catalysts.
Using Al2O3 supported Nb–Sb oxides, Catani and Centi in-
vestigated ethane ammoxidation between 480 and 540 C
with a contact time of 2.6 s (3). They obtained ethane to
�
acetonitrile selectivity of 50% , CO selectivity of >20% ,
(
8–10). A 10-g NH4-ZSM-5 sample obtained from VAW
1
Aluminum AG, Germany (Si/Al = 12), was exchanged with
E-mail: armorjn@apci.com.
Present address: Engelhard Corporation, 101 Wood Avenue, Iselin,
�
1 liter of 0.01 M cobalt acetate aqueous solution (pH = � 6)
�
NJ 08830-0770.
at 70–80 C for 24 h. After two identical exchanges, the
511
0021-9517/98 $25.00
Copyright �c 1998 by Academic Press
All rights of reproduction in any form reserved.

5
12
LI AND ARMOR
resulting zeolite slurry was filtered, washed with 1 liter
de-ionized water, and then filtered again. Finally, the zeolite
was dried at 110 C overnight. Elemental analysis showed
Selectivity of product,
P , S = y n
�
X
�
i i
y n ,
[2]
i
i
i
i
i
that the Co/Al atomic ratio of this catalyst was 0.49, or
9
8% of the cation exchange capacity. (Note that for diva-
2+
lent cations, such as Co , a Co/Al atomic ratio of 0.5 is where y and y are the mole fractions of products Pi
i
A
equivalent to 100% of its theoretical exchange capacity.) and C H , respectively; n and n are the number of car-
2
6
i
A
The cobalt loading of this catalyst was 3.8% by weight.
i 2 3
bon atoms in each molecule of product P and C H N, re-
Some zeolites were obtained commercially and others spectively, and all the terms were evaluated for the exit
were prepared in-house. Zeolite H-beta (Si/Al = 13) was stream. The major productsofC H ammoxidation reaction
2
6
obtained from PQ Corporation, Na-mordenite (LZ-M5), over Co-zeolite catalysts are C H N, C H , CO , and N .
2
3
2
4
2
2
Na-Y (LZ Y-52) (Si/Al = 2.5), and Na-A from Union Car- Other by-products also produced were very small quanti-
bide, and K, Na-ferrierite from Tosoh. ZSM-11, chabazite, ties of N O, C H N (propionitrile), and HCN, and these by-
2
3
5
offretite, and NU-87 were synthesized according to the pub- products are insignificant, compared to the major products.
lished procedures (11–14). Na or K zeolites were first con- For the conversion and selectivity calculations, all products
verted to NH4-zeolite (10) prior to metal ion exchange.
The Co/alumina-silica catalyst was prepared by exchang-
ing amorphous silica-alumina (12 wt% Al, obtained from
W.R. Grace & Co., Davison Division) with Co2+ in a 1-liter,
and by-products were included.
RESULTS
0
.02 M cobalt acetate solution. This catalyst had a cobalt
loading of 2.98 wt% . The CoO/Al2O3 catalyst was prepared Effect of Zeolite Topology
by impregnating � -alumina with a cobalt nitrate solution
The effect of zeolite topology on the catalytic activity and
using the incipient wetness technique; the preparation was
�
selectivity was tested over a variety of cobalt-exchanged
zeolites. The reaction results (conversion and selectivities)
along with catalyst compositions are summarized in Table 1.
Cobalt exchanged amorphous silica-alumina and impreg-
nated CoO on � -alumina are also included for compari-
son. Dramatic differences in catalytic performance were
dried at 110 C and subsequently calcined in 10% O2/He
�
mixture at 500 C for 1 h. The Co loading was calculated as
1
0 wt% .
The reaction runs were made using a microreactor sys-
tem operating in a steady state, plug-flow mode at atmo-
spheric pressure. The reactor was a U-shaped quartz tube
�
2+
0
0
00
observed for these zeolite catalysts. At 450 C, Co ex-
changed ZSM-5, beta, and NU-87 have comparable C2H6
conversions (27–38% ) and C2H3N selectivities (46–51% ).
Co-ZSM-11 has only a fraction the conversion (11% ), com-
pared to the first three catalysts, and the conversion over
Co–Y is even lower (8% ). Co-mordenite has a moderate
with 1/4 OD at the inlet section and 3/8 OD at the out-
let section. The catalyst was located in the outlet section
at the center of the electrical furnace which surrounds the
reactor tube. Quartz wool plugs were used to support and
secure the catalyst bed. The feed delivery system consists
of four flow channels (NH3, C2H6, O2/He mixture, and He),
each controlled by an independent mass flow controller,
and these channels merged before proceeding to the reac-
tor inlet. Typically, a total flow rate of 100 ml/min and a
catalyst weight of 0.2 g were used for each run. A catalyst
C2H6 conversion and C2H3N selectivity, 24 and 28% , re-
spectively. Interestingly, Co-ferrierite, the catalysts most
active for NO/CH4/O2 reaction (10), is a poor catalyst
for the C2H6 ammoxidation reaction (conversion = 2% ,
�
�
selectivity = 19% at 450 C). Co-offretite showed a mod-
was normally pretreated with flowing helium at 500 C for
erate conversion, but a low selectivity, to C2H3N. For most
of these zeolite catalysts, the selectivity to CO2 is normally
low (<20% ) and that to total C2 (C2H4 + C2H3N) is >70% .
However, on Co-A zeolite, a high selectivity to CO2
was observed, and the selectivity to C2H3N is low. As
1
h before a reaction run. The reactor effluent was analyzed
by two gas chromatographs in series, both equipped with
a thermal conductivity detector (TCD). Hydrocarbons, ni-
triles, CO2, and N2O were separated by a Porapak Q col-
umn, while N2, O2, and CO were separated by a molecular
sieve 5A column. The reactor system was also connected to
an on-line mass spectrometer to confirm the product identi-
fication. All routine product analyses were carrier out with
the GC technique. The conversion and selectivity are de-
fined as:
a reference, a nonzeolite catalyst, Co2 exchanged silica-
alumina, was tested for this reaction and found to be mod-
eratelyselective and somewhat active. The Al2O3 supported
CoO, however, did not produce any detectable amount
of C2H3N.
+
Conversion of C2H6,
�
!,�
!
Effect of Metal Cation
X
X
X =
yi ni
yAnA +
yi ni
,
[1]
The effect of metal cation on the catalytic performance
is summarized in Table 2. Cu-ZSM-5 is inactive for nitrile
i
i

AMMOXIDATION OF ETHANE TO ACETONITRILE
513
TABLE 1
a
Ethane Ammoxidation over Co-Zeolites
Total C2c
sel. (% )
Apparent TOF
� 1000 sec
d
Catalyst
name
Catalyst
comp.
Temp
( C)
Conv. of
C2H6 (% )
Sel. to
C2H3N (% )
Sel. to
C2H4 (% )
Sel. to
CO2 (% )
b
�
� 1
Co-ZSM-5
Co-beta
11.0; 0.49
450
450
450
450
450
450
450
450
400
38.2
35.3
26.7
11.1
8.4
48.7
50.8
46.3
39.2
60.0
27.5
18.9
7.9
28.5
22.9
35.6
44.6
17.8
55.0
63.0
59.5
14.3
20.6
22.1
14.1
12.2
15.6
15.4
18.0
26.9
57.1
77.2
73.7
81.9
83.8
77.8
82.5
81.9
67.4
42.9
5.3
8.5
4.8
2.7
0.4
1.5
0.1
0.3
0.05
3
.83%
12.9; 0.42
.32%
16.9; 0.49
.85%
30.2; 0.39
.79%
2.5; 0.58
3.2%
5.2; 0.40
.89%
8.3; 0.50
.3%
2.8; 0.39
.9%
1.0; 0.31
0.2%
2
Co-NU-87
Co-ZSM-11
Co–Y
2
1
1
Co-mordenite
Co-ferrierite
Co-offretite
Co-Linde A
23.6
2.2
4
4
33.7
1.6
8
28.6
1
4
50
38.6
10.6
9.3
43.8
65.4
46.3
9.9
53.1
89.9
0.4
1.0
Co-silica-alumina
CoO/Al2O3
4.7; 0.22
.0%
—; —
0%
500
24.5
3
450
4.6
0.0
65.1
30.2
65.1
0
1
a
Feed: 5% C2H6, 10% NH3, and 6.5% O2, He balance; F = 100 cc/min; 0.2 g catalyst.
The numbers are Si/Al ratio, Co/Al ratio, and Co loading as weight percent, respectively.
Total C2 = acetonitrile + ethylene.
b
c
d
The apparent TOFs are calculated based on the assumption that 100% of the cobalt is dispersed and each cobalt center is on an active site.
formation, with C2H4 and CO2 being the major prod- temperatures (15). It is obvious that Co is a metal of choice
2+
2+
3+
ucts. Ni , Mn , and Fe exchanged ZSM-5 resulted for this ammoxidation reaction.
in moderate acetonitrile selectivities (10–20% ) and C2H6
conversions compared to Co-ZSM-5. The relatively low
conversion of Fe-ZSM-5 compared to Mn-ZSM-5 and Ni-
Reaction Studies on Co-ZSM-5 Catalyst
ZSM-5 is probably due to its lower Fe content (1.0% by
Figure 1 shows the catalytic performance of Co-ZSM-5
weight). These differences are clearly apparent by looking as a function of temperature. The conversion of C2H6 in-
at the column of apparent TOFs (turnover frequencies) in creases with temperature at low temperatures but levels-
�
Table 2. Ag-ZSM-5 and the precious metal-ZSM-5 catalysts off at T > 425 C. The level-off of conversion is apparently
are basically inactive for nitrile formation. For Ag-ZSM-5, due to the depletion of O2 in the reactor as evidenced
�
the reaction products are C2H4 and CO2. Pd-ZSM-5 and Pt- bythe complete conversion ofO2 at T � 425 C. The selectiv-
ZSM-5 produce primarily CO2 by ethane combustion with ity to acetonitrile mildly decreases with increasing temper-
�
�
moderate conversions � 20% . Rh-ZSM-5 showed surpris- ature (from 60% at 350 C to 45% at 475 C). The selectivity
�
ing results; the conversion of C2H6 is extremely low (2% ) to C2H4 slightly increases with temperature at T < 400 C
at 450 C. Upon careful analysis of all products, including and tends to level-off at high temperatures. Interestingly,
�
N2 formation, we found ammonia oxidation to N2 and H2O the CO2 formation is relatively constant between 350 and
�
was a dominant reaction with Rh-ZSM-5, which resulted 475 C, and its selectivity is � 20% .
in a complete depletion of O2. Thus, C2H6 remained es-
The catalysts are very stable under the ammoxidation
sentially intact; there was very little CO2 formation. This conditions. A 5-day continuous run over a Co-ZSM-5 cata-
is consistent with our earlier report that Rh-ZSM-5 is an lyst revealed no appreciable change in catalytic perfor-
effective catalyst for removing NH3 in wet streams at low mance. Further, we discovered that water did not have any

5
14
LI AND ARMOR
TABLE 2
a
Ethane Ammoxidation over Metal-ZSM-5 Catalysts
Catalyst
comp.
Reaction
temp. ( C)
Conv. of
C2H6 (% )
Sel. to
C2H3N (% )
Sel. to
C2H4 (% )
Sel. to
CO2 (% )
Total C2c
sel. (% )
Apparent TOF
� 1000 sec
d
b
�
� 1
Metal
Co
11.0; 0.49
.83%
13.5; 0.65
.46%
11.0; 0.50
.92%
14.4; 0.17
.0%
12.7; 0.31
.34%
11.1; 0.37
.54%
14.1; 0.71
.96%
11.5; 0.23
.66%
14.2; 0.12
.55%
450
450
450
450
450
400
450
450
450
38.2
15.0
11.9
9.2
48.7
0.8
28.5
65.6
50.5
39.3
12.6
7.5
20.6
33.6
30.1
42.1
60.2
91.2
54.3
100.0
98.0
77.2
66.4
69.9
55.5
24.3
8.7
5.3
0.03
0.6
1.6
1.0
0.1
0
3
Cu
Ni
4
19.4
16.2
11.7
1.2
3
Fe
1
Mn
Pd
Ag
Rh
Pt
26.9
20.3
13.7
2.0
3
4
0.0
42.0
0.0
42.0
0.0
6
0.0
0
2
21.0
2.0
0.0
2.0
0.6
2
a
Reaction conditions: The feed consisted of 5% C2H6, 10% NH3, and 6.5% O2, balance by He. The total flow rate was 100 cc/min. A 0.2 g sample
was used for all runs.
b
The numbers are Si/Al atomic ratio, metal/Al atomic ratio, and metal loading as weight percent.
c
Total C2 = C2H3N + C2H4.
d
The apparent TOFs are calculated based on the assumption that 100% of the cobalt is dispersed and each cobalt center is on an active site.
influence on this reaction. Addition of 5% H2O in the mid- more than 10% H2O is produced as a by-product. The lack
dle of a normal reaction run (with 10% NH3, 5% C2H6, and of water inhibition for this reaction is not surprising. As will
6
.5% O2 as a feed) did not produce any change in catalytic be discussed later, ammonia adsorption is a necessary step
performance. Note, under our normal reaction conditions, for the C2H6 ammoxidation reaction, and water is a much
weaker Lewis base than ammonia and cannot compete with
ammonia for adsorption sites.
As shown in Fig. 2, the dependence of conversion and
selectivity on the C2H6 partial pressure was carried out
�
at 400 C with 0.1 g Co-ZSM-5 catalyst, where the par-
tial pressure of C2H6 was systematically varied with other
concentrations (10% NH3, 6.5% O2) and total flow rate
(
100 cc/min) was kept constant. Under this set of condi-
tions, all the reactants are in sufficiently high concentra-
tions during the course of the reaction so that the reaction
rate is not restricted by the availability of other reactants,
except C2H6. The C2H6 conversion decreases with increas-
ing [C2H6]. The selectivity to C2H3N is relatively constant
(
� 45% ), while the C2H4 and CO2 selectivities decrease and
increase, respectively, with increasing [C2H6].
The C2H3N formation rate (mmol/g/h) was calculated as
a function of [C2H6] from its log–log plot. Based on the
expression of the reaction rate,
n
m
p
r = d[C2H3N]/dt = k[C2H6] [NH3] [O2] ,
[3]
FIG. 1. Ethane ammoxidation over Co-ZSM-5 as a function of re-
action temperature. The feed consists of 10% NH3, 5% C2H6, 6.5% O2
balanced in He.
the empirical reaction order for C2H3N rate with respect

AMMOXIDATION OF ETHANE TO ACETONITRILE
515
FIG. 2. Ethane ammoxidation over Co-ZSM-5 as a function of C2H6
partial pressure. The feed consists of 10% NH3, 6.5% O2, and variable
levels of C2H6.
FIG. 4. Ethane ammoxidation over Co-ZSM-5 as a function of O2
partial pressure. The feed consists of 5% C2H6, 10% NH3, and variable
levels of O2.
becomes lower. The C2H3N rate was found to be first or-
der in [NH3] (m = 1). The CO2 rate change is parallel to
that of C2H6 at [NH3] < 10% , but decreases at higher NH3
levels. It is interesting to note that the ethene production
rate increases quite dramatically with increasing [NH3] by
an order of 0.78.
In the absence ofO2, no reaction productswere observed.
Upon introducing O2 into the reacting system, C2H4 and
C2H3N were produced along with some CO2 (see Fig. 4).
The C2H6 conversion increases quite linearly with the O2
level (6% at [O2] = 1% , 18% at [O2] = 6.5% ). The selec-
tivity to C2H3N reaches the maximum at [O2] = 4–5% . The
selectivity to C2H4 decreases with O2 level at low [O2] but
stabilizes at relatively high [O2]. CO2 selectivity increases
with the O2 level in the feed. The empirical reaction orders
of the products in O2 partial pressure are interesting. The
C2H3N and C2H4 formation rates are fractional order in O2,
C2H6 partial pressure (n) was found to be 0.5. Similar to
the C2H3N rate, the order for the C2H4 formation rate is
0
.62 and that for the CO2 rate is a negative number.
Figure 3 shows the C2H6 conversion and product selec-
tivities as a function of [NH3]. C2H6 conversion increase
monotonically with [NH3], and the increase appears lin-
ear below [NH3] < 10% . Similarly, the C2H3N selectivity
increases with the [NH3] in the feed. In contrast, the C2H4
selectivity is relatively constant with the change of NH3
concentration. With increasing [NH3], the CO2 selectivity
0
.69, and 0.43, respectively, and the rate of CO2 formation
increases rapidly with increasing the O2 level.
To check the possibility that C2H4 might be an interme-
diate hydrocarbon for the C2H6 ammoxidation, an exper-
iment with an C2H4/NH3/O2 feed mixture was conducted
under the same conditions. In the case of C2H4 ammoxida-
tion, two major products are formed, C2H3N and CO2, with
C2H3N being the dominant product (� 80% ). As shown in
Table 3, C2H4 ammoxidation is more efficient than C2H6
ammoxidation; the C2H3N yield for the C2H4 ammoxida-
tion reaction is more than twice of that for the C2H6 am-
moxidation reaction.
As a control experiment, we tested oxidative dehydro-
1
genation of C2H6 to C2H4 (C2H6 + O2 → C2H4 + H2O)
FIG. 3. Ethane ammoxidation over Co-ZSM-5 as a function of NH3
partial pressure. The feed consists of 5% C2H6, 6.5% O2, and variable
levels of NH3.
2
over the same catalyst (Co-ZSM-5) under the same reac-
tion conditions, except no NH was added (see Table 4). We
3

5
16
LI AND ARMOR
TABLE 3
sional channel structures with channel openings of either
0- or 12-member rings (or 10-ring, 12-ring, in short), and
1
a
Ethylene Ammoxidation on Co-ZSM-5
it appears that these features are important for the cataly-
sis of the C2H6 ammoxidation reaction. ZSM-5 is a three-
dimensional zeolite with an intersecting, 10-ring channel
system between the straight (5.3 � 5.6 A) and the sinusoidal
Reaction
Conversion
of C2H4
(% )
Selectivity
to C2H3N
(% )
Selectivity
to CO2
(% )
C2H3N
yield
(% )
temperature
�
(
C)
˚
(
5.1 � 5.5 A˚ ) channels. Beta is a three-dimensional zeo-
4
4
4
00
25
50
48.4
53.4
57.7
81.4
79.8
77.0
17.2
18.8
21.4
39.4
42.6
44.4
lite with mutually perpendicular intersecting 12-ring chan-
nels (5.5 � 5.5 and 7.6 � 6.4 A˚ ) (16). NU-87 has a two-
dimensional, 10-ring channel system. But these 10-rings are
not intersecting. Adjacent one-dimensional 10-ring chan-
nels are linked by short 12-ring channels to create a two-
dimensional network, but access to these bridging sections
is possible only via the 10-ring windows (17). The dimen-
sions of the 10-ring channels normal to [201] are alterna-
tively 4.6 � 6.2 and 4.8 � 5.9 A along each channel, and the
˚
a
Reaction conditions: The feed consisted of 5% C2H4, 10% NH3, and
.5% O2; balance by He. The total flow rate was 100 cc/min. A 0.2 g sample
was used for all runs.
6
�
found a very low C2H4 yield (3.6% at 450 C). Interestingly,
upon the addition of a small amount of NH3, in addition
to C2H3N formation, the C2H4 yield increased to 10.9% . It
appears that the presence of NH3 promotes the oxidative
dehydrogenation of C2H6 to C2H4.
˚
1
2-ring linkage is approximately 5.3 � 6.8 A. The above ze-
olite catalysts have Si/Al ratios between 11–17 with compa-
rable Co exchange levels (84–98% ).
2+
Co-ZSM-11 and Co–Y are two catalysts that fall in the
DISCUSSION
above structure category (multidimensional structure with
1
0- or 12-ring channels) but showed lower activities than ex-
Efficient conversion of C2H6 to C2H3N has not been pected. The topology of ZSM-11 resembles that of ZSM-5,
demonstrated over metal oxide catalysts that are typically except that the intersecting, 10-ring channels are straight
used for C3H6 or C3H8 ammoxidation to acrylonitrile. Per- in both directions. The lower activity of this Co-ZSM-11
haps, this can be attributed to the fundamental difference in catalyst may be due to the much smaller number of Co2+
molecular structure between acrylonitrile and acetonitrile sites in this zeolite, compared to the ZSM-5 catalyst (be-
and the tendency of the oxide catalysts to dehydrogenate C3 cause of its higher Si/Al ratio). The low activity of Co–Y
hydrocarbons. We have shown that for the C2H6 ammoxida- catalyst is not yet understood despite its three-dimensional
tion reaction, some metal-exchanged zeolite catalysts (e.g., open structure and abundant Co2 cations in the zeolite.
Co-beta) are much more efficient that metal oxide cata- This suggests that some other factors are also important for
lysts (5). Therefore, metal exchanged zeolites must exhibit this reaction.
+
some suitable properties to catalyze the C2H6 ammoxida-
Co-zeolites that do not exhibit the above structural fea-
tion. However, asshown in Table 1, not allzeolitesare equal; tures and showed lower activity include Co-OFF, Co-MOR,
dramatic differences in catalytic performance were found Co-FER, and Co-A. Offretite has a three-dimensional, in-
when they are used as supports for Co2 . These differences tersecting 12-ring (6.7 � 6.8 A˚ ), 8-ring (4.9 � 3.6 A˚ ) chan-
are clearly apparent by looking at the column of apparent nel systems, and mordenite is a one-dimensional 12-ring
+
TOFs.
(6.5 � 7.0 A˚ ) channel system (not including the 8-mem-
The three catalysts (Co-ZSM-5, Co-beta, and Co-NU- bered side pocket). Diffusional limitation may offer one
7), most effective for the ammoxidation reaction, share possible explanation of the low activity of these Co-zeolites.
8
some common features in zeolite topology, i.e., multidimen- Reaction 1 is a very fast reaction occurring on the ex-
changed cation sites and requires the availability of three
reactants, C2H6, NH3, and O2, within the zeolite channels.
The 8-rings, because of lower molecular diffusivity in these
TABLE 4
a
Ethane Oxidation (a Reference Reaction) over Co-ZSM-5
channels, may slow down the overall diffusivity, exerting
a rate-limiting influence. When this happens, the zeolite
is effectively a one-dimensional structure as far as the re-
action is concerned. A one-dimensional channel structure
does not permit effective mixing of the reactants which are
necessary for this reaction. Therefore, it is not difficult to
understand that Co-ferrierite is the least active catalyst. For
Reaction Conversion Selectivity Selectivity Yield of C2H4 yield for
b
temp.
C)
of C2H6
(% )
to C2H4
(% )
to CO2
(% )
C2H4
(% )
C2H6/NH3/O2
(% )
�
(
4
4
00
50
15.6
49.6
22.4
7.3
76.7
76.7
3.5
3.6
7.6
10.9
�
a
Co-A (a three-dimensional, 8-ring, cage structure), at 400 C
Feed composition: 5% C2H6, 6.5% O2, balance by He; total flow rate =
�
the conversion is very low, <2% , and at 450 C the conver-
1
00 cc/min. A 0.2 g catalyst was used for the runs.
b
Feed composition: 5% C2H6, 6.5% NH3, and 5% NH3.
sion increased, but the selectivity decreased. We suspect

AMMOXIDATION OF ETHANE TO ACETONITRILE
517
�
that at 450 C the framework of zeolite A was already of C2H4 formed (consumed + unconsumed) will be even
substantially dealuminated and, perhaps, partially col- larger. The origin of this positive effect by NH3 addition is
lapsed under the ammoxidation conditions, where several not yet understood.
percentages of H2O may exist as a product. Consequently,
Based on the product analyses and control reaction tests,
the activity shown at this temperature on Co-A may have the following key reaction steps are proposed for C2H3N
been due to the catalysis that occurred at the exterior or formation (Eqs. [5]–[7]). The first step is oxidative dehy-
the amorphous part of the zeolite, resulting in largely CO2 drogenation of C2H6 to C2H4. This reaction is promoted by
and C2H4.
NH3 and is relatively fast. This reaction step is evidenced
Zeolites, with appropriate structural and compositional by the substantial amount of C2H4 observed. The second
2+
features, are excellent supports for Co , e.g. ZSM-5, beta, step is the addition of C2H4 to adsorbed NH3 forming ad-
and NU-87, but are not unique for C2H6 ammoxidation. sorbed ethylamine, C2H5NH2, as a reactive intermediate.
2+
Co exchanged into amorphous silica-alumina gave a low This intermediate is quickly consumed by further reacting
�
C2H3N yield. Its conversion was barely detectable at 450 C with O2 (oxidative dehydrogenation) forming a more stable
�
and only 11% at 500 C. On the other hand, alumina- product, C2H3N (Eq. [7]),
supported CoO is inactive for producing C2H3N, suggesting
1
2
a fundamental difference between the cobalt species in a ze-
olite environment and that on a conventionaloxide support.
While the support phase is important for the ammoxida-
tion reaction, an effective metalcation asa catalyticcenter is
essential. The very different reaction results obtained over a
number of metal cations illustrate the importance of intrin-
C2H6 + O2 → C2H4 + H2O
[5]
[6]
[7]
C2H4 + NH3 ↔ C2H5NH2
C H NH + O → C H N + 2H O.
2
5
2
2
2
3
2
To check the validity of the last step, oxidation of mo-
sic selectivity of a metal cation. Na and H-ZSM-5 are inert noethylamine wasalso carried out over the same Co-ZSM-5
for the production of nitrile. Some metals, e.g., Rh, pref- catalyst. A feed consisting of � 2% monoethylamine and
erentially catalyze the NH3 oxidation reaction to mainly 7.5% O (balance by He) was passed over the catalyst at
2
�
N2 and, because of this parallel reaction, the formation of 375 and 400 C at a flow rate of 150 cc/min. Acetonitrile
the C–N bond is discouraged over these metal catalysts. In was observed as a dominant product. Thus, this experiment
2+
this regard, Co proved to be the most active and selective supports the proposal that ethylamine is a reactive inter-
species. It should also be pointed out that the chemical state mediate for nitrile formation. (Note, additional evidence
of cobalt, as an isolated divalent cation, is also important. for ethylamine will be reported in a subsequent manuscript
When cobalt exists as an oxide (CoO or Co3O4), it does not (18).)
provide any activity for C2H3N production.
2
The origin of the CO poses another interesting question.
An important observation is that a large amount of CO could come from the combustion of ethane, ethene,
2
C2H4 is formed during the ammoxidation reaction, and the or acetonitrile. Because C H is more reactive than C H ,
2
4
2
6
total C2 (C2H4 + C2H3N) selectivity is relatively constant we believe that CO is likely a secondary product rather
2
(
� 80% ) over a variety Co-zeolite catalysts. Because of the than a reaction product of C H combustion. It is postu-
2
6
inactive nature of alkanes versus alkenes, we believe that lated that C H , a primary product, undergoes parallel re-
2
4
the reactive hydrocarbon is C2H4 and that converting C2H6 actions: to C H N via ammoxidation (Eqs. [6] and [7]) and
2
3
to C2H4 via oxidative dehydrogenation is the first step of to CO via combustion (Eq. [8]). This postulation is sup-
2
its activation. This notion was further reinforced by the ported by the kinetic results, where the CO selectivity de-
2
results of the C2H4 ammoxidation reaction. The data in creaseswith increasing[NH ]. Higher [NH ]favorsreaction
3
3
Table 3 show that a higher concentration of C2H4 (com- [6] (C H /NH addition reaction) and discourages the par-
2
4
3
pared to instantaneous [C2H4] during the C2H6/NH3/O2 re- allel C H combustion reaction (Eq. [8]). In addition, high
2
4
action) results in a higher C2H3N formation rate, which is concentrations of NH (a strong Lewis base) would cover
3
entirely consistent with the mechanistic scheme (vide infra). most of the reactive sites with adsorbed NH and minimize
3
Thus, the oxidative dehydrogenation of C2H6 to C2H4 is the number of vacant sites for C H combustion. On the
2
4
a necessary step for its ammoxidation reaction. However, other hand, we also demonstrated that the C H N/O reac-
2
3
2
�
as shown in Table 4, NH3 plays a very important promoting tion also produced some CO at T > 400 C. However, we
2
role in thisstep. In the absence ofNH3, the C2H6/O2 reaction believe that only a small portion of the CO is produced via
2
produced mostly CO2 (77% ) and carbon deposition on the this route under ammoxidation conditions.
catalyst was found, as evidenced by the blackening of the
catalyst sample. A dramatic change was observed when 5% lieve the oxidative dehydrogenation of C H to C H can
2 4
On the basis of large amount of C H formation, we be-
2
6
2
4
NH3 was added to the feed. In addition to the formation of be easily carried out over Co-ZSM-5 under our reaction
C2H3N, a much larger amount of C2H4 was found. If C2H3N conditions. The last step, converting amine to nitrile, is a
is formed through the C2H4 intermediate, the total amount thermodynamicallyfavored reaction and wasdemonstrated

5
18
LI AND ARMOR
over Co-ZSM-5 catalysts. Therefore, neither of these reac- multidimensionality, proper channel opening (10- and 12-
tions can be a rate-limiting step. The second step (Eq. [6]), member rings), and sufficient ion exchange capacity are
amine formation on adsorption sites, is thus likely a rate- essential requirements. ZSM-5, beta, and NU-87 are the
determining step for this ammoxidation reaction. This re- best zeolite hosts. The active centers for this catalysis are
action is a reversible reaction and is shifted to the right the metal cations, and Co2 is found to be the most ef-
+
side with instantaneous removal of amine product via its fective one among the many transition metal ions studied.
conversion to nitrile.
Over a Co-ZSM-5 catalyst, all three feed components are
Other reactions also occur during the ammoxidation essential for C2H3N formation, and O2 is a driving force for
process (Eqs. [8]–[14]). Ammonia oxidation to N2 and the ammoxidation reaction. Without the presence of NH3,
H2O is a most significant side reaction (Eq. [9]). On a CO2 is the dominant product. We propose that C2H3N is
Co-ZSM-5 catalyst � 50% of the NH3 consumed goes to N2 formed via three key reaction steps, whereby C2H4, an reac-
�
at T > 400 C. On some catalysts, e.g. Rh-ZSM-5, NH3 oxi- tive intermediate, is formed by oxidative dehydrogenation
dation to N2 is a dominant reaction, which depletes all the of C2H6 and then adds onto an adsorbed NH3 molecule to
O2 in the feed, therefore shutting down the ammoxidation form an adsorbed amine species, which subsequently dehy-
reaction. Reaction [11] is known to occur on Co-ZSM-5 (9) drogenates to C2H3N.
and is expected to take place in this ammoxidation process.
�
Trace amounts of HCN (22 ppm at 450 C on Co-ZSM-5)
were observed, and it may serve as an intermediate for pro-
pionitrile formation (Eq. [13]).
REFERENCES
1
2
. Emig, G., and Martin, F., Catal. Today 1, 447 (1987).
. Guttmann, A. T., Grasselli, R. K., and Brazdil, J. F., U.S. Patent
4
,746,641 (1988).
C2H4 + 3O2 → 2CO2 + 2H2O
[8]
[9]
3
. Catani, R., and Centi, G., J. Chem. Soc., Chem. Comm., 1081 (1991).
4. Aliev, S. M., Sokolovskii, V. D., and Boreskov, G. B., USSR Patent
SU-738657 (1980).
. Li, Y., and Armor, J. N., Chem. Commun., 2013 (1997).
. Li, Y., and Armor, J. N., U.S. Patent 5,149,512 (1992).
7. Li, Y., and Armor, J. N., U.S. Patent 5,260,043 (1993).
. Li, Y., and Armor, J. N., Appl. Catal. B 2, 239 (1993).
9. Li, Y., and Armor, J. N., Appl. Catal. B 1, L21 (1992).
0. Li, Y., and Armor, J. N., Appl. Catal. B 3, L1 (1993).
3
2
NH3 + O2 → N2 + 3H2O
2
5
6
2
NH3 + 2O2 → N2O + 3H2O
[10]
[11]
[12]
1
N2O → N2 + O2
2
8
N2O + C2H6 → C2H4 + N2 + H2O
C2H4 + HCN → C3H5N
C2H3N + 5.5O2 → 4CO2 + N2 + 3H2O.
1
[13] 11. Kokotaolo, G. T., U.S. Patent 4,289,607 (1981).
1
1
2. Bourgogne, M., U.S. Patent 4,503,024 (1985).
3. Occelli, M. L., Pitz, G. P., Iyer, P. S., Walker, R. D., and Gerstein, B. C.,
Zeolites 9, 104 (1989).
2
[14]
1
1
1
4. Casci, J. L., Eur. Pat. Appl. 377291 (1990).
CONCLUSIONS
5. Li, Y., and Armor, J. N., Appl. Catal. B 13, 131 (1997).
6. Meier, W. M., Olson, D. H., and Baerlocher, Ch., “Atlas of Zeolite
Structure Types,” 4th ed., Elsevier, Amsterdam, 1996.
7. Shannon, M. D., Casci, J. L, Cox, P. A., and Andrews, S. J., Nature
We discovered that some metal-exchanged zeolites were
efficient catalysts for the ammoxidation of C2H6 to C2H3N.
The effectiveness of a Co-zeolite catalyst is greatly in-
1
353(3), 417 (1991).
fluenced by the zeolite topology. Zeolite structures with 18. Li, Y., and Armor, J. N., J. Catal., submitted.