Zeolite catalysts for the selective synthesis of mono- and diethylamines

Article abstract of DOI:10.1006/jcat.1998.2275

The kinetics and mechanism of ethylamine synthesis from ammonia and ethanol over several large pore acid catalysts are described. Mordenite produced higher monoethylamine yields than the zeolites beta, Y, mazzite, and amorphous silica-alumina. The reaction proceeds via the initial formation of ethylammonium ions, and alkylamines desorb with the assistance of ammonia and equilibrate with other ethylammonium ions before leaving the catalyst pores. The high yields of ethylamines with mordenite are related to the high acid strength of the catalyst stabilizing (alkyl)ammonium ions and so blocking the dehydration of ethanol. By choosing high ammonia partial pressures, reaction temperatures below 573 K (minimizing ethene elimination from ethylammonium ions), and subtle modifications of the parent mordenite material (EDTA leaching, silylation of the external surface) ethene selectivity was further decreased. These measures allowed us to prepare a catalyst on the basis of mordenite with a Si/Al ratio of 5 that showed 99% selectivity to ethyl amines at 60% conversion and that was stable for long times on stream.

Full text of DOI:10.1006/jcat.1998.2275

JOURNAL OF CATALYSIS 180, 258–269 (1998)
ARTICLE NO. CA982275
Zeolite Catalysts for the Selective Synthesis of Mono- and Diethylamines
Victor A. Veefkind and Johannes A. Lercher1
Faculty of Chemical Technology, University of Twente, P.O. Box 217, 7500 AE, Enschede, The Netherlands
Received June 19, 1998; revised September 2, 1998; accepted September 2, 1998
amines in general and to lower substituted amines (mono-
and dimethylamine) in particular. Such postsynthetic treat-
ments are claimed to deactivate the outer surface of the
mordenite crystallites and to reduce the size of the pore
mouth (6, 11, 12). Along these lines, the selectivity enhan-
cement to lower substituted amines can be explained by the
inability of the trimethylamine (TMA) to leave the pores of
the modified catalyst and by enhancement of the probabil-
ity that TMA in the mordenite pores disproportionates with
ammonia or monomethylamine (MMA) to dimethylamine
(DMA), the desired product (6, 12). A surprising side ef-
fect isthe significant decrease in formation ofdimethylether
(DME), the main side product, after silylation (12, 13).
The direct amination of ethanol to ethylamines has been
much less discussed in the literature, but reports indicate
that zeolites Y (FAU), ZSM-5 (MFI), erionite (ERI), mor-
denite (MOR), and beta (BEA) (14–16) are active catalysts.
The biggest challenge in ethylamine synthesis is to achieve
high selectivities with respect to ethanol use. The formation
of diethyl ether (DEE) and ethene should be avoided, as
ammonia and ethene are difficult to separate (complicat-
ing the recycle), and ethene may deactivate the zeolite via
oligomerization.
The kinetics and mechanism of ethylamine synthesis from am-
monia and ethanol over several large pore acid catalysts are de-
scribed. Mordenite produced higher monoethylamine yields than
the zeolites beta, Y, mazzite, and amorphous silica–alumina. The
reaction proceeds via the initial formation of ethylammonium ions,
and alkylamines desorb with the assistance of ammonia and equi-
librate with other ethylammonium ions before leaving the catalyst
pores. The high yields of ethylamines with mordenite are related to
the high acid strength of the catalyst stabilizing (alkyl)ammonium
ions and so blocking the dehydration of ethanol. By choosing high
ammonia partial pressures, reaction temperatures below 573 K
(minimizing ethene elimination from ethylammonium ions), and
subtle modifications of the parent mordenite material (EDTA leach-
ing, silylation of the external surface) ethene selectivity was further
decreased. These measures allowed us to prepare a catalyst on the
basis of mordenite with a Si/Al ratio of 5 that showed 99% selectiv-
ity to ethyl amines at 60% conversion and that was stable for long
c
times on stream.
ꢀ 1998 Academic Press
Key Words: large pore zeolites; ethylamine synthesis; shape se-
lective amination.
INTRODUCTION
Formation of ethene from ethanol is thermodynamically
favored under most reaction conditions applied, as illus-
trated in Fig. 1. In Fig. 1a the equilibrium product distri-
bution between monoethylamine (MEA), diethylamine
(DEA), triethylamine (TEA), ammonia, water, and etha-
nol at 1 bar is shown starting from 1 mol ethanol and 4 mol
ammonia. Formation of amines is favorable, although it
should be noted that thermodynamics limit the conversion
of ethanol to 97% at 573 K. Figure 1b also shows the equilib-
rium product distribution between the various molecules,
but unlike in Fig. 1a, the formation of ethene is now per-
mitted. One notes that ethanol can be almost completely
converted to ethene at temperatures above 550 K. Thus,
ethene formation has to be kinetically blocked in order to
achieve a useful ethanol utilization.
Alkylamines are widely used as intermediates in the syn-
thesis of fine chemicals (1, 2). Conventionally alkylamines
are produced via reductive amination of aldehydes or alky-
lation of ammonia over acidic silica/alumina (1, 2). Zeolites
have attracted increasing attention as catalysts for the latter
reaction route (3), as they display strong acidity and pro-
nounced shape selectivity favoring primary and secondary
amines.
Consequently, zeolite-catalyzed methylamine synthesis
has been extensively studied (4–10), including a recent ex-
haustive review on this subject (3). One of the most suc-
cessful zeolites used for this reaction is mordenite, and a
significant number of papers and patents on the catalysis
over mordenite have been published (see Ref. (3) and ref-
erences therein). Treatment of mordenite with silylating
agents has been shown to enhance the selectivity to methyl-
The present contribution aims at providing evidence for
the main elementary steps of the alkylation of ammonia
with ethanol using combined in situ IR spectroscopy and
kinetic measurements. This knowledge is used to tailor a
1 To whom correspondence should be addressed. Fax: 31-53-4894683.
E-mail: j.a.lercher@ct.utwente.nl.
258
0021-9517/98 $25.00
c
Copyright ꢀ 1998 by Academic Press
All rights of reproduction in any form reserved.

ZEOLITES FOR SELECTIVE ETHYLAMINE SYNTHESIS
259
FIG. 1. Dependence of the equilibrium product distribution on reaction temperature at NH3/EtOH = 4 and p = 1 bar (a) without ethene present
and (b) with ethene formation. (ᮀ) MEA, (4) DEA, (᭺) TEA, (᭜) ethene.
mordenite-based catalyst for achieving maximum ethyl- tetraethoxysilane was calculated to result in a weight gain
amine yield.
of 4% by externally deposited SiO2 (12). BEA with a Si/Al
ratio of 11 was obtained from PQ Corporation (19). FAU
with Si/Al = 2.7 was supplied by Ventron. The silica–
alumina was LA-C25W from Akzo–Nobel (Si/Al = 3.5).
The physicochemical data for the catalysts used are com-
EXPERIMENTAL
Brønsted acidic mordenites with a Si/Al ratio of 10
(HMOR20), 7.5 (HMOR15), and 5 (HMOR10) were ob- piled in Table 1. The BET surface area and the micropore
tained from the Japanese Catalysis Society (17). HMOR10- volume of the catalysts were measured on a Micromeritics
E was obtained by treating a Na-MOR (NaMOR10) with Accelerated Surface Area Porosimeter (ASAP 2400). The
EDTA solution, followed by calcination and ion exchange concentration of extraframework aluminum (EFAL) was
with NH4NO3. The reason for this treatment was the determined by 27Al MAS NMR (19). The concentration
anomalously low pore volume of 0.05 cm3/g for HMOR10, of strong Brønsted acid sites was calculated from the gravi-
which was restored to 0.10 cm3/g for HMOR10-E (18). metrically determined amount of irreversibly adsorbed am-
HMOR20-M, HMOR15-M, and HMOR10-EM were pre- monia after sorption and subsequent evacuation at 373 K
pared by adding tetraethoxysilane to a suspension of the for 10 h at a pressure of 10ꢂ6 mbar (18). These gravimet-
activated MOR in n-hexane (25 ml/g zeolite), followed by ric measurements were performed on a TGA/DSC system
intense stirring at room temperature, removal of the sol- consisting of a SETARAM TG-DSC 111 instrument con-
vent, and subsequent calcination in air. The amount of nected to a high vacuum system (20). The values for EFAL
TABLE 1
Physicochemical Properties of the Investigated Brønsted Acidic Mordenites
Specific
area
Micropor.
vol.
Brønsted
acid sites
(mol/g)
Calculated
Al content
(mol/g)
EFAL
(% )
Catalyst
(m2/g)
(cm3/g)
Si/Al
HMOR20-M
HMOR20
HMOR15-M
HMOR15
HMOR10-EM
HMOR10-E
HMOR10
HMAZ
353
390
390
350
360
280
130
330
758
514
0.15
0.21
0.17
0.15
0.12
0.10
0.05
0.12
0.34
0.11
10
10
7.5
7.5
6
6
5
10
2.7
11
10
n.d.
ꢃ10–15
11
ꢃ10–15
n.d.
8
n.d.
n.d.
38
1.1 ꢁ 10ꢂ3
1.3 ꢁ 10ꢂ3
1.5 ꢁ 10ꢂ3
1.7 ꢁ 10ꢂ3
1.9 ꢁ 10ꢂ3
2.0 ꢁ 10ꢂ3
2.1 ꢁ 10ꢂ3
ꢃ1.2 ꢁ 10ꢂ3
3.0 ꢁ 10ꢂ3
0.7 ꢁ 10ꢂ3
1.1 ꢁ 10ꢂ3
1.3 ꢁ 10ꢂ3
1.7 ꢁ 10ꢂ3
1.8 ꢁ 10ꢂ3
2.0 ꢁ 10ꢂ3
2.1 ꢁ 10ꢂ3
2.3 ꢁ 10ꢂ3
1.3 ꢁ 10ꢂ3
3.0 ꢁ 10ꢂ3
1.2 ꢁ 10ꢂ3
HFAU
HBEA
Note. n.d., not detected (<5% ).

260
VEEFKIND AND LERCHER
found with 27Al MAS NMR are consistent with the dif-
The chemicals used for these experiments were ethanol
ference between calculated Al concentration and the mea- (p.a. grade) obtained from Merck, ammonia gas (Praxair,
sured Brønsted acid site concentration. 99.999% pure), and a mixed gas containing 19.5% NH3 in
For adsorption/coadsorption measurements, a Bruker He (Praxair, 99.999% pure).
IFS88 FTIR spectrometer was equipped with a vacuum cell.
RESULTS
This high vacuum cell consisted of a stainless steel chamber
equipped with CaF2 windows and a resistance-heated fur-
nace in which a gold sample holder was placed. To analyze
desorbing gasses the system was equipped with a Balzers
Coadsorption of Ammonia and Ethanol
The formation of surface species under nonreactive
QMG 420 Mass spectrometer. The base pressure of the conditions was studied by coadsorption of ammonia and
system was 10ꢂ6 mbar. Reactants were introduced to the ethanol at ambient temperature. These coadsorption exper-
system with a pressure of 10ꢂ3 mbar, using a dosing valve. imentswere carried out at partialpressuresof1 ꢁ 10ꢂ3 mbar
Spectra were taken at a resolution of 4 cmꢂ1
.
for both components using a differentially pumped inlet
Under reaction conditions the IR spectra were recorded system. First, one compound was adsorbed until full cover-
in situ using a Nicolet 20SXB FTIR spectrometer in the age was reached and then the second was introduced, while
transmission absorption mode with a resolution of 4 cmꢂ1
.
maintaining the partial pressure of the first compound. Dur-
An in situ IR cell was used in combination with gas ing coadsorption experiments the same equilibrium state
chromatography to allow simultaneous analysis of sorbed was reached, regardless of the sequence of adsorption of
species and products (for details see Ref. (21)). Typically the two adsorbates. The spectra of ammonia sorbed on
3 mg of the catalyst was pressed into a self-supported wafer HMOR20, ethanol sorbed on HMOR20, and their coad-
and activated in He flow for 1 hr at 823 K and then cooled sorption are shown in Fig. 2. Also shown is the difference
to reaction temperature.
between the spectrum of the coadsorbed species and the
For measurements at higher conversion a quartz plug spectrum of ammonia sorbed on HMOR20. Upon sorp-
flow reactor was used in combination with a GC equipped tion of ethanol, bands appeared at 2984 and 2917 cmꢂ1
with FID and TCD. A Restek RTX amine column was (attributed to asymmetric and symmetric CH3 stretching
used for separation. Typical reaction conditions were 573 K vibrations) and bands at 2939 and 2880 cmꢂ1 (attributed
and 40 mbar ethanol and 160 mbar ammonia or 100 mbar to asymmetric and symmetric CH2 stretching vibrations).
ethanol and 800 mbar of ammonia, balanced with He to Additional broad bands at 3500, 3290, 2920, and 2400 cmꢂ1
atmospheric pressure. The specific conditions in each ex- and an intense band between 1900 and 1400 cmꢂ1 were ob-
periment are reported under Results.
served. These bands are attributed to O–H stretching and
FIG. 2. IR spectra of (a) ethanol, (b) ammonia, and (c) ammonia and ethanol adsorbed on H-MOR20. (d) Difference between (c) and (b). pNH3
pEtOH = 10ꢂ3 mbar, T = 300 K.
,

ZEOLITES FOR SELECTIVE ETHYLAMINE SYNTHESIS
261
deformation vibrations from strongly bound methanol
Upon coadsorbing ammonia onto a zeolite pre-equili-
molecules (22–25). Sorption of ammonia gave rise to brated with ethanol the bands attributed to ethanol in-
bands at 3378, 3300, 3200, 3050, 2964, and 2800 cmꢂ1 to- teracting with the Brønsted acid sites of the zeolite dis-
gether with an intense band at 1465 cmꢂ1, all characteris- appeared. All bands attributed to protonated ammonia
tic for sorbed ammonium ions (26, 27). The multiple N–H appeared with exception of the shoulder at 3378 cmꢂ1
stretching bands and the slightly asymmetric shape of the attributed to the free N–H band. The spectrum of this coad-
1465cmꢂ1 band suggest different acid sitesor more than one sorption complex obtained was identical to the spectrum
orientation of the ammonium ions with respect to the zeo- obtained when starting with the ammonia pre-equilibrated
lite lattice. The shoulder at 3378 cmꢂ1 is assigned to a free zeolite shown in Fig. 2.
N–H stretching vibration from ammonium ions. This shoul-
Subsequently, the sample was evacuated and the temper-
der disappears upon coadsorption with ethanol, which can ature was increased by 10 K minꢂ1 while monitoring the
also clearly be seen in the difference spectrum (negative adsorbates by IR spectroscopy and the desorbed products
band at 3378 cmꢂ1). This indicates that ethanol and am- by mass spectrometry. The results are shown in Fig. 3. Below
monium ions interact. The additional band at 3600 cmꢂ1 500 K desorption of ethanol was observed, followed by a re-
is assigned to the O–H stretching vibration of coadsorbed lease of ammonia starting between 550 and 600 K which was
ethanol. Its position in the spectrum is similar to that of al- accompanied by the reappearance of the bands of the zeo-
cohol O–H bands sorbed on alkali ion-exchanged zeolites lite OH groups. This confirms that the ethanol interacts with
(28). The ammonium N–H deformation band did not de- the ammonia but not directly with the Brønsted acid site.
crease upon ethanol coadsorption, indicating that ethanol
adsorbs on top of or next to ammonium ions, but does not amines from Brønsted acidic mordenites in all three
replace the ammonia from Brønsted acid sites. cases showed a reactive desorption of the amine, i.e., the
Temperature-programmed desorption (TPD) of ethyl-
FIG. 3. (a) IR spectra during TPD of coadsorbed species. (b) Surface coverage of MOR20 with (᭡) ethanol, (᭿) ammonia, and (᭜) Brønsted acid
sites during TPD of coadsorbed species.

262
VEEFKIND AND LERCHER
the ethylamine. With increasing alkyl substitution the onset
of the ammonia desorption shifted to higher temperatures
(from 550 K for NH3, 680 K for MEA, 700 K for DEA, to
740 K for TEA), indicating that the amines first undergo
decomposition (Hofmann elimination of ethylene from the
ethylamines) before ammonia desorbs. This is consistent
with the higher base strength and hence the higher stability
of the ammonium ions of the ethylamines compared to am-
monia. It should be emphasized at this point that the amine
decomposition limits the potential reaction temperature as
it opens a pathway to ethene formation.
Reaction of Ammonium Ions with Ethanol
In order to investigate the reaction pathway, NH4-MOR
was contacted with a stream of ethanol (40 mbar ethanol,
6 ml/min, 3 mg zeolite) at 573 K, while analyzing the zeo-
lite, adsorbates and reaction products with in situ IR spec-
troscopy. Upon contact with ethanol the ammonium ions
rapidly disappeared and a mixture of ethylammonium ions
appeared. Steady state was reached in less than 20 min.
Figure 5 shows the spectrum of the ammonia saturated
FIG. 4. Temperature-programmed desorption of (a) ammonia, MOR and after contacting this sample with 40 mbar EtOH
(b) MEA, (c) DEA, and (d) TEA from H-MOR20. Dashed line, ammonia;
solid line, ethene.
for 30 min. The bands attributed to the ethylammonium
ions (3005–2875 cmꢂ1, C–H stretching vibrations; 2850–
2400 cmꢂ1, combination bands typical for amines; 1610–
decomposition of the amine into ethylene and ammonia at 1395 cmꢂ1, N–H and C–H deformation vibrations) quickly
temperatures above 573 K, as was also described for other developed. By gas chromatography only ethene and di-
alkylamines (see Ref. (32)). The corresponding TPD traces ethyl ether (DEE) were detected as reaction product. At
are shown in Fig. 4. The relative intensity of the ethene peak longer time on stream also some other hydrocarbons, but
vs the ammonia peak correlated well to the substitution of no amines were observed.
FIG. 5. In situ IR spectra of (a) ammonium mordenite and (b) catalyst after contacting for 30 min with 40 mbar ethanol/He (T = 573 K), showing
adsorbed ethylammonium ions.

ZEOLITES FOR SELECTIVE ETHYLAMINE SYNTHESIS
263
acid sites present further in the catalyst bed. This was
largely prevented by presaturating the catalyst with am-
monia. Therefore, all experiments reported here were per-
formed after presaturation with ammonia.
Figures 6 and 7 compare the results for the amination of
ethanol. In Fig. 6 the rate to amines is plotted for each cata-
lyst as a function of temperature. The experiments were per-
formed with 30 mg catalyst and a feed of 40 mbar ethanol
and 160 mbar ammonia in He (total flow 10 ml/min). In
Fig. 7a, the rates to amines, DEE, and ethylene at 573 K
are plotted, and in Fig. 7b, the selectivities to the differ-
ent amines at 573 K are plotted. These experiments were
performed with a feed of 40 mbar ethanol and 160 mbar
ammonia, at a total flow of approx. 15 ml/min, but the cata-
lyst weight was varied to achieve a similar conversion over
all catalysts of approx. 18% . The overall rate of reaction
was the highest over FAU, but the MOR sample showed
the highest rate to amines in the temperature region 553–
633 K and the highest selectivity at 573 K. The selectivity
to ethylene was highest over FAU, while the selectivity to
DEE was highest over BEA and SiO2/Al2O3. From the se-
lectivities toward the different amines, it can be seen that
FAU had the highest selectivity to TEA and the zeolites
with a one-dimensional 12-ring system (MOR and MAZ)
the lowest, with BEA assuming an intermediate position.
SiO2/Al2O3 exhibited a relatively low TEA selectivity. The
MOR samples clearly showed the highest MEA yield of all
catalysts tested.
FIG. 6. Rate ofMEA formation vstemperature for different catalysts.
(᭿) HMOR15, (᭜) BEA, (᭡) MAZ, (ꢁ) FAU.
After 90 min purging with He at the same temperature,
ammonia was passed over the loaded catalyst. This resulted
in the appearance of amines in the gas phase as detected by
gas chromatography, indicating that the presence of ammo-
nia is indispensable for their catalytic formation. The rate of
amine release form the zeolite pores was two to three times
lower than the rate of formation of the methylammonium
ions. This suggests that the rate of ammonia-aided amine
release is rate determining. MEA was the favored product;
TEA was hardly detected which could be seen as a con-
sequence of the excess of ammonia under these particular
reaction conditions.
Table 2 compiles the results for ethanol amination over
different mordenite catalysts. All experiments were per-
formed over 25–30 mg of catalyst. Experiments 1–5 were
performed with pNH3 = 25 mbar, pEtOH = 50 mbar at a to-
tal GHSV of 13,000 hꢂ1. Experiments 6–7 were performed
at pNH3 = 40 mbar, pEtOH = 160 mbar at a total GHSV
of 8000 hꢂ1, and experiment 8 was performed with the
same partial pressures of the reactants but at a GHSV of
2500 hꢂ1. Experiments 1 and 2 show that higher reaction
Amination of Ethanol over Acid Zeolites
When introducing ammonia and ethanol simultaneously
to a freshly activated H-MOR sample, initially high rates
of ethene and DEE formation were observed. This was
attributed to a chromatographic effect in which initially am-
monia was retained at the entrance of the catalyst bed, giv-
ing the ethanol the chance to react with the free Brønsted
FIG. 7. (a) Rate of product formation and (b) amine selectivity over various catalysts at 15–21% conversion and 573 K.

264
VEEFKIND AND LERCHER
TABLE 2
Influence of Various Reaction Parameters on Conversion and Selectivity in Ethanol Amination
Amine
select.
(% )
Amine
distribution
(% /% /% )
Ether
select.
(% )
Reactant
ratioa
Conv.
(% )
Exp
Catalyst
T (K)
1
2
3
4
5
6
7
8
HMOR20
HMOR20
HMOR15
HMOR10
HMOR10-E
HMOR15
HMOR15-M
HMOR15-M
573
623
573
573
573
573
573
573
1E/2A
1E/2A
1E/2A
1E/2A
1E/2A
1E/4A
1E/4A
1E/4A
5
25
8
65
40
74
61
58
84
93
92
89/10/1
83/16/1
84/13/3
87/12/1
81/16/3
82/15/3
85/15/0
78/21/1
10
5
8
5
10
3
9
15
11
9
0
0
32
a E, ethanol; A, ammonia.
temperatures (expectedly) increased catalyst activity, but MOR materials this was the EDTA treated HMOR10) was
decreased the selectivity to amines, mainly due to higher used for these modifications. With this strategy HMOR10-
ethene make. Experiments 1 and 3–5 show that for the EM was prepared showing a 99% selectivity at 60% con-
mordenite-based catalysts, the activities of the catalysts in- version (250 mg HMOR10-EM, 11 ml/min 100/800 mbar
creased in the order HMOR20 < HMOR15 < HMOR10 < EtOH/NH3, 558 K).
HMOR10-E, i.e., parallel to the concentration of avail-
In general, the mordenite catalysts appeared white af-
able acid sites. HMOR10, which had an anomalously low ter reaction and within the time scale of the experiments
BET surface area and micropore volume, was less active (typically 3–4 h), an appreciable deactivation was not ob-
than the EDTA-treated sample (HMOR10-E). From ex- served. In order to test the long-term stability of these
periments 3 and 6 it can be seen that increasing the ammo- catalysts, an amination experiment over HMOR10-EM
nia/ethanol (N/R) ratio from 2 to 4 increased ethanol con- catalyst (100/800 mbar EtOH/NH3, 558 K, 0.3 kg amines
ꢂ1
version and amine selectivity. The modified, i.e., silylated, kg
h
ꢂ1 produced at 60% ethanol conversion) was per-
catalyst
catalyst showed slightly lower activity, but amine selectiv- formed for 75 h. The conversion and selectivity as a function
ity increased from 84 to 93% . DEE formation was negligi- of time on stream are depicted in Fig. 9. It can be seen that
ble when using HMOR15-M and TEA formation was also under these conditions the high selectivity was sustained at
severely reduced (compare experiments 6 and 7). Higher stable conversion. The amount of amines produced in this
conversions, achieved by lowering the GHSV, influenced period was approx. 25 kg amines/kg catalyst and the catalyst
the amine distribution, but did not influence the overall was still white after use. After 4 h time on stream (1.2 kg
amine selectivity of the catalysts (compare experiments 7 amines/kg catalyst produced), the FAU and BEA sample
and 8). To explore the influence of the ammonia/ethanol typically showed a slightly grey color.
ratio, two experiments were performed in which the am-
Because water is one of the reaction products, its influ-
monia partial pressure was varied relative to a constant ence on the catalyst performance was tested. This was done
ethanol pressure of 50 mbar at 573 K, and one in which
it was varied relative to an ethanol pressure of 100 mbar.
From these experiments an apparent reaction order of 0.94
in ammonia was obtained.
Optimization of the reaction temperature, the NH3/
ethanol ratio, and the catalyst allowed us to drastically in-
crease the yield to ethylamines at the expense of ethene
formation. The results of this optimization are compiled in
Fig. 8. Decreasing the reaction temperature enhanced the
ethylamine selectivity via a drastic decrease in the rate of
ethene formation. The consequentiallossin amine yield was
compensated by increasing the ammonia partial pressure.
As high acid site concentrations and silylation the outer sur-
face of the zeolites were seen to be beneficial to the reaction
rate and the amine selectivity (see also Refs. (12, 29)), a
material with the maximum acid site concentration (for the
FIG. 8. Optimization of MEA yield by changing catalyst, partial pres-
sures, and temperature.

ZEOLITES FOR SELECTIVE ETHYLAMINE SYNTHESIS
265
FIG. 11. Coadsorption complex of ethanol and ammonia on acidic
catalyst.
Fig. 3) shows that the release of ethanol does not affect the
Brønsted acid sites, but restitutes the free N–H stretching
band. The bands of the acidic hydroxyl groups of the ze-
olite are restored only after desorption of ammonia. Note
that the proposed coadsorption complex is analogous to the
coadsorption complex observed with methanol and ammo-
nia (30, 31). The interaction between the molecules in that
complex is rather weak. Consequently, it is not detectable
with IR spectroscopy under reaction conditions.
FIG. 9. Stability of HMOR10-EM (558 K, 100/800 mbar EtOH/NH3).
(ᮀ) MEA, (4) DEA, (᭺) TEA, (᭜) ethene, and (ꢁ) conversion.
by adding 33 mbar of water to the feed of a catalyst at stable
operation. As can be seen from Fig. 10, in which the vertical
line denotes the time at which water vapor was added to the
reaction mixture, the influence of water was negligible.
As ammonium ions and weakly adsorbed ethanol suf-
fice to produce alkylammonium ions, we have proposed
that under reactive conditions the alkylation of the ammo-
nium ion involves protonation of the alcohol by the ammo-
nium ion, followed by an immediate release of water and
the simultaneous formation of a C–N bond (for details see
Ref. (29)). The ethylammonium ions resulting from the
alkylation ofthe ammonium ionsare verystable, in line with
the high base strength of the substituted amines compared
to ammonia (32, 33). During the period that the catalyst
was exposed to helium or ethanol/helium, amines were not
observed in the gas phase. Additional ammonia was needed
to remove amines from the surface. This second step, the
ammonia-mediated release of amines into the gas phase,
is slower than the amine formation; i.e., the initial alkyla-
tion of the ammonium ion is fast and the desorption of the
formed alkylamine is rate determining. Also the fact that
the amination of ethanol is found to be practically first or-
der in ammonia strongly supports a mechanism in which the
ammonia-mediated release of amines is rate determining.
The results of the transient experiments for ethanol am-
ination described above are in excellent agreement with
those for methanol amination (27). In the latter case, the
rate of release of the amines from the surface was also much
lower than their initial rate of formation, and reaction or-
ders of approx. 1 (ammonia) and 0 (methanol) have been
found (9).
DISCUSSION
On the Mechanism of Ethanol Amination
As described above, coadsorption of ethanol and ammo-
nia leads to the same coadsorption complex between the
ammonium ion and ethanol regardless of the sequence of
introduction of the sorbate. The loss of the 3378 cmꢂ1 band,
assigned to a free N–H stretching vibration (27); the appear-
ance of a band at 3600 cmꢂ1, similar to alcohol adsorption
on alkali ions; and the fact that no ammonia is displaced
upon introduction of ethanol strongly indicate a coadsorp-
tion complex in which ethanol coordinates to NH+4 as de-
picted in Fig. 11. Additionally, TPD after coadsorption (see
For the ammonia-mediated release of the amines, two
types of mechanisms have been proposed: (i) an alkyl
scavenging-type reaction, in which a gas phase ammonia
molecule removes an alkyl group from a sorbed alkyl-
ammonium ion, yielding a alkylamine in the gas phase
and a sorbed ammonium ion depleted of one alkyl group,
and (ii) adsorption-assisted desorption (a.a.d.), in which an
FIG. 10. Influence of water addition (at t = 120 min, 558 K, 100/
800 mbar EtOH/NH3). (ᮀ) MEA, (4) DEA, (᭺) TEA, and (᭜) total
amine selectivity.

266
VEEFKIND AND LERCHER
creased from 23% over FAU to 10% over BEA and 1–2%
+
Scavenging: [Rx NH y ]+ + NH3(g) → [Rxꢂ1NH y+1
]ads + RNH2(g)
ads
over MAZ and MOR. It is speculated that due to the pore
structure in FAU with a three-dimensional channel system
with rings with an aperture of 0.74 nm and its even larger
supercages, diffusion of TEA out of the zeolite particle is
relatively facile; i.e., the routes to the outside of the crystal-
lite should be short and numerous and have few obstruc-
tions. In BEA, due to the channel structure outlined above,
it is more difficult for TEA to exit, but transport will be
most difficult in MOR and MAZ since these zeolites have
only a one-dimensional large-pore system.
A.a.d.:
[Rx NH y ]+ + NH3(g) → [NH4]+ + Rx NH yꢂ1(g)
where y = 4 ꢂ x
ads
ads
SCHEME 1. Mechanisms for ammonia-mediated release of amines from
acidic catalyst.
incoming ammonia or alkylamine deprotonates the chemi-
sorbed alkylamine which desorbs. Both possible pathways
are represented in Scheme 1 and are discussed in detail in
Refs. (13, 27).
Interesting to note is the significantly lower rate of DEE
formation over MOR, compared to the other catalysts. Mor-
denite has the smallest 12-ring of the samples investigated.
It has been shown previously (13) that the formation of
ethers proceeds in those circumstances via a concerted re-
action between two ethanol molecules over ammonium and
alkylammonium ions and depends, thus, on the available
pore volume during reaction. Under reaction conditions,
the sorbed amines decrease this available volume drasti-
cally. The effect of this will be most pronounced in morden-
ite, since it has the smallest pore diameter. A large external
surface area (pronounced meso- and macroporosity) will
consequently be unfavorable for low DEE selectivity. In
that context we would like to point to the similar perfor-
mance of amorphous silica/alumina and BEA which both
have a similarly large specific external surface area.
Differences between the Zeolites
Of the different zeolites investigated, mordenite gave the
highest yield to amines, and the rates to ethene and di-
ethyl ether were low compared to the other zeolites tested.
This is attributed to the high acidic strength of the morden-
ite at high acid site concentrations, compared to the other
zeolites, which helps to keep the acid sites covered with
ammonium ions. This prevents direct elimination of water
from ethanol yielding ethene. The much lower rate of DEE
formation over MOR is explained by the small pores of
MOR and low external acid site concentration disallowing
the spatially demanding reaction of two ethanol molecules
over a weak acid site. Both subjects will be discussed in
detail below.
FAU showed the highest ethanol conversion, but also
the by far highest yield to ethene. As it contains the high-
est concentration of tetrahedrally coordinated aluminum
(Brønsted acid sites) and in consequence the highest con-
centration of Al in next nearest neighborhood to another
tetrahedrally coordinated Al, its acid sites are among the
weakest of the materials studied (34, 35). This lower acid
strength is reflected in the fact that complete removal of am-
monia is achieved at temperatures lower than 673 K (com-
pared to 823 K for mordenite). As a consequence ammonia
is less stabilized and some free acid sites may be available
that catalyze direct ethanol dehydration.
Options to Enhance Selectivity
Reaction conditions. In Table 2 it can be seen that
upon increasing the temperature from 573 to 623 K the
The other zeolites tested had a higher Si/Al ratio and,
inferred from the temperature needed for complete ammo-
nia removal (723–823 K), a higher acid strength. The rate
of formation of ethene was rather similar for these zeolites.
MOR, however, has the highest acid site concentration and,
therefore, the highest rate to amines.
All the investigated zeolites have pores with a mini-
mum opening of 12-membered rings. FAU has a three-
dimensional channel system connecting the supercages;
BEA also has a three dimensional channel system, but
no supercages. MOR and MAZ have a one-dimensional
12-ring system and a one-dimensional 8-ring system too
small for molecules larger than MEA. For a schematic rep-
resentation of the molecular sieve channels see Fig. 12. The
amine product distribution seems to be related to the di-
FIG. 12. Zeolite structures, emphasizing the diameter of the 12-ring;
mensionality of the 12-ring system. Selectivity to TEA de- (a) BEA, (b) FAU, (c) MOR, and (d) MAZ.

ZEOLITES FOR SELECTIVE ETHYLAMINE SYNTHESIS
267
selectivity did not change dramatically. Selectivity to MEA 3 and 4 of Table 2, we see that they are not parallel. This
decreased from 89 to 83% , in favor of DEA, which went would indicate that formation of ethene from DEE is not a
from 10 to 16% , while TEA selectivity stayed low. Con- major pathway during amination of ethanol.
version increased from 5 to 25% . As DEA is a secondary
product from MEA an increase in DEA selectivity at the
Silylation of the outer surface of the mordenite crystallites.
As outlined before (6, 11–13), silylation of the outer surface
expense of MEA can be expected at increasing conversion.
of a molecular sieve generates a thin silica layer blanketing
The thermodynamic equilibrium composition (as shown in
the acid sites of the outer surface and decreasing the diam-
Fig. 1) is slowly shifted from DEA to MEA at increasing
eter of the pore mouth opening. With mordenite this has
temperatures, but this is not expected to have a significant
been shown to be very effective for enhancing the selectiv-
influence since (i) changes over the depicted temperature
ity to methylamines in general and to monomethylamine in
range (200 K) are not very large, (ii) conversions in the
particular. Thus, one might expect that the treatment has a
experiments shown are not high enough to approach equi-
similar positive effect upon ethylamine synthesis.
Table 2 (comparing experiments 6 and 7) shows that sily-
librium, and (iii) shape selectivity is imposed.
The reaction temperature has a more profound influence
lation drastically reduces the ether formation. The selectiv-
on the selectivity to amines than on the amine distribution.
ity to amines in general and the monosubstituted product
Table 2 also shows clearly that the experiment performed
in particular were significantly enhanced. Despite a slightly
at 623 K had a significantly lower selectivity to amines than
lower rate of total alcohol conversion, this results in an
the experiment at 573 K. With all acid sites covered by alkyl-
increase of the rate of mono-alkylamine formation, as illus-
ammonium ions under these conditions this suggests that
trated in Fig. 13. There, it can be seen that the formation
the decomposition of the ethylammonium ions, as found in
of MEA (primary product) is approximately first order in
the TPD experiments, poses a problem for ethanol amina-
the acid site concentration for the nonsilylated mordenite
tion at temperatures above 573 K. The rates of this reaction
series. Silylation of the outer surface, however, slightly en-
seem to approach the rates for the ammonia-mediated des-
hances the rate of formation of MEA. The decreased pore
orption pathways and consequently decrease the selectivity
mouth diameter hinders the higher substituted amines to
to amines. It should be noted that in the absence of ammo-
leave the pores, thereby generating a higher degree of alky-
nia this decomposition is quantitative and has been been
lation of alkylammonium ions. In methanol amination these
used to determine acid site concentrations (32). As a con-
were found to be more reactive than lower substituted am-
sequence all reactions discussed were therefore carried out
monium ions (12, 13). The smaller pore mouth could also
at or below 573 K.
enhance the concentration of reactants in the zeolite and
thereby enhance the reaction rate.
In order to maximize the selectivity to amines, not only
formation of ethylene via Hoffmann elimination has to be
Also the use of ethanol is more efficient since diethyl
suppressed, but also the elimination of water from ethanol,
ether is not formed over the silylated catalyst. As discussed
leading to ether or ethylene. Experiments with ethanol as
above, formation of ethers under amination conditions has
single feed indicate that the rate of ethene formation is
been found to proceed mainly over the sorbed ammonium
more than two orders of magnitude higher than the rate of
ions (13). As silylation causes a decrease in the size of
the pore mouth and the elimination of external acid sites,
it exerts two positive effects. The narrowing of the pore
amine formation (>10ꢂ4 mol gꢂ1 sꢂ1 vs 10ꢂ6 mol gꢂ1 ꢂ1).
s
Thus, only a small fraction of free acid sites suffice to cata-
lyze the conversion of a significant fraction of ethanol to
ethene. This can be minimized by (i) high ratios of ammo-
nia to ethanol (compare experiments 3 and 6 in Table 2),
(ii) a strongly acidic zeolite (high stability of the ammonium
ions; compare results for mordenite and zeolite Y), and (iii)
low reaction temperatures (compare experiments 1 and 2
in Table 2). In the latter case, three factors help to suppress
ethene formation; the thermodynamic equilibrium favors
ethanol at lower temperatures (see Fig. 1b), the relative ad-
sorption constant of ammonia is higher, and formation of
ethylene via Hoffmann elimination is minimized.
Another route to ethene could be the decomposition of
DEE. If formation of ethene from DEE were much faster
than Hoffman elimination, i.e., a major pathway, then DEE
and ethene selectivities should be more or less parallel.
However, when comparing DEE and ethene formation for
example over MOR and BEA or over MOR in experiments
FIG. 13. Influence of catalyst treatment on rate of MEA formation.

268
VEEFKIND AND LERCHER
opening effectively increases the substitution of the alkyl- alkylation of ammonium ions by ethanol and the second,
ammonium ions in the pores and decreases the internal rate determining, step is the ammonia-mediated release of
available pore volume for the concerted bimolecular DEE the formed amines (sorbed as alkylammonium ions) into
formation. On the other hand it also blocks strong acid sites the gas phase. The precursor to this reaction is proposed to
on the outer surface and so suppresses formation of alkyl- be a coadsorption complex in which ethanol coordinates
ammonium ions that may also catalyze DEE formation.
with its oxygen atom to the ammonium ion on the acid
site. After the formation of this complex, protonation of
the ethanol from the ammonium ion occurs, followed by
release of water and the formation of a C–N bond. These
steps are similar to those reported for methanol amination
(12, 29, 27).
Because the ethyl group is a better leaving group than the
methylgroup, ethene formation from ethanol, diethylether,
and the ethylammonium ammonium ion isfacile. Formation
of ethene is, thus, the main problem in solid acid-catalyzed
ethanol amination. It can be suppressed by applying high
ammonia partial pressures and low reaction temperatures,
and by choice of a proper catalyst, i.e., a molecular sieve that
has narrow pores and stabilizes (alkyl)ammonium ions well.
The influence of the ammonia concentration is twofold:
(i) ethanol amination has a positive order in p(NH3) of 0.94
for the formation ofaminesand (ii) high ammonia pressures
willminimize direct exposure ofethanolto acid sitesleading
to ethene formation.
Deactivation
Two factors are important in the prevention of coking.
(i) The suppression of olefin formation and (ii) the preven-
tion of olefin oligomerization to form stable species inside
the zeolite. The stability of the HMOR10-EM catalyst is
very good. No appreciable decrease in activity and selectiv-
ity after 75 h and no discoloration of the catalyst after use
were observed. We observed some formation of olefinic
species by IR spectroscopy, but no pronounced coking or
deactivation. The two reasons we propose for the high sta-
bilityofespeciallythe modified mordenite are (i) the overall
strength of the acid sites and (ii) the pore structure.
Coke precursor formation is greatly impeded by stabi-
lization of the ammonium ions due to the strength of the
acid sites. This effect is enhanced by the relatively large con-
centration ofhigher ethylammonium ionspresent inside the
pores of the silylated material and the one-dimensional na-
ture of the mordenite pores. These higher ethylammonium
ions are even more strongly adsorbed than ammonium ions,
providing a very effective shielding of the acid sites from
(i) ethanol and (ii) the small amounts of ethylene produced,
thereby inhibiting the formation of coke precursors.
Mordenite has often been described as being very sus-
ceptible to coke formation (36), mainly due to the one-
dimensional nature of the pore system in which a relatively
small number of obstructions can render a large portion
of the pore system inaccessible. Our results, however, in-
dicate that modified mordenite is one of the most stable
catalysts for amination reactions. The formation of larger,
stable coke species can be inhibited by the geometrical con-
straints due to pore structure. This space is even more con-
fined in the case of modified mordenite due to the increased
substitution of ammonium ions at the acid sites.
Low reaction temperatures are necessary, as TPD of
alkylamines shows that above 573 K ethylamines start to
decompose at an appreciable rate to ethene and ammonia.
For a given ammonia partial pressure lower temperatures
help to maintain the acid sites covered with ammonium
ions, preventing direct access of ethanol or DEE.
Comparison of the acid catalysts studied shows con-
clusively that the catalyst properties influence the overall
amine selectivity and the selectivity within the amines. With
respect to the first point, chemical composition and pore
geometry are found to be the crucial parameters. Zeolites
with a low Si/Al ratio, which are weaker solid acids and do
not stabilize ethylammonium ions well, show a higher ten-
dency to form ethene. This is attributed to the presence of
free acid sites which catalyze the formation of ethene due
to ethanol or DEE dehydration. It should be noted at this
point, however, that the rate of amination sympathetically
varies with the concentration of acid sites and that an opti-
mum Si/Al ratio exists as a consequence. This is exemplified
clearly by the low amine yield found with FAU (Si/Al = 2.7)
having the lowest yields and rates in ethylamines and the
high yields found with HMOR10-EM (Si/Al = 5). In that
context, the temperature needed for complete ammonia
removal during activation of the sample, as observed with
FTIR, seems to be a good indication of whether the chosen
catalyst will show a good amine selectivity. If complete am-
monia removal takes place below 723 K, the catalyst will
likely show high ethene selectvities.
FAU has much weaker acid sites and selectivity to olefins
is high. This fulfills one of the conditions for coke formation:
the presence of precursors in the form of olefinic species.
The oligomerization process can also be facilitated by the
higher availability of free acid sites. The specific shape of
the FAU structure with its large supercages facilitates the
formation of larger, more complex molecules which can
easily become trapped within the zeolite.
CONCLUSIONS
Amination of ethanol over acidic catalysts has been
The pore structure of the zeolite has a definite influence
shown to proceed in two steps. The first step consists of the on amine selectivity and amine distribution. If the acid sites

ZEOLITES FOR SELECTIVE ETHYLAMINE SYNTHESIS
269
are in a relatively confined space, as in MOR, which had 10. Ashina, Y., and Fukatsu, M., U.S. Patent 4,485,261 (1984).
11. Niwa, M., Kato, S., Hattori, T., and Murakami, Y., J. Chem. Soc.
Faraday Trans 1 80, 3135 (1984).
12. Gru¨ndling, Ch., Eder-Mirth, G., and Lercher, J. A., J. Catal. 160, 299
(1996).
13. Veefkind, V. A., Gru¨ndling, Ch., and Lercher, J. A., J. Mol. Catal. A
134, 111 (1998).
14. Kaeding, W. W., U.S. Patent 4,082,805 (1978).
15. Deeba, M., Ford, M. E., and Johnson, T. A., in “Catalysis of Organic
Reactions” (W. Blackburn, Ed.). Dekker, New York, 1990.
16. Shen, J.-P., Ma, J., Jiang, D.-Z., and Min, E.-Z., Chinese Chem. Lett. 5,
305 (1994).
the smallest channels of the zeolites investigated, the avail-
able space for ether formation over weakly acidic centers
will be lower and consequently the rate to DEE will de-
crease. The rate to TEA is greatly influenced by the dif-
fusional constraints imposed on the products. FAU, with
its open, three-dimensional pore system, had the highest
TEA selectivity(23% ), whereasMOR and MAZ, with their
one-dimensional pore system, produced much less TMA
(1–2% ).
The most promising way of enhancing MEA yield is sily- 17. Sawa, M., Niwa, M., and Murikami, Y., Zeolites 10, 532 (1990).
18. Pirngruber, G. D., Eder-Mirth, G., and Lercher, J. A., J. Phys. Chem.
lation of the outer surface of the mordenite. Although the
overall conversion drops slightly, selectivity to by-products
(especially diethyl ether) decreased and the MEA yield in-
101, 561 (1997).
19. Nivarthy, G. S., He, Y., Seshan, K., and Lercher, J. A., J. Catal. 176, 192
(1998).
creased. This is attributed to a decrease in available pore
volume for the bimolecular alcohol dehydration reaction
caused by increasing substitution of the alkylammonium
ions and the elimination of accessible Brønsted acid sites
or alkylammonium ions on the outer surface of the molec-
ular sieve crystals.
20. Stockenhuber, M., and Lercher, J. A., Microporous Mater. 3, 457
(1995).
21. Mirth, G., Eder, F., and Lercher, J. A., Appl. Spectrosc. 48, 194
(1994).
22. Blaszkowski, S. R., and van Santen, R. A., J. Phys. Chem. 99, 11728
(1995).
23. Zecchina, A., and Otero Arean, C., Catal. Rev. 35, 261 (1993).
24. Thursfield, A., and Anderson, M. W., J. Phys. Chem. 100, 6698
(1996).
25. Mirth, G., Lercher, J. A., Anderson, M. W., and Klinowski, J., J. Chem.
Soc. Faraday Trans. 86, 3039 (1990).
26. Little, L. H., “Infrared Spectra of Adsorbed Species,” p. 180. Academic
Press, London, 1966.
ACKNOWLEDGMENT
Thiswork wasperformed under the auspicesofNIOK, The Netherlands
Institute for Catalysis Research, Lab. Report UT-98-1-08.
27. Kogelbauer, A., Lercher, J. A., Steinberg, K. H., Roessner, F.,
Soellner, A., and Dimitriev, R. V., J. Chem. Soc. Faraday Trans. 88,
2283 (1992).
REFERENCES
1. Heilen, G., Mercker, H. J., Frank, D., Reck, R. A., and Ja¨ckh, R., 28. Gru¨ndling, Ch., Eder-Mirth, G., and Lercher, J. A., Res. Chem. Inter-
in “Ullmann’s Encyclopedia of Industrial Chemistry” (W. Gerhertz,
Ed.), 5th ed., Vol. A2, p. 1. VCH, Weinheim, 1985.
2. Turcotte, M. G., and Johnson, T. A., in “Kirk Othmer Encyclopedia of
mediates 23, 25 (1997).
29. Segawa, K., and Tachibana, H., in “Proceedings, 10th ICC,” p. 1273.
Elsevier, Amsterdam, 1993.
Chemical Technology” (J. J. Kroschwitz, Ed.), 4th ed., Vol. 2, p. 369. 30. Kogelbauer, A., Gru¨ndling, Ch., and Lercher, J. A., J. Phys. Chem.
Wiley, New York, 1992. 100, 1852 (1996).
3. Corbin, D. R., Schwarz, S., and Sonnichsen, G. C., Catal. Today 37, 71 31. Gru¨ndling, Ch., Veefkind, V. A., Eder-Mirth, G., and Lercher, J. A.,
(1997). Stud. Surf. Sci. Catal. 105, 591 (1997).
4. Ashina, Y., Fujita, T., Fukatsu, M., Niwa, K., and Yagi, Y., Stud. Surf. 32. Parillo, D. J., Adamo, A. T., Kokatailo, G. T., and Gorte, R. J., Appl.
Sci. Catal. 84, 28,779 (1986). Catal. 67, 107 (1990).
5. Shannon, R. D., Keane, M., Jr., Abrams, L., Staley, R. H., Gier, T. E., 33. Parillo, D. J., Gorte, R. J., and Farneth, W. E., J. Am. Chem. Soc. 115,
Corbin, D. R., and Sonnichsen, G. C., J. Catal. 113, 367 (1988).
6. Segawa, K., and Tachibana, H., J. Catal. 131, 482 (1991).
7. Ilao, M. C., Yamamoto, H., and Segawa, K., J. Catal. 161, 20 (1996).
8. Dingerdissen, U., Nagy, E., and Fetting, F., Chem. Ing. Technol. 63, 625
(1991).
12441 (1993).
34. Makarova, M. A., Zholobenko, V. L., Al-Ghefaili, K. M., Thompson,
N. E., Dewing, J., and Dwyer, J., J. Chem. Soc. Faraday Trans. 90, 1047
(1994).
35. Parillo, D. J., and Gorte, R. J., J. Phys. Chem. 97, 8786 (1993).
9. Chen, D. T., Zhang, L., Kobe, J. M., Yi, C., and Dumesic, J. A., J. Mol. 36. Magnoux, P., Cartraud, P., Mignard, S., and Guisnet, M., J. Catal. 106,
Catal. 93, 337 (1994). 242 (1987).