Cobalt-catalyzed amination of 1,3-propanediol: Effects of catalyst promotion and use of supercritical ammonia as solvent and reactant

Article abstract of DOI:10.1006/jcat.1999.2408

The catalytic synthesis of 1,3-diaminopropane from 1,3-propanediol and ammonia was studied in a continuous fixed-bed reactor in the pressure range 50 to 150 bar. The unsupported Co-based catalysts applied were characterized by N₂ physisorption, XRD, XPS, TPR, and ammonia adsorption using pulse thermal analysis and DRIFT spectroscopy. The latter investigations revealed that the best catalyst, 95 wt% Co-5 wt% Fe, contained only very weak acidic sites, unable to chemisorb ammonia. The absence of strong acidic and basic sites was crucial to suppress the various acid/base-catalyzed side reactions (retro-aldol reaction, hydrogenolysis, alkylation, disproportionation, dimerization, oligomerization). Other important requirements for improved diaminopropane formation were the use of excess ammonia (molar ratio NH₃/diol > 20) and the presence of the metastable β-Co phase. A small amount of Fe additive could efficiently hinder the transformation of this phase into the thermodynamically stable α-Co phase and thus prevent catalyst deactivation up to 10 days on stream. Application of supercritical ammonia almost doubled the selectivity to amino alcohol and diamine. The selectivity enhancement in the near-critical region is attributed to elimination of the interphase mass transport limitations and to the resulting higher surface ammonia concentration.

Full text of DOI:10.1006/jcat.1999.2408

Journal of Catalysis 183, 373–383 (1999)
Article ID jcat.1999.2408, available online at http://www.idealibrary.com on
Cobalt-Catalyzed Amination of 1,3-Propanediol: Effects
of Catalyst Promotion and Use of Supercritical
Ammonia as Solvent and Reactant
A. Fischer, M. Maciejewski, T. Bu¨rgi, T. Mallat, and A. Baiker1
Laboratory of Technical Chemistry, Swiss Federal Institute of Technology, ETH-Zentrum, CH-8092 Zurich, Switzerland
Received November 16, 1998; revised December 22, 1998; accepted December 22, 1998
presented mainly in the patent literature (13–18). The in-
formation available in patents provides only little insight
into this type of amination reactions. A typical example is
the amination of 1,3-propanediol with NH3, which was cho-
sen to illustrate the application range of a nickel–rhenium
catalyst (17, 18). There are no data available concerning
the product composition, except the conversion (45% ) es-
timated on the basis of water formed.
Amination of a simple aliphatic alcohol over a metal
catalyst is already a complex process which consists of
dehydrogenation, condensation, and hydrogenation steps
(Scheme 1) (12, 19). The first and last redox processes are
catalyzed by the metal, while the reaction of the interme-
diate carbonyl compound with NH3 or an amine to form
an imine or enamine does not require a metal catalyst, but
can be accelerated by acid/base catalysis (12, 20). Each in-
termediate and the product amine can take part in various
side reactions [for details, see (1, 4, 7, 11, 12)]. The synthesis
of a diamine from the corresponding diol requires repeti-
tion of the three major steps which multiplies the likelihood
of undesired side reactions, including oligomerization and
cyclization (21–23).
The catalytic synthesis of 1,3-diaminopropane from 1,3-pro-
panediol and ammonia was studied in a continuous fixed-bed reac-
tor in the pressure range 50 to 150 bar. The unsupported Co-based
catalysts applied were characterized by N2 physisorption, XRD,
XPS, TPR, and ammonia adsorption using pulse thermal analysis
and DRIFT spectroscopy. The latter investigations revealed that
the best catalyst, 95 wt% Co–5 wt% Fe, contained only very weak
acidic sites, unable to chemisorb ammonia. The absence of strong
acidic and basic sites was crucial to suppress the various acid/base-
catalyzed side reactions (retro-aldol reaction, hydrogenolysis, alky-
lation, disproportionation, dimerization, oligomerization). Other
important requirements for improved diaminopropane formation
were the use of excess ammonia (molar ratio NH3/diol > 20) and the
presence of the metastable -Co phase. A small amount of Fe ad-
ditive could efficiently hinder the transformation of this phase into
the thermodynamically stable -Co phase and thus prevent catalyst
deactivation up to 10 days on stream. Application of supercritical
ammonia almost doubled the selectivity to amino alcohol and di-
amine. The selectivity enhancement in the near-critical region is
attributed to elimination of the interphase mass transport limita-
tions and to the resulting higher surface ammonia concentration.
c
1999 Academic Press
Key Words: amination; -cobalt; -cobalt; 1,3-diaminopropane;
A further complication in the synthesisofprimaryamines
is that the product amines are significantly more reactive
than the reagent NH3. This effect may be illustrated by
the amination of 1,6-hexanediol. When using NH3, only
23% 1,6-diaminohexane was obtained at 220 C and 300 bar
(16). Amination with dimethylamine ran smoothly at at-
mospheric pressure and 195 C, affording 51% of the corre-
sponding diamine (24).
We have recently found that the application of supercrit-
ical ammonia (scNH3) as a solvent and reactant provides
a remarkable selectivity improvement in the amination of
aliphatic diols, compared with the performance of the same
catalyst under similar conditions, but applying subcritical
pressures (25). The aim of the present work was to obtain
deeper insight into this demanding type of amination re-
action and reveal the role of the catalyst in controlling the
product distribution. From economic and environmental
points of views, the direct synthesis of aliphatic diamines
1,3-propanediol; supercritical ammonia; effect of iron and lan-
thanum promotion.
INTRODUCTION
Amination of aliphatic alcohols for the synthesis of ali-
phatic amines is an industrially relevant process. Numer-
ous examples with good to high yields can be found in re-
views and books (1–12). The reaction is catalyzed by metal
hydrogenation–dehydrogenation-type catalysts such as
nickel, cobalt, and copper and bysolid acids(mainlyzeolites
and phosphates).
In comparison to reports on the amination of simple
aliphatic alcohols, studies on the transformation of diols
to the corresponding acyclic primary amines are scarce and
1 To whom correspondence should be addressed.
373
0021-9517/99 $30.00
c
Copyright
1999 by Academic Press
All rights of reproduction in any form reserved.

374
FISCHER ET AL.
washed, and dried at 100 C at reduced pressure (<10 kPa).
The material was calcined in air at 400 C for 2 h, mortared,
and sieved to a particular mesh size (0.14–0.4 mm). Subse-
quently the sample was activated in the reactor by reduc-
tion with H2 for 4 h at 335–420 C (Table 1). The reduction
temperature was chosen based on the maximum H2 con-
sumption rate in TPR (see Table 2).
Two fractions of the wet precipitate of the Co–Fe cata-
lyst (Table 1) were additionally treated with sodium ac-
etate or ammonium hydrogen phosphate, to modify their
acid/base properties. For this purpose, ca. 68 g of the wet
hydrated oxide–carbonate was mixed with 100 ml 0.1 M
aqueous NaOAc or (NH4)2HPO4 solution. The slurry was
mixed at 90 C for 30 min, filtered, and washed carefully
with hot water. The following steps were identical to those
of the general procedure.
SCHEME 1
from diols and NH3 would be an attractive alternative to
the presently applied methods (3, 26, 27).
EXPERIMENTAL
Materials
Catalyst Characterization
1,3-Propanediol (>97% , Fluka), 3-amino-1-propanol
(>98% , Fluka), NH3 (99.98% , Pan-Gas), H2 (99.999% ,
Pan-Gas), and N2 (99.995% , Pan-Gas) were used without
further purification.
Important properties of the various cobalt-based cata-
lysts used are collected in Table 1. Distilled water was ap-
plied in all catalyst preparations. According to a general
procedure, the metal nitrates were dissolved in 500 ml wa-
ter. About 100 g of an aqueous solution containing 20 wt%
(NH4)2CO3 was then dropped into the metal nitrate so-
lution over 1 h at room temperature until a pH of 7 was
reached. After 2 h stirring the precipitate was filtered,
Specific surface areas (SBET), mean cylindrical pore di-
ameters (hdpi), and specific pore volumes (Vp(N )) were de-
2
termined by N2 physisorption at 77 K using a Micromeritics
ASAP 2000 apparatus. Catalyst samples were first degassed
for 10 h at 100 C. The surface area was calculated in the rel-
ative pressure range 0.05 to 0.2, assuming a cross-sectional
area of 0.162 mm2 for the N2 molecule. The mean cylin-
drical pore diameter was determined using the relation
hdpi = 4Vp(N )/SBET
.
2
X-ray diffraction patterns were obtained on a Siemens
D5000 powder X-ray diffractometer using CuK radiation
(35 mA, 35 mV, Ni filter).
XPS measurements were performed with a Leybold
Heraeus LHS11 apparatus using MgK radiation. The
work function of the spectrometer was set so that the Au f7/2
line for metallic gold was located at 84.0 eV. Thin sample
films were pressed on a sample holder, evacuated in a load
lock, and subsequently transferred into the UHV chamber
for analysis. Pass energies of 151 and 63 V were used. Slight
sample charging was corrected using the C 1s line at
284.6 eV as an internal standard. Differential charging was
estimated by varying the potential of a tubus in the lens
system of the analyzer.
Adsorption of NH3 was studied using the pulse thermal
analysis technique on a Netzsch STA 409 thermal analyzer
(28, 29). NH3 pulses were injected with a Valco dual exter-
nal sample injection valve equipped with two 1-ml sample
loops. A quadrupole mass spectrometer (QMG 420, Balz-
ers) was used for analysis of the gas composition. After
calcination and reduction of the catalyst, NH3 pulses were
injected into the He carrier gas flow of 50 ml min 1, under
isothermal conditions.
TABLE 1
Preparation of Various Cobalt-Based Catalysts
Reduction
temperature
( C)
Metal contenta
(wt% )
Precursor
(mmol)
Catalyst
Co
Co(100)
Co(NO3)2 6 H2O(172)
335
335
Co–La
Co(88), La(12)
Co(NO3)2 6 H2O(171)
La(NO3)3 6 H2O(9.2)
Co–Fe
Co(95), Fe(5)
Co(52), Fe(48)
Co(NO3)2 6 H2O(171)
Fe(NO3)3 9 H2O(9.2)
335
340
420
Co–Fe-48
Co(NO3)2 6 H2O(171)
Fe(NO3)3 9 H2O(9.2)
Co–Fe–PO4 Co(95), Fe(5)b
Co(NO3)2 6 H2O(171)
Fe(NO3)3 9 H2O(9.2)
(NH4)2HPO4
c
Co–Fe–Na
Co(95), Fe(5)b
Co(NO3)2 6 H2O(171)
Fe(NO3)3 9 H2O(9.2)
NaOAcc
335
a Determined by ICP-AES.
b Bulk PO34 and Na+ contents were minor; for surface composition see
The same thermal analyzer system was used for deter-
mining the TPR profiles. Eighty milligrams of catalyst was
heated in a flow of 10% H2 in He, at a heating rate of 5 K
min 1. The formation of water (product of catalyst re-
Table 3.
c Co–Fe catalyst “precipitate” was treated with aqueous (NH4)2HPO4
or NaOAc solutions before drying and calcination.

Co-CATALYZED AMINATION OF 1,3-PROPANEDIOL
375
TABLE 2
Structural Properties of the Co-Based Catalysts Determined by N2 Physisorption, TPR, and XRD Analysis
a
b
c
d
Calcination
( C)
SBET
(m2 g
Vp,N
hdpi
TTPR
Phases detected by XRD
(crystallite size in nm)e
2
1
1
Catalyst
Co
Co–La
Co–Fef
)
(cm3 g
)
(nm)
( C)
400
400
—
400
400
400
400
400
400
54
37
108
35
12
6
68
133
91
0.34
0.43
0.58
0.31
0.10
0.02
0.15
0.67
0.6
21
34
14
29
43
15
7
338
356
—
383
—
CoO(12), Co3O4(22)
Co3O4(16)
Poorly crystalline
CoO(25), Co3O4(22)
-Co(31)
-Co(31)
Poorly crystalline
Co3O4(13)
Co–Fe
Co–Feg
Co–Feh
—
Co–Fe-48
Co–Fe–PO4
Co–Fe–Na
401
492
397
17
23
Co3O4(16)
a BET surface area.
b BJH cumulative desorption pore volume.
c Mean pore diameter hdpi = 4Vp,N /SBET
.
2
d Temperature of maximum hydrogen consumption.
e Mean crystallite size determined by XRD line broadening.
f Without calcination.
g Calcined and reduced.
h After amination reaction.
duction) was monitored by MS. For investigating the down flow mode. Standard conditions used were 8.0 g cata-
decomposition of NH3, 60 mg Co–Fe catalyst, prereduced lyst, 195 C, 135 bar, contact time 40,000 gs mol 1, molar
in H2 in a preceding cycle, was heated in a flow of 10% ratio of propanediol/NH3/H2-1/60/2.
1
NH3 in He at the heating rate of 10 K min
.
Concentrations in the liquid product mixture were deter-
The chemical composition of the catalysts was deter- mined by GC analysis using external calibration standards
mined by ICP-AES using an IRIS ICP-AES spectrometer (HP-5890A, FID detector, HP-1701 capillary column).
(Thermo Jarell Ash) in an inductively coupled Ar plasma Losses of the desired products due to possible evaporation
chamber.
DRIFT spectroscopic measurements were carried out on ual products were identified by GC–MS analysis.
a Perkin–Elmer 2000 FTIR spectrometer. A KBr back- A 50-ml high-pressure quartz cell was used to investigate
were minor, as confirmed in specific experiments. Individ-
ground spectrum was recorded at 50 C (100 scans with a whether the reaction mixture was in the supercritical state.
resolution of 1 cm 1) after heating the sample for 1 h in The experiment was performed with a mixture consisting
an Ar flow of 15 ml min 1. The catalyst was pretreated at of propanediol/NH3/H2 at a molar ratio of 1/60/2. Visual
300 C for 1 h in Ar to remove physisorbed water. Subse- inspection of the phase behavior at 130 bar and 200 C con-
quently, the sample was reduced with H2 at the same tem- firmed the existence of a substantially homogeneous super-
perature as applied before the catalytic tests (see above). critical phase. Note that the critical data on pure NH3 are
The background spectrum of the sample was recorded in Tc = 132.4 C and Pc = 114.8 bar (30). Conversion of alcohol
Ar at temperature steps of 50 C from 50 to 250 C. After Xa and selectivity Si of product amine i were defined as
being cooled to 50 C, the catalyst was flushed with NH3
(3600 ppm in Ar, 50 ml min 1) for 20 min. Spectra were
recorded at temperature steps of 50 C from 50 to 250 C.
Xa = 100 (Fa0 Fa)/Fa0,
Si = Fi /(Fa0 Fa),
where Fa0 and Fa correspond to molar flow rates of diol at
reactor inlet and outlet, respectively, and Fi represents the
molar flow rate of product i at reactor outlet.
Catalytic Amination
The apparatus consisted essentially of a dosing system
for the reactants, a high-pressure fixed-bed reactor, and
a gas/liquid separator. The reactor was constructed of
Inconel-718 tubing of 13-mm inner diameter and 38-ml vol-
ume. The temperature in the reaction zone was measured
with a thermocouple located in the center of the tube and
was regulated with a PID cascade controller. Total pressure
in the reactor system was set by a Tescom backpressure reg-
ulator. Liquid NH3 and 1,3-propanediol were dosed with
syringe pumps (ISCO D500). The reactor was operated in
RESULTS
Catalyst Characterization
Textural properties and surface composition of the
calcined catalysts. Some important properties of the Co-
based catalysts, obtained by N2 physisorption, TPR, and
XRD methods, are summarized in Table 2. All uncalcined

376
FISCHER ET AL.
catalysts were poorly crystalline. During calcination the
surface area and pore volume decreased considerably,
as emerges from the comparison of Co–Fe and Co–Fef
(Table 2). The unsupported materials after calcination
1
possessed relatively high surface areas up to 133 m2 g
.
Addition of Fe or La to Co as a second component chang-
ed considerably the pore volume, pore diameter, and
surface area. Similarly, treatments of Co–Fe after pre-
cipitation and before drying with ammonium hydrogen
phosphate (Co–Fe–PO4) or sodium acetate (Co–Fe–Na)
had a remarkable influence on the textural properties.
X-ray patterns of the calcined catalysts showed the pres-
ence of cobalt oxides, mainly Co3O4. The CoO phase was
absent after addition of 5 wt% La or treatments with
NaOAc or (NH4)2HPO4.
XPS analysis of Co–Fe–Na and Co–Fe–PO4 confirmed
the existence of Na and P, respectively, on the surface of
the modified Co–Fe catalysts. Sodium was identified by the
peak at 1071.1 eV (Na 1s) and phosphorus by the signal
at 132.5 eV (P 2p) which is indicative of phosphate (31).
The 2p3/2 line of cobalt was found at a binding energy of
780.17 eV, consistent with Co3O4 or Co2O3 (32, 33). Only
weak shakeup satellites for the 2p3/2 line indicated mostly
diamagnetic Co(III). The Fe 2p3/2 line at 710.5 eV is char-
acteristic of Fe3O4 or Fe2O3 (33).
The surface composition was estimated by integrating
the XPS signals of the different elements using formerly
determined sensitivity factors (34). The results are listed in
Table 3. The relative abundance of iron to cobalt at the sur-
face of the Co–Fe catalyst was almost twice as great as in the
bulk (determined by ICP-AES, see Table 1). The materials
treated with sodium acetate or ammonium hydrogen phos-
phate exhibited about the same Co/Fe ratio at the surface
as in the bulk.
FIG. 1. Temperature-programmed reduction of calcined Co–Fe, Co–
Fe–PO4, and Co–Fe–Na catalysts. Conditions: 10% H2 in He, heating rate
1
5 C min 1, gas flow rate 50 ml min
.
and Co–Fe–Na are similar to that obtained for the calcined
sample containing only Co. The weight loss observed dur-
ing reduction indicated that the two peaks represent the
reduction of Co3O4 to Co via CoO (35). Reduction of Co–
Fe–PO4 (and also Co–Fe-48) occurred in three steps. The
maxima of the biggest peaks of all catalysts are shown in
Table 2. TPR measurements confirmed that temperatures
listed in Table 1 for catalyst prereduction before amination
were sufficiently high to complete the reduction of oxides
to metals.
In additional sets of TPR experiments, NH3 (10 vol% in
He) was applied as a reducing agent. The reduction started
above 250 C and all peaks were shifted to higher temper-
atures, compared with the runs in which H2 was used. For
example, with Co–Fe the maximum in the TPR signal was
shifted from 383 to 470 C. The experiment with Co–Fe was
repeated with the prereduced catalyst to determine unam-
biguously the temperature range in which the metals are ac-
tive in NH3 decomposition. This reaction, indicated by the
formation of H2, was observed only from 250 C onward.
Therefore, decomposition of NH3 as a side reaction during
catalytic amination of propanediol at or below 210 C (see
later) can be excluded.
Characteristics of the reduced catalysts. Amination of
alcohols over metal catalysts is catalyzed by surface M0
sites. The calcined catalysts, which can only be considered
as “precursors” of the real active materials, were prere-
duced in situ before use. The reducibility of various Co-
based mono- and bimetallic catalysts and the influence
of treatments with acidic or basic salts were studied by
temperature-programmed reduction. Three examples are
shown in Fig. 1. The shapes of the reduction curves of Co–Fe
TABLE 3
The phase composition ofCo after prereduction and after
the amination reaction is shown in Fig. 2. After reduction
with H2 at 335 C, the catalyst consisted of hexagonal close-
packed -Co and face-centered cubic -Co (Fig. 2a). At
moderate temperatures -Co is metastable, which explains
the growth of the main reflection of the -Co phase at 47.5
(Fig. 2b) during the amination reaction (at 150–210 C for
12 h). Previous XRD studies on Co catalysts ascertained
XPS Analysis (in at.%) of Some Unreduced Co–Fe Catalysts
Element (sensitivity factor)
Catalyst
Co–Fe
Co–Fe–Na
Co–Fe–PO4
Co 2p(3.8)
Fe 2p(3.0)
Na 1s(2.3)
P 2p(0.39)
45.2
45.4
44.5
4.5
2.5
2.3
—
2.7
—
—
—
5.7

Co-CATALYZED AMINATION OF 1,3-PROPANEDIOL
377
FIG. 3. X-ray diffraction patterns of Co–Fe, Co–Fe–PO4, and Co–Fe–
Na after calcination and reduction.
FIG. 2. X-ray diffraction patterns of Co (a), Co–Fe (c), and Co–Fe-48
(d) after calcination and reduction, and Co (b) after use in amination.
the importance of the reduction temperature in the devel- Na at 50 C are collected in Table 4. The amount of NH3
opment of and phases. Only -Co was generated by H2 physisorbed on Co–Fe was small and it was completely
reduction at 400 C (36). The sluggish transition from - to removed by the He carrier gas within 5 min. Besides, no
-Co was observed between 340 and 380 C (37).
chemisorbed NH3 could be detected. A combined UPS and
Figures 2c and 2d illustrate the influence of Fe additive XPS study revealed earlier that the adsorption of NH3 on
on the phase composition of the reduced catalysts. Inter- Co is very weak at room temperature (38).
estingly, only -Co was detectable when the catalyst con-
Co–Fe–PO4 was the only catalyst that chemisorbed NH3.
taining 5 wt% Fe (Co–Fe) was reduced at 335 C. At this Its desorption occurred between 75 and 250 C, with a max-
temperature formation of the -Co phase is expected (37). imum rate at 170 C. Chemisorption of NH3 is a clear in-
Apparently, the presence of Fe favored formation of the dication for acidic sites introduced by the treatment with
metastable -Co phase. Moreover, restructuring during the (NH4)2HPO4. The enhanced capacity for physisorbed NH3
amination reaction was suppressed and the -Co phase was of this catalyst may simply be attributed to the higher sur-
still present after use at 150–210 C for 10 days (not shown face area, as compared with the untreated Co–Fe catalyst
in Fig. 2). The average size of Co crystallites in reduced (Table 4). On the other hand, neither physisorption nor
Co–Fe remained unaffected during the amination reaction chemisorption of NH3 could be detected on Co–Fe–Na, as
(see Co–Feg and Co–Feh in Table 2). Similarly, addition of expected for this base-treated material.
5 wt% La to Co prevented the restructuring of Co during
To our knowledge, no report has yet appeared on the
amination; these results are not shown here. These exper- DRIFT study of NH3 adsorption on Co. Due to the lack of a
imental observations reflect the excellent stabilizing effect reference spectrum, DRIFT spectra of NH3 adsorbed on Ni
of a small amount of Fe or La additive. An increase in the and Cu catalysts were used to allocate the adsorption bands
iron content to 48 wt% (Co–Fe-48) resulted in the devel- (39–41). Figure 4 shows the spectra of NH3 chemisorbed
opment of a new phase, a CoFe alloy (Wairauite), during on the prereduced Co–Fe–PO4 catalyst at different tem-
reduction by H2.
peratures. Adsorbed NH3 could be detected up to 250 C.
A partly different picture was obtained in the study of Vibrations at 1240 and 1610 cm 1 are ascribed to Lewis acid
Co–Fe catalysts treated with basic or acidic salts. The XRD centers ( s NH3, as NH3). The weaker deformation mode at
patterns of Co–Fe–Na (reduced at 335 C) and Co–Fe–PO4
(reduced at 440 C) disclosed the presence of both cobalt
phases, though the amount of -Co was very small in Co–
Fe–Na (Fig. 3). Structural changes during amination were
not observed for any of these two catalysts.
TABLE 4
Ammonia Adsorption Determined by Pulse Thermal Analysis
at 50 C, Together with the BET Surface Areas
Physisorption
Chemisorption
SBET
Adsorption of ammonia. Pulse thermal analysis (PTA)
was used for the quantitative study of NH3 physisorption
and chemisorption. This method provides correct, undis-
torted values by following the weight change during ad-
sorption and desorption (28, 29). The data measured on
the reduced samples of Co–Fe, Co–Fe–PO4, and Co–Fe–
1
1
1
Catalyst
Co–Fe
Co–Fe–PO4
Co–Fe–Na
(cm3 g
)
(cm3 g
)
(m2 g
)
0.4
0.8
0
0
1.5
0
12
31
6
a Both measurements were carried out on catalysts prereduced by H2.

378
FISCHER ET AL.
sorbed NH3 could be detected on the surface of Co–Fe and
Co–Fe–Na catalysts, in agreement with the PTA experi-
ments (Table 4).
Catalytic Amination
Choice of catalyst. Screening of several supported and
unsupported metal catalysts in an autoclave revealed that
only Co-, Ni-, and Cu-based catalysts were useful for the
amination of 1,3-propanediol. Ru/C and Pd/C were active
but unselective.
A parameter study was carried out in the fixed-bed re-
actor with the most promising catalysts Co, Co–La, and
Co–Fe. It was found that a small proportion of H2 in the
feed (1–5 mol% ) was sufficient to prevent the undesired
dehydrogenation reactions leading to the formation of ni-
triles and carbonaceous deposits (43). For the comparison
of Co, Co–Fe, and Co–La catalysts the pressure was set
at 135 bar and the temperature was varied between 190
and 210 C. Note that NH3 forms a supercritical fluid above
Tc = 132.4 C and Pc = 114.8 bar (30). A rather high mo-
lar excess of NH3 was employed to minimize oligomeriza-
tion of the intermediates and products. The existence of a
substantially homogeneous supercritical reaction mixture
was confirmed by separate experiments in a quartz cell (see
Experimental). Some representative data illustrating the
activity and selectivity of Co-based catalysts are listed in
Table 5. A considerable deactivation within 12 h on stream
was observed with unpromoted Co. It is likely that deacti-
vation is connected to the restructuring of Co, as observed
by XRD (Figs. 2a and 2b).
FIG. 4. DRIFT spectra of adsorbed ammonia on the reduced Co–Fe–
PO4 catalyst at different temperatures. Carrier gas: 50ml min 1 Ar. Spectra
obtained after subtraction of the background spectrum are shown.
1423 cm 1 is attributed to Brønsted populations ( as NH+4 ).
The weak NH+ around 1670 cm could not be clearly
1
s
4
identified. A multiplet was observed in the NH stretching
region (3352 cm 1, 3258 cm 1, 3216 cm 1, 3165 cm 1) that
is assigned to the asymmetric and symmetric N–H and to
the first overtone of 2 as NH3 vibrations (42). The reverse
adsorption signal at 3668 cm 1 indicates a decrease in the
amount of hydroxyl groups on the catalyst surface, because
NH3 was adsorbed on Brønsted sites.
Restructuring was negligible with bimetallic catalysts
containing 5 wt% Fe or La. Conversion and selectivity val-
ues presented in the following figures and tables could be
With increasing temperature NH3 desorbed from the sur-
face and the intensity of the signals decreased. No chemi-
TABLE 5
Catalyst Screening Tests for the Amination of 1,3-Propanediola
Selectivity (% )
Temperature
( C)
Molar ratio
diol/NH3/H2
Conversion
(% )
Catalyst
Aminol (3)
Diamine (4)
Cob
210
210
195
190
210
210
210
195
195
195
1/20/2
1/20/2
1/60/2
1/20/2
1/20/2
1/60/2
1/20/2
1/60/2
1/60/2
1/60/2
90
58
98
36
80
98
94
95
42
96
6
26
5
31
12
5
0
9
14
<3
11
21
12
12
11
23
11
34
7
Coc
Cob
Co–La
Co–La
Co–La
Co–Fe
Co–Fe
Co–Fe-48d
Co–Fe-48e
<3
1
a Conditions: 135 bar, contact time: 60,000 gs mol
.
b Values were determined after 3 h time on stream.
c Values were determined after 12 h time on stream.
1
d Contact time: 40,000 gs mol
e Contact time: 100,000 gs mol
.
1
.

Co-CATALYZED AMINATION OF 1,3-PROPANEDIOL
379
reproduced after 6–12 h on stream. Accordingly, Co–La and pounds (5 and 6). These aldehydes could not be detected
Co–Fe were chosen for further investigations.
in the liquid product mixture, likely due to their high reac-
tivity and volatility. Their existence was deduced from the
formation of alkylation products 9 and 10. The ratio of main
to by-products varied strongly with the reaction conditions,
as shown below.
Product distribution. On the basis of the major com-
ponents identified in the liquid product mixture by GC–
MS analysis, the important reactions occurring during am-
ination of 1,3-propanediol are depicted in Scheme 2. Be-
sides the key intermediate amino alcohol 3 and product
Influence of reaction parameters. Figure 5 depicts the
diamine 4, various compounds have been detected that influence of temperature on the conversion of 1,3-propane-
formed by fragmentation (7, 8), alkylation (9, 10, 14), and diol and on the yields of some key products over Co–Fe.
dimerization (11–13). The retro-aldol reaction via the
-
The conversion of 1,3-propanediol, which indicates the
hydroxyaldehyde 2 (21) produced reactive carbonyl com- consumption of the diol via the formation of a -hydroxy-
FIG. 5. Influence of temperature on the conversion of 1,3-propanediol (a) and on the yields of 3 (b), 4 (c), and degradation products 7–10 (d) over
Co–Fe, Co–Fe–Na, and Co–Fe–PO4. Standard conditions.

380
FISCHER ET AL.
FIG. 6. Effect of total pressure on the amination of 1,3-propanediol
over Co–La. Conditions:210 C, reactant molar ratio (diol/NH3/H2):1/20/2;
FIG. 7. Influence of NH3/diol molar ratio on the conversion of 1,3-
propanediol and product distribution over Co–La. Conditions: 210 C, con-
tact time: 60,000 gs mol 1; otherwise standard conditions.
1
contact time: 60,000 gs mol
.
Product distribution was strongly affected by the contact
time, as shown in Fig. 8. Longer contact times increased the
conversion and decreased the selectivities for 3 and 4. How-
ever, the yields of these products (i.e., conversion times se-
lectivity) remained almost constant. Interestingly, working
with a short contact time at higher temperature (20,000 gs
mol 1, 210 C, see Fig. 8), or with a longer contact time at
lower temperature (60,000 gs mol 1, 190 C, see Table 5)
aldehyde 2, increased by an order of magnitude in the tem-
perature range 150–180 C. The variation in the yields of
amino alcohol 3, diamine 4, and degradation products 7–10
with temperature is typical of a consecutive reaction series.
The effect of total pressure on the product distribution is
illustrated with the results obtained over the Co–La catalyst
(Fig. 6). There is a remarkable enhancement in selectivities
to 3 and 4 in the range of the subcritical–supercritical tran-
sition of NH3 [Pc = 114.8 bar (30)]. The change in selectivity
was minor below 100 bar and above 135 bar, and the varia-
tion in conversion was small over the whole pressure range.
It has been proved independently that at 200 C and 130 bar
the reaction mixture was in a homogeneous supercritical
phase.
The influence of NH3/diol molar ratio is shown in Fig. 7.
The higher this ratio, the higher are the conversion of diol
1 and the cumulative selectivity for aminol 3 and diamine
4. Simultaneously, the formation of dimers 11–13 was di-
minished, which is illustrated by the example of 13. As dis-
cussed in the Introduction, the reactivity (basicity) of NH3 is
markedly lower than that of the primary amines produced.
This reactivity difference leads to the formation of various
dimerization and oligomerization products (the aldehyde
intermediate reacts with an amine, instead of NH3, to pro-
duce a secondary amine). A relatively large NH3/diol molar
ratio (>20) is necessary to compensate this effect and im-
prove the selectivities to primary amines. A large excess
of NH3 also favors the condensation of NH3 with the alde-
hyde intermediate 2 by shifting the equilibrium toward the
adduct, and suppresses the disproportionation of primary
amines to secondary amines and NH3 (1, 4, 12).
FIG. 8. Effect of contact time on conversion of 1,3-propanediol (1)
and selectivities for 3-amino-1-propanol (3) and 1,3-diaminopropane (4)
over Co–La. Conditions:210 C, reactant molar ratio (diol/NH3/H2):1/20/2;
otherwise standard conditions.

Co-CATALYZED AMINATION OF 1,3-PROPANEDIOL
381
cohols phosphates require markedly higher temperatures
than metal catalysts (44, 45).
Treatment with NaOAc was less detrimental to the for-
mation of the amino alcohol intermediate, but the diamine
yield dropped above 180 C due to various side reactions,
mainly degradation (Fig. 5d).
DISCUSSION
Influence of Supercritical Ammonia
The one-step amination of 1,3-propanediol (1) to 1,3-di-
aminopropane (4) was studied over Co-based catalysts
under conditions where ammonia forms a supercritical
medium (above 132.4 C and 114.8 bar). At best, a 95 wt%
Co–5 wt% Fe catalyst (Co–Fe) afforded 32% diamine yield
at almost complete conversion. Compared with the ami-
nation of simple monofunctional alcohols with NH3 over
metal catalysts, this result is rather moderate (8, 46). How-
ever, excluding our work, there is no other report in the
open literature on the successful amination of 1. The likely
reason is not lack of interest in the product diamine, which
is a useful and versatile intermediate (3). Rather the com-
plexity of the process—eight major reaction steps including
many more elementary steps—hampers any easy success in
the production of 4.
The application of scNH3 as reactant and reaction
medium has been found to be essential for obtaining rea-
sonable selectivities. A change from subcritical to supercrit-
ical conditions by increasing the total pressure almost dou-
bled the selectivity to the amino alcohol intermediate 3 and
diamine 4 over Co–La (Fig. 6). An even more pronounced
influence of pressure was observed over the more selective
Co–Fe catalyst (25). The existence of a single supercriti-
cal phase (instead of liquid and gas phases) eliminated the
interphase mass transfer resistance (47, 48). It is very likely
that the positive changesin amination selectivityin the near-
critical region are due to the increased surface NH3 concen-
tration. Analysis of the main and side reactions in Scheme 2
suggests that high surface NH3 concentration favors the de-
sired reaction series leading to 4 and suppresses several side
reactions including dimerization, oligomerization, dispro-
portionation, retro-aldol reaction, and hydrogenolysis.
FIG. 9. Influence of temperature on the amination of 3-amino-1-
propanol (3) over Co–Fe. Standard conditions.
afforded almost identical conversions (34–36% ) and cumu-
lative selectivities for 3 and 4 (43% ). Attempts to improve
the diamine yield at high diol conversion by varying either
the temperature or the contact time resulted in rather sim-
ilar values.
Amination of 3-amino-1-propanol (3). Each interme-
diate and the product amine can take part in various side
reactions, as shown in Scheme 2. In a series of experiments
the amination of the key intermediate 3-amino-1-propanol
(3) was investigated. Starting from this intermediate instead
of 1, the number of necessary reaction steps is halved and
the selectivity to the diamine 4 is expected to rise consider-
ably. The results shown in Fig. 9 confirm this expectation.
The maximal yield of 50% 4 was remarkably higher than the
best value of 32% achieved under similar conditions in the
amination of the diol 1 (Fig. 5c). In both reactions, in the am-
ination of 1 and 3, the drop in diamine yield above 210 C
is due mainly to the enhanced formation of degradation
products (see Fig. 9 as an example).
Catalytic performance of Co–Fe treated with NaOAc or
(NH4)2HPO4. The effect of basic and acidic treatments of
Co–Fe on propanediol conversion and product distribution
Promotion of Co with La and Fe
The proper choice of catalyst was also important in
is shown in Fig. 5. Introduction of acidic sites by the phos- achieving reasonable and reproducible yields of 4. The best
phate treatment resulted in significant deactivation. A clear catalyst was developed by promoting Co with 5 wt% Fe.
indication is the lower conversions by 20–60% achieved at La as additive was less efficient. The stability of the Co–Fe
150–210 C, as compared with the performance of unmodi- catalyst is remarkable: no significant deactivation was ob-
fied Co–Fe (Fig. 5a). Similarly, the maximum in yields for 3 served even after 10 days of use at 150–210 C.
and 4 was shifted to higher temperatures ( 210 ) (Figs. 5b,
Only the metastable -Co phase formed during reduc-
5c). A possible explanation for the observed deactivation tion prior to the amination reaction, and no detectable
is that in part, metal phosphates were formed on the cata- restructuring occurred over several days on stream. For
lyst surface during calcination, and in the amination of al- comparison, unpromoted Co, which was less selective and

382
FISCHER ET AL.
to 4 includes metal-catalyzed redox and acid/base-catalyzed
condensation (addition–elimination) reactions. A former
investigation of the amination of monofunctional aliphatic
alcohols (49) demonstrated that the overall reaction rate
was limited by the abstraction of an -hydrogen. This is the
first elementary step during dehydrogenation of the alcohol
to the corresponding aldehyde. Assuming that the amina-
tion of an alkanediol can be divided into two “independent”
amination reactions, we can conclude that the presence of
strong acidic or basic sites cannot increase the overall rate
of the 1 → 4 transformation and, therefore, cannot improve
the selectivity to 4. On the contrary, these acidic and ba-
sic sites can accelerate the retro-aldol reaction leading to
5 and 6, the alkylation reactions resulting in 7–10 and 14,
and dimerization and oligomerization reactions affording
nonvolatile by-products.
The preparation conditions for Co, Co–Fe, and Co–La
(choice of precursors and precipitating agent, deposition
at neutral pH, etc.) were designed with the aim of mini-
mizing the concentrations of strong acidic and basic sites
on the surface of the final catalyst. It was hoped that the
model catalysts Co–Fe–Na and Co–Fe–PO4 would differ
from Co–Fe only by the presence of surface basic and acidic
sites and would enable us to confirm the above concept.
Unexpectedly, even the minor changes during preparation
[adsorption of NaOAc or (NH4)2HPO4 from dilute aqueous
solutions, followed by careful removal of excess salts with
hot water] resulted in remarkable changes in the physico-
chemicalproperties. The different properties(excludingthe
acid/base properties) of Co–Fe after modification are illus-
trated in Tables 2 and 3 and Figs. 1–3. It is probable that the
poor catalytic performance of Co–Fe–Na and Co–Fe–PO4
(Fig. 5) is partly due to these “structural” effects. Especially
in the case of Co–Fe–PO4, the significant amount of -Co
phase and the partial transformation of surface oxide to
inactive phosphate during calcination may also contribute
to the detrimental influence of phosphate on the catalyst
surface.
SCHEME 2. Important reactions occurring in the amination of 1,3-
propanediol with ammonia. The intermediates and products (except 2, 5,
and 6) were identified by GC–MS.
deactivated within a few hours, contained both - and -Co
phases, and the proportion of -Co phase increased parallel
with the loss of activity. Moreover, both - and -Co phases
existed in Co–La and Co–Fe treated with acidic or basic
salts. These three catalysts afforded considerably lower se-
lectivities and yields of 4. A further increase in iron content
from 5 to 48 wt% (Co–Fe-48) led to the formation of a Co–
Fe alloy (Wairauite), and this catalyst was also less efficient
than Co–Fe. Apparently, the small amount of Fe additive
can stabilize the metastable -Co phase, and the presence of
this single phase in the bulk seems to be a key requirement
for obtaining a stable and selective catalyst.
CONCLUSION
The amination of 1,3-propanediol with ammonia was car-
ried out in a continuous fixed-bed reactor in the tempera-
ture range 150–210 C and total pressure range 50–150 bar.
Promotion of the unsupported cobalt catalyst with iron
or lanthanum improved diamine selectivity significantly.
Best catalytic results were obtained with a 95 wt% Co–
5 wt% Fe catalyst containing only very weak acidic sites.
Role of Surface Basic and Acidic Sites
Treatment of the best catalyst Co–Fe prior to drying and The active metastable -cobalt phase tends to transform
calcination with aqueous NaOAc or (NH4)2HPO4 was to the thermodynamically more stable -cobalt phase un-
found to be detrimental to catalytic performance with re- der reaction conditions. Addition of a small amount of
spect to a broad range of reaction parameters. A feasible ex- iron suppresses this transformation, providing stable cobalt
planation for the negative influence of strong acidic and ba- catalysts. The absence of strong acidic and basic sites on
sic surface sites is based on a careful analysis of the reactions the catalyst was found to be crucial for the production
shown in Scheme 2. The main reaction path leading from 1 of 1,3-diaminopropane and for the suppression of various

Co-CATALYZED AMINATION OF 1,3-PROPANEDIOL
383
undesired reactions (retro-aldol reaction, hydrogenolysis, 14. Winderl, S., Haarer, E., Corr, H., and Hornberger, P., U.S. Patent
3,270,059 (1966).
15. Adam, K., and Haarer, E., U.S. Patent 3,520,933 (1970).
16. Boettger, H., Hoffmann, H., Toussaint, H., and Winderl, S., U.S. Patent
4,014,933 (1977).
17. Best, D. C., U.S. Patent 4,111,840 (1978).
alkylation, disproportionation, dimerization, oligomeriza-
tion) occurring in the reaction system. The use of supercrit-
ical ammonia as solvent and reactant proved to be benefi-
cial for formation of both amino alcohol and diamine. This
selectivity enhancement in the near-critical region may be 18. Best, D. C., U.S. Patent 4,123,462 (1978).
19. Jobson, E., Baiker, A., and Wokaun, A., J. Mol. Catal. 60, 399 (1990).
attributed to elimination of the interfacial mass transfer
leading to a higher surface concentration of ammonia.
The feasibility of a one-step synthesis of 1,3-diamino-
propane from 1,3-propanediol in supercritical ammonia has
been demonstrated. Considering the large number of ele-
mentary steps involved in this consecutive reaction series,
the best yields of 32% for the diamine and 8% for the valu-
able intermediate amino alcohol are attractive values. The
excellent stability of the best Co–Fe catalyst—no significant
deactivation up to 10 days on stream—is also a promising
feature of the process. However, a substantial further im-
provement in catalyst composition is required to achieve
yields useful for technical application.
20. March, J., “Advanced Organic Chemistry—Reactions, Mechanisms,
and Structure,” p. 898. Wiley, New York, 1992.
´
21. Sirokma´n, G., Molna´r, A., and Bartok, M., J. Mol. Catal. 19, 35 (1983).
22. Kijenski, J., Niedzielski, P. J., and Baiker, A., Appl. Catal. 53, 107
(1989).
23. Fischer, A., Mallat, T., and Baiker, A., Catal. Today 37, 167 (1997).
24. Vultier, R. E., Baiker, A., and Wokaun, A., Appl. Catal. 30, 167 (1987).
25. Fischer, A., Mallat, T., and Baiker, A., Angew. Chem. Int. Ed. 38, 351
(1999).
26. Smiley, R. A., in “Ullmann’s Encycl. Ind. Chem.” (B. Elvers,
S. Hawkins, M. Ravenscroft, J. F. Rounsavielle, and G. Schulz, Eds.),
Vol. A12a, p. 629. Chemie, Weinheim, 1985.
27. Carter, G. C., Doumaux, A. R., Kaiser, S. W., and Umberger, P. R., in
“Kirk–Othmer Encycl. Chem. Technol.” (F. M. Mark, D. F. Othmer,
C. G. Overberger, and G. T. Seaborg, Eds.), Vol. 8, p. 74. Wiley, New
York, 1992.
ACKNOWLEDGMENTS
28. Maciejewski, M., Mu¨ller, C. A., Tschan, R., Emmerich, W. D., and
Baiker, A., Thermochim. Acta 295, 167 (1997).
Thanks are due H. Bruder of Hoffmann–LaRoche Ltd., Kaiseraugst,
Switzerland, for studies of the phase behavior of the reaction mixtures.
Financial support from Lonza Ltd, Visp, Switzerland, is kindly acknowl-
edged.
29. Maciejewski, M., Emmerich, W. D., and Baiker, A., J. Therm. Anal.
Cal., in press.
30. Allamagny, P., “Encyclopedie de Gaz,” p. 951. L’Air Liquide, Elsevier,
Amsterdam, 1976.
31. Pelavin, M., Hendrickson, D. N., Hollander, J. M., and Jolly, W. L.,
J. Phys. Chem. 74, 1116 (1970).
REFERENCES
32. McIntyre, N. S., and Cook, M. G., Anal. Chem. 47, 2208 (1975).
1. Glaser, H., in “Methoden der Org. Chem. (Houben-Weyl)” (E. Mu¨ller, 33. Hirokawa, K., Honda, F., and Oku, M., J. Electron Spectrosc. 6, 333
Ed.), Vol. XI/1, p. 112. Georg Thieme, Stuttgart, 1957.
(1975).
2. Herman, R. G., in “Catalytic Conversion of Synthesis Gas and Al- 34. Wagner, C. D., Davis, L. E., Zeller, M. V., Taylor, J. A., Raymond,
cohols to Chemicals” (R. G. Herman, Ed.), p. 433. Plenum, New
York/London, 1984.
3. Heilen Mercker, H. J., Frank, D., Reck, R. A., and Ja¨ckh, R., in
“Ullmann’s Encycl. Ind. Chem.” (B. Elvers, S. Hawkins, M.
Ravenscroft, J. F. Rounsaville, and G. Schulz, Eds.), Vol. A2, p. 23.
Chemie, Weinheim, 1985.
4. Baiker, A., and Kijenski, J., Catal. Rev. Sci. Eng. 27, 653 (1985).
5. Vogt, P. F., and Gerulis, J. J., in “Ullmann’s Encycl. Ind. Chem.”
(B. Elvers, S. Hawkins, M. Ravenscroft, J. F. Rounsaville, and G. Schulz,
Eds.), Vol. A2, p. 37. Chemie, Weinheim, 1985.
R. M., and Gale, L. H., Surf. Interface Anal. 3, 211 (1981).
35. Castner, D. G., and Watsin, P. R., in “Catalyst Characterization Sci-
ence: Surface and Solid State Chemistry, Philadelphia, Pennsylvania,
August 26–31, 1984” (M. L. Deviney and J. L. Gland, Eds.), Vol. 288,
p. 144. ASC, Washington, DC, 1985.
36. Hofer, L. J. E., and Peebles, W. C., J. Am. Chem. Soc. 69, 2497 (1947).
37. Emmett, P. H., and Shultz, J. F., J. Am. Chem. Soc. 51, 3249 (1929).
38. Fukuda, Y., and Rabalais, J. W., J. Electron Spectrosc. Relat. Phenom.
25, 237 (1982).
39. Davydov, A. A., “Infrared Spectroscopy of Adsorbed Species on the
Surface of Transition Metal Oxides.” Wiley, Chichester 1990.
6. Deeba, M., Ford, M. E., and Johnson, T. A., in “Catalysis of Organic
Reactions” (D. W. Blackburn, Ed.), Vol. 40, p. 241. Marcel Dekker, 40. Jobson, E., Dissertation, Laboratory of Technical Chemistry, ETH-
New York, 1990.
No. 8974, Zu¨rich, 1989.
7. Roundhill, D. M., Chem. Rev. 92, 1 (1992).
8. Turcotte, M. G., and Johnson, T. A., in “Kirk–Othmer Encycl. Chem.
41. Jobson, E., Baiker, A., and Wokaun, A., J. Chem. Soc. Faraday Trans.
86, 1131 (1990).
Technol.” (F. M. Mark, D. F. Othmer, C. G. Overberger, and G. T. 42. Ramis, G., Busca, G., Lorenzelli, V., and Forzatti, P., Appl. Catal. 64,
Seaborg, Eds.), Vol. 2, p. 369. Wiley, New York, 1992.
243 (1990).
9. Visek, K., in “Kirk–Othmer Encycl. Chem. Technol.” (F. M. Mark, 43. Baiker, A., Monti, D., and Fan, Y. S., J. Catal. 88, 81 (1984).
D. F. Othmer, C. G. Overberger, and G. T. Seaborg, Eds.), Vol. 2, 44. Labadie, J. W., and Dixon, J. J., J. Mol. Catal. 42, 367 (1987).
p. 405. Wiley, New York, 1992.
45. Ford, M. E., and Johnson, T. A., in “Catalysis of Organic Reactions”
(D. W. Blackburn, Ed.), Vol. 40, p. 219. Marcel Dekker, New York,
1989.
46. Corbin, D. R., Schwarz, S., and Sonnichsen, G. C., Catal. Today 37, 71
(1997).
47. Savage, P. E., Gopalan, S., Mizan, T. I., Martino, C. J., and Brock,
E. E., AIChE J. 41, 1723 (1995).
48. Baiker, A., Chem. Rev. 99(2), 453 (1999).
49. Baiker, A., Caprez, W., and Holstein, W. L., Ind. Eng. Chem. Prod.
Res. Dev. 22, 217 (1983).
10. Amini, B., in “Kirk–Othmer Encycl. Chem. Technol.” (F. M. Mark,
D. F. Othmer, C. G. Overberger, and G. T. Seaborg, Eds.), Vol. 2,
p. 426. Wiley, New York, 1992.
11. Baiker, A., in “Catalysis of Organic Reactions” (J. R. Kosak and T. A.
Johnson, Eds.), Vol. 53, p. 91. Marcel Dekker, New York, 1994.
12. Mallat, T., and Baiker, A., in “Handbook of Heterogeneous Catalysis”
(G. Ertl, H. Kno¨zinger, and J. Weitkamp, Eds.), Vol. 5, p. 2334. VCH
Verlagsgesellschaft, Weinheim, 1997.
13. Schreyer, R., U.S. Patent 2,754,330 (1952).