Applied Catalysis A: Disproportionation of N-Propylamine On Supported Nickel Catalysts
Applied Catalysis A: Disproportionation of N-Propylamine On Supported Nickel Catalysts
Applied Catalysis A: Disproportionation of N-Propylamine On Supported Nickel Catalysts
catalysis
A
Applied Catalysis A: General 109 ( 1994) 263-275
Abstract
1. Introduction
In the selective hydrogenation of nitriles for the production of primary amines, the
occurrence of undesired condensation reactions leading to secondary and tertiary amines is
*Corresponding author. Tel. ( f 31-2246) 4524, fax. ( + 31-2246) 3615, e-mail VERHAAK@ECN.NL.
a major problem. It has been demonstrated that acidic sites on the catalyst support promote
the extent to which these reactions take place [ 1,2]. The use of a basic support and,
preferentially, the application of additives to supports commonly used in catalysis (‘y-
alumina and aluminosilicate) to reduce their intrinsic acidity are procedures that have been
adapted successfully to enhance the selectivity toward primary amines. If a catalyst is highly
selective toward primary amines, the condensation and deamination reactions leading to the
higher substituted di- and trialkylamines are apparently suppressed. The catalysts, as used
in the hydrogenation of nitriles [ 1,2], were therefore tested also with respect to their
performance in the condensation of a primary amine, i.e. the disproportionation of n-
propylamine.
At first sight the reaction seems to be relatively simple; the reaction of two propylamine
molecules yields dipropylamine and ammonia, as shown in the overall reaction 1:
2. Experimental
The supports and the supported nickel catalysts were prepared as described previously
[ 31. In short, precursors of the silica-supported nickel catalysts were prepared either by the
decomposition of urea at 90°C in a nickel containing aqueous suspension of silica or by wet
impregnation of a silica support (fumed silica, Aerosil200 from Degussa AG, surface area
about 200 m2/g) with an aqueous nickel nitrate solution.
A series of K-modified y-alumina samples was prepared by wet impregnation of pow-
dered y-alumina ( Aluminumoxide-C from Degussa AG, surface area about 100 m2/g) with
solutions containing ,various concentrations of KN03. Promoted alumina supports contain-
ing either sodium, calcium, or zinc were prepared in a similar way. The samples were dried
at 120°C and calcined for 5 h at 550°C. A series of supported nickel catalysts was obtained
from the promoted aluminas by wet impregnation of the support with a solution of nickel
nitrate. The unpromoted nickel catalysts were prepared by wet impregnation of several
calcined, alumina-based supports with an aqueous nickel nitrate solution. This way nickel
was applied onto y-alumina, aluminosilicate (Engelhard De Meern, product code A797-
M.J.F.M. Verhaak et al. /Applied Catalysis A: General 109 (1994) 263-275 265
06-002-01, surface area about 225 m2/g) and aluminophosphate (surface area about 125
m2/g, 131).
Magnesia-supported nickel catalysts were prepared through the precipitation of a nickel
magnesium mixed oxalate using the injection of ammonium oxalate into a solution con-
taining both nickel and magnesium nitrate at 50°C. The molar amount of ammonium oxalate
injected was equal to the molar amounts of nickel and magnesium together.
Unpromoted catalysts are encoded by appending the nickel weight loading to the first
character of the support (S for silica, A for alumina, M for magnesia, AS for aluminosilicate,
and AP for aluminophosphate). Promoted catalysts are encoded by adding the atomic
symbol of the promoting element to the code of the corresponding unpromoted catalyst.
The nickel loading was adjusted to 5 wt.-% in all catalysts (based on a complete reduc-
tion), except for the catalysts encoded S25U (U for urea decomposition) and M25 that
were loaded with 25 wt.-% of nickel [ 31. Samples of catalyst S25U were diluted with silica
to obtain formally a 5 wt.-% Ni-loaded catalyst in reduced form. To that end the dried
precursor was powdered in a mortar and diluted with Aerosil200 in the ratio precursor-to-
silica of l/4. The code of this catalyst is S25U-d. Catalyst M25 was diluted with magnesia
(Merck, p.a.) to obtain formally a 5 wt.-% Ni-loaded catalyst after reduction. The encoding
of this catalyst is M25-d.
The catalyst precursors were air-dried at 120°C overnight and subsequently powdered.
The powdered (diluted) precursors were pressed at 1000 kg/cm*, crushed, and finally
sieved. The sieve fraction 0.50-0.85 mm was used in all kinetic experiments.
two measurements. The same procedure was followed with decreasing bed temperatures.
The Arrhenius measurements were always performed in pure hydrogen saturated with 5.6
vol.-% n-propylamine.
It appeared that the selectivity of the process as well as the activity and stability of the
catalyst were influenced by the hydrogen concentration. To obtain information on the
influence of the hydrogen concentration on these parameters, also several isothermal exper-
iments were performed. In the isothermal measurements the reactor temperature was first
raised from room temperature to 150°C at a rate of lO”C/min. When this temperature was
attained, the catalytic performance was followed as a function of process time in various
hydrogen concentrations. Data points were collected every 20 min. A typical run consisted
of a gas-chromatographic (GC) analysis of a product sample (propylamine,,,) and of a
reference sample, withdrawn from the reactant gas entering the reactor (propylamine,,,) .
The conversion is defined as:
propylamine,,,
Conversion (%) = 1 - . 100%
( propylamine,,, 1
Product selectivities have been calculated from peak areas corrected for different response
factors in the flame ionization detector and normalized to moles of product. Since the carbon
balance was always near lOO%, the selectivity for product i is defined as:
Corrected area’
Selectivity’ (mol-%) = . 100%
Summed area of all peaks
In Table 1 the general test conditions adapted in the kinetic measurements are summarized.
Prior to the catalytic experiments it was verified that the bare, nickel-free supports were
inactive in the disproportionation reaction. In the temperature range 20-200°C none of the
supports was reactive. No experiments were performed above 200°C. Preliminary experi-
ments ruled out the existence of transport limitations. To that end the usual experimental
Table 1
General test conditions for the disproportionation of n-propylamine
checks were performe’d [ 51. The sieve fractions 0.212-0.50 mm and 0.50-0.85 mm pro-
duced identical conversion patterns indicating the absence of intra-particle transport limi-
tations. Variation of the bed depth and gas flow-rate at constant space velocity also resulted
in similar conversion patterns, and thus inter-particle heat and mass transfer limitations
were absent too.
The silica-supported nickel catalyst S25U-d was used to study the influence of the
hydrogen partial pressure on the activity, selectivity, and stability in the disproportionation
of n-propylamine. Only one sample of the catalyst was employed and between two meas-
urements the nickel phase was “cleaned” by treatment with pure hydrogen at 250°C for 2
h [ 61. Fig. 1 shows the conversion of n-propylamine at 150°C as a function of time measured
at several hydrogen partial pressures, ranging from 0 to 0.9 bar.
The data in Fig. 1 demonstrate that a lower hydrogen concentration resulted in increas-
ingly higher initial and final levels of activity. However, when the hydrogen concentration
was reduced to zero, the conversion of propylamine showed a severe drop. The removal of
all hydrogen from the feed resulted in a great loss of activity as well as an increased rate of
deactivation. While only in pure hydrogen the extent of deactivation was not significant (as
can be seen from the minor tilt of the curve), the extent of deactivation increased significantly
at lower hydrogen concentrations. The observation that the extent of deactivation is reduced
by the application of a sufficiently high hydrogen partial pressure agrees with the results
presented in other work [ 61. From the data presented in Fig. 1 it may be concluded that the
rate equation for disproportionation of n-propylamine has a negative order in hydrogen, i.e.
the reaction rate increases with decreasing hydrogen partial pressures. However, the pres-
268 M.J.F.M. Verhaak et al. /Applied Catalysis A: General 109 (1994) 263-275
10
1 94.4% 63.0% j 50.4% i 31.4% / 0%
75.6% i
0, . I .
0 10 20 30 40 50
Time (hours)
Fig. 1, Influence of the hydrogen concentration on the activity and stability of catalyst S25U-d in the dispropor-
tionation of n-propylamine as a function of time-on-stream. Numbers between lines denote the hydrogen concen-
tration.
ence of (some) hydrogen is needed to maintain activity, because in pure nitrogen both a
comparatively low catalytic activity and a fast rate of deactivation resulted.
At first sight it seems quite unexpected that the disproportionation reaction is sensitive
to changes in the hydrogen concentration. In the overall reaction equation for propylamine
disproportionation ( see reaction 1) hydrogen is not involved. The reason for this apparently
unusual behaviour will be given with the aid of the following experiments. The conversion
of n-propylamine was followed isothermally at 150°C in pure hydrogen for several hours
just like the previou,s measurement. In this experiment, however, the propylamine-saturated
hydrogen feed was replaced by propylamine-saturated nitrogen after approximately 10 h of
operation. Then, after another 10 h, the nitrogen stream was again replaced by hydrogen.
In between the gas switches the catalyst was not cleaned in hydrogen. Fig. 2 shows the
conversion pattern resulting from this experimental sequence.
The left-hand part of Fig. 2 displays the conversion of n-propylamine in a hydrogen-
containing gas stream. It is obvious that the catalyst exhibited a stable conversion level of
about 40%. The switch to nitrogen resulted in a severe drop of the conversion, as was
observed in the previous figure. When the carrier gas was switched back to hydrogen, a
steady conversion level was again attained, as shown in the right-hand part of Fig. 2. It
seems, however, that the catalyst had lost some of its activity, since the conversion level
shifted to the slightly lower value of 37%.
Studying the catalytic amination of long-chain aliphatic alcohols over copper catalysts,
Baiker and Richarz 1[7,8], and Baiker [ 91 observed a drop in the conversion of I-dodecanol
when hydrogen in the feed was replaced by nitrogen. Similar to our results, the catalytic
activity was recovered when the gas stream was switched back to hydrogen. They ascribed
their findings to the formation of copper nitride during the measurement in nitrogen. The
formation of copper nitride resulted in lower conversions. Based on the results on the
stability of nickel nitride in hydrogen [ 31, we assume that the presence of hydrogen brought
about a smooth tranlsformation of copper nitride into metallic copper with the concomitant
rise in activity. In the amination of 1-dodecanol with dimethylamine [ 81, the dispropor-
M.J.F.M. Verhaak et al. /Applied Catalysis A: General 109 (1994) 263-275 269
0 5 10 15 20 25 30 35
Time (hours)
Fig.2. Conversion of n-propylamine at 150°C in feeds of different composition. Data in left- and right-hand plot
were collected in 5.6% propylamine/H,, whereas data in the middle plot were collected in 5.6% propylamine/
N2.
tionation of the amine during the process was mentioned. The ammonia liberated in this
reaction was held responsible for the formation of copper nitride. Although nitride formation
occurs on exposure of either nickel or copper to ammonia at elevated temperatures in the
absence of hydrogen, it will be demonstrated that nitride formation is not solely responsible
for the drop in catalytic activity with nickel-based catalysts. To that end TF’SR was used,
as described extensively elsewhere [ 6,10,11].
After catalyst S25TJ-d had been operated in the disproportionation of propylamine at
150°C for 10 h, the sample was cooled to 100°C in the propylamine containing nitrogen-
hydrogen mixture, purged at this temperature for 30 min with pure nitrogen, and cooled to
room temperature in flowing nitrogen. Next, the temperature was raised to 400°C in 10
vol.-% H,/Ar while the hydrogen consumption was monitored with a thermal conductivity
detector. The resulting hydrogen consumption profile is plotted in Fig. 3.
The hydrogen consumption profile in Fig. 3 is dominated by a hydrogen consumption
peak at 185°C with a shoulder on the low-temperature side ( 152°C). Based on previous
results, the position of the peak at 185°C was assigned to the hydrogenation of surface nickel
carbide (a-carbon) [ 6,10-131.
It can be concluded that during the disproportionation of n-propylamine also carbon
deposition occurred, .in accordance with the results obtained with the hydrogenation of
acetonitrile [ 61. Because nickel nitride is not formed in the presence of hydrogen, it can be
concluded that the observed loss of activity with time-on-stream in the presence of hydrogen
(see Fig. 1) was caused exclusively by carbon deposition. During the measurements in
nitrogen both nickel nitride as well as extensive nickel carbide formation took place. This
explains why the catalytic activity did not regain its initial value after switching back to
hydrogen. The nickel carbide formed in nitrogen could not be removed completely in
hydrogen, since the measurement was performed at 15O”C, which is too low a temperature
to regenerate a nickel catalyst deactivated by carbon deposition [ 61. Nickel nitride, on the
other hand, is hydrogenated completely to nickel and ammonia at this temperature [ 31.
270 M.J.F.M. Verhaak et al. /Applied Catalysis A: General 109 (1994) 263-275
Temperature (“C)
Fig. 3. TPSR experiment in 10% H,/Ar with catalyst S25U-d after previous operation in the disproportionation
of n-propylamine at 150°C. The plot displays the hydrogen consumption (measured with a hot wire detector) as
a function of the catalyst temperature.
The selectivity of catalyst S25U-d as a function of time measured during the experiments
represented in Fig. 2 is shown in Fig. 4. This figure clearly shows that the drop in activity
after the hydrogen in the feed had been replaced by nitrogen was accompanied by a severely
decreased selectivity toward dipropylamine (DPA) . In hydrogen, a high selectivity toward
DPA was observed (more than 98 mol-%), whereas in nitrogen the DPA-selectivity dropped
to the rather low value of about 10 mol-%. When hydrogen was again substituted for
nitrogen, the DPA-selectivity recovered. Besides DPA, two other compounds were formed
in the hydrogen-lacking environment. Propylidene imine (encoded SCB- 1 in the legend,
r- j I
DPA
v SCB- I
P
-A 80 - SCB-2
iii8
@s!
a0
‘S
.- E
.? - 60 -
H . Hydrogen
f
40 -
0 5 10 15 20 25 30 35
Time (hours)
Fig. 4. Selectivity toward the products detected in the disproportionation of n-propyl-amine at 150°C over catalyst
S25U-d in feeds of different composition. Data in left- and right-hand plot were collected in 5.6% propylamine/
Hz, whereas data in the middle plot were collected in 5.6% propylamine/N,. DPA = dipropylamine and SCB-I
(SCB-2) = Schiff base 1 (2).
M.J. F.M. Verhaaket al. /Applied CatalysisA: GeneralIO9(1994) 263-275 271
CH,CH,CH=NH + CH,CH,CH2NH, =
0H’ H
CH,CH,CH-ti-CH,CH&H,
AH,
T
CH,CH,CH-N-CH,CH,CH, e CH&H$H=N-CH2CH,CH3 + NH,
AH,
7
CH,CH&H=N-CH2CH2CH3 + H, = CH,CH,CH,-N-CH,CH,CH,
In the previous section it was shown that during the disproportionation of n-propylamine
the presence of hydrogen was essential to hydrogenate the intermediate Schiff base formed
272 M.J.F.M. Verhaak et al. /Applied Catalysis A: General 109 (1994) 263-275
on the support. Furthermore, it was necessary to use large amounts of hydrogen in the feed
to suppress the extent of deactivation caused by carbon deposition. To minimize carbon
build-up on the catalyst during the process, all following experiments were performed in
hydrogen saturated with 5.6 vol.-% propylamine. To study the catalysts with respect to their
behaviour in the deamination of propylamine several Arrhenius measurements were per-
formed on the reduced catalysts in the temperature range of 75175°C. Fig. 6 shows the
Arrhenius plots of several catalysts as indicated in the legend of the figure. The data points
in Fig. 6 were all fitted to first order polynomials and the resulting straight lines all run in
parallel within the experimental error. This indicates that the kinetics of the reaction are the
same for all catalysts. The only difference between the various catalysts is the value of the
pre-exponential factor.
In Table 2 the parameters derived from the Arrhenius plots, the data on the acidity of the
catalysts (from ammonia desorption, taken from ref. [ 31)) and their selectivities in the
hydrogenation of acetonitrile (taken from refs. [ 1,2] ) are summarized.
Fig. 6 and Table 2 learn that the acidic catalysts all have large intercepts and that with
decreasing acidity the intercepts decrease also. For a particular catalyst, the intercept
obtained from the Arrhenius fit is thus a measure for its acidity. However, since the intercepts
are obtained only after a long, unreliable extrapolation to infinitely high temperatures, the
error in the value will be considerable. It was decided to use the conversion of n-propylamine,
measured in hydrogen at 125°C to quantify the acidity of the catalysts. In Fig. 7 the relation
between the acidity of the catalysts, as measured by the desorption of ammonia in the
temperature range of 20 to 150°C [3], as a function of the acidity as measured by the
conversion of propylamine at 125°C in hydrogen is shown.
It is obvious that with a decreasing acidity of the catalyst, as measured by the desorption
of ammonia, the conversion of propylamine decreased also. Whereas the acidic catalysts
AP5 and AS5 both converted more than 35% of propylamine at 125”C, the conversion of
propylamine over the basic catalysts AS5K-10 (containing 10 wt.-% K), ASNa, A5K-2
(containing 2 wt.-% K), A5K-3 (containing 3 wt.-% K), and M25-d was negligible ( < l-
2%).
A
AP5
.
AS5
0
S25U-d
A5
0
ASCa
. ASK-2
A5K-3
1000/T (i/K)
Fig. 6. Arrhenius plots of several catalysts in the disproportionation of n-propylamine. Catalysts as indicated in
the legend.
M.J.F.M. Verhaak et al. /Applied Catalysis A: General 109 (1994) 263-275 273
Table 2
Characterization and performance data for several nickel catalysts
4000
. ASK-IO
0 I
0 10 20 30 40 50
Fig. 7. Catalyst acidity, as measured by the rate of disproportionation of n-propylamine in hydrogen at 125°C. as
a function of the acidity, as measured by the TPD of ammonia (data from ref. [ 3I), for various supported nickel
catalysts.
The relation between the acidity of a catalyst and its activity in the disproportionation of
propylamine is not surprising since it was shown that the reaction is acid-catalyzed. We can
conclude that the activity in the conversion of (propyl)amine( s) is another tool to measure
the acidity of supported nickel catalysts. It has been demonstrated elsewhere that the
selectivity in the hydrogenation of acetonitrile was related directly to the acidity of the
catalyst, as measured by the desorption of ammonia [ l-31. A relation between the acidity
of the catalyst, as measured by its conversion of n-propylamine at 125”C, and its selectivity
toward the primary amine is therefore expected also. In Fig. 8 this relation is shown.
274 M.J.F.M. Verhaak et al. /Applied Catalysis A: General 109 (1994) 263-275
-I
0 10 20 30 40 50
Fig. 8. Relation between catalyst acidity, as measured by the rate of disproportionation of n-propylamine at 125°C
and the selectivity toward ethylamine in the hydrogenation of acetonitrile at 125°C (data from refs. Il.21 ).
Indeed, a linear relation exists. Again it is shown that the selectivity in the hydrogenation
of nitriles is determined by the acidity of the catalyst used to catalyze the reaction. It can be
concluded that the disproportionation of n-propylamine is a new and highly appropriate
method to measure the acidity of supported nickel catalysts.
4. Conclusions
The conversion of propylamine into dipropylamine was studied over a large number of
supported catalysts. By measuring the rate of this reaction at various hydrogen concentra-
tions, it was demonstrated that the reaction exhibits a negative order in hydrogen; at higher
hydrogen concentrations the conversion of propylamine decreased. In the absence of hydro-
gen, however, the conversion of propylamine showed a severe drop. It was concluded that,
although the rate was influenced negatively by hydrogen, its presence was necessary to
maintain catalytic activity and stability. In the absence of hydrogen, namely, the rate of
deactivation of the nickel catalysts was much higher than in the presence of (some)
hydrogen. Using TPSR it was shown that during operation the formation of surface nickel
carbide occurred. The application of high concentrations of hydrogen could suppress the
extent of carbon accumulation on the catalyst.
In the presence of hydrogen all catalysts were highly selective in the formation of
dipropylamine during the disproportionation of n-propylamine. However, when hydrogen
was removed from the feed, the selectivities changed significantly. Instead of DPA, mainly
unsaturated molecules, aldimines and Schiff bases, were formed. These observations could
be explained with the aid of a bifunctional mechanism and yielded further proof for the
existence of this mechanism.
The extent of conversion of propylamine was related to the acidity of the catalyst. The
existence of a linear relationship between the conversion of propylamine and the acidity of
a particular catalyst, as measured by the desorption of ammonia, proved the acid-catalyzed
M.J.F.M. Verhaak et al. /Applied Catalysis A: General 109 (1994) 263-275 215
nature of the reaction. Therefore, the high primary amine selectivity observed in the hydro-
genation of acetonitrile with basic nickel catalysts is due to the suppression of the
acid-catalyzed deamination of the product.
The disproportionation of n-propylamine was used to measure the acidity of a catalyst.
The method was applied successfully with a large number of nickel catalysts and can be
regarded as a new catalytic test reaction for the determination of the acidity of a catalyst.
Using this method for the measurement of catalyst acidity, it was shown again that the
selectivity toward ethyylamine in the hydrogenation of acetonitrile was determined by the
acidity of the catalyst.
Acknowledgement
The financial support of Engelhard Chemie b.v. (De Meern, The Netherlands) was
gratefully acknowledged.
References
1 M.J.F.M. Verhaak, A.J. van Dillen and J.W. Geus, Catal. Lett., submitted for publication.
2 M.J.F.M. Verhaak, PhD Thesis, ISBN 90-393-0243-X (1992).
3 M.J.F.M. Verhaak, A.J. van Dillen and J.W. Geus, Appl. Catal. A, 105 (1993) 251.
4 T.E. Jordan, Vapour Pressure of Organic Compounds, Interscience, New York, 1954.
5 J.R. Anderson and K.C. Pratt, Introduction to Characterization and Testing of Catalysts, Academic Press,
London, 1985.
6 M.J.F.M. Verhaak, A.J. van Dillen and J.W. Geus, J. Catal., 143 (1993) 187.
7 A. Baiker and W. Richarz, in T. Seiyama and K. Tanabe (Editors), New Horizons in Catalysis (Studies in
Surface Science and Catalysis, Vol. 7). Elsevier, Amsterdam, 1980, p. 1428.
8 A. Baiker and W. Richarz, hid. Eng. Chem., Prod. Res. Dev., 16 ( 1977) 261.
9 A. Baiker, Ind. Eng. Chem., Prod. Res. Dev., 20 (1981) 615.
10 J.G. McCarty and H. Wise, J. Catal., 57 (1979) 406.
11 H. Wise, J. McCarty and J. Oudar, in J. Oudar and H. Wise (Editors), Deactivation and Poisoning of
Catalysts, Wiley, New York, 1985, Chapter 1.
12 A. Baiker, D. Monti and Y. Song Fan, J. Catal., 88 ( 1984) 81.
13 J.R. Anderson and N.J. Clark, J. Catal., 5 ( 1966) 250.
14 J.M. Pommersheim and J. Coull, AIChE J., 17 (1971) 1075.
15 J. Volf and J. PaSek, in L. Cerveny (Editor), Catalytic Hydrogenation (Studies in Surface Science and
Catalysis, Vol. 27). Elsevier, Amsterdam, 1983, p. 105.