Znd. Eng. Chem. Prod. Res. Dev. 1961, 20, 615-618 TEMP' 269-C Pco * 303 kPo oETHANE
ETHYLENE
I
1
2
4 6 8 1 0 REACTOR PROFILE ( q . c o I 1
1
2
Figure 10. Reactor profile o f ethylene and ethane formation.
-
615
rates obtained from the initial reaction mixtures. Except at high CO conversions, the hydrocarbon product distribution was found to be independent of the PH,/Pcoin the range studied. The water-gas shift reaction plays an important role in determining the hydrocarbon product selectivity as the gases pass through the reactor. At high CO conversions, side reactions such olefin readsorption/hydrogenation and readsorption of carbon dioxide with subsequent reaction also occur.
Acknowledgment Dr. R. J. Madon of Exxon Corporation (New Jersey) supplied the iron catalyst used in this study. This contribution is greatly appreciated. Literature Cited Anderson, R. B. "Catalysis"; Vol. 4,Emmett, P. H., Ed.; Reinhold: New York,
40r
1956. Atwood, H. E.; Bennett, C. 0. Ind. Eng. Chem. Process Dev. 1979, 78,
163-1 70.
I
I I
INLET FLOW
I
I
1
2 0 4 accm
190-199.
LINE INDICATES 8 0 % CO "VERSION
1 I I
2
= 1011 kPa
PH2
I I
O0
Berty, J. M. Chem. Eng. Prog. 1974, 70, 78-84. BlyhoMer, B.; Neff, L. 0. J. Phys. Chem. 1962, 66, 1464-1469. Bohlbro, H. J. Catal. 1964, 3, 207-215. Dry, M. E.; Oosthulzen, 0.J. J. Catal. 1968, 7 7 , 18-24. Dry, M. E.; Shingles. T.; Boshoff, L. J.; Oosthuizen, G. J. J. Catal. 1969, 75,
TEMP = 269.C Pco = 303 kPa
I
,
I
1
1
J
4 6 8 1 0 1 2 REACTOR PROFILE ( g c a l l
Figure 11. Reactor profile o f COz formation.
subsequent hydrogenation is possible. It is also quite possible that the readsorbed olefins further react to form higher moIecular weight species (Dwyer and Somorjai, 1979),though this cannot be demonstrated from the results obtained. As shown in Figure 11,the reasorption of carbon dioxide with subsequent reaction also occurred at high CO conversions. Economic considerations usually demand high CO conversions, so the above results should be taken into account in plant design. Conclusion Our results demonstrate that the F-T product distribution (selectivity) cannot be evaluated by integrating the
Dry, M. E.; Shingles, T.; Boshoff. L. J. J. Catal. 1972, 25, 99-104. Dry, M. E. "The Rate of Carbon Deposition on Iron Catalyst in the FischerTropsch Reaction", unpublished manuscript supplied by author, 1980. Dwyer, D. J.; Simmons, 0.W. Surf. Sci. 1977, 64,617-632. Dwyer, D. J.; Somorjai, 0.A. J. Catal. 1978, 52, 291-301. Dwyer, D. J.; Somorjai, G. A. J. Catal. 1979, 36, 249-257. Feimer, J. L. MaSc. Thesis, Universlty of Waterloo, 1980. Hudgins, R. R. Paper 25-5,Seventh Canadian Symposium on Catalysis, Edmonton, Alberta, Canada, Oct 1960. Krebs, H. J.; Bonzel, H. P.; Gafner, kG. Surf. Sci. 1979, 88, 269-283. Madon, R. J.; Bucker, E. R.; Taylor, W. F. "Development of Improved Fischer-Tropsch Catalyst for Production of Liquid Fuels", Prepared for U.S. Department of Energy, Contract No. E(46-1b8008,July 1977. Mills, 0.A,; Steffgen, F. W. Catal. Rev.-Sci. 1973, 8(2),159-210. Podolski, W. F.; Kim, Y. G. Ind. Eng. Chem. Process. Des. Dev. 1974, 73,
415-421. Ponec, V. Catal. Rev.-Sci. 1978, 78. 151-171. Roberts, M. W.; Wood, D. R. J. Electron Spectr. Rekt. Phenom. 1977, 7 7 ,
431-437. Vannice, M. A. Catal. R e v d c l . Eng. 1976. 74, 153-191. Welsz, P. B.; Prater, C. D. Adv Cafal. 1954, 6 , 167.
Received for review September 29, 1980 Revised manuscript received April 17,1981 Accepted J u n e 12,1981
Catalytic Amination of Aliphatic Alcohols. The Role of Hydrogen as Inhibitor for Catalyst Deactivation Alfons Baiker Swiss Federal Institute of Technology (ETH), Department of Industrial and Engineering Chemistry, 8092 Zurich, Switzerland
The role of hydrogen in the catalytic amination of aliphatic alcohols is studied using the amination of ldodecanoi with dimethylamine on a supported copper catalyst as a model reaction. The experiments show that hydrogen acts as an inhibitor for catalyst deactivation. Temperature programmed desorption (TPD) and differential scanning calorimetry @SC) measurements indicate that the deactivation is caused by nitride formation on the copper surface. The nitride is formed by ammonia, originating from the simultaneously catalyzed disproportionation of the reactant amines. A similar deactivation mechanism is also likely to occur on other metallic amination catalysts, e.g., nickel and cobalt, as TPD and DSC investigations with these catalysts indicate.
Introduction In the past decade a considerable effort has been expended in the development of new catalytic processes for 0196-4321/81/1220-0615$01.25/0
the synthesis of long chain aliphatic amines. An economical way of synthesis is the catalytic amination of the corresponding aliphatic alcohols. Supported copper and 0 1981 American
Chemical Society
616
Ind. Eng.
Chem. Prod. Res. Dev., Vol. 20, No. 4,
1981
Table I. Physical Properties of Catalysts CUI
properties
i I1
Figure 1. Apparatus used for the catalyst deactivation studies: 1, tubular reactor; 2, thermostated air bath; 3, flow meters; 4, Deoxo purifier; 5, column packed with Drierite; 6, column packed with KOH pellets; 7, column packed with sodium pieces; 8, evaporator; 9, metering pump; 10, GC with gas sampling system; 11,cooler; 12, product condenser; T, thermocouples; P, pressure gauge.
cobalt catalysts proved to be very selective for this reaction, whereas nickel shows only moderate selectivity (Baiker and Richarz, 1977a). A striking feature of all recently developed continuous amination processes is that they are carried out with hydrogen in the feed. The reason for the use of hydrogen can only partly be understood by kinetic arguments and has mainly to be seen in its function as inhibitor for catalyst deactivation. In case the amination is carried out continuously and without hydrogen, a relatively fast deactivation of the supported copper catalyst is observed. The objective of the present work is to study fundamentally the cause of this deactivation and its inhibition by hydrogen. The amination of 1-dodecanol with dimethylamine is employed as a test reaction. Experimental Section The catalyst deactivation was studied in a fixed bed reactor. The apparatus (cf. Figure 1) comprised the metering system for dimethylamine, 1-dodecanol, hydrogen and nitrogen, an air bath for thermostating the reactor tube, and an analysis section. The reactor was constructed of thin wall (1.5 mm) stainless steel tubing of 20 mm inner diameter. Three stainless steel sheathed chromel-alumel thermocouples of 0.5 mm sheath diameter were used to measure the temperature in the catalyst bed. The temperature in the reactor tube could be maintained and regulated to within 1 "C. The liquid product mixture of the amination reaction was analyzed using a gas chromatograph (Perkin-Elmer 990, FID) and a 5% FFAP on Chromosorb G (HP 80/100) column, whereas for the analysis of the gaseous disproportionation products a gas chromatograph (Gow Mac 552, HWD) equipped with a 10% polyethylenimine on Poropak Q column was used. The deactivation studies were performed with a supported copper catalyst (Cu on y-alumina) which was prepared from Cu(N03),.3H20 solution and aluminum hydroxide. Catalyst preparation, pretreatment, and reduction were the same as described previously (Baiker and Richarz, 1978). The pure metal powders (Cu, Ni, and Co) used in the TPD and DSC investigations were prepared by a conventional precipitation method from the corresponding metal nitrates. Some properties of the catalysts employed are summarized in Table I.
.-
metal load, wt % 4 'I mean metal 21 particle size, nm BET surface 88 area, m2/g specific pore 0.458 volume, cm3/g geometric form spheres ( 3 1"
---P .--lo
Cu
?-alumina
Ni
CO
-
-
-
33
31
23
8
9
11
-
-
-
powder
powder
powder
The following reactant purities were quoted by the manufacturers: 1-dodecanol, >99%; dimethylamine. >97% (impurities monoethylakne and trimethylamine): The dimethylamine (>97%) was purified from carbon dioxide and water by passing it through two columns, one filled with KOH pellets and the other with pieces of sodium. Hydrogen (99.99%) was passed through a Deoxo catalytic hydrogen purifier (Engelhard Industries, Inc.) and then through a drying unit packed with Drierite. Nitrogen (99.99%) was purified by passing it through a column packed with Ascarite, a hot copper filled furnace, and finally the drying unit. The DSC curves shown in this paper are heating curves produced by a Mettler TA 2000 B instrument equipped with a data analysis unit. During the measurements the samples were kept under a purging dry nitrogen flow of 0.67 mL/s (STP) and an empty aluminum sample pan was used as reference material. The apparatus used for the TPD measurements was essentially the same as the one described by Cvetanovic and Amenomiya (1967). A mass spectrometer (Balzers, QMG 101 A) was employed for the detection of the desorbing gases. Helium (99.99%) purified by passage through a molecular-sieve trap cooled in liquid nitrogen was employed as the carrier gas. The reduced and outgassed (W4Pa, 570 K) metal catalyst samples were exposed to ammonia (100 kPa) for 2 h at 570 K. Before starting the TPD measurement the system was evacuated to loT4Pa for 1h at 493 K to remove the weakly adsorbed ammonia. For all measurements reported, the carrier gas flow was 1.67 mL/s (STP). X-ray diffraction patterns of the catalyst samples were obtained with a Norelco (Philipps) X-ray diffractometer using Cu K, radiation. The instrument was operated with a scanning speed of 0.5O/min. Corrections to the observed line breadth for instrumental broadening were made using the curves given by Rau (1963). Results and Discussion In an earlier investigation (Baiker and Richarz, 197713) it was found that if the amination reaction is carried out continuously and with hydrogen in the feed stream, the catalyst (Cu/ y-alumina) showed only a slight decrease in activity (about 2%) and no change in the selectivity after 1000 h on stream. The small loss of activity could be ascribed to sintering of the supported copper particles, as was evidenced by X-ray diffraction line broadening measurements which showed a slight increase of the copper particle size during the long term test. A completely different behavior of the catalyst is obtained for the same reaction, if the hydrogen in the feed is substituted by nitrogen, as illustrated in Figure 2. In this case a continuous decrease in catalyst activity is observed and in particular the selectivity to dimethyldodecylamine declines very rapidly. However, the deactivated catalyst starts to regenerate as soon as hydrogen is
Ind. Eng. Chem. Prod. Res. Dev., Vol. 20, No. 4, 1981 617 i o
1
d
8
I
*CONVERSloN
o SELECTIVITY
Hz 0.2
o
"
0.6 -
0.4
8
I
I
I
io
N2 H2-
20 TIME
30
,
40
50
"
i
i
"
"
.,,*
AMMONIA
0
0.2
0.4
0.8
0.6
1.0
CONVERSlON OF DIMETHYLAMINE
w
hours
Figure 2. Influence of hydrogen on activity and selectivity of Cu/yalumina catalyst during amination; molar feed rates (mol d): I-dodecanol, 1.9 X dimethylamine, 1.0 X IO4; hydrogen or nitrogen, 1.2 X IO4; catalyst load: 10 g; temperature: 510 K; total pressure: 100 kPa.
Figure 3. Product distribution observed with disproportionation of dimethylamine on Cu/y-alumina catalyst total molar feed rate: 7 x - 9 x IO4 mol s-l; molar fractions: dimethylamine, 0.1; hydrogen, 0.1; nitrogen, 0.8; catalyst load: 5 g; temperature: 473 K; total pressure: 100 kPa.
1
present in the feed again and returns to its original activity within a few hours. The origin of the observed reversible catalyst deactivation can be understood in the framework of the mechanism of the catalytic amination. Based on studies of the reaction path, it was earlier postulated (Baiker and Richaiz, 1977a) that the amination of aliphatic alcohols with a secondary alkylamine proceeds according to reaction scheme 1 on a RCHzCHzCH
-H
RCH,CHO
+HN(RI) 2 k
RCH,CHOHNIR,),
1
+HZ
0
RCHzCH,N(R,)z
2
(1)
copper catalyst. Step 1of this sequence was found to be rate determining under the amination conditions given (Caprez, 1980). The amination conducted with a,a-dideuterated 1-octanol showed a marked kinetic isotope effect and yielded no dideuterated amines in the product mixture. This result is in accordance with that reported earlier by Kliger et al. (1975) for the amination of a,adideuterated 1-octanol with ammonia on a molten iron catalyst. Reaction 1indicates that from a stoichiometric point of view, no additional hydrogen should be necessary if the reaction is carried out in a batch system. This point is substantiated by successful experiments performed without hydrogen in an autoclave (Baiker and Richarz, 1977a). In addition to the amination reaction discussed, the reactant amine can undergo disproportionation on amination catalysts under normal amination process conditions (Baiker and Richarz, 1977a). 2RzNH 6 RNHz R3N 2RNHz + RzNH + NH3
+
R3N + NH3 + RzNH + RNHZ (2) Experimental runs, carried out with dimethylamine and the Cu/ y-alumina catalyst yielded the product distributions presented in Figure 3. Figure 4 illustrates how the disproportionation activity of the Cu/y-alumina catalyst changes in case the hydrogen is substituted by nitrogen. As with the amination, a reversible catalyst deactivation is observed. From these findings, it is postulated that the specific reason for the deactivation of the copper catalyst is surface nitride formation caused by ammonia originating from disproportionation of the reactant amine (dimethylamine). NH3 + 3Cu + Cu3N + 3/2H2 (3) This postulated mechanism is consistent not only with the present results but also with earlier findings of Pommersheim and Coull (19711, who found that the rate of ethy-
4
6
8
TIME
,
hours
1
0
Figure 4. Influence of hydrogen on activity of Cu/y-alumina catalyst during disproportionation of dimethylamine; molar feed rates (mol s-'): dimethylamine, 1.7 X 10"; nitrogen, 1.1X IO4; hydrogen or nitrogen, 3.1 X catalyst load: 1g; temperature: 510 K; total pressure: 100 kPa.
lamine disproportionation on a copper catalyst strongly declined with increasing ammonia partial pressure. They reported that ammonia is strongly adsorbed and ties up some active sites which might otherwise act as hydrogenating-dehydrogenating centers. In order to confirm the postulated mechanism and to obtain some idea about possible regeneration of the deactivated catalyst with hydrogen, a series of differential scanning calorimetry (DSC) measurements were performed. For this purpose the reduced copper catalyst samples were exposed to an equimolar ammonia/nitrogen mixture at 520 K and atmospheric pressure. Figure 5a presents the DSC curves which were obtained for a reduced sample and for a sample exposed to the ammonia/nitrogen mixture. The DSC result of the reduced copper catalyst (dashed line) is in good accordance with the values for the specific heat capacity of copper reported in the literature, whereas the DSC curve of the sample which was exposed to ammonia indicates an exothermic decomposition of the copper nitride layer formed at about 660 K, according to Cu3N
-
3Cu + l/zNz
(4) The formation of nitrogen during the decomposition was confirmed by temperature programmed desorption measurements (cf. Figure 6). The number of nitrogen molecules desorbed per unit area of the copper sample amounted to 3.44 X 10l8 molecules/m2. Assuming that three prominent planes ((lll), (loo), (110)) are present in equal extent a mean number of 2.9 X 1019copper atoms/m2 is obtained. With these two numbers and the stoichiometry Cu3N one estimates that in the present case about 70% of the outermost atomic layer of copper was transformed to copper nitride.
618
Ind. Eng. Chem. Prod. Res. Dev., Vol. 20, No. 4, 1981
::1 -
0 5 t
0 4
_i
iy
;
I
0 3 600
500
700
800
TEMPERATURE , K
Figure 5. DSC measurementson copper (a) and nickel (b) catalyst previously expaed to ammonia (dashed l i e s were measured for pure reduced samples);exposure: equimolar ammonia/nitrogen mixture 100 kPa. 520 K, scan speeds: (a) 10 K/min; (b) 20 K/min; sample weights: (a) 26.8 mg; (b) 25.0 mg. 10
- 1
-
1
,- r .
I i
L
-
1 \PG
I
T-
1
7
0.0
550
600
650
700
750
800
TEMPERATURE , K
Figure. 6. Nitrogen desorption spectra of metal catalysts previously exposed to ammonia: (a) copper, (b) nickel and (c) cobalt. C/CM = normalized concentration of nitrogen in carrier gas; C M = concentration of nitrogen at peak maximum; exposure: ammonia 100 kPa, 2 h, 570 K; heating rates: (a) 11 K/min; (b) 10 K/min; (c) 6 K/min; sample weights: (a) 0.35 g; (b) 0.53 g; ( c ) 0.48 g.
Further evidence for the feasibility of surface nitride formation under the given conditions emerges from electron diffraction studies conducted by Tereao (1973) on evaporated copper films which were exposed to ammonia at temperatures from 470 to 720 K. He found that under these conditions Cu3N with a Reo3 type structure is formed. A similar deactivation mechanism is also likely to occur on other supported metal catalysts applied to the catalytic amination, in particular on those metals which show some activity for the disproportionation of the reactant amines as, for example, nickel and cobalt. The TPD studies of the ammonia adsorption on nickel and cobalt presented in Figure 6 indicate that on both
metals a surface nitride layer is formed in the presence of ammonia. Terao (1960) found, by electron diffraction studies conducted on nickel and cobalt films which were previously exposed to ammonia, that different types of nickel and cobalt nitride exist, depending on the degree of nitrogen insertion in the metal lattice. He reports for the metal nitrides of both metals a maximum nitrogen content which corresponds to Me3N. From the measured amounts of nitrogen desorbed (Ni: 8.03 X 1018,Co: 7.8 X 10l8molecules/m2, the mean number of metal atoms and the stoichiometry Me3N, one estimates that in both cases the transformation to metal nitride was confined to the two outermost atomic metal layers. Further evidence for the surface nitride formation on nickel comes from the DSC measurement shown in Figure 5b. In contrast to the copper nitride layer (cf. Figure 5a), the nickel nitride layer decomposes endothermally with the highest rate at about 750 K. No marked nitrogen desorption was observed if the catalyst samples were purged with hydrogen at 520 K after exposure to ammonia. This indicates that the formation of surface nitride can be suppressed by hydrogen. Conclusion The hydrogen applied in the catalytic amination of aliphatic alcohols acts as an inhibitor for catalyst deactivation. It inhibits catalyst deactivation caused by formation of an inactive nitride layer on the metal surface. The metal nitride is formed by ammonia originating from the simultaneously catalyzed disproportionation of the reactant amines. A deactivation mechanism similar to the one shown for the copper catalyst may occur on other supported metal amination catalysts. It is likely to occur on all metals which catalyze the disproportionation of the reactant amines and form metal nitrides in the presence of ammonia. Such metals are, for example, nickel and cobalt, as was evidenced by TPD and DSC studies. Acknowledgment The author acknowledges, with gratitude, the experimental assistance of Dr. W. Caprez, W. Kagi, and D. Monti. Literature Cited Baiker, A , , Richarz, W. Ind. Eng. Chem. Rod. Res. Dev. 1977a. 16, 261. Baiker, A,, Richarz, W. Filth Canadian Symposium on Catalysis, Calgary, 1977b; Preprints, p 298. Baiker, A , , Richarz, W. Synfh. Commun. 1978, 8 , 1. Caprez, W. Ph.D. Thesis No. 6676, Swiss Federal Institute of Technology, Zurich, 1980. Cvetanovic, R . J., Amenomiya, Y. Adv. Catal. 1967, 17, 103. Kliger, G. A. et al. Kinet. Catal. 1975, 76, 567, 571. Pommersheim, J. M., Coull, J. AIChEJ. 1971, 77, 1075. Rau, R. C. In “Encyclopedia of X-Rays and y-Rays”;Clark, G. L., Ed.; Reinhold: New York, 1963; p 184. Terao. N. Mem. Sci. Rev. M6t. 1960, 57, 95. Terao, N. C.R. Hebd. S6ances Acad. Sei. S6r. B 1973, 277, 595. Received for review April 6, 1981 Accepted July 17, 1981
Parts of this paper were presenbd in the Communication Session of the Seventh International Congress on Catalysis, Tokyo, June 30, 1980.
