Production of Acetonitrile and ther Low Molecular eight Nitril This work was part of a study of the reactions of low molecular weight hydrocarbons. The investigation resulted in the development of a new process for the production of acetonitrile by the reaction of Cp to Cg hydrocarbons with ammonia. Smaller amounts of higher molecular weight nitriles are also formed. Alkenes are more reactive than alkanes. Yields of from 10 to 40 weight YQ nitrile per pass are obtained with ultimate yields of 6OyQof theory. Typical operating conditions are: 940' F., atmospheric pressure, molar excess of hydrocarbon, 10 to 15 seconds' contact time, and catalyst of molybdenum oxide on activated alumina. As in most continuous processes, a large unit is capable of producing nitriles at a much lower cost than a smaller unit. The present market for acetonitrile is less than 500,000 pounds per year. Therefore, a plant large enough to achieve minimum nitrile costs would have to be designed to convert a portion of the nitriles to other products, such as acids or amides, until the market for nitriles has expanded to plant capacity. WILLIAM I. DENTON AND RICHARD B. BISHOP Research and Development Department, Socony- Vacuum Laboratories, Paulsboro, N . J .
ITRILES are conventionally produced by the reaction of an organic acid with ammonia over a dehydration catalyst ( I $ ) , by the dehydration of an amide with suitable catalysts or reagents (1, IO, I S ) , or by the dehydration of an ammonium salt of an organic acid with heat ( 1 1 ) or heat plus a catalyst (2, 7). Xethan01 plus hydrogen cyanide in the vapor phase over a highly porous catalyst such as activated charcoal, silica gel, alumina, or thoria yields acetonitrile (9),while ethanol plus ammonia over a reduced copper catalyst at 575" t o 650" F. gives a 40% yield of acetonitrile (6). Considerable work has been done on the reaction of acetylene and ammonia to produce acetonitrile using catalysts at about 650' F. with or without steam (6, 14). The reaction of alkenes with ammonia under pressure at about 650' F. in the presence of a cobalt or nickel catalyst yields acetonitrile and other nitrogen-containing products (15). A previous paper ( 5 ) described the production of aromatic nitriles from alkyl-substituted aromatic hydrocarbons and ammonia. This paper describes a process for the production of acetonitrile (CH3CN) from ammonia and alkenes. The process is covered in United States patents ( 4 , 8 ) . EXPERIiMENTAL
Figure 1 is a schematic flow diagram of the equipment used in this study. The hydrocarbon and ammonia were metered through rotameters, then mixed and passed through a preheater coil which was immersed in a molten salt bath held at the desired temperature. The preheated charge was then passed into a 350-ml. reactor filled with catalyst. Temperature control was obtained by immersing the reactor in a molten salt bath held a t the desired temperature. The products from the reactor were passed through a water condenser and then through a dry ice condenser. The condensate from the dry ice condenser was separated into a hydrocarbon layer and a n ammonia layer and each layer was distilled separately. Bottoms from the two distillations were combined and added to the condensate from the water condenser to give the total product from the run. This product was then analyzed for nitriles by fractionation on a 25-plate COIUmn.
It was found that acetonitrile could be recovered in two fractions: a 16% water (from reduction of hIoOa)-acetonitrile azeotrope boiling a t 77-78" C. and anhydrous acetonitrile boiling a t 82" C. These cuts were readily separated from propioni'
282
c,).
trile (boiling point 97" Individual components were identified by ultimate analysis of purified samples, refractive index, specific gravity and, in some cases, preparation of derivatives. I n this paper the terms "yield per pass" and "ultimate yield" are used. The yield per pass is expressed in weight per cent and is shown in the figures as the pounds of nitrile formed per 100 pounds of reactant passed through the reactor. Also shown in Table I1 are yields per pass in mole per cent. The ultimate yield is expressed as 100 times the moles of nitrile formed divided by the moles of reactant used up. Both yields are essential in evaluating a process, the ultimate yield indicating how efficiently the reactants are being used and the yield per pass indicating how much of the desired product is formed per pass. REACTANTS
Alkanes of low molecular weight are most reactive in this process. The hydrocarbon reactants investigated are listed in Table I. I n most cases the chief product is acetonitrile under the conditions employed, but smaller amounts of ammonium cyanide and nitriles of higher molecular weight are also formed. The exact mechanism of the reaction has not been established. With propylene as the hydrocarbon charge, one possible mechanism is the formation of propionitrile, which then breaks down to give acetonitrile and methane:
CHa-CH=CHt
+ KH3 +CH3CHzCN + 2132 + Hz --+ CH3CN + CN,
CHjCHzCN
(1) (2)
The feasibility of Reaction 1 is indicated by the fact that the nitrile product usually contains 5 to lOyo of propionitrile along with the acetonitrile. When a nickel catalyst is used a t 650" F. and moderate pressure, the product is predominantly propionitrile accompanied by small amounts of acetonitrile. This indicates that when the conditions are mild enough the propionitrile, which is first formed, survives as the major product. The second reaction has been studied b y charging propionitrile to the process at the normal operating conditions-ie., propionitrile and ammonia instead of propylene and ammonia. Under these conditions (975' F., 1.2 seconds, 2 moles of ammonia to 1 of nitrile), it was found that approximately 20% of the propionitrile was converted to acetonitrile, 3oy0 recovered un-
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 45, No. 2
Unit Processes
c
NITRILE TO
TABLE
PRODUCT
PURIFICATION
I.
Hydrocarbona Alkenes Ethylene Propylene 1-Butene 2-Butene Isobutylene Rutadienz'J 1-Octene Alkanes Ethane Propane n-Butane Isobutane Methylcyclohexaned Dimethylcyclohexane E n-Heptane
OF
$ ;:;:
proceeds. Therefore, before the various process variables could be studied, the length of time the catalyst would remain at a reasonable activity level had t o be determined. The study presented in Figure 2 shows that catalyst activity increases to a maximum 2 t o 4 hours after the run begins and then falls off to the initial activity after about 8 hours' usage. This peak is explained by the fact that Mooz is SPECTROMETER more active than MoOa. Thus, as ANALYSIS MoOa is reduced the activity rises to a peak and then gradually falls off as the catalyst becomes fouled with carbon. Both curves show the same trends, in spite of the fact t h a t major changes were made in three operating conditions and approximately 35 times as much reactant passed over the catalyst a t the short residence time. It was not determined whether this same Figure 1. Flow Diagram trend would be exhibited with an excess of hydrocarbon in the charge instead of an excess of ammonia. The regenerability of the catalyst KYDRoCARBoN was studied b y using the catalyst in various runs, regenerating, Yield per Pass, Lb. Nitrile per 100 Lb. and re-using. After every 25 runs, a check run under a standHydrocarbon ard set of operating conditions was made. This test showed no Am$ ";: $$:'&decline in activity after 100 cycles.
*:":'*
24 15 10 10
2 Trace Trace Trace
4 0.5
... 5 ...
3
3
3 4 2
1.5
3
1.5 0.5 0.5
...
... .. . .. .. ..
F'$z:
...
...
..,
0.5
2
0.5
...
...
1
... ...
1
0.5
.. .. .. ...
...
... ...
0.5 s
.
.
.
... ...
... ... . .. i"
...
... . .. ...
...
...
1
... ... ., .. ..
1.5
.2
PROCESS VARIABLES
The main process variables are temperature, molar ratio of the reactants, and residence time. Other variables are pressure and concentration of alkene in the hydrocarbon charge. I n order t o simplify the study of these variables, the operating conditions were arbitrarily standardized as follows: hydrocarbon charge, propylene; catalyst, 10% molybdic oxide on activated alumina; length of run, 3 hours. Some of the graphs presenting results obtained on individual variables compare curves in
2
Lowering the tewperature to 800" F. permitted most of the propionitrile t o come through unchanged, while at 850" F. 60'% propionitrile and 10% acetonitrile were recovered. With alkenes of higher molecular weight-e.g., butenesboth butyro- and propionitriles were found in addition t o acetonitrile. Alkanes are much less reactive than alkenes and probably are dehydrogenated t o alkenes, which then react with the ammonia. Hydrocarbons containing seven or more carbon atoms may formboth acetonitrile and aromatic nitriles, since a portion of the hydrocarbonis aromatized and can then be converted to niti-iles. D a t a on the results obtained with various reactants are summarized in Table I.
m
-I 0
1
P
8
P
d)--\
14-
W
&
5 e
IO
/
G 9 s i 2
i
\ e'
\ e'
\
\
\e
T.4BI.E
Mole % ammonia. in total charge
11.
OPERATISG VARIABLES
Operating ConditionMole % CaHa in hydrocarbon charge
~oiar5 ratio, NH3:CsHs
Pressure. Ib./sq. inch gage
Residence time, sec.
I. 67
95-100
75
95-100
62
46
Tyiip.,
F.
Yield per Pass Weight % b Mole r r / c c Ammonia Propylene Ammonia Propylene
-
Temperature
9 6
4 2 3.9 8.4 11 4 16 4 20 3
3:l
12.8 16.8 21.0 32.0
4 1 5,s 6.8 10.4
13.1 17.2 21.5 32.8
3.6:l
16.7 21. 8 24.7 33. 8
4.9 6.3
7.1 9.8
17 1 22 3 25.3 34.6
0
2:l
4 1 3 8 8 2 11 1 16 0 19.8
1.2 1.2 1.2 1.2 1 4 1.4
75
20
875 900 925
29.4 46.0 71.4
z:: 2
16.4
12.4 19.I 29.6
16.8 26.2 41.5
95-100
1:3
0
25 26
900 973
134.0 61.1
19.1 8.7
59.6 25.3
19,6 8.9
67
95-100
2:l
0
0.2 1.2 7 13 14 29
900 ROO 900 900 900 900
2.7 9.6 25.7 41.0 42.6 54.4
2.3 8. 2 22.0 35.0 35.8 46. G
1.1 4.0 10.7 17 0 17.7 22.5
2.4 8.4 22.5 3.5. 8 36 7 47.8
75
95-100
3:l
90
1.2 8.5 13
92 5 925 $125
86
925
4.3 16.4 28.1 28.8 49.8
6. 4 21.0 36.1 37.0 63.8
1.8 6.8 11.6 11.9 20.6
5.5 21.5 37.0 37.9 65.4
3 3
9.8 35.0 55.6
3 4 12.2 19.6
35
25
-4. Excess Animonia
92.5
15
B.
25
95-100
Excess Hydiocaibon
1:3
11.9
19.1
111. 3Iolar Ratio. R:l 2:l 1:1 1:2 1:3 1:9
10
1 2 1.2 1.2 1.2 1.2 1.2
B.
80 75 62
w
8 5:l
46 35
3.5:1 1.4:1 1.4:l 1:3 1:B
90 90 75 75 75 75
3:l
0
8.5:1
46 ..
33 10 6
35 35 35 35
75
95-100
33
50
7 7
1:I
0 275
28 24
1.4:1 1.4:l 1.4:I 1.1:l
50 75 280 350
21 20 18 27
90 67 50
95-100 95-100 35 35 46 80
v 75
35 46 95-100
17
35 95-100
80 75
2:1
Propylene Concentration. !w 8.5:l 8.5:l !N 3:l 90
B.
17
1:l.S 1:4.8
8
A.
A.
Shoit Residenre Time ROO 1.5 9.6 ROO 9.4 900 ROO 20.0 23.6 900 90 0 50.0
Long Residence Time 19 910 915 20 19 91,s 20 900 20 925 25 930 25 92.5
IT'.
284
%.i 7 8
1.4:1
33
11. Residence Time.
4
2 0 1,9
Pressure 9 60 9.50
9 50
90 .i
92.5
900 900 925 900
5 7 8 4 4 1 3.8 3 4 2 5
9.0 9.3 17.2 46.0 56.7 60.3 89.9
30.6 32.3 24.i 25.6 32,2
7..5 5.9
3.7 3.9 7.1 19.1 23.5 2.5.0 37.2
28 7 21 9 25 0
36 0 27 9 32 0
11 9 9 1 10 4
@36.R 28 6 32 8
37.1 24.2
31.2 9.9
15.4 10.0
32.0 10.1
32 46 18 19
18 25 10 6
13 14 7 7
18 26 11 6
8 0 8 1
2 6 9 6
31 4 33 1 ?,5 3
26.2 33.0
7.7 6.0
6 1 8 9
Excess Ammonia, Constant SHs-Hydrocarbon Ratio 20 91 5 9.3 32 3 3.9 19 910 9.0 30.6 3.7 14 910 25.0 32.0 10.4
Excess Hydrocarbon, Constant XHa-Hydrocarbon Ratio 0 7 975 44.0 0 7 975 75.7
9 8
6.4
18.2 31.8
6
2 2 8
33.1 31.4 32.8
10 0
6.6
Accurate t o apuroximately 1 0 . 2 100(wt. CH3CN CHsCH2CN)' Reactant indicated a t head of each column. Figures are I approximately 0.5%. ' wt. reactant charged Figures a r e I: approximately 0.5% with propionitrile calculated as being acetonitrile. 100 moles CHaCN Moles reactant charged
+
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 45, No. 2
Unit Prooesses
0
TEMPERATURE-*E
Figure 3. Effect of Temperature Mole %
NHa in
Charge
67
75 33
Mole % CsHs in Hydrocarbon 95 95 35
MOLE PERCENT &MMONIA IN CHARGE
Figure 4. Effect of Molar Ratio
Pressure Residence Time, Sec. 1.2 8 20
Lb./Sq.
I n c h Gage
Mole
0 90 75
which operating conditions have been varied widely. These data are presented t o show the effect of the primary variable under widely different conditions. However, care should be exercised in drawing any secondary conclusions from the data. Detailed data on the operating variables are shown in Table I1 and summarized in the following discussion. TEMPERATURE
Figure 3 shows that, as would be expected, increasing the temperature increases the yield of acetonitrile per pass. However, as the temperature is increased the amount of decomposition of the ammonia and propylene is increased and, from a practical standpoint, the advantages of the increased yield per pass must be weighed against the increased losses t o determine the optimum operating conditions. Temperature is much more critical a t long residence times (low throughputs). Thus, a t short residence times, the yield of acetonitrile per pass at temperatures of 975" F. and above may actually be lower than the yield of acetonitrile per pass obtained a t temperatures of 900' t o 920' F. and longer residence times.
%
CsHe in Hydrocarbon
Residence Time, Seo.
95 35
1 19-25
0 0
Temp., F. 900 920 av.
Pressure I nLb./Sq. c h Gag6 0
75-90
ammonia decomposition. The proper residence time may be determined by balancing reactor costs against the costs of additional recovery facilities. Thus, at short residence times large amounts of nitrile per volume of reactor are obtained but considerably more reactants have t o be recovered and recycled. At longer residence times, the reverse is true. 6
4
2
0 40
20
MOLAR RATIO OF REACTANTS 0 %
*
Changing the mole per cent of ammonia in the charge from 25 t o 90% has little effect on the yield of nitrile per pound of propylene charged (Figure 4). However, the yield of nitrile per pound of ammonia charged is much higher with smaller amounts of ammonia present (an excess of hydrocarbon). Therefore, an excess of hydrocarbon in the charge is preferable. Another advantage of using an excess of hydrocarbon is t h a t the alkenes of lower molecular weight seem t o be more stable than the ammonia under these reaction conditions and, therefore, the ultimate utilization of the ammonia is increased without excessively decreasing the effectiveness of the hydrocarbon. RESIDENCE TIME
Increasing the time required for one volume of reactants to pass through one volume of reactor (calculated assuming the reactor is empty) increases the yield of nitrile per pass (Figure 5 ) . Also shown on this figure is the yield of nitrile per pound of catalyst per hour a t various residence times (calculated from the original data). I n the range of 20 t o 7 0 mole yo ammonia in the charge, the curve indicates that the yield of nitrile per pound of catalyst is about the same regardless of whether an excess of hydrocarbon or an excess of ammonia is used. Therefore, an excess of hydrocarbon should probably be used to minimize
February 1953
I20
80
40
0 I
1
5
2
3
4
6
8
1
R E S I O E N C E TIME,SECONOS
Figure 5.
Effect of Residence Time
Mole .% NHa In Charge
Mole % CsH6 in Hydrocarbon
25 67
95 95
Temp.,
F.,
900
900
Pressure, Lb./Sq. Inch Gage 0 0
PRESSURE
No direct comparisons showing the effect of pressure are available. Some runs a t fairly similar operating conditions are shown in Table 11, IV. I n the range of 0 to 100 pounds, little difference was found in the yield per pass. The runs a t pressures above 100 pounds per square inch gage are not strictly
INDUSTRIAL AND ENGINEERING CHEMISTRY
285
dowi the reaction, the increase in Ihe ratio of ammonia to propglene compensates for the dilution effect and the same amount of propylene is converted to nitrile. No data are available on the dilution effect of propane a t constant ammonia-to-propylene ratios. However, it is likely that the amount of ammonia and propylme converted to nitrile would decrease as the amount of propane is increased. ULTIMATE YIELDS
1
1
70
I
80
00
1 0
MOLE P E R C E N T PROPYLENE IN HYDROCARBON CHARGE
Figure 6.
Effect of Propylene Concentration Mole
NHa in Charge
Residence Time, See.
Temp., F.
Pressure Lb./SCl. Inch. Gage
75
18
910
90
975
0
0 0
7
17
comparable, but are presented to give an indication of the results that may be obtained a t the higher pressure levels. They indicate that increasing the pressure a t constant residence time may decrease the yield of nitrile per pass. PROPANE-PROPYLENE RlIXTURES
Because pure propylene is considerably more expensive than propane-propylene mixtures which are available in oil refineries, it would be advantageous to use such a mixture rather than pure propylene. Therefore, a study was made of the effect of changing the proportions of propane and propylene while keeping the overall ratio of ammonia to hydrocarbon constant. This study
TABLE 111. TYPICAL MATERIALB A L A X C E ~ Charge
Reactant Aminonia Hydrocarbon!
14.8
5.6 5,4 0.5
100 0
wt. '.=
100.0
Wt. %
Yield per Pass 27.4e 20.38
Recovery 72 92
b;
LITERATURE CITEIP
73.4
... ... ...
Total
wiightbf
Utilirationd 42 94
reactant charged) X
100.
uu) X 100. 6 (Weight of nitrile divided by weight of ammonia charged) X 100. f Wt. 70recovery a n d utilization based o n total hydrocarbon rather t h a n d (Moles of nitrile divided b y moles of reactant used
propylene because recovered Dropane, while not as reactive in this process a8 propylene, is as yalu?h!e as t h e refinery s t r e a m charged. Y (Weight of nitrile divided by weight of propylene charged) X 100.
showed that diluting the propylene with propane did not affect the amount of propylene converted to acetonitrile per pass but did reduce the amount of nitrile formed per pound of ammonia passed (Figure 6). This indicates that although the propane is introduced as a diluent which might be expected to slow
-c?AYz
286
CONCLUSIOIV S
Alkenes and ammonia call react to form nitriles using a molybdena catalyst. Ethylene and propylene are more reactive than other alkenes or alkanes. Yields per pass of up to 0.8 pound of nitrile per pound of ammonia charged and ultimate yields of aiiout 60% of theoretical have been obtained. Temperature, molar ratio of reactants, and residence time are the most critical variables. Less important variables are pressure and concenti ation of allrene in the hydrocarbon charge. Typical operating conditions are: 940" F., an excess of hydrocarbon, 10 to 15 seconds' residence time, and a catalyst of 10% molybdic oxide on activated alumina.
Products
79 6 * 20 4
Hydrocarbon Ammonia Nitrile CZ a n d lighter Coke
The previous discussion has shown that operating conditions can be selected which give yields per pass of up to 0.4 pound of nitrile per pound of propylene or up t o 0.8 pound of nitrile per pound of ammonia charged. Generally speaking, an excess of hydrocarbon and relatively long residence times favor thcse higher yields per pass. Theoretically, approximately 1 pound of acetonitrile can be made from 1 pound of propylene, while from 1 pound of animonia it is possible to make 2.4 pounds of nitrile. However, the ultimate yield, because of the decomposition of the reactants which occurs siniultaneously with the nitrile reaction, is only 35 to 60y0 of theoretical, based on the ammonia, and 60 to 95% of theoretical, based on the propylene. These limits are the average of a number of material balances (data not shown) made 011 various runs. Table I11 presents a material balance on a typical run in which the weight per cent yield per pass is 27 and 20y0 and the ultiniate yields are 42 and 9470 of theoretical, based on ammonia and hydrocarbon, respectively.
Boehner, R,., and .indrews, C., J . Am. Chem. SOC.,38, 2503 (1916). Deem and Lazier (to E. I. du Pont de Nemours & Co.), C. 6 . Patent 2,149,280 (1939). Denton, W,, and Bishop, R., INU.ENG.C I i E M . , 42, 796 (1950). Denton, W., and Bishop, It. (to Socony-Vacuum Oil C o . ) , U. S. Patents 2,450,636 to 2,450,642 (1948). Ham, Tohoru, and Komatsu, Shigeru, Mem. CoZE. Sei. IC@o I m p . Unia., 8 A , 241 (1925). I. G . Farbenindustrie, Brit. Patent 295,276 (1927); 332,258 (1929); Ger. Patent 558,565 (1929). Linstead and Lowe (to Imperial Chemical Industries), U. S. Patent 2,054,088 (1936). Marisio, M., Denton, W., and Bishop, R . (to Socony-Vacuum Oil C o . ) ,Ibid., 2,450,675 to 2,460,678 (1948). i'-icodemus, Otto, Ger. Patent 463,123 (1928). Norris, J., and Sturgiu, B., J . Am. Chem. Soc., 6 1 , 1413 (1939). Ralston and IIarwood (to -4rmour and C o . ) , U. S. Patent 2,245,548 (1941). Reid, E. E., Am. Chem,. J . , 43, 162 (1910); J . Am. Chem. SOC., 38, 2120 (1916); 53, 321 (1931). Sowa, F., and Xieuwland, J. A., Ibid., 59, 1202 (1937). Stuer, B., and Grop, W. (to Chemical Foundation, Inc.), Brit. Patent 147,067 (1920); U. S. Patent 1,421,743 (1922). Teter, J. (to Sinclair Refining Co.), Ibid., 2,417.892, 2,417,893. 2,419,420, 2,429,865 (1947).
--
RKCEIVED for review September 5 , 1952.
C??W
INDUSTRIAL AND ENGINEERING CHEMISTRY
ACCEPTED December 10, 1952.
Vol. 45, No. 2
