832
Ind. Eng. Chem. Prod. Res. Dev. 1882, 27, 632-635
Hurwitz. M. D.; Conlon, L. E. Text, Res. J . 1958, 2 8 , 257-262. Hurwitz, M. D.;Cowan, S.L. J. Elisha Mitchell SOC. 1978, 94, 85-86. Hurwk, M. D.; Rubenstein, K. Abstracts of Papers, 155th National Meeting of the American Chemical Society, San Francisco, CA. April 1968;American Chemical Society: Washington, DC, 1966;ORGN 5. Kadowakl, H. Bull. Chem. SOC. J p n . 1936, 7 1 , 248-261; Chem. Absfr. 1936, 30, 5944'. Mauersberger, H. R.. Ed. "Matthews' Textile Fibers", 6th ed.; Wllay: New York. 1954. Meunier. L.; Guyot, R. Rev. G6n. Collohes 1929, 7 ,53-66; Chem , Absfr . 1929, 2 3 , 3i03. Meyer, U.: Muller, K.; Zolllnger, H. Text. Res. J . 1976, 46, 813-817. Nuessle, A. C.;Flneman, M. N.; Heiges, E. 0. J. Text. Res. J . 1955, 2 5 ,
Staudinger, H.; Ncdru, R. Chem. Ber. 1930, 638,721-724 Chem. Absfr. 1930, 2 4 , 5717. Steele, R. J. Appl. Polym. Sci. 1960, 4(10), 45-54. Steele, R. Am. Dyest. Rep. 1985, 54, P6-PI2. Steiger, F. H.; Wang, S. Y.; Hwwitz, M. D. Text. Res. J . 1961, 31, 327-339. Swenson, ti. A. I n "Methods In Carbohydrate Chemistry: Cellulose"; Whlstier, R. L., Ed.; Academic Press: New York, 1963; Vol. 3, pp 84-90. Vink, H. I n "Cellulose and Cellulose Derivatives"; Blkales, N. M.; Segal, L., Ed.; Wlley: New York, 1971;Part 4. Walter, H. C.; Buxbaum, J. K.; Green, L. Q. Text. Res. J . 1957, 27,
146-149.
Received for revieul February 11, 1982 Revised manuscript received June 17, 1982 Accepted July 16, 1982
24-40. Reeves, W. A.; Drake, G. L . Jr.; McMlllan, 0. J., Jr.; Guthrie, J. D. Text. Res. J . , 1955, 25, 41-46. Segal, L.; Tlmpa, J. D. Text, Res. J . 1973, 43, 185-194. Sorenson, 8. E. U S . Patent 2201 927, 1940; Chem. Absfr. 1940, 3 4 ,
6301'. Slhtola. H.; Kaila, E.; Laamanen, L. J. Polym. Sci. 1957, 23, 809-824.
Presented at the Cellulose, Paper, and Textile Division of the American Chemical Society, Atlanta, Mar 30-Apr 3, 1981.
Propylenamines by Cobalt-Catalyzed Nitrile Hydrogenations L. G. Duquette, P. E. Garrou," 0. E. Hartwell, and J. A. Kaufman Dow Chemical USA, New England Laboratory, Wayland, Massachusetts 01778
A process for the preparation of analogues to diethylenetriamine and triethylenetetramine based on Michael addition of acrylonitrile to ethylenediamine and subsequent hydrogenation has been studied. The Michael addition proceeds at 45 OC with no need for acid catalysis achieving a statistical distribution of products. Control of the products is possible only by altering the reactant mole ratios. The resultant nitriles are hydrogenated by commercial cobalt catalysts at 85 OC under 1500 psig H, to the corresponding primary amines and cyclic homopiperazine. Proper choice of the catalyst virtually eliminates the unwanted cyclic byproducts.
Introduction Aliphatic polyamines are employed in a wide spectrum of industrial applications including fungicides, chelating agents, surfactants/softeners and polyamide resins (Spitz, 1980). T h e polyethyleneamine series HzN(CHzCH2NH),CHzCHzNHzhave normally been the amines of choice in such applications. Such polyethylene polyamine compounds (ethylenediamine, diethylenetriamine, etc.) have been conventionally produced by reacting ethylene dichloride with ammonia at elevated temperatures. In general, relatively high yields of predominantly noncyclic polyethylene polyamines are obtained along with varying yields of cyclic piperazine materials. Such processes are generally employed throughout the industry. An alternate set of materials having similar properties and applications may be derived by the reduction of the Michael addition products of ethylenediamine (EDA) and acrylonitrile (VCN). Several hundred publications, patents, and reviews have concerned themselves with the reaction of amines with VCN, but we wish to describe, for the first time, the details of the specific reaction of EDA with VCN and the reduction of the resultant nitriles using typical industrial cobalt catalysts. Experimental Procedures Continuous Cyanoethylation of EDA. A 300-mL Parr stirred reactor was set up for "continuous stirred tank operation." The dip tube was adjusted so that the effective reactor volume was 58 mL. The stirrer rate was held constant at 250 rpm. In a typical reaction, EDA and VCN are fed separately (i.e., 63.6 and 13.2 g/h, respectively) 0196-4321/82/1221-0632$01.25/0
resulting in a residence time of 45 min/reactor volume over a period of 82 min. The effluent analyzed for 91% I, 9% 11. Altering the flow rates to adjust the mole ratio, as long as the residence time was 15 min, produced the composition diagram shown in Figure 1. HJ'J(CH2)2NH(CH2)&N I NC(CH2)2NH(CH,)ZNH(CH2)2CN I1 Hydrogenation of Cyanoethylated Products. A tubular, 316 stainless steel reactor ( 5 / 8 in. i.d. X 10 in. long) was packed with 45 g of catalyst, and the catalyst was activated by pretreatment with Hzas described by the vendor. Through the activated bed at a given temperature, and under 1500 lb of H2 was then flowed at a fixed rate a mixture of I or I1 and NH3. After a pass through the reactor bed, the product was analyzed by withdrawing a small aliquot for GC, and it was recycled over the catalyst until sufficient conversion had been achieved. Typically, 15-50-g samples of I or I1 could be processed (12 passes) in approximately 3 h. For specific catalysts and conditions, see Table I. Analytical Results GC analyses were performed on a Hewlett-Packard 5750 TC instrument. A 1.5 f t X l/g in. 0.d. 3% Silar 5CP on Chromasorb Q (100/120 mesh) column was used. The carrier gas (He) flowed through the column at a rate of 30 cm3/min. The temperature was programmed to run at 50 " C for 2 min, heat a t 16 "C/min up to 250 "C, and hold at 250 "C for 8 min. Under these conditions, the following 0 1982 American Chemical Society
Ind. Eng. Chem. Prod. Res. Dev., Vol. 21, No. 4, 1982 633
Table I. Conversion and Selectivity Data for Typical CN Reductions selectivity, w t react. no. 1 22 3 4 5 6
7
NH,/
reaction components
catalyst Co-0127T'
RCN, mol
I, NH, I, NH3 I, NH, I, NH, 11, NH, 11, NH, I, 11, EDA, NH, I I1 I11
G-619 G-628
KWH-1235h
G-61g
G-618 G-61g
no. of passes
conv./ total passes
IV
V
I11
%
rev. Michael EDA total
I
IV V
EDA + poly AN (not a At 85 'C, 1500 psig H,, flow of 186 mL/h, space velocity 0.15 g of liquid flow/g of cat. h. observed). C Flow rate 69 mL/h, space velocity 0.35 g of liquid/g of cat h. 88%(111) + 5% NH,(CH,),NH( CH,),NH(CH,),CN (the half hydrogenated product). e Cyanoethylhomopiperazine or its reduction products were not observed. f EDA/(I)/(II)= 25/50/25 wt % ratio. g G-61 (62% COO on Kieselguhr)and G-62 (33%COO on Al,O,) from United Catalysts, Inc. KWH-1235 (23% COO, 7% Mn, 0.4% Ag on Aluminosilicate)from Houdry-Huls. Co-0127T (39%COO on Kieselguhr) from Harshaw Catalysts.
Figure 1. Product distribution for varying ratios of EDA + VCN.
(CD8OD) 33.75 (CH&HzCHz), 40.79 (CH~CHZCHZNH~), 41.96 (CH,CH,NHz), 48.23 (CH*CH,CH2NH2), 53.01 (CHzCH2NHz). GC/MS ( t R = 7.5 min), m / e (relative intensity) 87(23), 56(11), 44(100), 42(13), 30(22), 28(15). N,N'-1,2-Ethanediylbis-l,3-propanediamine (V). 90MHz 'H NMR (C6D5) 1.02 ( s , ~ HNH2, , NH), 1.48 (q,4H, CHzCHzCHz),2.58 (m, 8H, CHzCHzCH2),2.64 (s, 4H, NHCHZCHZNH). 20-MHz 13C NMR (CDSOD) 33.22 (CH&!H,CH,), 40.15 (CHZNH,), 47.64 (NHCHZCHZNH), 48.25 (CHzCH2CHzNH2).GC/MS (tR = 10.6 min), m / e (relative intensity) 87(19), 44(100), 30(18). Toxicity Acrylonitrile and cyanoethylated amines are known toxic materials and should be handled cautiously with proper ventilation.
retention times were achieved: NH3,O.lO min; H20,0.35 min; EDA, 0.75 min; H2N(CH2)3NH2,1.77 min; 111, 3.61 min; IV, 7.57 min; I, 8.61 min; V, 10.49 min; 11, 13.64 min. NMR-MS Data 3-[(2-Aminoethylamino]propanenitrile(I). 90-MHz 'H NMR (C6D5) 1.18 (9,3H, NHz, NH), 2.15 (t,2H, CHzNHz), 2.46 (m, 6H, CH2's). 20-MHz 13C NMR (C6D5)18.72 (CHZNHJ, 42.06 (CH&HZNHz), 45.29 (CH&HZCN), 52.18 (CH2CN),86.57 (CN). GC/MS ( t =~8.6 min) m / e (relative intensity) 83(94),56(10),54(37),44(31),43(22),42(100), 41(13), 30(62), 28(37). 3,3'-(1,2-Ethanediyldiimino)bispropanenitrile (11). 90MHz lH NMR (C6D5) 1.38 (9, 2H, NH), 2.32 (t, 4H, CHZCH&N), 2.57 (9, 4H, CHZNH), 2.69 (t, 4H, CHZCN). 20-MHz 13CNMR (C6D5) 18.65 (CHZNH), 45.12 (CHzCH,CN), 48.63 (CHZCN), 119.14 (CN). GC/MS ( t =~13.6 min), m / e (relative intensity) 85(14), 84(12), 83(100), 56(19), 54(20), 44(12), 43(12), 42(47), 28(20). Hexahydro-lH-l,4-diazepine(111). 90-MHz 'H NMR (CHC13) 1.58 (9, 2H, NH), 1.78 (9, 2H, CH2CHzCHz),2.90 (s, 4H, NHCH2CH2NH), 2.95 (m, 4H, CHzCH2CHz). GC/MS (tR = 3.6 min), m / e (relative intensity) 99(27), 85(12), 70(20), 69(12), 58(12), 57(13), 56(21), 44(100), 43(48), 42(32), 41(19), 30(23), 28(28). N-(2-Aminoethyl)l,3-propanediamine(IV). 90-MHz 'H NMR (C6D5) 1.13 (s, 5H, NH2, NH), 1.46 (9,2H, CHzCHzCHz),2.54 (m, 8H, CHz's). 20-MHz 13C NMR
Results Michael Addition. A continuous stirred tank reactor was set up to operate at constant temperature with the EDA and VCN separately pumped into the system; the residence time in the reactor was controlled by the flow rate of the reactants. A complete description of such continuous stirred reactors has been given elsewhere (Levenspiel, 1972). We found, by analysis of the effluent stream, that the ratio of EDA/VCN/NC(CH,),NH(CHJzNH2 (I)/NC(CHz)zNH(CH,),NH(CHz),CN (11) independent of the residence time (15-45 min), the temperature (40-60 "C), or the presence of catalysts. Although a recent German patent (Ger. Offen., 1974) claims that a higher selectivity to I1 is achieved by the presence of an acid catalyst, we find no need for acid catalysis in this reaction. The reaction at 40-60 "C is highly exothermic, and no rate or selectivity advantages were observed using an HOAc catalyst as described in the patent. The only influence on the product composition was the molar ratio of the reactants EDA/VCN in the incoming stream. We verified equilibrium in our continuous reactor by examining several EDA/VCN concentrations in a static system (round bottom flask) for long periods of time (up to 10 h). All results were the same except that at temperatures greater than 60 "C we began to see products attributed to the addition of VCN to secondary amine groups. As shown in Figure 1,the reaction proceeds to a
_is----ciIC~. 2 0 30 '20 53 53 73 80 90100 3,
'401-
E
EC ;
-7
r'4010
'io\/CY
x~30
634
Ind. Eng. Chem. Prod. Res. Dev., Vol. 21, No. 4, 1982
-
Scheme I y2p--./ht$
EUA
+ -1,
IIC/"IIHN~~Z
I
VCN
I1 L
-VCN
~
i
' -AN
+ t I ~ ~ N H A / k j n "
i +VCN l
EUA
-VCN
-VCN
1.
-
+VCN
~
1
tVCN
i l I
Scheme I1
n
h
statistical mixture. Compound I can be prepared only after the EDA/AN molar ratio exceeds 0.5. Below the ratio of 0.5, the reaction mixture is composed only of VCN and 11. Dickerman and Simon (1957), in an attempt to produce I, reported briefly that EDA reacts with VCN at 31 molar ratio to give a 59% yield of I based on VCN. Figure 1 points out that a 1:l molar ratio would have produced a 50% yield based on EDA or VCN and thus better productivity per reaction could have been obtained. Separation of products by distillation of the effluent stream at various EDA/VCN/(I)/(II) ratios was easily achieved-EDA (32 "C/0.3 mm), (I)(103-105 "C/0.3 mm), (11) (173-4 "C/0.3 mm). Attempted distillation of I1 at higher pressures proved more difficult with the yield and purity suffering from some cracking (reversal of the Michael reaction) occurring a t higher temperatures. Reversibility of the cyanoethylation reaction has been discussed previously by American Cyanamid (1959), Butskus (1961), and Bruson (1949). Nitrile Reduction. The reduction of the cyanoethylated adducts (I and 11)is complicated by two factors. (1)The reversibility of the reaction, as was mentioned in the distillation of the nitriles, could alter the selectivity (see Scheme I). I1 can revert to I and VCN; I can revert to EDA and VCN. In addition, the VCN can polymerize and/or be hydrogenated in the presence of the metal catalysts. Attempts were made to minimize these effects by the maintenance of constant, low reaction temperatures. (2) The hydrogenation of RR'N(CH2),CN (n = 3,4) can be complicated by intramolecular condensations (Freifelder, 1971). In our case, reduction of I could lead to homopiperazine I11 as shown in Scheme 11. A series of commercially available cobalt catalysts that were known to reduce nitrile functionality under relatively mild conditions (1500 psig H2, 100 "C) were screened using NH3 as a reactive diluent. The use of NH, to prevent secondary amine formation is well documented (Freifelder, 1971). The results of these studies are shown in Table I. We have found't&at the rate of such nitrile reductions must be kept at a maximum to avoid reversal of the cyanoethylation reaction, while the residence time must be kept at a minimum to avoid hydrogenolysis of the N-C bonds. Attempts to hydrogenate in typical laboratory stirring autoclaves typically resulted in yield losses of 1&25% due to the aforementioned side reactions. Therefore, the reductions were performed in a modified trickle bed reactor in which the nitrile to be hydrogenated is pumped onto a vertical tube filled with catalyst at a constant H2pressure and is allowed to ''trickle- by gravity feed through the catalyst particles. This downflow "trickling" process produces the highest possible catalyst and hydrogen concentrations for the nitrile substrate, and, therefore, the
optimum conditions for hydrogenation. The utility of trickle bed reactors for other industrially significant reactions has been described by Satterfield (1980). Using Girder, Houdry-Hiila, or Harshaw commercially available cobalt catalysts, one obtains an approximately 90-100% conversion of the nitrile in approximately 3 h. (For descriptions of the catalysts, see footnotes in Table I). The major advantage of the Houdry-Huls catalyst was the marked inhibition of the intramolecular ring closure reaction to homopiperazine (see reaction 4 in Table I). We speculate that this catalyst might be valuable in other RR'(CH2),CN reductions where one seeks to inhibit the intramolecular reaction. The major difference between the catalysts is the preesence of Mn and Ag in the HoudryHills catalyst. The mechanism by which the added Ag and Mn inhibit the intramolecular reaction is intriguing and merits further study. From Table I, reactions 5 and 6 reveal that the reduction of I1 is slower than the reduction of I which is a t least in part due to the presence of 2CN functionalities per mole. Similar conversions and selectivities compared to I were obtained in the reduction of I1 when a 21/1 NH3/CN ratio was employed. At first glance, these data are deceiving since reaction 6 was carried out a higher space velocity (0.35 vs. 0.15 g of liquid/g of cat./h) then reaction 5. However, the higher NH3/VCN ratio of 21 vs. 7.5 really indicates a longer residence time for the nitrile in reaction 6. Reaction 7 was chosen to illustrate the reduction of a typical industrial reaction effluent having distilled off only excess VCN. In reactions 1-6, reversal of the cyanoethylation would produce VCN and EDA or I. The produced VCN could then react accordingly. VCN + NH3 HzNCH2CH2CN
-+
VCN EDA I VCN I --c I1 +
+
The lack of 1,3-propylenediamine in the product mixes is indicative of a faster reaction rate for VCN with primary amine vs. NH,. In reaction 7, analysis of the product stream indicates 36% of I1 reversed to I and VCN, and this VCN then reacted with the excess EDA to produce even more I. Again, as in reaction 1-6, the rate of reaction with primary amines (specifically EDA) is faster than with NH3. Such transcyanoethylations have been observed in aromatic nitrile reduction (Butskus, 1961), and by Whitmore et al. (19441, who observed that VCN reacts with aminopropionitrile approximately 20X faster than with NH,. Conclusions We have found that the reaction of VCN with EDA proceeds rapidly at 45 "C to achieve a statistical distribution of nitrile products without any need for acid catalysis. Hydrogenation (at 85 "C and 1500 psi H2) of such products over commercially available cobalt catalysts in a modified trickle bed reactor results in approximately 90% yields of the desired amines. If one hydrogenates the exiting effluent prior to removal of excess EDA, the expected product composition is altered due to transcyanoethylation to the EDA molecule. Intramolecular ring closure of CN(CH2)2NH(CH2)2NH2 (I) to give homopiperazine I11 can account for up to 10% of the products. However, a proper choice of catalyst can virtually eliminate the production of 111. Literature Cited Amerlcan Cyanamid "The Chemlstry of Acrylonhrlle", 2nd ed.,New York, 1959 p 22. Badische Anllln 8 Soda-Fabrlk AG, German Offen. 2 140 151, 1974. Bruson, H. A. "Organic Reactions", Adams, R., Ed., Wlley: New York, 1949, Vol. 5, p 79.
Ind. Eng. Chem. Prod. Res. Dev. Butskus, P. F. Russ. Chem. Rev. 1881, 30, 583. Dickerman, S. C.; Simon, J. J. Org. Chem. 1857,22, 259. Freifelder, M. "Catalytic Hydrogenation"; Wiley-Interscience: New York, 1971,Chapter 12. Levenspiel, 0. "Chemical Reaction Englneering", 2nd ed.;Wiley: New York,
1982,21, 635-639
835
Spltz, R. D. I n "Kirk/Othmer Encyclopedia of Chemical Technology", 3rd ed.; Wliey: New York, 1980,Vol. 7, p 580. Whitmore, F. C.; Mosher, H. S.; Adams, R. R.; Taylor, R. 8 . ; Chapin, E. c.; Yanko, W. J. Am. Chem. SOC.1844, 6 6 , 725.
1972. I) 101.
Receiued for reuiew February 1, 1982
SatterfG6 C. "hterogeneous Catalysis in Practice"; McGraw-Hill: New York, 1980, p 318.
Accepted March 29, 1982
Infrared Emission from Gas-Aerosol Reactions Raymond A. Mackay Department of Chemistry, Drexel Unlverslty, Philadelphia, Pennsylvania 19 104
The present study has been designed to produce infrared emission by means of an exothermic reaction between a liquid aerosol and a gas. A number of gas-aerosol systems employing acid-base reactions have been examined, and significant levels of radiation have been observed from the reaction of chlorosulfonic acid aerosol with gaseous ammonia and water. Other systems which were screened, including sulfuric acid-ammonia, octanoic acid-ammonia, and octytamlne-hydrogen chloride, have produced detectable levels of radiation. Some-methodsfor the investigation of emission from gas-aerosol reactions have been explored, and the results of these studies utilizing a chlorosulfonic acid aerosol system are presented.
Introduction Infrared radiation in the atmosphere above normal background levels can be produced in a variety of ways. For example, combustion gases (Handbook of Infrared Radiation, 1973) can produce significant amounts of radiation in the infrared region (2-20 pm). However, the total mass of material, and thus the radiant emittance, is small. In addition, the gas cloud rapidly cools and disperses. In order to significantly increase the amount of airborne material, an aerosol must be employed. At ambient temperatures (25 "C) the maximum of the blackbody emission curve is at about 1030 cm-I (9.7 pm), which means that a few degrees increase in temperature will produce an increase in spectral radiance on the order of 5 % at 5-10 pm. A highly conducting and absorbing substance may approximate a blackbody, but most real aerosol particles will at best be "grey" bodies with perhaps some superimposed structure (selectivity). The (equilibrium) thermal emittance of an aerosol particle will therefore be less than that of a blackbody at the same temperature, according to Kirchhoff s law. A second and more selective method could involve the production of infrared fluorescence. This would have the advantage of being selective as to wavelength range and would thus also require much less total energy input. Unfortunately, due to the rapidity with which vibrational energy degrades, it does not seem possible to employ this approach. In other words, in order to obtain infrared (IR)luminescence in any observable yield, a dilute gas and a high-energy IR laser are normally required. We have therefore focused attention on thermal emission. In order to accomplish the desired goal, an exothermic chemical reaction must be employed to raise the temperature of the aerosol particle above ambient. This reaction must ultimately involve either the reaction of a gas with the aerosol or the simultaneous generation of a very highly disperse co-aerosol which will rapidly coagulate with the c m e r aerosol and react. This is technically more difficult, and there would always be the problem of the highly disperse aerosol coagulating with itself. In the former case,
the gas could be generated as a volatile co-aerosol. A large number of chemical reaction studies have been performed on the upper end of the particle size spectrum with regard to the combustion of fuel sprays and dust clouds (Essenhigh and Fells, 1960). These involve particles in the super-micron range (10-1000 pm). Chemical reactions responsible for heterogeneous nucleation have also been investigated (Mirabel and Katz, 1974; Shen and Springer, 1976; Reiss et al., 1976), and the particle sizes here are below 0.1 pm. However, once aerosols are produced either by nucleation and growth in the atmosphere or by nebulization followed by settling and coagulation, most of these achieve a relatively stable existence in the 0.1-10 pm range as smokes, fogs, etc. (Friend, 1966; Heard and Wiffen, 1969). These systems are polydisperse, with number densities generally less than lo7 ~ m - ~ Nonethe. less, there have been relatively few studies of the reactions of gases with aerosol particles in this size regime. In principle, the reaction rate may be controlled by gas-phase diffusion, diffusion of reactants and/or products in the particle, by the bulk chemical reaction, or by processes occurring in the droplet interface region. Cadle and Robbins (1960) have developed equations for some limiting cases and applied them to the reaction of ammonia with sulfuric acid aerosols in the 0.2-0.9-pm range and to the reaction of NOz with a sodium chloride aerosol. In the former case, it was suggested that the rates in concentrated H2S04droplets were controlled by diffusion of reaction product in the particle. In dilute H2S04droplets, the rate was too fast to measure on their apparatus, and it was presumed that the rate was controlled by gas-phase diffusion (Cadle and Robbins, 1960; Robbins and Cadle, 1958). There have also been a number of studies of the metal-catalyzed oxidation of SOz in aqueous aerosols. Johnstone and Coughanour (1958), using a suspended 5001000-pm drop, concluded that the reaction was liquidphase controlled with all of the reaction occurring in the outer shell at high manganese concentration. A t lower catalyst concentration, the SO2penetrated to the center 0 1982 American Chemical Society
