NOVEMBER, 1964
REACTIOKS OF ETHYLEKE DIISOCYANATE
The reaction mixture was refluxed (132") and sampled at intervals. The cooled samples were inserted into a 0.1-mm. NaCl infrared cell and scanned in the 4.0-6.0-p region. The absorbance at Amax 4.45 /.I was compared with a plot of absorbance us. concentration determined from spectra of standard solutions of ethylene diisocyanate in chlorobenzene. B. Codistillation of Ethylene Diisocyanate with the Solvent .A solution of 1.0 g. (0.0068 mole) of I1 in either 49 g. of o-dichlorobenzene (180" at 760 mm.) or I-chloronaphthalene (226230" at 10C200 mm.) was distilled with and without phosgene passing through the distillation apparatus. The yields of ethylene diisocyanate according to the infrared spectra of the distillates were 40 and 5OcX, respectively. C . Dehydrochlorination with Triethylamine .-To a suspension of 78 g. (0.525 mole) of I1 in 1500 ml. of benzene a solution of 50.5 g. (0.50 mole) of triethylamine in 250 ml. of benzene was added at 30-35" over a period of 40 min. After standing for 30 min. at 45-50', 66.1 g. (96%) of triethylamine hydrochloride was removed by filtration. The filtrate was heated to 70-75" and about 10 g. of phosgene was added. The solution was cooled and filtered (1.3 g. of solids) and after evaporation of the solvent and distillation in vacuo, 38.6 g. (69%) of ethylene diisocyanate, b.p. 65-68' (7.5 mm.), b.p. 77-80', 16 mm.), lit.2 b.p. 75-76" (14 mm.), 122311 1.4472", were obtained. The diet,hylurethane obtained on treatment with excess ethanol The mixture melthad m.p. 107-108" (lit.'* m.p. 110'). ing point with a sample prepared from ethylene diamine and ethyl chl~roformate'~ showed no depression. Carbonylbisimidazolidinone-2 (XIII). A. From Ethyleneurea and 11.-A mixture of 1.72 g. (0.02 mole) of ethyleneurea and 2.97 g. (0.02 mole) of I1 in 60 ml. of ethylene dichloride was refluxed for 5 hr. with nitrogen passing through the reaction mixture to remove the generated hydrogen chloride. Evaporation afforded 3.5 g. (887G)of carbonylbisimidazolidinone-2: m.p. 218-222"; (infrared) 3.11, 5.78, 5.94, 6.72, 7.1, 7.4, 7.95, 8.8, and 9 . 4 /.I. Anal. Calcd. for C7H10x403: C , 42.42; H, 5.09; S , 28.27. Found: C, 42.62; H, 5.27; S , 28.32. B. From Ethyleneurea and Ethylene Diisocyanate.-A mixture of 0.101 g. (0.001175 mole) of ethyleneurea and 5.03 g. (0.0449 mole) of ethylene diisocyanate was heated to 68" (12 (12) T. Curtius, J . prakt. Chem., [ 2 ] 62,210 (1895). (13) E. Fischer,and H. Koch, Ann., 282, 222 (1886).
3347
mm.) when the formation of a precipitate was observed. Removal of the excess diisocyanate in vacuo gave 0.275 g. (calculated, 0.233 g.) of XIII. The infrared spectrum of the residue after trituration nith chloroform was identical with that of XI11 obtained according to A. Ethylenedicarbamoyl Chloride (XIV).-To 100 ml. of benzene saturated with hydrogen chloride, 0.574 g. (0.005 mole) of ethylene diisocyanate in 30 ml. of benzene was added dropwise with stirring. An immediat'e precipitation was observed. The excess hydrogen chloride was removed with nitrogen and filtration afforded 0.611 g. (72%) of ethylenedicarbamoyl chloride XIT': m.p. 95-97' (resolidified above 100" and remelted at, 145150'); ?: :A (infrared) 2.95 ( S H ) and 5.67 /.I (C=O). The infrared spectrum of the resolidified material in chloroform was identical with the infrared spectrum of 11. Anal. Calcd. for C ~ H ~ C I Z N Z O C1,~ : 38.20. Found: C1, 38.69. Dehydrochlorination of Methyl-2-imidazolidinone-N-carbonyl Chloride (IIa).-To a stirred suspension of 16.3 g. (0.1 mole) of IIa in 350 ml. of benzene there was added droptvise during 10-20 min. 10.0 g. (0.099 mole) of triethylamine. The temperature rose from 31.5 t o 35" by the heat of reaction. The mixture was stirred another hour a t 3 5 4 5 ' and worked up in t,he same way as applied to the ethylene diisocyanate preparation. From the 13.3 g. of pale yellow, crude concentrate there was obtained on distillation 8.7 g. (6970) of 1,2-diisocyanatopropane ( X I a ) , b.p. 81" (19 mm.) and 70" (12.5 mm.), n Z 41.4407. ~ Reaction of 1,2-Diisocyanatopropane with Aniline .-To 2.53 g. (0.02 mole) of 1,2-diisocyanatopropane in 40 ml. of benzene was added a solution of 1.87 g. (0.02 mole) of aniline in 30 ml. of benzene. After reflux for 16 hr. at 80", 0.109 g. of fine needles, m.p. 235-239", was filtered off. Evaporation of the solvent afforded 4.265 g. (97%) of methylimidazolidin-2-one-1-carboxanilide, m.p. 120-130", recrystallized (ethanol) m.p. 145-147". Anal. Calcd. for CIIHIJV&: C, 60.26; H, 5.98; N, 19.17 Found: C, 60.10; H, .5.98; N, 19.15. The higher melting (235") material according to its infrared spectrum is believed to be the straight-chain bisurea.
Acknowledgme3t.-The authors wish to express their gratitude t'o Mr. B. Tucker and Mr. F. Geremia for valuable assistance in the experiments and in the infrared determinat'ions.
Some Reactions of Ethylene Diisocyanate JAMES N. TILLEYAND A. A. R. S A Y I G H ~ Carwin Research Laboratories, The Upjohn Company, Sorth Haven, Connecticut 06473 Recesved January 8, 1964 Ethylene diisocyanate reacts with active hydrogen compounds such as primary and secondary amines, alcohols, and mercaptans to form the novel, cyclic 1: 1 adducts, 1-substituted 2-imidaxolidinones, in high yield, Even with excess of the active hydrogen compound (ethanol) there is a considerable yield of the 1: 1 cyclic adduct as opposed to the expected 1: 2 linear adduct, the bisurethan; however, trimethylene diisocyanate appears to have much less tendency to produce the analogous cyclic 1: 1 adducts (hexahydropyrimidones).
I n the course of our study of the reaction of ethylene diisocyanate (I), we have established that by far the preferred course of reaction of I with an active hydrogen compound (11) is the formation of the cyclic inonoadduct (111), a derivative of imidazolidinone, rather than the linear bisadduct (IV). The literature contains brief reference by Naegli and S ~ h r o e t e ronly ~ , ~ to the possibility of the existence of such cyclic compounds. However, there is abundant evidence for the formation of the linear b i s a d d u c t ~ ~such - ~ as the bisethylurethan (IVa) and the bisphenylurea (IVb) as shown by Curtius (1) T o whom inquiries should be directed. (2) C. Xaegeli and P. Lenderoff, H e h . Chem. Acta. 16, 49 (1932). (3) G . Schroeter and C. Seidler. J . prakt. Chem., 106, 165 (1923). (4) E. Fischer and H. Koch, Ann., 232, 222 (1886).
and Fischer. Presumably, these cyclic materials; I I I a and IIIb, were overlooked. Preliminary evidence that the cyclic products have formed came about by the observation during a kinetic study that the infrared spectrum of a reacting mixture of butanol with ethylene diisocyariate exhibited developinent of a carbonyl triplet, A$::"" 5.57, 5.69, 5.8 p , whereas under the same conditions hexaniethylene diisocyanate reacted only to forin a singlet, X$Ecane 5.79 p . Species such as IVa or the hexamethylene homolog would be expected to have a single carbonyl, whereas IIIa (and mixtures of IIIa with IVa) would have at least two carbonyl absorptions. (5) T. Curtius J . prakt. Chem.. 62, 210 (1895). (6) T. Curtius and Hechtenherg, zhzd., 106, 285 (1923)
TILLEY AND SAYIGH
3348 CHzNCO
I
0 II
0 -
CHzNCO
I
/C-XR NH
RXH I1
V
I11 CHaCHiOH
N VI RX
IIa, I&, IS'a b IIc, IIIc d
CZHO CsHJVH(CzH5)W C4H9S-
This was confirmed by preparing authentic pure I I I a and IVa independently' from ethyleneallophanoyl chloride (VI) and making a comparison with the products of the interaction of ethanol (IIa) and I under a variety of conditions. As shown in Table I, the cyclic product I I I a is produced even in large excess of ethanol, wherein one would expect only IVa to form.
VOL. 29
by cyclization with elimination of ethanol, with or without the attack of other isocyanate molecules, seems unlikely since the bisurethan is quite stable. There is no more than 1% cyclization of bisurethan to imidazolidinone during 1.5 hr. at 150' or 1 hr. at 190'. For comparison, trimethylene diisocyanate (VII) reacted with ethanol in benzene (16 hr. at 80') giving, at 1:2 mole ratio, practically pure bisurethan (IX) and complete isocyanate consumption. At a 1: 1 mole ratio, the infrared spectrum indicated that ca. half of the isocyanate was consumed in 16 hr., but urethan carbonyl was indicated at about 50% of that found in the 1 : 2 case. Since the only change during another 24 hr. of reaction was a slight decrease in isocyanate absorption and appearance of a weak additional carbonyl absorption, it is inferred that the reaction course is predoniinantly to the bisurethane (IX) and unreacted triinethylene diisocyanate, and perhaps only slightly to the cyclic allophanate, ethyl hexahydropyrimidone-2S-carboxylate (X). The isocyanate absorption still being considerable, X is ruled out as a major product, although statistically 33% of the ethanol could be incorporated as such and the remaining 67% as bisurethan (IX). Thus, cyclization appears to be considerably less important for the reaction with trimethylene diisocyanate than with ethylene diisocyanate.
VI11
TABLEI (IIa) ETHYLENE DIISOCYANATE ( I ) WITH ETHANOL IIa:I, moles
Solvent
16: 1
Ethanol
Total product compn., %" (m.p., O C . )
J
Pure product isolated (m.p., "C.b
II-a, 90-95 IS'a (108) (80-90) 2:l Benzene IIIa, 86 1:l Benzene IIIa, 100 I I I a (123-125) (112-125) a The infrared analytical method is described in the Experimental section.
The generality of t,he novel react'ion of ethylene diisocyanate with active hydrogen compounds has been established by similarly preparing t'he other new compounds IIIb-IIIf and characterizing them. As has been predicted* in a siniilar instance, the ease of reaction appears to be in the order R&H > RNH, > ROH > RSH with the mercaptan reaction requiring catalysis by triethylenediamine. The products all have a pair of carbonyl absorptions in the infrared. In view of the low reactivity of isocyanates toward carbamate S - H bondsJgthe reaction is unique yet may serve as another example, via an intermediate such as V, of the overwhelming kinetic effect incurred by intramolecular proximity, especially that which can yield five-inembered rings. lo While no intermediate such as V has been isolated, the products isolated strongly suggest it as being most likely. Preliminary formation of bisuret,han followed (7) A . A . R. Sayigh. J. N . Tilley, and H. Glrich, J . Org. Chem., 29, 3344 (1964). (8) R . G. Arnold. J. A . Nelson. and J. J. Verbanc, Chem. Rel;., 57, 49 (1957). (9) J. H.Saunders, Rubber Chem. Technol., 82,338 (1959).
/CHzNHCOOCzHs CHa
\CH~NHCOOC~H~ IX
HnNCHzCHzNHz
X
Experimental12 General synthesis of imidazolidinones is illustrated by the synthesis of I I I b . A solution of 1.86 g. (0.02 mole) of I I b in 75 ml. of benzene was added to 2.24 g. (0.02 mole) of I in 50 ml. of benzene (precipitate); the mixture was refluxed for 3 hr., cooled, and filtered. The solid, 3.308 g. (80% yield), m.p. 166", rerrystallized from ethanol, had m.p. 167-168" and the analysis agrees with I I I b (see Table 11). When this was repeated using instead 3.92 g. (0.04 mole) of IIb, 0.43 g. ( l l Y G )of ethanoiinsoluble material, m.p. 251-256" (lit.6 for IVb, m.p. 245O), was obtained together with 3.6 g. ( 6 l y G )of I I I b . With aziridine (10) Ozaki," el al., have been able to obtain ethylene urea easily by addition of ethylene diisocyanate to water, b u t a s the chain is lengthened, high dilution technique must be used to decrease polymer formation. Also, pertinent are the findings of R. N. Johnson and H. W. Woodburn [ J . Ow. Chem., 27, 3958 (1962)l who recently showed t h a t in the reaction of ethylenediamine with trifluoracetonitrile the cyclic amidine was the exclusive product in methanol. While this paper was in preparation, F. D'Angeli, el ai. [ J . Ow. Chem., 28, 1596 (1963)l. reported t h a t ethylene diisothiocyanate also reacts preferentially with cyclization t o t h e analogous thioimidiazolidin-2-one-l-thiocarboxyl derivatives. (11) S. Ozaki, T. Mukaiyama. and K. Uno, J . A m . Chem. SOC..79, 4358 (1957). (12) Elemental Analyses were made by Schwarzkopf Microanalytical Laboratory, Woodside, N. Y . All melting points were made on a FisherJohns apparatus. Spectra were taken on a Perkin-Elmer Model 21 infrared spectrophotometer.
REACTIONS OF ETHYLENE, DIISOCYANATE
KOVEMBER, 1964
3
I
W
d
h
E,
w i
U
6
e d
2
5
m
a
h
h
l
:
3 4
3349
( I I e ) or diethylamine (IIc) the reaction mixture ( 1 : 1 mole ratio) was allowed to stand a t room temperature, with the respective crude products being I I I e (solid) and IIIc, an oil difficult to crystallize. With butylamine ( I I f ) and butyl mercaptan ( I I d ) , the ( 1 : 1 mole ratio) reaction mixture in benzene was refluxed several hours, the latter with a trace of added triethylenediamine catalyst; the respective products being IIIf, a low-melting solid, and IIId, a very soluble solid. Physical and analytical data are in Table 11. Reactions with Ethanol (Table I ) . A . Ethanol Alone.-Into approximately 40 ml. of specially dried 2B ethanol was weighed 5.0 g. of I.' The solution was homogenized and immediately became very warm (6CL70"). It was then refluxed 15 min., cooled, and evaporated in uucuo to a residue which was examined in the infrared ( 5 % in chloroform). B. With a 1 : 1 or 2 : 1 Mole Ratio of Ethanol-I in Benzene.Benzene solutions were made up of 0.02 or 0.04 mole/50 ml. of solvent as in the general procedure and refluxed for 3 hr. or until disappearance of isocyanate, A,, 4.45 p . These were then evaporated to a residue and examined in the infrared by comparison of infrared spectra (5% chloroform) with those of a series of 5.60, 6.60, and 6.75 standard mixtures of IIIa and IVa using A,, Results are shown in Table I . p. Authentic IIIa and IIIc have been synthesized, respectively, from I I a and IIc by reaction with ethyleneallophanyl chloride ( V I ) as described by Sayigh, et al.' Authentic 1Va.-According to Fischer's p r ~ c e d u r e , 5.45 ~ g. (0.0502 mole) of ethyl chloroformate in 25 ml. of diethyl ether was added dropwise to a solution of 3.2 g. (0.05 mole, 91-937, purity) of ethylenediamine in 40 ml. of ethanol with agitation and ice cooling. The slightly damp filtered solid weighed 3.6 g. (calculated, 3.33 g. for amine hydrochloride). The filtrate was evaporated t o residue which was twice recrystallized from water, m.p. 107-109" (lit.4f5 m.p. 112, l l O o ) , 4.13 g. (81%). The infrared spectra (chloroform) of crude and purified material were 2.92 ( X H ) , 5.85 (C=O), other 6.60, virtually the same: A,, 7.25, 7.50, and 7.95 p . Thermal Stability of the Bisurethan.-A 0,1009-g. sample of the bisurethan was heated in an oil bath a t 150 f 5' for 90 min. (crystallized on cooling), resulting in a weight loss of 1.1 mg. (l.lyo). The weight loss corresponds to less t h m 5% theoretical conversion to the imidazolidone (theoretical loss of ethanol). The infrared spectrum (CHCI,) indicated, by the complete absence of the 5.6-p band, that not more than 1% of the imidazolidone had formed and the spectrum was essentially superimposable on that of the starting bisurethan. Heating at 190 f 5" for 60 min. caused a 46% weight loss (20% theoretical conversion), but the infrared spectrum indicated only a slight shoulder at 5.6 p indicative of about 1 % conversion, definitely less than 5% by comparison with standard mixtures. During 20 min. at 230" some needles sublimed in the test tube and the spectrum of the residue now showed 15y0 conversion, having the appearance of authentic mixtures of bisurethan and imidaxolidone. The sublimation suggested that weight loss was not a good criterion of extent of reaction. Reactions of Other Alcohols with I.-Methanol, butanol, 2propanol, and ethylene glycol have been treated in the same ways as ethanol with the observation that total crudes exhibit spectral proof of cyclic monoadduct formation, there being a pair of carbonyl absorptions (5.6 and 5.8 p ) in each case. Trimethylene diisocyanate (VII) was prepared by the method of Smith,13 35.2 g. of glutaric dihydrazide affording 12.6 g. (467% yield) of distilled material, b.p. 86.5" (14.5 mm.) to 88" (14 mm.), lit.1*J4b.p. 86" (14 mm.) and 98" (26 mm.), n 2 3 , 6 ~ 1.4485. Reactions with Ethanol. A.-In 25 ml. of benzene were placed 1.257 g. (ea. 0.01 mole) of trimethylene diisocyanate and 0.930 g. (cu. 0.02 mole) of absolute ethanol. The mixture was refluxed (80") and sampled a t 16 and 24 hr., from which the infrared spectra indicated complete consumption of isocyanate a t 16 hr. Evaporation afforded a clear oil, 2.064 g. (SOC/;,, which crystallized on standing within 2 days. The infrared spectrum was identical with that of the authentic bisurethan (see below). B.-In the same way, 1.268 g. (ca. 0.01 mole) of the diisocyanate and 0.467 g. (ea. 0.01 mole) of ethanol reacted in re(13) P. A. S. S m i t h . Ow. Reactions, 9 , 344 (1946). (14) W. Siefken, A n n . , 663, 75 (1949); Chem. Ab&..
4 4 , 115 (1950).
DA SETTIMO A N D SAETTONE
3350
fluxing benzene, with sampling a t 16, 24, and 40 hr. The spectra 5.87 p , which was of the reaction mixture showed carbonyl,, , A, about one-half as intense as that for A and remained unchanged 4.44 p ) from 16 to 40 hr. The isocyanate content was high ,,A,( being only about half consumed at 16 hr. At 24 and 40 hr., this had decreased slightly, and simultaneously a shoulder appeared a t 5.65 p , which could have been due to traces (possibly 5-107') of the cyclic trimethylene allophanate. On evaporation, a pungent (isocyanate) pale yellow oil, 1.653 g., was obtained, (infrared) 2.93, (3.06), 3.42,4.42, which did not crystallize:3':Af:
VOL.29
(5.65), 5.85, 6.58, 6.75, 7.85 to 8.25,8.75 and 9.6 p . The figures in parentheses could belong to the allophanate ( X ) , whereas all the others are found in the bisurethan ( I X ) but also possible in X . Authentic Trimethylene Bisethylurethan-The procedure for IVa was repeated using, however, 3.7 g. of trimethylene diamine with 5 . 5 g. of ethyl chloroformate. The crude (oil) product obtained on evaporation of the filtrate slowly crystallized. I t was distilled, b.p. 136" (0.1 mm.), affording 4.5 g. (8Oc/c) of a soft crystalline solid: m.p. 39-42' (lit.3m.p.42'); b.p. 210' (30 mm.); CHCl3 X,, (infrared) 2.94, 3.42, 5.87, 6.60, 8.05-8.25, 8.75, and 9.6 p .
Conversion of a 1,2-Glycol Monocarbamate into Two Isomeric 1-Piperidinecarboxylat es AXTONIODA S E T T I M O
AiYD JI.4RCO
F.
SAETTONE
Institute of Phar.naceutica1 Che nistry o j the C'niversity of Pisa,Pisa,Italy Receized June 1 , 1964 Interaction between piperidine and 2-hydroxy-3-phenyl-3-piperidinopropylcarbamate results in the formation of the two structurally isomeric l-piperidinecarboxylates of 3-phenyl-3-piperidino-1,2-propanediol.Both isomers react with SOClz to give the same l-chloro-3-phenyl-8-piperidino-2-propyl1-piperidinecarboxylate. The structures of these compounds were proved by unequivocal synthetic routes.
An investigation still underway in our laboratories dealt with the synthesis of several 3-amino-2-hydroxy3-phenylpropyl carbamates, which were obtained by the action of primary and secondary amines on trans2,3-epoxy-3-phenylpropyl carbamate (I, Chart I ) .
CHART I
ture, the isonieric erythro-1-piperidinecarboxylates I11 and IV being the only reaction products. These were fornied also from I1 and piperidine at reflux temperature. Compounds I11 and IV had different melting points and infrared spectra. and were transformed by acetic anhydride into the acetates IIIa and IVa. S t r x t x a l Investigation.-Hydrolysis of 11, 111, and IT' yielded the known crythro-3-phenyl-3-piperidino-l,2propanediol* (V), 1n.p. 93-95 O , thus indicating I11 and IV to differ only with respect to the position of the esterifying group. Furtherinore, both I11 and IV gave upon treatment with SOClz the same chlorinated derivative, erythro- l-chlo~o-3-phenyl-3-piperidino-2-propyl 1-piperidinecarboxylate (VI), whose structure was proved as shown in Chart 11. Reaction of VI with piperCHART
II
111,
H
0 X - O b
HCI,N.OH
Ilk - C - " c )
b
IV
-:-CH3 0
H 0
Nti
- 5 - C H3 0
Interaction between I and piperidine at rooni teniperature gave the expected erythro-2-hydroxy-3-phenyl-3piperidinopropyl carbamate' (11), no other reaction occurring than the opening of the epoxide ring. However, when the reaction was carried out at higher teniperature no I1 could be detected in the reaction mix-
,CH,Cl
"p.\ Ph
H
IX ( 1 ) trans epoxides are known to react with amines with inversion of configuration a t the point of attack i o yield the erythro isomers [ c j . R. E. Parker and N. S. Isaacs, Chem. Ree., 59, 737 (195911. We have therefore assigned the products of ring opening of I (which, being derived from trans-cinnamyl alcohol, is a trans epoxide) and their derivatives the erylhro configuration. (2) K. C. Tsou a n d N. H. Cromwell, J . O r g . Chem., 16, 1293 (1950). .knother 3-phenyl-3-piperidino-1,2-propanediol. m.p. 126", described in the literature IC/. K. Bodendorf a n d B. Binder, Arch. Phorm., 287, 453 (1954)1, is probably the threo isomer. Investigation is under way to clarify this point.