[CONTRIBUTION FROM
TlIE
DEPARTMENT OF CHEMISTRY, BROOKLYN COLLEGE]
THE ACTION OF AMMONIA ON BENZOIN DAVID DAVIDSOP;, MARVIS WEISS,
AND
MURRAY JELLING
Received August 23, 1937
The action of ammonia on benzoin has been studied previously by Laurent,' Erdmann,2 and Japp and Wilson.3 These investigators found that the reaction leads to three products-amarone (C2sH22N2, m.p. 245") [also called benzoinimide (Erdmann) and ditolane-azotide (Japp and Wilson)], benzoinam ((&&N20, m.p. 190"-220" depending on the rate of heating), and benzoinidam (CzsH23N02, m.p. 199"). With fused ammonium acetate at elevated temperatures, Japp and Wilson obtained a fair yield of amarone (55%), while with ammonium formate at 230' Leuckart4 reported a quantitative yield of amarone with only traces of by-products such as lophine and benzaldehyde. Of the products mentioned above only amarone has been assigned a structural formula. This Japp and Wilson3 represented as tetraphenylpyrazine (I). N
C~H,C-NH
I
I1
The present paper deals with an examination of the action of ammonia in acetic acid on benzoh6 In this medium it is found that benzoin yields mainly amarone (I) and 2-methyl-4,5-diphenylglyoxaline(11) together with dihydroamarone. Some light is thrown on the mechanism of this reaction by studying other acids particularly formic, and, further by testing the action of ammonia in acetic acid on esters of benzoin. This latter LAURENT,J. prakt. Chem., [l],36, 5 (1845). ERDMANN, Ann., 136,181 (1865). 3 JAPPAND WILSON, J. Chem. Soc., 49, 825 (1886). 4 L ~ J . prakt. ~ ~Chem., ~ [2],~41, 330 ~ (1890). ~ , 1 2
6 Previous papers from this laboratory dealing with the action of ammonia in acetic acid on carbonyl compounds: ( a ) DAVIDSON, J . Am. Chem. Soc., 48, 1821 (1936); (b) DAVIDSON, WSISS, AND JELLING, J. ORG. CHEM.,2, 319 (1937). 328
329
ACTION O F AMMONIA ON BENZOIN
reaction produces good yields of oxazoles (accompanied by subordinate amounts of glyoxalines), thus furnishing a new method of producing such cycles.6 THE MECHANISM OF THE ACTION O F AMMONIA ON BENZOIN
Since it is known that aniline converts benzoin into N-de~ylaniline~ (111), it is reasonable to assume that the primary product of the action
I11 of ammonia on benzoin is desylamine (IY). In this reaction it is probably the carbonyl group which is converted to the carbinamine group, the process involving condensation followed by tautomerization, thus : CsHsCO CCHSCHOH
--+
CeH& :NH
C ~ H ~ C H N HA.off, Z C~H~CHNHAC -
3
C6HjkHOH
C&f5C0
I
C6HbCO
-HI
Desylamine, in tim, may react in two ways: (1)undergo autocondensations to dihydroamarone (V) ; or (2) condense with acetic acid to form N-desylacetamide (VI). The latter has been shown to react with ammonia in acetic acid to form 2-methyl-4,5-diphenylglyoxaline(II),5bwhile Gabriels has reported that dihydropyrazines resembling V are readily oxidized to pyrazines. 6 Oxaeoles derived from benzoin have previously been prepared by ( a ) cyclodeJ. Chem. SOC.,103, 1331 hydration of N-desylbenzamide, MCKENZIEAND BARROW, (1913); ( h ) the condensation of benzoin with nitriles, JAPPAND MURRAY, ihid., 63, 469 (1893). 7 A recent study of this reaction has been made by CAMERON, Trans. Roy. SOC. Can., [3],23, Sect. 3 , s (1929);26, Sect. 3, 145 (1931). 8 Compare the behavior of other alpha-aminoketones, GABRIEL, Ber., 41, 1127 (1908); HILDESHEIMER, ibid. 43, 2796 (1910).
330
D. DAVIDSON, M. WEISS, AND M. JELLING
The formation of tetraphenyldihydropyrazine in the reaction of ammonia on benzoin in acetic acid is evidenced by the orange color which sets in soon after refluxing of the reaction mixture is begun. That air plays a r6le in its conversion to amarone is indicated by the following experiments (Table I) in which the heating was carried out (1) with air bubbling through the solution, (2) under normal conditions, and (3) in a closed tube. The amarone produced was separated by filtration, and the dihydroamarone remaining in tbe mother liquors was determined by adding nitric acid until the orange color was discharged, whereupon a second crop of amarone precipitated, while a fugitive green color was observed.* TABLE I FACTOR IN THE PRODUCTION OF AMARONE (In each experiment, 1.06 g. of benzoin, 10.0 g. of ammonium acetate, and 25 cc. of acetic acid were refluxed for one hour.)
AIR AS
A
1
Y I E L D OF A M A R O N E
I
lat crops
Air bubbled thru solution . . . . . . . . . . . . . . . . . Normal refluxing.. ......................... Closed tube.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0.41 0.35 0.27
CONDITIO RH
EFFECT OF
1
I
2nd crop (by "01).
~
TABLE I1 AMMONIUM ACETATEON
g.
1
0.14
Total, g.
0.55
8::;
~
0.55 0.51
YIELDS
(In each experiment 1.06 g . of benzoin was refluxed for one hour in 25 cc. of acetic acid with the amount of ammonium acetate indicated.) AMMONlU?d ACETATE E M P L O Y E D , G
3 5 10
1
1 1
I
Y I E L D OF A M A R O N E
~
~
1st crop, g.
0.22 0.31 0.35
I
1
2nd crop, 8 .
0.09 0.12 0.20
1 ~
Total, g.
0.31 0.43 0.55
1 j
Y I E L D OF GLYOXA L I N E , ct.
0.24 0.36 0.28
Table I1 displays the influence of the ammonia concentration on the total yield of amarone and the yield of the glyoxaline derivative. The mechanism outlined above is supported by an experiment in which benzoin was replaced by an equivalent quantity of desylamine hydrochloride, without influencing the yields of arnarone and 2-methyl-4,5,
* Unpublished studies in this laboratory indicate that the evanescent green color is due t o the action of nitrous acid on a colorless reduction product of amarone. This suggests that part of the dihyroamarone formed in the reaction disproportionates to amarone and the above-mentioned reduction product.
331
ACTION OF AMMONIA ON BENZOIN
diphenylglyoxaline. Replacement of the acetic acid by propionic acid did not affect the yield of amarone materially but led, of course, to another imidazole derivative; Le., 2-ethyl-4,5-diphenylglyoxaline. THE ACTION O F AMMONIA ON BENZOIN IN THE PRESENCE OF FORMIC ACID
In the presence of formic acid, the formation of pyrazines by the action of ammonia on benzoin was suppressed. Beside the expected 4,5-diphenylglyoxaline there appeared a fair a.mount of N-desylformamide (VII) . This result stands in sharp contrast to that of Leuckart4who obtained only amarone from the action of ammonium formate on benzoin at 230'. N-Desylformamide may be prepared by refluxing desylamine with formic acid. Like other N-desy1amidesls N-desylformamide is smoothly converted to the corresponding glyoxaline (VIII) by treatment with ammonia in acetic acid.
+ NH3 +
1
\\
,CH
+ 2Hz0
THE ACTION OF AMMONIA IN ACETIC ACID ON ESTERS OF BENZOIN
To test the idea that it is the carbonyl group in benzoin which is attacked by ammonia rather than the carbinol group, the latter function was esterified, and the resulting desyl esters were submitted to the action of ammonia in acetic acid. The conversion of N-desylamides to glyoxalines illustrated above suggested that the closely related desyl esters (IX) would yield the corresponding oxazoles (X) according to the equation:
CeH5CHOCOR
+ "s+ CsHbCO IX
1
CeH5C-O
\
//CR
+ 2H20.
CeHbC-N X
Tests of desyl acetate and benzoate demonstrate that oxazoles are formed in yields of better than 80 per cent., accompanied by the related glyoxalines. The oxazoles and glyoxalines may have a common intermediate (XI) which either undergoes cyclodehydration or suffers acyl migration from oxygen t o nitrogen. In the second instance an enol form (XII) of an
332
D. DAVIDSON, M. WEISS, AND M. JELLING
N-desylamide (XIII) would be formed; the latter could give rise to the glyoxaline.
X
XI
XI11 EXPERIMENTAL
The action of ammonia on benzoin i n acetic acid.-Example: A mixture of 1.06 g. of benzoin, 10 g. of ammonium acetate, and 25 cc. of glacial acetic acid was refluxed for one hour. The solution turned deep orange in color almost a t once, and the precipitation of amarone began within ten minutes. The cooled mixture was filtered, and the yellow precipitate was washed with small amounts of glacial acetic acid until the washings were practically colorless. This precipitate weighed 0.35 g. (36% theoretical) and melted a t 251" (corr.). It gave in concentrated sulfuric acid the orange-red color characteristic of amarone. The combined liquors were treated with concentrated nitric acid until the orange-red solution changed t o yellow, and a crystalline precipitate of amarone appeared; m.p. 246'; yield 0.20 g. (21% theoretical). This was separated, washed with acetic acid and then liberally with water. When recrystallized twice from glacial acetic acid, the amarone was obtained as large, colorless needles, m.p. 252' (corr.). Anal. Calc'd for C28H20N2: N, 7.3. Found: N, 7.1. A total of 100 cc. of water was added t o the mother liquor. This threw out a little oily matter which was removed by extracting twice with 25 cc. portions of ben2ene.t The final aqueous layer was clarified by filtering through a wet, folded filter, and then heated nearly t o boiling and rendered alkaline with concentrated ammonia, producing a precipitate of colorless needles of 2-methyl-4,5-diphenylglyoxaline melting at 237" (corr.); yield, 0.28 g. (24% theoretical). After recrystallization from a mixture of pyridine and water i t melted a t 240"; its melting point was not depressed by an admixture of the glyoxaline prepared from benzil and ~ a r a l d e h y d e . ~ ~ The action of ammonia on benzoin i n propionic acid.-To a solution of 5 g. of ammonium carbonate in 25 cc. of propionic acid was added 1.06 g. of benzoin. The mixture was refluxed for one hour and worked up as in the previous experiment. The precipitates of amarone weighed 0.28 g. (29% theoretical) and 0.20 g. (21% theoretical) and melted at 250" and 249" respectively. The glyoxaline precipitate weighed 0.16 g. (13% theoretical) and melted at 210" (corr.). After recrystallization from pyridine and water, i t melted a t 215" (corr.). Ana2. Calc'd for ClrHlsN~: S , 11.3. Found: N, 11.4. The action of ammonia on desylamine i n acetic acid.-A mixture of 1.24 g. of desylamine hydrochloride,@lOg. of ammonium acetate, and 25 cc. of glacial acetic acid was treated as in the case of benzoin (see above). The amarone precipitates weighed 0.39 g. (41% theoretical) and 0.16 g. (17% theoretical), and melted at 252" and 247", respectively. The glyoxaline fraction weighed 0.25 g. (21% theoretical) and melted at 237" (corr.).
t This
oily by-product may contain 2-methyl-4,5-diphenyloxazole. PSCHORR AND BRUGGBMANN, Ber., 36, 2740 (1902).
ACTION OF AMMONIA ON BENZOIN
333
T h e action of ammonia o n benzoin in the prcsence of formic acid.-To 10 cc. of formic acid (85-90%) was added 7.5 cc. of acetic anhydride, the heat of the reaction being moderated by means of a cold-water bath. Three grams of ammonium carbonate was then added t o the resulting mixture of anhydrous acetic and formic acids. After dissolution was complete, 1.06 g. of benzoin was added, and the mixture was refluxed for three hours. The addition of 100 cc. of water precipitated an oil which was removed by two extractions with 25 cc. portions of benzene. The aqueous layer was rendered ammoniacal, whereupon 0.40 g. (36% theoretical) of 4,5-diphenylglyoxaline, m.p. 232' (corr.), separated. Its melting point was unchanged by recrystallization from dilute pyridine or by mixing with a sample of the glyoxaline prepared from b e n d and he~amethylenetetramine.~A mixture with 2-methyl-4,5-diphenylglyoxaline, however, melted a t 205" (corr.). Evaporation of the benzene extract left 0.72 g. of a yellow oil from which there was obtained by three extractions with 100 cc. portions of boiling water, 0.34 g. (28% theoretical) of colorless crystals, m.p. 116-118". Recrystallization from a mixture of anhydrous ethyl acetate and ligroin yielded large, clear, hexagonal plates, melting at 122". This product was identified as N-desylformamide by conversion t o 4,5-diphenylglyoxaline as described below. A n a l . Calc'd for Cl~H13NO~ : N, 5.9. Found: N, 5.5. The residual oil was not identified, but may have contained 4,5-diphenyloxazole since it possessed the characteristic pepper-like odor of the oxazoles. Preparation of N-desy1jormamide.-A m x t u r e of 1.24 g. of desylamine hydro~ h l o r i d e ,1.0 ~ g. of sodium acetate, and the solution resulting from the addition of 7.5 cc. of acetic anhydride to 10 cc. of formic acid (85-90%) was refluxed for three hours. The addition of 100 cc. of water precipitated an oil which soon solidified. Yield, 0.91 g. (75% theoretical), melting at 120". On crystallization from a mixture of anhydrous ethyl acetate and ligroin i t was obtained in large, clear, hexagonal plates, melting a t 122". It is easily soluble in the common organic solvents except light petroleum fractions, and slightly soluble in boiling water. Like other N-desylamides its alcoholic solution turns yellow on the addition of 10% sodium hydroxide. Its conversion t o 4,5-diphenylglyoxaline is described below. Conoersion of Pi-desylformamide to 4,6-diphenylglyozaline.-A solution of 0.24 g. of N-desylformamide (see just above) and 1.5 g. of ammonium acetate in 10 cc. of acetic acid was refluxed for one hour. Dilution with 100 cc. of water yielded a light precipitate (0.007 9.) which appeared t o be crude amarone. From the mother liquor, ammonia precipitated glistening, white crystals, weighing 0.21 g. (95% theoretical) and melting at 231" (corr.) (mixture melting point). When the experiment was repeated with the N-desylformamide obtained from t h e action of ammonia on benzoin in the presence of formic acid (see above), 0.19 g. (86% theoretical) of 4,5-diphenylglyoxaline, m.p. 228" (corr.) was obtained. T h e action of ammonia o n desyl esters.--Co~aversion of desylbenzoate (benzoylbenzoin) to triphenyloxaaole (benziZam).-A mixture of 1.58 g. of benzoylbenzoin,lD 3.0 g. of ammonium acetate, and 25 cc. of acetic acid was refluxed for one hour and then diluted with 20 cc. of water, whereupon benzilam separated as a crystalline precipitate, which was collected by filtration and washed with 60% acetic acid (9 cc.); yield, 1.38 g. (93% theoretical); m.p. 116" (mixture melting point). A n a l . Calc'd for C21HISN0: N, 4.7. Found: N, 4.8. The filtrate was diluted with 50 cc. of water and clarified by filtering through a 1 0 ZININ, Ann., 104, 116 (1857). In the present study, the preparation of benzoyl benzoin was improved by employing pyridine as a solvent.
334
D. DAVIDSON, M. WEISS, AND M. JELLING
wet folded filter. Ammonia precipitated 0.05 g. (3% theoretical) of crude lophine, melting at about 250", which when crystallized from a pyridine-water mixture melted at 275" (corr.). Conversion of desylacelate (acetylbenzoin) to 2-rnethyl-/,b-diphenyloxazole.-A mixture of 6.35 g. of acetylbenzoin," 10.0 g. of ammonium acetate, and 25 cc. of acetic acid was refluxed for one hour. The addition of 100 cc. of water threw out an oil which was extracted with three 25-cc. portions of benzene. The benzene extract was filtered through a dry filter, the benzene was removed, and the residue was distilled in vacuum; yield, 4.8 g. (82% theoretical) of a viscous oil, b.p. 210-213" at 18 mm. (Japp and Murray6 reported the .b.p. of 2-methyl-4,5-diphenyloxazoleas 214" at 17 mm.) Anal. Calc'd for CMHISNO: N, 6.0. Found: N, 5.9. From the aqueous layer, ammonia precipitated 0.74 g. (13% theoretical) of 2-methyl-4,5-diphenylglyoxaline; m.p. 243" (corr.). SUMMARY
1. The action of ammonia on benzoin in boiling acetic acid yields amarone, dihydroamarone, and 2-methyl-4,5-diphenylglyoxaline. 2. Desylamine is assumed to be an intermediate product in this reaction. It behaves like benzoin toward ammonia in acetic acid. 3. In the presence of formic acid the formation of pyrazines is suppressed, the isolated products being N-desylformamide and 4,5-diphenyIglyoxaline. 4. In propionic acid, the behavior of benzoin toward ammonia is similar to that in acetic acid, amarone, dihydroamarone, and 2-ethyl-4,5-diphenylglyoxaline being obtained. 5. Esters of benzoin react with ammonia in acetic acid to form mainly 2-alkyl-4,5-diphenyloxazoles accompanied by subordinate amounts of the corresponding glyoxalines. 11
COBSONAND SALIANI, Organic Synthese's, 12, 1 (1932).
