4004
RICHARD N . MCDONALD A N D PETER A. SCHWAB [CONTRIBUTION FROM
THE
Molecular Rearrangements. BY
VOl. 85
DEPARTMENT O F CHEMISTRY, KANSAS STATE UNIVERSITY, MASHATTAN, KAN.]
11.
Chlorine Migration in the Epoxide-Carbonyl Rearrangement
RICHARD N. MCDONALD AND PETER A.
~CHWAB~
RECEIVEDJ U L Y 26, 1963 Peracid epoxidation of trans-a-chlorostilbene has not yielded the a-chloroepoxide, but rather the a-chloroketone, desyl chloride This result is explained by rearrangement of the intermediate a-chloroepoxide. Preferential chlorine migration is shown to occur by peracid epoxidations of trans-l-chloro-l-(p-tolyl)-2-phen~lethylene (Va) and trans-l-chloro-l-phenyl-2-(p-tolyl)-ethylene (1.b) The structures of the a-chloroketone products were determined by their ultraviolet absorption spectra and independent synthesis Various mechanisms are considered for this novel chlorine mibration in light of the results I t was also found t h a t basic hydrolysis of desyl chloride to benzoin and reconversion to desyl chloride with thionyl chloride proceeds with no isomerization of the functional groups
The literature contains a few reports of a-chlorocyclic a-chloroepoxides, but made little mention of the epoxides and no reports of a-haloepoxides with the meaning of the results. other halogens. The most frequently used and simplest trans- a-Chlorostilbene oxide7 appeared to suit our method of preparation is the perbenzoic acid oxidation needs to initiate a study of substitution and eliminaof vinyl chloride^.^ More recently, Walling and Fredtion reactions of a-haloepoxides. We attempted the e r i c k ~were ~ able to chlorinate photolytically ethylene Darzens reaction of benzaldehyde and benzal chloride and propylene oxides using t-butyl hypochlorite, and under various conditions used normally for this conchlorine has been reported to effect the same p r o ~ e s s . ~ densation with no success, the majority of the starting The Darzens condensation of methyl dichloroacetate materials being recovered in each case. and benzaldehyde has been reported to yield methyl 2Our attention then turned to peracid oxidation of chloro-2,3-epoxy-3-phenylpropanoate, while benzaldetrans-a-chlorostilbene (I) as a method of obtaining the hyde and benzal chloride are reported t o yield under a-chloroepoxide. However, the isolation of this comDarzens conditions a-chlorostilbene oxide.7 pound was not realized under any of the conditions The chemistry of these compounds has been studied used, but rather the only isolable product was the aeven less. Hydrolysis leading to a-hydroxy ketones is chloroketone, desyl chloride (11). This observation has mentioned in several of the above references. Prilesled to the discovery of preferential chlorine migration chajew3' reported the Tollens and permanganate oxidain the epoxide-carbonyl rearrangement, the details of tions of 1-chloro-1-heptene oxide and 2-chloro-2-octene which are the subject of this paper. oxide. However, these conditions probably involved Treatment of trans-a-chlorostilbenelo(I) with triinitial hydrolysis to the a-hydroxy ketone which then fluoroperacetic acid in methylene chloride solution with underwent oxidation. Russian workerss have studied disodium hydrogen phosphate as the insoluble base" the reactions of aliphatic a-chloroepoxides with various gave a mixture of desyl chloride (11) and the trifluoro0
c1
0 CFaCOaH CHzCI?
11
C&-C-CH-CsHa I1
0 f
I1
OCCFa I1
I
C~H~-C-CH-C~HF,
I11 0
CFaCOsH
I
II
0 OH HzO
I1
I
+ C~HS-C-CH-C~H~ IV
CI
I
d CaHj-C-CH-C6H6
H
c
I
I1
CeHsCOgH or CHaCOsH
thioamides to produce substituted thiazoles. Mouss e r ~ nhas ~ ~reported a few acid-catalyzed rearrangements and hydrogenation experiments on some ali( I ) For paper I in this series see R . N . McDonald a n d P . A. Schwah, J . A m . Chem Soc., 8 6 , 820 (1963). a preliminary communication of some of these results. (2) A portion of a dissertation t o be presented by P. A. Schwab t o the Graduate School of Kansas State University in partial fulfillment of t h e requirements for the degree of Doctor of Philosophy. ( 3 ) (a) M Mausseron. F Winternitz, and R. Jacquier, C o m p l . rend., 888, 1014 (1946); (h) M Mousseron, F Winternitz, and K Jncquier, Bull. soc. chim. F r a n c e . 81 (1947); (c) M Mousseron, F Winternitz, and R . Jacquier, i b i d . , 260 (1948). (d) M . Mousseran and R. Jacquier, ibid.,698 (19dO); (e) N .Prileschajew, Ber., 69, 194 (1926); ( f ) A A. Durgaryan a n d S. A Titanyan. Izvesl. A k a d l V a u k A r m y a n S S . R . , K h i m . N a u k i . 13, 263 (1960), Chem. A b s l r . . 6 6 , 210846 (1961). (4) C Walling and P S. Fredericks. J A m , Chcnr. Soc., 84, 3326 (1962). i.5) H Gross and A Rieche, German Patent 1,084,707 (1960); Chcm. A b s l r . , 66, 235fiZc (19f:l). (6) V. F Martynov and M. I . Titov, Z h . Obshch. K h i m . . 38, 319 (1962). (7) F. Barrow, Inaugural Dissertation, Strasbourg, 1909; see M . S. Newman and B. J. Magerlein, "Darzens Glycidic Ester CondenTation," OrR. Reaclions. 6, 113 (1949), ref. 31b. ( 8 ) A A . Durparyan, Izvesl A k a d . .Vauk A r m y a n . S.S.R., K h i m . Nauki, 14, R1 (1961), Chem. A b s l r . , 6 6 , 46.5,'(1962); and A . A. Durgaryan, S. A. Titanyan, and R.A . Kazaryan, Izvesl. A k a d . N a u k A r m y a n . S . S . R . , Khim. l ' n u k i , 14, 165 (1961); Chem. A b s l r . . 6 6 , 4741d (1962).
acetate of benzoin12 (111). Upon standing exposed to the atmosphere for several days, or when hydrolyzed with water, this mixture was completely converted to benzoin (IV) (76% yield based on truns-a-chlorostilbene). Using sodium carbonate as the insoluble base," trifluoroperacetic acid treatment gave desyl chloride (11) in 63y0yield. Treatment of I with peracetic acid in methylene chloride solution gave a 67y0 yield of 11, while perbenzoic acid in chloroform solution converted I to I1 in 74Y0.yield. All of the yields thus far reported are of crystallized, sharp melting product. To check these yield data and to determine the fate of the rest of the starting material the reaction product from a perbenzoic acid oxidation was shown by ultraviolet spectral analysis to contain some I as well as 11. The a-chloroketone I1 was readily separated from achloroolefin I by column chromatography or careful (9) Previous work on epoxide-carbonyl rearrangements has been reviewed by R. E. Parker and N . S. Isaacs, Chem. R e v . , 69, 737 (l9S9). (10) T W. J. Taylor and A . R . Murray, J . Chem. Soc., 2078 (1938). (11) W . D. Emmons and A. S. Pagano, J . A m . Chcm. Soc., 7 1 , 89 (1955). (12) Gas chromatographic and elemental analysis showed this mixture t o consist of 24% of desyl chloride and 76% of t h e trifluoroacetate of benzoin.
Dec. 20, 1963
CHLORINE MIGRATION I N EPOXIDE-CARBONYL REARRANGEMENT
crystallization. The yield of I1 in this manner was 75y0 while compound I was recovered in 24.5%. This shows that the conversion to I1 is quantitative within experimental error. The formation of the a-chloroketone, desyl chloride, can readily be explained in terms of the well known epoxide-arbonyl rearrangementg assuming the intermediacy of the a-chloroepoxide. This was considered reasonable in view of the nature of the reactants and the previous success in the isolation of a-chloroepoxides under similar conditions. The question remaining is which atom migrated, hydrogen or chlorine. Previous results on this and "related" molecular rearrangement^'^ certainly show that hydrogen is frequently found to migrate. On the other hand, the migration of the acetoxy group in the epoxide-carbonyl rearrangement of enol ester epoxides has been d e m ~ n s t r a t e d , which '~ lends support to the idea that chlorine might be the migrating group. l 5 T o decide this question of which atom was migrating we chose to study the peracid oxidations of trans-lchloro-l-(p-tolyl)-2-phenylethylene (Va) and trans-lchloro-1-phenyl-2- (p-tolyl)ethylene (Vb). With these two isomers we could distinguish between hydrogen and chlorine migration without significantly altering the path of rearrangement. 4-AIethyl-a-phenylacetophenone16 was prepared by the Friedel-Crafts reaction of phenylacetyl chloride with toluene and subsequently treated with phosphorus pentachloride in refluxing benzene to give the desired trans- 1-chloro-1-(p-tolyl)-2-phenylethylene (Va). When this vinyl chloride was treated with trifluoroperacetic acid in methylene chloride solution, using no insoluble base, a trifluoroacetate (VIb) was isolated and immediately hydrolyzed without purification to give a single product in 45y0 yield which was found to be methyl - a-hydroxy- a-phenylacetophenonel 7 (VIc) . Using sodium carbonate as the insoluble base, trifluoroperacetic acid gave a 58% yield of 4-methyl-achloro-a-phenylacetophenone (VIa).
Va, R = CHI, R ' = H b, R = H , R ' CH3
VI a , R = CHI, R ' = H, X H, X b, R = CH3, R' C, R CHI, R ' H, X H , R' = CHI, X d, R e , R = H , R ' = CH3, X H, R ' = CH3, X f, R
= = = = = =
C1 OCOCFi OH C1 OCOCFB OH
Also, treatment of Va with peracetic acid in methylene chloride solution gave a 71% yield of the rearranged product VIa, and perbenzoic acid in chloroform solution give a 73% yield of VIa. Since the yield of achloroketone in this latter oxidation-rearrangement is comparable to the yield of desyl chloride using this same peracid, this demonstrates that only chlorine migration has occurred. The structure of the chloroketone VIa was established by independent synthesis by sulfuryl (13) c. J. Collins, Q U ~ Y Rev. ~ . (London), 14, 357 (i960), reviews the pinacol and certain related rearrangements. (14) A. L. Draper, W . J. Heilman, W. E. Schaefer, H. J. Shine, and J. N. Shoolery, J . O I K . C h e m . , 27, 2727 (19621, and references cited therein. (1.5) The epoxide-carbonyl rearrangement is particularly well suited t o study the migration of various groups, e x . , the halogens and pseudo-halogens, due t o the relative ease of synthesis of the necessary a-substituted epoxides. Such studies are presently underway in our laboratories. (16) W. Mann, B e y . , 14, 1645 (1881). (17) R T. Arnold and R C . Fuson, J . A n . Chem. S O L . ,88, 1295 (1936).
4005
N
W
wavelength (rnfl). Figure 1
chloride chlorination of 4-methyl-a-phenylacetophenone. trans- 1-Chloro- 1-phenyl-2- (p-tolyl)-ethylene (Vb) was prepared in the following manner. Photolytic bromination of p-xylene was carried out in carbon tetrachloride giving a-bromo-p-xylene. This was then converted to a-cyano-p-xylene with sodium cyanide. l8 Hydrolysis of this nitrile gave p-tolylacetic acidla which was easily converted to its acid chloride with phosphorus trichloride. Friedel-Crafts reaction of this acid chloride with benzene gave the desired a-(p-tolyl)-acetophenone. l 9 Subsequent treatment with phosphorus pentachloride in benzene and distillation gave the expected trans-vinyl chloride Vb. As above, treatment of Vb with trifluoroperacetic acid, using no insoluble base, gave the trifluoroacetate VIe which was immediately hydrolyzed to the known a-hydroxy-a-(p-tolyl)-a~etophenone~~ (VIf) in 42y0 yield ; using sodium carbonate, a-chloro-a-(p-tolyl)acetophenone (VId) was formed in 53% yield. Reaction of Vb with peracetic acid and perbenzoic acid gave the same a-chloroketone VId in yields of 67 and 71%! respectively. Sulfuryl chloride chlorination of a-(ptolyl)-acetophenone likewise gave VId. As evident from the above results, in every instance only that a-chloroketone was isolated which necessarily involves chlom'ne migration. Also, no acid chlorides or aldehydes were found which would be the result of aryl or hydrogen migration. These facts together with quantitative conversions from the perbenzoic acid reaction (assumed also in the other Deracid reactions) demonstrate the high migratory abiiity of chlorine in these systems. The assignment of the trans configuration to Va and Vb is based On the Of their spectra with that Of known trans-a-chlorostilbene. These are shown in Fig, 1, The ultraviolet spectrum of cis-Va (18) E. F. J . Atkinson and J . F. Thorpe, J . C h e m . Soc., 1687 (1907). (19) H. Strassmann, B e r . , 29, 1229 (1889).
4006
RICHARD N. MCDONALD AND PETER A. SCHWAB
is also given.z0 The ultraviolet spectra of the product a-chloroketones are also in agreement with their assigned structures. 4-Methyl-a-chloro-a-phenylacetophenone (VIa) has its longer wave length absorption peak a t 258 mp (E 26,200) illustrating the p-tolyl group conjugated with the carbonyl, while its isomer a-chloroa-(p-tolyl)-acetophenone exhibits its longer wave length peak a t 247 mp ( E 24,500), very similar to that of desyl chloride. We believe certain of our results are useful in attempting to describe the mechanism by which chlorine migrates in these examples. To simplify the mechanistic possibilities we will make the assumption that this rearrangement is intramolecular. Two pieces of evidence can be cited to support this assumption: (1) acetoxy (a “pseudo-halogen”) has been shown to migrate intramolecularly in the thermal rearrangement of enol ester epoxides“ and gives stereospecific a-acetoxy ketone productz1and ( 2 ) our reactions with peracetic acid in acetic acid-methylene chloride solution containing added sodium acetate gave high yields of a-chloroketone and no evidence of benzoin acetates as products. With this assumption there appear to be three reasonable mechanistic pathways which will account for chlorine migration.2z The first that we might consider would be opening of the epoxide ring to give a carbonium ion followed by chlorine migration to yield the observed a-chloroketone product. The 1,2-shift of chlorine must be fast or we would have expected the carbonium ion
VI1
to be captured by acetate ion in the peracetic acid reactions. To maintain the intramolecular nature of the migration step, we may consider two methods of shifting the chlorine: (1) involvement of a chloronium ion, e.g., VIII, following a rotation about the central carboncarbon bond or ( 2 ) formation of an intimate ion pair, e.g., I X , followed by bond formation leading to product. In either case there would appear to be little driving force for return to the carbonium ion VII. A second mechanism that deserves mention would be a concerted process. However, this must involve considerably more C-0 bond breaking than C-Cl bond making in the transition state, which may be crudely represented as X. Such a structure would involve a high degree of strain energy and would infer that chlo-
VI11 IX X rine migration is facilitatingthe C-0 bond fission. This would then be similar to chlorine acting as a participating neighboring group in a modified solvolysis.z3 Since participation by neighboring chlorine has been shown to (20) The c i s and 17am isomers of Va were found t o be readily separated by or b y spinning-band distillation of the reaction product from 4-methyla-phenylacetophenoneand phosphorus pentachloride. (21) K . L. Williamson and W. S . Johnson, J . Org. C h e m . , 26, 4563 (1961). (22) We will not attempt t o differentiatebetween rearrangement of the achloroepoxide or its conjugate acid. (23) The term “modified solvolysis” is employed in that the departing anion is part of the same molecule. V.P.C.
Vol. 85
be very small in solvolyses24 and the strain energy necessary to achieve X would be great, this mechanism can be discarded a t present. Another point regarding the use of the “concerted” mechanism in epoxide-carbonyl rearrangements is that there must be a rotation about the C-C single bond of the epoxide of approximately 7 6 O Z 5 to bring the migrating group into a transoid position to the breaking C-0 bond. In order to allow for this there must be considerable, if not complete, rupture of the epoxide ring and this pathway becomes the same as mechanism 1. This point does not appear to have been considered before. The third mechanism worthy of consideration involves ion pair formation prior to epoxide ring opening, e.g., X I . Here the strain problems inherent in the
XI
concerted mechanism are not involved and the group that will migrate will be that which forms anion most readily. This seems to be the only mechanism of the three considered that accounts for the preferential chlorine migration and lack of competition by hydrogen or aryl. Whether internal return of X I to the achloroepoxide (or its conjugate acid) takes place is unknown as is the extent, if any, of epoxide C-0 bond fission in this process. The breaking of the particular epoxide C-0 bond in the discussion of the above listed mechanisms is explained by the following contributing structures to the possible hybrid of the a-chloroepoxidez6 (or its conjugate acid). Ion XI1 would be of greater stability than
H’;;c-c
