《有机化学》课程教学资源（文献资料）The Thermal, Aliphatic Claisen Rearrangement.pdf_大学文库

Chemical Reviews Volume 88. Number 8 December 1988 The Thermal, Aliphatic Claisen Rearrangement FREDERICK E. ZIEGLER steriing Ctmm!.qtry Labaatory. Yale UnivennV, New Haven. Connecticur 08511-8118 Receh.ed April 25. 1988 (Revised Manusnipt Received Seplember 8. 1988) Contents I. Introduction 11. Historical Overview 111. Mechanistic Aspects A. Kinetics B. Retro-Clalsen Rearrangement C. Competitive Rearrangements 1. [3.3] Claisen vs [2,3] Wmig 2. Diinylcarbinol Derivatives 3. Elimination D. Stereochemistry 1. Transition State 2. Vinyl Double-Bond Geometry 3. Secondary Allylic Alcohols 4. Tertiary Allylic Alcohols 5. RingBearing Substrates Rearrangement IV. Heteroatom Substituents A. The Vinyl Group 1. C, Hetero Substituents 2. C, Carbon Hetero Substituents 1. Oxygen Substituents 2. Silicon Substituents B. The Allylic Group V. Remote Asymmetry A. Acyclic Substrates B. Ring-Bearing Substrates VI. Consecutive Rearrangements A. Sequential Rearrangements 8. Tandem Rearrangements C. Iterative Rearrangements A. (1,llRearrangements 8. (1.2JRearrangements C. (1.41-Rearrangements D. (1.5lRearrangements E. (1.61-Rearrangements F. (2.41-Rearrangements G. (4.5l-Rearrangements H. (4.61-Rearrangements I. (5.61-Rearrangements VII. Synthetic Applications VIII. Biochemical Aspects IX. Concluding Remarks 1423 1424 1425 1425 1427 1427 1427 1428 1429 1429 1429 1429 1431 1433 1433 1435 1435 1435 1435 1436 1436 1437 1438 1438 1438 1440 1440 1440 1441 1442 1442 1442 1443 1444 1444 1444 1445 1446 1446 1448 1448 A. Frederick E. Ziegier received his B.S. from Fairiegh Dickinson University in 1960 and Ph.D. in 1964 from Columbia University where he studied under Gilbert Stork. As an NSF postdoctoral student. he spent 1 year in the laboratory of George Buchi at The Massachusetts Institute of Technology. He joined the Yale University facuHy in 1965 where he currently hokk the rank of Professor of Chemistry. His research interests include the synthesis of physiologically active natural products, the study of the stereochemistry of aganic reactions. and the development of new synthetic methods. I. Introduction This past year the diamond annivenary of the publication of Ludwig Claisen’s paper “Uber Umlagerung von Phenol-allyl-athern in C-allyl-phenole^ describing his now eponymous rearrangement’ was observed. And what a gem it has proved to he! Ironically, the majority of the text of the paper and all the experimental details dealt with the substance of the title while the first paragraph mentioned, in almost parenthetical fashion, the rearrangement of the 0-allylation product of acetoacetic ester 1 to its C-allylated isomer 2 upon distillation in the presence of ammonium chloride. Arguably, the aliphatic rearrangement has stimulated more interest in both its mechanistic and synthetic aspects than its aromatic counterpart. Today, the aliphatic Claisen rearrangement is but one member of the class of [3,3] sigmatropic rearrangements. The prototype for the rearrangement is the transformation of allyl vinyl ether 3 into 4-pentenal (4). 0009-2665/88/078&1423$06.50/0 0 1988 American Chemical Society

1424 Chemical Reviews, 1988, Vol. 88, No. 8 Ziegler dehydrohalogenation to form the archetypical allyl vinyl ether (3), which underwent successful rearrangement to aldehyde 4 at 255 “C. In addition, allyl isopropenyl ether (9) was prepared by acid-catalyzed elimination and was subjected to rearrangement to afford ketone 10. H‘ xo4 0- 8 - 2550c L 1 2 This review will deal with the history, mechanism, stereochemistry, and applications of the thermal, aliphatic rearrangement2 over the past 75 years, as recent publications have provided excellent summaries of the effect of catalysts on the rea~angement.~ While the contributions to this area are legion, an effort will be made to deal with both historical contributions and those reports that exemplify the scope of the reaction. Although heteroatom Claisen rearrangements will not be covered, examples will be provided as they apply to the discussions at hand. I I. Hlstorlcal Overvlew Bergmann and Corte (1935)4 and Lauer and Kilburn ( 1937)5 investigated the rearrangement of ethyl 0-cinnamyloxycrotonate (5) in the presence of ammonium chloride to determine if “transposition” of the allyl unit occurs, as had been established in the aromatic series? The former collaborators reported the formation of the “nontransposed” product 6 and “transposed” 7 while the latter investigators observed only the product of “transposition”. The formation of P-keto ester 7 provided access to a product formally derived from the s$’ C-alkylation of cinnamyl halides with acetoacetic ester anion. PhT 5 P.31 oTph C0,Et 7 Bergmann and Corte employed Claisen’s method7 of ammonium chloride catalyzed exchange of cinnamyl alcohol with ethyl 3-ethoxy-2-crotonate for the formation of 5 while Lauer and Kilburn used sodium cinnamylate and ethyl P-chlorocrotonate. The use of ammonium chloride in the rearrangement step soon disappeared, although it had been shown to have “a small, but significant, increase in rate” as a heterogeneous catalyst.8 While the @-keto esters provided access to y,b-unsaturated acids by Haller-Bauer cleavage (Le., retroClaisen condensation) and y,d-unsaturated ketones by acid hydrolysis, formation of y,d-unsaturated aldehydes had not been realized. In 1938, Hurd and Pollackga subjected @-bromoethyl allyl ether to base-promoted 9 10 In the early 1940s, Carrolllo investigated the basecatalyzed reaction of acetoacetic ester with allylic alcohols to produce olefinic ketones.ll In particular, the stereospecificity of the reaction was demonstrated in the case of the structural isomers cinnamyl alcohol (1 1) and phenylvinylcarbinol (13), each giving a transposed product. Aware of the results of Bergmann and Lauer, Carroll proposed a mechanism that invoked sN2’ displacement of hydroxide by acetoacetate anion. Kimel and Cope12 (1943) clarified the mechanism by demonstrating that acetoacetic acid esters derived from allylic alcohols undergo the rearrangement. Moreover, the use of diketene provided a reactive equivalent of acetoacetate that made the formation of substrates routine. Thus, this variation of the reaction provided nonacidic conditions, compared to those of Hurd, for the generation of y,d-butenyl methyl ketones. )!.& + EtOH + CO, CH3COCH2C02EWNaOE1 Ph *OH 12 or diketene; NaOEt 11 LPh + EtOH + CO, CH3COCH,C02EWNaOEt Ph or diketene; NaOEt 13 14 The generation of vinyl ethers by the dehydrohalogenation procedure of Hurd did not provide a general route for the derivatization of allylic alcohols. A solution to this problem was provided by Burgstahler and Nordin13 who adapted the mercuric acetate catalyzed exchange of alcohols with alkyl vinyl ethers14 to the formation of allyl vinyl ethers (15 - 16).15 These investigators were able to demonstrate that the rearrangement was successful in systems wherein at least one of the double bonds is contained in a ring (16 - 17). ROCH=CH, - Hg(OAc)z P A q? CHO 15 16 17 In 1967, Marbet and Saucy reported16 the acid-catalyzed exchange and rearrangement of seemingly labile tertiary allylic alcohols with 2,2-dimethoxypropane or 2-methoxypropene that resulted in the formation of methyl ketones.17 The conversion of linalool to gera-

Thermal, Aliphatic Claisen Rearrangement nylacetone laid the groundwork for a commercial synthesis of vitamin A alcohol. The first practical example of the preparation of y,b-unsaturated carboxylic acids via the aliphatic Claisen rearrangement was demonstrated by Arnold and co-workers18 in 1949.19 The allylic esters 18 and 20 of diphenylacetic acid underwent stereospecific rearrangement upon treatment with mesitylmagnesium bromide at ambient temperature. diradical Chemical Reviews, 1988, Vol. 88, No. 8 1425 Ph 20 21 Other variations employed sodium hydride as the base in refluxing toluene;18a9z0 however, a significant breakthrough was reported by Irelandz1 who, following Rathke's reportzz of the use of lithium dialkylamide bases for the generation of ester enolates, demonstrated that the method served as a means to achieve the Claisen rearrangement of acylated allylic alcohols at ambient temperature (22 - 23) and, as will be seen later, provided a method for the control of enolate geometry. Both the enolates and their 0-silyl ketene acetals underwent facile rearrangement.z3 Y l)L'cA'THF 2) TMSCI e 0 7 H30+ * 7 0 0 22 23 24 "'rf OTMS Although Ireland's contribution improved the formation of y,b-unsaturated acids, it was preceded by two independent contributions that realized amides and esters via the Claisen rearrangement. In 1964 Eschenm~ser~~~p~ adapted Meerwein's observationsz4c on the exchange of amide acetals with allylic alcohols, thereby facilitating the formation of y,b-unsaturated amides upon rearrangement (25 - 26). In a similar fashion, 1970 witnessed Johnson's reportz5 of the acid-catalyzed exchange of ethyl orthoacetate with allylic alcohols and the subsequent formation of y,b-unsaturated esters upon rearrangement (27 - 28). 0 II MezNC C02Me * MeC(OMe)?NMeZ, xylene 1 4OoC, 14h ''O I Me' OH 25 26 MeC(OEt), propionic acid m + OH OH 138'C, 3h 27 C02Et Et02C 28 0% - 100% diyl 0 Figure 1. Transition-state profile of the aliphatic Claisen rearrangement. ZZZ. Mechanistic Aspects A. Kinetics The ability of the Claisen rearrangement to give transposed structures led Hurd and Pollackgb to suggest a cyclic mechanism. The rearrangement of allyl vinyl ethers displays a negative entropyz6 and volumez7 of activation, both of which support a constrained transition state relative to ground-state geometries. Firstorder kinetics8Vz6 and the lack of crossover products8 argue for the intramolecularity of the reaction. The overall exothermicityz6 of the rearrangement of allyl vinyl ethers indicates an early transition ~tate.~J~ Using secondary deuterium isotope effects as a mechanistic probe, Gajewskiz9 has concluded that bondbreaking is more advanced than bond-making in the rearrangement of allyl vinyl ether itself. Thus, the transition state (TS) has been suggested to resemble more closely the diradical than the 1,4-diyl. Figure 1 (More O'Ferrall-Jencks diagram) locates the transition state for allyl vinyl ether above diagonal A (diradical > diyl) and below diagonal B (early, not late, TS). Dewar, using MIND0/3 calculations, has supported an early transition state for allyl vinyl ether with bondmaking being more advanced than bond-breaking, thereby requiring diyl character in the transition state.3o Substituents play an important role in affecting the rate of the Claisen rearrangement. Burrows and Car- enter,^^ using phenyl anion as a transition-state model, have predicted that ?r-donor substituents at C1, Cz, and C4 of allyl vinyl ether should increase the rate of the rearrangement, while substitution at C5 and c6 should cause deceleration. However, Dewar30 has argued that a C5-methoxy substituent should have a greater accelerating effect than a C,-methoxy group. The presence of electron-donating groups, e.g., EtO-, R3SiO-, and Me2N-, at Cz of the allyl vinyl ether causes a dramatic rate acceleration. Thus, the 2-(trimethylsily1)oxy (29; tllz = 210 f 30 min at 32 0C)21 and the 2-(tert-butyldimethylsilyl)oxy (31a; tljz = 107 min at 35 0C)3z derivatives rearrange with facility under near ambient conditions, while allyl vinyl ether (tllz = 1.7 X lo4 min at 80 0C)33 requires higher temperatures for rapid rearrangement. While both 29 and its 6-methyl congener 30a (tl,z = 150 f 30 min at 32 oC)zl rearrange

1426 Chemical Reviews, 1988, Vol. 88, No. 8 Ziegler kinetic parameters. Rate accelerations are observed for the C2-CN (krel = ill), C3-CN (krel = 270), and C4-CN (krel = 15.6) compounds while decelerations occur for the C1-CN (krel = 0.90) and C5-CN (krel = 0.11) isomers relative to allyl vinyl ether. The formation of anionic species increases the rate of the rearrangement. The enolates of allyl esters should be considered as the prototypes of strong C2 ?r-donors as they rearrange at ambient temperatures.21 Denmark has reportedm the first example of a carbanion-accelerated Claisen rearrangemen~~l The use of hexamethylphosphoramide (HMPA), as opposed to l&crown-G/THF, accomplishes the conversion of 38a - 39a at a lower temperature (50 "C) and in a higher yield (78%). This solvent-induced rate enhancement has been interpreted as ion-pair dissociation. Disubstitution at C1 (38b - 39b) causes a greater rate enhancement (20 "C, 15 min),42 similar to the silyl ketene acetal case. Phsoq] HMPA * / R R 38 PhSO, at nearly the same rate, the C4-alkyl-substituted isomer 3 1 b rearranges an order of magnitude more rapidly than This difference suggests a kinetic stabilization of the bond-breaking process by the alkyl group in 31b. R- I OY GTMS GTMS OTBS 29 30a, R,=R,=H 31a. R=H b, R,=Me, R,=H b, R=C,Hll c, R,=R,=Me OMe O4 O4 32 33 O4 34 Coates and C~rran~~ have measured the rates of rearrangement of the C4-, C5-, and C6-methoxy-substituted allyl vinyl ethers at 80 "C in benzene. The 4- methoxy derivative 32 rearranges 100 times faster than the parent while the 6-methoxy isomer 34 is 10 times faster, thereby demonstrating a strong kinetic stabilization in the former case and a vinylogous, kinetic anomeric effect in the latter.35 The observation of this effect is contrary to the Burrows-Carpenter model. In addition, the 5-methoxy isomer 33 is found to rearrange 40 times slower than the parent, in disagreement with the Dewar prediction. Coates and Curran have suggested a transition state for these systems with dipolar character (enolate-oxonium ion pair). When the solvent is changed from benzene to methanol, 32 and 34 show a 20- and 70-fold rate increase, respectively. In general, solvents have little effect on the rate of the rearrangement. Gaje~ski~~ has attributed the rate enhancement and the greater degree of bond-breaking in the transition state of the Ireland-Claisen rearrangement to the greater stability of the 2-[ (trimethylsilyl)oxy]-1-oxaallyl moiety over its oxaallyl counterpart (Figure 1). This conclusion derives from an examination of the heats of formation of the oxaallyl radicals and supports21b the finding that the relative rates of rearrangement of silyl ketene acetals 30 are 30c > 30b > 30a.37 The effect of the (trimethylsily1)oxy group is not general to all [3,3] sigmatropic rearrangements as 2-[ (trimethylsily1)- oxy]-3-methyl-1,5-hexadiene undergoes a Cope rearrangement with a half-life of 2 h at 210 "C. Although the Johnson and the Eschenmoser variants are conducted at elevated temperature, these conditions are required for the alcohol exchange reaction, not necessarily for the rearrangement. This point is amply demonstrated in the latter instance when ketene OJVacetals are generated by an alternative route (35 + 36 - 37).38 0 35 36 37 In an independent study, Carpenter and Burrows39 have synthesized the five isomeric cyano-substituted derivatives of allyl vinyl ether and have measured their 39 a, R=H b, R=Me 40 The Carroll rearrangement is accelerated by carbanion formation. Wilson43 has demonstrated that /3-keto ester 41, formed by 4-(dimethylamino)pyridine (DMAP) catalyzed addition of (E)-2-buten-l-ol to diketene,44 provides the @-keto acid 43 when treated with 2 equiv of lithium diisopropylamide (LDA) at -78 "C in THF followed by heating to reflux. Decarboxylation is readily accomplished in refluxing carbon tetrachloride to give the ketone in 95% yield. When 1 equiv of base is used, no reaction is observed. The thermal reaction in the absence of base requires heating at 200 "C, and the ketone is isolated in only 37% yield. In a similar fashion, BUCK& has observed acceleration in the rearrangement of 3-(allyloxy)-2-butenoic acid 44 prepared by alkoxide addition to the 3-chloro-2-butenoate. When the acid is treated with 1 equiv of KH in refluxing toluene for 2-6 h, the potassium carboxylate is stable. However, the use of 2 equiv of KH effects rearrangement via dianion 45 under the same conditions, affording ketone 46 in 68% yield upon acidification and decarboxylation. The rate enhancement occurs for substrates derived from secondary allylic alcohols, but not for primary allylic alcohols. Silyl ketene acetals prepared from secondary alcohols have been observed to rearrange faster than those derived from primary allylic alcohols.21b a-Allyloxy ketones have displayed remarkable rate accelerations. For example, Koreeda and Lueng~~~~ have generated the enolate 48a by conjugate addition of Me,CuLi to 2-(allyloxy)-2-cyclohexenone (47); rearrangement to acyloin 49a is complete in 15 min at 0 oC.46b The rate enhancement has been attributed to an allyl radical/oxyoxaallyl radical anion (semi-dione) pair. For comparison, the silyl enol ether 48b is slower

Thermal, Aliphatic Claisen Rearrangement Chemical Reviews, 1988, Vol. 88, No. 8 1427 ether as its tetracyanoethylene derivative and to the left by formation of the bisulfite adduct of the aldehyde. Decomposition of the bisulfite adduct reestablishes the equilibrium. These equilibria are presumably driven by the strain of the cyclopropane ring.49p50 7 2 equiv. LDATHF * 00 41 -78 'C -> 65 OC F * +CO2H I wo THF 42 43 L- &CO~H 2 equiv. KH 44 r 1 I) toluene, 120 "C 2) H30i L J 45 46 to rearrange, having tl/P = 1.6 h at 62.5 "C. In a related study, P~naras~~ has compared the relative rates of rearrangement of 2-(allyloxy)-3-methyl-2- cyclohexenone and its derivatives. In refluxing THF (65 "C), the parent ketone 50a has tll2 = 340 h, affording the diosphenol51, while the rearrangement of carbomethoxyhydrazone 50b to 52a is appreciably faster (tl = 22 h). The sodium salt of the hydrazone (50c) is the fastest of the three, rearranging to give 52b with t,2 = 1.5 h. This method has proved amenable to forming vicinal quaternary centers, and in the case of the carbomethoxyhydrazones 52, a subsequent Wolff-Kishner reduction can be conducted to remove the accelerating f~nctionality.~~ 47 48a. R=MeCuLi 49a, R=H b, R=TMS b, R=TMS R OH RN %la, R=O 51 52a, R=NN(H)C02Me b, R=NN(H)CO,Me b, R=NN(Na)CO,Me c, R=NN(Na)CO,Me B. Retro-Claisen Rearrangement The Claisen rearrangement, unlike its all carbon analogue the Cope rearrangement, is an irreversible reaction, except for several specially designed substrates. Vinylcyclopropanecarboxaldehyde (53) has been shown to be in rapid equilibrium with dihydrooxepine (54). Similarly, unsaturated aldehyde 55 forms a 7:3 equilibrium mixture with vinyl ether 56. The equilibrium is shifted to the right by trapping the vinyl U'CHO - u- 53 54 g+= 73 40 55 56 Oppolzer51 has observed that silica gel chromatography of aldehyde ester 57a provides recovered substrate (68%) in addition to unsaturated ester 58. During the same period, Boe~kman~~ observed that the rearrangement of 57a to 58 occurs quantitatively at room temperature in 24 h. However, the less strained homologue 59a, upon heating in refluxing toluene in the presence of a catalytic amount of HOAc, provides an equilibrium mixture of 59a and 60 (89:ll). Support for a sigmatropic rearrangement rather than a pathway invoking stepwise formation of carbocation intermediates follows from the observation that the stereoisomer 59b does not undergo rearrangement under conditions that are successful with 59a. However, BF3-Et,0 at room temperature is able to convert stereoisomer 57b into 58, ostensibly through a carbocation intermediate that may only be required for the isomerization (57b - 57a) and not necessarily for the rearrangement. These investigators have also observed acceleration in the BF3.Et20-catalyzed rearrangement at -78 "C in alkyl-substituted congeners of 59a. These observations have led Boeckman to suggest that the minor product, unsaturated ester 62, formed in the BF,.EkO-catalyzed Diels-Alder reaction (-78 "C) between cyclopentadiene 61 and methyl 2-acetylacrylate, may well arise from the major product of the reaction, norbornene 63, by way of the catalyzed retro-Claisen rearrangement.53 57a, R,=CHO, Rz=C02Me b, R1=CO2Me, Rz=CHO C0,Me 58 59a, Rl=CHO, R2=C02Me 60 b, Rl=COzMe, RAHO 61 Br COMe 62 63 C. Competitive Rearrangements 1. [3,3] Claisen vs [2,3] Wiffb Rearrangement Conceptually, a-allyloxy enolates of the type 65a can undergo either [3,3] sigmatropic rearrangement (anionic

1428 Chemical Reviews, 1988, Vol. 88, No. 8 oxy-Claisen rearrangement)46 or [2,3] Wittig rearrangement.54 Thomas55 has observed that ketone 67, upon exposure to t-BuOKlt-BuOH, undergoes a [2,3] Wittig rearrangement to "mainly" keto1 68. In contrast, K~reeda~~* has reported that the enolate of phenyl ketone 69 (from MH and MeOH) in toluene not only shows rate enhancement (M = K, -23 "C, tllz = 3.3 h; M = Na, 0 "C, tljz = 2.6 h; M = Li, 96.5 "C, tljz = 1.1 h) but also gives a ratio of Claisen to Wittig product (70:71) of >98:<2 (M = K, Na) and a ratio of -80:20 when M = Li. To ensure exclusive formation of the Claisen product, enolates need only be 0-silylated (65a - 65b) and rearranged to aldehydes (64b). Accordingly, the 0-trimethylsilyl enol ether of ketone 69 affords the 0-trimethylsilyl derivative of ketone 70 upon heating (71 "C, tljz = 0.5 h). Earlier, Sa10mon~~ had demonstrated the utility of the transformation 65b - 64b by preparing the 0-trimethylsilyl enol ethers with (TMS)C1/Et3N. The a-silyloxy aldehydes could be cleaved readily with methanolic periodic acid to afford P,y-unsaturated ketones. 2 i e g I e r 64 65 66 a, X=M, R=alkyl c, X=M. R=O-alkyl e, X=TMS, R=NR,' f. X=TMS, R=O-alkyl b, X=TMS. R=alkyl d, X=M, R=OM OH I 67 6e 69 'C 1 Surprisingly, the ester enolates of generic structure 65c do not undergo a [2,3] Wittig rearrangements7 while their carboxylate-derived dianions (65d) and dialkylamide anions (65e) do rearrange by this pathway.58 Exclusive Claisen rearrangement of these substances can be accomplished via the trimethylsilyl ketene acetals (65f - 64f). This procedure has been described independently by Nakai57 and Ra~cher~~ in thetransformation of ester 72a to its (2)-0-silyl ketene acetal 73 followed by Claisen rearrangement to the masked a-keto aldehyde 74. Significantly, when the C-silylated ester 72b is treated with tetra-n-butylammonium fluoride, formation of the [2,3] Wittig product occurs. On the other hand, the 0-silyl derivative 73 gives the starting ester 72a. This observation has led Nakai to suggest that a common "naked" anion is not involved in the two pathways but that separate C- and O-hypervalent silicon species are responsible for the dual reaction pathways. 2. Divinylcarbinol Derivatives The competitive rearrangement of 4-vinylallyl vinyl ethers has provided information on the relative rates 1) LDA e 'yC0zMe 2) TMSCl or TBSCl R 72a, R=H b, R=TMS 73 74 of substituted allyl residues. Scheme I illustrates such a study60 wherein the P-substitution pattern of the allylic residue of allyl vinyl ether 75 is systematically altered. The vinyl residue reacts twice as fast as the (E)-P-vinyl group (76a:77a) and 19 times faster than the (Z)+vinyl(76b:77b). At first hand, these data suggest that, in a competition between the (E)- and (Z)-p-vinyl groups, the E isomer should react 9.5 times faster (76b:77b to 76a:77a); the observed result is -3:l (76~77~). As has been suggested,61 each nonreacting group is a substitution for its reacting partner and need not offer additive substituent effects in each rearrangement. The presence of a 6-methyl substitution (E configuration) has a limited effect on the rate of rearrangement of vinyl ethers or silyl ketene acetals (cf., 29 vs 30a). Terminal 6,6-substitution with two methyl substituents completely favors formation of 76d over 77d. Parker and Farmar62 have uncovered a subtle selectivity in the rearrangement of the series of divinylcarbinol derivatives 79. The methyl substituent in 79a and 79b provides a small steric decelerationm while the less sterically demanding methoxyl group manifests itself as a decelerating C5-donor group in both 79c and 79d. These observations lend additional support to the Burrows-Carpenter model for C5-donors.31 For divinylcarbinols wherein one of the vinyl substituents is contained in a ring, rearrangement with the acyclic vinyl group is preferred when the acyclic unit is unsubstitued. Thus, vinyl ethers 82a and 82b give 81a and 81b, respectively, with high selectivity. The presence of an (E)-P-methylvinyl group retards acyclic rearrangement, but it still remains the major pathway for the rearrangement of 82d.64965 SCHEME I 75 75 76 77 76'77 a R1=R2=R3=H &Me 2. ' b Rl=R2=Rd=H. R,=Me 95'5 d R.=R,=H. R,=R.,=Me '00 0 c R,=R,=H R,=Rj=Me 78 22

Thermal, Aliphatic Claisen Rearrangement Chemical Reviews, 1988, Vol. 88, No. 8 1429 78 84a, R,=OH. Rz=H b, Rl=H, R,=OH 80 78180 a, R=Me, X=OEt 65135 b, &Me, X=NMez 74126 c, R=OMe, X=OEi 9515 d, R=OMe, X=NMe, 9713 0- 82 a, R=H, n=l b, R=H, n=2 c, R=Me, n=l d, R=Me, n=2 R 81 - 81/83 (CHZ)" dyH 0 8911 1 8811 2 50150 65/35 83 3. Elimination Alternative sigmatropic rearrangements are not the only irreversible processes that can compete with the Claisen rearrangement; elimination reactions are troublesome. This undesirable, competitive process is particularly acute when at least one olefin is contained in a ring. Cyclohex-2-en-1-01s have been particularly notorious in this regard. Ireland and co-workers@ have prepared the isomeric allylic alcohols 84a and 84b by reduction of the corresponding enone. The minor axial alcohol 84a, when subjected to mercuric ion catalyzed exchange with ethyl vinyl ether, undergoes elimination to dienic products. On the other hand, the major, equatorial allylic alcohol rearranges to aldehyde 85 without incident. An unfavorable transition state for the Claisen rearrangement in the former case may be the result of steric interactions between the angular methyl group and the forming C-C bond. When confronted with the problem of elimination, the use of an alternative strategy is often beneficial. Thus, allylic alcohol 86, when subjected to the Eschenmoser ketene 0,N-acetal variant, provides only a 45% yield of the desired amide 87a along with the products of disproportionation of the dihydropyridine, the immediate product of elimination. However, the Johnson orthoester route gives the ester 87b in 74% ~ield.~'?~~ D. Stereochemistry 1. Transition State The Claisen rearrangement is a suprafacial, con- *:Hz?oH 86 85 phcH*PR 87a, R=CONMe2 b, R=COzEi certed, nonsynchronous pericyclic process that may be considered phenomenolologically as an intramolecular SN2' alkylation. When the sp2-hybridized C1- and C6-positions of allyl vinyl ether are substituted to provide enantiotopic faces at both termini, the rearrangement can proceed through two pairs of stochastically achiral transition states to provide two racemic diastereomers bearing newly created centers of asymmetry at C2 and C3 of the products (Scheme 11). Thus, achiral allyl vinyl ether 91 can provide two enantiomeric chairlike transition states 88 and 90, both of which lead to the racemic diastereomer 89. Similarly, the enantiomeric boatlike transition states 92 and 94 provide racemic, diastereomeric aldehyde 93. The two transition states are inherently unequal in energy and the ratio 89:93 reflects the transition-state geometry. In a detailed study modeled after the Doering and Roth experiments that revealed the preferred chairlike transition state for the Cope rearrangement,69 Schmid26bic and his collaborators have examined the rate and stereochemistry of rearrangement of the four crotyl propenyl ethers 91a-d in the gas phase at 160 "C. All isomers show the expected negative entropy of activation (AS* = -10 to -15 eu) with enthalpies of activation ranging from 25 to 27 kcal/mol. Each isomer shows a clear preference for the chairlike transition state (91a, 95.9:4.1; 91b, 94.7:5.3; 91c, 95.54.5; 91d, 95.4:4.6). The E isomer 91a is found to rearrange an order of magnitude faster than the Z,Z isomer 91b, with the other two geometric isomers intermediate in rate. The E,E and Z,Z isomers rearrange through a chairlike transition state to give the threo isomer 89a (89b) as the major product; likewise, the Z,E and E,Z isomers give the erythro isomer 89c (89d) as the predominant stereoisomer. Since the four isomers 91 all proceed through the chairlike transition state, a change in the geometry of a single double bond exchanges the enantiotopicity of the faces of the double bond and leads to the opposite stereoisomer. Indeed, any pairwise change in olefin geometry for a given transition state, or single change of olefin geometry and change in transition state, results in the formation of the same dia~tereomer.'~ 2. Vinyl Double-Bond Geometry Before proceeding to other substituent effects and how they control the transition state of the Claisen rearrangement, it is appropriate to consider the methods that are available to control the geometry of the vinyl double bond. Unfortunately, no convenient

1430 Chemical Reviews, 1988, Vol. 88, No. 8 SCHEME I1 Ziegler Chair A r L Chair B methods are available for the selective preparation of propenyl ethers. The same difficulty exists with the Johnson orthoester method. The use of orthoesters derived from propionic acid derivatives and their higher analogues fail to give stereochemically defined ketene acetals.71 However, the ketene 0,N-acetal rearrangement does provide for selectivity. Sucrow and Richter72 have examined the Claisen rearrangement of the dimethyl acetal of N8-dimethylpropionaide with (E)- and (2)-crotyl alcohol (Scheme 111). Although the intermediate ketene 0,N-acetals are generated in situ and are not isolated, the assumption that a chairlike transition state is operable, coupled with a preferred axial orientation of the C1-methyl group of the (E)- ketene O,N-acetal(95, loo), correctly accounts for the stereochemistry of the products. The latter supposition is tenable as the dimethylamino group assists in delocalizing charge in the transition state and interacts with the C1 substituent when it is equatorially disposed.73 An alternative approach to the use of the ketene 08-acetals has been offered by the work of Fi~ini~~ who has employed 1-(N8-diethylamino)-l-propyne (101) as the propionate source; however, no stereochemical study was conducted. Recognizing that the (2)-ketene 0,Nacetals of Scheme I11 are the products of thermodynamic control, Bartlett and Hal~ne~~ have prepared the less stable, kinetic @)-ketene 0,N-acetal by the stepwise, cis addition of the crotyl alcohols across the triple bond of the Ficini ynamine (Scheme IV). The slow addition of the crotyl alcohol to the ynamine at 140 OC serves to make rearrangement, a trap for the kinetic addition product, competitive with isomerization. Protonation of the ynamine provides the ketene immonium cation 102, which adds alkoxide preferentially syn to the hydrogen atom via path A. In the case of the (2)-alcohol, a 2.51 ratio of 108 to 107 is obtained; the (E)-alcohol also preferentially follows path A through 105 leading to a 2:l ratio of 107 to 108. The Ireland variant21 of the Claisen rearrangement has proved the most adaptable for the control of vinyl 92 93 (racemic) 94 SCHEME I11 96 (E, equatorial) 95 (E, axial) NMe2 95% I CONMe, 97 (erythro) CONMe, Me 98 (threo) 13% I I 99 (Z, equatorial) 100 (Z, axial) olefii geometry. The deprotonation of esters by lithium dialkylamide bases developed by Rathke22-76 has proved amenable to the generation of specific enolates. Thus, treatment of butenyl propionates 110 and 114 (Scheme V) with lithium diisopropylamide (LDA) in THF under these kinetic conditions forms principally the (2)-lithium enolates, which upon silylation (tert-butyldimethylsilyl (TBS) gives the (E)-0-silyl ketene Rearrangement of silyl ketene acetal 109 provides an 87:13 mixture of acids 111 and 115 after desilylation while 116 gives an 89:ll ratio of 115 and 111. These two products can also be obtained by using 23% HMPA-THF (HMPA = hexamethylphosphoramide)

Thermal, Aliphatic Claisen Rearrangement Chemical Reviews, 1988, Vol. 88, No. 8 1431 TABLE I. Geometry of Enolate Formation OR2 R ox H OX H OR2 RICH~CO~R~ - 117 118 llQ entry 1 2 3 4 5 6 7 8 9 10 11 ester 117a 117b 117c 117d 117e 117a 117b 117c 117c 117f 117d X" TBS TBS TMS TBS TES TBS TBS TMS TES TES TBS solventb THF THF THF THF 23% HMPA-THF 23% HMPA-THF 23% HMPA-THF 23% HMPA-THF 23% HMPA-THF 23% HMPA-THF 23% HMPA-THF "TBS = tert-butyldimethylsilyl; TMS = trimethylsilyl; TES = triethylsilyl. *Enolates were generated with LDA, -70 to -78 "C. SCHEME IV rr NE!, NE!, I I 105 106 103 104 107 (threo) YCONEt, Me 108 (erythro) as an optimal solvent system for the generation of the therm~dynamic'~ (E)-lithium enolates ((2)-0-silyl ketene acetals 112 and 113). 2-(0)-Silyl ketene acetal 112 provides an 81:19 mixture of 115 and 111 while 113 affords an 86:14 mixture of 111 and 115. The major diastereomer in each rearrangement arises through the chairlike transition state, and once again, a single exchange of olefin geometry results in the other diastereomer becoming the major product. The erosion of diastereoselectivity can be attributed to two factors: the geometric integrity of the silyl ketene acetals and the selectivity of the chairlike vs boatlike transition state. Table I provides examples for the selectivity of enolate formation by the LDA/silylation procedure.82 The entries are listed in ascending bulk of the alcohol porSCHEME V 111 112 OSIR, 0 11 0 23% HMPA-THF 23% HMPA-THF I 116 OSIR, tion of the ester group for a given solvent. In general, enolate formation by the kinetic deprotonation procedure is somewhat more selective than the thermodynamic conditions. Because the silyl ketene acetal ratios are approximately equal to, or better than, the ratio of diastereomers 11 1 to 115, the chairlike transition state is virtually the exclusive pathway for rearrangement. 3. Secondary Allylic Alcohols Although the aliphatic Claisen rearrangement of secondary allylic alcohols had been recognized to provide E double bonds,83 Faulkner and Petersen@ have examined the selectivity of olefin formation as a function of C2 substituents. The vinyl ether rearrangement of vinyl ether 120a provides a 9O:lO ratio of (E)- to (2)-unsaturated aldehydes 121a. The congener 120b bearing an isopropyl rather than an ethyl substituent is more selective affording a 93:7 ratio of (E)- to (2)- olefinic aldehydes 121b. An increase in the steric bulk of the C2 substituent produces higher stereoselectivity. Thus, ketene 0,N-acetal rearrangement of 120c gives an E:Z ratio of 99.4:0.6 while the product from the 2- methoxypropene derivative of 2-methylpent-1-en-3-01 (120d) provides less than 1% of the (Z)-olefin. Simi-

1432 Chemical Reviews, 1988, Vol. 88, No. 8 Ziegler larly, Katzenellenbogena5 has reported less than 1 % of (Z)-olefin in the silyl ketene acetal rearrangement of 120e and Johnsona6 has observed >98% E selectivity in the orthoester rearrangement of 120f. This dramatic increase in selectivity observed in the C2,C4-substituted examples has been rationalized as the result of a pseudo-1,3-diaxial interaction in chairlike transition state 123 that leads to the (2)-olefin as opposed to the less congested chairlike transition state 122 that gives the (E)-~lefin.~~~~~~~~ Yo + + tR Z Z 120 a, %El, Z=H 121 b, R=i-Pr, Z=H c, R=Et. Z=NMe, d. R=Et, Z=Me e, R=n-C,H,3, Z=OTBS 1, R=CH,CH,CH(Me)=CH, /- . /f *\<<,.,i,\ n .*-.-7 .-’ R %. .-, &,.*. L*-o . /fl 122 123 When the secondary allylic alcohol has a substituent at the terminus of the double bond (i.e., C6), a center of asymmetry is destroyed during the rearrangement as a new one is created. This process has been often called “self-immolative” a9 and involves the “transfer of chirality”.2f In the sense that racemic substances bearing centers of asymmetry are chiral, and recognizing the concerted, suprafacial nature of the rearrangement, the transformation of Scheme VI is, by necessity, chiral throughout. In more modern terms, that which is transferred is stereogenicityw (Le., stereochemical information), and when practiced with enantiomerically pure allyl vinyl ethers (124), the rearrangement affords enantiomerically pure products (126). In the example of Scheme VI, the R,E enantiomer 124 bearing an equatorial R1 substituent undergoes bond formation on the si face of the allylic double bond to produce the R,E enantiomer 126. Conformational inversion of 124 leads to (R,E)-127. This conformation can undergo re bond formation through transition state 128 having the R1 substituent axial, resulting in the formation of S,Z enantiomer 120. Thus, the transition-state integrity may be monitored with enantiomerically pure reactants by measuring the enantiomeric excess of the dihydro aldehydes from reduction of 126 and 129. In the racemic series, the value (E - Z)/(E + 2) equals the enantiomeric excess that would be obtained using enantiomerically pure allylic alcohols. The chairlike vs boatlike transition state is not detectable in this case because there is no C1 substituent. Hill has observed the “transfer of chirality” in the vinyl ether rearrangement of enantiomerically pure cyclopent-2-en-l-ol.g1a The Eschenmoser and Johnson variants with enantiomerically pure (E)-pent-3-en-2-01 give products with 90% retention of enantiomeric purity as determined by optical rotati~n.~~~~~~~~ An ingenious, enantioconvergent variation on this theme has been executed by Chan.93a The enantiomers of propargyl alcohol 130 are prepared by resolution. The R enantiomer is reduced to the (R,Z)-allylic alcohol 131 while the S enantiomer is converted to the (S,- E)-allylic alcohol 132. Rearrangement to form the aldehyde, ester, or amide 133 occurs with “chirality transmission” of 94-99%. Thus, the (R,Z)-olefin exposes the re face while the (S,E)-olefin invokes the same re face, affording a single enantiomer, the (S,E)-olefin 133. Clearly, the other enantiomer, (R,E)-133, is accessible by exchanging the reduction procedure for each enantiomer of 130.94 The advent of the Sharpless kinetic resolution procedureg5 and the Midland asymmetric reduction of a,@-acetylenic ketones% has made a variety of secondary allylic alcohols readily available in both enantiomeric forms, thereby obviating the use of classical resolution. SCHEME VI (R, and R, lowest carbon priority) LR L J 125 126 124 127 128 129