Structural effects affecting the thermal electrocyclic ring closure of vinylallenes to alkylidenecyclobutenes

Article abstract of DOI:10.1021/ja9522241

The thermal electrocyclic ring closure of (2E,7E)-3,4,7-trialkylnona-2,4,5,7-tetraenes (divinyl-4,5-allenes) is regioselective, occurring at the most sterically congested vinylallene subunit to afford the trisubstituted alkylidenecyclobutene. Ring closure

Full text of DOI:10.1021/ja9522241

J. Am. Chem. Soc. 1996, 118, 1881-1891
1881
Structural Effects Affecting the Thermal Electrocyclic Ring
Closure of Vinylallenes to Alkylidenecyclobutenes
Susana Lo´pez, Jesu´s Rodr´ıguez,† Jose´ Garc´ıa Rey, and Angel R. de Lera*
Contribution from the Departamentos de Qu´ımica Orga´nica e Qu´ımica F´ısica, UniVersidade de
Santiago de Compostela (Gallaecia Fulget 1495-1995), 15706 Santiago, Spain
ReceiVed July 6, 1995. ReVised Manuscript ReceiVed December 22, 1995X
Abstract: The thermal electrocyclic ring closure of (2E,7E)-3,4,7-trialkylnona-2,4,5,7-tetraenes (divinyl-4,5-allenes)
is regioselective, occurring at the most sterically congested vinylallene subunit to afford the trisubstituted
alkylidenecyclobutene. Ring closure of both divinylallenes and vinylallenes displays high torquoselectivity (exclusive
formation of the (E)-alkylidenecyclobutenes) when the substituent at C4 is a sterically demanding alkyl group and
the substituent at C2 is a formyl group. Ab initio calculations clearly demonstrate the dominant steric influence of
the bulky C4 substituent in these selectivities. However, torquoselectivity appears to be enhanced if cyclization
involves loss of conjugation by a π system encompassing the C2 substituent.
The symmetry-based selection rules for electrocyclic reactions
(i.e. for the formation of a σ bond between the termini of a
fully conjugated π system or the reverse process) distinguish
only between disrotatory and conrotatory modes.1 They do not
allow prediction of torquoselectiVity, i.e. they do not distinguish
between the two possible disrotatory modes or the two possible
conrotatory modes for substituted polyenes.2-4 Early attempts
at torquoselectivity prediction were based on the steric effects
of the substituents.1c More recently, ab initio calculations of
the transition structures corresponding to the two alternative
directions of twist have been carried out.4 It is found that
3-substituted cyclobutenes 1 (four-electron systems5) with
different 3-substituents differ by as much as 20 kcal/mol in the
difference between the energies of the two transition structures
and that these large differences are best rationalized in terms
of the electronic nature of the substituents: for electron donors,
outward rotation (1 to 2 in Scheme 1) is favored because it
minimizes repulsive cyclic four-electron interactions and maxi-
mizes stabilizing cyclic two-electron interactions, while the
opposite electronic effects of electron acceptors favor inward
rotation to 3.4b However, analogous calculations for the six-
electron6 ring opening of bicyclo[4.2.0]octa-2,4-diene (4) to
Scheme 1
† Departamento de Qu´ımica F´ısica.
X Abstract published in AdVance ACS Abstracts, February 1, 1996.
(1) (a) Woodward, R. B.; Hoffmann, R. J. Am. Chem. Soc. 1965, 87,
395. (b) Woodward, R. B.; Hoffmann, R. Angew. Chem., Int. Ed. Engl.
1969, 8, 781. (c) Marvell, E. N. Thermal Electrocyclic Reactions; Academic
Press: New York, 1980. (d) Gajewski, J. J. Hydrocarbon Thermal
Isomerizations; Academic Press: New York, 1981.
(2) Houk, K. N. Stereoselective Electrocyclizations and Sigmatropic
Shifts of Strained Rings: Torquoelectronics, in Strain and Its Implications
in Organic Chemistry; de Meijere, A., Blechert, S., Eds.; Kluwer Academic
Publishers: Dordrecht, 1989; pp 25-37.
(3) The alternative term rotoselectiVity is also used; see, for example:
Trost, B. M.; Shi, Y. J. Am. Chem. Soc. 1993, 115, 12491.
(4) (a) Kirmse, W.; Rondan, N. G.; Houk, K. N. J. Am. Chem. Soc. 1984,
106, 7989. (b) Rondan, N. G.; Houk, K. N. J. Am. Chem. Soc. 1985, 107,
2099. (c) Houk, K. N.; Spellmeyer, D. C.; Jefford, C. W.; Rimbault, C. G.;
Wang, Y.; Miller, R. D. J. Org. Chem. 1988, 53, 2125. (d) Spellmeyer, D.
C.; Houk, K. N. J. Am. Chem. Soc. 1988, 110, 3412. (e) Kallel, E. A.;
Wang, Y.; Spellmeyer, D. C.; Houk, K. N. J. Am. Chem. Soc. 1990, 112,
6759. (f) Jefford, C. W.; Bernardinelli, G.; Wang, Y.; Spellmeyer, D. C.;
Buda, A. B.; Houk, K. N. J. Am. Chem. Soc. 1992, 114, 1157. (g) Nakamura,
K.; Houk, K. N. J. Org. Chem. 1995, 60, 686.
cycloocta-1,3,5-triene (5, R ) H) and the eight-electron6,7 ring
opening of compounds 5 to 1-substituted octa-1,3,5,7-tetraenes
(6) have found transition structures in which the relationships
between the substituents at the interacting termini and the
breaking or forming C-C bond are essentially independent of
the direction of twist; the consequent theoretical prediction of
just a small preference for the outward twist mode (due to the
steric effect of the substituent) in the ring opening of 7-substi-
tuted cycloocta-1,3,5-trienes 5 is in keeping with experimental
findings.7
As part of our research on structural influences on the
activities of retinoids as cell differentiation promoters, we
(5) For a review, see: Durst, T.; Breau, L. Cyclobutene Ring Opening
Reactions, in ComprehensiVe Organic Synthesis; Trost, B. M., Fleming, I.,
Paquette, L. A., Eds.; Pergamon Press: Oxford, 1991; Volume 5, Chapter
6.1, pp 675-697.
(6) Thomas, B. E.; Evanseck, J. D.; Houk, K. N. Isr. J. Chem. 1993, 33,
287.
(7) Thomas, B. E.; Evanseck, J. D.; Houk, K. N. J. Am. Chem. Soc.
1993, 115, 4165.
0002-7863/96/1518-1881$12.00/0 © 1996 American Chemical Society

1882 J. Am. Chem. Soc., Vol. 118, No. 8, 1996
Lo´pez et al.
Scheme 2a
a (a) n-BuLi, -78 °C to room temperature; (b) n-BuLi, PhCOCl, -78 °C to room temperature; (c) for 15a, t-BuLi, CuCN, -78 °C to room
temperature, 71%; for 15b, i-PrMgCl, CuBr‚SMe2, THF, -60 °C, 1 h, then 12 h at room temperature, 70%; for 15c, EtMgBr, CuBr‚SMe2, THF,
-60 °C, 1 h, then 12 h at room temperature, 60%; for 15d, MeMgBr, CuI‚LiBr, THF, 0 °C, 5 h, 70%; (d) TBAF, THF, room temperature; (e)
MnO2, CH2Cl2, room temperature. See supporting information for experimental procedures and spectroscopic characterization of these compounds.
introduced an allene group in the retinoid side chain.8,9 On
studying the thermal behaviour of these divinylallenes (7), we
discovered that, under relatively mild conditions, a four-electron
electrocyclic ring closure took place10a (cf.,11 9-10). Since in
some cases the reaction exhibited remarkable regioselectivity
(exclusive cyclization to alkylidenecyclobutenes 8) and tor-
quoselectivity10b (exclusive cyclization to the E-alkylidenecy-
clobutenal 8 when R2 ) CHO), we were prompted to a fuller
investigation of its dependence on steric and electronic effects.
We describe here the results of this investigation, which
included kinetic studies and comprehensive analysis of model
systems by means of ab initio calculations. We evaluated both
steric effects (by varying the bulk of the alkyl substituent on
the allene: t-Bu, i-Pr, Et, or Me), and the electronic effect of
the terminal R2 group (H, alkyl, CHO). Experimental results
and theoretical analysis indicate that in these systems torquo-
selectivity is mainly of steric origin but is further enhanced by
electronic effects if R2 is a formyl group.
Results and Discussion
Preparation of the Divinylallenes. The preparation of
substituted divinylallenes was based on the regioselective SN2′
displacement of propargylic benzoates by organocopper re-
agents.8,9,12 Incorporation of the tert-butyl group requires the
mixed heterocuprate dilithium di-tert-butyl cyanocuprate,12b,c but
the less bulky alkyl derivatives are best synthesized with the
Grignard reagent in the presence of stoichiometric amounts of
copper salts.12a,d
The lithium salt of TBDMS-protected (E)-3-methylpent-2-
en-4-yn-1-ol (12)13a was reacted with aldehyde 1113b to afford
propargylic alcohol 13 in 93% yield. Benzoylation of 13
afforded 14 in 88% yield. Addition of the benzoate 14 to (t-
Bu)2Cu(CN)Li2 in ether at -78 °C gave a 71% yield of the
TBDMS-protected 4,5-allenic alcohol 15a, in which the bulky
tert-butyl group is borne at position C4 (C12 of the retinoid side
chain9). Deprotection of 15a by treatment with TBAF14 gave
an 83% yield of the 4-tert-butyl-substituted divinylallenol 16a,
which was oxidized with MnO215 to divinylallenal 17a in 82%
yield (Scheme 2).
Addition of the benzoate 14 to a solution of the cuprate
obtained by reaction of i-PrMgCl or EtMgBr with CuBr‚SMe2
cooled to -60 °C afforded allenes 15b and 15c, respectively.
For the incorporation of the methyl group at C4 in 15d, benzoate
14 was added to a solution of the cuprate obtained by treating
(8) For leading references on the chemistry of vinylallenes, see: (a)
Okamura, W. H. Acc. Chem. Res. 1983, 16, 81. (b) Pasto, D. J. Tetrahedron
1984, 40, 2805. (c) Okamura, W. H.; Curtin, M. L. Synlett 1990, 1. For
monographs, see: (d) The Chemistry of Ketenes, Allenes and Related
Compounds, Part I and Part II; Patai, S., Ed.; John Wiley and Sons: New
York, 1980. (e) Landor, S. R. The Chemistry of the Allenes; Academic
Press: New York, 1982. (f) Schuster, H. F.; Coppola, G. M. Allenes in
Organic Chemistry; Wiley: New York, 1984.
(9) For a review on allenic retinoids, see: de Lera, A. R.; Chandraratna,
R. A. S.; Okamura, W. H. Synthesis and Studies of 12-s-cis Conforma-
tionally Locked Retinoids. In Chemistry and Biology of Synthetic Retinoids;
Dawson, M. I., Okamura, W. H., Eds.; CRC Press: Boca Raton, FL, 1990;
Chapter 9, pp 201-227.
(10) (a) Garc´ıa Rey, J.; Rodr´ıguez, J.; de Lera, A. R. Tetrahedron Lett.
1993, 34, 6293. (b) Lo´pez, S.; Garc´ıa Rey, J.; Rodr´ıguez, J.; de Lera, A. R.
Tetrahedron Lett. 1995, 36, 4669. Note that these communications used
the retinoid nomenclature and numbering system (IUPAC-IUB Joint
Commission on Biochemical Nomenclature, Eur. J. Biochem. 1982, 129,
1). Due to the non-retinoid nature of some of the compounds involved in
the work now reported, systematic nomenclature has been used throughout
this article.
(11) (a) Upon heating to 170 °C, the parent penta-1,2,4-triene gives,
among other dimeric products, the parent methylenecyclobutene. See:
Schneider, R.; Siegel, H.; Hopf, H. Liebigs Ann. Chem. 1981, 1812. We
thank Prof. Hopf for calling our attention to some of his early ideas on
periselectivity in vinylallenes: Hopf, H. Nachr. Chem. Techn. 1975, 23,
235. For thermal 4π electrocyclic ring closure of substituted vinylallenes,
see: (b) Gil-Av, E.; Herling, J. Tetrahedron Lett. 1967, 1. (c) Pasto, D. J.;
Kong, W. J. Org. Chem. 1989, 54, 4028. (d) Murakami, M.; Amii, H.;
Itami, K.; Ito, Y. Angew. Chem., Int. Ed. Engl. 1995, 34, 1476.
(12) (a) Erdik, E. Tetrahedron 1984, 40, 641. (b) Lipshutz, B. H.
Synthesis 1987, 325. (c) Lipshutz, B. H.; Sengupta, S. Org. React. N.Y.
1992, 41, 135. (d) Westmijze, H.; Meijer, J.; Bos, H. J. T.; Vermeer, P.
Recl. TraV. Chim. Pays-Bas 1976, 95, 299.
(13) (a) Marshall, J. A.; Tang, Y. J. Org. Chem. 1994, 59, 1457. (b)
Isler, O.; Huber, W.; Ronco, A.; Kofler, M. HelV. Chim. Acta 1947, 30,
1911.
(14) Corey, E. J.; Venkateswarlu, A. J. Am. Chem. Soc. 1972, 94, 6190.
(15) Fatiady, A. J. Synthesis 1976, 65.

Thermal Electrocyclic Ring Closure of Vinylallenes
J. Am. Chem. Soc., Vol. 118, No. 8, 1996 1883
Scheme 3a
Scheme 4
Scheme 5
a (a) Ph3P, CBr4, CH2Cl2, 0 °C; (b) n-BuLi, THF, -78 °C to room
temperature; (c) n-BuLi, THF, -78 °C to room temperature; (d) n-BuLi,
PhCOCl, THF, -78 °C to room temperature; (e) t-BuLi, CuCN, ether,
-78 °C to room temperature; (f) TBAF, THF, room temperature; (g)
MnO2, CH2Cl2, room temperature. See supporting information for
experimental procedures and spectroscopic characterization of these
compounds.
a solution of CuI/LiBr in THF with MeMgBr at 0 °C.
Functional group manipulations14,15 provided, uneventfully, the
alcohols 16b-d and aldehydes 17b-d in the yields shown in
Scheme 2. Finally, the unfunctionalized C3-tert-butyl-substi-
tuted allene 21 (Scheme 2) was prepared following the same
strategy, starting from 2-methylbut-1-en-3-yne (18).
The same general synthetic strategy afforded the 6-tert-butyl-
4,5-allenes (10-tert-butyl-11,7-retroretinoids9) 27-29 (Scheme
3). The propargylic benzoate precursor 26 was derived in 93%
yield from alcohol 25, which was itself obtained by condensation
of the protected aldehyde 2416 with the conjugated enyne 23.
The one-carbon homologation of aldehyde 11 to alkyne 23 was
carried out in 65% overall yield by the Corey-Fuchs proce-
dure,17 with dibromide 22 as intermediate.
NMR spectra were recorded periodically. Particularly helpful
for quantitative analysis was the disappearance of the allenyl
proton singlet and its replacement by the equivalent vinyl
singlet(s). The integration of these signals allowed calculation
of the mole fractions of starting material and product(s) at
successive times (see supporting information).
Regioselectivity and Kinetics of the Thermal Rearrange-
ment. The thermal rearrangement products of the divinylallenes
were identified by analysis of the 1H NMR spectra of the
reaction mixtures (the appearance of mutually coupled signals
between δ ∼3.0 ppm and δ 4.5 ppm, together with the vinyl
hydrogen singlet indicated a four-electron electrocyclization),
and for the tert-butyl-substituted divinylallenes these identifica-
tions were supported by the fact that, for stereochemical and
structural reasons, these t-Bu compounds, unlike other series
of vinylallenes,8 cannot undergo the alternative facile sigma-
tropic shifts ([1,3]-H, [1,5]-H, [1,7]-H) or six-electron electro-
cyclization.
The regioselectivity of the process was remarkable. Only
products arising from four-electron electrocyclizations at the
more congested trisubstituted vinylallene subunit were detected
(originating from the 3-s-cis conformer shown in Scheme 4 for
15-17 and from the 6-s-cis conformation shown in Scheme 5
for the positional isomers 27-29), regardless of the steric bulk
of the R1 substituent, the nature of the polar group R2, or the
location of R2 relative to the cyclization termini.
Analysis of the kinetic data for the C4-alkyl-substituted
divinylallenes (supporting information),18 showed a first-order
behavior, except where indicated in Table 1, which lists rate
constants and half-lives. For a given R1, the half-lives of the
protected and unprotected alcohols 15 and 16 were similar and
much shorter than those of the corresponding aldehydes (17)
(Table 1).19 The long half-lives (high activation energies) of
compounds 17 are probably due to the formation of compounds
32 involving loss of conjugation in the transition state. For
comparison with other pericyclic reactions, the activation
parameters of the reaction of 15a were calculated from the rate
constants determined (Figure 1 shows the dissappearance of 15a
at 90 °C) in the range 90-110 °C: plotting ln k against 1/T
(Figure 2) gave values of ∆Hq ) 26.6 ( 1.3 kcal/mol and ∆Sq
) +7.4 ( 0.4 cal/mol K (Ea ) 27.3 kcal/mol at 98.9 °C).18
The kinetics for the C6-tert-butyl series were more complex,
due to reversibility (Vide infra). Moreover, whereas the
reactions of 27 and 28 required heating at 100 °C, 29 rearranged
at 80 °C, which may be explained by reversing the argument
For kinetic analysis of the reactions, dilute solutions of the
divinylallenes in benzene-d6 were heated in sealed NMR tubes
at appropriate temperatures (range 90-120 °C), and their H
(18) We thank Prof. Okamura for generously providing a copy of the
kinetic analysis computer program.
(19) Our ab initio calculations (not shown) confirm previous findings
that the ionic canonical structures, thought to diminish electron density at
the diene terminus in 17a-d, do not contribute significantly to the ground
state structures of enals. For an in-depth treatment, see: Wiberg, K. B.;
Schreiber, S. L. J. Org. Chem. 1988, 53, 783.
1
(16) Torrado, A.; Iglesias, B.; Lo´pez, S.; de Lera, A. R. Tetrahedron
1995, 51, 2435.
(17) Corey, E. J.; Fuchs, P. L. Tetrahedron Lett. 1972, 3769.

1884 J. Am. Chem. Soc., Vol. 118, No. 8, 1996
Lo´pez et al.
Table 1. Kinetic and Stereochemical Characteristics the Thermal Rearrangement of Divinylallenes to Alkylidenecyclobutenes (Scheme 6)
k × 10-2
divinylallene
R1
R2
T (°C)
(h-1
)
τ
1/2 (h)
product
E/Z ratio
yield (%)
15a
16a
17a
15b
16b
17b
15c
16c
17c
15d
16d
17d
21
t-Bu
t-Bu
t-Bu
i-Pr
i-Pr
i-Pr
Et
Et
Et
Me
Me
Me
t-Bu
CH2OTBDMS
CH2OH
CHO
CH2OTBDMS
CH2OH
CHO
CH2OTBDMS
CH2OH
CHO
CH2OTBDMS
CH2OH
CHO
100
100
100
120
120
120
120
120
120
120
120
120
120
9.6
10.0
2.9
6.9
7.2
0.77
4.7
4.4
c
4.7
5.0
c
7.2
6.9
24.1
10.0
9.6
90.0
14.7
15.5
30a
31a
32a
30b
31b
32b
30c
31c
32c
30d
31d
32d
36
83:17
86:14
>99:1
80:20
82:18
>99:1
66:34
60:40
d
85
89
92
81
84
79a,b
75
79
5a,e
53a
67a
5a,e
80
14.6
13.7
50:50
58:42
d
H
5.2
13.3
91:9
a Unstable.21 b Yield for the total mass balance, after 40 h of reaction, with ∼72% of starting material still present; after longer reaction times,
1
signals for additional products were observed in the H NMR spectrum. c Complex kinetic behavior. d Not determined, due to the low yield of
alkylidenecyclobutene. e The yield of isolated (E)-alkylidenecyclobutene; the major product was an aldehyde of complex, still undetermined, structure.
Scheme 6
Starting from 15a, two isomers, E-30a and Z-30a (Scheme
6), were isolated in an 83:17 ratio. A strong NOE on H1′ after
saturation of the C2-tert-butyl group established that E-30a was
E with respect to the exocyclic double bond, while Z stereo-
chemistry and a predominant 1′-s-cis conformation for isomer
Figure 1. Plot of ln [15a] versus time for the rearrangement of 15a to
30a in C6D6 at 90 °C ([15a] measured by monitoring the H NMR
Z-30a were mainly supported by an NOE cross peak between
the C2-tert-butyl group and H3′. The high degree of stereose-
lectivity was attributed to the bulky C4-tert-butyl group favoring
the conrotatory motion in which the substituent at the allene
terminus (C6) rotates inward, as depicted in Scheme 6.20 This
explanation is supported by the E/Z ratios of 30b (80:20), 30c
(66:34), and 30d21 (50:50), which show that the predominance
of the “C6-inward” conrotatory mode decreases with decreasing
steric bulk of the C4 substituent. E/Z ratios similar to those of
the silyl ethers were observed upon reaction of the corresponding
alcohols 16a-d (Table 1). Isomer identification is based on
the fact that 1H NMR signals for H1 in (E)-alkylidenecy-
clobutenes resonate downfield (up to 0.4 ppm) from the
corresponding (Z)-alkylidenecyclobutene signals.
1
signals attributed to H6 of the reactant and H1′ of the product). Assuming
irreversible first-order kinetics, this plot implies a half-life of 21.16 h
and a rate constant k ) 3.27 × 10-2 h-1
.
1
The H NMR spectra of the crude final reaction mixtures
obtained from the C4-tert-butyl and C4-iso-propyl 4,5-allenals
17a and 17b each showed only the corresponding (E)-alkyli-
denecyclobutenal (E-32a or E-32b). The E geometry was
confirmed by NOE experiments. However, prolonged heating
(120 °C, 66 h) of the C4-ethyl and C4-methyl divinylallenals
17c and 17d produced only small quantities (∼5%) of the
alkylidenecyclobutenals E-32c and E-32d together with major
Figure 2. Arrhenius plot for the electrocyclization of 15a to 30a in
the range 90 to 110 °C.
(20) Note that an inward movement of the C6 substituent in the
enantiomer shown in Scheme 6 could also be described as a combined
outward movement of the C5 sp center and inward movement of the C2 sp2
center (outward-inward) for the conrotatory mode leading to the (E)-
alkylidenecyclobutenes.
applied to the C4-tert-butyl series: milder reaction conditions
are possible for 29 because of an increase in conjugation in the
transition state leading to trienal 35.
(21) After HPLC separation, the evaporation of solvent followed by rapid
acquisition of 1H NMR data in C6D6 and evaporation of deuterated solvent
resulted in complete loss of product. Characterization was therefore based
on the comparison of available NMR data with relevant data for E-30a
and Z-30a.
Stereo/Torquoselectivity. The E/Z ratios estimated from the
1H NMR spectra of the crude final reaction mixtures were
confirmed by HPLC whenever possible. The results for the
C4-alkyl series are listed in Table 1.

Thermal Electrocyclic Ring Closure of Vinylallenes
J. Am. Chem. Soc., Vol. 118, No. 8, 1996 1885
Scheme 7
components whose structure have not been completely deter-
mined but are unlikely to have arisen from the missing Z-32c
and Z-32d.
In the reactions of the C6-tert-butyl series 27-29 the electron-
acceptor group played no discriminant role. After short reaction
times, the alkylidenecyclobutenes 33-35 were all obtained in
E/Z ratios of ∼70:30. After longer reaction times, only the (E)-
alkylidenecyclobutenes E-34 and E-35 were isolated from the
reaction mixtures of 28 and 29, and although the mixture derived
from 27 could not be separated by HPLC, its qualitative behavior
was similar to that of 29 (described below). These results can
be explained, in keeping with the kinetic results, in terms of
reversible electrocyclization (Scheme 7). To prove reversibility,
pure Z-35, isolated from the short-time reaction mixture of 29,
was resubjected to the same reaction conditions (heating to 80
°C); 1H NMR monitoring clearly showed the development of a
mixture of 29, E-35, and Z-35 (in a 30:14:56 ratio after 5 h
and 33:49:18 ratio after 45 h; after 71 h, no Z-35 was detected,
and the 29/E-35 ratio was 33:67). Steric interactions between
the substituents of the cyclobutene in Z-33, Z-34, and Z-35 may
be responsible for the observed thermodynamic equilibration
to the more stable (E)-alkylidenecyclobutenes.
Figure 3. Computed energies of the vinylallenes 38, 39, and 44 as
functions of the torsion angle C2C3C4C5.
Chart 1
To determine whether the difference in torquoselectivity
between the reactions of 17 (a and b) and those of the
corresponding alcohols and silyl ethers was due to an electronic
effect of the formyl group22 or merely to its not impairing the
steric dominance of the C4-tert-butyl or C4-iso-propyl groups,
we studied the behavior of pentaene 21 (Scheme 2) which has
no polar end group. Heating 21 at 120 °C provided, after HPLC
purification, a 91:9 mixture of the highly unstable21 alkyli-
denecyclobutenes E-36 and Z-36 (Scheme 6, Table 1). Had
the steric effect been the only factor involved in the torquose-
lectivity of the electrocyclic ring closure, E-36 should have been
the only product obtained from 21. The especially high
torquoselectivity of the reactions of 17a and 17b may therefore
be attributed in part to an electronic effect of the formyl group.
Consequently, the uniform behavior of 27-29 is attributable
to the electronic effect of R2 being extinguished by the greater
distance between R2 and the cyclization site in these compounds.
Ab Initio Studies: (A) Regioselectivity. To investigate
further the regioselectivity of ring closure in divinylallenes, we
performed ab initio calculations23 on the substituted vinylallenes
38 and 39 (comprised in the model divinylallene 37) and 44
(Chart 1).
There is, however, another stable conformation with a dihedral
angle of 34.5°, closer to the 3-s-cis form required for cyclization.
In contrast, only a single stable conformer, with a dihedral angle
of 96.7°, was obtained for 38.
The electrocyclic conversion of the vinylallenes 38, 39, and
44 to the corresponding cyclobutenes was investigated at the
6-31G* level with complete optimization and, for reactants,
transition structures and products,23 single-point calculations
including electron correlation by second-order Møller-Plesset
perturbation theory (MP2).24 The main geometric parameters
are listed in Table 2, and the cartesian coordinates are available
in the supporting material. The transition structures are shown
in Figure 4, and the absolute and relative energies of the
electrocyclization processes are listed in Table 3.
In keeping with the regioselectivity observed in our experi-
ments with divinylallenes, the calculated barrier to cyclization
of 3-tert-butyl-4-methylhexa-1,2,4-triene (38) is lower than the
barriers to cyclization of either conformer of 4-methylhexa-
1,2,4-triene (39 and 39′) (Table 3). This difference is attributed
to repulsive interactions between the C3-tert-butyl and C4-methyl
groups in 38, facilitating the proximity of the cyclization termini
C2 and C5 in the transition state (Figure 4), in which the atoms
Conformational analysis of 38 and 39 using a 6-31G* basis
set and complete optimization24 provided potential energy curves
for rotation about the C3-C4 bond (Figure 3). For 39, the most
stable conformation is 3-s-trans (39′: C2C3C4C5 ) 180.0°), in
agreement with the results of calculations on similar systems.25
(22) The A values for formyl, methyl, and hydroxymethyl substituents
are 0.56-0.73, 1.70, and 1.76, respectively. See: Eliel, E. L.; Wilen, S.
H.; Mander, L. N. Stereochemistry of Organic Compounds; Wiley: New
York, 1994; pp 696-697.
(23) For reviews on transition-structure modeling discussing the use of
RHF wave functions, see: (a) Houk, K. N.; Li, Y.; Evanseck, J. D. Angew.
Chem., Int. Ed. Engl. 1992, 31, 682. (b) Eksterowicz, J. E.; Houk, K. N.
Chem. ReV. 1993, 93, 2439.
(24) Gaussian 92/DFT, Revision G.4. Frisch, M. J.; Trucks, G. W.;
Schlegel, H. B.; Gill, P. M. W.; Johnson, B. G.; Wong, M. W.; Foresman,
J. B.; Robb, M. A.; Head-Gordon, M.; Replogle, E. S.; Gomperts, R.;
Andres, J. L.; Raghavachari, K.; Binkley, J. S.; Gonzalez, C.; Martin, R.
L.; Fox, D. J.; Defrees, D. J.; Baker, J.; Stewart, J. J. P.; Pople, J. A.,
Gaussian, Inc., Pittsburgh, PA, 1993.
(25) Bond, D. J. Org. Chem. 1990, 55, 661.

1886 J. Am. Chem. Soc., Vol. 118, No. 8, 1996
Lo´pez et al.
Table 2. Main Geometric Parameters (Bond Lengths in Angstroms, Bond Angles in Degrees) of Substrates, Products, and Transition
Structures in the Electrocyclization of Vinylallenes to Methylenecyclobutenes (Chart 1)a
atomsb
38
TS
40
39
39′
TS
41
44
44′
TS
45
2-5
3.260
1.325
1.300
1.515
1.298
2.102
1.403
1.369
1.405
1.311
1.528
1.524
1.486
1.340
1.319
3.009
1.326
1.300
1.491
1.296
3.636
1.329
1.302
1.481
1.296
2.114
1.406
1.368
1.392
1.309
1.537
1.529
1.472
1.332
1.317
2.993
1.326
1.301
1.501
1.297
3.630
1.330
1.305
1.494
1.296
2.111
1.404
1.369
1.396
1.311
1.534
1.527
1.476
1.335
1.318
4-5
2-3
3-4
1-2
4-5-2
3-2-5
3-4-5
2-3-4
1-2-3
2-3-4-5
40.10
42.40
119.97
117.00
179.62
96.68
71.54
76.90
105.36
100.03
154.77
25.54
83.97
88.91
94.75
92.37
138.50
0.19
54.62
52.82
120.94
126.00
179.61
34.47
c
c
71.93
75.90
105.08
102.03
154.49
23.30
83.80
88.42
94.11
93.66
137.16
0.23
53.23
53.35
119.97
121.13
179.69
49.94
c
c
71.58
76.46
105.64
100.99
153.53
23.88
83.64
88.79
94.53
93.03
136.56
0.21
118.18
125.73
180.00
180.00
120.09
121.05
180.00
180.05
a At the MP2/6-31G*//RHF/6-31G* level. b See Chart 1. c Irrelevant for the 3-s-trans conformation.
Figure 4. Front (a) and top (b) views of the transition structures calculated at the RHF/6-31G* level for cyclization of vinylallenes 38, 39, and 44
to methylenecyclobutenes 40, 41, and 45, respectively.
Table 3. Total and Relative Energies of Reactants, Transition
Structures, and Products (Chart 1) in the Electrocyclization of
Vinylallenes to Methylenecyclobutenesa
more stable than the corresponding vinylallenes, stabilization
is substantially greater for the C3-tert-butyl-substituted vinyl-
allene 38 (14.9 kcal/mol) than for vinylallenes 39 (11.0 kcal/
mol) or 39′ (8.3 kcal/mol). This result contrasts with the
equilibrium observed in the parent system, in which the
vinylallene (42) and the methylenecyclobutene (43) are of
similar stability.11
The parameters determined for 3,4-dimethylhexa-1,2,4-triene
(44), a vinylallene with methyl substituents at both C3 and C4,
are intermediate between those found for the equivalent
conformers of 38 and 39. The consequent prediction that
cyclization of 15d, 16d, and 17d should take place with total
regioselectivity is, as we have seen, borne out experimentally.
(B) Torquoselectivity. The observed torquoselectivity of the
electrocyclic ring closure of divinylallene 15a and analogues
was studied further by performing MP2/6-31G*//RHF/6-31G*
calculations24 for the electrocyclic ring closure of the tetraenes
and triene depicted in Scheme 8.
The chief geometric parameters of the transition structures
for the conrotatory processes leading to the (E)- and (Z)-
alkylidenecyclobutenes are listed in Table 4; cartesian coordi-
nates can be found in the supporting information.
structureb
total energyc
relative energyd
38
-428.376 87
0.0
(t2-3-4-5 ) 96.7°)
TS
40
39′
(t2-3-4-5 ) 180.0°)
39
(t2-3-4-5 ) 34.5°)
TS
-428.334 71
-428.400 58
-271.709 23
26.4
-14.9
0.0
-271.704 96
2.7
-271.657 51
-271.722 47
-310.878 41
32.5
-8.3
0.0
41
44′
(t2-3-4-5 ) 180.0°)
44
(t2-3-4-5 ) 49.9°)
TS
-310.874 80
2.3
-310.830 31
-310.897 69
30.2
-12.1
45
a At the MP2/6-31G*//RHF/6-31G* level. b Dihedral angles in
parentheses. c In hartrees. d In kcal/mol.
forming the 4-membered ring are almost coplanar. Consistent
with this steric argument are the distances between the cycliza-
tion termini in the transition state, 2.102 and 2.114 Å for 38
and 39, respectively. Furthermore, although the energy values
in Table 3 clearly show that cyclization is thermodynamically
favored for both 38 and 39, the methylenecyclobutenes being
The geometries of the transition structures26 (Figure 5) are
roughly similar and are comparable to those computed for the
ring closure of vinylallene 38 (Figure 4). The substantial
bending of the allene group (C4C5C6 ranges from 18.7° for 50

Thermal Electrocyclic Ring Closure of Vinylallenes
J. Am. Chem. Soc., Vol. 118, No. 8, 1996 1887
Table 4. Main Geometric Parameters (Bond Lengths in Angstroms, Bond Angles in Degrees) of the Transition Structures Corresponding to
the Alternative Conrotatory Modes of the Electrocyclization of Vinylallenes to Alkylidenecyclobutenes (Scheme 8)a
46
48
50
52
54
atomsb
2-5
2-3
4-5
3-4
5-6
TS-Z
TS-E
TS-Z
TS-E
TS-Z
TS-E
TS-Z
TS-E
TS-Z
TS-E
2.064
1.404
1.371
1.407
1.320
2.111
1.406
1.372
1.401
1.320
2.095
1.404
1.369
1.398
1.318
2.129
1.406
1.372
1.393
1.319
2.061
1.413
1.380
1.399
1.315
2.103
1.415
1.379
1.395
1.314
2.052
1.400
1.374
1.406
1.320
2.102
1.399
1.372
1.403
1.319
2.094
1.413
1.380
1.397
1.306
2.107
1.414
1.379
1.396
1.306
3-2-5
4-5-2
2-3-4
3-4-5
4-5-6
2-3-4-5
72.47
77.74
104.47
99.33
159.91
24.78
71.50
76.59
105.53
100.55
149.20
24.92
71.94
76.83
105.23
100.67
156.04
23.72
71.40
75.87
105.94
101.81
148.09
23.25
72.94
77.40
103.99
100.08
161.34
23.89
72.04
76.41
104.86
101.10
151.54
24.28
72.89
77.80
104.28
99.17
159.42
24.45
71.77
76.55
105.31
100.25
150.12
25.45
72.34
76.62
104.72
100.92
158.19
23.76
72.00
76.31
104.97
101.15
156.41
24.27
a At the MP2/6-31G*//RHF/6-31G* level; E and Z refer to the geometry of the exocyclic double bond formed in the process. b The same numbering
system (see Scheme 8) was adopted for all the vinylallenes, irrespective of their substitution pattern.
Table 5. Total and Relative Energies, and C2-C5 Distances, of the
Alternative Transition Structures for the Electrocyclization of
Vinylallenes to (Z)- and (E)-Alkylidenecyclobutenes (Scheme 8)a
Scheme 8
d
e
structure
Eb
∆Eac
rC-C
∆rC-C
46
TS-Z
TS-E
48
TS-Z
TS-E
50
TS-Z
TS-E
52
TS-Z
TS-E
54
-544.644 08
-544.645 96
1.18
2.064
2.111
0.047
0.035
0.042
0.050
0.013
-427.141 44
-427.141 57
-0.08
2.43
2.095
2.130
-618.497 32
-618.501 20
2.061
2.103
-505.471 11
-505.473 52
1.51
2.052
2.102
TS-Z
TS-E
-541.358 99
-541.360 25
0.79
2.094
2.107
to 31.9° for 48) is not surprising, having previously been
observed in cumulenes.27c,28 What is of note is that steric
interactions between the substituent at C4 and the substituent at
the ensuing exocyclic double bond, which are negligible for
38, seem likely to play a predominant role in determining the
preferred direction of twist: when R1 is bulky (t-Bu or,
presumably, i-Pr), steric crowding appears to be greater in the
transition structure leading to the (Z)-alkylidenecyclobutene (TS-
Z) than in TS-E, the transition structure leading to the (E)-
alkylidenecyclobutene; this would seem to explain why, for R1
) t-Bu (46 in Scheme 8), the difference between the energies
of TS-E and TS-Z is 1.18 kcal/mol in favor of the former (Table
5). This value is in reasonable agreement with the E/Z isomer
ratio of 83:17 obtained on the thermal rearrangement of
divinylallene 15a to 30a in solution (Scheme 6, Table 1). For
the C4-methyl tetraene 48, the energy difference between TS-Z
and TS-E is negligible (0.08 kcal/mol). This again is in keeping
with the E/Z isomer ratio of 50:50 obtained upon heating the
corresponding divinylallene, 15d (Table 1). These calculations
thus support the conclusions drawn above from the experimental
results: that the preference for electrocyclization to the (E)-
alkylidenecyclobutene that is shown by divinylallenes with bulky
C4 substituents is due mainly to steric interaction between the
a At the MP2/6-31G*//RHF/6-31G* level. b In hartrees. c ∆Ea ) ETS-Z
- ETS-E in kcal/mol. d C2-C5 distance in Å. e ∆rC-C ) rC-C(TS-E) -
rC-C(TS-Z) in Å.
C4 substituent and the substituent on the ensuing exocyclic
double bond; this interaction causes the energy of the TS-Z
transition state to be higher than that of the TS-E transition state.
The experimental results had also suggested that the electronic
properties of a formyl group at the other cyclization terminus
also contributed to torquoselectivity (see above). Computational
investigations of this effect began with calculations of the energy
difference between TS-E and TS-Z for 50 and 52. As expected,
both differences were in favor of TS-E, by 2.43 and 1.51 kcal/
mol for 50 and 52, respectively (Table 5). The order 46 < 52
< 50 defined by these values and the value already obtained
for 46 confirms that the torquoselective effect of the C2
substituent is not primarily steric. However, the hypothesized
electronic effect does not seem to be due, as might be expected,
to the mixing of frontier orbitals to which torquoselectivity in
3-substituted cyclobutenes is attributed,4b even though the
geometries of the relevant parts of 46, 50, and 52 in the transition
states are similar to those calculated for 1.4 Firstly, the energy
differences between TS-E and TS-Z for 46, 50, and 52 differ
by no more than 1.25 kcal/mol, whereas orbital mixing is
thought to be responsible for observed differences of up to 20
kcal/mol for 1.4b Secondly, the fact that in the ring closure of
vinylallenes the sp center at C5 is transformed into an sp2 center
means that the C2 substituent has very much the same relation-
ship to the breaking or forming σ bond, regardless of whether
it rotates inwards or outwards, so that the “intended correlation”
of the orbitals involved in the breaking and forming bonds
cannot explain the stereoselectivity. Nor does interaction
(26) The process described here might qualify as a pseudopericyclic
reaction27 according to the definition of Lemal et al.:27a “a concerted
transformation whose primary changes in bonding encompass a cyclic array
of atoms, at one (or more) of which nonbonding and bonding atomic orbitals
interchange roles...”. For a detailed discussion and ab initio theoretical
analysis of pseudopericyclic reactions, see ref 27c.
(27) (a) Ross, J. A.; Seiders, R. P.; Lemal, D. M. J. Am. Chem. Soc.
1976, 98, 4325. (b) Henriksen, U.; Snyder, J. P.; Halgren T. A. J. Org.
Chem. 1981, 46, 3767. (c) Birney, D. M.; Wagenseller, P. E. J. Am. Chem.
Soc. 1994, 116, 6262.
(28) Trinquier, G.; Malrieu, J.-P. J. Am. Chem. Soc. 1987, 109, 5303
and references cited therein.

1888 J. Am. Chem. Soc., Vol. 118, No. 8, 1996
Lo´pez et al.

Thermal Electrocyclic Ring Closure of Vinylallenes
J. Am. Chem. Soc., Vol. 118, No. 8, 1996 1889
Scheme 9a
a Same reagents and reaction conditions as described in Scheme 2
for 15a, with the yields indicated in parentheses. See supporting
information for experimental procedures and spectroscopic characteriza-
tion of these compounds.
attributable to the absence of C4-C6 steric interaction having
resulted in TS-E occurring only very little earlier than TS-Z
(∆rC-C ) 0.013 Å, as against ∼0.040 Å in the series 46, 48,
50, 52; see Table 5)30 so that the effect of -CHO itself is
diminished. Experimental proof of the effect was nevertheless
obtained following synthesis of 54 from 59, itself obtained
following the general sequence used for 15a (Scheme 9): side-
by-side thermolysis of vinylallenol 59 and vinylallenal 54 at
90 °C led to a mixture containing a 66:34 ratio of the alcohols
E-60 and Z-60 as against an 83:17 ratio of the aldehydes E-55
and Z-55, in sound quantitative agreement with expectations
based on the computations.
Figure 6. Plot of ∆Ea versus σR° for the electrocyclic ring closures of
substituted divinylallenes to alkylidenecyclobutenes.
between the exocyclic diene and the formyl group in 50 seem
possible, since the diene is s-trans in the most stable conforma-
tions of both the (E)- and the (Z)-alkylidenecyclobutenes and
the formyl group is also s-trans throughout the reaction.
Interaction of the -CHO group with substituents or with the
forming σ bond having thus been ruled out, we surmised that
the extra torquoselectivity imparted by -CHO might have the
same origin as the long reaction times associated with the
presence of this group in 17a and 17b (Table 1): its conjugation
with the vinylallene polyene system. This hypothesis is
supported by the good correlation between TS-Z - TS-E energy
differences of 46, 50, and 52 and the values of the empirical
resonance parameter σR° for the corresponding R2 groups29
(Figure 6); similar correlation has been reported for the ring
opening of 3-substituted cyclobutenes (1) and 7-substituted
cycloocta-1,3,5-trienes (5).7 The cause of the relationship would
appear to be that, to judge from the ∆rC-C values listed in Table
5, the transition state is reached earlier during cyclization to
the (E)-alkylidenecyclobutene than during cyclization to the (Z)-
alkylidenecyclobutene; this must mean that conjugation between
the formyl group and the polyene system is maintained to a
greater extent in TS-E than in TS-Z, which must further lower
the energy of TS-E with respect to TS-Z. It may further be
noted that this mechanism appears to be modulated by the steric
effect of the substituent at C4, since ∆rC-C is greater for 46,
50, and 52 than for 48.
Experimental verification of the torquoselective effect of
-CHO in the absence of steric interaction between the C6
substituent and a bulky C4 substituent was frustrated by the
complexity of the reaction mixture obtained upon heating the
divinylallenal 17d. Computational verification was sought by
calculating transition structures for the ring closure of 4-tert-
butyl-3-methylhepta-2,4,5-trienal (54), in which the steric effect
of the C4-tert-butyl is reduced, relative to 50, by reducing the
size of the C6 substituent. The energy difference between TS-E
and TS-Z for 54 is 0.79 kcal/mol in favor of TS-E (Table 5).
We interpret this small difference as being the resultant of two
effects: (a) elimination of most of the “direct” steric effect of
the C4 substituent; and (b) the presence of a small effect due to
the formyl group. The smallness of this latter effect is
Summary and Conclusions
The thermal rearrangement of (2E,7E)-4-alkyl-3,7-dimeth-
ylocta-2,4,5,7-tetraenes (related to 11,7-retroretinoids9) to trisub-
stituted alkylidenecyclobutenes is peri-, regio-, and torquose-
lective if the alkyl substituent at C4 is large. A formyl group
at C2 has an additional torquoselective influence: in particular,
the single alkylidenecyclobutenal E-32 is obtained from divi-
nylallenals 17a and 17b. The effect of the formyl group is
ascribed to the reluctance of the starting vinylallenal to relinquish
extended π conjugation: because of this reluctance, the energy
of the transition structure leading to the (E)-alkylidenecy-
clobutenes, which ab initio calculations on 50 show to occur
earlier than the alternative leading to the (Z)-alkylidenecy-
clobutenes, is lower than that of the latter transition structure.
Experimental Section31
General Procedures:32 Computational Methods. Structures were
fully optimized at the Hartree-Fock level using the program GAUSS-
IAN 92 with the standard 6-31G* basis set. The correlated energies
of all optimized structures were then calculated at the MP2/6-31G*
(30) The value of the reaction progress4b at the transition state has been
calculated for two of the representative divinylallenes, 46 and 48. It is greater
for the latter and for the transition structures leading to the Z isomers: 46
f E-47, 54%; 46 f Z-47, 56%; 48 f E-49, 56%; 48 f Z-49, 59%. The
geometric parameters of the vinylallenes 46 and 48 and the alkylidenecy-
clobutenes E-47, Z-47, E-49, and Z-49 can be found in the supporting
information.
(31) Experimental procedures for the preparation of the vinylallenes, and
complete spectroscopic characterization of all compounds described in the
text, are provided in the supporting information. For alkylidenecyclobutenes,
selected 1H and complete 13C NMR spectral data are presented in abbreviated
form, in the Experimental Section. General experimental procedures are
also presented in the supporting information. The purities of all compounds
were evaluated by a combination of HPLC and 1H and 13C NMR analysis
before mass spectral determinations. As evidence of purity, NMR spectra
are included in the supporting information.
(29) (a) Katritzky, A. R.; Topsom, R. D. Chem. ReV. 1977, 77, 639. (b)
Ruff, F.; Csizmadia, I. G. Organic Reactions. Equilibria, Kinetics and
Mechanism. Studies in Organic Chemistry 50; Elsevier: Amsterdam, 1994;
Chapter 7, pp 161-209.
(32) Description of general procedures: de Lera, A. R.; Iglesias, B.;
Rodr´ıguez, J.; Alvarez, R.; Lo´pez, S.; Villanueva, X.; Padro´s, E. J. Am.
Chem. Soc. 1995, 117, 8220.

1890 J. Am. Chem. Soc., Vol. 118, No. 8, 1996
Lo´pez et al.
1
level. In each case, a set of possible rotamers was explored. Transition
structures were located and are characterized by a single negative
eigenvalue. Complete structural details in the form of Cartesian
coordinates for the RHF/6-31G*-optimized structures and transition
structures can be found in the supporting information.
15.0, 14.1, -5.0 (2×). Minor isomer (Z-30b): H NMR (CDCl3) δ
5.21 (1H, t, J ) 6.6 Hz), 5.17 (1H, s), 3.72 (1H, dd, J ) 10.0, 6.1 Hz),
3.59 (1H, dd, J ) 10.0, 7.1 Hz), 2.89 (1H, dd, J ) 7.1, 6.1 Hz), 2.75
(2H, d, J ) 6.6 Hz), 2.7-2.6 (1H, m), 1.94 (3H, s), 1.77 (3H, s), 1.06
(6H, d, J ) 7.0 Hz); 13C NMR (CDCl3) δ 148.9, 147.2, 138.4, 136.4,
130.6, 128.8, 127.7, 116.2, 65.4, 51.2, 39.9, 34.9, 32.9, 28.2 (2×), 27.9,
27.6, 25.8 (tBu), 21.4, 21.1, 19.7, 19.6, 18.1, 16.9, 14.7, -5.4 (2×).
Reaction of Vinylallenol 16b (Table 1). Purification by flash
chromatography (83:17 hexane/ethyl acetate) afforded compound 31b
(84% yield) as a mixture of isomers (E-31b/Z-31b, 82:18 ratio) which
were separated by HPLC. Major isomer (E-31b): 1H NMR (CDCl3)
δ 5.59 (1H, s), 5.26 (1H, t, J ) 6.3 Hz), 3.84 (2H, br), 3.44 (1H, br),
2.77 (2H, d, J ) 6.3 Hz), 2.6-2.5 (1H, m), 1.90 (3H, s), 1.84 (3H, s),
1.14 (6H, d, J ) 7.1 Hz); 13C NMR (CDCl3) δ 150.9, 142.8, 136.6
(2×), 131.3, 131.0, 127.8, 115.5, 62.1, 52.1, 39.8, 34.9, 32.8, 28.2 (2×),
27.6, 26.2, 21.1, 21.0, 19.6, 19.5, 15.2, 13.0. Minor isomer (Z-31b):
1H NMR (C6D6) δ 5.57 (1H, t, J ) 7.0 Hz), 5.45 (1H, s), 3.7-3.6
(2H, m), 3.0-2.8 (3H, m), 1.87 (3H, s), 1.84 (3H, s), 1.10 (6H, d, J )
7.0 Hz); 13C NMR (CDCl3) δ 150.1, 143.4, 136.6 (2×), 130.1, 127.9,
121.0, 110.5, 62.6, 52.5, 39.9, 34.9, 32.9, 31.5, 28.3 (2×), 27.7, 22.5,
20.9, 19.6, 19.5, 14.1, 13.9.
General Procedure for the Thermal Electrocyclic Reactions. A
solution of the vinylallene (∼10-4 M) in C6D6 (1.1 mL) was introduced
into a thick-walled NMR tube (rinsed with base, water, and acetone
and dried in an oven at 140 °C), and the tube was then flame-sealed
under argon, submerged in a thermostated oil bath for a measured time,
and rapidly cooled in a dry ice bath. The relative molar fractions of
starting material and product were determined by integration of the 1H
NMR signals attributed to the allenic and exocyclic vinyl protons,
respectively. The NMR tube was then broken open, contents were
transferred to a round-bottomed flask containing ether, and the solvents
were removed. The resulting residue was purified by flash chroma-
tography (to isolate the mixture of isomers for which the yield is given)
1
and/or HPLC (using the same eluant); isomer ratios obtained by H
NMR integration were confirmed by weight when the isomers were
stable. In order to rule out the possibility of acid-induced rearrange-
ments, selected divinylallenes (15a, 16a, 17a) were treated with Cl3-
Fe‚SiO2 or p-TsOH in C6D6 solution for 24 h at room temperature;
p-TsOH deprotected silyl ether 15a, and the remaining experiments
led to unaffected starting material.
Reaction of Vinylallenal 17b (Table 1). Purification by HPLC
(98:2 hexane/ethyl acetate) afforded the very unstable compound E-32b
as a single isomer (>99:1 ratio) in 79% yield: 1H NMR (CDCl3) δ
9.16 (1H, d, J ) 6.1 Hz), 5.88 (1H, s), 5.32 (1H, t, J ) 6.4 Hz), 3.84
(1H, d, J ) 6.1 Hz), 2.79 (2H, d, J ) 6.4 Hz), 2.7-2.5 (1H, m), 1.88
(3H, s), 1.69 (3H, s), 1.19 (6H, d, J ) 7.0 Hz); 13C NMR (CDCl3) δ
202.0, 155.9, 137.4, 136.4, 134.2, 131.2, 130.8, 128.2, 119.8, 63.1,
39.9, 34.9, 32.9, 28.3 (2×), 27.6, 26.7, 21.1, 20.9, 19.7, 19.6, 14.3,
14.0.
Reaction of Vinylallene 15a (Table 1). Purification by flash
chromatography (hexane) afforded compound 30a (85% yield) as a
mixture of isomers (E-30a/Z-30a, 83:17 ratio) which were separated
by HPLC. Major isomer (E-30a): 1H NMR (CDCl3) δ 5.64 (1H, s),
5.21 (1H, t, J ) 6.4 Hz), 4.07 (1H, dd, J ) 10.0, 3.8 Hz), 3.51 (1H,
dd, J ) 10.0, 8.5 Hz), 3.26 (1H, dd, J ) 8.5, 3.8 Hz), 2.79 (2H, d, J
) 6.4 Hz), 2.02 (3H, s), 1.85 (3H, s), 1.19 (9H, s); 13C NMR (CDCl3)
δ 150.7, 145.7, 137.8, 136.8, 131.5, 130.7, 127.7, 116.0, 66.2, 51.2,
39.8, 34.9, 33.9, 32.9, 29.2 (tBu), 28.3 (2×), 27.7, 25.9 (tBu), 19.8,
Reaction of Vinylallene 15c (Table 1). Purification by flash
chromatography (hexane) afforded compound 30c (75% yield) as a
mixture of isomers (E-30c/Z-30c, 66:34 ratio) which were separated
1
1
19.6, 18.2, 15.3, 15.0, -5.4 (2×). Minor isomer (Z-30a): H NMR
by HPLC. Major isomer (E-30c): H NMR (CDCl3) δ 5.48 (1H, s),
(CDCl3) δ 5.21 (1H, t, J ) 6.3 Hz), 5.14 (1H, s), 3.72 (1H, dd, J )
10.1, 5.8 Hz), 3.60 (1H, dd, J ) 10.1, 6.7 Hz), 2.81 (1H, dd, J ) 6.7,
5.8 Hz), 2.73 (2H, d, J ) 6.3 Hz), 1.95 (3H, s), 1.71 (3H, s), 1.12 (9H,
s); 13C NMR (CDCl3) δ 151.0, 146.0, 137.4, 136.1, 132.0, 129.0, 127.7,
115.9, 65.3, 50.7, 39.8, 34.9, 33.2, 32.9, 29.3 (tBu), 28.4 (2×), 27.5,
25.9 (tBu), 19.9, 19.5, 18.2, 17.4, 15.0, -5.4 (2×).
5.22 (1H, t, J ) 6.4 Hz), 4.11 (1H, dd, J ) 9.8, 3.9 Hz), 3.54 (1H, dd,
J ) 9.8, 9.0 Hz), 3.39 (1H, dd, J ) 9.0, 3.9 Hz), 2.78 (2H, d, J ) 6.4
Hz), 2.07 (2H, q, J ) 7.6 Hz), 1.91 (3H, s), 1.84 (3H, s), 1.06 (3H, t,
J ) 7.6 Hz); 13C NMR (CDCl3) δ 147.2, 145.9, 139.3, 136.8, 131.4,
130.9, 127.8, 114.2, 66.3, 52.6, 39.9, 35.0, 32.9, 28.3 (2×), 27.7, 25.9
(tBu), 19.7, 19.6, 18.2, 17.8, 15.0, 13.5, 12.2, -5.4 (2×). Minor isomer
1
(Z-30c): H NMR (CDCl3) δ 5.24 (1H, t, J ) 6.8 Hz), 5.24 (1H, s),
Reaction of Vinylallenol 16a (Table 1). Purification by flash
chromatography (83:17 hexane/ethyl acetate) afforded compound 31a
(89% yield) as a mixture of isomers (E-31a/Z-31a, 86:14 ratio) which
were separated by HPLC. Major isomer (E-31a): 1H NMR (CDCl3)
δ 5.70 (1H, s), 5.28 (1H, t, J ) 6.2 Hz), 3.8-3.7 (2H, m), 3.35 (1H,
br), 2.77 (2H, d, J ) 6.2 Hz), 1.95 (3H, s), 1.82 (3H, d, J ) 0.9 Hz),
1.20 (9H, s); 13C NMR (CDCl3) δ 152.4, 142.5, 136.6, 136.5, 131.3,
131.1, 127.9, 116.8, 61.9, 51.2, 39.8, 34.9, 34.1, 32.9, 29.2 (tBu), 28.3
3.77 (1H, dd, J ) 10.0, 6.0 Hz), 3.60 (1H, dd, J ) 10.0, 7.5 Hz), 2.98
(1H, dd, J ) 7.5, 6.0 Hz), 2.77 (2H, d, J ) 6.8 Hz), 2.2-2.1 (2H, m),
1.88 (3H, s), 1.81 (3H, s), 1.01 (3H, t, J ) 7.3 Hz); 13C NMR (CDCl3)
δ 149.0, 145.0, 138.8, 136.6, 131.0, 129.6, 127.9, 116.8, 65.7, 51.6,
39.9, 35.1, 32.9, 28.4 (2×), 27.7, 25.9 (tBu), 20.7, 19.8, 19.6, 18.3,
16.4, 13.3, 13.0, -5.4 (2×).
Reaction of Vinylallenol 16c (Table 1). Purification by flash
chromatography (83:17 hexane/ethyl acetate) afforded compound 31c
(79% yield) as a mixture of isomers (E-31c/Z-31c, 60:40 ratio) which
were separated by HPLC. Major isomer (E-31c): 1H NMR (CDCl3)
δ 5.56 (1H, s), 5.27 (1H, t, J ) 6.4 Hz), 3.9-3.8 (2H, m), 3.49 (1H,
br), 2.78 (2H, d, J ) 6.4 Hz), 2.12 (2H, q, J ) 7.4 Hz), 1.88 (3H, s),
1.84 (3H, d, J ) 1.0 Hz), 1.09 (3H, t, J ) 7.4 Hz); 13C NMR (CDCl3)
δ 147.3, 144.1, 137.9, 136.7, 131.5, 131.0, 127.9, 115.1, 62.2, 52.6,
39.9, 34.9, 32.9, 28.3 (2×), 27.6, 19.7, 19.6, 17.9, 15.2, 12.5, 12.3.
Minor isomer (Z-31c): 1H NMR (CDCl3) δ 5.30 (1H, s), 5.27 (1H, t,
J ) 6.2 Hz), 3.79 (1H, dd, J ) 11.0, 4.7 Hz), 3.7-3.6 (1H, m), 3.05
(1H, br), 2.78 (2H, d, J ) 6.2 Hz), 2.21 (2H, q, J ) 7.7 Hz), 1.87 (3H,
s), 1.83 (3H, s), 1.02 (3H, t, J ) 7.7 Hz); 13C NMR (CDCl3) δ 147.2,
146.2, 137.4, 136.4, 130.6, 131.2, 127.9, 117.5, 62.4, 51.7, 39.9, 34.2,
32.9, 28.3 (2×), 27.7, 20.7, 19.8, 19.6, 16.4, 13.0, 12.6.
1
(2×), 27.7, 19.7, 19.5, 15.6, 13.8. Minor isomer (Z-31a): H NMR
(CDCl3) δ 5.3-5.2 (2H, m), 3.79 (1H, dd, J ) 10.8, 4.2 Hz), 3.7-3.6
(1H, m), 2.9-2.8 (1H, m), 2.8-2.7 (2H, m), 1.95 (3H, s), 1.73 (3H,
s), 1.15 (9H, s); 13C NMR (CDCl3) δ 152.4, 144.5, 136.3, 136.0, 131.6,
129.5, 127.9, 116.3, 62.1, 50.8, 39.8, 34.9, 33.4, 32.9, 29.3 (tBu), 28.4
(2×), 27.5, 19.9, 19.5, 17.4, 14.3.
Reaction of Vinylallenal 17a (Table 1). Purification by HPLC
(98:2 hexane/ethyl acetate) afforded compound E-32a as a single isomer
(> 99:1 ratio) in 92% yield: 1H NMR (CDCl3) δ 9.12 (1H, d, J ) 6.1
Hz), 6.00 (1H, s), 5.30 (1H, t, J ) 6.6 Hz), 3.76 (1H, d, J ) 6.1 Hz),
2.77 (2H, d, J ) 6.6 Hz), 1.92 (3H, s), 1.68 (3H, d, J ) 1.0 Hz), 1.24
(9H, s); 13C NMR (CDCl3) δ 202.2, 157.6, 136.5, 136.3, 134.5, 131.3,
130.1, 128.1, 121.1, 62.6, 39.7, 34.9, 34.3, 32.8, 29.1 (tBu), 28.2 (2×),
27.5, 19.7, 19.5, 14.8, 14.4.
Reaction of Vinylallenal 17c (Table 1). Purification by HPLC (98:2
hexane/ethyl acetate) afforded, among other products, the highly
unstable compound E-32c in 5% yield: 1H NMR (C6D6) δ 9.32 (1H,
d, J ) 6.0 Hz), 5.95 (1H, s), 5.55 (1H, t, J ) 6.5 Hz), 3.90 (1H, d, J
) 6.0 Hz), 2.87 (2H, d, J ) 6.5 Hz), 2.0-1.8 (4H, m), 1.86 (3H, d, J
) 1.0 Hz), 0.96 (3H, t, J ) 7.1 Hz).
Reaction of Vinylallene 15d (Table 1). Purification by HPLC
(hexane) afforded the unstable compound 30d (53% yield) as a mixture
of isomers (E-30d/Z-30d, 50:50 ratio). E-30d: 1H NMR (C6D6) δ
5.74 (1H, s), 5.54 (1H, t, J ) 6.4 Hz), 4.28 (1H, dd, J ) 10.0, 4.1 Hz),
Reaction of Vinylallene 15b (Table 1). Purification by flash
chromatography (hexane) afforded compound 30b (81% yield) as a
mixture of isomers (E-30b/Z-30b, 80:20 ratio) which were separated
by HPLC. Major isomer (E-30b): 1H NMR (CDCl3) δ 5.52 (1H, s),
5.21 (1H, t, J ) 6.2 Hz), 4.09 (1H, dd, J ) 10.0, 3.9 Hz), 3.53 (1H,
dd, J ) 10.0, 8.4 Hz), 3.34 (1H, dd, J ) 8.4, 3.9 Hz), 2.77 (2H, d, J
) 6.2 Hz), 2.5-2.4 (1H, m), 1.94 (3H, s), 1.84 (3H, s), 1.12 (3H, d,
J ) 7.0 Hz), 1.11 (3H, d, J ) 7.0 Hz); 13C NMR (CDCl3) δ 149.4,
145.9, 138.4, 136.8, 131.4, 130.8, 127.7, 114.6, 66.2, 52.1, 39.9, 34.9,
32.9, 30.7 (2×), 28.3, 27.6, 26.1, 25.9 (tBu), 20.9, 19.7, 19.6, 18.2,

Thermal Electrocyclic Ring Closure of Vinylallenes
J. Am. Chem. Soc., Vol. 118, No. 8, 1996 1891
3.78 (1H, dd, J ) 10.0, 8.5 Hz), 3.59 (1H, dd, J ) 8.5, 4.1 Hz), 2.97
(2H, d, J ) 6.4 Hz), 2.00 (3H, d, J ) 0.9 Hz), 1.94 (3H, s), 1.67 (3H,
Hz), 2.34 (3H, s), 2.25 (1H, dd, J ) 14.2, 10.3 Hz), 1.95 (3H, s), 1.18
(9H, s); 13C NMR (CDCl3) δ 190.8, 156.5, 154.2, 151.3, 135.0, 129.0,
127.2, 116.5, 112.9, 48.9, 39.9, 34.9, 33.8, 32.8, 31.7, 29.8, 29.0 (tBu),
28.3, 20.9, 19.4, 16.7, 15.6. Minor isomer (Z-35): 1H NMR (CDCl3)
δ 9.98 (1H, d, J ) 8.4 Hz), 6.05 (1H, d, J ) 8.4 Hz), 5.23 (1H, s),
3.01 (1H, dd, J ) 8.3, 7.6 Hz), 2.3-2.2 (2H, m), 2.20 (3H, s), 1.95
(3H, s), 1.11 (9H, s); 13C NMR (CDCl3) δ 190.9, 159.8, 154.7, 152.9,
145.6, 143.5, 129.2, 120.8, 113.6, 48.2, 40.2, 34.9, 33.8, 32.9, 30.9,
29.4 (tBu), 21.3, 19.4, 19.4, 18.5, 15.0 (2×).
1
s), 1.60 (3H, d, J ) 1.2 Hz). Z-30d: H NMR (C6D6) δ 5.66 (1H, s),
5.60 (1H, t, J ) 6.3 Hz), 3.93 (1H, dd, J ) 9.8, 5.8 Hz), 3.78 (1H, dd,
J ) 9.8, 7.5 Hz), 3.27 (1H, br), 2.96 (2H, d, J ) 6.3 Hz), 1.98 (3H, s),
1.85 (3H, s), 1.79 (3H, s), 1.67 (3H, s).
Reaction of Vinylallenol 16d (Table 1). Purification by HPLC
(83:17 hexane/ethyl acetate) afforded the unstable compound 31d (67%
yield) as a mixture of isomers (E-31d/Z-31d, 58:42 ratio). Major
isomer (E-31d): 1H NMR (C6D6) δ 5.73 (1H, s), 5.55 (1H, t, J ) 6.4
Hz), 3.91 (1H, dd, J ) 10.8, 3.6 Hz), 3.80 (1H, dd, J ) 10.8, 5.9 Hz),
3.40 (1H, br), 2.96 (2H, d, J ) 6.4 Hz), 1.94 (3H, d, J ) 1.0 Hz), 1.74
(3H, s), 1.68 (3H, s). Minor isomer (Z-31d): 1H NMR (C6D6) δ 5.61
(1H, s), 5.57 (1H, t, J ) 6.6 Hz), 3.68 (2H, d, J ) 5.5 Hz), 3.06 (1H,
br), 2.96 (2H, d, J ) 6.6 Hz), 1.95 (3H, d, J ) 1.2 Hz), 1.75 (3H, d,
J ) 1.0 Hz), 1.71 (3H, s).
Reaction of Vinylallenal 17d (Table 1). Purification by HPLC
(96:4 hexane/ethyl acetate) afforded, together with a major component
of complex, still undetermined structure, the highly unstable compound
E-32d in 5% yield: 1H NMR (C6D6) δ 9.38 (1H, d, J ) 5.0 Hz), 5.93
(1H, s), 5.59 (1H, t, J ) 6.2 Hz), 3.97 (1H, d, J ) 5.0 Hz), 2.93 (2H,
d, J ) 6.2 Hz), 1.91 (3H, s), 1.67 (3H, s), 1.65 (3H, s).
Reaction of Vinylallenol 59. The sample was maintained at 90 (
0.1 °C during the entire experiment (k ) 1.15 × 10-1 h-1; τ1/2 ) 6.0
h). The reaction mixture was evaporated, and the residue was purified
by flash chromatography (85:15 hexane/ethyl acetate) to afford
compound 60 (91% yield) as a mixture of isomers (E-60/Z-60, 66:34
ratio), which were separated by HPLC. Major isomer (E-60): 1H
NMR (CDCl3): δ 5.12 (1H, q, J ) 6.8 Hz), 3.81 (2H, d, J ) 3.8 Hz),
3.13 (1H, br), 1.92 (3H, s), 1.64 (3H, d, J ) 6.8 Hz), 1.16 (9H, s); 13
C
NMR (CDCl3) δ 151.8, 140.3, 138.8, 106.1, 62.3, 50.2, 34.1, 29.3 (tBu),
13.7. Minor isomer (Z-60): 1H NMR (CDCl3) δ 4.77 (1H, q, J ) 7.2
Hz), 3.78 (1H, dd, J ) 10.9, 4.3 Hz), 3.66 (1H, dd, J ) 10.9, 4.3 Hz),
2.87 (1H, t, J ) 4.3 Hz), 1.94 (3H, s), 1.83 (3H, d, J ) 7.2 Hz), 1.22
(9H, s); 13C NMR (CDCl3) δ 152.0, 143.1, 138.5, 104.9, 62.2, 50.8,
33.2, 29.6 (tBu), 15.5, 14.0.
Reaction of Vinylallene 21 (Table 1). Purification by HPLC (83:
17 hexane/ethyl acetate) afforded the unstable compound 36 (80% yield)
as a mixture of isomers (E-36/Z-36, 91:9 ratio) which were separated
Reaction of Vinylallenal 54. The sample was maintained at 90 (
0.1 °C during the entire experiment (k ) 3.7 × 10-2 h-1; τ1/2 ) 18.5
h). The solvent was evaporated, and the residue was purified by column
chromatography (98:2 hexane/ethyl acetate) to afford compound 55
(90% yield) as a mixture of isomers (E-55/Z-55, 83:17 ratio), which
were separated by HPLC. Major isomer (E-55): 1H NMR (CDCl3) δ
9.15 (1H, d, J ) 5.8 Hz), 5.36 (1H, q, J ) 6.6 Hz), 3.58 (1H, d, J )
5.8 Hz), 1.93 (3H, s), 1.55 (3H, d, J ) 6.6 Hz), 1.20 (9H, s); 13C NMR
(CDCl3) δ 202.5, 156.0, 134.9, 133.0, 110.5, 60.6, 34.1, 28.9 (tBu),
14.8, 12.9. Minor isomer (Z-55): 1H NMR (CDCl3) δ 9.04 (1H, d, J
) 5.8 Hz), 4.72 (1H, q, J ) 7.2 Hz), 3.37 (1H, d, J ) 5.8 Hz), 1.98
(3H, s), 1.87 (3H, d, J) 7.2 Hz), 1.21 (9H, s); 13C NMR (CDCl3) δ
202.2, 161.7, 129.6, 128.0, 111.4, 62.0, 29.6, 29.4 (tBu), 22.6, 14.0.
1
by HPLC. Major isomer (E-36): H NMR (CDCl3) δ 5.75 (1H, s),
5.20 (1H, t, J ) 6.4 Hz), 2.89 (2H, s), 2.80 (2H, d, J ) 6.4 Hz), 1.96
(3H, s), 1.82 (3H, s), 1.24 (9H, s); 13C NMR (C6D6) δ 151.1, 140.7,
137.1, 136.9, 133.6, 131.0, 128.0, 116.8, 40.2, 39.5, 35.2, 34.1, 33.1,
29.3 (tBu), 28.5 (2×), 27.8, 19.9, 19.8, 15.6, 14.2. Minor isomer (Z-
1
36): H NMR (CDCl3) δ 5.24 (1H, t, J ) 6.5 Hz), 5.06 (1H, s), 2.74
(2H, d, J ) 6.5 Hz), 2.51 (2H, br), 1.97 (3H, s), 1.73 (3H, s); 13C
NMR (C6D6) δ 150.9, 142.9, 136.5, 135.1, 132.6, 129.6, 128.0, 117.5,
40.2, 39.1, 35.1, 33.2, 30.3, 29.5 (tBu), 28.6 (2×), 27.9, 20.1, 19.9,
17.6, 16.2.
Reaction of Vinylallenol 28. The sample was maintained at 100
( 0.1 °C during the entire experiment. This reaction exhibited non-
first-order kinetics; the isomer ratio is given for a reaction time of 19
h. The solvent was removed, and the residue was purified by flash
chromatography (hexane) to afford starting 28 and compound 34 in
70% joint yield. Purification by HPLC afforded a mixture of
Acknowledgment. We thank the DGICYT (Grant PM91-
0085), FIS (Contract 95/1534), and Xunta de Galicia (Grant
XUGA20904B95; fellowship to J.G.R.) for generous financial
support; Prof. J. M. Aurrecoechea for helpful discussions; and
CESGA (Centro de Supercomputacio´n de Galicia) for generous
allocation of time on the Fujitsu VP2400 computer. We
dedicate this paper to Professor William H. Okamura in honor
of his 55th birthday.
1
compounds E-34 and Z-34 in 67:33 ratio. Major isomer (E-34): H
NMR (CDCl3) δ 5.58 (1H, t, J ) 7.1 Hz), 5.58 (1H, s), 4.24 (2H, d,
J ) 7.1 Hz), 3.43 (1H, dd, J ) 10.6, 6.0 Hz), 2.51 (1H, dd, J ) 14.2,
6.0 Hz), 2.24 (1H, dd, J ) 14.2, 10.6 Hz), 1.90 (3H, s), 1.89 (3H, s),
1.17 (9H, s); 13C NMR (CDCl3) δ 150.2, 148.6, 146.0, 136.9, 135.6,
128.4, 126.3, 113.1, 59.6, 48.4, 40.0, 34.8, 33.9, 32.9, 32.1, 29.9, 29.1
1
(tBu), 28.3, 20.8, 19.4, 15.7, 15.1. Minor isomer (Z-34): H NMR
Supporting Information Available: General experimental
procedures, preparation of substrates, complete spectroscopic
characterizations, kinetic analyses, 1H NMR or 13C NMR spectra
for the compounds described in the text, and RHF/6-31G*
geometries of transition structures and products in the form of
cartesian coordinates (116 pages). This material is contained
in many libraries on microfiche, immediately follows this article
in the microfilm version of the journal, can be ordered from
the ACS, and can be downloaded from the Internet; see any
current masthead page for ordering information and Internet
access instructions.
(CDCl3) δ 5.58 (1H, t, J ) 7.1 Hz), 5.12 (1H, s), 4.20 (2H, br), 2.96
(1H, t, J ) 7.1 Hz), 2.3-2.2 (2H, m), 1.90 (3H, s), 1.72 (3H, s), 1.12
(9H, s); 13C NMR (CDCl3) δ 149.4, 149.1, 142.2, 138.4, 135.9, 128.1,
126.7, 114.3, 59.7, 48.1, 40.4, 34.9, 33.0, 33.0, 31.1, 29.4 (tBu), 29.4,
21.3, 19.5, 17.6, 14.1, 14.0.
Reaction of Vinylallenal 29. The sample was maintained at 80 (
0.1 °C during the entire experiment. This reaction exhibited non-first-
order kinetics; the isomer ratio is given for a reaction time of 17 h.
The solvent was evaporated under vacuum, and the residue was purified
by HPLC (96:4 hexane/ethyl acetate) to afford, in 63% yield, starting
29 and compound 35 in 69:31 ratio. Major isomer (E-35): 1H NMR
(CDCl3) δ 10.01 (1H, d, J ) 8.2 Hz), 5.93 (1H, d, J ) 8.2 Hz), 5.70
(1H, s), 3.50 (1H, dd, J ) 10.3, 8.5 Hz), 2.52 (1H, dd, J ) 14.2, 6.6
JA9522241