Conformational restraint in thermal rearrangements of a cyclobutane: 3,4-dicyanotricyclo[4.2.2.02,5]decane

Article abstract of DOI:10.1021/ja030050e

There being no study of a cyclobutane so fused to another cyclic system that an antiperiplanar conformation of the related diradical be precluded, the system in the title has been synthesized and studied for its thermal behavior. In comparison to the thermal behavior of unconstrained 1,2-dicyanocyclobutane in stereomutation and fragmentation to acrylonitrile, the constrained system shows a ～10-fold higher ratio of stereomutation to fragmentation to the three cis-1,4-bis-β-cyanovinylcyclohexanes. In these diolefins, a stereochemical correlation between the two olefinic fragmentation products is preserved. Revealed in the thermal rearrangement of isomer trans-1 is a surprising excess (78%) of cis-1-cis-β-cyanovinyl-4-transβ -cyanovinyl-cyclohexane, cis, trans-2, the result of zero internal rotations within the diradical-in-caldera prior to fragmentation (retention of configuration). Similarly, to a comparably striking extent, anti,cis-1 gives trans,trans-2 as its major product (71%), again by zero internal rotations.

Full text of DOI:10.1021/ja030050e

Published on Web 08/08/2003
Conformational Restraint in Thermal Rearrangements of a
Cyclobutane: 3,4-Dicyanotricyclo[4.2.2.02,5]decane
W. von E. Doering* and JoAnn Peters DeLuca
Contribution from the Department of Chemistry and Chemical Biology, HarVard UniVersity,
Cambridge, Massachusetts 02138-2902
Received January 27, 2003; E-mail: doering@chemistry.harvard.edu
Abstract: There being no study of a cyclobutane so fused to another cyclic system that an antiperiplanar
conformation of the related diradical be precluded, the system in the title has been synthesized and studied
for its thermal behavior. In comparison to the thermal behavior of unconstrained 1,2-dicyanocyclobutane
in stereomutation and fragmentation to acrylonitrile, the constrained system shows a ∼10-fold higher ratio
of stereomutation to fragmentation to the three cis-1,4-bis-â-cyanovinylcyclohexanes. In these diolefins, a
stereochemical correlation between the two olefinic fragmentation products is preserved. Revealed in the
thermal rearrangement of isomer trans-1 is a surprising excess (78%) of cis-1-cis-â-cyanovinyl-4-trans-
â-cyanovinylcyclohexane, cis,trans-2, the result of zero internal rotations within the diradical-in-caldera
prior to fragmentation (retention of configuration). Similarly, to a comparably striking extent, anti,cis-1 gives
trans,trans-2 as its major product (71%), again by zero internal rotations.
anti,cis-1 and trans-1 in Figure 1.6 Although this system is more
Introduction
flexible, and therefore less desirable, considerably less strain is
A new factor that may influence distribution among products
from not-obviously concerted, purely thermal rearrangements
of cyclobutanes is explored. A prior history can be reconstructed
from references given in a recent contribution.1 The possibility
that fragmentation of a cyclobutane to two olefins might require,
or at least be favored by, an antiperiplanar conformation in the
intermediate diradical (the “caldera”) has been suggested by
Dewar et al.2 and Halevi.3 More recent theoretical work of
Doubleday on cyclobutane has refuted this suggestion that an
antiperiplanar conformation be required but still finds a small
barrier to cleavage from synclinal conformations, in contrast to
no barrier from the antiperiplanar.4
placed on the 2,5-bond. While a dihedral angle of as much as
50° can be entertained in the related diradical resulting from
cleavage of the 3,4-bond, a 180° twist to an antiperiplanar
conformation is still precluded. The two cyano groups are
introduced because their radical-stabilizing ability of ∼8-11
kcal mol-1 each promises substantial control over the identity
of the bond being broken. They also minimize steric repulsions
and otherwise unwelcome competing reactions that often
accompany the use of vinyl, for example, as a stabilizing group.
A further important advantage is the existence for comparison
of the thorough kinetic studies of trans- and cis-1,2-dicyano-
cyclobutane, the parent, reference compounds (trans- and cis-
3; Scheme 1).7
The specific system studied in this contribution is a cyclobu-
tane so constructed as to preclude access to antiperiplanar
conformations of the 1,4-diradical-like intermediates, yet flexible
enough to allow access to synclinal conformations. The device
selected involves fusion of a bicyclic system to the cyclobutane
ring. Although a bicyclo[2.2.1]heptane system at first had
seemed ideal [indeed trans-5 (see Scheme 5 below) is known],5
subsequent worries about the influence of the strain in that
system (15 kcal mol-1) on the identity of initial bond-cleavage
(2,5- versus 3,4-) prompted the selection of a bicyclo[2.2.2]-
octane system (8 kcal mol-1 of strain) as represented by
Results
Preparation of anti,cis-1 and trans-1 was uncomplicated but
for our failure to find conditions for the transformation of the
cis-diacid into the cis-diamide (Scheme 2). This inconvenience
required two epimerizations instead of one in the final procedure,
as shown. No significant amount of the third isomer, syn,cis-1,
or its precursors appeared at any time, consistent with a higher
MM2-calculated steric energy of 56.3 kcal mol-1 compared to
that of trans-1 (50.9 kcal mol-1) and anti,cis-1 (52.4 kcal
mol-1).
(1) Doering, W. v. E.; Cheng, X-h.; Lee, Y-h.; Lin, T-z. J. Am. Chem. Soc.
2002, 124, 11642-11652.
The three products of fragmentation (2 in Figure 1) were
obtained for purposes of assignment of structure by heating a
sample of trans-1 at 299 °C for ∼16 h. Although trans,trans-2
(2) Dewar, M. J. S.; Krishna, S.; Calmer, H. W. J. Am. Chem. Soc. 1974, 96,
5240-5242; Dewar, M. J. S.; Kirschner, S.; Kollmar, H. W.; Wade, L. J.
Am. Chem. Soc. 1974, 96, 5242-5244; Dewar, M. J. S.; Kirschner, S. J.
Am. Chem. Soc. 1974, 96, 5246-5248.
(3) Halevi, E. A. HelV. Chim. Acta 1975, 58, 2136-2150.
(4) Doubleday, C. H., Jr. J. Phys. Chem. 1996, 100, 15083-15086.
(5) Tabushi, I.; Yamamura, K.; Yoshida, Z. J. Am. Chem. Soc. 1972, 94, 787-
792.
(6) DeLuca, JoAnn Peters, Ph.D. Dissertation, Harvard University, 1986; Diss.
Abstr. Int. B 1987, 47, 2434.
(7) Guyton, C. A., Ph.D. Dissertation, Harvard University, 1980; Diss. Abstr.
Int. B 1980, 41, 1370; pp 153-185.
9
10608
J. AM. CHEM. SOC. 2003, 125, 10608-10614
10.1021/ja030050e CCC: $25.00 © 2003 American Chemical Society

Thermal Rearrangements of a Cyclobutane
A R T I C L E S
Scheme 2
Scheme 3
Figure 1. Thermal interconversion of anti,cis-1 and trans-1 and the three
products of their fragmentation, trans,trans-2, cis,trans-2, and cis,cis-2, are
shown along with their specific rate constants (all k × 10-6 s-1) at 310.6
°C in tert-butylbenzene. Relative MM2 steric energies in twist-boat
conformations are also indicated (StEn in kcal mol-1). (Note the omission
of syn,cis-1, consistent with its higher StEn of 56.3 kcal mol-1). The
minimum number of internal rotations (rot.) required by each path is also
given.
Scheme 1
Although the expected “irreversibility” of the product dienes
was confirmed by the failure to detect any 1 upon heating a
9.9:1 mixture of cis,trans- and cis,cis-2 at 299 °C, trans,trans-2
appeared at a slow rate: 4.5% after 5.6 h (k ) ∼2 × 10-6 s-1),
by which time the ratio of cis,trans-2 to cis,cis-2 had decreased
to 4.7:1. If the dimerization of unsubstituted acrylonitrile in the
temperature range 150-170 °C be relevant, this slow intercon-
version among the dienes likely occurs by way of reentry into
the caldera of diradicals as a first step in an otherwise
thermodynamically thwarted cyclodimerization.
For the kinetic study, thermal rearrangements of trans-1 and
anti,cis-1 are run to low conversion (15-20%) in tert-
butylbenzene over the temperature range 256-311 °C. The data
are given in Supporting Information, Table SI-1. Specific rate
constants are evaluated by extrapolation to zero time and also
by application of a multivariable program by courtesy of the
late Professor Wolfgang R. Roth, the latter results being given
in Table 1.
was cleanly separated by HPLC, the other two stereoisomers
were obtained as a mixture in a ratio high enough (1:9) to permit
confident identification of the olefinic-hydrogen portion of their
NMR spectra. Geometrical assignments were made by analogy
with the spectrum of acrylonitrile [5.64, (m, J1,2 ) 17.0 Hz,
H-1); 6.23, (m, J1,2 ) 17.0 Hz, J2,3 ) 3.5 Hz, H-2 [trans to
H-1]); 6.05, (m, J1,3 ) 11.5 Hz, H-3 [cis to H-1])]. Details are
given in the Experimental Section and in the dissertation of
DeLuca.6 A compound identical to trans,trans-2 was also
obtained synthetically from a mixture of cis- and trans-1,4-bis-
hydroxymethylcyclohexane via the mixture of dialdehydes,
followed by reaction with cyanomethylenetriphenylphosphorane
(Scheme 3).
Discussion
The rates of thermal reorganizations of 1 and unsubstituted,
1,2-dicyanocyclobutane (3) are of the same order of magnitude
9
J. AM. CHEM. SOC. VOL. 125, NO. 35, 2003 10609

A R T I C L E S
Doering and DeLuca
Table 1. Specific Rate Constants for the Stereomutation of trans-1 (ktc) and anti,cis-1 (kct), and Their Fragmentation to the Mixture of
Olefins 2 (kto and kco, Respectively)
a
a
a
b
b
b
b
T,
°
C
ktc
kct
kto
ktc
kct
kto
kco
Keq
256.5
269.8
281.1
291.4
300.4
310.4
0.276 ( 0.011
0.834 ( 0.040
2.204 ( 0.094
4.559 ( 0.187
10.77 ( 0.276
21.75 ( 0.711
0.74
2.20
5.71
11.63
27.06
53.70
0.103 ( 0.010
0.256 ( 0.037
0.924 ( 0.087
1.968 ( 0.177
4.303 ( 0.260
16.55 ( 0.649
0.316 ( 0.007
0.849 ( 0.016
0.103 ( 0.067
0.001 ( 0.015
2.69
12.33 ( 0.38
28.95 ( 0.80
31.72 ( 0.62
70.00 ( 1.10
4.30 ( 0.31
16.17 ( 0.57
0.47 ( 0.56
6.27 ( 0.90
2.57
2.42
Arrhenius Parameters
log ktc
log kto
log ktc
log kct
(50000 ( 400)c
RT ln 10
(56900 ( 3400)d
51000e
RT ln 10
50000e
RT ln 10
14.55 -
14.56 -
(14.00 ( 0.33) -
(16.4 ( 1.32) -
RT ln 10
a Specific rate constants (in units of 10-6 s-1) are calculated from the data in Table SI 1 starting from trans-1 and using a simplified kinetic scheme of
four rate constants: trans-1 in reversible reactions with anti,cis-1 (ktc and kct) and the sum of all three dienes 2 (kto and k-to, the ratio k-to/kto being arbitrarily
set to 1 × 10-5; the equilibrium constants, kct/ktc, at the various temperatures being set by the linear regression line obtained from the data in column 9).
b Calculated from the data in Table SI 1 at 256.5, 300.4, and 310.4 °C starting from both of trans-1 and anti,cis-1, and using a kinetic scheme of four rate
constants involving the reversible stereomutation of trans-1 and anti,cis-1 (ktc and kct) and their irreversible fragmentation to the sum of all three dienes 2
(kto and kco, respectively). c Calculated from the values in columns 1 and 2. d Calculated from the values in columns 1 and 4. e Calculated from the values
in columns 1, 5, and 6, respectively.
Table 2. Specific Rate Constantsa for the Stereomutation of trans- and cis-1,2-Dicyanocyclobutane (trans-3 and cis-3; k1 and k2,
Respectively), and Fragmentation to Acrylonitrile (4, k3 and k4, Respectively; See Scheme 1)
T,
°
Cb
k1
k2
k3
k4
225.0 ( 1.2
246.1 ( 0.2
257.0 ( 0.2
278.8 ( 0.1
289.0 ( 1.0
0.083 ( 0.007
0.667 ( 0.039
1.49 ( 0.10
9.10 ( 0.35
22.5 ( 1.3
0.216 ( 0.011
1.48 ( 0.11
3.69 ( 0.33
22.2 ( 0.8
0.338 ( 0.051
3.07 ( 0.17
7.41 ( 0.82
43.4 ( 2.0
107 ( 8
0.654 ( 0.039
4.47 ( 0.41
10.5 ( 0.4
63.1 ( 1.9
165 ( 11.
53.1 ( 3.4
Arrhenius Parameters
log k1
log k2
log k3
log k4
(48200 ( 900)
(47600 ( 500)
(49400 ( 1000)
RT ln 10
(47600 ( 1100)
(14.1 ( 0.4) -
(14.2 ( 0.2) -
(15.2 ( 0.4) -
(14.7 ( 0.5) -
RT ln 10
RT ln 10
RT ln 10
a Specific rate constants in units of 10-6 s-1
.
b Mean T ) 257.0 °C.
(see Tables 1 and 2, 257 °C). Apparently the fusion of the
bicyclic system does not constitute a substantial perturbation
on the basic reaction. Neither is there any significant difference
in the activation parameters reported for stereomutation in Tables
1 and 2.
nar conformations is sterically prohibited, the reverse is true.
The ratio Fr/St from trans-1 has become ∼0.4 (Table 1, 256.5-
300.4 °C). This striking reversal by a factor of ∼10 is consistent
with the original stimulus for the present work. It confirms that
an antiperiplanar conformation for diradicals from 3, but not
from 1, may be ideal for processes leading to the two planar
olefinic products, since fragmentations from the antiperiplanar
geometry would require minimal geometrical adjustment beyond
lengthening of the 3,4-bond. It is also consistent with the
conclusion from Doubleday’s calculations that the exit channel
from the cyclobutane-derived caldera-defining fragmentation
from the antiperiplanar conformation is downhill with no
discernible barrier.4
Although increasing stability of the radical appears to be a
significant factor influencing the ratio, Fr/St,9 in this series, 1,2-
cyanocyclobutane occupies a position intermediate between
cyclobutane (Fr/St ≈ 3, Ea ) 61.8 kcal/mol-1) and a divinyl-
cyclobutane (Fr/St ≈ 2, Ea ) 32.5 kcal/mol-1).10 It is unlikely
that an explanation for the factor ∼10 is to be found here.
Lack of accessibility to the antiperiplanar conformations not
Discussion of fragmentation begins with a qualitative response
to its surprisingly minor role [except at the highest temperature
to be discussed below; see Table SI 1, 310.4 °C]. The
quantitative aspects are compromised to some extent by the
inability of our analysis by gas chromatography to determine
with high quantitative accuracy the very small amounts of dienes
2 produced. This limitation is particularly marked in the behavior
of anti,cis-1, where the extent of fragmentation is so small over
the lower range of temperature (256.5-300.4 °C) that no useful
information can be developed (see data in Table SI 1) but is
conservatively estimated to be less than 2% of the total reaction.
The reference compounds for comparison are the 1,2-
dicyanocyclobutanes (3), which have uninhibited access to all
conformational minima open to the intermediate diradicals prior
to fragmentation. The results of a thorough examination made
by Guyton in connection with her study of the stereochemistry
of the dimerization of acrylonitrile (4) are recorded in Table
2.8 Fragmentations (Fr) from trans-3 and cis-3 are favored over
stereomutation (St) by factors (Fr/St) of 4.07 and 3.03,
respectively. In the present example where access to antiperipla-
(8) Doering, W. v. E.; Guyton, C. A. J. Am. Chem. Soc. 1978, 100, 3229-
3230.
(9) Doering, W. v. E.; He, J.; Shao, L. J. Am. Chem. Soc. 2001, 123, 9153-
9161.
(10) Doering, W. v. E.; Ekmanis, J. L.; Belfield, K. D.; Kla¨rner, F.-G.; Krawzcyk,
B. J. Am. Chem. Soc. 2001, 123, 5532-5541.
9
10610 J. AM. CHEM. SOC. VOL. 125, NO. 35, 2003

Thermal Rearrangements of a Cyclobutane
A R T I C L E S
Scheme 4
Scheme 5
°C, 0.588 at 279 °C, and 0.580 at 289 °C, in good agreement
with the 61% result from fragmentation.7,8
In the present example, where correlation is strictly main-
tained by linkage of the two olefins of fragmentation, the major
diene formed from trans-1 over the temperature range, 256.5-
300.4 °C is cis,trans-2, the product of “zero internal rotations”
prior to fragmentation. [Strictly two internal rotations (one about
each of carbon bonds 2,3 and 4,5sor any other comparable
combination) also suffice]. At the highest temperature (310.4
°C), the ratio Fr/St increases: in trans-1 to 0.77 from 0.39 (
0.05 and in anti,cis-1 to a value of 0.09 from undetectable (not
necessarily zero). From the data in Figure 1, trans-1 is seen to
fragment to cis,trans-2 (∼78%; k4/k3 + k4 + k5), the result of
zero internal rotations prior to cleavage. The remaining dienes,
trans,trans-2 and cis,cis-2, 14% and 8%, respectively, are the
product each of a single rotation. (The process of double rotation
(two single rotations) is potentially observable as racemization
of optically active trans-1 but has not been explored in this
investigation.) The major product of fragmentation of anti,cis-1
is trans,trans-2 (71%), again the result of zero rotations. Lesser
products are cis,trans-2, the result of a single rotation (25%),
and cis,cis-2 (4%), the result of a double rotational process (two
single internal rotations). In both instances the stereochemistry
of the predominant diene has resulted from a zero rotational,
Woodward-Hoffmann forbidden process.
The experimental outcome might be thought consistent with
the incursion of a significant contribution from cleavage of the
(wrong) 2,5 bond, for which a higher activation enthalpy has
been estimated (see above). Were this process also to have
contributed to fragmentation of trans-1 in the lower temperature
range, the actual value of Fr/St in cleavage of the 3,4-bond
would become even smaller than that proclaimed above. In this
connection, a cursory investigation of the more highly strained
bicyclo[2.2.1] system, syn,cis-3,4-dicyanotricyclo[4.2.1.02,5]-
nonane (syn,cis-5), has revealed only cis,cis-6, the product of
fragmentation by a zero-rotational process (Scheme 5). Sug-
gestive of wrong-bond cleavage as the rationalization for the
favoring of zero-rotation are the activation parameters extract-
able from the data in columns 1 and 4 of Tables 1 and SI-2. In
both instances, the activation energy for fragmentation is ∼6-7
kcal mol-1 higher than that for stereomutation, while the
frequency factors are higher by ∼log A ) 2.5.12 We have no
only removes any contribution to fragmentation from this mode,
but it also removes a large segment of phase space from which
reclosure (stereomutation) is normally barred. The expected
overall effect is a decrease in the lifetime of the caldera, and a
relative favoring of stereomutation. Whatever ultimately ac-
ceptable explanation may emerge, the barring of the antiperipla-
nar conformation to 1 has led to a significant decrease in
fragmentation relative to stereomutation.
trans-1 and anti,cis-1, among the many cyclobutanes in the
literature that have been fragmented, are the first to retain
stereochemical correlation between the pairs of olefins formed
in the fragmentation. It has therefore become possible to deduce
the minimum number of internal rotations needed to describe
distribution among the three dienes of fragmentation. Cyclobu-
tane, labeled as it has been by Guyton in her work on
acrylonitrile (4) and its dimers 3, serves to illustrate the
limitations of the information available from simple cyclobu-
tanes. Fragmentation of cis-1,2-dicyano-anti,cis-3,4-dideuterio-
cyclobutane (cis-3-anti,cis-2d of Scheme 4) generates the two
acrylonitriles, trans-4 and cis-4, by way of three intermediate
diradicals formally resulting from zero, one, or two internal
rotations about C2-C3 and C4-C1 bonds. Because each
stereoisomer can result from two of these three diradicals, no
conclusions about their relative contributions can be drawn (three
variables, two observables!). The 61% retention of configuration
(trans-4) observed experimentally in the fragmentation of cis-
3-anti,cis-2d does not, for example, allow any conclusion to
be drawn about the relative amounts formed by zero and one
internal rotation.
Nonetheless, the stereochemistry of fragmentation should be
consistent with that observed in the reverse, more informative
dimerization of cis-R,â-dideuterioacrylonitrile to six 1,2,3,4-
tetradeuterio-1,2-dicyanocyclobutanes in which zero rotational
processes can and have been unequivocally established (which
see).8 From the given data, the fraction of retained configuration,
(e0 + t1)/(e0 + 2t1 + e2),11 can be calculated to be 0.607 at 257
(11) From the three equations, c + t + o ) 1, (where c, t, and o are the respective
fractions of cis-1, trans-1, and the sum of the three dienes 2 emerging
from the hypothetical caldera-in-common (example taken from Figure 1),
c/o ) k1/(k3 + k4 + k5), and t/o ) k2/(k6 + k7 + k8), values for the three
fractions can be obtained.
(12) These activation parameters are not the result of an extended, dedicated
study and should be accepted with caution.
9
J. AM. CHEM. SOC. VOL. 125, NO. 35, 2003 10611

A R T I C L E S
Doering and DeLuca
Chart 1
then leads to the two obserVable products in the ratio 13/7 (65%/
35%), while entry from anti,cis-1 produces its two obserVable
products in the ratio 80/7 (92%/8%). In other words, the ratios
of observable products are determined by differences in the
nonobserVable, identity reactions. (We reiterate that our inability
adequately to define the ratios among the three dienes 2, let
alone to find that their ratios are independent of educt as required
for full compatibility with a common intermediate, makes this
discussion in its specific terms purely hypothetical.) If the same
hypothetical analysis were to be applied to Guyton’s experiments
with the acyclic analogues, 3 and 4, the lower diagram in Figure
2 results. Herein lies a proposition that may only be resolvable
by theoretical calculation, for example a` la Doubleday.4
The major surprise revealed by these geometrically con-
strained cyclobutanes, trans-1 and anti,cis-1, in conjunction with
earlier studies of the thermal chemistry of deuterium-labeled
1,2-dicyanocyclobutanes, is a substantial disfavoring of frag-
mentation relative to stereomutation. For trans-1 (relative to
trans-3), the Fr/St ratio is diminished by a factor of about 10,
while for anti,cis-1 (relative to cis-3), the Fr/St ratio is
diminished by a factor of about 30. More questions are raised,
perhaps, than are answered. Might a look at the severely
geometrically restricted, isomeric 2,5-dicyanobicyclo[4.2.2.02,5]-
octane, in which stereomutation is barred and fragmentation
almost forced to be concerted, give useful information (Chart
1, 7)? Would examination of the rate of racemization of trans-
1, a two-rotational process overall, shed light on the extent to
which a caldera-in-common is approached (Chart 1, 8, ent-8)?
If trans-1,2-dicyano-cis-3,4-dideuteriocyclobutane (trans-3-cis-
2d) were to lead to a different composition of acrylonitriles
(trans- and cis-4) from that afforded by cis-3-anti,cis-2d, could
inferences be made about the degree to which a common steady-
state caldera had been achieved? Might comparison of the
activation parameters of the stereomutation of 2,3-dicyanobicyclo-
[2.2.2]octane and 2,3-dicyanobicyclo[2.2.1]heptane (9 and 10)
help resolve doubts about which bond is cleaved in the
generation of dienes 2 from 1? Could a look at the behavior of
the various 7,8-dicyanobicyclo[4.2.0]octanes (11) add further
insight?
Figure 2. Diagram of the fractions in percent of cis- and trans-cyclobutanes
and combined olefinic products of fragmentation (2 and 4, respectively)
emerging from the hypothetical caldera of diradical intermediates in the
thermal transformation of cis-1 and trans-1 (upper pie) and cis-3 and trans-3
(lower pie).
way of credibly estimating any difference in the relief of energy
of strain in the bicyclooctane and bicycloheptane systems
beyond noting that their respective energies of strain are 9.6
and 15.3 kcal mol-1
.
Although the favoring of zero rotation may not be associated
with confidence with the behavior of the bis-cyanomethyl
diradical prior to termination by fragmentation, the observation
brings into view the possibility that otherwise nonobservable,
zero rotational processes may be critical for a complete
description of the exit channels from calderas of diradicals.
It seems worth exploring a hypothetical situation in which
entry channels connect trans-1 or anti,cis-1 in a caldera-in-
common. In earlier thinking, for example, refs 13 and 14 it was
held that a common intermediate would require identical
distributions among the products of the exit channels. On this
basis, maximum values for participation of a caldera-in-common
were deduced. In the present example, described in Figure 1,
trans-1 undergoes stereomutation to anti,cis-1 (65%) and
fragmentation to combined dienes 2 (35%). In sharp contrast,
anti,cis-1 undergoes stereomutation to trans-1 (92%) and
fragmentation to dienes 2 to the extent of only 8%. Without
the inclusion of the identity reactions (e.g., trans-1 returning
to trans-1), this system would appear to be far from having a
single caldera common to the two educts.
Experimental Section
General Methods. NMR spectra were recorded on a JEOL-FX270
(270 MHz) or a Varian CFT20 (80 MHz) spectrometer in CDCl3 unless
otherwise stated, chemical shifts and coupling constants (J) being
reported in ppm (δ) from tetramethylsilane, and hertz (Hz), respectively.
Infrared spectra (IR) were recorded on a Perkin-Elmer model 598
spectrophotometer and reported in reciprocal centimeters (cm-1).
Quantitative gas chromatographic analysis (GC) was effected on a
Perkin-Elmer model 990 instrument with flame ionization detector and
With its inclusion, a caldera-in-common becomes qualitatively
tenable. Solution of the three equations transforming the
experimental results may be expressed pie-graphically as the
upper diagram in Figure 2.11 Entry into the caldera from trans-1
(13) Doering, W. v. E.; Sachdev, K. J. Am. Chem. Soc. 1974, 96, 1168-1187.
(14) Doering, W. v. E.; Mastrocola, A. R. Tetrahedron, Suppl. 1 1981, 17, 329-
344.
9
10612 J. AM. CHEM. SOC. VOL. 125, NO. 35, 2003

Thermal Rearrangements of a Cyclobutane
A R T I C L E S
Hewlett-Packard 3380S integrator, using a 50-ft OV-225 capillary
column. HPLC employed a Waters ALC/GPC 501 instrument fitted
with a model 6000 solvent delivery system, a model UK6 injector,
and a refractive index detector: column A, Waters µPorasil (25 cm);
and column B, DuPont Zorbax Sil (25 cm). Melting points were
determined without correction in a Hershberg apparatus.
before having a precipitate removed by filtration. The diamide was
washed with benzene, methylene chloride, and water: 75-90% of
theoretical yield, dec >220 °C, varied from 75 to 95% of theoretical:
IR (mineral oil mull) 3340, 3180, 1665, 1400, 1205, 1105, 720.
Ethyl polyphosphate17 (4.6 g) and 0.46 g of crude diamide above
were boiled under reflux in 20 mL of CHCl3 for 24 h. Filtered through
glass wool, and stirred with 50 mL of 25% aqueous K2CO4 for 1 h,
the chloroform solution was successively washed with water, dried over
MgSO4, and concentrated to an oil, which partially crystallized when
triturated with hexane. Recrystallization from methanol/water gave 0.24
Tricyclo[4.2.2.02,5]decane-cis-3,4-dicarboxylic Acid. Into an im-
mersion-well photolysis apparatus fitted with a magnetic stirrer, a reflux
condenser capped with a CaSO4-filled drying tube, and a 450-W
Hanovia mercury lamp contained in a Pyrex, water-cooled sleeve was
placed a solution of 7.3 g of bicyclo[2.2.2]oct-2-ene [prepared in three
steps from cyclohexa-1,3-diene and maleic anhydride following the
procedure of Grob et al.15 and Cimarusti and Wolinsky16 (for details
see ref 6)] in pentane (150 mL) and methylene chloride (250 mL)
containing maleic anhydride (35 g) and acetophenone (21 g). The
photolysis vessel, cooled in a bath refrigerated at 5 °C, was irradiated.
The photolysis was interrupted twice during the first 24 h to allow
filtration of the reaction mixture and cleaning of the sleeve. After a
total of 43 h, the filtrate was concentrated under vacuum to a maroon
liquid, which was extracted with boiling ether. The ethereal extract
was filtered and concentrated to an oil, which was then boiled under
reflux with 7.5% aqueous K2CO3 (200 mL) for 2.5 h. The resulting
mixture was washed with ether, acidified, and extracted thrice with
ether. The combined ethereal extracts were washed with water, dried
over MgSO4, and concentrated to an oil, which crystallized upon
trituration with a mixture of benzene and acetone: 2.75 g; mp 188-
1
g (62%) of trans-1: mp 105.5-107.0 °C; H NMR (270 MHz) 3.7
(m, 1H), 2.9 (m, 1H), 2.7 (m, 1H), 2.3 (m, 1H), 1.9-1.3 (m, 10H); 13
C
NMR (1H decoupled) 41.06, 37.50, 27.30, 24.82, 24.66, 24.44, 24.25,
21.42, 20.61 (quaternary C atoms not seen above baseline); IR (CH2-
Cl2) 2950, 2895, 2890, 2250, 1485, 1465.
anti,cis-3,4-Dicyanotricyclo[4.2.2.02,5]decane (anti,cis-1). A solution
of 0.18 g of trans-1 in 3 mL of dried DMSO, to which 0.3 mL of
freshly prepared 1.5 N sodium methoxide had been added, was stirred
for 5 min, poured into a separatory funnel containing 50 mL of ice
water, and extracted twice with 15 mL each of benzene. The benzene
extracts, washed with water and dried over MgSO4, were concentrated
to an oil (0.18 g), which was separated into two fractions by
chromatography on silicagel with hexane/ethyl acetate. That of shorter
retention time is trans-1, while that of longer is anti,cis-1: mp 121-
1
122 °C; H NMR (270 MHz) 3.55 (m, 2H), 2.97 (m, 2H), 1.6 (m,
10H); 13C NMR (1H decoupled) 40.31, 26.68, 25.06, 23.98, 20.61
(quaternary C atoms not seen above baseline); IR (CH2Cl2) 2950, 2890,
2250, 1485, 1465.
1
190 °C dec; H NMR (80 MHz) 3.5 (m, 2H), 2.7 (m, 2H), 1.6 (m,
10H); 13C NMR (1H decoupled) 128.40, 41.87, 37.53, 25.79, 25.04,
21.58; IR (mineral oil mull) 3300-2600, 1710, 1285, 1130, 880.
Tricyclo[4.2.2.02,5]decane-trans-3,4-dicarboxylic Acid. A solution
of 0.75 g of the cis-diacid immediately above and 2 mL of concentrated
H2SO4 in 20 mL of methanol was boiled under reflux for 3 h, poured
into 50 mL of water, and extracted with ether. The ethereal extract
was washed successively with water, saturated NaHCO3, and water,
then dried over MgSO4, and finally concentrated to crystalline dimethyl
tricyclo[4.2.2.02,5]decane-cis-3,4-dicarboxylate: 0.71 g; mp 66-68 °C;
1H NMR (80 MHz) 3.7 (s, 6H), 3.5 (m, 2H), 1.9-1.4 (m, 10H); IR
(mineral oil mull) 1720, 1275, 1180, 1040, 990.
The Three cis-Bis-1,4-(2′-cyanoethenyl)cyclohexanes. A solution
of 25 mg of trans-1 and 5 mg of diphenylamine in 2.0 mL of tert-
butylbenzene was divided into four equal quantities; each was placed
into ammonia-treated, silanized Pyrex tubes, which were degassed and
sealed in vacuo. The ampules were heated for 945 min in the fluidized
bath at 299 °C. Analysis of the recombined solutions by capillary GC
(190 ° C, 12 psi He) revealed five compounds: cis,cis-2 (retention
time, 11 min), trans-1 (rt, 13 min), cis,trans-2 (rt, 18 min), trans,trans-2
(rt, 38 min), and anti,cis-1 (rt, 48 min). Separation by HPLC (column
B, 4% ethyl acetate in hexane, 1.0 mL/min) afforded four fractions:
trans-1 (rt, 10 min), a 1:9 mixture of cis,cis-2 and cis,trans-2 (rt, 15
min), trans,trans-2 (rt, 24 min), and anti,cis-1 (rt, 33 min): 1H NMR
(270 MHz) cis,cis-2 6.55 (dd, J ) 10.9, J ) 9.9), 5.32 (dd, J ) 10.8,
J ) 0.5), 2.85 (m), 1.6 (m); cis,trans-2 6.77 (dd, J ) 16.5, J ) 6.3),
6.49 (dd, J ) 10.9, J ) 9.9), 5.35 (dd, J ) 16.5, J ) 1.6), 5.31 (dd, J
) 10.9, J ) 0.7), 2.85 (m), 2.49 (m), 1.6 (m); trans,trans-2 6.73 (dd,
J ) 16.5, J ) 6.3, 2H), 5.33 (dd, J ) 16.5, J ) 1.6, 2H), 2.41 (m,
2H), 1.7-1.6 (m, 8H); IR (CHCl3) 2940, 2870, 2230, 1730, 1630, 1450,
970.
Freshly prepared 3 N sodium methoxide (3 mL) was added to a
solution of 2.71 g of the diester above in 10 mL of dimethyl sulfoxide
(DMSO). After 30 min of stirring, the solution was poured into ice
water, acidified with 6 N HCl, and extracted three times with ether.
After being washed with water, dried (MgSO4), and concentrated, the
ether solutions afforded a residual yellow oil (2.76 g), which was
refluxed for 2 h with 50 mL of 5% aqueous NaOH. Workup in the
usual fashion gave a residue, from which 1.33 g of tricyclo[4.2.2.02,5]-
decane-trans-3,4-dicarboxylic acid was obtained after crystallization
from CHCl3 (filtrate contains mainly the cis starting material): mp 208-
Reaction of Cyclohexane-1,4-dicarboxaldehyde with Cyanom-
ethylenetriphenylphosphorane. The dialdehyde was prepared by the
addition of a solution of 1.2 g of cis- and trans-bis-1,4-(hydroxymethyl)-
cyclohexane in 3 mL of acetone to a reagent prepared from 10.2 g of
chromium trioxide and 16.4 mL of anhydrous pyridine.18 After 15 min
of being stirred, the supernatant solution was decanted from the tarry
solid, which was extracted twice with ether. The combined ether
solutions were washed with 5% aqueous NaOH, water, 0.5 N HCl,
water and brine and finally dried and concentrated to give 0.5 g of
crude dialdehyde. Solid cyanomethylenetriphenylphosphorane was
prepared by boiling 12 g of triphenylphosphine and 3.5 g of chloro-
acetonitrile in 50 mL of benzene for 8 h and treating the resulting
crystalline phosphonium salt in 50 mL of water with NaOH to
precipitate 3.7 g of the phosphorane.
1
215 °C (dec); H NMR (80 MHz, acetone-d6) 3.65 (m, 1H), 3.55 (m,
1H), 2.60 (m, 1H), 2.50 (m, 1H), 1.8-1.6 (m, 10 H); 13C NMR (1H
decoupled, acetone-d6) 175.56, 174.78, 41.63, 40.63, 39.44, 37.83,
26.17, 26.06, 25.98, 25.79, 22.66, 21.96; IR (mineral oil mull) 2720,
2640, 1690, 1310, 1255, 1245, 950, 730.
trans-3,4-Dicyanotricyclo[4.2.2.02,5]decane (trans-1). A mixture of
0.50 g of the diacid prepared immediately above, 1.0 mL of dry THF,
and 2 mL of thionyl chloride in 20 mL of anhydrous benzene was
refluxed for 1.5 h, concentrated in a vacuum to a partly crystalline
solid, which was extracted with several small portions of anhydrous
benzene. The combined benzene extracts were introduced by filtration
through glass-wool into a round-bottomed flask fitted with a septum
and a condenser capped with a CaSO4 drying tube. After the stirred
mixture was briefly cooled in an ice bath, anhydrous ammonia was
bubbled through it for 15 min. The now viscous mixture rested overnight
A benzene solution (25 mL) of dialdehyde (0.5 g) and phosphorane
(2.7 g) was boiled under reflux for 4 h, concentrated to a brown oil,
(15) Grob, C. A.; Ohta, M.; Renk, E.; Weiss, A. HelV. Chim. Acta 1958, 41,
1191-1197.
(16) Cimarusti, C. M.; Wolinsky, J. J. Am. Chem. Soc. 1968, 90, 113-120.
(17) Pollman, W.; Schramm, G. Biochim. Biophys. Acta 1964, 80, 1-7.
(18) Ratcliffe, R.; Rodehorst, R. J. Org. Chem. 1970, 35, 4000-4002.
9
J. AM. CHEM. SOC. VOL. 125, NO. 35, 2003 10613

A R T I C L E S
Doering and DeLuca
Table 3. Thermal Rearrangement of a Mixture of cis,trans-2 and
cis,cis-2 at 199 °C
which was extracted thrice with ether. The ether extracts were filtered,
and again concentrated to an oil. This process, intended to remove
triphenylphosphine oxide, was repeated. The major product, isolated
by chromatography on silica gel (hexane/ethyl acetate) and purified
by crystallization from methanol/water, was a trans-bis-1,4-(trans-2′-
cyanoethenyl)cyclohexane: 1H NMR (80 MHz) 6.54 (dd, J ) 16.5, J
) 6.5, 2H), 5.27 (dd, J ) 16.4, J ) 1.4, 2H), 2.3-1.0 (m, 8H); IR
(CHCl3) 2985, 2940, 2860, 2235, 1635, 1115. A minor product was
shown to be identical to trans,trans-2 by co-injection on capillary GC
time, h
cis,trans-2
cis,cis-2
trans,trans-2
ratioa
0
3.03
5.61
90
85
78
10
12
17
0
2.6
4.5
9.0:1
7.1:1
4.6:1
a The ratio of cis,trans-2 to cis,cis-2.
1
butylbenzene containing 2.8 mg of diphenylamine was divided into
aliquots (∼200 µL) and placed in Pyrex tubes as noted above. After
being heated in the fluidized alumina bath at 199 °C for various periods,
the tubes were removed and their contents analyzed by capillary GC,
each three times. Results are given in the Table 3. No trace of trans-1
or anti,cis-1 was seen.
column, and by co-injection on HPLC column B: H NMR (80 MHz,
C6D6) 6.05 (dd, J ) 16.8, J ) 6.4, 2H), 4.63 (dd, J ) 16.8, J ) 1.6,
2H), 1.5 (br m), 0.9 (m).
Kinetics of the Rearrangement of trans-1 and anti,cis-1. In a
typical experiment, a stock solution of trans-1 (0.25%), hexacosane
(0.13%), and diphenylamine (0.05%) in tert-butylbenzene was prepared.
Aliquots (200 µL) were degassed and sealed in vacuo in Pyrex ampules
(100 × 4 mm i.d.) prepared by first soaking in aqueous ammonia (20%)
overnight, rinsing with water and acetone, and drying and then treating
with dichlorodimethylsilane (20% in benzene), rinsing with benzene,
acetone, and drying. The sealed tubes were heated in a Tecam FB-07
fluidized alumina bath, the temperature being monitored by an iron-
constantan thermocouple and Leeds and Northrop 8686 millivolt
potentiometer. For the 256 °C temperature, heating was by insertion
into the vapors of boiling biphenyl.
Acknowledgment. Support from NSF Grant 7698 is grate-
fully acknowledged. We express our gratitude to the late
Professor Wolfgang R. Roth for his independent calculations
of the kinetic results. Our warmest appreciation goes to Professor
John E. Baldwin for his many helpful comments.
Supporting Information Available: Unrefined kinetic data
from the thermal transformations of trans- and cis-1 are
contained in Table SI-1, while those from syn,cis-5 are found
in Table SI-2 (PDF). This material is available free of charge
at the Internet at http://pubs.acs.org.
Analysis of the contents of the ampules was by GC, column A (190°,
12 psi He). The analyses reported in Table SI-1 were the averages of
three separate GC analyses.
Thermal Rearrangement Among Dienes 2. A solution of ∼5 mg
of a 9.9:1 mixture of cis,trans-2 and cis,cis-2 in 1.5 mL of tert-
JA030050E
9
10614 J. AM. CHEM. SOC. VOL. 125, NO. 35, 2003