Fate of the intermediate diradicals in the caldera: Stereochemistry of thermal stereomutations, (2 + 2) cycloreversions, and (2 + 4) ring-enlargements of cis- and trans-1-cyano-2-(E and Z)-propenyl-cis-3,4-dideuteriocyclobutanes

Article abstract of DOI:10.1021/ja0206083

This paper addresses the decades-old problem of gaining a measure of intellectual control over the fate of the diradical intermediate in not-obviously-concerted thermal rearrangements. It focuses mainly on the stereochemistry of the thermal rearrangement of cis- and trans-1-cyano-2-trans-propenylcyclobutane to the related ring-enlarged products, 4-cyano-3-methylcyclohexenes. The complete stereochemical profile is revealed by the incorporation of a pair of cis deuterons to serve as a stereochemical lighthouse. The striking result (besides providing a further example of the inapplicability of the orbital symmetry rules of Woodward and Hoffmann to not-obviously-concerted reactions) is the predominance of the same stereoisomer regardless whether starting from the cis or trans educt. This preference is rationalized by a simple conceptual scheme based on two premises of the behavior of the diradical as intermediate: removal of the diradical from the caldera of rotationally labile conformations occurs whenever the two radical centers come within bonding distance in an appropriate orientation of orbitals; relative internal rotational rates are in the order, cyanomethyl faster than methallyl, faster than internal rotation about the backbone.

Full text of DOI:10.1021/ja0206083

Published on Web 09/07/2002
Fate of the Intermediate Diradicals in the Caldera:
Stereochemistry of Thermal Stereomutations,
(2 + 2) Cycloreversions, and
(2 + 4) Ring-Enlargements of cis- and trans-
1-Cyano-2-(E and Z)-propenyl-cis-3,4-dideuteriocyclobutanes
W. von E. Doering,*,1 Xueheng Cheng, Kyuwang Lee, and Zisen Lin
Contribution from the Department of Chemistry and Chemical Biology, HarVard UniVersity,
Cambridge, Massachusetts 02138-2902
Received April 29, 2002
Abstract: This paper addresses the decades-old problem of gaining a measure of intellectual control over
the fate of the diradical intermediate in not-obviously-concerted thermal rearrangements. It focuses mainly
on the stereochemistry of the thermal rearrangement of cis- and trans-1-cyano-2-trans-propenylcyclobutane
to the related ring-enlarged products, 4-cyano-3-methylcyclohexenes. The complete stereochemical profile
is revealed by the incorporation of a pair of cis deuterons to serve as a stereochemical lighthouse. The
striking result (besides providing a further example of the inapplicability of the orbital symmetry rules of
Woodward and Hoffmann to not-obviously-concerted reactions) is the predominance of the same
stereoisomer regardless whether starting from the cis or trans educt. This preference is rationalized by a
simple conceptual scheme based on two premises of the behavior of the diradical as intermediate: removal
of the diradical from the caldera of rotationally labile conformations occurs whenever the two radical centers
come within bonding distance in an appropriate orientation of orbitals; relative internal rotational rates are
in the order, cyanomethyl faster than methallyl, faster than internal rotation about the backbone.
Introduction
cycloreversion] to two molecules of ethylene, through studies
of labeled derivatives to reveal stereomutation, and finally the
The class of “diradical” thermal reorganizations is divided
into concerted and not-obviously-concerted subclasses on the
basis of the criterion of energy of concert. The former strictly
obey the Woodward-Hoffmann rules of orbital symmetry and
follow stereochemical paths under full intellectual control.
Enthalpies of activation in the second class are under modest
intellectual control, but stereochemistry is anything but. Most
of the studies in the second class have involved cyclopropanes
and cyclobutanes, the strain energies of which bring the reactions
into a manageable range of temperatures. This work is no
exception and is concerned with the stereochemistry of three
modes of reaction of a suitably substituted cyclobutane.
Cyclobutanes undergo a variety of purely thermal reactions,
the rates of which are influenced almost exclusively by tem-
perature and pressure alone (“no-mechanism”, “not-obviously-
concerted” thermal reorganizations). Their study has been an
exemplar of the productive interaction of mechanistic organic
chemists, earlier with statistical kineticists1 and later with
calculational organic chemists.2
introduction of vinyl groups to add ring-enlargement, the number
of exit channels brought under observation has increased
dramatically. These developments may be followed in a
sequence of references.3 Coupled with comparable developments
in cyclopropanes, these works have intensified an awareness
of the rudimentary level of our intellectual control over the
distribution and stereochemistry of products in multichanneled,
not-obviously-concerted thermal reorganizations. Meanwhile
calculational control has advanced rapidly in parallel with
extraordinary advances in computer power.
About twenty years ago, we began to examine the systems
cis- and trans-1-cyano-2-(cis-propenyl)cyclobutane (1 and 2)
and cis- and trans-1-cyano-2-(trans-propenyl)cyclobutane (5 and
(3) (a) Gerberich, H. R.; Walters, W. D. J. Am. Chem. Soc. 1961, 83, 3935-
3939; 4884-4888. (b) Chikos, J. S. J. Org. Chem. 1979, 44, 780-784.
Chikos, J. S.; Annamalai, A.; Keiderling, T. A. J. Am. Chem. Soc. 1986,
108, 4398-4402. (c) Goldstein, M. J.; Cannarsa, M. J.; Kinoshita, T.; Koniz,
R. F. Stud. Org. Chem. (Amsterdam) 1987, 31, 121-133. Cannarsa, M. J.
Ph.D. Dissertation, Cornell University, 1984; Diss. Abstr. Int. B 1984, 45,
1468 (Order No. 84-15,310). (d) Berson, J. A.; Tomkins, D. C.; Jones, G.,
II. J. Am. Chem. Soc. 1970, 92, 5799-5800. Jones, G., II; Fantina, M. F.
J. Chem. Soc., Chem. Commun. 1973, 375-377. Jones, G., II; Chow, V.
L. J. Org. Chem. 1974, 39, 1447-1448. (e) Doering, W. v. E.; Guyton, C.
A. J. Am. Chem. Soc. 1978, 100, 3229-3230. Guyton, C. A. Ph.D.
Dissertation, Harvard University, 1980; Diss. Abstr. Int. B 1980, 41, 1370.
(f) Hammond, G. S.; DeBoer, C. D. J. Am. Chem. Soc. 1964, 86, 899-
905. (g) Berson, J. A.; Dervan, P. B. J. Am. Chem. Soc. 1973, 95, 267-
271; 1976, 98, 5937-5968. (g) Jordan, L. M. Ph.D. Dissertation, Yale
University, 1974; Diss. Abstr. Int. B 1975, 35, 5332-5333. (h) Baldwin,
J. E.; Burrell, R. C. J. Am. Chem. Soc. 2001, 123, 6718-6719.
From the first studies of unlabeled cyclobutane, which
revealed the single exit channel of fragmentation [(2 + 2)
* To whom correspondence may be addressed. E-mail: doering@
chemistry.harvard.edu.
(1) Lin, M. C.; Laidler, K. J. Trans. Faraday Soc. 1968, 64, 927-944, and
cited references.
(2) Doubleday, C., Jr. J. Phys. Chem. 1996, 100, 15083-15086, and cited
references.
9
11642
J. AM. CHEM. SOC. 2002, 124, 11642-11652
10.1021/ja0206083 CCC: $22.00 © 2002 American Chemical Society

Intermediate Diradicals in the Caldera
A R T I C L E S
Scheme 1
Structure of Educts and Products. Although this section
is presented in considerable detail because the structure of the
cyclobutanes is critically important to the reliability of the
conclusions, a trusting reader may pass over it and proceed
directly to the Thermal Reorganizations section. The preparation
of cis-1-cyano-2-(cis-propenyl)-3,4-dideuteriocyclobutane, 1ee,
is accomplished by the sequence of reactions in Scheme 1
(experimental details are given only for the dideuterium
compounds, although identical procedures were first developed
for the nondeuterated compounds). Preparation of the corre-
sponding trans-1-cyano-2-(cis-propenyl)-3,4-dideuteriocyclobu-
tane, 2ze, was effected by base-catalyzed equilibration of 1ee.
Conversion of the cis-propenyl group in 1ee into its correspond-
ing trans-propenyl isomer 5ee was accomplished photochemi-
cally. A fourth isomer, 6ze, was prepared from 5ee by base-
catalyzed equilibration. As was to be expected, the trans isomers,
2ze and 6ze, were favored at equilibrium by a factor of ∼3
(Figure 3).
Identification of H-1, H-2, H-7, H-8, and H-9 in the 1H NMR
spectra of all four stereoisomers follows in a straightforward
manner from chemical shifts, integrated areas, coupling patterns,
and decoupling experiments. Identification of the cis and trans
configuration of the propenyl groups is based on the coupling
constants of H-7 and H-8: ∼10 Hz for 1, 1ee, 2, and 2ze; ∼16
Hz for 5, 5ee, 6, and 6ze. Identification of the remaining
hydrogen atoms and assignments of stereochemistry are based
on lanthanide induced shifts (LIS) with reagent, Eu(fod)3.9 We
find relative bound shifts, as extensively developed indepen-
dently by Davis et al.,10 in practice more convenient to use than
absolute bound shifts.11
Figure 1. Specific rate constants at 198 °C for stereomutation and
fragmentation of 1 and 2 (10-6 s-1). Designations e and z refer to the relation
of cyano (first) and cis-propenyl (second) to the pair of cis deuterons as
reference.
6),4 in an extension of the work of Sachdev on 1-cyano-2-
isopropenylcyclopropane5 and that of Mastrocola on 1-cyano-
2-vinylcyclobutane.6
To render visible all internal rotational possibilities, first,
optical activity with disappointing results, and later, two cis
deuterium atoms as the chiral element have been employed
(transformation “of interrelations among enantiomers into
interrelations among diastereomers”). Although deuterium has
long been used as a probe of the stereochemistry of fragmenta-
tion,7 it has not been used in this way as a probe for stereo-
mutation or ring-enlargement.8
Reactions of the (Z)-propenyl system are limited to stereo-
mutation and to fragmentation to piperylene (3) and acrylonitrile
(4), ring-enlargement being thwarted by steric inhibition of the
generation of the apposite allyl radical in the necessarily cis-
configuration required for generation of a strain-free cyclohex-
ene ring (Figure 1). The (E)-propenyl system additionally may
undergo ring-enlargement to the 4-cyano-3-methylcyclohexenes,
7 and 8 of Figure 2.
(6) Doering, W. v. E.; Mastrocola, A. R. Tetrahedron 1981, 37, Suppl. 1, 329-
344.
(7) See for example: Srinivasan, R.; Hsu, J. C. N. J. Chem. Soc., Chem.
Commun. 1972, 1213-1214. Wang, Y.-s.; Chickos, J. S. J. Org. Chem.
1987, 52, 4776-4781.
(8) Doering, W. v. E.; Yamashita, Y. J. Am. Chem. Soc. 1983, 105, 5368-
5372.
(9) Tris(1,1,1,2,2,2,3,3-heptafluoro-7,7-dimethyl-4,6-octadionato)europium.
(10) (a) Willcott, M. R., III; Lenkinski, R. E.; Davis, R. E. J. Am. Chem. Soc.
1972, 94, 1742-1744. (b) Davis, R. E.; Willcott, M. R., III. J. Am. Chem.
Soc. 1972, 94, 1744-1745. (c) Doering, W. v. E.; Birladeanu, L.
Tetrahedron 1973, 29, 499-512. (d) Davis, R. E.; Willcott, M. R., III;
Lenkinski, R. E.; Doering, W. v. E.; Birladeanu, L. J. Am. Chem. Soc.
1973, 95, 6846-6848.
(4) Of the several advantages offered by the cyano groupsacidifying effect,
LIS-friendlinesssnot the least is its low steric value: Vo¨gtle, F.; Neumann,
P.; Zuber, M. Chem. Ber. 1972, 105, 2955-2962.
(5) Doering, W. v. E.; Sachdev, K. J. Am. Chem. Soc. 1974, 96, 1168-1187;
1975, 97, 5512-5520.
9
J. AM. CHEM. SOC. VOL. 124, NO. 39, 2002 11643

A R T I C L E S
Doering et al.
Figure 3. Interactions among the four cyanopropenylcyclobutanes in this
work. Relative responses in chemical shifts to the lanthanide shift reagent,
Eu(fod)3, (H-1 ) 1.000) of the hydrogen atoms in the nondeuterated
compounds 1, 2, 5, and 6 (see also Table 1) are indicated. At the bottom is
drawn the average geometry assumed for the nitrile-europium coordinate
bond and the parameters in the McConnell-Robertson equation governing
pseudocontact shifts (distances in Å).
Chemical shifts of all hydrogen atoms in compounds 1, 2, 5,
and 6 were recorded at 11, incrementally increased concentra-
tions of the shift reagent. The data were plotted as the differences
in chemical shifts of H-1 through H-8 against H-9, which was
well separated in all the spectra. Linear regression afforded
slopes that define the relative response of each type of hydrogen
atom to the shift reagent. The intercepts at zero concentration
of shift reagent are differences between the conventional
chemical shifts and that of H-9 and provide a source of the few
chemical shifts that are otherwise obscured by unresolved
overlaps. These latter values are recorded in italics in Table 1.
In the next step, the relative response factors vis-a`-vis H-9 were
normalized against H-1, the most responsive of the hydrogen
atoms, being set equal to 1.000. These “experimental” factors,
∆ex, are given in both Figure 3 and Table 1.
As might be expected intuitively, the hydrogen atoms next
closest to the coordinating cyano group after H-1 all show large
response factors ranging from 0.576 (H-4 in 2ze) to 0.683 (H-7
in 5ee). The stereochemical relation of the cyano and propenyl
groups is indicated not only by the response factors for H-7
(larger, 0.660 and 0.683, in the cis compounds, 1ee and 5ee,
respectively; smaller, 0.323 and 0.290, in the trans compounds,
2ze and 6ze, respectively) but also from the response factors
for H-2 (smaller, 0.431 and 0.442, in 1ee and 5ee, respectively;
larger, 0.641 and 0.616, in 2ze and 6ze, respectively).
Figure 2. Specific rate constants at 198 °C for stereomutation, fragmenta-
tion, and ring-enlargement of 5 and 6 (10-6 s-1). Designations e and z
refer to the relation of cyano (first) and trans-propenyl (second) to the pair
of cis deuterons as reference.
(11) (a) Raber, D. J.; Johnston, M. D., Jr.; Schwalke, M. A. J. Am. Chem. Soc.
1977, 99, 7671-7673. (b) Raber, D. J.; Johnston, M. D., Jr.; Perry, J. W.;
Jackson, G. F., III. J. Org. Chem. 1978, 43, 229-231. (c) Raber, D. J.;
Janks, C. M.; Johnston, M. D., Jr.; Raber, N. K. Org. Magn. Reson. 1981,
15, 57-76.
9
11644 J. AM. CHEM. SOC. VOL. 124, NO. 39, 2002

Intermediate Diradicals in the Caldera
A R T I C L E S
Table 1. NMR Chemical Shifts and Lanthanide Induced Shifts
(Eu(fod)3) in CDCl3 of Cyanopropenyl Cyclobutanes
Table 2. Correlations in the cis-Propenyl Compounds, 1 and 2,
and in the trans-Propenyl Compounds, 5 and 6, of Calculated
Values of LIS Response Factors of the Pairs of Conformations
with Experimental Values
1ee
1
2ze
2
5ee
5
6ze
6
H-1 δa
3.27 3.29 2.77 2.77 3.22 3.24 2.72 2.76
1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
correlationa
ab
bb
cb
σ
σa
σb
r c
b
∆
ex
eqax1/axeq1d vs 1e 0.188 0.578 0.010 0.014 0.003 0.003 0.997
eqax1/axeq1 vs 2 0.714 -0.256 0.150 0.136 0.331 0.311 0.654
eqeq2/axax2 vs 2 0.795 -0.019 0.005 0.022 0.008 0.007 0.993
eqeq2/axax2 vs 1 0.213 0.257 0.177 0.150 0.391 0.345 0.587
eqax5/axeq5 vs 5 0.142 0.662 -0.002 0.019 0.005 0.005 0.996
eqax5/axeq5 vs 6 0.545 -0.060 0.115 0.148 0.308 0.284 0.623
eqeq6/axax6 vs 6 0.784 0.003 -0.004 0.011 0.002 0.002 0.998
eqeq6/axax6 vs 5 0.344 0.146 0.184 0.159 0.385 0.386 0.585
c
∆
δ
1.000
1.000
1.000
1.000
cl
H-2
H-3
H-4
H-5
H-6
H-7
H-8
H-9
3.49 3.51 3.55 3.56 3.16 3.17 3.14 3.18
0.424 0.431 0.643 0.641 0.440 0.442 0.618 0.616
∆
∆
δ
ex
0.410
2.30d
0.291
0.306
2.34
0.605
2.19
0.276
0.278
2.25
0.416
2.21
0.302
0.309
2.32
0.603
2.10
0.277
0.278
2.19
cl
∆
∆
δ
ex
cl
∆
∆
δ
0.398
0.410
0.576
0.605
0.391
0.416
0.589
0.603
ex
a Correlation by eq 3 (∆j′ ) a(∆j(calc)/conformation R) + b(∆j(calc)/conformation â
+ c). b Best fit values by least squares linear regression. c The correlation
)
cl
2.22 2.24 2.20 2.23 2.16 2.17 2.12 2.16
0.578 0.592 0.413 0.408 0.592 0.602 0.430 0.428
coefficient, r. d Calculated values, ∆j(calc) e Experimental values, ∆exp (from
.
∆
∆
δ
ex
Table 1).
0.604
0.427
0.622
0.423
cl
2.03 2.07 1.84 1.86 2.11 2.13 1.83 1.87
0.438 0.446 0.305 0.302 0.466 0.475 0.308 0.304
McConnell-Robertson equation. As a consequence, H-1 is not
included in the subsequent calculation. The second discrepancy
involves H-3 and arises from the existence of two conformations
of the cyclobutane ring. Experimental values, ∆j, are therefore
compared to an optimized combination of the calculated
theoretical values. This combination is made by the linear
regression of eq 3, H-1 being omitted. The results are shown in
Table 1, while values of ∆j′ are given in Table 2.
∆
∆
δ
ex
0.445
0.298
0.470
0.293
cl
5.72 5.72 5.36 5.36 5.71 5.71 5.44 5.44
0.661 0.660 0.325 0.323 0.664 0.683 0.279 0.290
∆
∆
δ
ex
0.650
0.310
0.671
0.304
cl
5.61 5.62 5.50 5.49 5.57 5.57 5.54 5.54
0.188 0.188 0.175 0.176 0.315 0.318 0.248 0.277
∆
∆
δ
ex
0.188
0.171
0.306
0.277
cl
1.59 1.60 1.66 1.66 1.72 1.73 1.66 1.67
0.198 0.200 0.221 0.220 0.095 0.105 0.103 0.096
∆
∆
ex
0.193
0.227
0.109
0.096
cl
Examination of Table 2 reveals far higher correlation coef-
ficients for the assigned structures than for their alternatives.
Compare, for example, eqeq2/axax2 versus 2 and versus 1.
Further to be noted is the essentially exclusive favoring of the
equatorial-equatorial conformation over the axial-axial in
compounds 2 and 6: a ) 0.795 and b ) -0.019, and a )
0.784 and b ) 0.003, respectively. It seems justified also to
conclude that both conformations contribute to the equilibrium
structure of 1 and 5, for which a ) 0.188 and b ) 0.578, and
a ) 0.142 and b ) 0.662, respectively.13 The advantage in
having the propenyl group equatorial is correspondingly 75%
in cis-propenyl and 82% in trans-propenyl. This bias is fully
consistent with the narrower profile of the cyano group.
Thermal Reorganizations. The kinetics of the various
thermal transformations were brought to light by heating samples
in the gas phase in sealed ampules. The kinetics were determined
at four temperatures, 192.0, 198.0, 207.2, and 217.8 °C.
Analyses of products were by GC, the relative areas being
converted into relative concentrations by application of inde-
pendently determined response factors. Samples of undeuterated
1 and 2 were used to determine specific rate constants for
stereomutation and fragmentation to acrylonitrile (4) and cis-
piperylene (3). The ratio of these latter two rates should have
been equal, but was generally ∼0.8, presumably owing to some
polymerization of acrylonitrile that we were unable to eliminate.
In the final calculations of specific rate constants, those for
formation of acrylonitrile (k9 + k10) were adjusted to equal the
values for cis-piperylene (k7 + k8); that is, the ratio was assumed
to be 1.00 (Figures 1 and 2). Otherwise recoveries were 100%
within experimental uncertainties. The double rotational pro-
cesses, for example of 1ee to 1zz, cannot be distinguished
explicitly from two-step processes such as 1ee to either 2ze or
2ez, then to 1zz. However, if the two direct interconversions
(k5 and k6 of Figure 1) were omitted from the kinetic model,
agreement between calculated values of (1 - e-kt) and
a Chemical shifts in parts per million with reference to tetramethylsilane.
b Sensitivity of H-j to LIS relative to H-1 ) 1.000. c In those instances
where the chemical shift could not be extracted from the normal NMR
spectrum, a value in italics was estimated by extrapolation of the LIS data
to zero concentration of the shift reagent (see Figure 3).
Further analysis involves a quantitative comparison of the
experimental values theoretically calculated by the McConnell-
Robertson equation (eq 1),12 which evaluates the pseudocontact
LIS of hydrogen atoms in terms of two variables, Φj, the angle
between the line connecting H-j with the principle magnetic
axis of the substrate-metal complex, and rj, the distance
between H-j and the lanthanide atom. The lanthanide is
generally assumed to be collinear with the C-CN group and is
here taken to have a nitrogen-lanthanide separation R of 2.5
Å (see Figure 3).10d,11 For each H-j, values of Φj and rj are
estimated from Dreiding-stereochemical models in which the
four-atom system, H-2, C-2, C-4, and C-6, is maintained
coplanar; that is, C-H-2 and the olefinic group are eclipsed.
Cyclobutane being nonplanar, each of the four stereoisomers
may exist in two conformations. The angle, R, is assumed equal
to 20°. The resulting absolute values obtained by application
of eq 1, ∆δj, are then converted to values relative to H-1, ∆j(calc)
by eq 2. When these values are plotted against the corresponding
,
∆δj ) K(3 cos2 Φj - 1)/(rj)3
j(calc) ) [(3 cos2 Φj - 1)/(rj)3]/[(3 cos2 Φ1 - 1)/(r1)3] (2)
(1)
∆
∆j′ ) a[∆j(calc/conformation R)] + b[∆j(calc/conformation â)] + c (3)
experimental values, ∆ex, all but two of the hydrogen atoms
fall on a straight line. The first exception is H-1, which appears
too responsive. This discrepancy is likely due to a contact shift
superimposed on the pseudocontact shift calculated by the
(13) For a successful application to conformational equilibria in some acyclic
nitriles, see ref 12a; and in cyanocyclohexane, ref 15.
(12) Seux, R.; Morel, G.; Foucand, A. Tetrahedron Lett. 1972, 1003-1006.
9
J. AM. CHEM. SOC. VOL. 124, NO. 39, 2002 11645

A R T I C L E S
Doering et al.
Table 3. Molar Percent Concentrations of Products from Thermal Reorganizations of 1ee and 2ze at 198.0 °C
timea
1ee
1zz
2ze
2ez
3ez
3zz
4e
4z
“5ee” d
0.0b,c
1.0
2.0
97.42 ( 0.76
94.64
90.81
1.46 ( 0.16
1.89
2.23
0.42 ( 0.32
1.37
2.52
0.21 ( 0.32
0.71
1.35
0.33 ( 0.34
0.91
1.95
0.17 ( 0.21
0.48
1.05
0.31 ( 0.30
0.84
1.80
0.20 ( 0.23
0.54
1.20
0.00 ( 0.08
0.00
0.10
4.0
6.0
85.81
78.46
2.62
3.55
4.30
5.89
2.43
3.46
2.91
5.12
1.69
3.11
2.70
4.74
1.90
3.49
0.23
0.43
8.0
73.08
3.69
8.09
4.09
5.71
3.76
5.37
4.10
0.69
0.0b,c
3.0
6.0
0.66 ( 0.22
2.42
3.40
0.33 ( 0.20
1.25
1.83
90.40 ( 0.71
84.72
80.41
7.93 ( 0.08
8.21
8.46
0.00 ( 0.29
1.90
3.45
0.00 ( 0.53
0.97
1.93
0.00 ( 0.45
1.75
3.21
0.00 ( 0.36
1.11
2.17
9.0
12.0
4.63
5.20
2.60
3.06
74.48
69.41
8.67
8.74
5.62
7.69
3.44
5.42
5.27
7.41
3.79
5.70
a Time in hours. b Calculated by linear extrapolation to zero time of the data. c Standard deviation by linear least squares regression. d Identity supported
by GC retention time alone.
Table 4. Calculated Specific Rate Constants at 198.0 °C of
Stereomutation and Fragmentation of 1-Cyano-2-(Z)-propenyl-
cis-3,4-dideuteriocyclobutanes, 1ee, 2ze, 1zz, and 2ez (Figure 1)
stereomutation
fraction
fragmentation
a
k1
k3
3.35 ( 0.42
0.348
0.180
k7
k8
k9
k10
k11
k12
k13
k14
2.13 ( 0.29
1.73 ( 0.29
1.28 ( 0.19
1.99 ( 0.28
1.42 ( 0.18
1.90 ( 0.23
1.05 ( 0.31
1.17 ( 0.23
1.76 ( 0.22
k2
k4
k5
k6
1.70 ( 0.15
0.88 ( 0.07
1.21 ( 0.18
0.76 ( 0.05
0.177
0.091
0.126
0.079
a Specific rate constants in units of 10-6 s-1
.
experimental concentrations was much poorer than when
optimized values for k5 and k6 were included. We believe the
resulting values of k5 and k6 to be the least accurate in this work.
Quantitative analysis by NMR of purified samples of the
products from 1ee and 5ee allowed dissection of the specific
rate constants from the nondeuterated materials into diastereo-
meric pairs. The ratios of 3ze to 3ee are complicated by some
subsequent rearrangement of deuterium in an otherwise degen-
erate 1,5-hydrogen shift (in cis-piperylene but not in trans-
piperylene). The resulting perturbation in chemical shift of the
relevant vinyl hydrogen atoms by deuterium appears as shoul-
ders on the peaks of the terminal vinyl hydrogen atoms. Details
of the correction for this complication are available in the
dissertation of Cheng, as are all other parts of the experimental
work.14 A sample of data from the rearrangements of 1ee and
2ze at 198.0 °C is given in Table 3. The complete set of data
is available in Cheng’s dissertation.14
The derived specific rate constants are given in Table 4 and
Figure 1. The precision we were able to attain at four
temperatures, 192.0, 198.0, 207.2, and 217.8 °C, was insufficient
to allow the temperature coefficients to be determined with an
accuracy sufficient to warrant dissection of the ratios of the exit
channels into differences in enthalpy and entropy. The results,
which are not discussed here in detail but are available,14 are
in unexceptional agreement with the values expected of
stabilization in cyanomethyl and allyl radicals. Arrhenius
parameters for the disappearance of 1ee are Ea ) 38.1 ( 1.1
kcal mol-1 and log A ) 12.6 ( 0.5, and of 5ee are Ea ) 38.9
( 1.2 kcal mol-1 and log A ) 12.8 ( 0.5.
Figure 4. Estimated enthalpies of formation of the cyclobutane educt,
products, and diradical transition region for trans-1-cyano-2-trans-propenyl-
cyclobutane (6) in kcal mol-1
.
Table 5. Calculated Specific Rate Constants at 198.0 °C of
Stereomutation, Fragmentation, and Ring-Enlargement of
1-Cyano-2-(E)-propenyl-cis-3,4-dideuteriocyclobutanes, 5ee, 6ze,
5zz, and 6ez (Figure 2)
stereomutation
fraction
fragmentation
ring-enlargement
a
k1 2.03 ( 0.24 0.348
k3 0.86 ( 0.08 0.180
k7 1.84 ( 0.22
k8 0.99 ( 0.12
k9 1.68 ( 0.20
k10 1.14 ( 0.11
k11 1.07 ( 0.21
k12 0.45 ( 0.05
k13 0.59 ( 0.07
k14 0.92 ( 0.13
k15 0.08 ( 0.01
k16 0.19 ( 0.03
k17 0.98 ( 0.15
k18 0.23 ( 0.01
k19 0.06 ( 0.01
k20 0.26 ( 0.01
k21 0.46 ( 0.04
k22 0.19 ( 0.01
k2 0.82 ( 0.04 0.177
k4 0.35 ( 0.04 0.091
k5 1.08 ( 0.17 0.126
k6 0.52 ( 0.04 0.079
a Specific rate constants in units of 10-6 s-1
.
Alder reaction of acrylonitrile and trans-piperylene, a reaction
that cis-piperylene does not undergo. The structure including
stereochemistry of the dideuterium analogues is assured by
analysis of the LIS spectra (for details, see Cheng14). The
kinetics are followed in the same manner as described above.
The resulting specific rate constants are given in Table 5 and
Figure 2.
Ring-enlargement is a third mode among the reactions of 5ee
and 6ze. The resulting 4-cyano-3-methylcyclohexenes are
identified by comparison with samples prepared by the Diels-
Discussion
The three modes of reaction, stereomutation, fragmentation,
and ring-enlargement, are initiated by cleavage of the 1,2-bonds,
the weakest, to the corresponding conformationally labile
(14) Cheng, X.-h. Ph.D. Dissertation, Harvard University, 1988; Diss. Abstr.
Int. B 1990, 50, 3472.
9
11646 J. AM. CHEM. SOC. VOL. 124, NO. 39, 2002

Intermediate Diradicals in the Caldera
A R T I C L E S
Scheme 2
diradicals in the caldera. The thermochemistry among the
products (Figure 4) varies markedly.15-17 The products of
fragmentation, acrylonitrile and piperylene, lie ∼23 kcal mol-1
below the enthalpy of formation estimated for the generic
diradical, while the products of stereomutation have estimated
enthalpies of formation ∼40 kcal mol-1 below. The cyano-
methylcyclohexenes resulting from ring-expansion are the most
stable, lying some 61 kcal mol-1 below. A glance at Figure 2
shows that these three exit channels comprise 34%, 48%, and
18% of the reaction of 5ee and 42%, 36%, and 23% of 6ze,
respectively. We conclude that distribution among the three
modes is not correlated in any recognizable way to the
thermochemistry. Although fragmentation may be favored by
entropy, this factor seems also not to dominate the outcome.
Stereomutation and fragmentation comprise 65% and 35%,
respectively, of the reaction of 1ee and 53% and 47%,
respectively, of the reaction of 2ze. The system does not
participate in ring-enlargement to a significant degree owing
most likely to adverse steric repulsions between the cis-methyl
group and the ethano link in the necessarily cisoid (“tucked
under”) conformations required for generation of a cis-cyclo-
hexene. In contrast, cis- and trans-1-methyl-2-cis-propenylcyclo-
butanes have been found first by Jordan3g and most recently in
full detail by Baldwin and Burrell3h to give a small amount of
ring-enlargement: 2.8% from the cis isomer and 3.3% from
the trans isomer. The failure of the cis-propenyl compounds
1ee and 2ze may be due to an unfavorable thermochemistry.
This factor has been examined by molecular mechanical
estimations of the relative steric energies of cisoid and transoid
conformations of the proximate, coplanar allylic radicals. On
the assumption that the differences in heats of formation of the
transition regions parallel the differences among the selected
conformations, the cis-propenyl group in its cisoid orientation
is 3.78 and 2.80 kcal mol-1 higher in steric energy than in its
transoid orientation in the (E,Z) and (Z,Z) diastereomers,
respectively. By contrast, the differences in the trans-propenyl
systems, 5ee and 6ze, are 1.08 and 0.98 kcal mol-1, respectively
(Table 5).
observable consequence stems from these three “backbone”
conformations, each differs in the modes of exit available to it.
The antiperiplanar set may engage only in fragmentation, while
the (+) synclinal set may also engage in stereomutation.20 Ring-
enlargement is precluded in the (E,E) (+) synclinal set by the
thermochemical improbability of generating a trans-cyclohexene
and in the (Z,E) (+) synclinal set by the longer separation
[reducible by internal rotation about the C-3-C-2 bond (vide
infra)]. In the (-) synclinal set, the cis conformation can engage
in all three modes of exit, while ring-enlargement remains barred
to the trans set.
The calderas of diradicals can be divided into two non-
interconvertible subsets differing in configuration of the allylic
radical (Scheme 2; illustrated for 5ee only). One is composed
of (Z,E)-1,3-disubstituted allylic radicals, the other, of (E,E)-
1,3-disubstituted allylic radicals. Thermal interconversion within
the caldera between the two sets requires cis-trans isomerization
of the allylic radicals and is rendered highly improbable within
the lifetime of the caldera by an enthalpy of activation estimated
to be ∼16 kcal mol-1 18
.
As a consequence, crossing between
systems 1/2 and 5/6 is not observed.19
Stereomutation among the cyclobutanes 1 and 2, and 5 and
6, is governed by Onsager’s demonstration that (at equilibrium)
k1 × k4 × k1 × k4 ) k3 × k2 × k3 × k2, whence k1/k2 ) k3/k4
) K, the equilibrium constant, and k1/k3 ) k2/k4 ) Ra, the
relative rotational propensities of cyano versus (Z)- and (E)-
propenyl.21 In the systems 1/2 and 5/6, the equilibrium constants,
K, have the values 1.97 and 2.47 at 198.0 °C, respectively,
Each of these two subsets is comprised of three conformations
staggered about the C-3-C-4 bond and interconnected by
rotation about that bond by a barrier presumed to be ∼3 kcal
mol-1, the barrier to rotation in butane. Although no direct,
(15) Heats of formation of 3, 4, cyclobutane, and 3-methylcyclohexene are taken
from Pedley, Naylor, and Kirby. Replacement of hydrogen by cyano is
estimated following ref 16 in Doering and Wang.
corresponding to ∆∆G ) -0.63 and -0.85 kcal mol-1
,
respectively. These values are in good agreement with those
differences in steric energy between isomers 1 and 2, and 5
(16) Pedley, J. B.; Naylor, R. D.; Kirby, S. P. Thermochemical Data of Organic
Compounds, 2nd ed; Chapman and Hall: London, 1986.
(17) Doering, W. v. E.; Wang, Y.-h. J. Am. Chem. Soc. 1999, 121, 10967-
10975.
(18) Korth, H.-G.; Trill, H.; Sustmann, R. J. Am. Chem. Soc. 1981, 103, 4483-
(20) In connection with synclinal (gauche), (-) and (+) designate helicity about
4489.
the back, C-3-C-4, bond.
(19) The small amount of material questionably suggested to be 5ee from 1ee
in Table 3 is supported by GC retention time only.
(21) Doering, W. v. E.; Sachdev, K. J. Am. Chem. Soc. 1974, 96, 1168-
1187.
9
J. AM. CHEM. SOC. VOL. 124, NO. 39, 2002 11647

A R T I C L E S
Doering et al.
Scheme 3
between C-1 and C-2 and either to the stereoisomeric cyclobu-
tane (5ee to 6ze) or to a nonreaction (5ee to 5ee).
In the diradical conceptual scheme, nonreactions of the cyclo-
butane involving ring-opening and ring-closing without leaving
an observable trace inevitably contribute to the full picture of
the exit channels accessible from the caldera and the confor-
mational gyrations of the diradical leading to them. Although
not observable as a stereomutation of a cyclobutane, the non-
reaction can be observed in principle in the reverse, the cyclo-
dimerization of (E)- or (Z)-deuterioacrylonitrile and (E)- or
(Z)-deuterio-cis-piperylene to the corresponding cyclobutanes.
Although such a study has not been made, in the related, more
symmetrical system, the thermal dimerization of cis-dideuterio-
acrylonitrile, the two processes involving no internal rotations
prior to closure of the intermediate diradicals to the dicyano-
cyclobutanes amount to 38% of the stereochemical outcome,
despite being Woodward-Hoffmann forbidden!3c
The zero rotational processes become observable, if not in
stereomutation, at least in principle in ring-enlargement and
fragmentation. The latter mode of reaction might appear to offer
a favorable opportunity, but, as can be seen in Figure 5, the
desired k0-rot is not separately observable because it is composed
of both k10 and k7, two uncorrelated rate constants each of which
also consists of two components, k0-rot and kPr-rot, and k0-rot
and kCN-rot, respectively. There is no way of circumventing this
limitation, although, as noted previously, relevant insight into
all six processes can be obtained from the reverse reaction of
and 6, calculated by Allinger’s MM2 program for conformations
of lowest energy: -0.38 and - 0.57 kcal mol-1, respectively.22
The relative rotational propensities, Ra, have the values 1.94
and 2.35 in the systems 1/2 and 5/6, respectively. These values
are comparable to those found in 1-cyano-2-vinylcyclobutane
(1.48),6 1-cyano-2-isopropenylcyclopropane (2.20),5 and 1-cy-
ano-2-trans-propenylcyclopropane (2.36)23 and somewhat larger
than the rotational propensity of methyl versus trans-propenyl,
1.41, elucidated by Baldwin and Burrell in the system 1-methyl-
2-trans-propenylcyclobutane.3h Although there is no theoretical
reason k1/k2 should be equal to k1/k3, the near identity of their
values may simply be coincidental or an indication that the
relevant channels of exit from the caldera reflect the differences
in the thermochemical stability of the products. The double
rotational processes make smaller contributions to stereomuta-
tion (zero rotational processes not being measurable).
The “diradical” as a mechanistic conceptual scheme for the
not-obviously-concerted reactions consists of a rate-determining
entry of the educt into a caldera, rapid conformational changes,
and the sudden death of the diradical by covalent bond
formation, leading to exit channels that determine the structure
and relative proportion of products. Entry into the caldera is
achieved by bond-breaking to noninteractive singlet diradicals.
This step is considered not to be achieved simply by reaching
a higher level stretching vibration that corresponds in energy
to that of the incipient diradical. Such an excited vibrational
state is still one of the educt and only becomes a state of the
diradical when coupled to some further motion such as an
internal rotation that destroys any residual bonding between C-1
and C-2. At that point, orbital overlap vanishes and with it rule
by Woodward and Hoffmann. An example, shown in Scheme
3, is that from 5ee in the (-) synclinal conformation. The cyano
group has rotated to an extent that the saturated hydrogen atom
at C-1 is facing the sp2 orbital of C-2, which may also be
orthogonal to the sp2 orbital of C-1. The propenyl group has
remained in a cis conformation without further rotation. (Its
rotation simultaneous with the readjustments required to ac-
commodate the planarity of the allylic radical seems less
attractive.) Subsequent rotation of the cyano group in either
direction leads to the reestablishment of a bonding relation
cyclodimerization. However, values for (k0 - k2) and (kCN
-
kPr) can be extracted (Figure 5). These are positive in all
instances. We are struck by the preponderance of the individual
(as opposed to coupled) products that have involved no rotation
prior to cleavage. Thus, of the piperylenes formed from 1ee
and 2ze, 62% and 64%, respectively, have resulted from no
internal rotation prior to cleavage. The comparable values from
5ee and 6ze are 65% and 71%, respectively. A similar result is
seen in the acrylonitrile half of the fragmentation: 1ee and 2ze,
60% and 61%, respectively; 5ee and 6ze, 58% and 60%,
respectively. Strictly, these values pertain not just to zero
rotations but to an even number of rotations generally.
If a statistical model is explored in which the set is restricted
to zero, single, and double rotations prior to cleavage, their
relative contributions are given by
(1 - x)2 + 2x(1 - x) + x2
where x is the chance of a single rotation, and (1 - x) is the
chance of no rotation. In the case of piperylene, the mean of
the four values, now construed as the sum of the fractions of
zero and double rotations, 0.657, corresponds to a value of
(1 - x) ) 0.78 (x ) 0.22), whereas with respect to acrylonitrile,
the mean value 0.597 corresponds to (1 - x) ) 0.72 (x ) 0.28).
In this model cyano rotation again appears to be favored over
propenyl, but by a small factor of only 1.27.
Ring-enlargements present a further example where applica-
tion of Woodward-Hoffmann rules leads to discordant predic-
tions of the major products. Both 5ee and 6ze lead to 8ez as
the major product, the more so from cis isomer 5ee by a
forbidden process rs than from the trans isomer 6ze by an
allowed process is. In toto, 6ze leads to the four possible
products in the order 8ez, allowed-si (47%) as the major product,
7zz, forbidden-sr (27%), and 8ze, allowed-ar (20%) in com-
(22) Allinger, N. L. J. Am. Chem. Soc. 1977, 99, 8127-8132 (As made available
in Cambridge Chem 3D plus 6.0).
(23) Barsa, E. A. Ph.D. Dissertation, Harvard University, 1976; Diss. Abstr.
Int. B 1977, 37, 5077 (pp 99-132).
9
11648 J. AM. CHEM. SOC. VOL. 124, NO. 39, 2002

Intermediate Diradicals in the Caldera
A R T I C L E S
Figure 6. Distribution of products in % from 1-cyano-2-trans-propenyl-
cyclobutane (this paper; 198 °C; NC) and 1-methyl-2-trans-propenylcyclo-
butane (Baldwin and Burrell; 275 °C; H3C in parentheses; D ) H).3h
methyl and the higher temperature of reaction (275.0 versus
198.0 °C).
Of the three modes of reaction open to the cyclobutanes, ring-
enlargement is the simplest in the sense that only cisoid (Z,E)
diradicals can participate, as noted above (Scheme 2). Both
cisoid and transoid conformations are competent to engage in
stereomutation and fragmentation, but having no way of
examining the behavior of the two conformations independently,
and being at a loss to understand the relative distribution among
the three modes of reaction, we focus a detailed exploration of
the diradical conceptual scheme on ring-enlargements.
The striking features of the ring-enlargement are the pre-
dominance of a single stereoisomer as noted above and the
favoring of the thermodynamically trans-cyanomethylcyclo-
hexenes and trans-dimethylcyclohexenes over cis- in our system
and that of Baldwin and Burrell, respectively,3h and also in the
cyclopropane series.23
Figure 5. Internal rotations involved in the four paths for fragmentation
for 1ee. Their relation to the experimental rate constants are then indicated.
At the bottom, the only analytical conclusions give the differences in rate
constants (10-6 s-1 at 198 °C) between the zero and double rotational
processes (k0 - k2) and the cyano and propenyl processes (kCN - kPr) from
the educts, 1ee, 2ze, 5ee, and 6ze.
The qualitative aspects of ring-enlargement find reasonable
accommodation within the framework of the diradical hypothesis
outlined above. (In what follows, the text and figures are a feeble
substitute for models in hand.) The initial entry into the caldera
by breaking bond C-1-C-2 in 5ee and 6ze is seen as a further
progression of the torsion already inherent in the C-3-C-4 bonds
of the nonplanar cyclobutanes. The cisoid conformation of lower
enthalpy of formation, ax.eq of 5ee (-) leads to the cis (Z,E)-
diradical, (-) synclinal-1, of Figure 7. To the extent that some
further internal rotation in the direction of (-) synclinal-1/2 is
required to reach a nonbonded diradicalsa sufficient condition
for loss of Woodward-Hoffmann controlsthat of the cyano
group is preferred to that of the methallyl group in recognition
of the widely documented observation of its greater rotational
propensity, and in the guess that the propenyl group is fully
occupied in becoming the coplanar allyl radical responsible for
the reactions occurring at these lower temperatures. This is not
to say that rotation of the propenyl group during the bond-
breaking process to a similar synclinal diradical does not occur,
only that it is a less likely play.
parable amount, and 7ee, forbidden-ai (6%) as the minor
product, while the cis isomer 5ee leads to 8ez, forbidden-sr
(66%), 7zz, allowed-si (13%), 8ze, forbidden-ai (16%), and 7ee,
again the minor product, allowed-ar (5%) (see Figure 5). The
fact is both educts show much the same distribution of products
within likely experimental uncertainties without regard to any
dictat by orbital symmetry rules of Woodward and Hoffmann.
This disregard is also apparent in the major products of ring-
enlargement of optically active cis- and trans-1-cyano-2-
propenylcyclopropane,23 which closely parallel the cyclobutane
results of this paper. More satisfying are the closely similar
results obtained by Baldwin and Burrell in their awesome
resolution of the system of cis- and trans-1-methyl-2-propenyl-
cyclobutanes by means of optical activity as the informant.3h
These are reproduced in Figure 6 in parentheses. The small
differences from the cyano results are trivial in comparison to
the overall agreement. The indicated compression seems not
unreasonable in light of the smaller rotational propensity of
9
J. AM. CHEM. SOC. VOL. 124, NO. 39, 2002 11649

A R T I C L E S
Doering et al.
Figure 8. Four of a large number of paths for the formation of the four
1-cyano-2-methylcyclohexenes from the two conformations of trans-1-
cyano-2-trans-propenylcyclobutane 6ze. The designations (-) and (+) refer
to the helical chirality of the conformations, not optical chirality. Steric
Figure 7. Four of a large number of paths for the formation of the four
1-cyano-2-methylcyclohexenes from the two conformations of cis-1-cyano-
2-trans-propenylcyclobutane 5ee. The designations (-) and (+) refer to
the helical chirality of the conformations, not optical chirality. Steric energies
energies (StEn) from molecular mechanical calculations are in kcal mol-1
.
(StEn) from molecular mechanical calculations are in kcal mol-1
.
In similar fashion, the lower energy conformation, eq.eq, of
6ze (-) leads preferentially to the cis (Z,E)-diradical, (-)
synclinal-2 (Figure 8), which, in contrast to 5ee (-), requires a
rotation of the cyanomethyl group to produce (-) synclinal-1,
and then the thermodynamically more stable 8ez (+). The higher
energy conformation, ax.ax, of 6ze (+) similarly requires an
internal rotation, but the less favorable one of the methallyl
group, and then leads directly without a further internal rotation
of the cyanomethyl to the more stable product 8ze (-).
It is a quality of the (-) synclinal diradicals that the cyano-
methyl radical is more or less equally well situated to form a
four-membered or a six-membered ring by bonding to one or
the other of the terminal atoms of the methallyl radical.
Proceeding without further internal rotations beyond the torsions
needed to accommodate the formation of the six-membered ring
(in its higher energy, ax.ax conformation) leads to the thermo-
dynamically more stable product, 8ez (+). The minor product,
7zz (+), is not only the thermodynamically less stable, but
requires for its formation an internal rotation of the cyanomethyl
group by way of (-) synclinal-2 and 7zz (-).
Breaking the weakest bond in the higher energy conformation,
eq.ax, of 5ee (+) leads to cis, (E,E)-diradical, (+) synclinal-5
and gains entry to the minor pair, 8ze (-) and 7ee (-). In
contrast to (-) synclinal diradicals, (+) synclinal diradicals have
a substantially longer separation between the cyanomethyl
radical and the terminal atom of the methallyl radical, bonding
between which is required for formation of six-membered rings.
As a consequence, internal rotations are required in order to
come within bonding distance. Shortening is achievable by a
180° rotation of the methallyl radical about the C-2-C-3 bond
(rotation about the backbone returns the system to the major
(-) synclinal set, in which no such rotation is required). The
result is (+) synclinal-4. Proceeding with no further rotation
via 7ee (+) as indicated in Figure 7 leads to the thermodynami-
cally less favored 7ee (-). To reach the thermodynamically
more favored 8ze (-) via 8ze (+), a further rotation, that of
the cyanomethyl group, to (+) synclinal-3 is required. But note
that, according to the premised ordering of internal rotations,
the cyanomethyl group will already have undergone 180°
rotation to a significant extent prior to the required 180° rotation
of the methallyl group to (+) synclinal-4.
Were (-) synclinal-1/2 an intermediate in common from both
5ee (-) and 6ze (-), and the sole way to 8ez (+) and 7zz (+),
the ratio of these two products would be the same from both
educts. That this ratio is larger (5.1) from 5ee and smaller (1.74)
from 6ze is qualitatively ascribable to the path from 5ee to the
thermodynamically favored trans 8ez (+) being direct, whereas
that from 6ze requires a full 180° rotation of the cyanomethyl
group about the C-1-C-4 bond. The opposite is true for the
generation of 7zz (+). Here a C-1-C-4 rotation is required of
(-) synclinal-1 from 5ee and leads to the thermodynamically
higher energy, cis configuration of the cyanomethylcyclohexene,
whereas the path from 6ze (-) is direct. The change in ratio of
8ez (+) to 7zz (+) from 5.10 (66/13) starting from 5ee (-) to
1.74 (47/27) starting from 6ze (-) can be translated into free
energy values of 1.82 and 0.62 kcal mol-1, respectively.
Simplistically, they can then be separated into a thermodynamic
favoring of 1.22 kcal mol-1 and the disfavoring factor of a
cyanomethyl rotation of 0.60 kcal mol-1
.
We view this attempt to understand the stereochemistry of
ring-enlargement as a plausible, possibly misguided beginning,
but believe it is based on reasonable premises: the life of a
diradical is terminated whenever the two radicals come within
apposite bonding distance and orientation; internal rotational
gyrations make some stereochemical exploration possible within
9
11650 J. AM. CHEM. SOC. VOL. 124, NO. 39, 2002

Intermediate Diradicals in the Caldera
A R T I C L E S
two sidearms were placed 49.0 g (0.50 mol) of freshly sublimed maleic
anhydride, 19 mL (14.0 g, 0.12 mol) of acetophenone, and 2600 mL
of ethyl acetate (both dried with a 4 Å molecular sieve). One of the
sidearms was fitted with a calcium sulfate drying tube; the other, with
a gas inlet tube attached to an acetylene tank. Acetylene was purified
by passing it through two traps, one packed with anhydrous sodium
hydroxide pellets, the second filled with concentrated sulfuric acid. The
photolysis flask was cooled in a dry ice/acetone bath to -65 °C. The
lamp housing was cooled by circulating methanol that was maintained
at -10 °C by a FLEXI-COOL refrigerator. The reaction solution was
degassed by bubbling nitrogen through for 30 min, then saturated with
acetylene by passing the gas through the solution for 1 h. The solution
was irradiated with a 700 W, high-pressure Hanovia mercury lamp.
After the lamp was turned on, the flow of acetylene was kept around
10 psi, and a moderate flow of nitrogen maintained efficient stirring
of the reaction mixture. The reaction was monitored by removing small
portions of the reaction mixture at hourly intervals, evaporating the
solvent, and measuring the 1H NMR spectrum in CDCl3. After 26 h of
irradiation, there was no further increase in the ratio of signal intensities
of the product and maleic anhydride. Removal of the solvent by rotary
evaporator left 52.0 g of a yellow solid, which was recrystallized from
chloroform to yield 25.4 g (0.205 mol) of colorless leaflets of A
the limits of the lifetime of any particular diradical; and the
likelihood of internal, product-altering rotations is in the
decreasing order cyanomethyl > methallyl > backbone. Under
this scheme, products are determined by a principle of least
rotational involvement. The larger the number of internal
rotations (and the slower their rates) required of an initial
diradical to reach a conformation appropriate to collapse to a
designated product, the longer will have been the lifetime of
the diradical and the smaller its contribution to the total
distribution among products. The scheme makes clear why
random distribution from a common caldera is not seen in these
types of rearrangement. It focuses as does the theoretical analysis
of Doubleday on internal rotations as the decisive factor in
reactions with a common diradical as intermediate. It may
perhaps be generously described as a poor man’s trajectory
analysis!
Experimental Section
General Methods. NMR spectra were recorded on JEOL-270,
Bruker AM250, and Bruker WM300 and -500 spectrometers in CDCl3
unless otherwise stated. NMR chemical shifts and coupling constants
(J) are reported in ppm (δ), from tetramethylsilane, and hertz (Hz),
respectively. Infrared spectra (IR) were recorded on a Perkin-Elmer
model 337 grating spectrophotometer and reported in reciprocal
centimeters (cm-1). KBr sample plates were used for all the measure-
ments. Quantitative gas chromatographic analysis (GC) was effected
on a Perkin-Elmer model 990 instrument with an Autolab model
6300-01 digital integrator. Preparative GC employed a Gelmac 300
instrument. The following columns were used in the purification and
separation of samples: Column A: Carbowax 20M on Anakrom ABS
59/60, 2m × 0.25 in.; Column B: 20% DBTCB on Chromsorb P
60/80, 3m × 0.25 in.; Column C: 20% TCEPE on Chromsorb P
60/80, 3m × 0.25 in.; Column D: 20% Carbowax 20M on Chromsorb
P 60/80, 3m × 0.25 in. Melting points (mp) are measured with a
Hershberg apparatus; boiling points (bp) are uncorrected.
1
(41%): mp 86-88 °C; H NMR 6.48 (s, 1H), 4.04 (s, 1H).
(b) trans-3-trans-4-Dideuteriocyclobutane-1,2-dicarboxylic Acid
Anhydride (B). Deuteration was performed in a hydrogenation
apparatus at atmospheric pressure. A mixture of 0.40 g of 5% palladium
on barium sulfate, 42.0 g of A, and 150 mL of ethyl acetate (distilled
from calcium hydride) was stirred under deuterium at atmospheric
pressure and room temperature. Uptake of deuterium was complete after
8 h (3000 mL). Filtration followed by removal of solvent gave a
colorless solid residue which was recrystallized from chloroform to
give 39.5 g (0.309 mol) of B as colorless leaflets (88%): mp 78-80
1
°C; H NMR 3.51 (s, 1H), 2.39 (d, 1H). No cis-deuterated product
being detectable by 1H or 2H NMR analysis, a stereoisomeric purity of
∼99% was indicated for the deuterated product.
(c) trans-3-trans-4-Dideuterio-cis-2-hydroxymethylcyclobutane-
carboxylic Acid Lactone (C). Into a 250 mL, three-necked, round-
bottomed flask equipped with a pressure-equalizing dropping funnel
and a nitrogen gas inlet tube were placed 3.9 g (0.103 mol) of sodium
borohydride and 50 mL of tetrahydrofuran (THF; distilled from LiAlH4).
The flask was cooled in a dry ice/acetone bath at -65 °C. Under
nitrogen atmosphere and with stirring, a solution of 13.0 g (0.10 mol)
of B in 100 mL of dry THF was added over a period of 30 min. The
cooling bath was removed, and the mixture was continuously stirred
for a further 2 h at room temperature. The reaction was treated by slow
addition of 1.0 M HCl until all the solid had dissolved. The mixture
was extracted three times with chloroform (100, 70, and 50 mL portions,
respectively). The combined organic layers were washed with saturated
aqueous NaCl, dried with anhydrous NaSO4, and concentrated. Distil-
lation of the residue under reduced pressure gave 6.5 g (0.057 mol) of
C as a colorless liquid (57%): bp 65-66 °C/0.7 Torr; 1H NMR δ 4.28
(m, 2H), 3.11 (s, 1H), 2.12 (s, 2H).
General Method of Kinetics. Samples were transferred by micro-
syringe and sealed at 10-3 mm in dimethyldichlorosilane-treated Pyrex
ampules, with an equal volume of tridecane as internal standard and
2% of diphenylamine as inhibitor of polymerization. For heating, the
ampules were suspended in the vapors of pure compounds boiling under
reflux as previously described.21 Temperatures were monitored with
an iron-constantan thermocouple and a Leeds and Northrop No. 8686
millivolt potentiometer. The temperature gradient was found to be less
than 0.3 °C across the section of the heating bath where the ampules
were suspended, while temperature fluctuation during any run was less
than 0.3 °C. The time lag between the insertion of the ampules and the
attainment of the final vapor temperature was ∼1 min. Four sets of
reactions were run at 192.0, 198.0, 207.2, and 217.8 °C.
After specified intervals of time, ampules were removed and analyzed
quantitatively by capillary GC for the areas of products relative to
tridecane. These were converted to relative concentrations by application
of predetermined GC response factors. The products were separated
and isolated by preparative GC and analyzed quantitatively by NMR
to determine the ratios among the pairs of diastereomers. Concentrations
were converted to molar percent concentrations for purposes of
calculation. Samples of undeuterated 1 and 2 were used to determine
specific rate constants for stereochemical interconversion and frag-
mentation to cis-piperylene and acrylonitrile. The further dissection of
these rate constants was accomplished by partitioning between the pairs
of diastereomers by NMR as a function of reaction time and extrapola-
tion of the resulting linear change in the ratios to zero time.
(d) trans-3-trans-4-Dideuterio-cis-2-hydroxymethylcyclobutane-
carbaldehyde Hemiacetal (D). A 500 mL, three-necked flask, equipped
with a pressure-equalizing dropping funnel, a magnetic stirrer, and a
gas inlet tube, was flame-dried with nitrogen before use. A solution of
10.0 g (0.088 mol) of C and 100 mL of methylene chloride (distilled
from CaH2) was introduced, degassed, and cooled to -65 °C in a dry
ice/acetone bath. Under a nitrogen atmosphere and with stirring, 100
mL of 1.0 M isobutylaluminum hydride (DIBAL) in hexane was added
from the dropping funnel to the lactone over a period of 2 h. Stirring
was continued for a further 3 h at -65 °C. The reaction was monitored
by thin-layer chromatography (TLC: silica gel plate; ether/hexane )
4:1, detected by 20% SbCl6 in CCl4). At -65 °C, a solution of 1.0 M
HCl was added slowly to terminate the reaction and dissolve the solid.
The aqueous layer was separated and extracted with two 50 mL portions
(R)-1-Cyano-cis-2-(cis-propenyl)-trans-3,trans-4-dideuteriocyclo-
butane (1ee) (see Figure 1). (a) Cyclobut-3-ene-1,2-dicarboxylic Acid
Anhydride (A). Into a 3000 mL immersion-well photolysis flask with
9
J. AM. CHEM. SOC. VOL. 124, NO. 39, 2002 11651

A R T I C L E S
Doering et al.
of methylene chloride. The combined organic layers were neutralized
with aqueous NaHCO3, washed with brine, dried over anhydrous
Na2SO4, and concentrated. Fractional distillation of the residue afforded
7.0 g of D as a colorless liquid (69%): bp 80-82/0.5 Torr; H NMR
5.33 (s, 1H), 3.87 (m, 2H), 2.89 (m, 2H), 1.66 (d, 2H).
combined organic extracts were washed with water until neutral, dried
over anhydrous Na2SO4, and concentrated to a residue, which was
purified by vacuum transfer at 0.001 mmHg. Analysis by capillary GC
indicated an equilibrium ratio of 2ze to 1ee of 2.87 ( 0.05 (four runs).
The mixture was separated and each isomer was purified by preparative
GC (column D, 130 °C). 2ze: 1H NMR 5.50 (m, 1H, J ) 10), 5.36
(m, 1H, J ) 10), 3.55 (t, 1H), 2.27 (t, 1H), 2.20 (t, 1H), 1.87 (t, 1H),
1.66 (d, 3H); 2H NMR 2.20 (overlapping peaks); IR 3010(s), 2980(s),
2240(s), 2230(s), 2200(m), 1650(w), 1450(m), 1300(m), 970(m),
930(m), 900(m), 720(s).
1
(e) trans-3-trans-4-Dideuterio-cis-2-(cis-propenyl)cyclobutyl-
methanol (E). A Wittig reagent was prepared from 30.0 g of ethyl-
triphenylphosphonium iodide suspended in 150 mL of THF at -65 °C
and a 1.0 M solution of tert-butyllithium in hexane. Forced by a gentle
stream of nitrogen, a degassed solution of 7.0 g of D in 10 mL of dry
THF was added to the freshly prepared Wittig reagent through a
stainless steel needle. After completion of the addition, the mixture
was stirred for 8 h at room temperature, cooled again to -65 °C, and
acidified with aqueous 1.0 M HCl. Separated solid was extracted twice
with ether, as was the aqueous filtrate. The combined ether solutions
were neutralized with saturated NaHCO3, washed with brine, dried over
anhydrous MgSO4, and concentrated. Distillation of the residue under
reduced pressure gave 4.9 g of E as a colorless liquid (64%): bp 43-
The equilibrium ratio obtained with nondeuterated 2/1 is 2.97 (
0.06, while that from 2ze/1ee effected in DMSO-d6 (quenching with
D2O showed incorporation of deuterium exclusively at C1 as expected)
is 3.37 (one run).
(R)-1-Cyano-cis-2-(trans-propenyl)-trans-3,trans-4-dideuterio-
cyclobutane (5ee). A 30 mL Pyrex test tube with outlet attached to a
drying tube was loaded with 75 µL of GLC-purified 1ee, 140 µL of
acetophenone, and 25 mL of benzene (both dried with 3 Å molecular
sieves). Clamped next to the water-cooled lamp-housing of a 450 W,
high-pressure Hanovia mercury lamp, the reaction mixture was irradi-
ated until monitoring by capillary GC indicated the beginning of a
decline in the concentration of 5ee, at which time the ratio 5ee/1ee
was 3-4. (Extrapolation to infinite time gave a ratio of 6.5-7.0.)
Removal of benzene followed by separation and purification on column
D at 130 °C gave pure 5ee as a colorless liquid: 1H NMR 5.71 (m,
1H, J ) 16), 5.57 (m, 1H, J ) 16), 3.22-3.16 (m, 2H), 2.17-2.10
(m, 2H), 1.72 (d, 3H); 2H NMR 2.23 (s, 1H), 1.86 (s, 1H); IR 3020(m),
2970(s), 2240(s), 2210(s), 1665(w), 965(s), 920(w).
1
50 °C/0.4 mmHg; H NMR 5.57 (m, 2H), 3.69 (m, 2H), 3.40-3.55
(m, 1H), 2.65 (m, 1H), 1.70 (m, 1H), 1.60 (d, 3H).
(f) trans-3-trans-4-Dideuterio-cis-2-(cis-propenyl)cyclobutane Car-
baldehyde (F). A solution of 4.9 g (0.038 mol) of E in 10 mL of
CH2Cl2 (freshly distilled from CaH2) was added dropwise to a mixture
of 19.6 g (0.052 mol) of pyridinium dichromate and 40 mL of CH2Cl2
at room temperature over a period of 20 min. After having been stirred
for 16 h, the reaction mixture was treated with 20 mL of water and
extracted with 20 mL portions of ether and pentane. After removal of
solid material by filtration, the combined organic solutions were washed
with saturated aqueous NaHCO3 and brine, then dried over anhydrous
MgSO4. The concentrated residue was distilled in vacuo to give 3.9 g
(0.031 mol) of F as a colorless liquid (yield 81%): bp 37-45 °C at
(R)-1-Cyano-trans-2-(trans-propenyl)-cis-3,cis-4-dideuterio-
cyclobutane (6ze). The procedure for epimerization of 1ee to 2ze
described above applied to 5ee led to an equilibrium mixture, 6ze/5ee
) 3.19 ( 0.04 (nondeuterated analogue, 3.05 ( 0.04). 6ze: 1H NMR
5.54 (m, 1H, J ) 16), 5.44 (m, 1H, J ) 16), 3.14 (t, 1H), 2.72, (t, 1H),
1
1.5 mmHg; H NMR 9.75 (s, 1H), 5.50 (m, 2H), 3.60 (m, 1H), 3.30
(m, 1H), 2.30 (m, 1H), 1.80 (m, 1H), 1.70 (d, 3H).
2
(g) trans-3-trans-4-Dideuterio-cis-2-(cis-propenyl)cyclobutanecar-
baldehyde Oxime (G). To a mixture of 3.9 g of F and 4.3 g of
hydroxylamine hydrochloride in 50 mL of water was added 21.5 g of
40% aqueous K2CO3 with stirring that was continued for 2 days.
Concentration, extraction with two 20 mL portions of ether, washing
with brine, drying over anhydrous MgSO4, and concentration gave 2.5
g of crude oxime G, used further without purification: 1H NMR 8.60
(br m, 1H), 5.52 (m, 2H), 3.20 (m, 2H), 2.00 (m, 2H), 1.80 (d, 3H).
(h) (R)-1-Cyano-cis-2-(cis-propenyl)-trans-3,trans-4-dideuterio-
cyclobutane (1ee). To a solution of 3.2 g of N,N-carbonyldiimidazole
in 30 mL of CH2Cl2 (freshly distilled from CaH2) was added over a
period of 20 min a solution of 2.5 g of crude oxime G in 5 mL of
CH2Cl2. After being stirred for 2 h, the mixture was concentrated to a
residue, three ether extractions (100 mL each) of which were combined,
washed with brine, dried over anhydrous MgSO4, and concentrated to
3.1 g of a pale yellow liquid. Analysis by GLC (column A, 150 °C)
revealed three components, 1ee (90%), 5ee (10%), and a mixture
of 2ze and 6ze (<1%): 1H NMR (sample of 1ee purified by GLC)
5.72 (m, 1H, J ) 10), 5.61 (m, 1H, J ) 10), 3.49 (t, 1H), 3.27 (t, 1H),
2.12 (t, 1H), 1.83 (t, 1H), 1.66 (d, 3H); H NMR 2.15 (overlapping
peaks); IR 2980(s), 2950(s), 2230(s), 2200(m), 1660(w), 1450(m),
970(s), 930(w).
(R)-1-Cyano-cis-2-methylcyclohex-3-ene (7) and (R)-1-Cyano-
trans-2-methylcyclohex-3-ene (8) (see Figure 8). A solution of 0.77
g of hydroquinone, 5.5 g of acrylonitrile (purified by fractional
distillation), and 50 mL of piperylene (Aldrich, 65-70% purity) was
placed in a 200 mL, high-pressure reactor equipped with a heater and
mechanical stirrer. Having been heated and stirred for 14 h at 125 °C,
the reaction mixture was washed with 10% aqueous NaOH, neutralized
with 10% aqueous HCl, washed with brine, and dried over anhydrous
Na2SO4. Fractional distillation afforded 11.8 g (97% based on acrylo-
nitrile) of a colorless liquid (bp 39 ( 2 °C at 0.12 mmHg). Analysis
by capillary GC revealed 7 (40%), 8 (47%), and two minor products.
Separation and purification by GC (column C, 140 °C) afforded 8
(retention time (tR), 56 min) and 7 (tR, 70 min). 7: 1H NMR 5.71 (m,
1H), 5.49 (m, 1H), 2.92 (m, 1H), 2.47 (m, 1H), 2.25 (m, 1H), 2.10-
2.00 (m, 2H), 1.85 (m, 1H), 1.18 (d, 3H). 8: 1H NMR 5.69 (m, 1H),
5.49 (m, 1H), 2.45 (m, H), 2.32 (m, 1H), 2.13-2.06 (m, 3H), 1.84 (m,
1H), 1.17 (d, 3H).
2
2.22 (t, 1H), 2.03 (t, 1H), 1.59 (d, 3H); H NMR 2.32 (overlapping
peaks); IR 3010(s), 2960(s), 2230(s), 2200(m), 1660(m), 975(m),
920(m), 720(s).
Acknowledgment. The support of this work by the National
Science Foundation (Grants 76-24300, 80-19427, and 85-18451)
and by the Norman Fund in Organic Chemistry in memory of
Ruth Alice Norman Weil Halsband is gratefully acknowledged.
Support to the Department of Chemistry and Chemical Biology
of Harvard University for its NMR Facility comes from the NIH
(S10-RR01748 and S10-RR04870) and from the NSF (CHE
88-14019).
(R)-1-Cyano-trans-2-(cis-propenyl)-cis-3,cis-4-dideuteriocyclobu-
tane (2ze). To a solution of 2.3 mg (0.20 mmol) of freshly sublimed
potassium tert-butylate in 1.0 mL of dimethyl sulfoxide (dried over
CaH2), cooled to 15 °C, was added slowly by syringe a twice degassed
sample (50 µL) of 1ee, purified by GC (column A, 120 °C), under
nitrogen and with stirring. After 15 min of stirring, 5 µL of water was
injected to quench the reaction. After a further 5 min of stirring, 1 mL
of water and 2 mL of pentane were added. The separated aqueous layer
was extracted three times with 1.0 mL portions of pentane. The
JA0206083
9
11652 J. AM. CHEM. SOC. VOL. 124, NO. 39, 2002