A m-benzyne to o-benzyne conversion through a 1,2-shift of a phenyl group

Article abstract of DOI:10.1021/ja0213672

Pyrolysis of two differently labeled versions of 3-phenylphthalic anhydride shows that a m-benzyne can form the related o-benzyne through shift of a phenyl group. The highest energy point in the process is the transition structure for a reverse carbon-hyd

Full text of DOI:10.1021/ja0213672

Published on Web 05/01/2003
A m-Benzyne to o-Benzyne Conversion through a 1,2-Shift of
a Phenyl Group
Michael E. Blake, Kevin L. Bartlett, and Maitland Jones, Jr.*
Contribution from the Department of Chemistry, Princeton UniVersity,
Princeton, New Jersey 08544
Received November 20, 2002; Revised Manuscript Received March 14, 2003; E-mail: mjjr@princeton.edu
Abstract: Pyrolysis of two differently labeled versions of 3-phenylphthalic anhydride shows that a m-benzyne
can form the related o-benzyne through shift of a phenyl group. The highest energy point in the process
is the transition structure for a reverse carbon-hydrogen insertion in an intermediate benzopentalene.
With the minor addition of an intermediate alkyne formed through a Roger Brown rearrangement, the original
mechanism for formation of acenaphthalene accommodates the labeling results.
Scheme 1. Pyrolytic Conversion of 1,3-Diiodobenzene to
3-Hexene-1,5-diyne (1)
Introduction
Almost 40 years ago, Fisher and Lossing suggested that one
benzyne might be forming another, when they found 3-hexen-
1,5-diyne (1), a product characteristic of p-benzyne, upon
pyrolysis of 1,3-diiodobenzene.1 As the authors made clear,
however, this observation does not necessarily require a 1,2-
hydrogen shift converting m-benzyne into p-benzyne. For
example, stepwise processes, in which only one iodine is initially
lost, could produce either p-benzyne or 1 without the interme-
diacy of m-benzyne. A more likely mechanism for the conver-
sion of m- to p-benzyne is either an initial ring-opening to yield
a transient vinylidene followed by rapid rearrangement to 1 or
a concerted ring-opening and hydrogen migration, directly
transforming m-benzyne into 1 (Scheme 1),2,3 and thence to
p-benzyne. Convincing experimental evidence has yet to be
found that suggests that o-benzyne can isomerize into either
m- or p-benzyne, or vice versa.4
adjacent hydrogen atom were changed to a substituent that could
migrate more easily, these isomerizations may be observable.8
We chose to study the rearrangements of phenyl-substituted
The heats of formation at 298 K (∆Hf°,298) for o-, m-, and
p-benzyne are 106 ( 3, 122 ( 3, and 138 ( 3 kcal/mol,
respectively,5 indicating that a cascade from p- to o-benzyne
would be relentlessly exothermic. Of course, the high barrier
for similar 1,2-hydrogen shifts in simple alkyl6 and aryl7 radicals
is likely to be present substantially in the benzynes as well and
may kinetically prevent these isomerizations. However, if the
(4) (a) The isomerization of one o-benzyne into another o-benzyne via an
apparent hydrogen migration has been observed in the pyrolytic generation
of 1,6-13C2-o-benzyne as a minor pathway.4b Whether or not this proceeds
via another benzyne isomer, or even whether it is intramolecular, is a matter
of speculation. It is quite possible that, at the highest temperatures studied
(830 °C), the participation of m- and/or p-benzyne is simply not observable.
Several papers have approached this question both theoretically and
experimentally,4c and there is some evidence for the interconversion of the
parent benzynes over or through very high barriers. Changing the migrating
group to phenyl introduces the possibility of a mechanism in which the π
system of the phenyl interacts with the σ orbital of the benzyne, thus making
the rearrangement process easier. As a referee has quite correctly implied,
the price of this increased ease of investigation is lack of relevance to the
parent system, which we are explicitly not considering here. (b) Barry,
M.; Brown, R. F. C.; Eastwood, F. W.; Gunawardana, D. A.; Vogel, C.
Aust. J. Chem. 1984, 37, 1643. (c) Diau, E. W.-G.; Casanova, J.; Roberts,
J. D.; Zewail, A. H. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 1376. Wang,
H.; Laskin, A.; Moriarty, N. W.; Frenklach, M. Proc. Combust. Inst. 2000,
28, 1545. Moskaleva, L. V.; Madden, L. K.; Lin, M. C. Phys. Chem. Chem.
Phys. 1999, 1, 3967.
(1) Fisher, I. P.; Lossing, F. P. J. Am. Chem. Soc. 1963, 85, 1019.
(2) (a) Sander, W. Acc. Chem. Res. 1999, 32, 669. (b) Bapat, J. B.; Brown, R.
F. C.; Bulmer, G. H.; Childs, T.; Coulston, K. J.; Eastwood, F. W.; Taylor,
D. K. Aust. J. Chem. 1997, 50, 1159.
(3) (a) The reaction energies and barriers for the direct 1,2-hydrogen shifting
processes constituting the p- to m- to o-benzyne cascade have been
calculated for parent benzyne at the RCCSD(T)/6-31G(d,p)//B3LYP/6-31G-
(d) and G2M//B3LYP/6-31G(d) levels of theory.3b The computed barrier
for formation of o-benzyne from m-benzyne (56.3 kcal/mol) is 4.4 kcal/
mol lower than the barrier for formation of p-benzyne (60.7 kcal/mol).
This result would predict that pyrolysis of m-benzyne would predominantly
yield products from o-benzyne, rather than from p-benzyne. In addition,
neither stepwise nor concerted ring-opening and hydrogen migration
processes, as shown in Scheme 1, were considered. The concerted
H-migration and ring-opening process converting m-benzyne into 3-hexen-
1,5-diyne is predicted to have a barrier of approximately 50 kcal/mol at
the B3LYP/6-31G(d,p) level.2a (b) Moskaleva, L. V.; Madden, L. K.; Lin,
M. C. Phys. Chem. Chem. Phys. 1999, 1, 3967.
(5) Wenthold, P. G.; Squires, R. R.; Lineberger, W. C. J. Am. Chem. Soc.
1998, 120, 5279.
(6) Wilt, J. W. In Free Radicals; Kochi, J. K., Ed.; Wiley: New York, 1973;
Chapter 8.
(7) Brooks, M. A.; Scott, L. T. J. Am. Chem. Soc. 1999, 121, 5444.
9
10.1021/ja0213672 CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 6485-6490
6485

A R T I C L E S
Blake et al.
Scheme 2. Formation of o-Benzyne and Its Potential
Interconversion with m-Benzyne
Scheme 3. Mechanism for Conversion of 3-Phenyl-o-benzyne (3)
into Benzopentalene (6) and Acenaphthalene (9)
o-benzyne9 (3), made through pyrolysis of 3-phenylphthalic
anhydride (2) (Scheme 2). The phenyl substituent can be
reasonably expected10 to undergo more facile migration than
hydrogen.11 Indeed, 1,2-migrations of phenyl groups to radicals
are known in aromatic systems.12,13 However, by entering the
potential equilibration from the o-benzyne side, we do take the
risk that the lower energy o-benzyne will not be able to climb
the energetic hill to the higher energy m-benzyne before passing
on to its ultimate fate.
Scheme 4. Predicted Labeled Products from Pyrolysis of
[3-13C]-3-Phenylphthalic Anhydride (2a)
Results and Discussion
The flash vacuum pyrolysis of 2 has been used previously14
to study the thermal chemistry of 3, which yields cyclopent[a]-
indene (benzopentalene, 6) as the primary rearrangement
product, with acenaphthalene (9) being formed by a subsequent
rearrangement of 6. The generally accepted mechanism is shown
in Scheme 3.15
To observe phenyl migration in 3, a label must be placed
that would be scrambled by the isomerization of o- to m-benzyne
and back.16 Our initial experiment involved 13C-labeling the ipso
carbon in 3, which would scramble via o- to m-benzyne
isomerization, to give labeled 6 and 9 as shown in Scheme 4.
In the products formed following o- to m-benzyne intercon-
version, the label ends up as one of the two triply bonded
(8) In fact, the migration of a TMS group in a p-benzyne to form a m-benzyne
has been suggested in the high-temperature pyrolysis of a TMS-substituted
enediyne: Johnson, G. C.; Stofko, J. J., Jr.; Lockhardt, T. P.; Brown, D.
W.; Bergman, R. G. J. Org. Chem. 1979, 44, 4215.
(9) (a) It may seem more sensible to generate 4, rather than 3, because the
desired phenyl shift would then be exothermic. Unfortunately, it can always
be argued that stepwise decomposition of any potential precursor to 4 could
conceivably bypass 4 and still give 3. Accordingly, we started instead with
labeled 3 and hoped that the endothermic isomerization to 4 would compete
with the formation of the ultimate products. (b) In ref 2b, Brown et al.
formed 4 by pyrolysis (1000-1100 °C) of diallyl biphenyl-2,6-dicarboxylate
and observed only a small yield of products nominally deriving from
3-phenyl-o-benzyne. Approximately 85% of the products were unidentified
aromatic compounds.2b
carbons in the resultant o-benzyne, which are equilibrated via
2-phenylcyclopentadienylidene carbene.17,18 Thus, in the primary
rearrangement product, 6, there are two positions that should
each have 50% of the label. In the secondary product, 9, an
additional symmetry element is gained, as the two positions
become equivalent, and the 1-position of 9 is predicted to receive
100% of the label.
It has been suggested18 that the scrambling of the two triply
bonded carbons in o-benzyne is not the result of equilibration
via cyclopentadienylidene carbene, but occurs through a step-
wise decomposition of the anhydride. Fortunately, it matters
not for this experiment whether the benzyne or cyclopentadi-
enylidene carbene is formed initially, as there is no simple
mechanism that accomplishes the label scrambling shown in
Scheme 4 directly from the cyclopentadienylidene carbene. It
is certainly possible to construct alternative schemes that do
generate 4 from 2 in a stepwise manner without the intermediacy
of 3. However, the reformation of 3 from 4 is not subject to
(10) Wilt, J. W. In Free Radicals; Kochi, J. K., Ed.; Wiley: New York, 1973;
p 346 ff.
(11) We completely ignore the question of whether the phenyl migration occurs
in one or two steps.
(12) Preda, D. V.; Scott, L. T. Org. Lett. 2000, 2, 1489.
(13) Preda, D. V.; Scott, L. T. Polycyclic Aromat. Compd. 2000, 19, 119.
(14) (a) Reference 2b. (b) Brown, R. F. C.; Choi, N.; Eastwood, F. W. Aust. J.
Chem. 1995, 48, 185. (c) Brown, R. F. C.; Choi, N.; Coulston, K. J.;
Eastwood, F. W.; Wiersum, U. E.; Jenneskens, L. W. Tetrahedron Lett.
1994, 35, 4405.
(15) Scheme 3 is essentially the mechanism given14 by Brown, with one addition
and one omission. The cis/trans isomerization step is not present (although
obviously implied) in refs 2b, 14b, or 14c. It is included here in Scheme
3 for clarity. Also, in refs 2b and 14b, an additional possible pathway
between 6 and 9 is presented, single bond scission in 6 to yield a diradical.
This step is not included in Scheme 3, as it seems highly unlikely that the
required diradical intermediate could compete energetically with the
vinylidene, especially given the well-known rearrangement between vi-
nylidenes and terminal acetylenes.
(16) In ref 2b, Bapat, et al. also studied the FVP (1020 °C) of [2,3-13C2]-3-
phenylphthalic anhydride and did not report any products of phenyl
migration in the doubly labeled o-benzyne intermediate 3. It is possible
that phenyl migration did not occur in 3 under their conditions or that the
doubly labeled products resulting from a small amount of scrambled 3 were
overlooked. Their pyrolyses were performed on samples that were diluted
with unlabeled 2, such that only 10% was doubly labeled.2b
(17) (a) See ref 4b. (b) Brown, R. F. C.; Coulston, K. J.; Eastwood, F. W.;
Vogel, C. Aust. J. Chem. 1988, 41, 1687.
(18) Wentrup, C.; Blanch, R.; Briehl, H.; Gross, G. J. Am. Chem. Soc. 1988,
110, 1874.
9
6486 J. AM. CHEM. SOC. VOL. 125, NO. 21, 2003

m-Benzyne to o-Benzyne Conversion
A R T I C L E S
Figure 2. Isotopic enrichment over natural abundance in 13C-labeled 6,
produced by pyrolysis of 2b at 900 °C and ∼0.1 Torr.
Scheme 6. Synthesis of [6-13C]-3-Phenylphthalic Anhydride (2b)
Figure 1. Isotopic enrichment over natural abundance in 13C-labeled 6,
produced by pyrolysis of 2a at 900 °C and ∼0.1 Torr.
Scheme 5. Synthesis of [3-13C]-3-Phenylphthalic Anhydride (2a)
to positions 3 and 12.22 Although the first wave of Scott
rearrangements cannot populate position 1, such a transfer will
become important in future discussions of other labeling
experiments. Accordingly, we see position 1 as having 6-7%
excess label, which means that 12-14% of the reaction involved
symmetrization through m-benzyne 4.
these concerns. However we get to 4, the return to 3 must
involve a benzyne-to-benzyne rearrangement.
The necessary labeled precursor, [3-13C]-3-phenylphthalic
anhydride (2a), was synthesized (Scheme 5) following the
method of Sissman19 as elaborated by Brown.14b A label was
introduced at the ipso position by starting with labeled benzal-
dehyde, which was itself made from commercially available
labeled benzoic acid.
A second labeling experiment was subsequently performed,
in which the “para” position in the anhydride precursor was
13C-labeled (2b). The synthesis of 2b was accomplished by a
simple Wittig reaction on cinnamaldehyde (Scheme 6).
The standard mechanism predicts that the label should appear
in the 5 position without scrambling and, if scrambling via
m-benzyne 4 occurs, in the 3 position as well (Scheme 3).
Pyrolysis of 2b at 900 °C and ∼0.1 Torr yielded acenaphthalene
in which the excess label was distributed as shown in Figure 2.
The isotopic enrichments shown in Figure 2 clearly show
that Scheme 3 does not account for all chemical processes
occurring in the pyrolysis. The scrambling of positions 4 and 5
in 9 is not predicted by the mechanism given in Scheme 3.
Fortunately, there is a simple addition to the mechanistic picture
that accounts for the observed scrambling. It has been shown
that under pyrolytic conditions, 1-phenyl-1-buten-3-yne under-
goes scrambling of the acetylenic carbons via a vinylidene by
a process consisting of both a hydrogen migration and a styryl
migration.23 This process is analogous to the acetylenic scram-
bling observed in 13C-labeled phenylacetylene, the Roger Brown
rearrangement.24 Thus, 7 and 8, the two vinylidenes shown in
Scheme 3, can reversibly form the corresponding alkyne by
In the analysis of our pyrolysate, we chose to examine only
the labeled acenaphthalene, 9. Under our conditions, the major
product,14 6, was not isolated; rather, it dimerized quickly when
brought to room temperature.14c Given that three possible
isotopomers of 6 are predicted in Scheme 4, there are six
possible isotopomers of the dimer. In accordance with this
prediction, we found the 13C spectrum of the labeled benzo-
pentalene dimer to be too complex to be evaluated. No attempts
were made to perform low-temperature NMR spectroscopy on
6 itself, or to trap 6 to give an analytical sample,14c because, in
theory, all necessary information can be derived from the 13C
spectrum of labeled 9.
When 2a was sublimed through a quartz tube at 900 °C and
∼0.1 Torr, the isolated acenaphthalene was 13C-enriched as
shown in Figure 1.
As can be seen in Figure 1, most of the label appears at the
9 position, exactly where the conventional mechanism would
place it. Yet there is significant enrichment at position 1,
indicating the some symmetrization has occurred, most likely
through m-benzyne 4, as shown in Scheme 4. The baseline
enrichment of positions that “cannot” receive a 13C label seems
to be about 2-3%.20 In addition, at temperatures g900 °C,
significant label “leaks” to other positions, exactly as predicted
by the Scott “6-5” benzenoid ring-contraction-ring-expansion
mechanism.21 In this process, the skin of acenaphthalene rotates
about the central quaternary carbon 11 four positions clockwise
or counterclockwise. At 900 °C, the Scott rearrangement appears
to be transferring approximately 1% of the label at position 9
(20) Position 11 of 9 (the central carbon) should not receive any 13C-enrichment,
either by the scrambling shown in Scheme 4 or by the ratcheting mechanism
in 9.21 The reported enrichment percentages are based on the 13C NMR
spectrum acquired with a 120 s delay, in which the longest 13C T1 in 9 was
estimated to be approximately 12 s. The overestimation of 13C-label in
positions that should not receive any enrichment is likely the result of errors
in comparing the relative integrations.
(21) Scott, L. T.; Roelofs, N. H. Tetrahedron Lett. 1988, 28, 6857.
(22) Professor Scott has kindly informed us that pyrolysis of 2a-13C-labeled
acenaphthalene above 900 °C reveals exactly that: specific leakage of the
label to positions 3 and 12 in the first generation product. Racoveanu, A.;
Scott, L. T., personal communication with permission to cite.
(23) Schulz, K.; Hofmann, J.; Zimmermann, G.; Findeisen, M. Tetrahedron Lett.
1995, 36, 3829.
(19) Sissman, E. E.; Murray, R. J.; McChesney, J. D.; Houston, L. L.; Pazdernik,
(24) Brown, R. F. C.; Eastwood, F. W.; Harrington, K. J.; McMullen, G. L.
Aust. J. Chem. 1974, 27, 2393.
T. L. J. Med. Chem. 1976, 19, 148.
9
J. AM. CHEM. SOC. VOL. 125, NO. 21, 2003 6487

A R T I C L E S
Blake et al.
true diradicals, and a multireference method would require a
(16/16) active space for the benzyne isomerization and a (14/
14) active space for the cis/trans isomerization, putting the
computational cost too high. In addition, the benzyne isomer-
ization does not lend itself to approximation by model compound
investigation. Thus, although a computational value for the
activation enthalpy for the phenylbenzyne isomerization is
obviously of great interest, we are unable (at present) to provide
one. For the ethynylbenzofulvene isomerization (10a to 10b),
an approximate enthalpic barrier of ∼41 kcal/mol was obtained
from multireference calculations on a model compound, which
gives the bracketed numbers shown for that process in Figure
3.26 Second, although the equilibrium between terminal alkynes
and vinylidenes has become very commonly invoked in explain-
ing the pyrolytic chemistry of alkynes,23,24,27 it appears that, at
least at the level of theory employed here, the vinylidene
structures (7 and 8) produced from migration of the hydrogen
in ethynylbenzofulvenes 10a and 10b are not even minima. For
7b, 8a, and 8b, B3LYP/6-311+G** optimizations resulted in
energetic minima which, upon inclusion of zero-point energy
and integrated heat capacity, were higher in enthalpy than their
corresponding transition structures for formation of 10a or 10b.28
Vinylidene 7a was not even found to be an energetic minimum.
Rather, the energy given in Figure 3 for 7a corresponds to the
relative electronic energy of an inflection point where the slope
of energy versus atomic displacement is, while still negative,
closest to zero.29 That is, the C-H insertion reactions that
produce benzopentalene (6) and acenaphthalene (9), from trans-
(10a) and cis-ethynylbenzofulvene (10b), respectively, each
proceed over a single transition structure through species that
are electronically vinylidenes, but which are not energy minima.
Attempts were made to identify the energetic minima on either
side of the transition state, but IRC calculations failed. That no
local minimum exists for these vinylidene “intermediates” has
been noticed before by Cioslowski et al.30
Figure 3. Computed enthalpies at 900 °C, relative to 3-phenyl-o-benzyne
(3), in kcal/mol. Nonitalicized numbers are at B3LYP/6-311+G**; italicized
numbers are at CCSD(T)/cc-pVDZ.
either H-migration or vinyl migration prior to formation of
acenaphthalene by C-H insertion. In addition, this terminal
acetylene scrambling could occur either before or after the cis/
trans isomerization of the exocyclic alkene. It is this process
that accounts for the nearly perfect scrambling of the label
between positions 4 and 5 in the pyrolysis of 2b.
Position 3 shows a slight enrichment of 6%, 3-4% above
the baseline, which corresponds to 6-8% scrambling through
m-benzyne. Interestingly, position 1 also shows nonnegligible
enrichment. Although it is fed by Scott rearrangement from both
positions 4 and 5,21,22 that process should only account for an
additional 1-2%, based on the first labeling experiment. Thus,
we cannot fully account for the amount of label observed in
position 1.
In addition to the labeling studies detailed above, we analyzed
the proposed mechanistic pathway computationally. All geom-
etries were optimized at the B3LYP/6-311+G** level of theory,
and the electronic energies were refined by performing CCSD-
(T)/cc-pVDZ single point calculations for each stationary point.
The computational results are summarized in Figure 3.
Several points are worth mentioning. First, because of the
size of the system, believable enthalpies were not available for
the two processes that require significantly diradicaloid transition
structures or intermediates - the o- to m-benzyne isomerization
(3 to 4) and the cis/trans isomerization of ethynylbenzofulvene
(10).25 The restricted density functional theory (DFT) method
B3LYP we employed for our geometry optimizations fails for
(25) (a) Both the o- and the m-benzynes have some diradical character, and
RB3LYP will almost certainly overestimate the zwitterionic nature of the
wave function, resulting in predictable geometric deviations and less
predictable energetic fluctuations.25b This failure is a quantitative, as opposed
to the qualitative, failure of restricted DFT to compute the wave functions
of pure diradicals, such as the transition structures and/or intermediates
involved in the o- to m-phenylbenzyne isomerization and the terminal alkyne
π-bond rotation. The performance of DFT for the vinylidenic structures
should be much better, as they are well described by a single configuration.25c
Regardless of the failures of DFT to give reliable energetics, the energies
should be improved by the use of CCSD(T). (b) Cramer, C. J.; Nash, J. J.;
Squires, R. R. Chem. Phys. Lett. 1997, 277, 311. (c) Chen, W.-C.; Yu,
C.-h. Chem. Phys. Lett. 1997, 277, 245.
(26) The model compound used to approximate the terminal alkyne isomerization
was ethynylfulvene. Two 1-A′ Cs geometries were optimized at the (10/
10)CASSCF/6-311+G** level of theory: one with the alkyne in the plane
of the molecule, and one with the alkyne rotated 90°. Vibrational analysis
showed the planar geometry to be a minimum and the rotated geometry to
be a transition state (exactly one mode with an imaginary frequency). The
electronic energies were refined by performing (10/10)CASPT2/6-311+G**
single point calculations at the CASSCF optimized geometries. The
CASSCF zero-point energies and integrated heat capacities were added to
the CASPT2 energies to give a value of ∆Hq ) 40.8 kcal/mol for rotation
about the fulvene double bond in ethynylfulvene.
(27) See, and references therein: (a) Brown, R. F. C. Eur. J. Org. Chem. 1999,
3211. (b) Brown, R. F. C. Recl. TraV. Chim. Pays-Bas 1988, 107, 655.
(28) The electronic energy minima for these vinylidenes are extremely shal-
low: 0.2 kcal/mol for 7b, 0.2 kcal/mol for 8a, and 0.1 kcal/mol for 8b, all
at the B3LYP/6-311+G** level of theory. The CCSD(T)/cc-pVDZ relative
electronic energies for these B3LYP saddle point geometries are very
similar: 0.3 kcal/mol for 7b, 0.2 kcal/mol for 8a, and 0.1 kcal/mol for 8b.
(29) This inflection point was obtained by performing a relaxed scan of the
C-C-H angle involving the two vinylidenic carbons and the migrating
hydrogen atom, with 1° increments. At the approximate inflection point,
the C-C-H angle was 91.6°, and the electronic energy (B3LYP/
6-311+G**) was -462.042301 hartrees. The slope at this point was
approximately -0.005 kcal/mol per degree.
9
6488 J. AM. CHEM. SOC. VOL. 125, NO. 21, 2003

m-Benzyne to o-Benzyne Conversion
A R T I C L E S
zofulvene, and that the required hydrogen atom migration has
virtually not even begun in the transition structures, they could
also be viewed as fairly “early” transition structures.
It is always possible, of course, that systematic error inherent
in the optimization method is resulting in artifactual behavior
and that the vinylidenes are, in fact, minima. Lineberger and
co-workers found a 3.4 kcal/mol classical barrier for hydrogen
migration in vinylvinylidene, employing CISD/D95(d,p).32 They
also estimated the lifetime of singlet vinylvinylidene to be 20-
200 fs by analysis of the photoelectron spectrum of its negative
ion.32a On the other hand, Schaefer and co-workers found that,
at the highest level investigated (CCSD(T)/TZ2P//CCSD/TZ2P),
ethynylvinylidene has a vanishingly small classical barrier for
formation of diacetylene.33 A fairly systematic comparison of
several DFT methods25c with the estimated coupled-cluster
complete basis set limiting value34 found that, at B3LYP/aug-
cc-pVTZ, the classical barrier height for the isomerization of
parent vinylidene to acetylene was overestimated by 1.0 kcal/
mol (with B3LYP/aug-cc-pVTZ, the barrier was 3.9 kcal/mol,
whereas the coupled-cluster CBS limit was 2.9 kcal/mol).35
Therefore, the method error associated with B3LYP apparently
works to raise the barrier associated with hydrogen migration
in vinylidenes rather than lower it, and the finding that the
vinylidenes in Figure 3 are not minima is probably not
artifactual. As mentioned earlier, this observation is not new.30
Comparison of the computed potential energy surface with
the experimental results shows an overall qualitative agreement.
The retro C-H insertion from benzopentalene (6) to form trans-
ethynylbenzofulvene (10a) is computed to be the highest point
along the reaction path by about 2 kcal/mol. This result is
consistent with the observation that benzopentalene is the major
product of the pyrolysis.14 The final C-H insertion step to form
acenaphthalene (9) is computed to be slightly higher than the
barriers for carbon scrambling in ethynylbenzofulvene (10):
about 2 kcal/mol with CCSD(T) and 3 kcal/mol with B3LYP.
This calculation is in qualitative agreement with the observation
that positions 4 and 5 are scrambled in the acenaphthalene
product (see Figure 2).36 Although we were unable to calculate
the barrier for phenyl migration in o-benzyne, the observation
of 6-14% scrambling through m-benzyne indicates that the
barrier is slightly higher than that for the benzopentalene retro
Figure 4. Two views of the transition structures for scrambling the two
acetylenic carbons in 10a (left) and 10b (right). Geometries were optimized
at B3LYP/6-311+G**. Bond lengths are in Ångstroms, and angles are in
degrees.
Scheme 7. Mechanism by Which the Acetylenic Positions Are
Scrambled in the Pyrolysis of 2b
In addition to the C-H insertion reactions, the process by
which the two acetylenic carbons in 10a and 10b are equilibrated
(indicated by the curved arrows in Figure 3) was also found to
occur through a single transition structure, rather than the two-
step process shown in Scheme 7 and postulated in refs 23 and
24. The transition structures for the scrambling processes in 10a
and 10b are shown in Figure 4.
The structures shown in Figure 4 clearly show that the
“styryl” migration (in this case, a “benzofulvenyl” migration)
is a nonplanar process in which the migrating sp2 carbon orients
its p orbital along the long axis of the alkyne, allowing a
favorable interaction between the exocyclic fulvene π-bond and
one of the π* molecular orbitals in the alkyne. This interaction
is also evident in the length of the triple bond, which at 1.274
Å is considerably longer than those in 10a and 10b (1.207 and
1.208 Å).31 These transition structures could be considered “late”
in that the terminal carbon of the alkyne is closer to the sp2
migrating carbon (1.598 and 1.588 Å) than is the internal carbon
(1.778 and 1.789 Å). That is, the benzofulvenyl migration is
more than half complete. Yet when it is considered that there
is no energetic minimum prior to the scrambled ethynylben-
(31) This lengthening is also more significant than in the transition structure
connecting vinylidene with acetylene. At the B3LYP/6-311+G** level of
theory, the C-C distance in the transition structure is 1.250 Å, whereas it
is 1.199 Å in acetylene. Thus, the triple bond is lengthened by 0.051 Å in
the transition structure connecting vinylidene and acetylene, but by 0.067
Å in the transition structure for acetylenic carbon scrambling in trans-
ethynylbenzofulvene, 10a.
(32) (a) Gunion, R. F.; Ko¨ppel, H.; Leach, G. W.; Lineberger, W. C. J. Chem.
Phys. 1995, 103, 1250. (b) The CISD/D95(d,p) calculations reported in
ref 28a gave a nonplanar transition structure for hydrogen migration in
vinylvinylidene. They report that a planar transition structure optimization
resulted in a second-order saddle point. We observed no such tendency
toward nonplanarity in the hydrogen migration transition structure, in that
all three transition structures we found for hydrogen migration reactions
were planar and had exactly one mode with an imaginary frequency. (c)
In ref 33, Schaefer and co-workers calculated classical barriers of 2.7 kcal/
mol at CISD/DZP and 2.6 kcal/mol at CISD/TZ2P for hydrogen migration
in ethynylvinylidene, but 0.05 kcal/mol at CCSD(T)/TZ2P//CCSD/TZ2P.
It is possible that the barrier for hydrogen migration in vinylvinylidene
obtained by CISD/D95(d,p)32a would drop when higher correlated methods
and/or larger basis sets were used.
(33) Collins, C. L.; Hu, C.-H.; Yamaguchi, Y.; Schaefer, H. F., III. Isr. J. Chem.
1993, 33, 317.
(34) Chang, N.-Y.; Shen, M.-Y.; Yu, C.-h. J. Chem. Phys. 1997, 106, 3237.
(35) When using the 6-311+G** basis set, we found that the overestimation
by B3LYP is even larger. The classical barrier for the hydrogen migration
in vinylidene was computed to be 4.4 kcal/mol at B3LYP/6-311+G**,
1.5 kcal/mol larger than the coupled cluster CBS limit.
(30) Cioslowski, J.; Schimeczek, M.; Piskorz, P.; Moncrieff, D. J. Am. Chem.
Soc. 1999, 121, 3773. We thank Professor L. T. Scott for alerting us to
this reference. Sander, W.; Exner, M.; Winkler, M.; Balster, A.; Hjerpe,
A.; Kraka, E.; Cremer, D. J. Am. Chem. Soc. 2002, 124, 13072.
9
J. AM. CHEM. SOC. VOL. 125, NO. 21, 2003 6489

A R T I C L E S
Blake et al.
Flash Vacuum Pyrolysis of Anhydrides 2a and 2b.14 Typically,
50 mg of anhydride (2a or 2b) was sublimed into a hot quartz tube
(750-1100 °C) under reduced pressure (∼100 mTorr). The pyrolysate
was condensed on a dry-ice-cooled surface. The receiving vessel was
washed with CH2Cl2 (10 mL), and the solvent was removed. The
reaction mixture was purified by flash column chromatography. The
column was eluted first with hexane (250 mL), giving acenaphthylene
9 as a yellow solid possessing NMR spectroscopic properties identical
to those reported by the Brown group.2b 1H NMR (500 MHz; CDCl3):
δ (ppm) 7.71 (2H, dd, J ) 8, 1 Hz, H5,6), 7.59 (2H, dd, J ) 7, 1 Hz,
H3,8), 7.44 (2H, dd, J ) 8, 7 Hz, H4,7), 7.01 (2H, s, H1,2). 13C NMR
(125 MHz; CDCl3): δ (ppm) 139.5 (C9,10), 128.1 (C12), 128.0 (C11),
129.2 (C1,2), 127.6 (C4,7), 127.1 (C5,6), 124.0 (C3,8). The column
was then treated with a mixture of CH2Cl2 in hexane (1:20; 800 mL),
giving the dimer of benzopentalene 6 as a yellow oil possessing NMR
spectroscopic properties identical to those reported by the Brown
group.2b 1H NMR (500 MHz; CDCl3): δ (ppm) 7.33-7.21 (4H, s, H1,-
10; H2,9), 7.18-7.08 (4H, s, H3,8; H4,7), 6.89 (2H, t, J ) 2 Hz, H11,-
12), 6.49 (2H, s, H5,6), 3.79 (2H, br d, J ) 2 Hz, H5a,5b), 3.22 (2H,
br d, J ) 4 Hz, H11a,11b). 13C NMR (125 MHz; CDCl3): δ (ppm)
155.7 (C4a,6b), 151.9 (C4b,6a), 150.1 (C10a,12a), 134.1, 129.8 (C11a,-
11d), 128.3, 123.6, 122.5, 120.8 (C5,6), 116.7 (C11,12), 58.0 (C5a,-
5b), 40.8 (C11b,11c).
C-H insertion step. Manipulation of the Eyring equation37 gives
a value of ∆∆G1173q ) 6.4 kcal/mol for 6% scrambling and 4.2
kcal/mol for 14%. Therefore, we can provide an estimate of
46-52 kcal/mol for the barrier to phenyl migration in o-benzyne.
Conclusion
The labeling results described in these two sets of experiments
show that m-benzene 4 is able to form o-benzyne 3 through
shift of a phenyl group. Although they are highly suggestive,
they do not absolutely demand that 3 be able to form 4 through
a phenyl shift. They also reveal the need to elaborate slightly
Roger Brown’s original mechanism for the formation of
acenaphthalene from 3-phenylphthalic anhydride to include,
ironically, an alkyne formed through a Roger Brown rearrange-
ment.
Experimental Section
General. NMR spectra were recorded on Varian Mercury 300,
Varian-Unity INOVA 400, and Varian-Unity INOVA 500 NMR
spectrometers. Chemical shifts are reported in parts per million (ppm,
δ) and were indirectly referenced to TMS (tetramethylsilane) by
employing known solvent resonances as internal standards. Coupling
constants (J) are reported in Hertz (Hz).
Acknowledgment. We thank the Universiy of Washington
Chemistry Department for generous grants of computer time.
This work was funded by the National Science Foundation
through grant CHE-0073373. We thank Professor L. T. Scott
for useful comments and unpublished results, and Professor
Robert A. Pascal, Jr. for helpful advice.
Computational Methodology. All geometry optimizations and
vibrational analyses were performed using the B3LYP38 density
functional method and the standard Pople basis set 6-311+G**. All
structures (except for 7a) were shown to be electronic minima (no
vibrational modes with imaginary frequencies) or transition structures
(exactly one vibrational mode with an imaginary frequency). The
energies of the B3LYP optimized geometries were refined by perform-
ing CCSD(T)39 single-point calculations, using Dunning’s cc-pVDZ
basis set.40 The zero-point and thermal corrections from the B3LYP
vibrational analyses were used without scaling to convert the B3LYP
and CCSD(T) electronic energies to enthalpies at 1173 K. All of the
calculations were carried out with the Gaussian 98 package of electronic
structure programs.41
Supporting Information Available: Full experimental details
on the synthesis of all labeled materials for pyrolysis. Included
for each reported calculated stationary point are Cartesian
coordinates (Ångstroms), the CCSD(T)/cc-pVDZ electronic
energy (hartrees), the B3LYP/6-311+G** electronic energy
(hartrees), symmetry, and, from the B3LYP vibrational analysis,
zero-point energy (ZPE) correction (hartrees), thermal correction
to enthalpy (hartrees), and lowest energy vibration and its
symmetry (negative implies a vibration with an imaginary
frequency) (PDF). This material is available free of charge via
the Internet at http://pubs.acs.org.
(36) The ∆∆Hq ) 2-3 kcal/mol difference between scrambling and acenaph-
thalene formation would not actually account for the extent of scrambling
at 900 °C, which is roughly 95%. Manipulation of the Eyring equation
gives k1/Ik2 ) exp(∆∆Gq/RT). 95% scrambling corresponds to a relative
rate of 19:1, so at 1173 K, ∆∆Gq ) 6.9 kcal/mol. An error of 1-2% in the
NMR integrations could stretch this to a relative rate of 10:1, but that still
comes to ∆∆Gq ) 5.4 kcal/mol. Thus, while the computational results agree
qualitatively, there is a slight quantitative disagreement. Of course, the
reported relative enthalpies at 900 °C neglect a sizable contribution from
entropy. When the relative Gibbs free energies at 900 °C are compared,
JA0213672
q
q
the ∆∆G1173 ) 4.9 and 5.6 kcal/mol with B3LYP, and ∆∆G1173 ) 4.4
and 4.1 kcal/mol with CCSD(T). In addition, a statistical effect favors
scrambling, because it can occur from either of the two isomers of 10.
(37) For scrambling of 6-14%, the relative reaction rates are 15.7:1 and 6.1:1,
(41) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.;
Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng,
C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.;
Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon,
M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.7; Gaussian,
Inc.: Pittsburgh, PA, 1998.
and thus ∆∆G1173 ) 6.4 and 4.2 kcal/mol, respectively.36
(38) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (b) Lee, C.; Yang, W.;
Parr, R. G. Phys ReV. B 1988, 37, 785. (c) Miehlich, B.; Savin, A.; Stoll,
H.; Preuss, H. Chem. Phys. Lett. 1989, 157, 200.
q
(39) (a) Cizek, J. AdV. Chem. Phys. 1969, 14, 35. (b) Purvis, G. D.; Bartlett, R.
J. J. Chem. Phys. 1982, 76, 1910. (c) Scuseria, G. E.; Janssen, C. L.;
Schaefer, H. F., III. J. Chem. Phys. 1988, 89, 7382. (d) Scuseria, G. E.;
Schaefer, H. F., III. J. Chem. Phys. 1989, 90, 3700. (e) Pople, J. A.; Head-
Gordon, M.; Raghavachari, K. J. Chem. Phys. 1987, 87, 5968.
(40) Woon, D. E.; Dunning, T. H., Jr. J. Chem. Phys. 1993, 98, 1358.
9
6490 J. AM. CHEM. SOC. VOL. 125, NO. 21, 2003