Effect of para substituents on the rate of bond shift in arylcyclooctatetraenes

Article abstract of DOI:10.1021/jo010526h

The rate constants for bond shift (k_BS) in phenylcyclooctatetraene (1b) and its p-nitro and p-methoxy analogues (1a and 1c, respectively) in THF-d₈ were determined by dynamic NMR spectrometry to be identical, but k_BS is eight times greater at 280 K relative to 1b when the para substituent is cyclooctatetraenyldipotassium (2^2-/2K⁺). These results are discussed in the context of (a) possible, intrinsically small substituent effects (as determined by ¹³C chemical shifts in the ground state (GS)) for 1a-c and (b) differences in steric interactions and resonance stabilization between the ground and BS transition state (TS). The latter factor was modeled by employing HF/3-21G(*) ab initio molecular orbital calculations of the GS and ring inversion TS. It is concluded that k_BS is unchanged in 1a-c because the potentially greater π interaction in the BS TS is counterbalanced by a greater degree of twist between the aryl and COT rings resulting from increased steric hindrance relative to the GS. However, π interaction assumes a greater importance in the TS of 2^2-/2K⁺ owing to a decreased HOMO-LUMO energy gap compared to 1a-c, particularly when the counterions are solvated. This causes a decrease in the inter-ring twist angle and, together, these changes are responsible for the observed increase in k_BS in 2^2-/2K⁺. The effect of substituents on a possible contribution of heavy atom tunneling to the reaction mechanism is also discussed.

Full text of DOI:10.1021/jo010526h

5572
J . Org. Chem. 2001, 66, 5572-5579
Effect of P a r a Su bstitu en ts on th e Ra te of Bon d Sh ift in
Ar ylcycloocta tetr a en es
Stuart W. Staley* and J oanne D. Kehlbeck
Department of Chemistry, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213
staley@andrew.cmu.edu
Received May 23, 2001
The rate constants for bond shift (kBS) in phenylcyclooctatetraene (1b) and its p-nitro and p-methoxy
analogues (1a and 1c, respectively) in THF-d8 were determined by dynamic NMR spectrometry to
be identical, but kBS is eight times greater at 280 K relative to 1b when the para substituent is
cyclooctatetraenyldipotassium (22-/2K+). These results are discussed in the context of (a) possible
intrinsically small substituent effects (as determined by 13C chemical shifts in the ground state
(GS)) for 1a -c and (b) differences in steric interactions and resonance stabilization between the
ground and BS transition state (TS). The latter factor was modeled by employing HF/3-21G(*) ab
initio molecular orbital calculations of the GS and ring inversion TS. It is concluded that kBS is
unchanged in 1a -c because the potentially greater π interaction in the BS TS is counterbalanced
by a greater degree of twist between the aryl and COT rings resulting from increased steric
hindrance relative to the GS. However, π interaction assumes a greater importance in the TS of
2
2-/2K+ owing to a decreased HOMO-LUMO energy gap compared to 1a -c, particularly when
the counterions are solvated. This causes a decrease in the inter-ring twist angle and, together,
these changes are responsible for the observed increase in kBS in 22-/2K+. The effect of substituents
on a possible contribution of heavy atom tunneling to the reaction mechanism is also discussed.
Sch em e 1. Rin g In ver sion (top ) a n d Bon d Sh ift
(bottom ) in COT
Substituent effects on allowed electrocyclic reactions
have received considerable attention in recent years.1 In
contrast, few experimental or theoretical investigations
of such effects have been reported on formally symmetry-
forbidden pericyclic processes,2,3 despite the fact that the
latter are intrinsically interesting and of practical use
synthetically.4
Carpenter employed a Hu¨ckel molecular orbital (HMO)
model to predict that substituent effects on forbidden
pericyclic processes will be larger than those on closely
related allowed reactions.5 This model also predicts that
both π-electron donors and acceptors accelerate the rate
of a forbidden pericyclic process through π interactions.
However, there is little evidence regarding the validity
of these hypotheses.
decreases in activation energies that were quantitatively
in accord with the measured values. The greater sub-
stituent effect expected on the basis of the HMO model
for a forbidden reaction compared to a corresponding
allowed reaction was confirmed for electron acceptors, but
just the opposite was found for electron donors.2
The bond shift (BS) reaction in cyclooctatetraenes
(COTs) (Scheme 1) is a well-known example of a formally
forbidden electrocyclic reaction. Substituted cyclooctatet-
raenes represent attractive compounds for the investiga-
tion of Carpenter’s postulate.5 The formally antiaromatic
nature of the BS transition state (TS) implies that (at
the HMO level) COTs will possess a high-lying highest
occupied MO (HOMO) and a low-lying lowest unoccupied
MO (LUMO). As a result, any perturbation in π-electron
density by a substituent would be expected to accelerate
the rate of the reaction because the antiaromatic desta-
bilization of the TS can be mitigated either by increasing
or decreasing the π-electron density.5b
One relevant study was reported by Spellmeyer et al.,2
who found that the rate of disrotatory electrocyclization
of an o-xylylene increases on substitution by both donors
and acceptors, in accord with the prediction of Carpen-
ter’s model. These authors also modeled this reaction by
employing substituted butadienes at the RHF/3-21G//
UHF/3-21G level and computed substituent-induced
* To whom correspondence should be addressed.
(1) (a) Hrovat, D. A.; Beno, B. R.; Lange, H.; Yoo, H.-Y.; Houk, K.
N.; Borden, W. T. J . Am. Chem. Soc. 1999, 121, 10529. (b) J ackman,
L. M.; Fernandes, E.; Heubes, M.; Quast, H. Eur. J . Org. Chem. 1998,
2209. (c) Wiest, O.; Montiel, D. C.; Houk, K. N. J . Phys. Chem. A 1997,
101, 8378.
(2) Spellmeyer, D. C.; Houk, K. N.; Rondan, N. G.; Miller, R. D.;
Franz, L.; Fickes, G. N. J . Am. Chem. Soc. 1989, 111, 5356.
(3) Kless, A.; Nendel, M.; Wilsey, S.; Houk, K. N. J . Am. Chem. Soc.
1999, 121, 4524, 7278.
(4) (a) Cohen, T.; Bhupathy, M.; Matz, J . R. J . Am. Chem. Soc. 1983,
105, 520. (b) Chantrapromma, K.; Ollis, W. D.; Sutherland, I. O. J .
Chem. Soc., Perkin Trans. 1 1983, 1049. (c) Bauld, N. L.; Harirchian,
B.; Reynolds, D. W.; White, J . C. J . Am. Chem. Soc. 1988, 110, 8111.
(5) (a) Carpenter, B. K. Tetrahedron 1978, 34, 1877. (b) Zoeckler,
M. T.; Carpenter, B. K. J . Am. Chem. Soc. 1981, 103, 7661.
The TSs of some formally forbidden pericyclic reactions
that have been studied theoretically have been shown to
10.1021/jo010526h CCC: $20.00 © 2001 American Chemical Society
Published on Web 07/18/2001

Rate of Bond Shift in Arylcyclooctatetraenes
J . Org. Chem., Vol. 66, No. 16, 2001 5573
be partially or almost fully biradicaloid in character with
little overlap between the reacting orbitals.3 An impor-
tant advantage of COT for the study of pericyclic reac-
tions is that the strong overlap between the reacting
orbitals is maintained throughout the BS reaction, the
D8h TS of which was calculated to be a singlet at the
MCSCF/6-31G*//HF/6-31G* level of theory.6
common donors and acceptors and has led to the discov-
ery of some unexpected results.
Trindle and Wolfskill performed AM1 semiempirical
MO calculations of formyl-, methoxy-, and methylene ion-
substituted COTs.7 These workers predicted that a meth-
oxy substituent decreases the barrier to ring inversion
(∆GqRI) but that a formyl group increases it (Scheme 1).
Their calculations also suggested that strong electron-
donor substituents are more effective than strong electron-
withdrawing substituents at flattening the tub confor-
Resu lts
Syn th esis. The syntheses of phenylCOT (1b) and its
p-methoxy analogue (1c) were straightforward. Com-
pound 1b, first prepared by Cope11 in 25% yield, was
synthesized by DeKock’s modification12 of Cope’s proce-
dure. The first step involves nucleophilic attack of an
organolithium reagent on COT to form a cyclooctatatri-
enyl anion. Subsequent abstraction of a proton with a
second equivalent of organolithium affords the substi-
tuted COT dianion, which can be cleanly oxidized to the
arylCOT with oxygen.
mation of COT, thereby reducing ∆Gq
.
RI
Kinetic studies on substituted COTs have shown
substantial steric effects of substituents on the RI and
BS energy barrier relative to COT.8 As the COT ring
flattens toward the transition state, steric crowding
between a substituent and the hydrogens at the vicinal
positions increases, especially across the single bonds.
An increase in steric interactions has been shown to
increase ∆Gq by several kcal mol-1 in homologous
BS
compounds.9 Conversely, destabilization of the ground
state has been shown to reduce ∆Gq in sterically
BS
crowded COTs.10
The substituents calculated by Trindle and Wolfskill
(OCH3 and CHO) span a large range of π-electronic
perturbations. However, they also differ in steric effects
and, more importantly, electronegativity. It has been our
experience that such differences will significantly affect
the rate of BS.8b,c
Paquette and co-workers first reported the synthesis
of 1c by the Cope method in <2% yield.13 A more recent
synthesis by Harmon and Streitwieser, which employed
the coupling of bromoCOT with the cuprate of anisole,14
gave a much higher yield (80%) but is less convenient.
The Cope/DeKock procedure has the advantage that the
synthesis of bromoCOT is not required, and a separation
of cyclooctatriene side products is avoided. Consequently,
the latter method was employed to afford 1c in 64% yield.
We were concerned about the stability of phenyllithium
substituted in the para position by a strong electron-
withdrawing group when subjected to the higher tem-
peratures required for the Cope/DeKock reaction.15 Con-
versely, an acceptor in the para position of an aryl halide
is known to facilitate the Stille reaction.16 Thus, Stille
coupling of 1-iodo-4-nitrobenzene with tributylstannyl-
cycloctatetraene cleanly afforded p-nitrophenylCOT (1a )
in 70% yield.17
To explore the effect of substituents on forbidden
electrocyclic processes, we decided to investigate the BS
in COTs by employing para donor- and acceptor-substi-
tuted phenylCOTs (1a -c). These compounds have identi-
cal steric and hybridization effects at the position of
substitution on the COT ring yet represent a potentially
wide range of π-electronic effects.
Although steric and hybridization effects often domi-
nate ∆∆GqBS, it is reasonable to expect that π delocal-
ization could also play a measurable role. We recently
measured a 6-fold increase in kBS on going from 1,4-
dicyclooctatetraenylbenzene (2) to its dipotassium salt
(22-/2K+) at 280 K.10a This acceleration is undoubtedly
due to a π-electronic effect. The present study was
designed to relate this observation to the effects of other
(11) Cope, A. C.; Kinter, M. R. J . Am. Chem. Soc. 1951, 73, 3424.
(12) Miller, J . T.; DeKock, C. W.; Brault, M. A. J . Org. Chem. 1979,
44, 3508.
(13) Paquette, L. A.; Malpass, J . R.; Barton, T. J . J . Am. Chem. Soc.
1969, 91, 4714.
(14) Harmon, C. A.; Streitweiser, A., J r. J . Org. Chem. 1973, 38,
549.
(6) Hrovat, D. A.; Borden, W. T. J . Am. Chem. Soc. 1992, 114, 5879.
(7) Trindle, C.; Wolfskill, T. J . Org. Chem. 1991, 56, 5426.
(8) (a) Paquette, L. A. Acc. Chem. Res. 1993, 26, 57. (b) Staley, S.
W.; Grimm, R. A.; Martin, G. S.; Sablosky, R. A. Tetrahedron 1997,
53, 10093. (c) Staley, S. W.; Grimm, R. A.; Sablosky, R. A. J . Am. Chem.
Soc. 1998, 120, 3671.
(9) (a) Anderson, J . E.; Kirsch, P. A. J . Chem. Soc., Perkin Trans. 2
1992, 1951. (b) Grimm, R. A., Ph.D. Thesis, Carnegie Mellon Univer-
sity, 1997.
(10) (a) Staley, S. W.; Kehlbeck, J . D.; Grimm, R. A.; Sablosky, R.
A.; Boman, P.; Eliasson, B. J . Am. Chem. Soc. 1998, 120, 9793. (b)
Staley, S. W.; Kehlbeck, J . D. Org. Lett. 1999, 1, 565.
(15) Parham, W. E.; Bradsher, C. K. Acc. Chem. Res. 1982, 15, 300.
(16) (a) Milstein, D.; Stille, J . K. J . Am. Chem. Soc. 1979, 101, 4992.
(b) Farina, V.; Krishnamurthy, V.; Scott, W. J . The Stille Reaction,
Wiley: New York, 1998; pp 12-16.
(17) For other Pd-catalyzed couplings involving the COT ring, see:
Siesel, D. A.; Staley, S. W. J . Org. Chem. 1993, 58, 7870.

5574 J . Org. Chem., Vol. 66, No. 16, 2001
Staley and Kehlbeck
Ta ble 1. Kin etic Da ta for Bon d Sh ift in Ar ylCOTs in
Ch em ica l Sh ift Assign m en ts. The following general
procedure was employed for the assignment of H and
1
THF -d 8
13C NMR signals for 1a -c, 2, and 22-/2K+. The eight
carbons in the neutral ring resonate in the region of δ
125-150, while the five signals for the C2v dianion ring
appear at δ 83-96 due to the higher electron density at
these carbons.18 The peaks for C2-C4 and C6-C8 were
identified on the basis of exchange broadening as a result
of BS, although, except for C2, assignments were not
made for individual carbons. The most upfield of these
signals was assigned to C2 by analogy with similar
compounds,19 while C5 was assigned based on the absence
of exchange broadening and an increased peak area
relative to C1. The proton signals of the aryl ring were
easily assigned and the details are given below. The
signals for C10 and C11 were assigned by heteronuclear
correlation spectroscopy (HETCOR) on the basis of
coupling to H10 and H11, respectively. The remaining
quaternary signals were identified by C-H long-range
correlation or cross-relaxation correlation experiments.
Additional assignments for each compound are given
below.
a,b
b,c
compound
C6H5-COT (1b)
T range (K) kBS
∆Gq
BS
262-293
260-320
262-289
271-298
5.3 15.4 ( 0.1
6.8 15.3 ( 0.1
6.5 15.3 ( 0.1
7.5 15.2 ( 0.1
43.3 14.3 ( 0.1
p-CH3O-C6H4-COT (1c)
p-NO2-C6H4-COT (1a )
p-COT-C6H4-COT (2)d
p-COT-C6H4-COT/2K+ (22-/2K+)d 248-284
a
b
In s-1
.
At 280 K. c In kcal mol-1; calculated from van’t Hoff
d
plot. Reference 10a.
A cross-correlation experiment confirmed the assignment
of H10 based on a significant NOE with H2. C9 and C12
were assigned on the basis of long-range correlations
3
(corresponding to J CH(trans) ≈ 8 Hz) with H11 and H10
,
respectively, while the same experiment with longer
3
delays (corresponding to J CH(cis) ≈ 5 Hz) led to the
correlation H10 with C1.
Kin etics. Rate constants for bond shift (kBS) were
obtained at a number of temperatures where line broad-
ening could be measured in the slow exchange region.
Kinetic data were obtained for duplicate samples of 1a -c
in THF-d8. 13C line broadening was measured for one
nonoverlapped signal (δ 133.4) in 1a , for six nonover-
lapped signals (δ 133.7, 133.2, 133.0, 132.4, 132.2, 126.7)
in 1b, and for four nonoverlapped signals (δ 132.9, 131.4,
130.8, 125.7) in 1c. Values of kBS were determined over
27, 31, and 60 degree temperature ranges for 1a -c,
respectively. The kinetics for BS in 1c were determined
by employing three NMR spectrometers operating at
different magnetic field strengths in order to extend the
temperature range for line broadening measurements.
Although the COT 13C signals had coalesced and were
emerging from the baseline by 50 °C, five distinct signals
were not observed. Rate constants and free energies of
activation at 280 K are listed in Table 1. Complete tables
of kinetic data and the corresponding van’t Hoff plots are
given in the Supporting Information.
1
The downfield aryl doublet in the H spectrum of 1a
was assigned to H11 due to the electron-withdrawing
character of the nitro group, whereas the upfield doublet
was assigned to H10. Long-range correlation spectroscopy
3
(COLOC) with delays corresponding to a J CH(trans) value
of approximately 8 Hz led to the correlation of H11 and
H10 with C9 and C12, respectively. The latter assignment
is supported by the chemical shift of C1 of nitrobenzene
(δ 148.5).20
The phenyl proton signals of 1b (H10-H12) were as-
signed on the basis of their integration and apparent
multiplicities. They were then correlated with coupled
13C signals by HETCOR and COLOC in order to assign
C10-C12 and C9, respectively.
1
The upfield aryl doublet in the H spectrum of 1c was
assigned to the proton ortho to the methoxy group (H11
due to the π-donor character of the methoxy substituent.
Likewise, the downfield aryl doublet was assigned to H10
)
Discu ssion
.
A COLOC experiment with delays corresponding to a
3J CH(trans) value of approximately 8 Hz led to the correla-
tion of H11 with C9 and H10 with C12. The literature value
for δ C1 of anisole (δ 159.9)21 supports the assignment of
C12.
A perturbation of the π-electron density within the
COT ring might be expected to affect the energy required
for BS due to the antiaromatic nature of the transition
state. Surprisingly, none of the neutral substituted COTs
(1a -c, 2) show an experimentally significant variation
in kBS (Table 1). In contrast, the dianion ring in 22-/2K+
causes an 8-fold enhancement of the rate of BS in the
neutral COT ring at 280 K relative to 1b.
The singlet at δ 7.31 in the spectrum for 2 was assigned
to H10 due to its multiplicity and integrated area. A cross-
correlation experiment was employed to assign H2 based
on a significant NOE with H10. The 3.9 Hz coupling
constant measured for H2 at -40 °C also supports this
assignment. Long-range correlation experiments with
We have considered three possible explanations for
these interesting but unexpected results. (1) Resonance
effects in 1a -c and 2 might be intrinsically too small to
be detected in either the ground state (GS) or the BS TS
while 22-/2K+, being a charged species, is a markedly
superior donor. (2) Resonance effects (in the Hartree-
Fock (HF) model) in the GS of 1a -c and 2 might be
substantial but of similar magnitude to those in the TS
3
delays corresponding to J CH(trans) ≈ 8 Hz correlated H10′
with C9 (which is the same as H10 with C9′), while the
same experiment with longer delays (corresponding to
3J CH(cis) ≈ 5 Hz) led to the correlation of H2 and H10 with
C9 and C1, respectively.
1
The doublets in the H spectrum for 22-/2K+ at δ 7.34
and therefore their effects on ∆Gq cancel. (3) The
BS
and 6.85 were assigned to H10 and H11 by analogy to 2.
absence of any effect in 1a -c and 2 might be related to
heavy-atom tunneling or configuration interaction (CI).
Each of these possibilities is discussed and a reasonable
explanation is offered.
In t r in sica lly Sm a ll Su b st it u en t E ffect s. NMR
chemical shifts have long been one of the principal probes
of substituent electronic effects. Among the nuclei most
widely employed are the para 13C in monosubstituted
(18) Boman, P.; Eliasson, B. Acta Chem. Scand. 1996, 50, 816.
(19) Boman, P.; Eliasson, B.; Grimm, R. A.; Martin, G. S.; Strnad,
J . T.; Staley, S. W. J . Am. Chem. Soc. 1999, 121, 1558.
(20) Silverstein, R. M.; Bassler, G. C.; Morrill, T. C. Spectrometric
Identification of Organic Compounds, 4th ed.; Wiley: New York, 1981;
pp 265-6.
(21) Ewing, D. F. Org. Magn. Reson. 1979, 12, 499 and references
cited.

Rate of Bond Shift in Arylcyclooctatetraenes
J . Org. Chem., Vol. 66, No. 16, 2001 5575
Ta ble 2. 13C Ch em ica l Sh ifts for Ar ylCOTs in THF -d 8
a n d Der ived σ Con sta n ts
a
+ a
compound substituent δ C2
δ C9
σp
σp
σ13 a
1a
1b
2
NO2
H
COT
OCH3
132.8 147.9
126.7 141.0
128.7 140.0
0.78
0
0.74
0
1.01
0
0.15b
0.1b
0.1b
1c
125.7 132.5 -0.27 -0.65 -0.74
2
2-/2K+
COT2-/2K+ 122.2 129.4 -0.7b -1.0b -1.1b
a
b
Reference 27d unless indicated otherwise. This work; (0.1.
F igu r e 1. HF/3-21G-optimized GS (left) and RI TS (right) of
1b.
benzenes,22 19F in para-disubstituted benzenes,23 and 13C
at the â position in substituted styrenes.24 To address
the possibility that some substituent effects are too small
to exert an influence on bond shift, the 13C NMR spectra
for 1a -c, 2, and 22-/2K+ were fully assigned in order to
compare the relative magnetic and electronic environ-
ments of the â and para carbons (C2 and C9, respectively)
in these compounds.
Due to the distance of C9 and C2 from the position of
substitution, the chemical shifts of these carbons (espe-
cially C2) should primarily reflect resonance (π delocal-
ization) effects of the substituents on these carbons, as
opposed to inductive and/or electric field effects. This
qualitative interpretation of 13C chemical shifts in aro-
matic systems follows interpretations by previous authors
who have concluded that 13C chemical shifts are useful
probes of π-electron density.24
The change in chemical shift for C9 (∆δC9) on going
from 1a to 1c is 15.4 ppm (Table 2). This corresponds to
a difference in charge density that does not manifest itself
in a measurable difference in kBS (Table 1). Conversely,
∆δC9 on going from 1c to 22-/2K+ is only 3.1 ppm, yet
the rate constant for bond shift increases by over 6-fold.
Similarly, ∆δC2 on going from 1a to 1c is 7.1 ppm while
δC2(1c) - δC2(22-/2K+) is only 3.5 ppm, yet the rate
constant for bond shift is appreciably increased relative
to 1b only in the dianion. This analysis appears to
preclude the explanation that the rate of bond shift in
1a -c and 2 is unchanged because electronic effects are
intrinsically too small while the COT2- ring in 22-/2K+
is a much superior donor. Note that these measurements
reflect electronic effects in the GS, and it is not known
whether a similar trend applies to the TS.
included in the present study, one cannot expect to get
more than an approximate correlation with even a single-
parameter, let alone a DSP, treatment. Average values
of σp, σp+, and σ13 for p-COT and p-COT2-/2K+ obtained
by interpolation or extrapolation of the correlation lines
for plots of δC2 and δC9 against the corresponding σ
values for the p-phenyl substituents in 1a -c are given
in Table 2. The values derived using δC2 do not vary by
more than 0.15 from those using δC9 and the correlations
+
involving σp and σp are comparable. The NMR-derived
parameter σ13 has only been defined for the â carbon of
styrenes.27d
The derived σ values indicate that COT2-/2K+ in THF-
d8 is only a moderately stronger donor than methoxy in
the GS. However, substituent constants for COT2- are
undoubtedly dependent on the cation and the solvent, i.e.,
on the degree and type of ion pairing.
Com p en sa tin g Effects in th e Gr ou n d a n d Tr a n si-
tion Sta tes. To assess both steric and electronic factors,
HF/3-21G(*)29 geometry optimizations and harmonic fre-
quency analyses30 were performed on the GS and RI TS
of 1a -c, 2, and 22-/2K+ (Figure 1 and Table 3). The RI
TS (Scheme) is a reasonable model for steric interactions
in the BS TS, which was not calculated owing to the large
multiconfigurational wave function necessary to repre-
sent it accurately.6,31,32
Two models were investigated for 22-/2K+, the first of
which involved a complete geometry optimization of the
(27) (a) Relles, H. M.; Schluenz, R. W. J . Org. Chem. 1972, 37, 1742.
(b) Posner, T. B.; Hall, C. D. J . Chem. Soc., Perkin Trans. 2 1976, 729.
(c) Corne´lis, A.; Lambert, S.; Laszlo, P.; Schaus, P. J . Org. Chem. 1981,
46, 2130. (d) Slater, C. D.; Robinson, C. N.; Bies, R.; Bryan, D. W.;
Chang, K.; Hill, A. W.; Moore, W. H., J r.; Otey, T. G.; Poppelreiter, M.
L.; Reisser, J . R.; Stablein, G. E.; Waddy, V. P., III.; Wilkinson, W. O.;
Wray, W. A. J . Org. Chem. 1985, 50, 4125. (e) Aun, C. E.; Clarkson, T.
J .; Happer, D. A. R. J . Chem. Soc., Perkin Trans. 2 1990, 645.
(28) Datta, S.; De, A.; Bhattacharyya, S. P.; Medhi, C.; Chakravarty,
A. K.; Brunskill, J . S. A.; Fadoujou, S.; Fish, K. J . Chem. Soc., Perkin
Trans. 2 1988, 1599.
(29) (a) Binkley, J . S.; Pople, J . A.; Hehre, W. J . J . Am. Chem. Soc.
1980, 102, 939. (b) Gordon, M. S.; Binkley, J . S.; Pople, J . A.; Pietro,
W. J .; Hehre, W. J . J . Am. Chem. Soc. 1982, 104, 2797. (c) Pietro, W.
J .; Francl, M. M.; Hehre, W. J .; DeFrees, D. J .; Pople, J . A.; Binkley,
J . S. J . Am. Chem. Soc. 1982, 104, 5039. (d) Dobbs, K. D.; Hehre, W.
J . J . Comput. Chem. 1986, 7, 359.
(30) Frisch, M. J .; Trucks, G. W.; Head-Gordon, M.; Gill, P. M. W.;
Wong, M. W.; Foresman, J . B.; J ohnson, B. G.; Schlegel, H. B.; Robb,
M. A.; Replogle, E. S.; Gomperts, R.; Andreas, J . L.; Raghavachari,
K.; Binkley, J . S.; Gonzalez, C.; Martin, R. L.; Fox, D. J .; DeFrees, D.
J .; Baker, J .; Stewart, J . J . P.; Pople, J . A. GAUSSIAN 92, Revision A;
Gaussian, Inc., Pittsburgh, PA, 1992.
(31) Since Karadakov et al. found that the CASSCF/6-31G*//HF/6-
31G correlation energy for the D4h RI TS of COT is only 3% greater
than that for the D2d GS, we conclude that relative values of ∆ERI can
be reliably calculated at the HF level of theory; Karadakov, P. B.;
Gerratt, J .; Cooper, D. L.; Raimondi, M. J . Phys. Chem. 1995, 99,
10186.
A more quantitative measure of the substituent effect
of p-COT and p-COT2-/2K+can be obtained by plotting
the chemical shifts in Table 2 against appropriate Ham-
mett-type substituent constants. Previous workers have
reported excellent correlations of σ constants, particularly
+ 25
σp
,
with 13C chemical shifts for the para carbon of
substituted benzenes21,24b,26 and the â carbon of substi-
tuted styrenes.27
Dual substituent parameter (DSP) treatments have
also been employed for these nuclei with good success.23,28
However, in view of the minimal number of substituents
(22) Taft, R. W.; Price, E.; Fox, I. R.; Lewis, I. C.; Andersen, K. K.;
Davis, G. T. J . Am. Chem. Soc. 1963, 85, 3146 and references cited.
(23) Craik, D. J .; Brownlee, R. T. C. Prog. Phys. Org. Chem. 1983,
14, 1 and references cited.
(24) (a) Lauterbur, P. C. Tetrahedron Lett. 1961, 274. (b) Spiesecke,
H.; Schneider, W. G. J . Chem. Phys. 1961, 35, 731. (c) Hehre, W. J .;
Taft, R. W.; Topsom, R. D. Prog. Phys. Org. Chem. 1976, 12, 159. (d)
Ewing, D. F. In Correlation Analysis in Chemistry. Recent Advances;
Chapman, N. B., Shorter, J ., Eds.; Plenum: New York, 1978; Chapter
8.
(25) Swain, C. G.; Lupton, E. C., J r. J . Am. Chem. Soc. 1968, 90,
4328.
(26) (a) Nelson, G. L.; Levy, G. C.; Cargioli, J . D. J . Am. Chem. Soc.
1972, 94, 3089. (b) Schulman, E. M.; Christensen, K. A.; Grant, D. M.;
Walling, C. J . Org. Chem. 1974, 39, 2686.
(32) (a) Wenthold, P. G.; Hrovat, D. A.; Borden, W. T.; Lineberger,
W. C. Science 1996, 272, 1456. (b) Koseki, S.; Toyota, A. J . Phys. Chem.
A 1997, 101, 5712. (c) Glukhovtsev, M. N.; Bach, R. D.; Laiter, S. J .
Mol. Struct. (THEOCHEM) 1997, 417, 123. (d) Andre´s, J . L.; Castan˜o,
O.; Morreale, A.; Palmeiro, R.; Gomperts, R. J . Chem. Phys. 1998, 108,
203.

5576 J . Org. Chem., Vol. 66, No. 16, 2001
Staley and Kehlbeck
Ta ble 3. Rela tive En er gies a n d Tor sion a l An gles for th e
HF /3-21G(*)-Op tim ized Gr ou n d Sta te a n d Rin g In ver sion
Tr a n sition Sta te of Ar ylCOTs
a
b
c,d
e
compound substituent state ∆(E+ZPE)RI ω1 ∆ω1
ω2 cos2 ω1
1a
1b
1c
2f
NO2
GS
TS
GS
TS
GS
TS
GS
TS
43.3 1.1
0.53
0.21
0.51
0.21
0.55
0.23
19.17
18.93
19.14
18.99
18.91
18.61
62.9 0.1
44.4 (0)
63.0 (0)
42.4 2.0
61.4 1.6
43.1 1.3 43.1 0.53
61.7 1.3 43.2 0.22
42.8 1.6 45.5 0.54
58.8 4.2 45.4 0.27
40.1 4.3 44.7 0.59
57.2 5.8 45.7 0.29
H
OCH3
COT
2
2
2-/2K+ f,g COT2-/2K+ GS
TS
2-/2K+ f,h COT2-/2K+ GS
TS
F igu r e 2. Diagram showing interaction (in hartrees) between
the HF/3-21G(*) frontier molecular orbitals of the COT and aryl
rings in the GS and RI TS of 1b, 1c, and 22-/2K+ (solvated
a
b
In kcal mol-1 relative to the ground state. ω(C2C1C9C10); in
K+). These calculations employed the aryl and COT or COT2-
2K+ substructures from the corresponding GS or RI TS.
/
degrees. c ∆ω1,GS ) ωGS(1b) - ωGS
.
∆ω1,TS ) ωTS(1b) - ωTS.
d
e ω(C2′C1′C9′C10′), the twist of the second COT ring in 2 or the
COT2- ring in 22-/2K+; in degrees. f The COT2- ring and the
neutral COT ring (or its C1C2 double bond) are nearly coplanar in
greater degree of sterically induced inter-ring twist
relative to the GS.
g
h
the minimum energy conformation. Fully optimized. Solvated
K+; the potassium ions were fixed 2.400 Å above and below the
center of the ring (approximately as found by X-ray diffraction
[refs 33, 34]), and all other parameters were optimized.
In contrast, ∆ω1 for 22-/2K+ (solvated K+) is 1.5° larger
in the TS than in the GS. In accord with this result,
∆(E+ZPE)RI is calculated to be 0.32 kcal mol-1 smaller
than for 1b. Why does the COT2- substituent increase
kBS while the methoxy group, which is only a moderately
weaker donor in the GS, has no effect? An insight can be
gained by employing second-order perturbation theory.
In eq 1, the interaction energy (∆E) between the COT
and COT2- (π) HOMO and the aryl (π*) LUMO in weakly
coupled aryl and COT or COT2- rings is inversely
proportional to the LUMO-HOMO energy gap (ꢀL-ꢀH)
and directly proportional to the square of the resonance
ion triplet. In the second, the potassium ions were fixed
above the center of the COT2- ring at a distance (2.400
Å) similar to that found by X-ray diffraction (2.32-2.46
Å) for contact ion pairs of COT2- 33 or substituted COT2- 34
and K+ solvated with glyme or diglyme and all other
parameters were optimized. The counterions are located
2.312 and 2.400 Å from the plane of the ring in these
two structures, which we designate as 22-/2K+ and 22-
/
2K+ (solvated K+), respectively.
2
integral ( π|H ′|π* ) as well as to cos2 ω. Although it is
The value of ω(C2C1C9C10) (ω1) in 1a -c and 2 is about
19° greater in the RI TS than in the GS (Table 3). This
is due primarily to increased steric interactions between
the ortho aryl hydrogens and the COT hydrogens at C2
and C8 as CCC in the COT ring increases from ca. 126°35
in the ground state to ca. 135° in the RI TS.
likely that these two rings are more strongly coupled than
is appropriate for the quantitative application of a second-
order perturbation model, the relative values of ∆E that
are predicted by eq 1 should nevertheless be valid.
2
π|H ′|π*
As a working hypothesis, we assume that the twist
angle between the aryl and COT or COT2- ring (ω1 and
ω2, respectively [Table 3]) primarily reflects π resonance
interactions between the two rings if steric effects remain
constant. Thus, the fact that the COT2- ring is more
twisted with respect to the phenylene ring in the RI TS
of 22-/2K+ than in the GS (ω2 in Table 3) is taken to
indicate that there is more π resonance with the neutral
COT ring in the GS. Further, the reduction in twist for
the neutral compounds (1a , 1c, and 2) relative to 1b (∆ω1)
is the same or slightly less in the TS than in the GS,
suggesting that resonance due to the substituent either
increases less in the TS (consistent with the small
increase in ∆(E+ZPE)RI for 1a and 1c relative to 1b) or
remains relatively constant (as in 2) in these two states.
∆E )
(1)
ꢀ1 - ꢀ2
If the resonance integral remains unchanged, ∆E is
calculated to increase by 136% on going from the GS to
the TS in 22-/2K+ (solvated K+) (Figure 2) due to the
decrease in ꢀL - ꢀH. In contrast, the stabilization of 1b
and 1c increases by only 119 and 121%, respectively. This
is due to the fact that, since the decrease in ꢀL is
essentially the same for all of the substituents, the
change in ꢀL- ꢀH is proportionately much greater for the
donor with the highest-lying HOMO (C6H4COT2-).
This type of perturbation argument has, for example,
been employed previously to explain the effects of sub-
stituents on the rates of 1,3-dipolar cycloaddition36 and
Diels-Alder37 reactions. Sustmann and Schubert con-
cluded that “introducing a substituent into a diene has
a greater effect on the rates of [Diels-Alder] reactions
of TCNE [tetracyanoethene] than on those of MA [maleic
anhydride]” due to a smaller HOMO-LUMO separation
in the TS of TCNE.37 The Diels-Alder TSs for MA and
TCNE in their comparison are analogous to the GS and
BS TS, respectively, in our study.
Thus, the absence of a substituent effect on ∆Gq in these
BS
compounds can be attributed to a fortuitous cancellation
of two effects: (a) an increase in π delocalization in the
RI TS relative to the GS assuming no change in ω1 and
(b) a decrease in π delocalization in the TS owing to a
(33) (a) Kinsley, S. A.; Streitwieser, A., J r.; Zalkin, A. Organometal.
1985, 4, 52. (b) Hu, N.; Gong, L.; J in, Z.; Chen, W. J . Organomet. Chem.
1988, 352, 61.
(34) (a) Goldberg, S. Z.; Raymond, K. N.; Harmon, C. A.; Templeton,
D. H. J . Am. Chem. Soc. 1974, 96, 1348. (b) Kinsley, S. A.; Streitwieser,
A., J r.; Zalkin, A. Acta Crystallogr., Sect. C: Cryst. Struct. Commun.
1986, 42, 1092.
(36) Sustmann, R.; Trill, H. Angew. Chem., Int. Ed. Engl. 1972, 11,
838.
(35) Claus; K. H.; Kru¨ger, C. Acta Crystallogr., Sect C: Cryst. Struct.
Commun. 1988, 44, 1632.
(37) Sustmann, R.; Schubert, R. Angew. Chem., Int. Ed. Engl. 1972,
11, 840.

Rate of Bond Shift in Arylcyclooctatetraenes
J . Org. Chem., Vol. 66, No. 16, 2001 5577
This analysis can be extended to the numerator of eq
1. Thus, if ꢀL - ꢀH remains unchanged, ∆E is calculated
to decrease by only 50% due to the decrease in cos2 ω on
going from the GS to the TS in 22-/2K+ (solvated K+)
(Table 3), whereas decreases of 60 and 58% are calculated
for 1b and 1c, respectively. Although this model employs
only the HOMO and LUMO, it illustrates how a strong
π-orbital interaction in 22-/2K+ (solvated K+) can cause
a decrease in inter-ring twist that reinforces this interac-
tion and increases kRI or kBS while a weaker donor (C6H4-
OCH3 in 1c) has no observable effect.
The TS for BS should be stabilized even more by
resonance than that for RI owing to its higher HOMO
and lower LUMO. Furthermore, ion pairing for COT2-
/
2K+ might be even looser in THF solution than in the
crystal structures that we employed as a model.33,34 Both
factors would decrease the calculated barriers reported
in Table 3 and tend to bring their relative values more
into line with the relative values for ∆Gq in Table 1.
F igu r e 3. Arrhenius plot for bond shift in 1c; r ) 0.942. The
symbols represent measurements on 200 (O), 300 (0), and 500
MHz ([) NMR spectrometers.
BS
However, this conclusion is dependent on whether BS
occurs by passage over a classical barrier. A complete
analysis requires that tunneling through the barrier also
be considered.
An Arrhenius plot for BS in 1c gives a straight line
(Figure 3). However, this does not exclude a contribution
from tunneling for at least two reasons. First, the kinetics
were measured only down to 271 K; second, theoretical
treatments of quantum tunneling reactions (such as
proton transfer) by Borgis and Hynes43 and by Kim and
co-workers44 have shown that the appearance of classical
Arrhenius behavior can result primarily from the clas-
sical bath contribution to the rate constant.
The energetics of BS in COTs are dominated by the
ring-flattening step,6 which we assume to be the origin
of the observed Arrhenius behavior of kBS. This is in
accord with the calculations of Andre´s et al., who
employed CASSCF imaginary frequencies and MP2-
CASSCF barrier heights (including vibrational zero point
energy corrections) to compute tunneling corrections for
COT of 1.04 and 15.2 for RI and BS, respectively, at 200
K.31d These authors concluded that “the experimental rate
constants [for BS] mainly provide information about the
RI process.”
The question of the electronic effect of substituents on
the rate of tunneling has been addressed recently by
Mackenzie and co-workers,45 who studied the intra-
molecular 4σ + 2π dyotropic reaction (2H transfer) of
N,N-diarylpyrazolines. These authors found that rear-
rangement occurs over twice as fast when aryl )
p-MeC6H4 than when aryl ) p-ClC6H4. They also found
strong evidence for H-atom tunneling in the latter
compound but not for the former.
The contribution of tunneling is dependent on both the
height and width of the barrier,40,46 as well as on the
symmetry of the potential energy wells for the initial and
final states.47 Simply stated, the probability of tunneling
at a given temperature is greater through a high, thin
Hea vy Atom Tu n n elin g. Carpenter was the first to
propose that heavy-atom tunneling (HAT) contributes to
the mechanism of bond shift in COT.38 HAT can occur
only if the atomic positions of the interconverting species
are nearly the same, thereby causing the barrier separat-
ing them to be narrow. The distance carbon atoms in D4h
COT have to move in the bond shift process (∆rBS) is
about half the difference in length between a CC double
bond and a CC single bond (1.32 and 1.48 Å, respec-
tively),39 or ca. 0.08 Å. Since ∆rBS is less than the de
Broglie wavelength of 0.1 Å calculated for a CH particle
with a kinetic energy of 15.0 kcal mol-1 40
of HAT is reasonable.
, the postulation
HAT in processes such as BS in COT has been termed
vibrationally assisted tunneling (VAT) by Dewar et al.41
in order to describe tunneling that originates from a
vibrationally excited state. In COT, the unfolding of the
tub conformation to a planar or nearly planar RI TS can
be viewed as a large amplitude vibration that must take
place before tunneling can occur. The tunneling step
would then involve the movement of CH groups concomi-
tant with π-bond shift.
Because the tunneling rate constant is temperature
independent, a significant contribution from a tunneling
mechanism can cause a curved Arrhenius plot.40 Tun-
neling might also manifest itself in an unusually small,
commonly negative, entropy of activation (∆SqBS).40 To
determine a reasonably reliable value for ∆SqBS, we
measured the rate of bond shift for 1c over a 50 degree
temperature range and calculated ∆Sq to be -10 ( 5
BS
cal mol-1 K-1 from a van’t Hoff plot (Supporting Informa-
tion). This is in good agreement with the value of -9.7
cal mol-1 K-1 determined by Naor and Luz42 for COT in
liquid crystalline solvents measured over a temperature
range of 195 degrees, as well as with expectations based
on a tunneling contribution to the mechanism of BS.38,40
(43) Borgis, D.; Hynes, J . T. Chem. Phys. 1993, 170, 315.
(44) Dakhnovskii, Y.; Bursulaya, B.; Kim, H. J . J . Chem. Phys. 1995,
102, 7838.
(45) (a) Mackenzie, K.; Howard, J . A. K.; Mason, S.; Gravett, E. C.;
Astin, K. B.; Shi-Xiong, L.; Batsanov, A. S.; Vlaovic, D.; Maher, J . P.;
Murray, M.; Kendrew, D.; Wilson, C.; J ohnson, R. E.; Preiss, T.;
(38) Carpenter, B. K. J . Am. Chem. Soc. 1983, 105, 1700.
(39) Peterson, M. L.; Staley, S. W. Struct. Chem. 1998, 9, 305.
(40) Bell, R. P. The Tunnel Effect in Chemistry; Chapman and
Hall: London, 1980, Chapter 2.
(41) (a) Dewar, M. J . S.; Merz, K. M., J r.; Stewart, J . J . P. J . Chem.
Soc., Chem. Commun. 1985, 166. (b) Dewar, M. J . S.; Merz, K. M., J r.
J . Phys. Chem. 1985, 89, 4739.
Gregory, R. J . J . Chem. Soc., Perkin Trans.
2 1993, 1211. (b)
Mackenzie, K.; Howard, J . A. K.; Siedlecka, R.; Astin, K. B.; Gravett,
E. C.; Wilson, C.; Cole, J .; Gregory, R. J .; Tomlins, A. S. J . Chem. Soc.,
Perkin Trans. 2 1996, 1749. (c) Mackenzie, K.; Astin, K. B.; Gravett,
E. C.; Gregory, R. J .; Howard, J . A. K.; Wilson, C. J . Phys. Org. Chem.
1998, 11, 879.
(42) Naor, R.; Luz, Z. J . Chem. Phys. 1982, 76, 5662.

5578 J . Org. Chem., Vol. 66, No. 16, 2001
Staley and Kehlbeck
Ta ble 4. Bon d Len gth s for COT in th e HF /3-21G(*)-Op tim ized Rin g In ver sion Tr a n sition Sta tes for Ar ylCOTsa
b
compd
1a
r12
r23
r34
r45
r56
r67
r78
r18
r18 - r12
∆rBS
1.3271
1.3230
1.3276
1.3274
1.3283
1.3289
1.4760
1.4758
1.4764
1.4764
1.4762
1.4757
1.3226
1.3218
1.3229
1.3229
1.3233
1.3252
1.4745
1.4742
1.4740
1.4741
1.4735
1.4738
1.3217
1.3228
1.3218
1.3218
1.3219
1.3221
1.4759
1.4765
1.4758
1.4758
1.4758
1.4759
1.3228
1.3270
1.3232
1.3230
1.3233
1.3235
1.4881
1.4893
1.4901
1.4895
1.4912
1.4914
0.1610
0.1663
0.1625
0.1621
0.1631
0.1625
0.07754
0.07765
0.07760
0.07765
0.07747
0.07715
1b
1c
2
2
2
2-/2K+ c
2-/2K+ d
a
b
d
In angstroms. Average single bond length - average double bond length. c Fully optimized. Solvated K+; see footnote g in Table
3.
barrier separating isoenergetic potential energy wells
than through a short, broad barrier between minima of
different energies. We expect that the heights of the
classical BS barriers in arylCOTs will parallel those of
the RI barriers in Table 3. However, differences between
substituents might be enhanced owing to greater π
interaction between the substituent and the COT ring
in the classical BS TS as a result of its higher HOMO
and lower LUMO.
The width of the barrier is dependent on the distance
the CH groups move during interconversion between
planar bond-alternant structures. In arylCOTs, the posi-
tion of C1 is essentially fixed by the mass of the aryl
group. Consequently, ∆rBS for C2 and C8 will be about
twice as large as in COT with smaller differences for C3-
C7, significantly decreasing the contribution of tunneling
experimental error on the rate of BS relative to the
parent phenylCOT (1b). In contrast, kBS for 22-/2K+ is
eight times greater than that for 1b at 280 K. These
results are unexpected on the basis of classical ground-
state substituent effects, as exemplified by NMR-derived
values of σp and σp for p-NO2, p-OCH3, and p-COT2-
/
+
+
2K+. The values of σp and σp for p-COT2-/2K+ (deter-
mined to be -0.7 and -1.0, respectively, from 13C NMR
chemical shifts) indicates that it is a significantly stron-
ger donor than p-OCH3. However, these values are
undoubtedly highly dependent on the degree and type of
ion pairing involved.
Geometry optimization of the GS and RI TS at HF/
3-21G(*) suggests that the absence of a substituent effect
on kBS in 1a -c and 2 results from the fortuitious
cancellation of (a) an increase in π delocalization in the
RI TS relative to the GS assuming no change in the
COT-aryl dihedral angle and (b) a decrease in π delo-
calization in the TS relative to the GS due to increased
sterically induced twisting in the former. In contrast,
p-COT2-/2K+ is a strong enough donor that the effect of
increased π delocalization in the TS outweighs the rate-
retarding effect of increased sterically induced twisting.
Heavy-atom tunneling is expected to be less important
for BS in arylCOTs than in COT itself. We were unable
to assess the effect of configuration interaction (CI) on
kBS owing to the size of the multiconfigurational wave
functions required. Since CI is known to be important in
calculating the BS TS of COT,6,31,32 the above conclusions
need to be tested at the CI level when this is computa-
tionally feasible.
to kBS
.
Table 4 gives the CC bond lengths for the RI TS of
arylCOTs optimized at HF/3-21G(*). Note that there is
little variation in ∆rBS ((0.0001 Å) within the series 1a -
c, 2, and 22-/2K+ whereas ∆rBS for 22-/2K+ (solvated K+)
is 0.0003-0.0005 Å less than for the above compounds.
This is probably a result of “leakage” of π electron density
from the dianion ring to the neutral COT ring in the RI
TS of 22-/2K+ (solvated K+) and might contribute to the
larger value of kBS in the latter. On the other hand, r18
-
r12 in 22-/2K+ (solvated K+) is as large as or larger than
for the other arylCOTs. Since C2 and C8 would have to
tunnel the greatest distance, there is no clear evidence
that the BS barrier is narrower in 22-/2K+ (solvated K+)
at the HF/3-21G(*) level of theory.
We do not expect symmetry to play a major role in
determining the value of kBS in arylCOTs. This is because
BS changes the RI TS of 1b into an identical structure,
so the classical potential energy surface (PES) is sym-
metrical for all angles of twist between the phenyl and
planar COT rings. The C2 symmetry of the RI TS of 1b
is broken when p-phenyl substituents with various
degrees of twist are introduced, but the local symmetry
at C1 of the COT ring, and therefore the PES, should not
be greatly affected. Further, torsional vibrations of the
phenylene-substituent bond occur on a much shorter
time scale than HAT.48
Exp er im en ta l Section
Gen er a l. Diethyl ether and tetrahydrofuran were distilled
from sodium/benzophenone ketyl. Preparative column chro-
matography was performed on 40 µM silica gel. NMR spectra
(CDCl3 solutions) were obtained at 300 and 75 MHz for 1H
and 13C, respectively, unless otherwise noted. Elemental
analyses were performed by Midwest Microlab, Inc., India-
napolis, IN.
Syn th esis. P h en ylcycloocta tetr a en e (1b) was prepared
by the method of DeKock and co-workers.12 Final purification
was achieved by HPLC on a C18 column on elution with 95/5
1
acetonitrile/water. H NMR (THF-d8, 25 °C): δ 7.0-7.2 (5 H,
m), 5.8-6.1 (7 H, m). 13C NMR (THF-d8, 25 °C): δ 142.7 (C1),
141.1 (C9), 133.7, 133.2, 133.0, 132.9 (C5), 132.4, 132.2, 128.9
(C11), 128.2 (C12), 126.8 (C10), 126.7 (C2). The italicized values
are those for C3, C4, and C6-C8, which undergo exchange due
to bond shift.
Su m m a r y
The p-NO2, p-OCH3, and p-COT substituents in 1a , 1c,
and 2, respectively, in THF-d8 have no effect within
p-(Meth oxyp h en yl)cycloocta tetr a en e (1c). One-centi-
meter pieces of clean lithium wire (1.39 g, 0.20 mol) were added
to a dry three-necked round-bottomed flask equipped with an
addition funnel and a reflux condenser and charged with 100
mL of ether under argon. 4-Bromoanisole (0.598 g, 32.0 mmol)
in 60 mL of diethyl ether was added dropwise over 1 h at a
rate sufficient to maintain reflux. The mixture was heated to
reflux for an additional 1 h and transferred via a cannula to
(46) (a) O’Ferrall, R. A. M.; Kouba J . J . Chem. Soc. B 1967, 985. (b)
Parr, C. A.; Truhlar, D. G. J . Phys. Chem. 1971, 75, 1844. (c) Liu, Y.-
P.; Lynch, G. C.; Truong, T. N.; Lu, D.; Truhlar, D. G.; Garrett, B. C.
J . Am. Chem. Soc. 1993, 115, 2408. (d) Houk, K. N.; Li, Y.; McAllister,
M. A.; O’Doherty, G.; Paquette, L. A.; Siebrand, W.; Smedarchina, Z.
K. J . Am. Chem. Soc. 1994, 116, 10895.
(47) de la Vega, J . R. Acc. Chem. Res. 1982, 15, 185.
(48) (a) Borgis, D.; Hynes, J . T. J . Chem. Phys. 1991, 94, 3619. (b)
Borgis, D.; Hynes, J . T. J . Phys. Chem. 1996, 100, 1118.

Rate of Bond Shift in Arylcyclooctatetraenes
J . Org. Chem., Vol. 66, No. 16, 2001 5579
a vigorously stirred solution of COT (0.111 g, 10.6 mmol) in
50 mL of ether under argon. The resulting dark-orange
solution was stirred for 24 h in order to complete the formation
of the dianion. Oxygen was then passed over the reaction
mixture for 2 h. Saturated aq NH4Cl was added, and the
organic layer was separated and washed with 2 × 50 mL of
water. After extraction of the combined aqueous layers with
2 × 50 mL of ether, the combined organic layers were dried
(MgSO4) and concentrated. The residue was chromatographed
on silica gel with 10% CH2Cl2/hexane to afford 350 mg (64%)
of 1c, mp 64.5-65 °C (lit.13 64-65 °C). Final purification was
achieved by HPLC on a C18 column by eluting with acetonitrile.
1H NMR (THF-d8, 25 °C): δ 7.05 (2 H, d, J ) 9.0 Hz), 6.59 (2
H, d, J ) 9.0 Hz), 5.8-6.2 (7 H, br m) and 3.6 (3 H, s); 13C
NMR (THF-d8, 25 °C): δ 160.0 (C12), 141.0 (C1), 132.5 (C9),
132.9, 132.2, 131.9 (C5), 131.8, 131.4, 130.8, 127.9 (C10), 125.7
(C2), 114.1 (C11), 55.0 (C13). The italicized values are those for
C3, C4, and C6-C8, which undergo exchange due to bond shift.
Tr ibu tylsta n n ylcycloocta tetr a en e (3) was prepared by
the method employed for the synthesis of trimethylstannylCOT
by Stone and co-workers.49 n-Butyllithium (15.0 mL, 2.0 M in
hexane) was added over 2 min to a stirred solution of
bromocyclooctatetraene (5.00 g, 27.3 mmol) in ether (15 mL)
at -80 °C. The reaction mixture was stirred for an additional
1.5 h at -80 °C, 8.14 g (25.0 mmol) of tributylchlorostannane
was added, and the mixture was allowed to warm to room
temperature over 30 min. The mixture was hydrolyzed with
20 mL of satd aq NH4Cl and washed with 2 × 20 mL of satd
aq NH4Cl. After extraction of the aqueous layers with 2 × 50
mL of ether, the combined organic layers were dried (MgSO4)
and distilled to afford a yellow oil, bp 80 °C (0.1 mm) in 68%
Kin etic Meth od s: Gen er a l. All kinetics were measured
for at least two independent samples that had been sealed
under argon in 5 mm NMR tubes after five freeze-pump-
thaw cycles. The temperature of the sample was determined
from the difference in the proton chemical shifts of methanol
for temperatures <295 K according to the method of Van
Geet.50 The temperature accuracy is estimated to be (1 °C
due to improvements made to the variable temperature system
as recommended by Allerhand.51 The free induction signals
were sampled with 32K data points. After zero filling and
Fourier transformation, the spectral resolution was typically
in the range of 0.3-0.5 Hz/point.
Lin e Br oa d en in g. Bond shift rate constants (kBS) were
calculated at a number of temperatures from 13C signals where
line broadening could be determined. Line widths for the
pairwise exchanging carbons (C2 and C8, C3 and C7, C4 and
C6) could each, in principle, be measured, but kBS was
determined only from nonoverlapped signals. Rate constants
for bond shift were calculated by eq 5,52 where ∆νexch is the
exchange broadening (in Hz) of a signal in the slow exchange
kBS ) π∆νexch
(5)
region, i.e., the observed line width at half-height (∆ν1/2
)
corrected for the intrinsic line width at half-height. This
correction was determined as follows. From spectra at low
temperature, where bond shift exchange broadening was
absent, it was found that the line width of C5 was very similar
to those of the other proton-bearing carbons in the neutral COT
ring. At higher temperatures, the line width of C5, which does
not undergo exchange, was subtracted from that of the
exchange-broadened signals to give ∆νexch. All line widths were
determined by a least-squares fit of the peak to a Lorentzian
line shape.
1
yield. H NMR (CDCl3): δ 5.8-6.2 (7 H, broad m), 1.47 (6 H,
m), 1.36 (6 H, m), 0.91 (15 H, m). Calcd for C20H34Sn:
61.10%, H 8.72%. Found: C 61.14%, H 8.90%.
C
Uncertainties in the rate constants were estimated from
replicate measurements because differential error analysis
yielded unreasonably small uncertainties in the rate constants.
p-(Nitr op h en yl)cycloocta tetr a en e (1a ). To a stirred
solution of 1-iodo-4-nitrobenzene (0.249 g, 10.0 mmol), tris-
(dibenzylideneacetone)palladium(0) (23 mg, 0.025 mmol Pd),
and triphenylarsine (31 mg, 0.10 mmol) in 10 mL of THF at
room temperature was added 3.26 g (10 mmol) of tributyl-
stannylcyclooctatetraene in 10 mL of THF. After 24 h, the
reaction mixture was diluted with 10 mL of EtOAc, and 10
mL of 50% satd aq KF was added. The reaction mixture was
stirred for 15 min and filtered, and the organic layer was
washed with 20 mL of H2O and 20 mL of brine. The combined
aqueous layers were extracted with 20 mL of EtOAc, and the
combined organic layers were dried (MgSO4) and concentrated.
The residue was chromatographed on silica gel with 10% CH2-
Cl2/hexane to afford a yellow solid, mp 33-34 °C, in 70% yield.
Final purification was achieved by HPLC on a C18 column by
The error in ∆Gq was determined by propagation of the
BS
errors from the rate constants. The maximum error was found
to be (10% when kBS was obtained from line width measure-
ments.
Com p u ta tion a l Meth od s. Ab initio molecular orbital
calculations and geometry optimizations were performed using
the GAUSSIAN 92 series of programs30 at the Hartree-Fock
(HF) level of theory with the 3-21G(*) basis set.29 Fully
optimized geometries of the GSs and RI TSs were shown to
have zero and one imaginary frequency, respectively, by
analytical frequency analysis.
Ack n ow led gm en t. We thank the National Science
Foundation for partial support of this work and Intel
Corporation for the donation of a computer to Carnegie
Mellon University. We also thank Scott Vignon for
performing several calculations and Professor Hyung
Kim for enlightening discussions.
1
elution with acetonitrile. H NMR (THF-d8, 25 °C): δ 8.15 (2
H, d, J ) 9.0 Hz), 7.58 (2 H, d, J ) 9.0 Hz), 5.8 (7 H, br m);
13C NMR (THF-d8, 25 °C): δ 147.9 (C9), 147.2 (C12), 140.9 (C1),
134.4, 133.0, 132.9 (C5), 132.8 (C2), 132.6, 132.4, 132.3, 127.6
(C10), 124.2 (C11). The italicized values are those for C3, C4 and
C6-C8, which undergo exchange due to bond shift. IR (KBr):
3078, 3002, 1286, 1540 cm-1. Calcd for C14H11NO2: C 74.65%,
H 4.91%, N 6.22%. Found: C 74.44%, H 4.96%, N 6.32%.
1,4-Dicycloocta tetr a en ylben zen e10a (2). 1H NMR (THF-
d8, 25 °C): δ 7.31 (4 H, s), 5.8 (14 H, br m); 13C NMR (THF-d8,
25 °C): δ 142.2 (C9), 140.2 (C9), 133.6, 133.3, 133.1, 132.9 (C5),
132.6, 132.5, 128.7 (C2), 126.7 (C10). The italicized values are
those for C3, C4, and C6-C8, which undergo exchange due to
bond shift.
Su p p or tin g In for m a tion Ava ila ble: Rate constants and
temperatures for bond shift in 1a -c, 2, and 22-/2K+ (Tables
S1-S5) and the corresponding van’t Hoff plots (Figures S1-
S5); HF/3-21G(*) energies and Z-matrixes for 1a -c, 2, 22-/2K+,
and 22-/2K+ (solvated K+) (Tables S6-S17). This material is
available free of charge via the Internet at http://pubs.acs.org.
1,4-Dicyclooctatetr aen ylben zen edipotassiu m (22-/2K+).
The synthesis and NMR chemical shifts for this compound
have been reported previously.10a
J O010526H
(50) Van Geet, A. L. Anal. Chem. 1970, 42, 679.
(51) Allerhand, A.; Addleman, R. E.; Osman, D. J . Am. Chem. Soc.
1985, 107, 5809.
(49) Cooke, M.; Russ, C. R.; Stone, F. G. A. J . Chem. Soc., Dalton
Trans. 1975, 256.
(52) Sandstro¨m, J . Dynamic NMR Spectroscopy; Academic Press:
London, 1982; pp 14-18.