Reduction of benzylidene dibenzo[ a, d ]cycloheptenes: Over-reduction of antiaromatic dianions to aromatic tetraanions

Article abstract of DOI:10.1021/jo200227t

The antiaromaticity of a series of dianions of p-substituted benzylidene dibenzo[a,d]cycloheptenes was examined through calculated measures of antiaromaticity. The nucleus-independent chemical shifts (NICS) and magnetic susceptibility exaltation both show

Full text of DOI:10.1021/jo200227t

ARTICLE
pubs.acs.org/joc
Reduction of Benzylidene Dibenzo[a,d]cycloheptenes:
Over-Reduction of Antiaromatic Dianions to Aromatic Tetraanions
Blakely Tresca, MacDonald Higbee, and Nancy S. Mills*
Department of Chemistry, Trinity University, San Antonio, Texas 78212-7200, United States
S Supporting Information
b
ABSTRACT: The antiaromaticity of a series of dianions of p-substituted benzylidene
dibenzo[a,d]cycloheptenes was examined through calculated measures of antiaroma-
ticity. The nucleus-independent chemical shifts (NICS) and magnetic susceptibility
exaltation both showed substantial antiaromatic character in the benzannulated
tropylium anion. When the antiaromaticity was normalized for the area of the ring,
these tropylium anions were shown to be among the most antiaromatic anions in the
chemical literature. Attempts to make the dianion through reduction with lithium or
potassium gave the tetraanion as the only species observable in the 1H NMR spectrum.
Quench of the reaction mixture with trimethylsilyl chloride or D2O confirmed the
presence of the tetraanion, but only as a small portion of the reaction mixture, with the
major product being unreacted starting material. The failure to observe starting material
was attributed to similarities in the structures of the starting material and anion radical
(first reduction), allowing rapid electron transfer between them. The inability to see the dianion (second reduction) could be the
result of the very small HOMOꢀLUMO gap anticipated for highly antiaromatic species, which would allow access to diradical
species. The magnitude of the HOMOꢀLUMO gap was determined by the difference between the HOMO and LUMO energies
from geometry optimization and the lowest energy transition from TD-DFT calculations. The HOMOꢀLUMO gap for the
benzylidene dibenzocycloheptatriene dianions was shown to be much smaller than the HOMOꢀLUMO gap of species for which
1H NMR spectra had been observed.
’ INTRODUCTION
Herein we report the attempt to prepare and characterize
dianions of substituted benzylidene dibenzocycloheptenes (1)
through reduction with lithium or potassium. We anticipated that
the electron-donating (destabilizing) group on the benzyl sub-
stituent would enhance the antiaromaticity of the dibenzotropy-
lium anion just as electron-withdrawing groups on the benzyl
substituent of the analogous benzylidene fluorene dications had
enhanced the antiaromaticity of the fluorenyl system.7 We
observed instead the over-reduction of the desired dianion
(12ꢀ) to an unexpected tetraanion (14ꢀ), which contains an
aromatic dibenzotropylium trianion (Scheme 1). Of particular
interest was our inability to observe the dianion in the 1H NMR
spectrum of the reaction mixture. There are two possibilities for
this result: that the dianion was sufficiently unstable that it was
immediately reduced to the tetraanion or that the antiaromatic
dianion had a sufficiently small HOMO/LUMO gap to give the
dianion appreciable diradical character. The presence of the
tetraanion is confirmed through comparison of its experimental
and calculated chemical shifts and through mass spectral data of
its quench products. The evidence for the small HOMO/LUMO
gap comes from the ΔEHOMOꢀELUMO from the geometry
optimization and from the lowest energy electronic transition
from TD-DFT calculations.
Aromaticity has occupied a central role in organic chemistry
because the stability associated with the aromatic sextet has
dictated the formation of products from a wide variety of
reactions. Interest in the aromatic sextet and related expansion
to other species with a H€uckel number of electrons in a
conjugated π-system led to the development of a number of
nonbenzenoid aromatic species. These species were determined
to be aromatic based on a number of properties associated
with benzene. Their stability was evaluated through energetic
measures, such as aromatic stabilization energy,1 through struc-
tural measures, such as planarity and lack of bond length
alternation,2 and through magnetic measures, such as the diatropic
shift of protons on the periphery of the ring system,3 nucleus-
independent chemical shifts (NICS),4,5 and magnetic suscept-
ibility exaltation.6
Much less attention has been paid to the preparation of
antiaromatic systems because of their anticipated instability.
We have been able to prepare and characterize experimentally
a number of antiaromatic dications of fluorenyl systems7ꢀ11 and
have extended those studies to antiaromatic indenylidene
dications.12,13 More recently, we have prepared antiaromatic
dianions containing dibenzotropylium anions14 as well as those
containing heterocyclic antiaromatic systems such as the dianion
of dixanthyidene.15
Received: January 30, 2011
Published: May 24, 2011
r
2011 American Chemical Society
5539
dx.doi.org/10.1021/jo200227t J. Org. Chem. 2011, 76, 5539–5546
|

The Journal of Organic Chemistry
ARTICLE
Scheme 1. Reduction of 1
Table 1. Nucleus-Independent Chemical Shifts, NICS(1)zza, for 1a2ꢀꢀ1f2ꢀ
NICS(1)zz, 7-ring, ppm
NICS(1)zz, 6- ring, ppm
ΣNICS(1)zz for all rings, ppm
ΣNICS(1)zz/area2, ppm/Å2
1a2ꢀ
1b2ꢀ
1c2ꢀ
1d2ꢀ
1e2ꢀ
1f2ꢀ
108.02
115.79
116.74
118.62
118.40
125.91
42.41
48.56
49.51
51.29
51.46
56.72
192.85
212.92
215.77
221.21
221.32
239.35
0.609
0.668
0.677
0.694
0.695
0.751
a Shifts calculated with the GIAO method with basis set B3LYP/6-311þG(d,p) on geometries optimized at the B3LYP/6-31G(d) level. Ghost atom
placed 1 Å above the ring. NICS evaluated using only the component of the magnetic tensor parallel to the π-system.17
Table 2. Magnetic Susceptibilitya, X, and Magnetic Suscept-
’ RESULTS AND DISCUSSION
ibility Exaltation, Λ, for 1a2ꢀꢀ1f2ꢀ
Calculated Measures of Antiaromaticity for 12ꢀ. Nucleus-
Xb
Λb
Λ/area2
Independent Chemical Shifts. The ring current model of aro-
maticity predicts a magnetic environment in the center of an
aromatic ring that is opposite in sign to that of the periphery,
resulting in protons that are shifted downfield when on the
periphery of an aromatic species and would be shifted upfield if
present in the center of the aromatic ring. These effects would be
opposite for an antiaromatic species, and these predictions have
1a2ꢀ
1b2ꢀ
1c2ꢀ
1d2ꢀ
1e2ꢀ
1f2ꢀ
ꢀ33.52
ꢀ17.23
ꢀ7.63
10.68
2.94
129.82
141.58
145.52
149.37
147.95
150.42
0.410
0.444
0.456
0.469
0.465
0.472
1
0.73
been given experimental support in the H NMR characteriza-
a Shifts calculated with the CSGT method with basis set B3LYP/6-
311þG(d,p) on geometries optimized at the B3LYP/6-31G(d) level.
See Supporting Information for details of the calculation. b csgt/ppm.
tion of [18]annulene and its dianion.16 Nucleus-independent
chemical shifts, NICS, are based on this phenomenon, with the
calculation of the magnetic shielding tensor for a ghost atom
placed in the center of the ring system.4 NICS for an aromatic
system have a negative sign; those for an antiaromatic system
have a positive sign.
to the sum of the magnetic susceptibility for the individual bonds
in the system, giving in essence the magnetic susceptibility for the
nondelocalized system. The difference between the two suscept-
ibility values is the exaltation due to the ring current (see
Supporting Information for details of the calculations). As is
true for NICS, a negative value of Λ is associated with aromatic
systems and a positive value with antiaromatic systems. The
calculated magnetic susceptibilities, X, of 1a2ꢀꢀ1f2ꢀ and the
magnetic susceptibility exaltation Λ are given in Table 2. Com-
parison of the ΣNICS(1)zz and Λ shows similar trends in the
effect of the substituent on the antiaromaticity of the dibenzo-
tropylium anion. The magnitude of both Λ6 and ΣNICS(1)zz22 is
related to the square area of the system under examination. For
similarly sized species such as 1a2ꢀꢀ1f2ꢀ, it is reasonable to
compare the NICS and Λ values directly. However, to place
these values in a larger context by comparison with other
aromatic and antiaromatic species, each value needs to be
normalized by the square area of the ring system. We have
demonstrated a linear relationship between the ΣNICS(1)zz/sq
area and Λ/sq area,22 and that relationship is shown in Figure 1,
The calculated NICS(1)zz values for the dibenzotropylium
systems of 1a2ꢀꢀ1f2ꢀ are given in Table 1. The positive values
support the prediction of antiaromaticity in these anionic diben-
zotropylium systems, with larger values indicating greater anti-
aromaticity. In general, electron-withdrawing substituents cause
a decrease in the antiaromaticity due to the stabilization of the
anionic system. The methoxy substituent is behaving more as an
inductively withdrawing group than a resonance-donating group,
as seen by the decreased antiaromaticity. The methyl-substituted
dianion is the most antiaromatic system in the series.
Magnetic Susceptibility Exaltation. A second magnetic mea-
sure of aromaticity and antiaromaticity arises from the additional
response to a magnetic field of a species with a ring current. This
additional magnetic susceptibility is called magnetic susceptibility
exaltation, Λ.6,18ꢀ21 The magnetic susceptibility, X, can be
calculated with the CSGT (continuous set of gauge trans-
formations) method for the entire species. This is then compared
5540
dx.doi.org/10.1021/jo200227t |J. Org. Chem. 2011, 76, 5539–5546

The Journal of Organic Chemistry
ARTICLE
Scheme 2. Preparation of p-Substituted 1
benzylic protons and the protons on the dibenzotropylium rings
in Figure 3. The relationship for the phenyl protons is fairly linear
for both the dianions and tetraanions, as would be expected since
in both systems the amount of charge would be basically the
same. However, there is no correlation for the dibenzotropy-
lium/benzylic protons in the dianion, with a reasonable correla-
tion for the tetraanion, R = 0.941.
Figure 1. Relationship between ΣNICS(1)zz/sq area and Λ/sq area for
1aꢀf2ꢀ and a set of mono- and dianions,22 3,4-dibenzophenathrene
dianion, and planar cyclooctatetraene, calculated as described in Tables 1
and 2.
which includes a set of representative anions from ref 22 (see
Supporting Information). Notice that 1a2ꢀꢀ1f2ꢀ are the most
antiaromatic anions in this group which includes the dianion of
3,4-dibenzophenanthrene, vide infra.23 The plot also includes
planar (D4h) cyclooctatetraene, which is presumed to be the
transition state for inversion of the tub-shaped geometry of
cyclooctatetraene.24
1
The experimental H NMR chemical shifts and shifts calcu-
lated for Li4[1b,dꢀf] are given in Table 3. The chemical shifts
calculated for Li2[1b,dꢀf] can be found in the Supporting
Information. The proton assignments were made by a combina-
tion of COSY data and comparison with calculated shifts.
Identification of Li41d in the Reaction Mixture. The reac-
tion mixture after ∼1 h of sonication was quenched with either
(CH3)3SiCl or D2O in dry THF. Tetrasilylated or deuterated
material was present in the reaction mixture, but it was only a
small percentage of the silylated or deuterated material. Present
in much larger concentration was unreacted starting material,
followed by several trisilylated/deuterated products, then tetra-
substituted product, and disubstituted product (see Table 4).
Attempted Preparation of 1a2ꢀꢀ1f2ꢀ through Reduction
with Lithium or Potassium. Benzylidene dibenzocyclohepta-
trienes 1bꢀf were synthesized through reaction of a substituted
phenyl Grignard reagent with dibenzosuberenone, followed by de-
hydration (Scheme 2). The trifluoromethyl substituted compound
1a was prepared through reaction of the Wittig product from
1-bromo-4-trifluoromethylbenzene with dibenzosuberenone.
The reduction of neutral precursors 1aꢀf with lithium and
potassium was attempted according to the method of Rabinovitz
and Scott25 and monitored by 1H NMR spectroscopy. Within 10 min
of sonication, the solution turned brown, and all peaks dis-
appeared, with no peaks visible until after one hour of sonication.
The 1H NMR spectrum of the reduction by lithium of 1d after
one hour is shown in Figure 2. The COSY spectrum of the
reaction mixture as well as the spectra from reduction of 1b, e,
and f are given in the Supporting Information. Neither 1a nor 1c
gave clean enough reduction products for identification of the
species present.
1
The H NMR spectrum of the reaction mixture in Figure 2
clearly shows no unreacted starting material (see Supporting
Information for the full spectrum) nor does it show substantial
amounts of anything but Li41d. We did not anticipate seeing the
mono- or trianion radicals in the spectrum of the reaction
mixture, but the absence of spectral evidence for the starting
material and dianion apparent from the quenches was of concern.
Spectral Invisibility of 1d in the Presence of an Anion
Radical. The dominance of unreacted starting material in the
mixture of products from quench with (CH3)3SiCl or D2O was
of concern to us because it should have been apparent in the
NMR spectrum of the reaction mixture. We considered the
possibility of inadvertent quench of the anionic species by
adventitious water but saw substantial amounts of starting
material even with the most stringently anhydrous conditions.
A possible explanation is based on in the changes seen in the 1H
NMR spectrum during reduction of cyclooctatetrane, COT, to
the dianion with lithium or potassium.28 The electron transfer
between the anion radical and dianion of COT was rapid,
resulting in broadening of the NMR signal, with the singlet of
the dianion becoming sharp when two equivalents of metal were
consumed. That is, while both anion radical and dianion existed
in solution, the signal was broad. The spectrum of unreacted
COT remained sharp until it was consumed. The explanation for
the differing behavior of the two diamagnetic species involved
We assumed that the reduction would take place by one
electron transfers giving initially the anion radical followed by the
antiaromatic dianion 12ꢀ, which would presumably exist as Li21,
vide infra. We therefore assumed that the species present after
one hour was the dianion since the anion radical would be
1
invisible in the H NMR spectrum due to broadening.26 The
linear correlation between the calculated and experimental shifts
seen previously for antiaromatic dications and dianions14 would
support the identification of the dianion. However, the agree-
ment was particularly poor for Li2[1b,dꢀf] (Figure 3a) but
substantially better for Li4[1b,dꢀf] (Figure 3b). For both sets of
data, the shifts were calculated with the inclusion of solvent using
the polarization continuum method, PCM.27 We have presented
the data with the protons on the phenyl rings distinct from the
5541
dx.doi.org/10.1021/jo200227t |J. Org. Chem. 2011, 76, 5539–5546

The Journal of Organic Chemistry
ARTICLE
Figure 2. 1H NMR spectrum of the reaction mixture from reduction of 1d with lithium. TMS as reference. Peak at δ3.57 ppm due to residual THF-d7.
Figure 3. Comparison of experimental 1H NMR chemical shifts from reduction of 1b,dꢀf with shifts. Solvent was included in the calculations through
use of the polarization continuum method, PCM. (a) Li21b,dꢀf. (b) Li41b,dꢀf.
their geometry. Neutral COT is tub-shaped, while the COT
dianion is planar. Because electron transfer between two species
with similar geometries is rapid, the broadening of the signal of
the COT dianion suggested that the anion radical must also be
planar. We have calculated the geometries of 1 and 1•ꢀ, and they
are very similar and different from the geometry of Li41 (see
Figure 4 for 1 and 1•ꢀ and Supporting Information). Thus,
electron transfer between 1 and 1•ꢀ would be expected to be
rapid, resulting in broadening of the signal for starting material.
Spectral Invisibility of Li21d. Our inability to see any inter-
is based on stability. Molecules with large HOMOꢀLUMO gaps
have been shown to be stable and unreactive; those with small
HOMOꢀLUMO gaps have been shown to be chemically
reactive.29,30 The theoretical basis between the HOMOꢀLUMO
gap and stability was articulated by Pearson, defined as hardness,
η,31 and considered as a measure of aromaticity.32 Minsky et al.33
and Shenhar25 have reported the effect of a small HOMOꢀLU-
MO gap on the NMR spectra of a series of dianions of polycyclic
aromatic hydrocarbons.
There are two methods for obtaining information about the
HOMOꢀLUMO gap available through computation. The geo-
metry optimization at an appropriate level provides the energies
of the HOMO and LUMO. The energy for the first vertical
excited state, from the HOMO to the LUMO, would approx-
imate the HOMOꢀLUMO gap and can be calculated with time-
1
pretable H NMR signal during the reduction of 1d until the
formation of Li41d can be understood if the dianion was
sufficiently antiaromatic that it possessed a very small HOMOꢀ
LUMO gap, allowing for diradial formation. The relationship be-
tween the HOMOꢀLUMO gap and aromaticity/antiaromaticity
5542
dx.doi.org/10.1021/jo200227t |J. Org. Chem. 2011, 76, 5539–5546

The Journal of Organic Chemistry
ARTICLE
Table 3. Experimentala and Calculatedb 1H NMR Chemical
Shifts for Li41b and Li4[1dꢀf]
HOMOꢀLUMO gap caused by different substituents on the
phenyl ring vary in an analogous way for each type of optimization
(see Supporting Information). In addition, NICS(1)zz values
calculated for the geometries obtained with the two methods also
show a linear relationship (see Supporting Information for a
complete discussion of dependence on basis set and functional).
Thus, it is appropriate to use the geometries optimized at this lower
level to obtain HOMOꢀLUMO energies as well as magnetic
properties.
Li41b
Li41d
Li41e
calc.
Li41f
calc.
exp.
calc.
exp.
calc.
exp.
exp.
1,10
2,9
3,8
4.7
5,6
R
6.91 7.186 6.99 8.093 6.91 8.100 6.888 8.086
5.50 5.602 5.46 5.542 5.47 5.560 5.461 5.650
5.37 5.829 5.39 5.878 5.41 5.815 5.294 5.827
6.98 7.541 7.02 7.475 7.01 7.527 6.999 7.530
5.68 5.965 5.71 6.027 5.71 5.973 5.714 5.991
4.45 5.380 4.49 5.284 4.49 5.265 4.401 5.399
6.85 6.388 6.82 6.478 6.83 6.627 6.844 6.320
3.98 3.579 3.85 3.727 3.87 3.512 3.730 2.778
6.34 5.658 6.33 5.956 6.32 5.890 6.232 5.799
6.07 5.604 6.04 5.653 6.04 5.596 5.887 5.632
5.06 4.577
A more troublesome issue deals with the association of the
counterion. Anions in solution are known to have close associa-
tions with their counterions, either as contact or solvent sepa-
rated ion pairs. It is clear from the relationship of the calculated
and experimental shifts for the tetraanions that they exist as
Li4[1aꢀf], although we have not determined whether these are
contact or solvent separated ion pairs. Monolithiated benzocyclo-
heptatrene is believed to exist as a contact ion pair, based on
chemical shift changes as a function of changes in temperature
and counterion.38 While we cannot confirm that 1a2ꢀꢀ1f2ꢀ
exist as Li2[1aꢀf], it is reasonable to assume that this is the case.
Thus, it would also be reasonable to expect that the properties of
the lithiated species should be examined rather than those of the
dianion. There are two primary problems. There are no reports in
the literature for the determination of NICS values and Λ on
lithiated species. Indeed, it is difficult to determine the appropriate
manner to include both the counterion and ghost atom necessary
for the evaluation of NICS and difficult to determine the
appropriate reference system that would include the counterion
for the calculation of Λ. These problems are more difficult when
one also includes the molecules of THF providing solvation for the
counterion. The calculations also become prohibitively expensive,
particularly TD-DFT, asthe molecules become more complicated.
However, the work of Rabinovitz et al.25,39,40 has demon-
strated that the magnitude of the HOMOꢀLUMO gap calcu-
lated on dianions without counterions explains changes in their
NMR spectra. Since this is the question we are examining, we will
use the same process. We have previously been able to obtain 1H
NMR spectra for antiaromatic dications and dianions, which
suggests that their HOMOꢀLUMO gaps must have been suffi-
ciently great to prohibit much diradical character. Three of those
o
o
m
m
p
subst
1.877 1.6330
a Spectra in THF-d8, temperature = 10 °C. b Chemical shifts calculated
with the GIAO method with B3LYP/6-311þG(d,p) for geometries
optimized at the B3LYP/6-31G(d) level; solvent (THF) included with
the PCM method.
Table 4. Amounts of Di-, Tri-, and Tetrasilylated Product
from Li41d,a from Reaction with Trimethylsilyl Chloride
trisilylated trisilylated trisilylated tetrasilylated
disilylated product-a product-b product-c
product
percentage of all
silylated product
4% 4% 22% 62%
8%
a From quench of a reduction mixture after 1 h of sonication. The H
1
NMR shifts for the reduction mixture correspond to those in Figure 2.
species are dication 22þ41 and dianions 32ꢀ14 and 42ꢀ 42
. We also
consider the HOMOꢀLUMO gap of the dianion of 3,4-dibenzo-
phenathrene, 52ꢀ, to make a link to the previous work of Minsky
et al.,23 which describes the effect of the HOMOꢀLUMO gap
on its NMR spectrum. The HOMOꢀLUMO gap calculated
on geometries optimized at the B3LYP/6-31g(d) level for
1a2ꢀꢀ1f2ꢀ and 22þ, 32ꢀ, 42ꢀ, and 52ꢀ are reported in Table 5,
along with the calculated energy for the first excited state for a
subset of these species. The magnitude of the energies of
ΔEHOMOꢀELUMO and the lowest energy electronic transition
demonstrate that the HOMOꢀLUMO gap is indeed smaller for
1a2ꢀꢀ1f2ꢀ than for antiaromatic 22þ, 32ꢀ, and 42ꢀ, whose
NMR spectra we have been able to determine. The reduction of
3,4-dibenzophenanthrene to 52ꢀ with lithium, sodium, and
potassium failed to give an NMR spectrum with observable 1H
and 13C signals. This was attributed to a “large population of the
excited triplet state with two unpaired electrons”. This would be
expected for a system with the very small HOMOꢀLUMO gap
shown in Table 5.
Figure 4. Calculated structures of 1d and its monoanion radical. (a)
Starting material (1d). (b) First reduction product.
dependent density functional theory, TD-DFT. We have
demonstrated a good relationship between the experimental
and calculated electronic spectra for antiaromatic dications.34
The challenge for the calculation of properties of anions
comes from several issues. While DFT methods are generally
very good for larger molecular systems, questions have been
raised about their use for anions because of a self-interaction
error in approximate functionals.35,36 The result is that the
HOMO energy is positive, even for stable anions, indicating
that a valence electron is unbound. A possible solution to this
problem lies in the use of a long-range corrected version of the
B3LYP functional.37 We have determined that the geometries
calculated for 1a2ꢀꢀ1f2ꢀ using B3LYP/6-31g(d) and the
long-range corrected functional LC-BLYP/6-31þg(d) give
HOMOꢀLUMO gap energies that demonstrate the same
relative changes in 1a2ꢀꢀ1f2ꢀ. That is, the changes in the
It remains to link the HOMOꢀLUMO gap to the antiaroma-
ticity of 1a2ꢀꢀ1f2ꢀ. As Figure 1 shows, the substituted dianions
of 1 are more antiaromatic than antiaromatic dianions we and
5543
dx.doi.org/10.1021/jo200227t |J. Org. Chem. 2011, 76, 5539–5546

The Journal of Organic Chemistry
ARTICLE
Table 5. ΔEHOMOꢀELUMOa and Efirst excitationb for 1a2ꢀꢀ1f2ꢀ and 22þ, 32ꢀ, 42ꢀ, and 52ꢀ
1a2ꢀ
1b2ꢀ
1c2ꢀ
1d2ꢀ
1e2ꢀ
1f2ꢀ
22þ
32ꢀ
42ꢀ
52ꢀ
ΔEHOMOꢀELUMO
1.584
1.412
1.196
s1.310
0.679
1.306
1.316
1.858
0.951
2.089
0.983
2.925
1.492
1.159
Efirst excitation
a Geometries optimized at B3LYP/6-31G(d). b Energies in eV calculated with B3LYP/6-31G(d) and the TD-DFT method, on geometries optimized at
the B3LYP/6-31G(d) level.34
others have previously reported (see Supporting Information for
the species represented in Figure 1). The NICS(1)zz/square area
and Λ/square area for 52ꢀ, whose lack of NMR signals supported
the presence of a substantial triplet population, suggests that
1a2ꢀꢀ1f2ꢀ might also possess substantial diradical character. We
will be examining the preparation of 1a2ꢀꢀ1f2ꢀ by deprotonation
to allow evaluation of a diradical species and to characterize it as
triplet vs singlet in a cleaner reaction medium.
over10 min to the refluxing flask. The flask was allowed to reflux for 2 h; the
solution became metallic gray; and most of the Mg was consumed.
Dibenzosuberenone (10.0 mmol) was dissolved in dry Et2O (50 mL)
and was added dropwise over 10 min to the refluxing flask. The reaction
mixture was allowed to reflux for 2 h. The reaction was quenched with sat.
aqueous NH4Cl (25 mL) and then extracted with 2 ꢁ 10 mL of Et2O. The
combined organic layers were washed with 3 ꢁ 20 mL of water and then
3 ꢁ 20 mL of brine. The solvent was removed under vacuum to give a
yellow oil. The yellow oil was dissolved in EtOH (20 mL), and H2SO4 in
glacial acetic acid (20 mL, 20%) was added. The oil and acid were allowed
to stir for 2 h, and then ice (50 mL) was added. The acid was neutralized
with solid NaHCO3, and the mixture was extracted with 3 ꢁ 20 mL of
Et2O. The combined organic layers were washed with 3 ꢁ 20 mL of sat.
aqueous NaHCO3, 3 ꢁ 20 mL of water, and 3 ꢁ 20 mL of brine. The
solvent was removed under vacuum to give a yellow oil. Purification in a
silica flash column with hexanes gave a white powder. Yield: 54%.
Melting Point and Spectral Data for 1c and 1f. 1c, mp
93ꢀ97 °C. 1H NMR (400 MHz, CDCl3) δ 3.758 (s, 3H), 6.454 (s, 1H),
6.660 (dt, J = 9 Hz, 1 Hz, 2H), 6.894 (dt, J = 8.4 Hz, 1 Hz, 2H), 6.922
(d, J = 12 Hz), 6.986 (d, J = 12 Hz, 1 H), 7.139ꢀ7.331 (m, 5H),
7.374ꢀ7.454 (m, 2H), 7.526 (bd, J = 7.5 Hz, 1H). 13C NMR (100 MHz,
CDCl3) δ 55.14, 113.30, 126.80, 126.93, 127.13, 128.29, 128.72, 128.83,
128.85, 129.13, 129.43, 130.38, 131.17, 131.44, 131.71, 134.38, 134.97,
137.69, 140.37, 142.79, 158.34. Calculated for C23H16O: C, 89.00; H,
5.85. Found: C, 88.82; H, 6.03.
’ SUMMARY
Dianions of 1 are among the most antiaromatic dianions
reported in the literature, in terms of their NICS(1)zz and
magnetic susceptibility exaltation, when normalized by the
square of the ring area. Attempts to prepare the dianions by
reduction with lithium or potassium resulted in over-reduction to
tetraanions. Identification of the major species in the 1H NMR
spectrum of the reaction mixture as tetraanion was made through
1
comparison of the calculated H NMR shift of the tetraanion
with the experimental spectrum. The relationship was substan-
tially more linear than that of the calculated chemical shifts of the
dianion with the experimental shifts. In both cases, comparison
was made for chemical shifts calculated for the lithiated species in
THF. However, quench of reaction mixtures which appear to
contain only tetraanion with (CH3)3SiCl or D2O revealed that
the major product was starting material, with evidence of mono-,
di-, tri-, and tetrasubstituted product. The lack of NMR spectral
evidence for unreacted starting material was rationalized through
the similarity of the structures of starting material and mono-
anion radical, leading to rapid electron transfer between the
neutral and negative species. The lack of spectral evidence for
dianion was rationalized by its apparent large antiaromaticity that
would result in a small HOMOꢀLUMO gap, allowing for
diradical character. The explanation was supported by a compar-
ison of ΔEHOMOꢀELUMO and by the lowest energy electron
transition calculated by TD-DFT methods for 12ꢀ, with other
antiaromatic dications and dianions, and by comparison to the
behavior of the dianion of 3,4-dibenzophenathrene.
1f, mp 112ꢀ114 °C. 1H NMR (400 MHz, CDCl3) δ 2.248 (s, 3H),
6.466 (s, 1H), 6.808 (d, J = 8.1 Hz, 1H), 6.917 (bd, J = 9.6, 2 H), 6.981
(d, J = 11.7H, 2 H), 7.109ꢀ7.222 (m, 2H), 7.257ꢀ7.315 (m, 3H),
7.361ꢀ7.461 (m, 2H), 7.520 (bd, J = 7.2 Hz, 1H). 13C NMR (100 MHz,
CDCl3) δ 21.14, 126.87, 126.94, 127.15, 128.28, 128.58, 128.69, 128.78,
128.84, 129.08, 129.10, 131.19, 131.38, 132.20, 133.96, 134.33, 134.90,
136.46, 137.63, 141.40, 142.65 Calculated for C23H18: C, 93.84; H, 6.16.
Found: C, 93.49; H, 6.24.
Preparation of Dianions. Benzylidene cycloheptatriene, 1 (10 mg,
∼0.040 mmol), was added to a specially designed NMR tube insert (see
Supporting Information) under Ar. Lithium wire (1 g, 150 mmol) was
cleaned of oxide and mineral oil and added to the insert, followed by
THF-d8 (0.7 mL). The solution was degassed by freezeꢀpumpꢀthaw
and the insert flame-sealed under vacuum. The insert was inverted to
bring the metal in contact with 1, and reduction was initiated by
sonication at 10 °C. Reduction appeared to begin after less than 5 min,
with the development of a dark brown-orange solution. The products
were monitored with NMR spectroscopy at 10 °C. The 1H NMR spectra
in Figure 2 and in the Supporting Information were taken after 1 h.
Quench Studies. Trimethylsilyl chloride (0.25 mL, 2.0 mmol) or
D2O (0.10 mL, 5.5 mmol) was added to dry THF (5 mL) and cooled to
ꢀ78 °C in a septum capped flask. Previously prepared reduction samples
were transferred to the quench solution with stirring, and the solution
’ EXPERIMENTAL SECTION
Preparation of 1bꢀf. . The general procedure for formation of
1b,43 1c, 1d,43 1e,44 and 1f is shown below for the unsubstituted benzyli-
dene dibenzocycloheptene 1d.
5H-Benzylidene-dibenzo-(a,c)-cycloheptatriene, 1d. Mg (16.5 mmol)
was added to dry Et2O (50 mL) and heated to reflux. Benzyl bromide
(15.0 mmol) was dissolved in dry Et2O (20 mL) and was added dropwise
5544
dx.doi.org/10.1021/jo200227t |J. Org. Chem. 2011, 76, 5539–5546

The Journal of Organic Chemistry
ARTICLE
was allowed to warm to room temperature. The solution changed from
dark brown-orange to clear or pale yellow. The reaction mixture was
neutralized with 10% HCl (2 mL); water (40 mL) was added. The
solution was extracted with 3 ꢁ 10 mL of Et2O, and the combined
organic layers were washed with 2 ꢁ 10 mL of sat. aq. NaHCO3. The
solvent was removed under vacuum to give a white powder. The samples
were purified with a silica plug with hexanes, and solvent was removed
under vacuum. Solutions of the quench products in Et2O (1%) were
analyzed via GC/mass spectrometry (see Supporting Information for
details and mass spectral data).
Computational Methods. Geometries were optimized at B3LYP/
6-31G(d) density functional theory levels with the Gaussian 03 program
package.45 The chemical shifts were calculated at B3LYP/6-311þG(d,p)
using the GIAO method on the geometries optimized with lithium
counterions with the Gaussian 09 program.46 The best agreement
between experimental and calculated shifts was found when the effect
of solvent was included in the calculation through the polarization
continuum method, PCM, with the ε value of 7.4257, the default value
for THF in Gaussian 03. The nucleus-independent chemical shifts
(NICS(1)zz)4,17 were obtained from the chemical shift tensor perpendicular
to the ring for a ghost atom placed 1 Å above the center of each ring on
geometries optimized without counterions. Magnetic susceptibility
exaltation was determined from the magnetic susceptibility calculated
for the substituted 12ꢀ using the CSGT method with basis set B3LYP/
6-311þG(d,p) by subtraction of the magnetic susceptibility for the localized
system. Areas were calculated from the Cartesian coordinates for each
ring system oriented in the xy plane. See Supporting Information for
specific details of the calculation. TD-DFT calculations were done with
the B3LYP/6-311þG(d,p) on the geometries optimized at the B3LYP/
6-31G(d) level. See the Supporting Information for the discussion of the
validity of geometry optimization with B3LYP/6-31G(d) vs that with
functionals with long-range correction, such as LC-BLYP/6-31þG(d).
(2) Krygowski, T. M.; Cyranski, M. K. Chem. Rev. 2001, 101,
1385–1419.
(3) Mitchell, R. H. Chem. Rev. 2001, 101, 1301–1316.
(4) Schleyer, P. v. R.; Maerker, C.; Dransfeld, A.; Jiao, H.; Hommes,
N. J. v. E. J. Am. Chem. Soc. 1996, 118, 6317–6318.
(5) Chen, Z.; Wannere, C. S.; Corminboeuf, C.; Puchta, R.; Schleyer,
P. v. R. Chem. Rev. 2005, 105, 3842–3888.
(6) Gayoso, J.; Ouamerali, O. Rev. Roum. Chim. 1981, 26, 1035–1040.
(7) Do, C.; Hatfield, J.; Patel, S.; Vasudevan, D.; Tirla, C.; Mills, N. S.
J. Org. Chem. 2011, 76, 181–187.
(8) Dahl, B. J.; Mills, N. S. J. Am. Chem. Soc. 2008, 130, 10179–10186.
(9) Dahl, B. J.; Mills, N. S. Org. Lett. 2008, 10, 5605–5608.
(10) Mills, N. S.; Benish, M. A. J. Org. Chem. 2006, 71, 2207–2213.
(11) Mills, N. S.; Tirla, C.; Benish, M. A.; Rakowitz, A. J.; Bebell,
L. M.; Hurd, C. M. M.; Bria, A. L. M. J. Org. Chem. 2005, 70, 10709–
10716.
(12) Mills, N. S.; Cheng, F. E.; Baylan, J. M.; Tirla, C.; Hartmann,
J. L.; Patel, K. C.; Dahl, B. J.; McClintock, S. P. J. Org. Chem. 2011, 76,
645–653.
(13) Mills, N. S.; Llagostero, K. B.; Tirla, C.; Gordon, S.; Carpenetti,
D. J. Org. Chem. 2006, 71, 7940–7946.
(14) Piekarski, A. M.; Mills, N. S.; Yousef, A. J. Am. Chem. Soc. 2008,
130, 14883–14890.
(15) Black, M.; Woodford, C.; Mills, N. S. J. Org. Chem. 2011in
revision.
(16) Oth, J. F. M.; Woo, E. P.; Sondheimer, F. J. Am. Chem. Soc.
1973, 95, 7337–7345.
(17) Fallah-Bagher-Shaidaei, H.; Wannere, C. S.; Corminboeuf, C.;
Puchta, R.; Schleyer, P. v. R. Org. Lett. 2006, 8, 863–866.
(18) Haberditzl, W. Angew. Chem., Int. Ed. Engl. 1966, 5, 288–298.
(19) Dauben, H. J. J.; Wilson, D. J.; Laity, J. L. J. Am. Chem. Soc. 1968,
90, 811–813.
(20) Schmalz, T. G.; Norris, C. L.; Flygare, W. H. J. Am. Chem. Soc.
1973, 95, 7961–7967.
(21) Dauben, H. J.; Wilson, T. D.; L., L. In Nonbenzenoid Aromati-
city; Snyder, J. P., Ed.; Academic Press: New York, 1971; Vol. 2, pp
167ꢀ180.
’ ASSOCIATED CONTENT
(22) Mills, N. S.; Llagostera, K. B. J. Org. Chem. 2007, 72, 9163–9169.
(23) Minsky, A.; Meyer, A. Y.; Poupko, R.; Rabinovitz, M. J. Am.
Chem. Soc. 1983, 105, 2164–2172.
S
Supporting Information. Details on the calculation of
b
magnetic susceptibility, ring areas, and Cartesean coordinates,
frequencies, and energies of the optimized geometries of
1a2ꢀꢀ1f2ꢀ, Li2[1bꢀf], with and without explicit THF, and
Li41b and Li4[1dꢀf], and Li41d, with and without explicit THF,
at varying levels of theory; 1H NMR spectra of reduction of 1b,
1d (full spectrum), 1e, and 1f; 13C NMR, COSY, and HMQC
spectra of the reduction mixture of 1d; calculated chemical shifts
of dianions of Li41b and Li41dꢀf; mass spectra of the tetra-
silylated product and tetradeuterated product; diagram of the
insert used for reduction of 1. This material is available free of
charge via the Internet at http://pubs.acs.org.
(24) Nishinaga, T.; Ohmae, T.; Iyoda, M. Symmetry 2010, 2, 76–97.
(25) Shenhar, R.; Beust, R.; Hagen, S.; Bronstein, H. E.; Willner, I.;
Scott, L. T.; Rabinovitz, M. J. Chem. Soc., Perkin Trans. 2 2002, 449–454.
(26) Sharp, R. R. Nucl. Mag. Res. 1995, 24, 469–513.
(27) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005, 105,
2999–3093.
(28) Katz, T. J. J. Am. Chem. Soc. 1960, 82, 3785–3786.
(29) Aihara, J. J. Phys. Chem. A 1999, 103, 7487–7495.
(30) Aihara, J. Phys. Chem. Chem. Phys. 1999, 1, 3193–3197.
(31) Pearson, R. G. Acc. Chem. Res. 1993, 26, 250–255.
(32) Zhou, Z.; Parr, R. G. J. Am. Chem. Soc. 1989, 111, 7371–7379.
(33) Minsky, A.; Meyer, A. Y.; Hafner, K.; Rabinovitz, M. J. Am.
Chem. Soc. 1983, 105, 3975–3981.
’ AUTHOR INFORMATION
(34) Mills, N. S.; Levy, A.; Plummer, B. F. J. Org. Chem. 2004, 69,
6623–6633.
(35) Roesch, N.; Trickey, S. B. J. Chem. Phys. 1997, 106, 8940–8941.
(36) Galbarith, J. M.; Schaefer, H. F., III J. Chem. Phys. 1996,
105, 862–864.
(37) Jensen, F. J. Chem. Theory Comput. 2010, 6, 2726–2735.
(38) Sethson, I.; Johnels, D.; Edlund, U. Acta Chem. Scand. 1990,
44, 1029–1031.
(39) Rabinovitz, M.; Ayalon, A. Pure Appl. Chem. 1993, 65, 111–118.
(40) Benshafrut, R.; Shabtai, E.; Rabinovitz, M.; Scott, L. T. Eur. J.
Org. Chem. 2000, 1091–1106.
Corresponding Author
nmills@trinity.edu
’ ACKNOWLEDGMENT
We thank the Welch Foundation (Grant W-794), the National
Science Foundation (Grants CHE-0126417, CHE-0553589,
CHE-0957839, and CHE-0948445), and the Dr. John A. Burke, Jr.
research fund for their support of this research and Dr. Cheryl
Stevenson for helpful discussions.
(41) Mills, N. S. J. Org. Chem. 2002, 67, 7029–7036.
(42) Black, M.; Woodford, C.; Mills, N. S. J. Org. Chem. 2011,
76, 2286–2290.
’ REFERENCES
(1) Schleyer, P. v. R.; Puehlhofer, F. Org. Lett. 2002, 4, 2873–2876.
(43) Bergmann, E. D.; Solomonovici, A. Synthesis 1970, 2, 183–189.
5545
dx.doi.org/10.1021/jo200227t |J. Org. Chem. 2011, 76, 5539–5546

The Journal of Organic Chemistry
ARTICLE
(44) Fox, H. H.; Gibas, J. T.; Lee, H. L.; Boris, A. J. Med. Chem. 1964,
7, 790–792.
(45) Frisch, M. J. T., G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.;
Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.;
Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.;
Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa,
J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.;
Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.;
Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.;
Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.;
Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich,
S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.;
Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.;
Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.;
Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham,
M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.;
Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian
03, revision B.03 ed.; Gaussian, Inc.: Pittsburgh PA, 2004.
(46) R., A.; Frisch, M. J.; T., G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; C., J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson,
G. A.; N., H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; J. B.,
Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; K. T.; Fukuda, R.;
Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; K., O.; Nakai, H.;
Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; O., F.; Bearpark, M.;
Heyd, J. J.; Brothers, E.; Kudin, K. N.; S., V. N.; Kobayashi, R.; Normand,
J.; Raghavachari, K.; R., A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi,
M.; R., N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.;
A., C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; A. J. A.;
Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; M., K.;
Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; D., J. J.; Dapprich, S.;
Daniels, A. D.; Farkas, O.; F., J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J.
Gaussian 09; Gaussian, Inc.: Wallingford CT, 2009.
5546
dx.doi.org/10.1021/jo200227t |J. Org. Chem. 2011, 76, 5539–5546