Benzocyclobutenone enolate: An anion with an antiaromatic resonance structure

Article abstract of DOI:10.1039/a905868k

Benzocyclobutenone enolate (1a) was synthesized by deprotonating its conjugate acid with fluoride ion in the gas phase using a Fourier transform mass spectrometer. Reactions of 1a were explored and C-alkylation was found to be greatly preferred over oxygen attack. Thermodynamic properties (ΔH○_acid(1) = 360.3 ± 2.1 kcal mol^-1, EA(benzocyclobutenyl radical) = 1.90 ± 0.10 eV, and BDE(C-H) = 90.5 ± 3.1 kcal mol^-1) also were measured. The results are contrasted to ab initio calculations on cyclobutanone, benzocyclobutenone, and cyclobutenone enolates. Natural bond orbital and Bader analyses are reported too. Given our gas phase and computational results, we were able to devise conditions for the generation and trapping of 1a in tetrahydrofuran despite previous failings reported in the literature.

Full text of DOI:10.1039/a905868k

View Article Online / Journal Homepage / Table of Contents for this issue
Benzocyclobutenone enolate: an anion with an antiaromatic
resonance structure†
Katherine M. Broadus and Steven R. Kass*
Department of Chemistry, University of Minnesota, Minneapolis, MN 55455, USA
Received (in Cambridge, UK) 20th July 1999, Accepted 24th August 1999
Benzocyclobutenone enolate (1a) was synthesized by deprotonating its conjugate acid with fluoride ion in the gas
phase using a Fourier transform mass spectrometer. Reactions of 1a were explored and C-alkylation was found
to be greatly preferred over oxygen attack. Thermodynamic properties (∆HЊacid(1) = 360.3 2.1 kcal molϪ1,
EA(benzocyclobutenyl radical) = 1.90 0.10 eV, and BDE(C–H) = 90.5 3.1 kcal molϪ1) also were measured. The
results are contrasted to ab initio calculations on cyclobutanone, benzocyclobutenone, and cyclobutenone enolates.
Natural bond orbital and Bader analyses are reported too. Given our gas phase and computational results, we were
able to devise conditions for the generation and trapping of 1a in tetrahydrofuran despite previous failings reported
in the literature.
Introduction
Enolates are important nucleophiles for forming carbon–
carbon bonds in synthetic and biochemical transformations.1,2
These ions also pose interesting questions in regard to their
charge distributions and reactivities. For a typical enolate,
the major resonance contributor represents a structure in which
the charge resides on oxygen. The conjugate base of benzo-
cyclobutenone (1a) is unusual in that this resonance form is a
derivative of cyclobutadiene, an antiaromatic cyclic 4π electron
system. The destabilizing character of this resonance structure
is expected to influence the overall structure, reactivity and
thermochemistry of 1a.
Several years following the first reported synthesis of benzo-
cyclobutenone (1),3 evidence for its enolate was obtained when
the dimer 2 was isolated upon treatment of 1 with sodium
hydride [eqn. (1)].4–6 Carrying out this reaction in the presence
of one equivalent of benzaldehyde led to the recovery of acid 3
in a 50% yield. This product was postulated to arise from initial
trapping of benzocyclobutenone enolate (1a) followed by sub-
sequent cyclization and ring opening steps [eqn. (2)]. Detection
of trace amounts of lactones 4 and 5 is consistent with the
mechanism. Renewed efforts to trap benzocyclobutenone enol-
ate derivatives were reported almost thirty years later.7 In situ
trapping experiments with lithium tetramethylpiperidide
(LiTMP) and benzaldehyde at 0 ЊC afforded lactone 6 [eqn. (3)].
Formation of this product was proposed to go through a
phenide intermediate which was intercepted by quenching
with deuterium oxide. Furthermore, 2,2-dimethylbenzocyclo-
butenone undergoes this reaction suggesting that the inter-
mediate is not a benzocyclobutenone enolate.
oxygen reactivity for any given electrophile because the ionic
products (which are what is detected) have the same mass-to-
charge ratio. Collection of the neutral products of an ion–
molecule reaction requires a tremendous effort;8 consequently,
a series of reagents which yield unique ionic products with oxy
anions and carbanions have been developed.9–14 Perfluoroben-
zene, perfluoropropylene and perfluorinated toluene are three
such examples. These reagents react in a similar manner and
differ only slightly in the percentage of carbon versus oxygen
attack (Scheme 1). In the former case formation of an adduct
followed by the loss of one or more molecules of hydrogen
fluoride typically takes place whereas the latter pathway leads
to OϪ/F exchange. Thermodynamic (i.e., the keto vs. enol
energy difference) and frontier molecular orbital arguments
have been offered to account for the observed product ratios
While an enormous amount of enolate chemistry has been
studied in solution, gas-phase experiments have advantages in
that one can probe the intrinsic reactivity and thermochemistry
of a species. Unfortunately, mass spectrometry does not allow
one to easily address the intriguing question of carbon versus
† In memory of Robert R. Squires.
J. Chem. Soc., Perkin Trans. 2, 1999, 2389–2396
This journal is © The Royal Society of Chemistry 1999
2389

View Article Online
Scheme 1 General reaction of an enolate with perfluoropropylene.
with cyclic and acyclic enolates.11,13,14 Regrettably, the per-
Table 1 Summary of bracketing studies for benzocyclobutenone eno-
late (1a)
fluorinated hydrocarbons react predominantly through carbon.
An analogous reagent which differentiates between the two
pathways and displays higher oxygen reactivity has not been
developed.15,16
∆HЊacid
/
Proton
transfer
H/D
Exchange
Ref. acid
kcal molϪ1 a
In order to gain an understanding of benzocyclobutenone
enolate, we have carried out gas-phase experiments to explore
its thermochemistry and reactivity. Ab initio calculations have
been used to examine structural changes and charge redistribu-
tion upon anion formation in comparison to the enolates of
cyclobutanone (7) and cyclobutenone (8). We also report suc-
cessful trapping of benzocyclobutenone enolate in solution.
D2O
MeOD
t-BuOD
392.9 0.1
383.5 0.7
374.6 2.1
366.1 2.2
362.2 2.2
361.8 2.5
360.1 2.1
358.7 2.2
No
No
No
No
No
No
No
Yes (1)
N.a.
No
N.a.
N.a.
(CH3)2C᎐NOD
᎐
o-FC6H4NH2
CF3CH2OD
CF3C(CH3)2OH
Pyrrole
No
Yes b
Yes b
Yes
a Acidity values taken from refs. 26 and 27. MeOD value from S. E.
Barlow, T. T. Dang and V. M. Bierbaum, J. Am. Chem. Soc., 1990, 112,
6832. b Proton transfer is observed in both the forward and reverse
directions.
Experimental
Benzocyclobutenone, cyclobutanone and 1-phenylpropan-2-
one were prepared by published methods.17–19 Gas-phase
experiments were carried out in a dual cell model 2001 Finnigan
Fourier transform mass spectrometer (FTMS) equipped with a
3.0 T superconducting magnet. Benzocyclobutenone was
deprotonated upon reaction with fluoride, which was generated
by electron ionization of carbon tetrafluoride at 6 eV. The
resulting enolate was isolated by ejecting undesired ions using a
SWIFT waveform,20 transferred to the second cell and cooled
with an argon pulse (2 × 10Ϫ5 Torr). Neutral reagents were
introduced into the second cell through slow leak valves, and
the reactions of interest were monitored as a function of time.
Ab initio calculations were carried out using GAUSSIAN9421
on Unix-based workstations or Cray supercomputers. All struc-
tures were optimized at the HF and MP2 levels of theory with
the 6-31ϩG(d) basis set. Vibrational analysis at the HF and
MP2 levels was performed to ensure that energy minima struc-
tures were found. Zero-point energy corrections were made
using an empirical scaling factor of 0.9135 (HF) and 0.9646
(MP2).22 Temperature corrections from 0 to 298 K were carried
out by scaling the harmonic frequencies by 0.8929 (HF) and
0.9427 (MP2).22 The B3LYP functional was also employed for
certain species and the frequencies were used without adjust-
ment. Natural bond orbital (NBO) and Bader’s topological
analyses were carried out to examine charge and structural
changes.23–25
146.6, 134.9, 127.2, 122.1, 120.1, 59.3, Ϫ2.7; IR (neat) 3065,
2956, 2896, 1757, 1579, 1460, 1438, 1251, 1146, 1087, 959, 868,
ϩ
845, 753 cmϪ1; HRMS-EI M calcd for C11H14OSi 190.0814,
᝽
found 190.0802.
Results and discussion
Benzocyclobutenone enolate is readily generated in the gas
phase by the reaction of 1 with fluoride ion [eqn. (4)]. The
thermodynamic stability of this species can be assessed by
measuring its proton affinity and the electron affinity of the
corresponding radical. The former value was initially bracketed
by observing the occurrence or nonoccurrence of proton trans-
fer with a series of standard reference acids (Table 1).26,27 In
particular, benzocyclobutenone enolate was found to deproton-
ate trifluoroethanol, 2-(trifluoromethyl)propan-2-ol, and pyr-
role, but not less acidic compounds. In addition, upon reaction
with dimethyl oxime-OD ((CH ) C᎐NOD), 1a undergoes a
᎐
2
3
2-(Trimethylsilyl)benzocyclobutenone (9)
single hydrogen/deuterium exchange. Equilibrium measure-
ments with 2-(trifluoromethyl)propan-2-ol enabled us to obtain
a refined acidity value for 1 [eqn. (5)]. The bimolecular reaction
A solution of lithium tetramethylpiperidide was freshly pre-
pared by reacting tetramethylpiperidine (1.7 ml, 10 mmol) and
n-butyllithium (4 ml, 2.5 M, 10 mmol) in 150 ml of THF at 0 ЊC
for 30 minutes. The solution was then cooled to Ϫ78 ЊC and
chlorotrimethylsilane (20 ml, 0.16 mol) was added followed by
benzocyclobutenone (1.0 g, 8.2 mmol). After three hours of
stirring at Ϫ78 ЊC, the solution was allowed to slowly warm to
room temperature. The reaction was quenched with 1 M HCl
and diluted with hexanes after a period of 18 h. The organic
layer was washed excessively with water to remove THF, dried
with MgSO4 and concentrated by rotary evaporation. The
trapped product, 9, (0.32 g, 21%) was obtained after separation
from 1 (1 (9):1.5 (1)) by MPLC using 10% ethyl acetate in
hexanes. 1H NMR (300 MHz, CDCl3) δ 7.45–7.20 (m, 4H), 3.84
(s, 1H), 0.07 (s, 9H); 13C NMR (75 MHz, CDCl3) δ 191.2, 153.5,
rates in both directions, k1 and kϪ1, were obtained using eqn.
(6), where kobs = the slope of a plot of ln [anion] vs. time (in
kobsRxT
9.66 × 1018 [P]
(6)
k1 or kϪ1
=
2390
J. Chem. Soc., Perkin Trans. 2, 1999, 2389–2396

View Article Online
seconds), Rx = the pressure gauge correction for the neutral28 (a
function of the polarizability29), T = 300 K, and P = the pres-
sure of the neutral compound (Torr). Data were collected over
approximately two half-lives (3–10 s) following an argon pulse
(10Ϫ5 Torr) and a 500 ms vibrational cooling delay. The values
obtained, k1 = (2.50 0.16) × 10Ϫ10 and kϪ1 = (2.50 0.55) ×
10Ϫ9 cm3 moleculeϪ1 sϪ1, represent the average ( standard devi-
ation) of six and nine measurements, respectively, taken on
three separate days. Given the uncertainty associated with
accurate pressure readings, a 50% error in both rates has been
assumed for further data analysis.30 The value of ∆GЊrxn
(Ϫ1.42 0.15 kcal molϪ1) was combined with a calculated
∆SЊrxn (Ϫ3.70 e.u.),31 and the reported acidity of 2-
(trifluoromethyl)propan-2-ol (360.1 2.1 kcal molϪ1) to give
∆HЊacid (1) = 360.3 2.1 kcal molϪ1. This result is in good
agreement with theory which predicts a value of 359.2 kcal
molϪ1 (MP2/6-31ϩG(d) at 298 K).
Based on the bracketing results, trifluoroethanol also
appeared to be a reasonable acid for an equilibrium acidity
measurement. However, we found that in addition to proton
transfer, the reaction with this alcohol gives a trace of CF3Ϫ and
an m/z 169 product. The last of these species was identified as a
CHF3ؒCF3CH2OϪ cluster based upon high resolution mass
measurement and the observation of CF3CH2OϪ upon low
energy fragmentation (SORI).32 This cluster is not generated in
the FTMS upon reaction of the alcohol with trifluoromethyl
anion so benzocyclobutenone enolate appears to be involved in
its formation. A mechanism is outlined in Scheme 2. Proton
Table 2 Electron affinity bracketing results for benzocyclobutenone
radical
Ref. cmpd
EA/eVa
Electron transfer
Maleic anhydride
p-NO2C6H4COCH3
p-NO2C6H4CHO
3,5-di-CF3C6H3NO2
3,5-di-t-Bu-o-quinone
o-CH3-p-quinone
p-NCC6F4CN
1.44
1.57
1.691 0.087
1.79 0.10
1.804 0.087
1.852 0.087
1.89
1.91
2.00
0.10
0.10
No
No
No
No
No
No
No
Yes
Yes
0.10
0.10
0.10
p-Quinone
p-NO2C6H4NO2
a Values taken from refs. 26 and 27.
CF3CH2OϪ was produced, and except for the last case, proton
transfer was observed. 2-Fluoroethanol, however, reacts with
1a in similar fashion to trifluoroethanol and affords FϪؒHOCH2-
CH2F (Scheme 3). In this case ring expansion to a five- or
seven-membered lactone can, in principle, take place (pathways
a and b, respectively). Either way, this reaction is inefficient
compared to the one with 2,2,2-trifluoroethanol, probably
because the initial proton transfer is 11 kcal molϪ1 endothermic
in the former case and only 1.5 kcal molϪ1 endothermic in
the latter instance. 2-(Trifluoromethyl)propan-2-ol does not
react in an analogous fashion, presumably because of steric
hindrance.
To further examine the stability of benzocyclobutenone eno-
late, the electron affinity of the corresponding radical (1r) was
investigated by examining electron transfer reactions between
1a and a series of reference compounds.33 These experiments
were monitored as a function of time to ensure that any
observed electron transfer was due to the reaction of 1a with
the given reagent. The results are given in Table 2 and enable us
to assign EA(1r) = 1.90 0.10 eV, which compares favorably
with a directly calculated value of 1.76 eV at the UB3LYP/6-
31ϩG(d) level. An isogyric reaction with cyclobutanone
enolate (EA(7r) = 1.84 eV) at the same level of theory provides
even better agreement with experiment [1.94 eV, eqn. (7)]. Our
Scheme 2 Proposed mechanism for the reaction of 1a with trifluoro-
ethanol.
transfer leads to an ion–molecule complex which can dissociate
or the newly formed trifluoroethoxide may attack the carbonyl
carbon. Ring expansion via a tetrahedral intermediate (or
transition state) with concomitant expulsion of trifluoromethyl
anion leads to 10. Subsequent separation of the two com-
pounds gives CF3Ϫ, while a solvent switching reaction with tri-
fluoroethanol affords the m/z 169 cluster. The driving force for
this process is the relief of strain in benzocyclobutenone and it
is facilitated by the fact that a benzylic carbon, which is capable
of stabilizing a(n incipient) radical or charged center, is the
migrating group. The generality of this reaction was briefly
explored with the enolates of cyclobutanone, acetone, aceto-
phenone and 1-phenylpropan-2-one. In each case no CHF3ؒ
electron affinity and acidity measurements can be combined
to provide a value of 90.5 3.1 kcal molϪ1 for the C–H bond
energy in benzocyclobutenone [eqn. (8), Table 3].
BDE(HX) = ∆HЊacid(HX) Ϫ IP(H) ϩ EA(X)
(8)
The reactivity of benzocyclobutenone enolate was also exam-
ined and particular attention was paid to the nucleophilic site
(i.e., C- vs. O-attack). Perfluorobenzene reacts with 1a solely
through the carbon end of the ambident nucleophile to afford
Scheme 3 Reaction of benzocyclobutenone enolate with 2-fluoroethanol.
J. Chem. Soc., Perkin Trans. 2, 1999, 2389–2396
2391

View Article Online
Table 3 Comparison of the thermochemical properties of benzocy-
clobutenone. Parenthetical values correspond to computational results.
Acidities are at the MP2/6-31ϩG(d) (298 K) and electron affinities are
at the UB3LYP/6-31ϩG(d) (0 K) levels of theory
∆HЊacid
/
BDE/
Cmpd
kcal molϪ1
EA/eV
kcal molϪ1
Benzocyclobutenone 360.3 2.1
1.90 0.10
90.5 3.1
(1)
(359.2)
350.2 2.5
(1.76 and 1.94)a
1-Phenylpropan-
2-one
Cyclobutanone (7)
367.2 4.1
(364.3)
1.84 0.07 (1.66) 96.0 4.4
(0.89)
31ϩG(d) level], undoubtedly because of the antiaromatic
Cyclobutenone (8)
Benzocyclobutene
Ethylbenzene
a See text for details. b Unpublished results, M. C. Hare and S. R. Kass.
(381.4)
nature of the enol form; in this regard it is interesting to note
that at the same level of theory, ∆HKE = 55.2 kcal molϪ1 for
cyclobutenone [eqn. (11)]. Given the small C/O selectivity dif-
ference between 1a and 7a, their similar HOMO energies and
disparate keto-enol tautomerization energies, we conclude that
the HOMO–LUMO interaction is the key factor in determin-
ing the regioselectivity in this system. This conclusion is in
accord with a previous analysis of enolate ions by Zhong and
Brauman.16 Finally, it is worth mentioning that condensed-
phase studies show parallel behavior for small cyclic enolates
(i.e., alkylation of four- to six-membered rings occurs preferen-
tially at carbon).39
386.2 2.4b
379.8 2.4
<1.11
<98
84.6
0.80 0.13
2
an adduct Ϫ HF ion while perfluoropropylene gives 98.5%
C- and 1.5% O-attack.34 This difference is consistent with the
higher percentage of carbon reactivity previously reported for
hexafluorobenzene.11,13 It is interesting to note that if 1a is given
excess kinetic energy via a chirp35 excitation, a small amount of
oxygen alkylation is observed with C6F6 (~2–5% upon excit-
ation of 1a with 15 eV for 100–300 ms concurrent with a
10Ϫ5 Torr pulse of argon) and the proportion of this reaction
channel increases to approximately 10–20% with C3F6.
Additional reactivity studies with benzocyclobutenone eno-
late were carried out. The product ion obtained with neopentyl
nitrite [eqn. (12)] is consistent with the behavior of other eno-
There are two general explanations in the literature for the
observed regioselectivity of enolates with perfluorinated hydro-
carbons.11,13 Both correlate well with most of the reported data;
however, there are some anomalous results. One explanation
involves the energy difference between the keto and enol tauto-
mers (∆HKE) of the carbonyl compound.11 A thermodynamic
argument works in this case because the kinetically controlled
ratios have been shown to be a reflection of the initial reaction
step energetics (i.e. carbanion vs. oxy anion attack). Moreover,
alkylation and protonation are sufficiently similar energetically
that this difference can be tolerated.36,37 In general, carbon
selectivity has been observed when ∆HKE is large (~30–40
kcal molϪ1), whereas a smaller difference (∆HKE = 10–15 kcal
molϪ1) results in oxygen reactivity. Alternatively, interactions of
the enolate’s highest occupied molecular orbital (HOMO) and
the perfluorinated reagent’s lowest unoccupied molecular
orbital (LUMO) have been used to explain the carbon and
oxygen regioselectivity.13 The following trend has been observed
for C3F6 (LUMO = Ϫ4.53 eV): higher HOMO energies (> Ϫ1.7
eV) lead to predominantly oxygen attack whereas enolates with
lower HOMO energies (< Ϫ1.9 eV) display carbon reactivity. In
the case of C6F6 (LUMO = Ϫ4.83 eV), the values are shifted up
by 0.10 eV.
The small amount of oxy anion reactivity for 1a under
thermal conditions was not unexpected given that perfluoro-
propylene has been reported to react with cyclobutanone eno-
late entirely through carbon in an ICR and in a 96 (C):4 (O)
ratio in the higher pressure regime of a flowing afterglow.11,14
We repeated this measurement in our FTMS and observed ~5–
10% O-alkylation. This preference for carbon attack can be
accounted for by two main factors. First, the keto tautomer
is favored by 19.9 kcal molϪ1 [eqn. (9), MP2/6-31ϩG(d)] and
lates.40 Likewise, dimethyl disulfide reacts with 1a to afford an
α-thiomethyl enolate [eqn. (13)].41 The transformation with sul-
fur dioxide is interesting in that it slowly leads to the formation
of an adduct ion which has lost two hydrogen atoms. Experi-
ments with 1a-d1, produced by hydrogen–deuterium exchange,
suggest a benzyne-type product such as shown in eqn. (14).
Carbonyl sulfide and carbon disulfide only give adduct ions
upon reaction with 1a and no products are detected with
nitrous oxide.
Given the stability of benzocyclobutenone enolate in the gas
phase, the ability to generate the anion in solution seemed like a
manageable task. Our gas-phase acidity measurement for 1 can
be used to estimate a pKa value in dimethyl sulfoxide (DMSO)
if we assume that the difference between 1 and 1-phenylpropan-
2-one is the same in both phases. This is not an unreasonable
approximation for compounds that give delocalized ions since a
linear correlation with unit slope has been found between ∆HЊacid
(gas phase) and pKa(DMSO).42 Given that ∆∆HЊacid(1 Ϫ 1-
phenylpropan-2-one) = 10.1 kcal molϪ1 or 7.2 pK units and that
second, the HOMO energy of cyclobutanone enolate is a
relatively low Ϫ1.84 eV.38 The situation for benzocyclobutenone
enolate is further biased towards carbon reactivity. The HOMO
energy (Ϫ1.9 eV) is slightly lower and the keto form is much
more favored [eqn. (10), ∆HKE = 37.6 kcal molϪ1 at the MP2/6-
2392
J. Chem. Soc., Perkin Trans. 2, 1999, 2389–2396

View Article Online
Table 4 Structural changes upon deprotonation for 1, 7 and 8 at the
MP2/6-31ϩG(d) level of theory
Fig. 1 Changes in the C1–C2 bond upon deprotonation in benzo-
cyclobutenone (1), cyclobutanone (7), and cyclobutenone (8).
Bond or anglea
1 (Cs) 1a (Cs) 7 (Cs) 7a (C1) 8 (Cs) 8a (C2)
5). In contrast, the bonds emanating from the anionic site show
substantially increased s-character (C1–C2: 32.4% - 21.8% (1)
vs. 38.7%-32.7% (1a); C2–C3: 24.7%-31.0% (1) vs. 30.6%-
33.2% (1a) and undergo large bond contractions [0.141 (C1–
C2) and 0.058 Å (C2–C3)]. Overall, Bader’s analysis of the
bond order, and ellipticity changes provide a consistent theme
upon going from 1 to 1a: the C1–C2 and C2–C3 bonds become
C᎐O
1.218 1.250 1.218 1.274 1.218 1.236
1.567 1.426 1.531 1.397 1.567 1.518
1.524 1.466 1.555 1.516 1.523 1.422
1.403 1.428 1.555 1.561 1.358 1.422
1.500 1.572 1.531 1.555 1.499 1.518
1.395 1.371
1.400 1.436
1.410 1.391
1.405 1.428
1.391 1.390
᎐
C1–C2
C2–C3
C3–C4
C1–C4
C4–C5
C5–C6
C6–C7
C7–C8
C3–C8
C2–H
stronger and the C1–C4, C3–C4, and C᎐O bonds weaken
᎐
(Table 5).
The bond localization index (L(dcc)) can be used to assess the
distortion in the benzene moiety.51 This parameter is a simple
summation of absolute deviations of the individual bond
lengths from the average bond length. For example, L(dcc)
equals zero for D6h benzene and 0.36 for a hypothetical
cyclohexatriene. The aromatic ring in benzocyclobutenone is
almost entirely delocalized (L(dcc) = 0.03) but undergoes some
fixation upon deprotonation to 1a (L(dcc) = 0.14). This results in
bond alternation (C5–C6 and C7–C8 lengthen while C4–C5,
C6–C7, and C3–C8 shorten) and accompanying changes in
the bond order and ellipticity. The alternation in the benzene
moiety, however, is less significant than in the four-membered
ring.
1.096 1.087 1.095 1.092 1.096 1.095
<C2–C1–C4
<C1–C2–C3
<C2–C3–C4
<C1–C4–C3
<C1–C2–H
89.9
83.9
95.5
90.7
88.5
91.5
92.7
87.7
92.4
87.3
90.5
87.3
91.4
94.8
86.9
87.0
89.8
82.6
96.7
91.0
91.0
84.9
99.2
84.9
114.1 133.6 116.0 133.2 114.1 128.2b
a Bond lengths and angles are in angstroms and degrees, respectively.
b The hydrogen atom is bent 43Њ out of the plane of the ring.
the pKa of the latter compound is 19.4,43 we estimate that the
pKa of benzocyclobutenone is 27. In qualitative accord with
this prediction, we found that 1 can be deprotonated with
lithium tetramethylpiperidide and trapped in situ by chloro-
trimethylsilane at Ϫ78 ЊC under dilute conditions [eqn. (15)].
Comparable structural changes occur in the four-membered
ring of cyclobutanone (7) upon anion formation. The bonds
emanating from the anionic center (C2) contract and are
strengthened, while the C3–C4, C1–C4, and C᎐O bonds expand
᎐
and are weakened, although not to as large an extent; the C1–
C2–C3 angle opens up as well. In contrast, deprotonation of
cyclobutenone leads to a C2 anion in which the C2–C3 bond
contracts by 0.101 Å, the C3–C4 expands by 0.064 Å, and the
C2 and C4 hydrogens lie on the opposite side and 43Њ out of the
C2–C3–C4 plane. Smaller changes are found in the C1–C2 and
The worked up reaction mixture contained starting material,
trapped product (9) and dimer 2 in an 11:8:1 ratio.44 This result
does not reflect optimized experimental conditions, but instead
illustrates that benzocyclobutenone enolate can be trapped in
solution at low temperatures (Ϫ78 ЊC) contrary to previous
efforts carried out at higher temperatures (≥0 ЊC).6,7 Moreover,
this method offers a possible alternative synthesis for α-substi-
tuted benzocyclobutenones45 which are of potential interest in
the synthesis of anticancer agents and macromolecules.46,47
With our experimental knowledge of benzocyclobutenone
enolate, we turned to theory to elucidate the pertinent factors
contributing to its stability. Natural bond orbital (NBO)23,24
and Bader25 analyses were carried out on MP2/6-31ϩG(d)
optimized structures of 1, cyclobutanone (7), cyclobutenone (8)
and their corresponding enolates. These formalisms furnish
atomic charges as well as descriptive properties of the bonds
such as hybridization, ellipticity, and electron density at the
critical point (ρc) which was used to calculate bond orders.48,49
Initial examination of the structural differences of 1 and 1a
(Table 4) provides a platform for understanding the NBO and
Bader results. Upon deprotonation of 1 (at C2), the remaining
hydrogen moves into the plane of the bicyclic system allowing
the resulting charge to be aligned with the π system of the
carbonyl group and the benzene ring. To accommodate these
changes, the C1–C2–C3 angle enlarges by 7.6Њ and the C1–C4,
C᎐O bond lengths (and in comparison to 1a and 7a) such that
᎐
8a clearly resembles an allylic anion rather than an enolate.
Consistent with this view, the HOMO of cyclobutenone enolate
is that of an allyl anion and not an enolate.
Fig. 1 summarizes the C1–C2 bond in 1a, 7a, and 8a, which is
the most critical one in assessing the cyclobutadiene character
in 1a and 8a. In the former species there clearly is double bond
character between C1 and C2 in that the changes upon depro-
tonation of its conjugate acid are virtually the same as in
cyclobutanone enolate (i.e., the bond lengths decrease by 0.13–
0.14 Å, the bond orders increase by 0.41–0.44, and the elliptici-
ties increase by 0.29–0.37). Cyclobutenone enolate, on the other
hand, is anomalous. Deprotonation of its conjugate acid leads
to relatively little change in the C1–C2 bond.
Additional insight can be obtained by examining the NBO
and Bader charges on the individual atoms (Fig. 2 and Table 6).
The absolute values are quite different for the two methods but
the relative electron populations are in good accord and provide
a coherent picture. Deprotonation of cyclobutanone leads to a
build up of charge at the acidic site and delocalization on to the
more electronegative oxygen atom. The carbonyl carbon also is
less electron deficient in the electron rich enolate. Similar results
are seen for 1a, but less charge is delocalized on to the carbonyl
group because it is further spread out over the benzene ring.
Perhaps the most unexpected result is the charge at C5 and C7.
None of the canonical resonance structures support a build up
of electron density at these positions and thus changes in the
σ-framework contribute to the overall charge distribution.
Once again cyclobutenone enolate is quite different from
benzocyclobutenone and cyclobutanone enolates. In this case
C3–C4 and C᎐O bonds elongate by 0.072, 0.025 and 0.032 Å.50
᎐
The C1–C4 bond is less restricted and consequently increases
more, this also helps alleviate the antiaromatic character in the
system. These changes are further reflected in the NBO hybrid-
ization which shows diminished s-character in the bonds (Table
J. Chem. Soc., Perkin Trans. 2, 1999, 2389–2396
2393

View Article Online
Table 5 Natural bond orbital hybridizations and Bader’s topological parameters (ρc, ∇2pc, and ε) for 1, 7, 8 and their corresponding enolates
Baders topological parameters
Cmpd
Bond
NBO s-character
ρc
∇2pc
ε
BO
1 (1a)
C᎐O
33.7–41.6 (33.8–40.0)
32.4–21.8 (38.7–32.7)
24.7–31.0 (30.6–33.2)
30.1–30.0 (27.7–29.0)
34.2–30.1 (27.6–29.6)
39.8–33.6 (41.3–35.4)
34.8–35.0 (34.3–3.43)
34.9–35.0 (36.1–36.2)
35.0–34.6 (34.9–34.6)
38.7–34.2 (38.9–35.1)
26.7–100 (36.5–100)
33.2–41.9 (33.4–38.4)
33.6–23.1 (38.7–35.4)
24.6–24.2 (29.6–27.4)
24.2–24.6 (23.4–24.3)
33.6–23.1 (28.0–25.5)
27.1–100 (34.9–100)
33.7–41.7 (32.2–42.1)
33.4–22.1 (34.1–28.9)
24.2–29.6 (33.9–34.2)
35.4–34.3 (34.3–34.2)
34.2–28.7 (34.1–29.5)
26.8–100 (35.7–100)
0.397 (0.371)
0.235 (0.294)
0.244 (0.266)
0.315 (0.300)
0.261 (0.225)
0.308 (0.318)
0.306 (0.286)
0.302 (0.310)
0.305 (0.289)
0.310 (0.308)
0.267 (0.260)
0.396 (0.353)
0.252 (0.308)
0.234 (0.245)
0.234 (0.229)
0.252 (0.236)
0.267 (0.260)
0.397 (0.380)
0.238 (0.250)
0.248 (0.291)
0.333 (0.291)
0.266 (0.250)
0.268 (0.258)
0.273 (Ϫ0.347)
Ϫ0.518 (Ϫ0.735)
Ϫ0.548 (Ϫ0.612)
Ϫ0.853 (Ϫ0.793)
Ϫ0.617 (Ϫ0.463)
Ϫ0.840 (Ϫ0.864)
Ϫ0.822 (Ϫ0.727)
Ϫ0.806 (Ϫ0.825)
Ϫ0.815 (Ϫ0.741)
Ϫ0.846 (Ϫ0.820)
Ϫ0.904 (Ϫ0.888)
0.295 (Ϫ0.224)
Ϫ0.593 (Ϫ0.788)
Ϫ0.509 (Ϫ0.536)
Ϫ0.509 (Ϫ0.487)
Ϫ0.593 (Ϫ0.514)
Ϫ0.908 (Ϫ0.848)
0.273 (0.176)
0.082 (0.073)
0.040 (0.333)
0.007 (0.166)
0.197 (0.122)
0.090 (0.050)
0.172 (0.254)
0.210 (0.177)
0.194 (0.275)
0.207 (0.225)
0.190 (0.225)
0.014 (0.054)
0.066 (0.054)
0.046 (0.420)
0.009 (0.098)
0.009 (0.036)
0.046 (0.067)
0.017 (0.061)
0.078 (0.041)
0.030 (0.124)
0.170 (0.233)
0.327 (0.233)
0.112 (0.124)
0.012 (0.053)
2.55 (2.16)
0.90 (1.31)
0.95 (1.09)
1.50 (1.36)
1.06 (0.84)
1.44 (1.53)
1.42 (1.25)
1.38 (1.45)
1.41 (1.27)
1.45 (1.44)
1.10 (1.05)
2.53 (1.92)
1.00 (1.44)
0.89 (0.96)
0.89 (0.86)
1.00 (0.90)
1.10 (1.05)
2.55 (2.29)
0.91 (0.99)
0.97 (1.29)
1.69 (1.29)
1.09 (0.99)
1.11 (1.04)
᎐
C1–C2
C2–C3
C3–C4
C1–C4
C4–C5
C5–C6
C6–C7
C7–C8
C3–C8
C2–H
7(7a)
8(8a)
C᎐O
᎐
C1–C2
C2–C3
C3–C4
C1–C4
C2–H
C᎐O
᎐
C1–C2
C2–C3
C3–C4
C1–C4
C2–H
Ϫ0.528 (Ϫ0.552)
Ϫ0.561 (Ϫ0.722)
Ϫ0.923 (Ϫ0.772)
Ϫ0.642 (Ϫ0.552)
Ϫ0.909 (Ϫ0.831)
Table 6 Natural population analysis and Bader charges (parenthetical
values) for 1, 7, 8 and their corresponding enolates
Atoma
1
1a
7
7a
8
8a
O
Ϫ0.49
Ϫ0.67
Ϫ0.50
Ϫ0.78
Ϫ0.50
Ϫ0.61
(Ϫ1.14) (Ϫ1.26) (Ϫ1.15) (Ϫ1.32) (Ϫ1.15) (Ϫ1.24)
C1
C2
C3
C4
C5
C6
C7
C8
0.56
(1.08)
Ϫ0.02
0.44
(0.97)
Ϫ0.31
0.57
(1.06)
Ϫ0.04
0.37
(0.88)
Ϫ0.36
0.52
(1.08)
Ϫ0.06
(0.06) (Ϫ0.33)
0.10 0.00
(0.02) (Ϫ0.11)
Ϫ0.06 Ϫ0.45
0.52
(1.00)
Ϫ0.45
(0.06) (Ϫ0.23)
Ϫ0.02 Ϫ0.03
(Ϫ0.04) (Ϫ0.03)
Ϫ0.15 Ϫ0.07
(Ϫ0.07) (Ϫ0.09)
0.05 Ϫ0.08
(0.04) (Ϫ0.07)
0.02 Ϫ0.09
(0.02) (Ϫ0.10)
0.02 Ϫ0.06
(0.02) (Ϫ0.08)
0.03 Ϫ0.13
(0.03) (Ϫ0.11)
(0.03) (Ϫ0.30)
0.02 Ϫ0.10
(0.04) (Ϫ0.12)
Ϫ0.01 Ϫ0.11
(0.03) (Ϫ0.13) (Ϫ0.01) (Ϫ0.33)
Fig. 2 Charge redistribution upon deprotonation of benzocyclo-
butenone (1), cyclobutanone (7), and cyclobutenone (8). Both the NBO
and Bader charges (parenthetical values) represent the difference
between the enolate and neutral and have the contributions from the
hydrogens added into the carbon atoms to which they are attached.
MP2/6-31ϩG(d) structures and wave functions were employed.
a Charges on hydrogen atoms have been summed in to the carbons to
which they are attached.
only 10–15% of the charge is delocalized on to the carbonyl
group as opposed to 45–50% in 7a and 23–30% in 1a. The two α
carbons (C2 and C4) are equivalent in the anion and bear 70–
80% of the charge, exactly what one would expect for an allylic
ion.
Benzocyclobutenone is a stronger acid than cyclobutanone
by 6.9 5 kcal molϪ1 and this difference is well reproduced by
ab initio calculations at the MP2/6-31ϩG(d) level [eqn. (16),
Table 3]. On the other hand, 1-phenylpropan-2-one, an acyclic
model, is more acidic than 1 by 10.1 3 kcal molϪ1 [eqn. (17)].
This latter difference is not surprising given that benzo-
cyclobutene (cyclic) is less acidic than ethylbenzene (acyclic)
by 6 3 kcal molϪ1, and suggests that there is little, if any,
2394
J. Chem. Soc., Perkin Trans. 2, 1999, 2389–2396

View Article Online
destabilization in 1a as a result of cyclobutadiene character in
the anion. This view is consistent with the structure of 1a, its
charge distribution, and Bader parameters such as the bond
critical point, bond order, and ellipticity. It also is interesting to
note that while resonance delocalization accounts for about two
thirds of the stabilization in acetaldehyde enolate (i.e., the
rotational barrier is 36 kcal molϪ1 at the MP2/6-31ϩG(d)//HF/
6-31ϩG(d) level52 but ∆∆HЊacid(ethane Ϫ acetaldehyde) = 54
kcal molϪ1) the contribution in 11a, and presumably 1a, is much
larger (i.e., the rotational barrier about the PhCHϪ–CHO bond
is 26.7 kcal molϪ1 compared to ∆∆HЊacid(ethylbenzene Ϫ phenyl-
acetaldehyde) = 31 kcal molϪ1).53
American Chemical Society, the Minnesota Supercomputer
Institute, and the University of Minnesota – IBM Shared
Research Project is gratefully acknowledged.
References
1 B. M. Trost and I. Fleming, Comprehensive Organic Synthesis:
Selectivity, Strategy and Efficiency in Modern Organic Chemistry,
Pergamon Press, New York, 1991.
2 B. A. Lorsbach and M. J. Kurth, Chem. Rev., 1999, 99, 5149.
3 M. P. Cava and K. Muth, J. Am. Chem. Soc., 1960, 82, 652.
4 Reported deuterium incorporation into benzocyclobutenone via 1a
was later retracted (see ref. 5). H. Hart and R. W. Fish, J. Am. Chem.
Soc., 1960, 82, 749.
The conjugate base of cyclobutenone (8a) is the simplest
5 H. Hart, J. A. Hartlage, R. W. Fish and R. R. Rafos, J. Org. Chem.,
1966, 31, 2244.
6 D. J. Bertelli and P. Crews, J. Am. Chem. Soc., 1968, 90, 3889.
7 T. Matsumoto, T. Hamura, Y. Kuriyama and K. Suzuki,
Tetrahedron Lett., 1997, 38, 8985.
8 M. E. Jones, S. R. Kass, J. Filley, R. M. Barkley and G. B. Ellison,
J. Am. Chem. Soc., 1985, 107, 109.
9 J. E. Bartmess, R. L. Hays and G. Caldwell, J. Am. Chem. Soc.,
1981, 103, 1338.
10 M. D. Brickhouse and R. R. Squires, J. Am. Chem. Soc., 1988, 110,
2706.
enolate which can have cyclobutadiene character.54 In contrast
to benzoannelation of cyclobutanone enolate, introduction of a
double bond is highly disfavored [eqn. (18) and (19)]. At the
11 M. D. Brickhouse and R. R. Squires, J. Phys. Org. Chem., 1989, 2,
389.
12 R. N. Hayes, R. P. Grese and M. L. Gross, J. Am. Chem. Soc., 1989,
111, 8336.
13 I. L. Freriks, L. J. de Koning and N. M. M. Nibbering, J. Am. Chem.
Soc., 1991, 113, 9119.
14 I. L. Freriks, L. J. de Koning and N. M. M. Nibbering, J. Phys. Org.
Chem., 1992, 5, 776.
15 Trifluoroacetyl chloride has been proposed as a reagent which gives
a higher percentage of O-alkylation. However, the percentage of
oxygen reactivity is based on detection of chloride which can,
in principle, arise from either C- or O-attack. B. D. Wladkowski,
J. L. Wilbur, M. Zhong and J. I. Brauman, J. Am. Chem. Soc., 1993,
115, 8833.
16 M. Zhong and J. I. Brauman, J. Am. Chem. Soc., 1996, 118, 636.
17 O. Abou-Teim, M. C. Goodland and J. F. W. McOmie, J. Chem.
Soc., Perkin Trans. 1, 1983, 2659.
18 S. Takano, K. Inomata, K. Samizu, S. Tomita and M. Yanase,
Chem. Lett., 1989, 1283.
19 M. Krumpolc and J. Rocek, Org. Synth., 1990, Coll. Vol. VII, 114.
20 T. C. L. Wang, T. L. Ricca and A. G. Marshall, Anal. Chem., 1986,
58, 2935.
21 M. J. Frisch, G. W. Trucks, H. B. Schlegel, P. M. W. Gill, B. G.
Johnson, M. A. Robb, J. R. Cheeseman, T. Keith, G. A. Peterson,
J. A. Montgomery, K. Raghavachari, M. A. Al-Laham, V. G.
Zakrewski, J. V. Ortiz, J. B. Foresman, J. Cioslowski, B. B. Stefanov,
A. Nanayakkara, M. Challacombe, C. Y. Peng, R. Y. Ayala, W.
Chen, M. W. Wong, J. L. Andres, E. S. Replogle, R. Gomperts,
R. L. Martin, D. J. Fox, J. S. Binkley, D. J. Defrees, J. Baker,
J. P. Stewart, M. Head-Gordon, C. Gonzalez and J. A. Pople,
GAUSSIAN94 Revisions A, B, C; Gaussian, Inc., Pittsburgh, PA,
1995.
22 J. A. Pople, A. P. Scott, M. W. Wong and L. Radom, Isr. J. Chem.,
1993, 33, 345.
23 A. E. Reed, L. A. Curtiss and F. Weinhold, Chem. Rev., 1988, 88,
899.
MP2/6-31ϩG(d) level of theory, benzocyclobutenone is com-
puted to be 22 kcal molϪ1 more acidic than cyclobutenone and
the corresponding radical also is substantially more stabilized
(i.e., ∆EA(1r Ϫ 8r) = 0.9 eV). The former difference is consider-
ably smaller than that between benzocyclopropene and cyclo-
propene (34.5 kcal molϪ1) at the same level of theory.55 This
represents a leveling effect and is not surprising since cyclobu-
tenone is much more acidic than cyclopropene (∆∆HЊacid = 38.7
kcal molϪ1). As for the relative lack of stability of 8a, this is
due to its inability to distribute the charge on to oxygen via
resonance structures b and c and the resulting distortion to an
allylic-type ion (resonance structure d).
Conclusion
Benzocyclobutenone enolate was generated in the gas phase by
deprotonation of 1. The measured thermochemistry (∆HЊacid
=
360.3 2.1 kcal molϪ1, EA = 1.90 0.10 eV and BDE = 90.5
3.1 kcal molϪ1) suggests that there is little, if any, antiaromatic
destabilization of 1a as a result of a resonance structure with
cyclobutadiene character. In fact, an analysis of the ion’s struc-
ture using NBO and Bader’s parameters reveals that 1a’s over-
all stability is a result of its ability to distort and alleviate the
4π electron interaction, and to distribute the charge over the
benzene ring and the electronegative oxygen atom. Anion 1a
displays carbon reactivity with perfluorinated reagents under
thermal conditions, but can be induced to react through oxygen
upon excitation. Despite previous reported failures, benzo-
cyclobutenone enolate was trapped in situ with an electrophile
in the liquid phase.
24 F. Weinhold and J. E. Carpenter, in The Structure of Small
Molecules and Ions, Plenum, New York, 1988.
25 R. F. W. Bader, in Atoms in Molecules: A Quantum Theory, Oxford
University Press, Oxford, 1990.
26 All thermodynamic data, unless otherwise noted, comes from
S. G. Lias, J. E. Bartmess, J. F. Liebmann, J. L. Holmes, R. D. Levin
and W. G. Mallard, J. Phys. Chem. Ref. Data, 1988, 17, Suppl. 1 [or
the slightly updated form available on a personal computer, NIST
Negative Ion Energetics Database (Version 3.00, 1993); NIST
Standard Reference Database 19B] or ref. 27. Note, 4.184 kJ = 1 kcal
and 1 eV = 23.06 kcal and 96.48 kJ.
27 J. E. Bartmess, Negative Ion Energetics Data in Secondary Negative
Ion Energetic Data, November 1998, National Institute of
Standards and Technology, Gaithersburg, MD 20899 (http://
webbook.nist.gov).
28 J. E. Bartmess and R. M. Georgiadis, Vacuum, 1983, 33, 149.
29 K. J. Miller, J. Am. Chem. Soc., 1990, 112, 8533.
30 The error in kϪ1 is related to the magnitude of the rate and the gauge
reading for the neutral pressure. A minimal pressure of the ketone
Acknowledgements
Support from the National Science Foundation, the donors of
the Petroleum Research Foundation, as administered by the
J. Chem. Soc., Perkin Trans. 2, 1999, 2389–2396
2395

View Article Online
was required to slow the reaction and enable it to be followed over
two half-lives. The resulting pressure (ca. 1–1.5 × 10Ϫ8 Torr) was not
much greater than the background pressure (0.3 × 10Ϫ8 Torr), so the
error in the ion gauge reading was larger than usual.
44 Ketone 9 was found to be stable to the acidic workup conditions.
45 B. L. Chenard, C. Slapak, D. K. Anderson and J. S. Swenton,
J. Chem. Soc., Chem. Commun.. 1981, 179.
46 D. N. Hickman, K. J. Hodgetts, P. S. Mackman, T. W. Wallace and
J. M. Wardleworth, Tetrahedron 1996, 52, 2235.
47 Z. Y. Wang, L. Kuang, X. S. Meng and J. P. Gao, Macromolecules,
1998, 31, 2935.
31 G. H. Herzberg, Molecular Spectra and Molecular Structure:
Infrared and Raman Spectra of Polyatomic Molecules, D. Van
Nostrand, Princeton, NJ, 1945; Vol. 2.
32 J. W. Gauthier, T. R. Trautman and D. B. Jacobson, Anal. Chim.
Acta., 1991, 246, 211.
33 J. J. Grabowski, J. M. Van Doren, C. H. DePuy and V. M. Bierbaum,
J. Chem. Phys., 1984, 80, 575.
34 Aside from the ions at m/z 247 (adduct ϪHF) and m/z 147
(perfluoroacetone enolate), two additional products were detected at
m/z 169 and 219 (6% and 8%, respectively, relative to the base peak
at m/z 247). The m/z 219 species could arise from decarbonylation of
the adduct ϪHF ion and subsequent loss of CF2 would give the
anion at m/z 169.
35 A. G. Marshall and D. C. Roe, J. Chem. Phys., 1980, 73, 1581.
36 T. B. McMahon and P. Kebarle, Can. J. Chem. 1985, 63, 3160.
37 T. B. McMahon, T. Heinis, G. Nicol, J. K. Hovey and P. Kebarle,
J. Am. Chem. Soc., 1988, 110, 7591.
38 We have used the electron affinity of the enolate’s corresponding
radical as a measure of the ion’s HOMO energy. The computed
MP2 HOMO energy gives a similar value.
39 G. J. Heiszwolf and H. Kloosterziel, Recl. Trav. Chim. Pays-Bas,
1970, 89, 1153.
40 G. K. King, M. M. Maricq, V. M. Bierbaum and C. H. DePuy,
J. Am. Chem. Soc. 1981, 103, 7133.
48 Bond order is calculated as a function of the electron density at
the critical point, ρc. Bond order = exp{6.458* (ρc Ϫ 0.2520)}. See:
K. B. Wiberg, R. F. Bader and C, D. H. Lau, J. Am. Chem. Soc.,
1987, 109, 985.
49 Ellipticity is a measure of the anisotropy of the charge distribution.
A
larger number suggests more p character. For example,
ethane = 0.0, benzene = 0.23 and ethylene = 0.45.
50 The extended C1–C4 bond might suggest a ring-opened α-ketene
phenide isomer; however, this structure is 23.6 kcal molϪ1 higher in
energy than 1a at the MP2/6-31ϩG(d)//HF/6-31ϩG(d) level of
theory.
51 D. Kovacek, Z. B. Maksic and I. Novak, J. Phys. Chem., 1997, 107,
1147.
52 M. C. Baschky, PhD Thesis, Unversity of Minnesota, 1994.
53 We estimate that 11 is 1 kcal molϪ1 more acidic than phenyl-
acetaldehyde.
54 M. Hare, PhD Thesis, University of Minnesota, 1998.
55 L. Moore, R. Lubinski, M. C. Baschky, G. D. Dahlke, M. Hare,
T. Arrowood, Z. Glasovac, M. Eckert-Maksic and S. R. Kass,
J. Org. Chem. 1997, 62, 7390.
41 J. J. Grabowski and L. Zhang, J. Am. Chem. Soc. 1989, 111, 1193.
42 R. W. Taft and F. G. Bordwell, Acc. Chem. Res., 1988, 21, 463.
43 F. G. Bordwell, J. E. Bartmess, G. E. Drucker, Z. Margolin and
W. S. Matthews, J. Am. Chem. Soc., 1975, 97, 3226.
Paper 9/05868K
2396
J. Chem. Soc., Perkin Trans. 2, 1999, 2389–2396