Computational assessment of the electronic structure of 1-azacyclohexa-2,3,5-triene (3δ2-1H-pyridine) and its benzo derivative (3δ2-1H-quinoline) as well as generation and interception of 1-methyl-3δ2-1H-quinoline

Article abstract of DOI:10.1002/chem.200305000

Treatment of a solution of 3-bromo-1-methyl-1,2-dihydroquinoline (9) and [18]crown-6 in furan or styrene with KOtBu followed by hydrolysis afforded a mixture of 1-methyl-1,2-dihydroquinoline (10) and 1-methyl-2-quinolone (11). If the reaction was performed in [D₈]THF and the mixture was immediately analysed by NMR spectroscopy, 2-tert-butoxy-1-methyl-1,2-dihydroquinoline (17) was shown to be the precursor of 10 and 11. The structure of 17 is evidence for the title cycloallene 7, which arises from 9 by β elimination of hydrogen bromide and is trapped by KOtBu to give 17 so fast that cyclo-additions of 7 with furan or styrene cannot compete. Since this reactivity is unusual compared to the large majority of the known six-membered cyclic allenes, we performed quantum-chemical calculations on 8, which is the parent compound of 7, and the corresponding isopyridine 6 to assess the electronic nature of these species. The ground state of 6 was no longer an allene (6a) but the zwitterion 6b. In the case of 8, the allene structure 8a is more stable than the zwitterionic form 8b by only ≈1 kcal mol^-1. These results suggest a high reactivity of 6 and 8 towards nucleophiles and explains the behaviour of 7. In addition to the ground states, the low-lying excited states of 6 and 8 were considered, which are represented by the diradicals 6c and 8c and, as singlets, lie above 6b and 8a by 19.1 - 24. 8 and 14.4-17.7 kcal mol^-1, respectively.

Full text of DOI:10.1002/chem.200305000

FULL PAPER
Computational Assessment of the Electronic Structure of 1-Azacyclohexa-
2
2
2
,3,5-triene (3d -1H-Pyridine)and Its Benzo Derivative (3 d -1H-Quinoline)
2
as well as Generation and Interception of 1-Methyl-3d -1H-quinoline**
Jan C. Schˆneboom, Stefan Groetsch, Manfred Christl,* and Bernd Engels*[a]
Dedicated to Professor Wolfgang Malisch on the occasion of his 60th birthday
Abstract: Treatment of a solution of
-bromo-1-methyl-1,2-dihydroquinoline
9) and [18]crown-6 in furan or styrene
with KOtBu followed by hydrolysis
afforded a mixture of 1-methyl-1,2-dihy-
droquinoline (10) and 1-methyl-2-qui-
nolone (11). If the reaction was per-
KOtBu to give 17 so fast that cyclo-
additions of 7 with furan or styrene
cannot compete. Since this reactivity is
unusual compared to the large majority
of the known six-membered cyclic al-
lenes, we performed quantum-chemical
calculations on 8, which is the parent
compound of 7, and the corresponding
isopyridine 6 to assess the electronic
nature of these species. The ground state
of 6 was no longer an allene (6a) but the
zwitterion 6b. In the case of 8, the allene
structure 8a is more stable than the
3
(
zwitterionic
form
8b
by
only
À1
ꢀ1 kcalmol . These results suggest a
high reactivity of 6 and 8 towards
nucleophiles and explains the behaviour
of 7. In addition to the ground states, the
low-lying excited states of 6 and 8 were
considered, which are represented by
the diradicals 6c and 8c and, as singlets,
lie above 6b and 8a by 19.1 ± 24.8 and
formed in [D ]THF and the mixture was
8
immediately analysed by NMR spectros-
copy, 2-tert-butoxy-1-methyl-1,2-dihy-
droquinoline (17) was shown to be the
precursor of 10 and 11. The structure of
Keywords: allenes ¥ eliminations ¥
À1
17 is evidence for the title cycloallene 7,
14.4 ± 17.7 kcalmol , respectively.
quantum-chemical calculations
¥
which arises from 9 by b elimination of
hydrogen bromide and is trapped by
strained molecules ¥ zwitterions
Introduction
Experimental studies of six-membered cyclic allenes[1, 2] that
contain a nitrogen atom have been performed for the
symmetrical isodihydropyridines 1,[ the unsymmetrical iso-
3]
[4]
dihydropyridine 2, its borane complex 3, the isopyridines 5
and 6[ and the cephalosporine derivatives 4.
5]
[6]
[1]
Akin to cyclohexa-1,2-diene, the 1-azacyclohexa-3,4-di-
enes 1 can be trapped by activated olefins.[ In contrast,
attempts to generate 1-methyl-1-azacyclohexa-2,3-diene (2)
did not furnish products that proved the existence of 2.
However, the intermediacy of its borane complex 3 has been
secured by the isolation of cycloadducts of 3 with furan and
3]
styrene.[4] The liberation of the cephalosporins 4 proceeds
under astoundingly mild conditions and their interception,
even with nonactivated olefins and acetylenes, takes place
[
a] Prof. Dr. M. Christl, Prof. Dr. B. Engels,
Dipl.-Chem. J. C. Schˆneboom, Dr. S. Groetsch
Institut f¸r Organische Chemie, Universit‰t W¸rzburg
Am Hubland, 97074 W¸rzburg (Germany)
Fax : (‡49)931-8884606
[6]
with high efficiency.
Isopyridines of types 5 and 6 have been postulated by
Shevlin et al.[ as intermediates in reaction sequences that
start with the addition of carbon atoms onto the respective
pyrroles. The structure of the products as well as preliminary
quantum-chemical calculations support the dipolar nature of
5]
E-mail: christl@chemie.uni-wuerzburg.de
bernd@chemie.uni-wuerzburg.de
[
**] Cycloallenes; Part 18. For Part 17 see: ref. [9].
5
and 6, namely, the zwitterions 5b and 6b are more likely to
Supporting information for this article is available on the WWW under
http://www.chemeurj.org or from the author: Cartesian coordinates
for the structures 6a ±c and 8a ±c, obtained by various theoretical
approaches, are provided.
[5]
be the ground state than the allenes 5a and 6a. More
recently, Yavari et al.[ have theoretically studied the zwitter-
ionic form 6b and the triplet diradical 6c, but did not
7]
3
Chem. Eur. J. 2003, 9, 4641 ± 4649
DOI: 10.1002/chem.200305000
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4641

M. Christl, B. Engels et al.
FULL PAPER
compound.[11, 12] However, the reaction of 9, dissolved in
furan or styrene, with KOtBu in the presence of [18]crown-6
did not furnish any product that arose by participation of
furan or styrene. After the usual workup, the result was a
mixture of 1-methyl-1,2-dihydroquinoline (10) and 1-methyl-
2
-quinolone (11) as the major components in a 1:2 ratio
(
Scheme 1). A similar mixture was formed when the reaction
was carried out in THF as the solvent in the absence of furan
and styrene. In this case, 9 was consumed at an appreciable
rate, even in the absence of [18]crown-6. To characterise 10
Scheme 1.
consider more closely the allene form 6a and the singlet
1
2
diradical 6c. The possible intervention of derivatives of 3d -
H-quinoline (8) in the thermolysis of 1-(2-amino-5-chloro-
[13, 14]
1
and 11, we prepared these compounds by known routes
and completely assigned their H and C NMR spectra.
When 12 was generated in the absence of activated olefins,
the acetal 13 resulted with high efficiency. Obviously, tert-
butoxide added to the allene terminus of 12 that carries the
oxygen atom (Scheme 2).[ The liberation of 3d -pyran (14),
1
13
phenyl)propargyl carboxylates has recently been advanced by
Frey et al.[
8]
2
Herein, we report the generation of 1-methyl-3d -1H-
quinoline (7). Since the consecutive products of 7 were
surprising, we carried out a computational study of the parent
heterocycle 8. To assess the effect of the benzo group of 8, we
included the isopyridine 6 in our calculations, the principal
goal of which was to answer the question as to the ground
state of these species, whether they are allenes (6a, 8a) or
zwitterions (6b, 8b). We have recently shown that, to
understand the nature of systems with a strained allene
moiety, the calculation of the respective global energy
minimum alone is not sufficient and the electronic structures
of the ground and the first excited state for the planar
10]
2
[9]
geometry have to be taken into account as well. Thus, a
second objective was the determination of the energetic
distance of 6a and 8a as well 6b and 8b from the singlet
diradical states 6c and 8c, respectively.
Regarding experiment and theory, the present work
expands our previous one, in which we investigated the oxa
and carba analogues of 5 ± 8 (O or CH instead of NR, see
Scheme 2.
2
compounds 14 and 18 in Figure 4).[ In this context, the
present work also aims at a detailed understanding of the
factors that govern the electronic and geometric structures in
9]
the monocycle related to 12, from 3-bromo-4H-pyran by
KOtBu yielded 4-tert-butoxy-4H-pyran (15) as the only
identified product, regardless of whether activated olefins
were present or not (Scheme 2).[ Again, the nucleophile
attacked an allene terminus.
6
and 8 as compared to 14 and 18. Based on our model, we are
able to present a complete picture of the differences that arise
from benzannulation and heteroatom substitution.
9]
These findings made us assume that the desired cyclic
allene 7 had formed from 9 by b elimination of hydrogen
bromide and was rapidly trapped by KOtBu to give the N,O-
acetal 17 eventually (Scheme 3). In that case, furan and
styrene would have been inferior compared to the nucleophile
in competition for 7. Grignon-Dubois and Meola[ have
mentioned 17, but did not characterise it. They gave a
plausible postulation of the formation of 17 on treatment of
N-methylquinolinium iodide (16) with KOtBu and took
advantage of its sensitivity towards molecular oxygen to
prepare the quinolone 11. We repeated this experiment with
THF as the solvent, filtered the complete mixture through
Results and Discussion
2
Generation and interception of 1-methyl-3d -1H-quinoline
7): On treatment of 3-bromo-2H-chromene with potassium
14]
(
tert-butoxide (KOtBu) in the presence of activated olefins, we
2
recently obtained cycloadducts of 3d -chromene (2,3-didehy-
dro-2H-1-benzopyran, 12),[ the oxa analogue of 7 and 8.
This encouraged us to test whether 7 can be generated by the
same procedure, particularly because 3-bromo-1-methyl-1,2-
dihydroquinoline (9), the starting material, is a known
10]
4642
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 4641 ± 4649

Electronic Structure of Cycloallenes
4641±4649
is already approximated in 1-oxacyclohexa-2,3-diene, the
dihydro derivative of 14, which clearly should have a chiral
ground state, but is attacked by KOtBu[ and enolates
18]
[19]
either at the central or a terminal allene atom. By analogy,
these findings support the ground state of 7 to be rather polar,
and this is indeed corroborated by the calculations presented
below.
We recently reported that cycloadducts of 2 with activated
olefins could not be obtained. However, if the lone pair of the
nitrogen atom of 2 was withdrawn from service, as in the
borane complex 3, cycloadditions became feasible.[ This
recipe could not be applied to 7, since it is too weak a base to
bind borane.
4]
Scheme 3.
Computational details: For a reliable description of systems
which possess diradical character, a multireference treatment
is essential in most cases. Since the planar diradical species
appeared to be important for the understanding of the
chemistry of cyclic allenes,[ MR-CI and CASPT2 single-
point energy calculations were employed in the present work.
The geometric parameters of the stationary points were
optimised with analytical gradients of the complete active
space SCF (CASSCF) method and of the density functional
theory (DFT). To obtain the minima of a given multiplicity,
we performed optimisations without symmetry constraints.
The geometric parameters of the planar singlet species were
basic Al O (activity IV) and obtained 10 and 11 in a 1.0:1.3
2
3
ratio. When we performed the reaction in [D ]DMSO and
6
1
13
immediately thereafter recorded the H and C NMR spectra
of the mixture, they each exhibited one set of signals, which
match perfectly with the structure of 17. Moreover, the
9]
1
H NMR spectrum of the mixture measured immediately after
combining 9 and KOtBu in [D ]THF exhibited only one set of
8
signals, which we also ascribe to 17. In view of the different
1
solvents in which the H NMR spectra have been determined,
the chemical shifts of the products formed from 9 ([D ]THF)
8
and 16 ([D ]DMSO) are in excellent agreement, and hence
6
the products are the same, namely 17.
optimised with C symmetry constraints.
s
The transformation of 17 to 10 and 11 apparently occurs
when a hydrolytic workup is performed. Under these con-
ditions, the tert-butoxy group is probably replaced by a
hydroxy group and the resulting pseudo-base undergoes the
redox process to afford 10 and 11. Disproportionations of that
kind are well-known.[ The fact that the above product
mixtures contained invariably more 11 than 10 is readily
explained by the susceptibility of the latter to autoxidation,
which is the key of the efficient preparation of 2-quinolones
and related compounds from quinolinium salts and KOtBu
and exposure of the resulting N,O-acetals to air.[
For the geometry optimisations that employed DFT
methods, we used the B88 or B3 exchange expressions[
in combination with the correlation functionals by Lee, Yang
and Parr (LYP).[ The unrestricted ansatz was used for all
DFT computations. The singlet diradical species were calcu-
lated and their structures optimised with broken-spin sym-
metry determinants. Although the DFT methodology em-
ployed in the present work is a single reference method, it has
proven to be capable of describing the geometries of diradical
systems similar to 6 and 8 in our previous work.[ Never-
theless, a careful validation of the used functional and basis set
against reliable multireference treatments is required from
case to case. In this work, MR-CI and CASPT2 results from 6
were used to check the validity of the present DFT approach.
The nature of the various stationary points was analysed by
calculation of the DFT harmonic frequencies. The vibrational
analysis at the DFT level was also utilised to determine the
thermal corrections necessary for obtaining the enthalpy
values. All DFT calculations were performed with the
Gaussian98 program package.[
20, 21, 22]
23]
15]
9]
14]
Thus, the formation of 10 and 11 on treatment of 9 with
KOtBu clearly supports the intermediacy of the cyclic allene 7.
However, cycloadditions of 7 onto activated olefins do not
proceed under these conditions since KOtBu traps 7 much
more efficiently. The high sensitivity toward this nucleophile
as well as the site of attack are evidence for a ground state of 7
that is strongly polar if not entirely zwitterionic (7b) in nature.
Six-membered cyclic allenes that definitely or presumably
have a chiral ground state are attacked by nucleophiles at the
central carbon atom of the allene moiety: cyclohexa-1,2-
24]
A comparison of different choices for the active space in the
CAS treatment showed that an adequate description in Cs
symmetry (compounds 6b and 6c) includes the three occu-
pied a'' (p) orbitals with the corresponding number of virtual
orbitals and the nonbonding a' (s) orbital mainly located at
the central allene carbon, thus defining an [8,7]-CASSCF
2
[1]
[3a]
diene, 1,1-dimethyl-3d -1H-naphthalene (1, R ˆ C H ),
6
5
1
-oxacyclohexa-3,4-diene[16] and 3d -1H-naphthalene.
2
[17]
However, if the ground state is zwitterionic, as calculations
[9]
demonstrate for 14 in solution, that is 14b, or if the
À1
zwitterionic state is only a few kcalmol less stable than the
chiral state, as in the case of 12 (energy difference of
method. The active space for the C -symmetrical species 6a
1
À1
[9]
3.2 kcalmol in solution), KOtBu interacts with an allene
was chosen as the three highest occupied and three lowest
unoccupied molecular orbitals (MOs) of the a irreducible
representation, leading to a [6,6] active space. The CASPT2
single-point energy calculations utilised [6,6]-CASSCF wave
functions as a reference. The CASSCF as well as the CASPT2
terminus exclusively (Scheme 2). This is readily explained
since these steps can be considered as an addition to a quasi-
carbonyl group (12b) and as conjugated addition to an a,b-
unsaturated quasi-carbonyl compound (14b). Such a situation
Chem. Eur. J. 2003, 9, 4641 ± 4649
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4643

M. Christl, B. Engels et al.
FULL PAPER
calculations were performed with the MOLCAS program
package.[
which possesses no allene character. The deviations from
planarity are very small in the case of BLYP computations
(average of the above dihedral angles ˆ 1718) and even
negligible according to the B3LYP result. We assume that
the deviation from planarity obtained with the BLYP func-
tional is a symmetry-breaking effect, which is why this
geometry was not considered any further. At the CASPT2
level, single-point calculations favour the planar structure
25]
The MR-CI approach used in this study is based on an
individually selecting multireference CISD algorithm[ to
reduce the dimension of the CI eigenvalue problem. The
contribution of the configurational state functions neglected
26]
in the CI procedure was taken into account by the Buenker±
Peyerimhoff extrapolation scheme.[
27, 28]
The influence of
À1
higher excitations was estimated by the normalised form of
the usual Davidson correction.[ In the following, these
calculations are abbreviated as MR-CI ‡ Q. They were
performed with the DIESEL-CI program package[ and
utilised the orbitals obtained from a [6,6]-CASSCF calcula-
tion. Thermal corrections were obtained from DFT
over the allenic one by 1 ± 2 kcalmol . Since CASPT2
29]
calculations represent a more sophisticated approach than
the CASSCF method, the allenic structure can be ruled out as
energy minimum of 6 in the ground state. On the basis of this
finding, structure 6a seems to be an artefact and only
represents a minimum if dynamic correlation effects are
neglected. Obviously, in the case of the ground state of 6,
which has a closed-shell single-determinant nature, the DFT
methods perform better than CASSCF calculations, since they
offer a more balanced description of dynamic and non-
dynamic correlation effects. The results of the CASSCF and
DFT calculations are in much better agreement for the species
6b and 6c, where the larger [8,7] active space could be used.
In general, the geometries optimised with DFT methods gave
lower absolute single-point energies at the CASPT2 level, and
30]
(
UB3LYP/cc-pVDZ) frequency calculations.
For the CASPT2 and the MR-CI calculations, we employed
the cc-pVDZ basis set of Dunning.[ Geometry optimisations
were performed with the cc-pVDZ or 6-31G(d) basis sets.
While a multireference approach is necessary for an
accurate computation of the energetics, a qualitative descrip-
tion of the electronic situation can already be obtained on the
basis of less sophisticated treatments. This can be seen from
the orbitals resulting from the CASSCF and the DFT
calculations, which do not differ significantly. This shows that
the electron densities provided by the various treatments are
qualitatively equal. Consequently, the natural resonance
theory (NRT)[ analysis based on the DFT result was
employed to obtain qualitative insights into electron-density
distributions.
31]
we will focus on these structures in the following.
Thus, in accordance with previous studies,[
5, 7]
we find the
2
zwitterion 6b to be the equilibrium structure of 3d -1H-
pyridine (6) in the ground state. The striking structural feature
of 6b is the small magnitude of the angle C1-C2-C3 (ꢀ1118),
which originates from the electronic repulsion of the lone pair
in a s orbital at C2 and the neighbouring CÀC bonds. This
32]
2
Calculations of 3d -1H-pyridine (6): The optimised geome-
™chemically∫ intuitive representation of the electronic struc-
ture is corroborated by the analysis of the frontier orbitals
(Figure 2) and the strongly dominating configuration of the CI
expansion (coefficient 0.967) (Table 1): the electronic state is
tries of allene 6a, the zwitterion 6b, and the singlet diradical
1
6
c are depicted in Figure 1. The [6,6]-CASSCF calculation
predicts an unsymmetrical species 6a as global minimum. The
average of its dihedral angles C1-C2-C3-H3 and C3-C2-C1-
H1 (see Figure 1 for the numbering of atoms) is 1598 and the
angle at the central allene carbon atom is 1228, indicating a
highly strained allenic structure. In contrast, DFT calculations
result in a geometry much closer to the planar arrangement,
1
of A' symmetry, with the non-bonding s orbital 18a' being the
HOMO and three occupied a'' orbitals constituting a pos-
itively charged aromatic p system. This shows that the p
orbital of the nitrogen atom arranged perpendicularly to the
molecular plane is integrated into the ring p system.
Figure 1. Optimised CÀC and CÀN bond lengths [ä] and bond angles [8] computed for the isopyridine 6. From bottom to top, the values given were obtained
1
by CASSCF/cc-pVDZ, BLYP/cc-pVDZ and B3LYP/cc-pVDZ . [a] No stationary point on the A surface was located with B3LYP/cc-pVDZ; [b] [6,6]-
CASSCF; [c] [8,7]-CASSCF.
¹
2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 4641 ± 4649
4644

Electronic Structure of Cycloallenes
4641±4649
1
À1
At the MR-CI ‡ Q level, 6c is calculated to lie 23.9 kcalmol
(
1.04 eV) above 6b, while the CASPT2 calculation yields a
À1
relative energy of ‡19.1 kcalmol (0.83 eV). The corre-
3
sponding triplet diradical 6c is computed to be 27.3 (MR-
À1
CI ‡ Q) and 21.3 (CASPT2) kcalmol higher in energy than
[
7]
3
6
‡
b. Yavari et al. have reported a relative energy for 6c of
À1
37.9 kcalmol at the QCISD/6-31G(d)//HF/6-31G(d) level.
The optimised singlet and triplet diradical structures 6c are
virtually identical. Most notably, the angle C1-C2-C3 is
calculated to be ꢀ1278 and is thus significantly larger than
in the zwitterionic structure 6b (Figure 1). This larger value is
in agreement with the much smaller electronic repulsion in 6c
as compared to 6b, since the s orbital is only singly occupied.
The results of the DFT meth-
Figure 2. Frontier orbitals (FO) of different states of the isopyridine 6.
ods (B3LYP/BLYP) for the rel-
Table 1. Dominating configurations of the CI expansion for three stationary points of the isopyridine 6.
ative energies of the diradical
states are in remarkably good
agreement with the ab initio
Stationary point Symmetry, State Configuration
Coefficient
6
6
6
b
c
c
C
C
C
s
s
s
, 1A'
j (1a') ... (18a') ; (1a'') (2a'') (3a'') >
2
2
2
2
2
0.9673
À 0.1629
À 0.1235
0.1139
2
2
2
2
2
1
3
j (1a') ... (18a') ; (1a'') (2a'') (4a'') >
approaches: 6c: ‡23.0/24.8, 6c:
2
2
2
2
2
j (1a') ... (18a') ; (1a'') (3a'') (5a'') >
À1
‡
23.7/25.0 kcalmol . Table 2
also contains values for DDH2
98
2
2
2
1
1
1
1
j (1a') ... (18a') ; (1a'') (2a'') (3a'') (4a'') (5a'') >
, 1A''
2
2
1
2
2
2
1
j (1a') ... (17') (18a') ; (1a'') (2a'') (3a'') (4a'') >
0.9732
2
2
1
2
2
1
2
and DDG298 calculated at the
B3LYP level of theory. These
results indicate that thermal and
entropic effects cause only a
j (1a') ... (17') (18a') ; (1a'') (2a'') (4a'') (5a'') >
À 0.1293
2
2
1
2
2
1
1
1
j (1a') ... (17') (18a') ; (1a'') (2a'') (3a'') (4a'') (5a'') >
0.1255
2
2
1
2
1
2
1
1
j (1a') ... (17') (18a') ; (1a'') (2a'') (3a'') (4a'') (6a'') >
0.1067
0.9918
À 0.1277
, 3A''
j (1a') ... (17') (18a') (1a'') (2a'') (3a'') (4a'') >
2
2
1;
2
2
2
1
2
2
1
2
2
1
2
j (1a') ... (17') (18a') ; (1a'') (2a'') (4a'') (5a'') >
slight stabilisation of the dirad-
À1
ical states by 1 ± 2 kcalmol
.
Relative free energies DDG298
The partial atomic charges obtained from a Natural
may also be estimated by adding the free energy corrections
obtained from DFT calculations to the electronic energies
resulting from the MR-CI ‡ Q approach. On this scale, the
Population Analysis (C1: À0.029; C2: À0.317; C3: À0.205)
indicate that the ™central allene∫ carbon atom C2 is, in fact, a
nucleophilic centre. This is in contrast with the prototypical
allene moiety with an electrophilic central allene carbon
atom–a situation that is also found in strained carbocyclic
allenes, such as 1,2-cyclohexadiene or 1,2,4-cyclohexatriene
1
diradical minima are predicted to lie 22.8 ( 6c) and 24.9
3
À1
( 6c) kcalmol above the zwitterion 6b.
2
Calculations of 3d -1H-quinoline (8)and explanation of the
(
isobenzene 18, see Figure 4).[9]
The first excited singlet state of 6 in the planar arrangement
reactivity of its N-methyl derivative (7): Since the DFT
approaches gave results in fair agreement with the more
sophisticated ab initio methods and we were mainly interested
in the qualitative changes of the electronic structure that arise
from benzannulation, only the DFT methods were utilised for
the investigation of the quinoline 8.
1
1
is of A'' symmetry and corresponds to the s,p diradical 6c.
The strongly dominating configuration in the CI expansion
shows that it results from the excitation of an electron from
the non-bonding s orbital (18a') into an antibonding MO
(
4a'') of the cyclic p-electron system (Figure 2), which now
The optimised geometries of the allenic form 8a, the
zwitterion 8b and the diradical 8c were calculated by the
1
hosts seven electrons. As expected for this unfavourable
electronic situation, all theoretical approaches predict this
species to lie significantly higher in energy than 6b (Table 2).
B3LYP method and are presented in Figure 3. In contrast to
the situation of the isopyridine 6, the ground state equilibrium
structure of 8 possesses some allene character (8a), as is
apparent from the average of the dihedral angles C1-C2-C3-
H3 and C3-C2-C1-H1 of 1668, the rather short bonds C1ÀC2
(1.380 ä) and C2ÀC3 (1.378 ä) and the relatively large angle
À1
Table 2. Relative energies[kcalmol ] for various states of isopyridine 6,
as calculated by different methods.
16c
36c
C1-C2-C3 of 1178 (see Figure 3 for the numbering of atoms).
The length of the bond C4ÀC5 (1.425 ä) is increased with
Method
6a
6b
B3LYP/cc-pVDZ
BLYP/cc-pVDZ
±
0
0
0
0
0
‡ 23.0
‡ 24.8
‡ 23.9
‡ 19.1
‡ 20.2
‡ 23.7
‡ 25.0
‡ 27.3
‡ 21.3
‡ 21.8
respect to the corresponding bond in 6 (1.385 ä), which is a
consequence of the aromatic delocalisation of the p bond in 8.
À 0.4
[
a]
MR-CI ‡ Q/cc-pVDZ//DFT
±
CASPT2/cc-pVDZ//DFT[
a]
±
‡ 1.3
For the planar arrangement (C symmetry), we have consid-
s
CASPT2/cc-pVDZ//CASSCF[
b]
ered the zwitterionic and diradical states as for the isopyridine
. The calculation of the harmonic frequencies characterised
the zwitterion 8b as a saddle point of first order, which serves
as transition state for the racemisation of the allene species.
The geometrical parameters of the diradical 8c show trends
Thermochemistry:
B3LYP/cc-pVDZ: DDH298
B3LYP/cc-pVDZ: DDG298
6
±
±
0
0
‡ 21.4
‡ 22.0
‡ 22.1
‡ 21.3
[
a] B3LYP/cc-pVDZ. [b] See text.
Chem. Eur. J. 2003, 9, 4641 ± 4649
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4645

M. Christl, B. Engels et al.
FULL PAPER
2
Figure 3. Optimised CÀC and CÀN bond lengths [ä] and bond angles [8] computed [B3LYP/6 ± 31G(d)] for various stationary points of 3d -1H-quinoline
(
8).
analogous to those of the isopyridine state 6c and will not be
discussed further.
The calculated relative energies of the different stationary
points of 8 are summarised in Table 3. According to these
data, the allene 8a is the most stable species, which will,
however, undergo a quick racemisation through the zwitter-
and C3 in Figure 3). Therefore, the attack of KOtBu at the
allene terminus next to the nitrogen atom is not astounding,
giving rise to 17 eventually. In other words, the reactivity of 7
towards nucleophiles is obviously similar to that of quinoli-
nium salts, such as 16. In this respect, 7 and the chromene 12
show the same behaviour. However, in the case of 12, furan
and styrene react faster than KOtBu and furnish cyclo-
adducts.[ On the basis of the calculations, the different
sensitivity of these intermediates towards nucleophiles may
be caused by different relative energies of the allene
structures and the zwitterions, which represent the transition
states for the racemisation of the allene enantiomers. The
smaller this difference is, the greater should be the zwitter-
ionic character of the allene ground state. Whereas the free
10]
À1
Table 3. Relative energies[kcalmol ] for various states of quinoline 8, as
calculated by DFT methods.
Method
8a
8b[a]
18c
38c
B3LYP/6 ± 31G(d)
BLYP/6 ± 31G(d)
0
0
‡ 1.3
‡ 2.7
‡ 14.4
‡ 17.7
‡ 15.4
‡ 18.0
thermochemistry:
energy difference of 12a and 12b in solution amounts to
B3LYP/6 ± 31G(d): DDH298
B3LYP/6 ± 31G(d): DDG298
0
0
‡ 0.8
‡ 1.4
‡ 14.9
‡ 14.8
‡ 14.2
‡ 13.2
À1 [9]
ꢀ
3 kcalmol , that of 8a and 8b in the gas phase is only
À1
ꢀ
1 kcalmol (Table 3) and most probably the zwitterion 8b
[
a] Transition state.
will become the global minimum in a polar solution.
À1
ion 8b with a barrier of about 1 ± 3 kcalmol . Thus, compared
with the isopyridine 6, benzannulation in 8 stabilises the
allene form relative to the zwitterion. This can probably be
attributed to the increased length of the C4 ± C5 bond, which
effectively increases the length of the tether across the allene
termini and thereby reduces the ring strain in 8a. Interest-
Comparison of the isopyridine 6 with the pyran 14 and the
isobenzene 18: The free energies of the different states of the
isopyridine 6 are compared with those of two isoelectronic
species, namely the pyran 14 and the isobenzene 18, which
have recently been investigated by means of MR-CI, CASPT2
and DFT methods,[ in Figure 4. The variations in the
electronic structures of these species can be best understood,
9]
1
ingly, the diradical 8c is calculated to lie only 14 ±
8 kcalmol above the zwitterion 8b. The comparison with
À1
1
when the nature of the electronic ground state S and the first
0
the parent diradical 6c shows that the gap between the ground
and the first excited singlet state is reduced by 7 ± 9 kcalmol .
excited singlet state S1 for the planar arrangement (Cs
symmetry) are analysed.
À1
The same effect is calculated for the triplet diradical. Since the
change of the geometric parameters on going from 8b to 8c is
similar to that on going from 6b to 6c, electronic factors are
likely to cause this difference in the energy gaps between
zwitterionic and diradical structures. The possible nature of
these factors will be discussed below.
Regarding their zwitterionic states 6b and 14b, respective-
ly, 6 and 14 exhibit closely related structural and electronic
features. In the planar arrangement, the zwitterionic A' state
1
represents the electronic ground state of both species.
However, 14b serves as the transition state for the racemi-
À1
sation of the allene state and is calculated to lie 1.0 kcalmol
A resemblance of the energies of the allene 8a and the
zwitterion 8b, which should also hold for the N-methyl
derivative 7, has already been calculated for the pyran 14 and
was taken there as the reason for the rapid reaction with
KOtBu and the attack of the latter at an allene terminus.[ The
arguments advanced there should also apply to the formation
of 17 from 7, even in the presence of furan or styrene. As in the
case of 6 (Figure 2), the LUMO of 7 should have large
coefficients in positions 2 and 4 of the quinoline system (C1
above the latter. The question arises as to why in the case of 6,
6b is more stable than the allene structure 6a. A qualitative
explanation to this question is offered by an analysis of the
electron densities, calculated by the DFT method, by means of
the Natural Resonance Theory (Table 4). Accordingly, the
two Kekul e¬ forms, which localise the formal positive charge at
the heteroatom, are the dominating resonance structures of
6b, summing up to 35% weight. In 14b, however, these
structures contribute only 22% to the resonance hybrid.[
9]
9]
4646
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 4641 ± 4649

Electronic Structure of Cycloallenes
4641±4649
energy difference between the
2
2
configurations s and p* . Ow-
ing to its higher s character, the
s orbital has the lower orbital
energy, favouring the zwitter-
2
ionic configuration s . On the
other hand, electron repulsion
and the avoidance of charge
separation favours the diradical
1
1
configuration s p* . According
to the relative state energies
calculated for the isopyridine 6,
the zwitterionic state is the
ground state and the diradical
corresponds to the first excited
singlet state. Such systems are
classified as strongly hetero-
symmetric diradicaloids.[ The
34]
underlying reason for this elec-
tronic situation stems from the
shape of the frontier orbitals as
the p* orbital has a strong
antibonding and hence desta-
bilising contribution from the
Figure 4. Comparison of the relative state energies of the isopyridine 6 (this work) with those of the isobenzene
1
8 and pyran 14 (taken from ref. [9]). Relative DG298 values were obtained by adding the thermal corrections from
a DFT vibrational analysis to the MR-CI ‡ Q calculated electronic energies.
nitrogen atom. The weaker the
destabilisation of the p* orbital,
Table 4. Natural resonance theory (NRT) analysis of the minimum
the smaller will be the energy gap between the zwitterionic
and the diradical state, and vice versa.
From this, the differences in the energy gaps between 6b/6c
and 8b/8c can be understood. As to the planar structure of the
quinoline 8, the p* orbital should be stabilised with respect to
the s orbital, as compared to the situation in 6b/6c, since it is
delocalised into the benzo subunit, whereas such an effect
should not be possible for the s orbital. This is corroborated
by the calculated energy difference between the p* and the s
structures 6b and 14b (C
s
, 1A') of the isopyridine 6 and the pyran 14,
respectively.
[
a]
X ˆ NH (6b)
18%
11%
17%
11%
14%
21%
7%
8%
5%
6%
[
b]
X ˆ O (14b)
[
a] This work. [b] Taken from reference [9].
3
3
orbital of 3.42 and 3.26 eV for 6c and 8c, respectively, at the
B3LYP/6-31G(d) level. As a consequence, the calculated
energy gap between the zwitterionic and the diradical state is
These findings mirror the chemical experience that nitrogen,
being less electronegative than oxygen, is the better p donor
and thus provides greater stabilisation of the zwitterion.
As can be seen from the CI wavefunctions discussed above
À1
significantly reduced by almost 10 kcalmol in 8 (Table 3) as
compared to 6 (Table 2).
The correlation of the relative state energies with the
relative energies of the p* orbital and the s orbital is
becoming even clearer, when the isoelectronic systems 14 and
18 are considered (Figure 4).[ Similar to 6, the pyran 14 is a
strongly heterosymmetric diradicaloid, characterised by a
closed-shell, zwitterionic electronic ground state for the
planar geometry. However, the destabilising contribution of
the oxygen atom to the p* orbital is less pronounced as it is
obvious from the energy difference between the p* and the s
orbital at the B3LYP/6-31G(d) level in the triplet diradicals,
(
Table 1), the most important configurations in the CI
expansions arise from the differential occupation of two
MOs, the non-bonding s orbital (18a' in 6b/6c) and the first
antibonding p* orbital (4a'' in 6b/6c, Figure 2). A clear
picture of the electronic structure is provided by the simplest
CI approximation to the electronic structure of this system, in
which only the three lowest-lying singlet configurations and
the lowest-lying triplet configuration are taken into account.
They can be constructed from the different occupations of the
above-mentioned non-bonding s orbital and first antibonding
p* orbital with two electrons. These four configurations have
9]
3
3
which is only 2.99 eV in 14c but 3.42 eV in 6c. This agrees
with the qualitative notion that oxygen is a poorer p donor
than nitrogen, and thus its contributions to the bonding and
antibonding p-frontier orbitals are smaller. In agreement with
been previously analysed for the general case of a planarised
and bent allene.[
1, 33]
The situation here, where the two critical
MOs do not interact as a result of the symmetry but differ in
their energy, is usually discussed in the general formalism of
heterosymmetric diradicaloids.[34] Which electronic state is
lowest in energy depends on the electronegativity difference
between the two orbitals s and p*, a measure of which is the
these arguments, the energy gap between the zwitterionic and
À1
the S diradical state is smaller by ꢀ10 kcalmol in 14 as
1
compared to 6. The most intriguing species, however, is the
isobenzene 18. Owing to the membership of the methylene
group within the ring, the antibonding character of the p*
Chem. Eur. J. 2003, 9, 4641 ± 4649
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4647

M. Christl, B. Engels et al.
FULL PAPER
3-Bromo-1-methyl-1,2-dihydroquinoline (9): The published procedure[12]
was scaled up and modified. Lithium aluminium hydride (1.10 g,
orbital is further reduced, as shown by the energy difference
of the p* and the s orbital of 1.60 eV in the triplet diradical. In
this case, commonly referred to as a weakly heterosymmetric
diradicaloid,[ the state ordering is reversed and, at the
2
3
9.0 mmol) was added in small portions to a stirred suspension of
-bromo-1-methylquinolinium iodide (10.0 g, 28.6 mmol) in anhydrous
34]
diethyl ether (200 mL), kept at room temperature under nitrogen, within
1 h. Afterwards, cold water (08C, 270 mL) was cautiously added with
continued stirring. The layers were separated, the aqueous layer was
extracted with ether (3 Â 50 mL) and the combined organic layers were
planar geometry, the diradical state (18c) represents the
2
ground state S , whereas the zwitterion with the s config-
0
uration (18b) now dominates the S state.
1
dried with MgSO
4
and concentrated in vacuo. The residual yellow oil was
filtered through basic Al
2
3
O (activity IV) with diethyl ether, and the solvent
was evaporated in vacuo to give 9 as an orange oil (5.76 g, 90%), which
1
could be stored under nitrogen at À308C over several weeks. H NMR
Conclusion
(600 MHz, CDCl
3
): d ˆ 2.79 (s, 3H; CH ), 4.26 (d, J2,4 ˆ 1.6 Hz, 2H; H2),
3
6
.53 (brd, J7,8 ˆ 8.2 Hz, 1H; H8), 6.70 (td, J5,6 ˆ J6,7 ˆ 7.4, J6,8 ˆ 0.9 Hz, 1H;
H6), 6.72 (brt, J2,4 ˆ 1.6 Hz, 1H; H4), 6.87 (dd, J5,6 ˆ 7.4, J5,7 ˆ 1.6 Hz, 1H;
H5), 7.16 ppm (ꢀtd, average of J6,7 and J7,8 ˆ 7.8 Hz, J5,7 ˆ 1.6 Hz, 1H; H7);
the assignment is based on H,H COSY and NOESY experiments [NOEs
The reactivity of the six-membered cyclic allenes towards
nucleophiles seems to be closely connected to the structure
associated with the global energy minimum. In the case of the
isobenzene 18 and its benzo derivative 3d -1H-naphthalene,
the allene forms are by far the most stable structures.[ Both
species can be trapped by activated olefins with formation of
cycloadducts,[ even in the presence of KOtBu.
though the zwitterionic state (12b) of the chromene 12
approaches the allene form (12a) to within a few kcalmol ,
inter alia between following signal pairs: 2.79 (CH
3
) ± 6.53 (H8), 6.72
2
(H4) ± 6.87 (H5)]. These data are in accord with the scant information
published.[
10.0 (C8), 114.9 (C3), 117.4 (C6), 121.7 (C4a), 126.3 (C5), 128.4 (C4), 129.3
C7), 143.8 ppm (C8a); the assignment is based on C,H COSY experiment.
11, 12] 13
C NMR (151 MHz, CDCl ): d ˆ 36.9 (CH ), 58.5 (C2),
9]
3
3
1
(
35]
[17, 36]
Al-
Treatment of 9 with KOtBu
À1 [9]
Furan, styrene or THF as the solvent: K Ot Bu (700 mg, 6.24 mmol) was
added in small portions, under nitrogen, to a stirred solution of 9 (673 mg,
the formation of cycloadducts is still the major process, in
3
.00 mmol) and [18]crown-6 (794 mg, 3.00 mmol) in furan (10 mL) at room
spite of the presence of KOtBu.[ However, the pyran 14
10]
[9]
temperature within 5 min, whereby the mixture warmed and turned brown.
Stirring was continued for 2 h and then water (5 mL) was added to the
mixture. The layers were separated, the aqueous layer was extracted with
and the quinoline 7 do not furnish any cycloadducts but only
products resulting from the interception by KOtBu. If the
energetics of 8 can be applied to the N-methyl derivative 7, the
free energy difference between allene form 7a and the
ether (3 Â 20 mL), then the combined organic layers were dried with
1
MgSO
4
and concentrated in vacuo. As shown by a H NMR spectrum, the
residual dark oil contained 1-methyl-1,2-dihydroquinoline (10) and
1-methyl-2-quinolone (11) as major components in a ratio of 1:2. If styrene
or anhydrous THF was used as the solvent instead of furan, the result was
virtually the same. In the case of THF, 18-crown-6 was not added and the
yield was determined to be 16% by means of an internal standard.
À1
zwitterion 7b amounts only to ꢀ1 kcalmol in the gas phase.
Such a small preference of the allene form (14a) over the
zwitterion (14b) was also calculated for the pyran 14.[9] On
consideration of the solvent effect, the zwitterion 14b is even
1
the more stable form in solution,[9] which may be valid for 7b
[D
8
]THF as solvent, immediate analysis by H NMR spectroscopy: K OtBu
(
55 mg, 0.49 mmol) was added to a solution of 9 (30.0 mg, 0.134 mmol) in
as well. Thus, the high polarity of the ground state seems to be
in agreement with a particular sensitivity towards the KOtBu.
Whether this property is accompanied by a low reactivity
towards activated olefins is an open question, which could be
answered by the generation of 7 and 14 in the absence of
KOtBu. However, there is no promising method to reach that
goal. This is also true for the N-methylisopyridine 5, the
intermediacy of which was supported by experimental evi-
dence,[5b] but under conditions that are unsuitable for a test to
prove whether or not 5 can be intercepted by an activated
1
[
D
8
]THF (0.7 mL) in an NMR tube. The H NMR spectrum recorded
immediately afterwards showed that 9 had been completely consumed and
that only one product had been formed, the signals of which are in perfect
agreement with the structure of 2-tert-butoxy-1-methyl-1,2-dihydroquino-
1
line (17). H NMR (200 MHz, [D
8
]THF): d ˆ 1.23 [s, 9H; OC(CH
3 3
) ], 3.05
(
s, 3H; NCH
3
), 5.38 (d, J2,3 ˆ 5.1 Hz, 1H; H2), 5.83 (dd, J3,4 ˆ 9.4, J2,3 ˆ
5
.1 Hz, 1H; H3), 6.56 ± 6.74 (m, 3H; H4, H6, H8), 7.03 (d, J5,6 ˆ 7.4, 1H;
H5), 7.11 ppm (t, average of J6,7 and J7,8 ˆ 7.5 Hz, 1H; H7); since the
reaction mixture was analysed, the line width was not small enough for the
resolution of coupling constants over four bonds. The assignment is in
agreement with that of the signals of 10. After having been kept at room
1
temperature for two weeks, the sample gave a H NMR spectrum that
olefin or not.
showed, as judged on the basis of N-CH
3
1
3
signals (d ˆ 3.00, 3.05, 3.09,
.62 ppm), four major components, among them the quinolone 11 and still
7.
Treatment of 1-methylquinolinium iodide with KOtBu
Experimental Section
THF as solvent, formation of 10 and 11 and conversion of the mixture into
pure 11: A suspension of 1-methylquinolinium iodide (5.00 g, 18.4 mmol)
and KOtBu (2.07 g, 18.4 mmol) in anhydrous THF (100 mL) was stirred
under nitrogen for 30 min at room temperature. Then the mixture was
NMR measurements: Bruker AC200 and DMX600 instruments were used.
Solvent signals were taken as internal standards.
3
-Bromo-1-methylquinolinium iodide: This salt was prepared according to
2 3
filtered through basic Al O (activity IV) with diethyl ether as the washing
[
11]
Kreevoy et al. Since virtually no NMR spectroscopic data are published,
we report them here: H NMR [600 MHz, (CD
CH
8
solvent. The filtrate was concentrated in vacuo to give a brown oil (2.03 g),
which contained 10 and 11 as the major components in a ratio of 1.0:1.3. For
the oxidation of 10 to 11, a procedure of Grignon-Dubois and Meola,[
described for the conversion of N,O-acetals, such as 17, was utilised. Thus,
silica gel (5 g) was added to a solution of the brown oil in dichloromethane
and the surplus solvent was removed in a rotary evaporator. The silica gel
was exposed to air for 4 h and then extracted with acetone (100 mL).
1
3
)
2
SO]: d ˆ 4.63 (brs, 3H;
14]
3
), 8.09 (ddd, J5,6 ˆ 8.1, J6,7 ˆ 7.0, J6,8 ˆ 0.8 Hz, 1H; H6), 8.30 (ddd, J7,8 ˆ
.9, J6,7 ˆ 7.0, J5,7 ˆ 1.4 Hz, 1H; H7), 8.40 (dm, J5,6 ˆ 8.1 Hz, 1H; H5), 8.51
(
dm, J7,8 ˆ 8.9 Hz, 1H; H8), 9.65 (brdd, J2,4 ˆ 1.9 Hz, 1H; H4), 9.91 ppm
(
brdd, J2,4 ˆ 1.9 Hz, 1H; H2); the assignment is based on H,H COSY and
NOESY experiments [NOEs inter alia between following signal pairs: 4.63
1
3
(
[
1
(
CH
151 MHz, (CD
29.52 (C4a), 130.7 (C6), 135.6 (C7), 137.1 (C8a), 148.0 (C4), 151.3 ppm
C2); the assignment is based on a C,H COSY experiment.
3
) ± 8.51 (H8), 4.63 (CH
3
) ± 9.91 (H2), 8.40 (H5) ± 9.65 (H4)]; C NMR
Evaporation of the acetone in vacuo yielded 1-methyl-2-quinolone (11,
1
3
)
2
SO]: d ˆ 45.3 (CH
3
), 114.4 (C3), 119.2 (C8), 129.48 (C5),
1.69 g, 58%) as a reddish brown solid. H NMR (600 MHz, CDCl
3
): d ˆ
3.73 (s, 3H; CH
3
), 6.71 (d, J3,4 ˆ 9.5 Hz, 1H; H3), 7.24 (ddd, J5,6 ˆ 7.7, J6,7 ˆ
7.3, J6,8 ˆ 1.0 Hz, 1H; H6), 7.37 (brd, J7,8 ˆ 8.4 Hz, 1H; H8), 7.56 (brdd,
4648
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 4641 ± 4649

Electronic Structure of Cycloallenes
4641±4649
J
1
5,6 ˆ 7.7, J5,7 ˆ 1.4 Hz, 1H; H5), 7.57 (ddd, J7,8 ˆ 8.4, J6,7 ˆ 7.3, J5,7 ˆ 1.4 Hz,
H; H7), 7.67 ppm (d, J3,4 ˆ 9.5 Hz, 1H; H4); the assignment is based on a
NOESYexperiment [NOE inter alia between the following signal pair: 3.73
[7] I. Yavari, F. Nourmohammadian, D. Tahmassebi, J. Mol. Struct.
(Theochem) 2001, 542, 199 ± 206.
[8] L. F. Frey, R. D. Tillyer, S. G. Ouellet, R. A. Reamer, E. J. J. Grabow-
ski, P. J. Reider, J. Am. Chem. Soc. 2000, 122, 1215 ± 1216.
[9] B. Engels, J. C. Schˆneboom, A. F. M¸nster, S. Groetsch, M. Christl, J.
Am. Chem. Soc. 2002, 124, 287 ± 297.
(
(
1
CH
C8), 120.7 (C4a), 121.8 (C3), 122.1 (C6), 128.7 (C5), 130.6 (C7), 138.9 (C4),
40.1 (C8a), 162.3 ppm (C2); the assignment is based on C,H COSY
3
) ± 7.37 (H8)]; 13C NMR (151 MHz, CDCl
3
): d ˆ 29.4 (CH
3
), 114.1
spectrum.
[10] M. Christl, S. Drinkuth, Eur. J. Org. Chem. 1998, 237 ± 241.
[11] R. M. G. Roberts, D. Ostovi c¬ , M. M. Kreevoy, J. Org. Chem. 1983, 48,
2053 ± 2056.
1
13
These H and C NMR data are in agreement with those in reference [14],
where the assignment is less specific, however. In addition, the H NMR
1
chemical shifts of reference [14] are systematically smaller by ꢀ0.3 ppm.
[12] J. W. Bunting, N. P. Fitzgerald, Can. J. Chem. 1985, 63, 655 ± 662
[13] E. A. Braude, J. Hannah, R. Linstead, J. Chem. Soc. 1960, 3249 ± 3257.
[14] M. Grignon-Dubois, A. Meola, Synth. Commun. 1995, 25, 2999 ± 3006.
[15] E. F. V. Scriven in Comprehensive Heterocyclic Chemistry, Vol. 2
(
3 2
CD ) SOas the solvent , immediate analysis by NMR spectroscopy: K OtBu
(
10 mg, 0.089 mmol) was added to a solution of 1-methylquinolinium
SO (0.7 mL) in an NMR tube. The
3 2
iodide (17.0 mg, 0.0627 mmol) in (CD )
(
Eds.: A. J. Boulton, A. McKillop), Pergamon Press, Oxford, 1984,
NMR spectra recorded immediately afterwards showed that the quinoli-
nium salt had been completely consumed and that only one product had
pp. 165 ± 314.
1
[16] M. Schreck, M. Christl, Angew. Chem. 1987, 99, 720 ± 721; Angew.
Chem. Int. Ed. Engl. 1987, 26, 690 ± 692.
been formed. The H NMR spectrum is in agreement with that obtained
1
from product 17, prepared from 9. H NMR [200 MHz, (CD
[
3
)
2
SO]: d ˆ 1.15
), 5.44 (d, J2,3 ˆ 4.4 Hz, 1H; H2), 5.73
dd, J3,4 ˆ 9.6, J2,3 ˆ 4.4 Hz, 1H; H3), 6.27 (d, J3,4 ˆ 9.6, 1H; H4), 6.43 ± 6.53
m, 2H; H6, H8), 6.91 (dd, J5,6 ˆ 7.5, J5,7 ˆ 1.5 Hz, 1H; H5), 7.01 ppm (ꢀtd,
[
[
17] S. Groetsch, J. Spuziak, M. Christl, Tetrahedron 2000, 56, 4163 ± 4171.
18] R. Ruzziconi, Y. Naruse, M. Schlosser, Tetrahedron 1991, 47, 4603 ±
3 3 3
s, 9H; OC(CH ) ], 2.92 (s, 3H; NCH
(
(
4
610.
1
3
[19] B. Jamart-Gr e¬ goire, S. Mercier-Girardot, S. Ianelli, M. Nardelli, P.
average of J6,7 and J7,8 ˆ 7.7 Hz, J5,7 ˆ 1.5, 1H; H7); C NMR [50 MHz,
CD , 35.4 (NCH ), 66.9 (C2), 78.8 [OC(CH ],
SO]: d ˆ 31.3 (OC(CH
10.6 (C8), 116.0 (C6), 120.1 (C4a), 123.1 (C3), 125.0 (C4), 126.6 (C5), 128.5
C7), 142.3 ppm (C8a). The assignments of the signals in these spectra are
just guesses based on the data of 10.
-Methyl-1,2-dihydroquinoline (10): The original procedure[13] was modi-
Caub e¡ re, Tetrahedron 1995, 51, 1973 ± 1984.
(
3
)
2
3
)
3
3
3 3
)
[
[
[
[
[
20] A. D. Becke, Phys. Rev. A 1988, 38, 3098 ± 3100.
21] A. D. Becke, J. Chem. Phys. 1993, 98, 1372 ± 1377.
22] A. D. Becke, J. Chem. Phys. 1993, 98, 5648 ± 5652.
1
(
23] C. Lee, W. Yang, R. G. Parr, Phys. Rev. B. 1988, 37, 785 ± 789.
24] GAUSSIAN98, Revision A.7, M. J. Frisch, G. W. Trucks, H. B.
Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G. Zakr-
zewski, J. A. Montgomery, Jr, R. E. Stratmann, J. C. Burant, S.
Dapprich, J. M. Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O.
Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi, B. Mennucci, C.
Pomelli, C. Adamo, S. Clifford, J. Ochterski, G. A. Petersson, P. Y.
Ayala, Q. Cui, K. Morokuma, D. K. Malick, A. D. Rabuck, K.
Raghavachari, J. B. Foresman, J. Cioslowski, J. V. Ortiz, A. G. Baboul,
B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.
Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y.
Peng, A. Nanayakkara, C. Gonzalez, M. Challacombe, P. M. W. Gill,
B. G. Johnson, W. Chen, M. W. Wong, J. L. Andres, M. Head-Gordon,
E. S. Replogle, J. A. Pople, Gaussian, Inc., Pittsburgh, PA (USA)
1
fied to resemble the preparation of 9. Thus from 1-methylquinolinium
iodide (3.00 g, 11.1 mmol), the product 10 (1.22g, 76%) was obtained as a
yellow oil, which was sensitive to air, but could be stored without
1
decomposing at À308C under nitrogen. H NMR (600 MHz, CDCl
3
): d ˆ
2
.74 (s, 3H; CH
3
), 4.00 (dd, J2,3 ˆ 3.9, J2,4 ˆ 1.9 Hz, 2H; H2), 5.79 (dt, J3,4 ˆ
9
(
6
J
.8, J2,3 ˆ 3.9 Hz, 1H; H3), 6.32 (dt, J3,4 ˆ 9.8, J2,4 ˆ 1.9 Hz, 1H; H4), 6.47
brd, J7,8 ˆ 7.8 Hz, 1H; H8), 6.60 (td, J5,6 ˆ J6,7 ˆ 7.4, J6,8 ˆ 1.1 Hz, 1H; H6),
.84 (dd, J5,6 ˆ 7.4, J5,7 ˆ 1.6 Hz, 1H; H5), 7.05 ppm (ꢀtd, average of J6,7 and
7,8 ˆ 7.6, J5,7 ˆ 1.6 Hz; H7); the assignment is based on H,H COSY and
NOESY experiments [NOEs inter alia between the following signal pairs:
.74 (CH ) ± 6.47 (H8), 6.32 (H4) ± 6.84 (H5)]. The data published
2
3
[
37]
13
previously contain less information. C NMR (151 MHz, CDCl
7.1 (CH ), 51.7 (C2), 109.7 (C8), 117.0 (C6), 122.2 (C3), 122.3 (C4a), 126.5
C5), 126.6 (C4), 128.9 (C7), 146.0 ppm (C8a); the assignment is based on a
C,H COSY spectrum.
3
): d ˆ
3
3
1
998.
(
[
25] MOLCAS, Version 4, K. Anderson, M. R. A. Blomberg, M. P. F¸l-
cher, G. Karlstrˆm, R. Lindh, P. ä. Malmqvist, P. Neogr a¬ dy, J. Olsen,
B. O. Roos, A. J. Sadley, M. Sch¸tz, L. Seijo, L. Serrano-Andr e¬ s,
P. E. M. Siegbahn, P.-O. Widmark, 1997, University of Lund (Sweden).
26] M. Hanrath, B. Engels, Chem. Phys. 1997, 225, 197 ± 202.
[
[
[
Acknowledgement
27] R. J. Buenker, S. D. Peyerimhoff, Theor. Chim. Acta 1974, 35, 33 ± 58.
28] R. J. Buenker, S. D. Peyerimhoff, Theor. Chim. Acta 1975, 39, 217 ±
This work was supported by the Deutsche Forschungsgemeinschaft and the
Fonds der Chemischen Industrie. Discussions with members of the
Graduiertenkolleg 690 Elektronendichte: Theorie und Experiment were
very helpful.
2
28.
[
[
29] S. R. Langhoff, E. R. Davidson, Int. J. Quant. Chem. 1974, 8, 61 ± 72.
30] DIESEL-MRCI, B. Engels, M. Hanrath, 1997, Universit‰t Bonn
(
Germany), 1997.
[
[
31] T. H. Dunning, Jr., J. Chem. Phys. 1989, 90, 1007 ± 1023.
32] a) NBO, Version 4.0, E. D. Glendening, J. K. Badenhoop, A. E. Reed,
J. E. Carpenter, F. Weinhold, Theoretical Chemistry Institute, Uni-
versity of Wisconsin, Madison, WI (USA), 1994; b) E. D. Glendening,
F. Weinhold, J. Comput. Chem. 1998, 19, 593 ± 609; c) E. D. Glenden-
ing, F. Weinhold, J. Comput. Chem. 1998, 19, 610 ± 627; d) E. D.
Glendening, J. K. Badenhoop, F. Weinhold, J. Comput. Chem. 1998,
19, 628 ± 646.
[
[
1] R. P. Johnson, Chem. Rev. 1989, 89, 1111 ± 1124.
2] M. Balci, Y. Taskesenligil, Advances in Strained and Interesting
Organic Molecules 2000, 8, 43 ± 81.
[
3] a) M. Christl, M. Braun, E. Wolz, W. Wagner, Chem. Ber. 1994, 127,
1
137 ± 1142; b) W. Wagner, Dissertation, Universit‰t W¸rzburg (Ger-
many), 1994.
[
[
4] S. Drinkuth, S. Groetsch, E.-M. Peters, K. Peters, M. Christl, Eur. J.
Org. Chem. 2001, 2665 ± 2670.
5] a) C. J. Emanuel, P. B. Shevlin, J. Am. Chem. Soc. 1994, 116, 5991 ±
[33] B. Lam, R. P. Johnson, J. Am. Chem. Soc. 1983, 105, 7479 ± 7483.
[34] a) M. Klessinger, J. Michl, Excited States and Photochemistry of
Organic Molecules, VCH, Weinheim, 1994; b) V. Bona cœ i c¬ -Kouteck y¬ , J.
Kouteck y¬ , J. Michl, Angew. Chem. 1987, 99, 216 ± 236; Angew. Chem.
Int. Ed. Engl. 1987, 26, 170 ± 189.
5
992; b) W. Pan, P. B. Shevlin, J. Am. Chem. Soc. 1997, 119, 5091 ±
5
094.
[
6] a) R. L. Elliott, A. K. Takle, J. W. Tyler, J. White, J. Org. Chem. 1993,
8, 6954 ± 6955; b) R. L. Elliott, N. H. Nicholson, F. E. Peaker, A. K.
[35] M. Christl, M. Braun, G. M¸ller, Angew. Chem. 1992, 104, 471 ± 473;
Angew. Chem. Int. Ed. Engl. 1992, 31, 473 ± 476.
5
Takle, J. W. Tyler, J. White, J. Org. Chem. 1994, 59, 1606 ± 1607;
c) R. L. Elliott, N. H. Nicholson, F. E. Peaker, A. K. Takle, C. M.
Richardson, J. W. Tyler, J. White, M. J. Pearson, D. S. Eggleston, R. C.
Haltiwanger, J. Org. Chem. 1997, 62, 4998 ± 5016.
[36] M. Christl, S. Groetsch, Eur. J. Org. Chem. 2000, 1871 ± 1874.
[37] J. W. Bunting, W. G. Meathrel, Can. J. Chem. 1972, 50, 917 ± 931.
Received: March 27, 2003 [F5000]
Chem. Eur. J. 2003, 9, 4641 ± 4649
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4649