Reactivity and endo-exo selectivity in Diels-Alder reaction of o- quinodimethanes. An experimental and DFT computational study

Article abstract of DOI:10.1016/S0040-4020(00)00126-5

endo-exo Selectivity in Dieis-Alder cycloadditions of several o- quinodimethanes (1-4) with acrylonitrile, 2-(5H)-furanone and N- methylmaleimide has been investigated in acetonitrile solution. Transition structures of the cycloaddition of the parent o-QDM (1), (E,E)-1,8-dimethyl- o-QDM (2), isoindene (3) and 2,3-dihydronaphthalene (4) with acrylonitrile and maleimide were located at both HF/6-31G* and B3LYP/6-31G* methods. Theoretical data reproduce fairly well both experimental absolute reaction rates and diastereoisomer ratios. The high endo selectivity has been rationalized mainly as a result of solvation effects (acetonitrile, PCM model) and reactant deformations. The latter is due to steric interactions. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0040-4020(00)00126-5

TETRAHEDRON
Pergamon
Tetrahedron 56 (2000) 2547–2559
Reactivity and endo–exo Selectivity in Diels–Alder Reaction of
o-Quinodimethanes. An Experimental and DFT
Computational Study
*
`
Cristiana Di Valentin, Mauro Freccero, Mirko Sarzi-Amade and Riccardo Zanaletti
`
Dipartimento di Chimica Organica, Universita di Pavia, V. le Taramelli 10, 27100 Pavia, Italy
Received 7 December 1999; revised 25 January 2000; accepted 10 February 2000
Abstract—endo–exo Selectivity in Diels–Alder cycloadditions of several o-quinodimethanes (1–4) with acrylonitrile, 2-(5H)-furanone and
N-methylmaleimide has been investigated in acetonitrile solution. Transition structures of the cycloaddition of the parent o-QDM (1), (E,E)-
1,8-dimethyl-o-QDM (2), isoindene (3) and 2,3-dihydronaphthalene (4) with acrylonitrile and maleimide were located at both HF/6-31Gء
and B3LYP/6-31Gء methods. Theoretical data reproduce fairly well both experimental absolute reaction rates and diastereoisomer ratios.
The high endo selectivity has been rationalized mainly as a result of solvation effects (acetonitrile, PCM model) and reactant deformations.
The latter is due to steric interactions. ᭧ 2000 Elsevier Science Ltd. All rights reserved.
Introduction
stereochemical outcome of the reaction. In fact a-phenyl-o-
QDM and (E,E)-a-alkoxy-a0-phenyl-o-QDM react with
dimethylmaleate and crotonate by exo orientation, showing
a 90% exo diastereoselectivity.9,10 The mild conditions used
ꢀT Ͻ 80ЊC† ‘exclude the possibility of a reversible DA
reaction and one is left with the conclusion that at least in
some cases exo approach to aryl substituted o-QDMs does
occur’.1 a-Alkyl-o-QDMs seem to react with dienophiles
with a lower degree of endo selectivity, but this statement
is based on a very limited number of substrates.10
Xylylenes, otherwise known as o-quinodimethanes (o-QDMs),
are a peculiar class of dienes, because they are much more
reactive than related ‘classical dienes’ (such as butadiene
and cyclopentadiene) in Diels–Alder cycloaddition reac-
tions (DA) with olefins, and they seem to show higher
endo selectivity.
o-QDMs have been widely used as intermediates1 in the
synthesis of lignans,2 terpenes, anthracyclines alkaloids,3
steroids and other natural products.4–6 The study of their
reactivity, applied to organic synthesis, is still a hot issue,
as documented in a very recent review by Martın. Thus,
understanding the factors that control the stereochemistry of
this DA reaction is of both synthetic and theoretical rele-
vance. Nevertheless, the only computational investigation
on o-QDMs DA cycloadditions is quite dated.8
Secondary orbital interactions (in particular with highly
reactive electron poor dienophiles such as maleic anhy-
dride),11 steric repulsion12 and even hydrogen bonding1
(for a-hydroxy-o-QDMs) have been suggested as con-
trolling factors of the diastereofacial selectivity in these
cycloaddition reactions.
7
´
More recently Wagner and coworkers pointed out the
importance of the structure of photogenerated o-xylylenols
(E, Z isomers), arising from a twisted triplet o-xylylenol
intermediate, in the control of the stereoselectivity of their
Diels–Alder reactions.13 The author demonstrated that the
diastereoisomer adduct ratio of a-hydroxy-a0-phenyl-o-
QDM addition to maleic anhydride (endo:exoˆ4:1)14
actually is the result of competing endo Diels–Alder reac-
tion on two isomeric o-xylylenols and not the outcome of
competing endo–exo additions on the same geometrical
isomer.
On the other hand, the literature data on o-QDMs endo–exo
selectivity is far from being clearly rationalized. The endo–
exo ratio has been reported to be dependent both on the
nature of the dienophiles and on the substituents at the
o-QDM a-positions. a-Hydroxy, a-alkoxy and a-acetoxy-
o-QDMs cycloadd to dienophiles such as maleic anhydride
and dimethyl maleate by endo addition.9 The presence of a
phenyl group9 at the a position of the o-QDM reverses the
Keywords: o-quinodimethanes; Diels–Alder reactions; diastereoselection;
transition states.
* Corresponding author. Tel.: ϩ39-382-507668; fax: ϩ39-382-507323;
e-mail: freccero@chifis.unipv.it
The latter work emphasizes the obvious observation that the
same stereochemical outcome in a,a0-disubstituted-o-QDM
0040–4020/00/$ - see front matter ᭧ 2000 Elsevier Science Ltd. All rights reserved.
PII: S0040-4020(00)00126-5

2548
C. Di Valentin et al. / Tetrahedron 56 (2000) 2547–2559
DA reaction can be rationalized either in terms of endo (or
exo) competing cycloadditions on two geometrical
isomers or in terms of a competition between endo and
exo cycloaddition on only one geometrical isomer. Thus,
an investigation focused on the controlling factors of the
DA diastereoselectivity for a,a0-disubstituted-o-QDMs
has to choose carefully the derivatives whose DA reactions
involve just one geometrical isomer and not an isomer
mixture (i.e. EE, EZ, ZE and ZZ).
and 2,3-dihydronaphthalene (4) (see Scheme 1) have been
generated in situ by a 1,4 elimination triggered by fluoride
anion (with solid potassium fluoride, at 80ЊC, for the less
reactive dienophiles, or an acetonitrile solution, at room
temperature of tetrabutyl ammonium fluoride) added to an
acetonitrile solution of the corresponding [a-[o-[a-(tri-
methylsilyl)alkyl]phenyl]alkyl]trimethyl ammonium iodide,
according to the method of Ito, Nakatsuka and Saegusa.10
As already reported, o-quinodimethanes generated in situ
dimerize in absence of reactive dienophiles to produce
spiro dimers. The generation of o-QDMs in the presence
of various dienophiles provided cycloadducts in fairly
good yields.10
o-Quinodimethanes with a-alkyl substituents represent
appealing substrates to study the diastereoselectivity
problem in DA reactions for at least two reasons: (i) they
allow one to evaluate experimentally (through the adducts
stereochemistry) the endo–exo approach of the dienophiles,
while at the same time, (ii) alkyl substituents should not
appreciably modify the endo–exo orienting electronic
effects typical of the o-QDM moiety. Thus, the study of
such substrates should allow an evaluation of the role of
the o-QDM moiety in the control of endo–exo selectivity.
Cycloadditions of o-QDM 1
o-QDM 1 generated in situ from [o-((trimethylsilyl)-
methyl)benzyl] trimethylammonium iodide (5) can be
trapped efficiently by acrylonitrile, 2-(5H)-furanone and
N-methyl maleimide to give DA cycloadducts 9, 10 and
11, respectively (Scheme 2).
For the above reasons, we considered worthwhile an experi-
mental and computational investigation centered on the
comparison of reactivity of three a,a0-dialkylsubstituted
o-QDMs [namely, (E,E)-1,8-dimethyl-o-QDM (2), iso-
indene (3) and 2,3-dihydronaphthalene (4)] with electron
poor dienophiles [i.e. acrylonitrile, 2-(5H)-furanone and
N-methylmaleimide (maleimide in the computational
study)] with that of the parent o-QDM 1 with the same
dienophiles.
Cycloadditions of (E,E)-1,8-dimethyl-o-QDM (2)
a,a0-Dimethyl-o-QDM (2) generated in situ from a 3:2
diastereoisomeric mixture of [a-[o-[a-(trimethylsilyl)ethyl]-
phenyl]ethyl]trimethylammonium iodide (6)10 (Scheme 1)
reacted with N-methyl maleimide to give a single cyclo-
adduct (13endo) (Scheme 2).
The aim of our investigation was firstly to assess unambig-
uously the stereochemical outcome by using o-QDMs
having a well known configuration and then to reproduce
at the B3LYP/6-31Gء level of theory the absolute reaction
rates as well as the endo–exo selectivity of the above
o-QDMs, in order to quantitatively evaluate the controlling
effects of the diastereoselectivity of such DA cyclo-
additions.
The endo stereochemistry and the cis relationship between
the two methyl groups of 13endo have been unambiguously
established by X-ray analysis.15 Since it looks highly
unlikely that this product comes from a complete exo dia-
stereoselective cycloaddition on the much less stable dia-
stereomeric (Z,Z)-a,a0-dimethyl-o-QDM [notice (see
below) that (Z,Z)-cyclic,disubstituted-o-QDM, i.e. 3 and 4,
afforded only endo adducts with N-methylmaleimide], we
can safely assume that it derives from the endo cyclo-
addition on (E,E)-a,a0-dimethyl-o-quinodimethane (2).
To the best of our knowledge this is the first attempt to
rationalize the endo–exo selectivity of o-QDM Diels–
Alder cycloadditions with a computational investigation at
DFT level of theory.
The complete (E,E) diastereoselectivity in the formation of
2, i.e. the absence of (E,Z) isomers even starting from a
mixture of diastereoisomeric iodides 6, is noteworthy.
This peculiarity was already underlined by Ito, Nakatsuka
and Saegusa10 in the cycloaddition of 2 with dimethyl
fumarate.
Experimental
Generation of o-QDMs and their trapping with
dienophiles
The cycloaddition reaction of 2 with acrylonitrile affords a
12endo–12exo (86:14) mixture (Scheme 2). The stereo-
chemistry of cycloadducts 12endo and 12exo has been
o-QDM (1), (E,E)-1,8-dimethyl-o-QDM (2), isoindene (3)
Scheme 1.

C. Di Valentin et al. / Tetrahedron 56 (2000) 2547–2559
2549
Scheme 2.
assigned on the basis of the 1H NMR coupling constants and
NOE effects.
group, and consequently its cis relationship to the C-1
methyl group, is supported by couplings involving the
benzyl proton H-4. In fact, this proton occurs as a multiplet
The ring proton H-2 in 12endo (3.02–3.10 ppm), a to the
cyano group (Fig. 1), appears as a ddd (J2,3axˆ12.2 Hz,
J1,2ˆ4.9 Hz, J2,3eqˆ3.4 Hz) whose largest J (with H-3ax)
is certainly the result of an axial–axial coupling while the
other two Js indicate axial–equatorial relationships. These
observations establish H-2 as axial, the CN group as
equatorial and the neighboring benzyl methyl group at
C-1 as axial. The equatorial position of the C-4 methyl
with an axial–axial coupling to H-3ax, that is, J3ax,4
ˆ
11.1 Hz. The relative configuration of the C-1 and C-4
stereogenic centers in 12endo was confirmed by NOE
experiments: irradiation of the proton at d 1.85 (H-3ax)
gave rise to NOE effect for the signals of both the methyl
groups, Me-1 (2.5%) and Me-4 (2.0%), as indicated in Fig. 1.
In the exo isomer (12exo) the H-2 proton (2.78 ppm)

2550
C. Di Valentin et al. / Tetrahedron 56 (2000) 2547–2559
Figure 1. Selected NOE enhancements on 12 and 14 endo–exo isomers.
appears as a ddd showing a small J2,3eqˆ3.8 Hz and two
relatively large coupling constants (J1,2ˆ8.0 Hz and
J2,3axˆ9.0 Hz). This evidence suggests that H-2 is involved
in an axial–equatorial as well as two axial–axial couplings.
Therefore, H-2 is axial while the geminal CN group and the
Me group at C-1 are equatorial and, thus, trans to each other.
Consistently, irradiation of Me-5 (at d 0.80) in 14exo
induced a signal enhancement for the close aromatic proton
H-6 (2.5%), but did not lead to any measurable NOE for the
H-3endo signal.
Cycloadditions of isoindene (3)
The relative configuration of the C-4 stereogenic center (in
particular the axial orientation of Me-4) in adduct 12exo
was established by observation that irradiation of the proton
at d 2.78 (H-2) gave rise to NOE effects for the signals of
both Me-1 (1.9%) and Me-4 (2.7%) as indicated in Fig. 1.
Thus, Me-1, H-2 and Me-4 are cis to each other.
Isoindene (3) generated in situ from (1-(trimethylsilyl)-
indan-3-yl)trimethyl ammonium iodide (7) (Scheme 1),
adds to N-methyl maleimide to give a single endo
(15endo) cycloadduct (Scheme 2). Similarly, N–H male-
imide gave a single endo cycloadduct. Saturation of
H-12syn (Scheme 2) increased the signal intensity of both
H-4 and H-11 by 3.5% thus proving the endo stereo-
chemistry of this cycloadduct.
The cycloaddition of 2 with 2-(5H)-furanone afforded a
mixture of adducts, 14endo–14exoˆ94:6 (Scheme 2),
whose structure, in particular endo–exo stereochemistry,
3 adds to acrylonitrile to afford a 16endo–16exo (77:23)
mixture in poor yields (14%). Indene is the main product
(80% yield) of the reaction. The product distribution analy-
sis suggests that the DA cycloaddition reaction does not
compete efficiently with [1,5]-H shift. No dimers were
detected in the crude reaction. This is the case, as well,
for the cycloaddition of 3 with 2-(5H)-furanone, where no
DA cycloadducts were detected in the reaction mixture.
Long range couplings allowed an easy choice between the
endo and exo structure in the case of adducts 16. H-9anti in
16endo is coupled with H-3endo through a long range W
coupling constant (J3endo,9antiˆ2.7 Hz) whereas it does not
show any measurable coupling constant with H-2 (which
occupies an exo position). On the other hand, in 16exo, a
W coupling path connects H-2 (now in endo position) and
H-9anti and consistently the related long range coupling
constant (J2,9antiˆ2.5 Hz) is observed.
1
has been assigned on the basis of H NMR data associated
with the result of decoupling and NOE experiments. The 1H
NMR spectrum of the more abundant of these adducts, i.e.
14endo, exhibited two doublets at d 0.82 and 1.68, respec-
tively, which are attributable to Me-5 and Me-10 (see
Scheme 2). Irradiation of the higher field doublet made
the multiplet centered at d 2.20 collapse to a doublet
(J4,5ˆ5.0 Hz) while irradiation of the lower field doublet
led to high simplification of the multiplet centered at d
2.38. These observations demonstrate that these two multi-
plets can be attributed to H-5 and H-10, respectively. Proton
H-3endo (d 3.05, dd) and H-3exo (d 3.45, dd) are coupled
to each other (J3endo,exoˆ9.3 Hz) and both, as shown by
decoupling experiments, to H-4 (multiplet at d 2.28,
J3endo,4ˆ6.3 and J3exo,4ˆ9.3 Hz).
The endo stereochemistry of this adduct rests firmly on
NOE experiments (see Fig. 1). In fact, saturation of
Me-5 (at d 0.82) gave rise to increase in signal intensity
of H-3endo (1.6%), of the aromatic proton H-6 (2.8%),
and of H-4 (2.0%) while saturation of H-3endo (d 3.05)
brought about a significant NOE effect in the Me-5 signal
(2.7%).
Cycloadditions of 2,3-dihydronaphthalene (4)
2,3-Dihydronaphthalene (4) generated in situ from a 7:3
diastereoisomeric mixture of (1-(trimethylsilyl)-1,2,3,4-
tetrahydronaph-4-yl)trimethylammonium iodide (8) (Scheme
1), reacted with N-methyl maleimide. A single cycloadduct,

C. Di Valentin et al. / Tetrahedron 56 (2000) 2547–2559
2551
Figure 2. o-QDMs 1–4 optimized geometries at B3LYP/6-31Gء level.
i.e. 17endo, was obtained in fairly good yield (67%). Analo-
gously to N–Me maleimide, maleimide cycloadds to 4 with
a complete endo selectivity. NOE experiments and X-ray
analysis15 did not leave any doubt about its endo stereo-
chemistry. Irradiation of H-4 and H-11 (Scheme 2) brought
about an increase in the H-12syn and H-13syn signal
intensity by 3.0%.
H10-syn by 3.3%. No NOE effects were detected on H-9syn
and H-10syn on irradiation of H-2 in 18exo.
Cycloaddition of 4 with 2-(5H)-furanone gave pure 19endo.
No exo isomer was detected in the crude reaction mixture by
1H NMR analysis. Irradiation of H-3endo led to enhance-
ment of the H-5 signal by 2.3% in accord with the endo
stereochemistry of this adduct.
The cycloaddition with acrylonitrile affords an adduct
mixture, 18endo–18exoˆ84:16, in good yields (80%), in
agreement with the results already published by Ito
(8:1).10 The stereochemistry of the endo cycloadduct has
been assigned on the basis of the finding that irradiation
of H-2 brought about an NOE enhancement of H-9syn and
Computational Methods
The energies and geometries of the reactants (Fig. 2),
adducts as well as of the transition structures (TSs) (Figs.
Figure 3. Exo and endo transition structures (TSs) for the DA cycloadditions of acrylonitrile with o-QDMs 1, 2, 3 and 4, respectively, optimized at B3LYP/6-
31Gء level. Forming bond lengths in the optimized TSs at B3LYP/6-31Gء and HF/6-31Gء levels are given in bold and italic characters, respectively. All bond
˚
lengths are in angstroms (A).

2552
C. Di Valentin et al. / Tetrahedron 56 (2000) 2547–2559
Figure 4. Exo and endo transition structures (TSs) for the DA cycloadditions of maleimide with o-QDMs 1, 3 and 4, respectively, optimized at B3LYP/6-31Gء
level. Forming bond lengths in the optimized TSs at B3LYP/6-31Gء and HF/6-31Gء levels are given in bold and italic characters, respectively. All bond
˚
lengths are in angstroms (A).
3 and 4) were calculated with both HF/6-31Gء method and
with density functional theory (DFT) using the Becke3-LYP
functional16,17 and the 6-31Gء basis set. All calculations
were performed with the Gaussian 94 suite of programs.18
Critical points have been characterized by diagonalizing the
Hessian matrices calculated for the optimized structures;
transition structures have only one negative eigenvalue
(first-order saddle points) with the corresponding eigen-
vector involving the expected formation of the two new
carbon bonds C1C2 and C3C4.
activation parameters, obtained from gas phase vibrational
frequencies, are listed in Table 2.
The contribution of solvent effects to the activation free
enthalpy of the reactions under study are calculated via
the self-consistent reaction field (SCRF) using the Tomasi
model (interlocking spheres) by single point calculations
(i.e. with unrelaxed gas-phase geometries of reactants and
TSs) at the B3LYP/6-31Gء level.19 Table 2 contains the
solvent effect on activation Gibbs free energy, computed
according to Tomasi PCM model for acetonitrile as solvent
(eˆ35.9).
In order to produce theoretical activation parameters, vibra-
tional frequencies (in the harmonic approximation) were
calculated for all the optimized B3LYP/6-31Gء structures
and used, unscaled, to compute the zero point energies, their
thermal corrections, the vibrational entropies, and their
contributions to activation enthalpies, entropies and free
enthalpies. The computed electronic energies for the reac-
tants and transition structures (TSs) and the thermodynamic
Computational Results and Discussion
Experimental data show the endo attack on 2, 3 and 4 as
highly predominant even for a dienophile such as acrylo-
nitrile (86, 77 and 84% endo adducts, respectively), which
has always shown a negligible endo selectivity with classi-
cal dienes.20 Somewhat surprisingly diastereoselectivity
does not significantly change on changing the o-QDM
derivative, in particular on passing from QDM 2 [with
acyclic (E,E)-dimethyl substitution] to QDM 3 [with cyclic
(Z,Z)-dimethylene substitution].
Table 1. Forming bond length and dihedral angles (a and b) in TSs
(B3LYP/6-31Gء) with maleimide
a
a
TS
C4C5
C10C11
ab
bb
28
30
32
29
31
33
exo
2.464
2.437
2.481
2.473
2.433
2.486
2.461
2.423
2.472
2.462
2.410
2.476
112.77
99.08
103.95
116.39
103.46
110.04
99.86
102.97
106.67
102.39
102.23
103.82
Complete endo selectivity is observed experimentally with a
more reactive dienophile such as N–Me maleimide for all
o-QDMs 2–4.
endo
In order to clarify the controlling factors of such a high endo
selectivity we carried out ab initio calculation at both HF/
6-31Gء and B3LYP/6-31Gء level of theory, for the cyclo-
addition reactions of 1–4 with acrylonitrile and maleimide,
a
˚
Bond length in A.
b Angles in degree (see text and Fig. 4 for the definition of a and b
angles).

C. Di Valentin et al. / Tetrahedron 56 (2000) 2547–2559
2553
Table 2. Calculated activation parameters in gas-phase (at 298 K) for o-QDM 1–4 Diels–Alder reactions, with acrylonitrile (20–27) and maleimide (28–33) at
the B3LYP/6-31Gء (HF/6-31Gء) leveld
Structures
E (Hartree)
DE±
DDE±a,c
0.83 (0.74)
DH±
DS±
DG±
DDG±b,c
exo TS 20
endo TS 21
exo TS 22
endo TS 23
exo TS 24
endo TS 25
exo TS 26
endo TS 27
exo TS 28
endo TS 29
exo TS 30
endo TS 31
exo TS 32
endo TS 33
Ϫ480.4283011
Ϫ480.4269891
Ϫ559.0623486
Ϫ559.0611666
Ϫ518.5475298
Ϫ518.5465850
Ϫ557.8559783
Ϫ557.8544747
Ϫ669.0287195
Ϫ669.0307918
Ϫ707.1480376
Ϫ707.1495153
Ϫ746.4549065
Ϫ746.4587941
5.77 (26.72)
6.60 (27.47)
5.40 (26.68)
6.14 (27.29)
6.80 (25.13)
7.39 (25.64)
11.21 (32.58)
12.15 (32.92)
3.86 (23.77)
2.56 (21.95)
4.83 (23.01)
3.90 (20.87)
10.23 (31.16)
7.79 (27.35)
7.88
8.63
6.92
7.55
8.46
9.01
12.82
13.63
5.89
4.57
6.46
5.55
11.79
9.28
Ϫ28.90
Ϫ29.36
Ϫ35.64
Ϫ35.93
Ϫ33.44
Ϫ33.71
Ϫ32.35
Ϫ32.32
Ϫ29.61
Ϫ30.98
Ϫ34.35
Ϫ35.47
Ϫ33.80
Ϫ34.65
16.50
17.39
17.54
18.26
18.42
19.06
22.46
23.26
14.72
13.80
16.70
16.11
21.87
19.61
0.89 (0.76)
–
0.72 (0.64)
0.74 (0.62)
0.59 (0.51)
0.64 (0.50)
0.94 (0.34)
0.80 (0.14)
–
Ϫ1.30 (Ϫ1.82)
Ϫ0.93 (Ϫ2.14)
Ϫ2.44 (Ϫ3.81)
Ϫ0.92 (Ϫ1.45)
–
Ϫ0.59 (Ϫ1.73)
Ϫ2.26 (Ϫ3.67)
–
a DDE±ˆDEe±ndoϪDEe±xo: endo–exo relative electronic activation energy.
b DDG±ˆDGe±ndoϪDGe±xo: endo–exo relative Gibbs free activation energy.
c A negative value means a more stable endo TS, vice versa a positive value means a more stable exo TS. Electronic energies (in Hartree) of the reactants:
acrylonitrile: EˆϪ170.8315508, N–H maleimide: EˆϪ359.4289178, 1: EˆϪ309.6059506, 2: EˆϪ388.2394072, 3: EˆϪ347.7268126, 4: Eˆ
Ϫ387.042294.
d Energies in kcal molϪ1, entropy in eu; standard state (298 K) of the molar concentration scale (gas in ideal mixture at 1 mol lϪ1, Pˆ1 atm); DE± is the
electronic activation energy; DH±, DG± are the molar enthalpy and free Gibbs energy, DS± is the molar activation entropy. For conversion from 1 atm
standard state to 1 mol lϪ1 standard state (both for gas-phase) the following contributions need to be added to standard enthalpy, free Gibbs energy, and
entropy, respectively: ϪRT, RT ln R0T, ϪR ln R0TϪR, where R0 is the value of the R constant given in l atm molϪ1KϪ1. For a reaction with AϩBˆC
stoichiometry, the corrections for DH±, DG± and DS± are RT, ϪRT ln R0T, R ln R0TϩR. At 298 K the corrections amount to 0.59 and Ϫ1.90 kcal molϪ1
for DH± and DG±, respectively, and 8.34 eu for DS±. A further correction of R ln 4 to DS± (and consequently RT ln 4 to DG±) has to be added to take into
account statistic effects.
which show respectively, the lowest and the highest endo
selectivity.
deviation from planarity. UB3LYP/6-31Gء calculation
reveals that the o-QDM triplet shows a planar structure
with an aromatic ring. In fact the diradical species does
not show any single–double bond alternation of the inner
We were able to locate both the endo and exo TSs for the
cycloadditions of 1–4 with acrylonitrile (Fig. 3) as well as
for the reactions of 1, 3 and 4 with parent maleimide (Fig.
4). Our aim was to produce not only geometries but also
reliable relative (endo–exo) and absolute energies, in order
to properly describe the competing endo–exo pathways
taking into account the solvent effect as well.
˚
bonds (C2C3ˆ1.405, C3C4ˆ1.407 A). The spin density
(0.85) is mainly located on the exo CH2 groups of such a
diradical species. The electronic energy is 22.28 kcal molϪ1
higher than that of the closed shell o-QDM 1.
Transition structures of o-QDMs 1–4 with acrylonitrile
Looking at the TS geometries, the most significant para-
meters that change with the introduction of the electronic
correlation (at DFT level) are the forming bond distances. A
comparison between HF/6-31Gء and DFT calculations
reveals that DFT TSs (Fig. 3) always exhibit longer incipi-
ent bond lengths than the corresponding HF TSs and that
this difference is more pronounced for the longer C1- -C2
bond than for the shorter C3- -C4 bond (see Fig. 3 for
numbering). As a result, forming bond length asynchro-
nicity predicted by HF calculations for TSs of cyclo-
additions between 1–4 and acrylonitrile is enhanced on
passing to DFT calculations. At the latter theory level the
forming bond length asynchronicity spans the range from
First, for sake of clarity, we will discuss the transition
structures and endo–exo selectivity for the reactions of
QDMs with each dienophile, one at a time. Then, the
absolute rate constants and the endo–exo selectivity are
addressed in comparison with experimental data.
o-QDMs structures
Full geometry optimization was carried out for the o-QDMs
1–4. Both o-QDMs 1 and 3 have planar structures (Fig. 2).
The deviation from planarity of the diene moiety, evaluated
from the dihedral angle C1C2C7C8 (Fig. 2 for numbering), is
moderate in 4 (12Њ) and much more pronounced in 2 (28.0Њ).
It is accompanied by a slight conrotatory rotation around the
C1C2 and C7C8 bonds. The optimized geometries show a
clearcut single–double bond alternation (for instance
˚
˚
0.52 A, for exo TS 26 to Ϸ0.35 A for TSs 21, 24 and 25.
The forming bond lengths in these TSs are longer than those
of the forming bonds in the corresponding TSs of the
cycloaddition between the less reactive cyclopentadiene
and acrylonitrile at both HF/6-31Gء20 and B3LYP/
6-31Gء21 theory levels, suggesting that o-QDM TSs, in
accord with the Hammond postulate, are more reactant
like than cyclopentadiene TSs.
˚
C1C2ˆ1.353, C2C3ˆ1.462, C3C4ˆ1.351 A in 1 and C1C2ˆ
˚
1.362, C2C3ˆ1.443, C3C4ˆ1.360 A in 3), which decreases
in the order 1Ͼ2Ͼ4Ͼ3. The stability of SCF solution for the
optimized structures 1–4 (at the B3LYP/6-31Gء level) has
been successfully tested. Such stability, at this level of
theory, indicates that the lowest energy wavefunction for
all o-QDMs studied is a closed shell singlet and not a
diradical, even for o-QDMs 2 and 4 which show significant
Other characteristic geometrical features of these TSs, such
as the sliding motion of the acrylonitrile moiety underneath

2554
C. Di Valentin et al. / Tetrahedron 56 (2000) 2547–2559
the diene moiety on going from both endo and exo TSs for
the reactions of 1 and 2 to those for the reactions of 4 and 5,
are commented in detail for the related TSs of maleimide
reactions.
approach geometry as discussed above but also they should
give rise to reactant deformations (in the TSs) different for
the exo as compared to the corresponding endo TSs.
Contribution of these deformations to the electronic acti-
vation barrier can be very easily obtained by calculating
the energy necessary to deform the reactant geometries to
those they assume in the TS, i.e. by single point calculation
on the reactant deformed to their TS geometries. Thus, we
computed, at B3LYP/6-31Gء level, the contribution of the
reactant deformation (Edef) to the electronic activation
barriers (DE±) for the cycloaddition reactions of o-QDM
1, 3 and 4 with maleimide. The deformation term always
favors selectively the endo approach over the exo one and it
contributes to the DDE± (DEe±xoϪDEe±ndo) by 0.66, 0.57 and
2.28 kcal molϪ1 for the cycloaddition of maleimide with 1,
3 and 4, respectively. Such contribution represents 50, 61
and 93% of the endo–exo electronic energy difference. Thus
we can confidently say that reactant deformation (caused by
steric repulsions) in cycloaddition reactions involving
o-QDMs and a reactive dienophile such as maleimide, is
one of the main controlling factor of the endo–exo selec-
tivity in the gas-phase.
Transition structures of o-QDMs with maleimide
Three couples of endo–exo isomeric TSs (Fig. 4) were
located. They are all concerted and perfectly synchronous
at HF and slightly asynchronous at B3LYP level of theory,
although 28–31 TSs derive from symmetrical reactants.
Asynchronicity in TSs arising from symmetric reactants is
not a novelty in computational studies of Diels–Alder
cycloadditions and it can be much more dramatic than
that observed in our reactions.22,23 Potential surface in the
region of the TS is probably flat with the respect to
the asymmetric stretching vibration that involve the two
incipient bonds.
In order to describe some relevant aspects of TS geometries
it is useful to focus our attention on the dihedral angles
between the forming bond plane and both the diene (i.e.
aˆC10C9aC5aC5–C10C5C4C11) and dienophile heavy atom
plane (bˆC10C5C4C11–C1C11C4C3). In the TSs the carbon
skeleton of both the QDM-diene and dienophile moieties are
roughly confined to a plane.
And now a typical question concerning the DA cyclo-
addition reactions: is there evidence that the secondary
orbital interactions play an important role in favoring the
endo selectivity?
As for exo TSs, on going from 28 to 30 and 32 one can
observe a sliding motion of the maleimide moiety under-
neath the quinodimethane moiety accompanied by a clock-
wise (downward) rotation of the maleimide plane around the
C4–C11 bond. The sliding motion reduces both the a and b
angles but in the case of the latter angle its effect is counter-
acted and overcome by the opposite, i.e. widening, effect of
the downward maleimide rotation. Actually, there is a
decrease by Ϸ13 and 9Њ in the a angle value in 30 and
32, respectively, while the b angle widens by 3 and 7Њ. It
seems that sliding is larger in 30 than in 32 while the
rotation is the same in these TSs. Obviously, maleimide
rotation also changes the dihedral angle between the diene
and dienophile heavy atom planes; the angle between these
planes is 167Њ in 28 and Ϸ183–184Њ in TSs 30 and 32 where
they are almost parallel to each other.
A quantitative answer requires a factorization of activation
energy and we are still working in this field.24 Nevertheless,
there are few geometrical evidences which cast serious
doubt on their role: (i) the carbon atoms C5a and C9a (Fig.
4) are always slightly anti pyramidalized with respect to the
face attacked by the dienophile; (ii) the pyramidalization of
such carbons is the same in endo and exo TSs;† (iii) the
cycloaddition reaction of isoindene has the best endo TS
geometry (smallest a and b angles of the series, i.e.
aˆ103Њ and bˆ102Њ in the endo TS) for a maximization
of secondary orbital interactions, but actually, it is the less
endo selective (at the B3LYP level, see Table 2).
Although, no direct comparison to the DA cycloaddition of
butadiene or cyclopentadiene with maleimide is possible,
since no data are reported in the literature for the latter
reactions, the forming bond lengths in TSs 28 and 29 are
longer than those in TSs for the cycloaddition between
isoprene and maleic anhydride25 at the same level of theory.
Thus, as emphasized above for TSs of o-QDM cyclo-
additions with acrylonitrile, also TSs for cycloadditions of
o-QDMs with maleimide are more reactant-like than those
for DA reactions of classical dienes.
The maleimide sliding motion as well as clockwise (now
upward) rotation are observable also for endo TSs, i.e. on
passing from 29 to 31 and 33. Once again the former geo-
metry change leads to a decrease in the a angle value by 13
and 6Њ in 31 and 33, respectively. Also in endo TSs sliding
motion and clockwise rotation have opposite effects on the
b angle value but now the former gives rise to widening and
the latter to narrowing. Moreover, they compensate each
other almost exactly as shown by the remarkable substantial
constancy of the b angle in endo TSs. The dihedral angle
between the diene and dienophile heavy atom planes is
Ϸ39Њ in 29, 26Њ in 31 and 34Њ in 33 (Table 1).
Absolute reactivity
In order to assess the reliability of the computational
approach in the reproduction of the experimental reactivity,
we made a direct comparison of the absolute computed
It is quite evident that these dienophile motions take place in
order to better accommodate the steric congestion of the
aliphatic bridge and to lessen as far as possible the related
non-bonded repulsive interactions.
†
Pyramidalization of C5a and C9a carbon atoms, measured by C6C5aC9aC5
and C9C9aC5aC10 respectively, is: Ϫ173.2, 173.2Њ in 28; Ϫ172.8, 172.8Њ in
29; Ϫ176.7, 176.6Њ in 30; Ϫ177.3, 177.1Њ in 31; Ϫ171.2, 171.2Њ in 32;
Ϫ172.3, 172.1Њ in 33.
The effect of steric repulsions should influence not only the

C. Di Valentin et al. / Tetrahedron 56 (2000) 2547–2559
Table 3. Solvent effect on activation Gibbs free energy (dDG±MeCN) in kcal molϪ1 f
2555
a
b
d
c
c,e
Structures
dDG±MeCN
DG±MeCN
DG±MeCN,app
DG±exp
DDG±MeCN
DDG±exp
(kcal molϪ1
)
(kcal molϪ1
)
(kcal molϪ1
)
(kcal molϪ1
)
(kcal molϪ1
)
(kcal molϪ1
)
exo TS 20
endo TS 21
exo TS 22
endo TS 23
exo TS 24
endo TS 25
exo TS 26
endo TS 27
exo TS 28
endo TS 29
exo TS 30
endo TS 31
exo TS 32
endo TS 33
Ϫ1.43
Ϫ2.04
Ϫ0.64
Ϫ1.61
Ϫ1.45
Ϫ2.27
Ϫ1.22
Ϫ2.49
Ϫ1.14
Ϫ1.20
Ϫ1.47
Ϫ1.91
Ϫ0.96
Ϫ2.12
15.07
15.35
16.90
16.65
16.97
16.79
21.24
20.77
13.58
12.60
15.23
12.33
20.91
17.49
14.78
15.9
–
ϩ0.28
–
16.35
Ϫ0.25
Ϫ1.07
Ϫ0.72
Ϫ0.98
–
16.46
–
Ϫ0.18
20.55
20.1
11.4
–
Ϫ0.47
12.50
Ϫ0.98
12.23
Ϫ1.03
ϽϪ2.0
ϽϪ2.0
17.49
14.7
Ϫ3.42
a DG±MeCN includes the dDGM±eCN correction (DGM±eCNˆDG±ϩdDG±MeCN).
±
exoϪDG±endo†=RT
b
Contribution of both endo and exo pathways: DG±appˆDGe±xoϩRT ln‰1ϩeꢀDG
Š
if the exo TS is the most stable, or DG±appˆDGe±ndoϩ
Ϫ1
±
endoϪDG±exo†=RT
RT ln‰1ϩeꢀDG
Š
if the endo TS is the most stable.
Ϫ1
c A negative value means a more stable endo TS, vice versa a positive value means a more stable exo TS.
d Measured by flash photolysis in acetonitrile solution.
e Calculated from experimental endo–exo adduct ratios.
f Electrostatic solvent effect according to the Tomasi model (acetonitrile, eˆ35.9) at the B3LYP/6-31Gء level of theory. It is given as the difference between
the solvent effect on the TSs and that on the reactants) and comparison between the theoretical absolute (DG±MeCN), and relative activation Gibbs free energy
(DDGM±eCNˆDGM±eCN,endoϪDGM±eCN,exo) with the experimental data ꢀDG±exp; DDGe±xp, respectively.
Gibbs free activation energies (DG±) with the experimental
values (DG±exp). The latter were obtained from the measured
second-order rate constant for DA reaction of 1 and 4,
generated in situ by flash photolysis (from 2-indanone and
1,2,3,4-tetrahydro-1,4-methanonaphthalen-9-one, respec-
tively)26 with acrylonitrile, maleimide and N-methylmalei-
mide. These data will be reported in detail elsewhere.26 The
kinetic experimental data for the cycloadditions involving 3
were not measured due to the lack of a suitable and efficient
photochemical precursor. The second-order reaction rates of
the cycloaddition involving the parent o-QDM 1 and 2,3-dihy-
dronaphthalene 4 with acrylonitrile were measured in aceto-
nitrile at 25ЊC (kˆ1.1(^0.2)×101, 1.3(^0.1)×10Ϫ2 MϪ1 sϪ1).
In a similar fashion we measured second-order reaction
rates of the reaction of the parent o-QDM 1 and 2,3-dihydro-
naphthalene 4 with N–H maleimide (kN–Hˆ2.9(^0.1)×104,
kN–Hˆ1.1(^0.1)×102 MϪ1 sϪ1, for 1 and 4, respectively).
N–Me maleimide is roughly 10 times more reactive than
the parent maleimide, showing second-order rate constants
of 2.5(^0.1)×105 and 1.8(^0.1)×103 MϪ1 sϪ1, for the
cycloaddition reaction of 1 and 4, respectively.
TSs, respectively). Thus, if we want to compare theoretical
and experimental absolute values we cannot neglect solvent
effects, in particular electrostatic effects. Indeed, evaluation
of electrostatic solvent effect through a PCM model
(Tomasi) by single point calculation at the B3LYP/6-
31Gء level for polar acetonitrile (dielectric constantˆ35.9)
demonstrated that both endo and exo TSs are stabilized
with respect to the reactants by roughly 0.6–2.5 kcal molϪ1
(see Table 3). The theoretical absolute apparent activation
free energies, obtained taking into account both the endo
and exo pathways (kappˆkendoϩkexo), after correction with
the solvent effect (DG±MeCN,app) reproduce fairly well the
experimental values (DG±exp). The results are summarized
in Table 3.
endo–exo Controlling factors
The computed gas-phase relative activation free energies
DDG± (DGe±ndoϪDG±exo, see Table 2) indicate a preference
for the exo attack in acrylonitrile cycloadditions (by 0.6–
0.9 kcal molϪ1), and for endo addition in maleimide (by
0.6–2.3 kcal molϪ1) reactions with o-QDMs 1–4. The
predicted distereoselectivity is substantially similar at both
HF/6-31Gء and B3LYP/6-31Gء level of theory, even if
without electronic correlation the preference for the endo
TSs is slightly higher. The exo computational selectivity of
the acrylonitrile gas-phase cycloadditions (see Table 1) is in
contrast with the experimental results in acetonitrile solu-
tion (see DDG±exp in Table 3, evaluated from the experi-
mental endo–exo ratio) which show the endo attack to be
preferred. On the other hand, the endo selectivity in male-
imide cycloadditions with 1–4, computed in gas-phase, is
underestimated. However, it is well known that solvent
effect can influence, usually by increasing, the endo–exo
selectivity.27 Thus, in order to computationally match the
experimental results it is necessary to take into account the
solvent effects.
The corresponding activation free energies DG±exp (aceto-
nitrile solution, Table 3) for the DA cycloaddition
1ϩacrylonitrile, 1ϩN–H maleimide, 4ϩacrylonitrile and
4ϩN–H maleimide are as follows: 15.9, 20.1, 11.4,
14.7 kcal molϪ1. Comparison between these experimental
values in solution and the corresponding theoretical gas-
phase DG± values shows that the latter are always higher
than the former by 0.6 –4.9 kcal molϪ1 (Table 2). Notwith-
standing such discrepancy, the reactivity order 1Ͼ4 is well
reproduced computationally.
However, it should be emphasized that there is always an
increase in the dipole moment on going from the reactants
(e.g. mˆ3.88 D for acrylonitrile, 1.49 D for maleimide,
0.12 D for 1, 0.47 D for 3 and 0.57 D for 4), to TSs
(mˆ4.18, 4.43, 4.67 and 4.71 D for 21, 23, 25, and 27

2556
C. Di Valentin et al. / Tetrahedron 56 (2000) 2547–2559
The introduction of electrostatic solvent effect in calcu-
lations not only increases the predicted reactivity of the
systems under study, as described in the preceding section,
but it changes the relative endo–exo selectivity as well. In
fact the endo TSs are stabilized in comparison with the exo
TSs by 0.06–1.27 kcal molϪ1 (Table 3). Selective endo
stabilization by solvent effect in DA cycloadditions has
already been described computationally, e.g. by
Sustmann28,29 and Ruiz–Lopez.27,30 Anyway, even with
the introduction of the solvent effect the computational
endo–exo ratio is systematically underestimated by about
do not appear to be particularly important in the control of
the endo–exo selectivity, at least in Diels–Alder cyclo-
additions under study.
Noteworthy is the ability of the correlated calculation (DFT)
to predict fairly well, with the inclusion of the solvation
effects, the absolute and relative (endo–exo) reactivity of
o-QDMs in DA cycloaddition reactions.
Experimental
0.5 kcal molϪ1
.
Melting points are uncorrected. Elemental analyses were
made on a Carlo Erba CNH analyzer, model 1106. Infrared
spectra were recorded as KBr discs, nujol or films on a
Perkin–Elmer FT1000 spectrophotometer. 1H and 13C
NMR spectra were recorded in CDCl3 solutions (unless
otherwise stated) on a Bruker AE 300 spectrometer with
Me4Si as internal standard. Protons were correlated by
decoupling and COSY experiments, while protons were
Such a discrepancy could be attributed to a systematic
underestimation of the solvent effect at PCM level, because
the dielectric continuum model is approximate and it is less
accurate in polar solvent such as acetonitrile.28
It is clear from the results reported above that to describe in
detail the reactivity and the endo–exo selectivity of o-QDM
DA cycloadditions one must pay attention to solvent effect.
In particular for cycloaddition involving acrylonitrile as
dienophile the solvent becomes the controlling factor of
the endo–exo selectivity, reverting the exo selectivity
computed in the gas-phase.
1
correlated to carbons by H–13C heterocorrelated spectra.
GC analyses were performed with a DANI 6500, PTV injec-
tor, CP-Sil 19CB (25 m) capillary column using H2 as a
carrier. Preparative column chromatography was performed
on Merck Silica gel 60 (79–230 mesh) using cyclohexane–
ethyl acetate mixtures (from 95:5 to 60:40) unless otherwise
stated.
Similar stabilization of the endo pathway by solvent effect is
at work in the DA cycloadditions of o-QDMs 1, 3 and 4 with
maleimide as well, but the computational analysis of the DA
cycloaddition reaction with such dienophile suggests that
other endo orienting effects emerge clearly already in the
gas-phase. We found, as described in the previous section,
that steric effects, as a result of introduction of a methano or
ethano bridge in o-QDM, not only give rise to significant
changes in the geometry of approach between the two reac-
tion partners but also can be held responsible of an ener-
getically more expensive reactant deformation in exo as
compared to the corresponding endo TSs. This deformation
effect can account for a large part of the endo–exo energy
difference.
[o-((Trimethylsilyl)methyl)benzyl]trimethylammonium
iodide (5), [a-[o-[a-(trimethylsilyl)ethyl]phenyl]ethyl]tri-
methylammonium iodide (6), [1-(trimethylsilyl)-1,2,3,4-
tetrahydronaphth-4-yl] trimethylammonium iodide (8)
were prepared according to published literature.10 Adducts
9, 18endo and 18exo have been already characterized.10
[1-(Trimethylsilyl)-3-dimethylamino]indan. To a solution
of N,N-dimethyl-1-indanylamine31 (4.5 g, 27.90 mmol) in
diethylether (90 ml), 1.6 M n-BuLi in hexane (45 ml,
72.0 mmol) was added at room temperature. The solution
was stirred for 22 h at room temperature. A mixture of
TMSCl (9 ml, 70.91 mmol) and triethylamine (1 ml) was
added to the resulting reddish solution all at once at 0ЊC.
After being stirred at room temperature for 2 h, the mixture
was quenched with cold aqueous sodium carbonate and
extracted with ether. The ether extract was washed with
brine and evaporated to give a mixture of [1-(trimethyl-
silyl)-3-dimethylamino]indan, [1-(dimethylamino)-7-trimethyl-
silyl]indan and [1,1-(bis(trimethylsilyl))-3-dimethylamino]indan.
The mixture was dissolved in 10 M sulfuric acid (50 ml) and
acetic acid (70 ml) and heated at 50ЊC for 8 h.
Conclusion
The cycloaddition of o-QDMs to electron-poor olefins has
been rationalized in the frame of a concerted mechanism. In
fact, the involvement of a triplet diradical o-QDMs is
unlikely because such a species is more than 22 kcal molϪ1
less stable than the closed shell o-QDM 1, and above all, it is
even less stable than the TSs of the concerted cycloadditions
with both acrylonitrile and maleimide.
The reaction mixture was made alkaline by addition of solid
sodium carbonate and extracted with ether. The ether extract
was washed with brine and dried over MgSO4. Evaporation
of the solvent followed by vacuum distillation afforded 3.8 g
(58.3% yield) of the desired product as a colorless oil [bp
The endo selectivity observed experimentally in o-QDMs
cycloaddition reactions with acrylonitrile and more reactive
dienophiles such as maleimide has been explained on the
basis of a computational study at B3LYP/6-31Gء level as a
result of a selective endo electrostatic stabilization by a
highly polar solvent such as acetonitrile, which is able
even to revert the exo selectivity with acrylonitrile in the
gas-phase. For a more reactive and rigid dienophile the
preference for an endo approach (in the gas-phase) has
been attributed mainly to deformation effects of both
reagents in the TSs. Secondary orbital interactions if present
1
103ЊC (0.3 mmHg)]: IR (nujol) 1249, 899, 838 cmϪ1. H
NMR (CDCl3) d 0.11 (s, 9H, TMS), 1.93 (ddd, 1H,
Jˆ13.0, 8.0, 3.9 Hz), 2.25 (s, 6H), 2.31 (ddd, 1H, Jˆ13.0,
10.0, 8.0 Hz), 2.50 (dd, 1H, Jˆ10.0, 3.9 Hz), 4.32 (t, 1H,
Jˆ8.0 Hz), 7.05–7.18 (m, 4 H). Anal. Calcd for C14H23NSi:
C, 72.04; H, 9.93; N, 6.00; Si, 12.03. Found: C, 71.95; H,
9.92; N, 6.05.

C. Di Valentin et al. / Tetrahedron 56 (2000) 2547–2559
2557
[1-(Trimethylsilyl)-indan-4-yl] trimethylammonium iodide
(7). A solution of [1-(trimethylsilyl)-3-dimethylamino]-
indan (820 mg, 3.52 mol) and methyl iodide (1.6 ml,
25.6 mmol) in acetonitrile solution was refluxed for 2 h.
A large amount of ether was added to precipitate
[1-(trimethylsilyl)-indan-4-yl]trimethylammonium iodide
which was washed with ethyl acetate to obtain 1.1 g (83%
yield) of 7 as a white solid: mp 181^1ЊC (from ethyl
(12endo and 12exo). To a degassed solution of 6 (230 mg,
0.59 mmol) in 4 ml of acetonitrile–acrylonitrileˆ1:1
mixture, a degased solution of TBAF (240 mg, 0.76
mmol) in 10 ml of acetonitrile was added over 1 h at room
temperature. The resulting solution was stirred for an addi-
tional hour. The reaction mixture was diluted with ether and
insoluble material was filtered off.
acetate); IR (nujol) 1467, 1250, 840 cmϪ1
.
1H NMR
The filtrate was concentrated, and column chromatographed
on silica gel to give 50 mg (49%, yield) (eluentˆcyclo-
hexane–ethyl acetateˆ8:2) of a mixture of 12endo and
12exo (84:16) as an oil. endo–exo Ratio was evaluated by
1H NMR on the reaction mixture. Adducts 12endo and
12exo were characterized in mixture.
(CDCl3) d 0.80, (s, 9H), 2.72–2.89 (m, 2H), 3.34 (s,
9H), 3.55 (m, 1H), 5.24 (m, 1H, Jˆ8.1 Hz), 7.21–7.78
(m, 4H). Anal. Calcd for C15H26NISi: C, 48.00; H, 6.98;
N, 3.73; I, 33.81; Si, 7.48. Found: C, 48.04; H, 7.01; N,
3.74.
1
12endo. H NMR (CDCl3) d 1.40 (d, 3H, Jˆ6.8 Hz), 1.47
Cycloaddition of o-quinodimethanes with dienophiles
(general procedures)
(d, 3H, Jˆ7.1 Hz), 1.84 (ddd, 1H, Jˆ13.4, 12.2, 11.1 Hz),
2.22 (dddd, 1H, Jˆ13.4, 6.3, 3.4, 1.0 Hz), 2.87–2.98 (m,
1H) 3.02–3.10 (ddd, 1H, Jˆ12.2, 4.9, 3.4 Hz), 3.11–7.22
(m, 1H), 7.11–7.20 (m, 4H).
Method A: This method is exemplified by synthesis of
3a,4,9,9a-tetrahydro-3H-naphtho[2,3-C]furan-1-one.
12exo. 1H NMR (CDCl3) d 1.35 (d, 3H, Jˆ7.1 Hz), 1.51 (d,
3H, Jˆ7.1 Hz), 1.92–1.99 (ddd, 1H, Jˆ13.5, 6.2, 3.8 Hz),
2.12 –2.18 (ddd, 1H, Jˆ13.5, 9.0, 5.7 Hz), 2.87–2.98 (m,
1H, signal overlapped to 12endo isomer), 2.78 (ddd, 1H,
Jˆ9.0, 8.0, 3.8 Hz), 3.11–7.22 (m, 1H, signal overlapped to
12endo isomer), 7.15–7.24 (m, 4H).
(3ar, 9ac)-3a,4,9,9a-Tetrahydro-3H-naphtho[2,3-C]furan-
1-one (10). To a deoxygenated solution of 5 (100 mg,
0.27 mmol) and 2-(5H)-furanone (260 mg, 3.10 mmol)
solid KF (100 mg, 1.73 mmol) was added. The resulting
suspension was boiled for 5 h. The reaction mixture was
cooled, diluted with ether and the insoluble material was
filtered off.
12endoϩ12exo. IR (nujol) 2236, 750 cmϪ1; Anal. Calcd for
C13H15N: C, 84.28, H, 8.16, N, 7.56. Found: C, 84.35; H,
8.19, N, 7.52.
The filtrate was concentrated, and purified by column
chromatography on silica gel (eluentˆcyclohexane–ethyl
acetateˆ6:4) to give 10 (28 mg, 54% yield) as a white
crystalline solid: mp 135^1ЊC; IR (nujol) 1756, 1151,
2,4,9-Trimethyl-(3ar,4c,9c,9ac)-3a,4,9,9a-tetrahydro-benzo-
[f]isoindole-1,3-dione (13endo). The reaction gave only the
endo adduct as a white crystalline solid (86% yield) purified
by column chromatography on silica gel (eluentˆcyclo-
hexane–ethyl acetateˆ7:3). Mp 154^1ЊC; IR (nujol)
1019, 755 cmϪ1 1H NMR (CDCl3) d 2.59 (dd, 1H,
;
Jˆ14.4, 5.8 Hz), 2.86 (dd, 1H, Jˆ14.4, 5.8 Hz), 2.93–
3.14 (m, 4H, Jˆ3.0, 15.0 Hz), 3.86 (dd, 1H, Jˆ2.9,
8.6 Hz), 4.47 (t, 1H, Jˆ8.6 Hz), 7.09–7.21 (m, 4H). Anal.
Calcd for C12H12O2: C, 76.57, H, 6.43; O, 17.00. Found: C,
76.48, H, 6.41.
1
1698, 1283, 755 cmϪ1.; H NMR (CDCl3) d 1.67 (d, 6H,
Jˆ7 Hz), 2.47 (s, 3H), 2.91–3.02 (m, 2H), 3.10–3.15 (m,
2H), 7.11–7.20 (m, 4H); 13C NMR (CDCl3) 14.1, 24.1, 32.6,
45.0, 123.6, 126.8, 138.3, 177.1. Anal. Calcd for
C15H17NO2: C, 74.05, H, 7.04, N, 5.76; O, 13.15. Found:
C, 74.05; H, 7.04, N, 5.76.
Method B: This method is exemplified by the synthesis of
2-methyl-3a,4,9,9a-tetrahydro-benzo[f]isoindole-1,3-dione.
2-Methyl-(3ar,9ac)-3a,4,9,9a-tetrahydro-benzo[f]iso-
indole-1,3-dione (11). To a degassed solution of 5 (100 mg,
0.27 mmol) and N-methylmaleimide (40 mg, 0.36 mmol) in
1 ml of acetonitrile was added a degassed solution of TBAF
(91 mg, 0.29 mmol) in 5 ml of acetonitrile over 40 min at
room temperature. The resulting pink solution was stirred
for 1 h, the reaction mixture was diluted with ether and the
insoluble materials were filtered off.
2-Cyano-4,5-benzobicyclo[2.2.1]eptane (16endo and
16exo). According to the Procedure B, 16endo and 16exo
were prepared in low yield (12 mg, 14%) starting from 7
(200 mg, 0.53 mmol). endo–exo Ratio (77:23) was evalu-
1
ated by H NMR on the reaction mixture. Adducts 16endo
and 16exo were isolated by column chromatography on
silica gel (eluentˆcyclohexane–ethylacetateˆ9:1), both as
colorless oil.
The filtrate was concentrated, and column chromatography
on silica gel of the residue gave 41 mg (69%) (eluentˆ
cyclohexane–ethylacetateˆ7:3) of 11 as a white crystalline
1
16endo. IR (nujol) 2246, 742 cmϪ1. H NMR (CDCl3) d
1.46 (ddd, 1H, Jˆ12.0, 4.1, 2.7 Hz), 1.61 (dt, 1H, Jˆ9.5,
1.5 Hz), 1.93 (ddd, 1H, Jˆ9.5, 4.5, 2.7 Hz), 2.39 (ddd, 1H,
Jˆ12.0, 10.5, 3.9 Hz), 3.12 (ddd, 1H, Jˆ10.5, 4.1, 3.9 Hz),
3.48 (m, 1H) 3.64 (m, 1H), 7.13–7.38 (m, 4H). Anal. Calcd
for C12H11N: C, 85.19, H, 6.52, N, 8.28. Found: C, 85.30; H,
6.50, N, 8.24.
1
solid: mp 138^1ЊC; IR (nujol) 1693, 1281, 769 cmϪ1; H
NMR (CDCl3) d 2.77 (s, 3H), 2.92 (dd, 2H, Jˆ3.0, 15 Hz),
3.12 (dd, 2H, Jˆ3.0, 15.0 Hz), 3.28 (t, 2 H, Jˆ3.0 Hz),
7.11–7.20 (m, 4H). Anal. Calcd for C13H13NO2: C, 72.54,
H, 6.09, N, 6.51; O, 14.87. Found: C, 72.59; H, 6.10, N,
6.55.
16exo: IR (nujol) 2241, 735 cmϪ1. 1H NMR (CDCl3) d 1.43
(ddd, 1H, Jˆ12.0, 10.5, 2.5 Hz), 1.59 (dt, 1H, Jˆ9.5,
1,4-Dimethyl-2-cyano-1,2,3,4-tetrahydronaphthalene

2558
C. Di Valentin et al. / Tetrahedron 56 (2000) 2547–2559
1.5 Hz), 1.93 (ddd, 1H, Jˆ9.5, 4.5, 2.5 Hz), 2.73 (dd, 1H,
Jˆ12.0, 3.9 Hz), 2.94 (m, 1H), 3.40 (m, 1H) 3.64 (m, 1H),
7.10–7.28 (m, 4H). Anal. Calcd for C12H11N: C, 85.17, H,
6.55, N, 8.28. Found: C, 85.28; H, 6.47, N, 8.20.
the Procedure A, a single endo adduct was recovered as a
white crystalline solid (14 mg, 17%), which was purified by
column chromatography on silica gel (eluentˆcyclohex-
ane–ethyl acetateˆ7:3). Mp 130^2ЊC; IR (nujol) 1752
1
1175, 765 cmϪ1; H NMR (CDCl3) d 1.49–1.58 (m, 2H),
2-Methyl-(3ac,9ac)-3a,4,9,9a-tetrahydro-4r,9c-methano-
benzo[f]isoindole-1,3-dione (15endo). Starting from 7
(150 mg, 0.39 mmol) and N–Me maleimide (66.5 mg,
0.59 mmol) according to Method B, 15endo was obtained
as a single adduct in white crystals (47 mg, 69% yield) from
cyclohexane–ethylacetateˆ7:3; mp 155^1ЊC; IR (nujol)
1.72–1.91 (m., 2H), 2.94–3.01 (m, 2H), 3.01–3.06 (m, 1H),
3.48–3.51 (m, 1H), 3.58–3.63 (m, 1H), 4.28–4.33 (m, 1H),
7.18–7.28 (m, 4H). Anal. Calcd for C14H14O2: C, 78.48, H,
6.59, O, 14.93. Found: C, 78.33; H, 6.63.
1
1698, 1283, 755 cmϪ1.; H NMR (CDCl3) d 1.91 (dt, 1H
Acknowledgements
Jˆ9.5, 1.5 Hz), 2.10 (dt, 1H, Jˆ9.5, 1.5 Hz), 2.28 (s, 3H),
3.47 (m, 2H), 3.82 (m, 2H), 7.08–7.19 (m, 4H); 13C NMR
(CDCl3) 23.5, 46.1, 47.4, 122.6, 126.9, 141.8, 176.8. Anal.
Calcd for C14H13NO2: C, 73.99, H, 5.77, N, 6.16; O, 14.08.
Found: C, 74.07; H, 5.71, N, 6.18.
Financial support from MURST and CNR is gratefully
acknowledged.
References
2-Methyl-(3ac,9ac)-3a,4,9,9a-tetrahydro-4r,9c-ethano-
benzo[f]isoindole-1,3-dione (17endo). Starting from 8
(125 mg, 0.32 mmol) and N–Me maleimide (71 mg,
0.64 mmol) according to Method B, 17endo was obtained
as a single adduct in white crystals (52 mg, 67% yield) from
cyclohexane–ethylacetateˆ7:3; mp 195^1ЊC; IR (nujol)
1. Charlton, J. L.; Alauddin, M. M. Tetrahedron 1987, 43, 2873.
2. Mann, J.; Piper, S. E.; Yeung, L. K. P. J. Chem. Soc., Perkin
Trans. 1 1984, 2081.
3. Kametani, T.; Fukumoto, K. Med. Res. Rev. 1981, 1, 23.
4. Quinkert, G.; Stark, H. Angew. Chem., Int. Ed. Engl. 1983, 22,
637.
1
1695, 1283 cmϪ1; H NMR (C6D6) d 1.08 (m, 2H), 1.25
(m, 2H), 2.25 (s, 3H), 2.38 (m, 2H), 3.30 (m, 2H), 6.96–
7.01 (m, 4H); 13C NMR (C6D6) 24.1, 24.7, 45.0, 125.5,
127.6, 139.2, 177.6. Anal. Calcd for C15H15NO2: C, 74.67,
H, 6.27, N, 5.81; O, 13.26. Found: C, 74.59; H, 6.23, N, 5.84.
5. Kametani, T.; Nemoto, H. Tetrahedron 1981, 37, 3.
6. Nemoto, H.; Fukumoto, K. Tetrahedron 1998, 54, 5425.
´
7. Segura, J. L.; Martın, N. Chem. Rev. 1999, 99, 3199 and refer-
ences therein.
8. Kametani, T.; Honda, T.; Ebisawa, Y.; Ichikawa, H. Tetrahedron
1985, 41, 3643.
9. Durst, T.; Kozma, E. C.; Charlton, J. C. J. Org. Chem. 1985, 50,
4829.
10. Ito, Y.; Nakatsuka, M.; Saegusa, T. J. Am. Chem. Soc. 1980,
102, 863.
11. Ginsberg, D. Tetrahedron 1983, 39, 2995.
12. Charlton, J. L.; Alauddin, M. M.; Penner, G. H. Can. J. Chem.
1986, 64, 793.
13. Wagner, P. J.; Sobczak, M.; Park, B.-S. J. Am. Chem. Soc.
1998, 120, 2488.
14. Arnold, B. J.; Mellows, S. M.; Sammes, P. G.; Wallace, T. W.
J. Chem. Soc., Perkin Trans. 1 1974, 401.
4,9-Dimethyl-(3ar,9ac)-3a,4,9,9a-tetrahydro-3H-naphtho-
[2,3-C]furan-1-one (14endo and 14exo). Starting from 6
(100 mg, 0.26 mmol), 2-(5H)-furanone (231 mg, 2.75
mmol), KF (100 mg, 1.72 mmol) in 3 ml of acetonitrile
according to the Procedure A, two adducts (18 mg, 33%)
were recovered as white crystalline solids, 14endo (94%)
and 14exo (6%) purified by column chromatography on
silica gel (eluentˆcyclohexane–ethyl acetateˆ6:4).
14endo. Mp 146^1ЊC; IR (nujol) 1757, 1170, 759 cmϪ1; 1H
NMR (C6D6) d 1.39 (d, 3H, Jˆ7.1 Hz), 1.71 (d, 3H,
Jˆ6.8 Hz), 2.95–3.23 (m, 4H), 3.48 (dd, 1H, Jˆ9.3,
1
6.4 Hz), 4.15 (t, 1H, Jˆ9.3 Hz), 7.19–7.30 (m, 4H). H
15. Bovio, B. Personal communication.
NMR (CDCl3) d 0.82 (d, 3H, Jˆ7.1 Hz), 1.68 (d, 3H,
Jˆ6.8 Hz), 2.15–2.25 (m, 2H), 2.32–2.43 (m, 2H), 3.05
(dd, 1H, Jˆ9.3, 6.4 Hz), 3.45 (t, 1H, Jˆ9.3 Hz), 7.05–
7.15 (m, 4H). Anal. Calcd for C14H16O2: C, 77.75, H,
7.46, O, 14.80. Found: C, 77.68; H, 7.49.
16. Becke, A. D. J. Chem. Phys. 1993, 98, 1372.
17. Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785.
18. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.;
Petersson, G. A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham,
M. A.; Zakrzewski, V. G.; Ortiz, J. V.; Foresman, J. B.;
Cioslowski, J.; Stefanov, B. B.; Nanayakkara, A.; Challacombe,
M.; Peng, C. Y.; Ayala, P. Y.; Chen, W.; Wong, M. W.; Andres,
J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.; Fox, D. J.;
Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; Head-
Gordon, M.; Gonzalez, C.; Pople, J. A. Gaussian 94, Revision
E.3; Gaussian, Inc.: Pittsburgh, PA, 1995.
14exo. Mp 158^1ЊC; IR (nujol) 1752, 1177, 746 cmϪ1; 1H
NMR (C6D6) d 1.38 (d, 3H, Jˆ7.1 Hz), 1.70 (d, 3H,
Jˆ6.8 Hz), 2.88–3.15 (m, 3H), 3.72 (dd, 1H, Jˆ9.3,
1
6.6 Hz), 4.42 (t, 1H, Jˆ9.3 Hz), 7.08–7.25 (m, 4H). H
NMR (CDCl3) d 0.80 (d, 3H, Jˆ7.1 Hz), 1.55 (d, 3H,
Jˆ6.8 Hz), 2.15–2.30 (m, 2H), 2.40–2.48 (m, 2H), 3.15
(dd, 1H, Jˆ9.3, 6.4 Hz), 3.62 (t, 1H, Jˆ9.3 Hz), 7.05–
7.15 (m, 4H). Anal. Calcd for C14H16O2: C, 77.75, H,
7.46, O, 14.80. Found: C, 77.70; H, 7.40.
19. Tomasi, J.; Persico, M. Chem. Rev. 1997, 94, 2027.
20. Jorgensen, W. L.; Lim, D.; Blake, J. F. J. Am. Chem. Soc.
1993, 115, 2936.
21. Branchadell, L. Int. J. Quant. Chem. 1997, 61, 381.
22. Chen, J. S.; Houk, K. N.; Foote, C. S. J. Am. Chem. Soc. 1998,
120, 12302.
23. McCarrick, M. A.; Wu, Y.; Houk, K. N. J. Org. Chem. 1993,
58, 3330.
(3ac,9ac)-3a,4,9,9a-tetrahydro-4r,9c-ethano-3H-naphtho-
[2,3-C]furan-1-one (19endo). Starting from 8 (150 mg,
0.38 mmol), 2-(5H)-furanone (323 mg, 3.84 mmol), KF
(150 mg, 2.59 mmol) in 3 ml of acetonitrile according to

C. Di Valentin et al. / Tetrahedron 56 (2000) 2547–2559
2559
24. Di Valentin, C.; Rastelli, A.; Gandolfi, R. In preparation.
25. Beno, B. R.; Houk, K. N.; Singleton, D. A. J. Am. Chem. Soc.
1996, 118, 9984.
28. Karcher, T.; Sicking, W.; Sauer, J.; Sustamann, R. Tetrahedron
Lett. 1992, 33, 8027.
29. Sustamann, R.; Sicking, W. Tetrahedron 1992, 48, 10293.
`
`
`
26. Freccero, M.; Gandolfi, R.; Sarzi-Amade, M. In preparation.
30. Ruiz-Lopez, M. F.; Assfeld, X.; Garcıa, J. I.; Mayoral, J. A.;
Salvatella, L. J. Am. Chem. Soc. 1993, 115, 8780.
`
27. Cativiela, C.; Garcıa, J. I.; Mayoral, J. A.; Salvatella, L. Chem.
Soc. Rev. 1996, 209.
31. Brewster, J. H.; Buta, J. G. J. Am. Chem. Soc. 1966, 88, 2233.