Stereospecific nucleophilic trapping of encounter complexes between photoexcited 1-cyanonaphthalene and norbornadiene

Article abstract of DOI:10.1002/(SICI)1099-1395(199802)11:2<101::AID-POC978>3.0.CO;2-G

Photo-induced electron donor-acceptor reactions between 1-cyanonaphthalene (CNN) and norbornadiene (N) generate products of several structure types. Methanol adducts (1-3) formed in polar solvents are rationalized via the radical cation, N^+., and stereospecific (exo-) nucleophilic attack by methanol. In less polar solvents, CNN and N form [2 + 2]-cycloadducts, exclusively on the exo-face of N. In non-polar solvents containing methanol, CNN, N and methanol combine to form 1:1:1 adducts, containing the sensitizer on the endo- and the methoxy groups on the exo-face. The formation of these products is rationalized via the trapping of encounter complexes of different geometries. Any rearrangement of the norbornenyl system can be eliminated, since neither tricyclyl nor 7-methoxynorbornenyl structures are formed. Apparently, the alcohol captures an endo-encounter complex of CNN and N by attack from the exo-face, similar to the attack of methanol on N^+..

Full text of DOI:10.1002/(SICI)1099-1395(199802)11:2<101::AID-POC978>3.0.CO;2-G

JOURNAL OF PHYSICAL ORGANIC CHEMISTRY, VOL. 11, 101–108 (1998)
Stereospeci®c nucleophilic trapping of encounter
complexes between photoexcited 1-cyanonaphthalene
and norbornadiene
Hengxin Weng and Heinz D. Roth
Department of Chemistry, Rutgers University, New Brunswick, New Jersey 08855-0939, USA
Received 25 March 1997; revised 31 July 1997; accepted 31 July 1997
ABSTRACT: Photo-induced electron donor–acceptor reactions between 1-cyanonaphthalene (CNN) and norborn-
adiene (N) generate products of several structure types. Methanol adducts (1–3) formed in polar solvents are
‡
Á, and stereospecific (exo-) nucleophilic attack by methanol. In less polar
rationalized via the radical cation, N
solvents, CNN and N form [2 + 2]-cycloadducts, exclusively on the exo-face of N. In non-polar solvents containing
methanol, CNN, N and methanol combine to form 1:1:1 adducts, containing the sensitizer on the endo- and the
methoxy groups on the exo-face. The formation of these products is rationalized via the trapping of encounter
complexes of different geometries. Any rearrangement of the norbornenyl system can be eliminated, since neither
tricyclyl nor 7-methoxynorbornenyl structures are formed. Apparently, the alcohol captures an endo-encounter
‡
Á.
complex of CNN and N by attack from the exo-face, similar to the attack of methanol on N
1998 John Wiley &
Sons, Ltd.
KEYWORDS: encounter complexes; 1-cyanonaphthalene; norbornadiene; stereospecific; nucleophilic trapping
INTRODUCTION
attention for the past two decades. Both systems contain
two identical groups, either ethene units or cyclopropane
moieties, held rigidly in orientations allowing the study
of through-space or through-bond interactions, respec-
tively.7 The valence isomers have also received attention
because of their potential for the storage of solar energy8
or as the basis of an optical memory system.9 The
interconversion may proceed via radical cations or via an
exciplex. Recently, we reported results of the electron
donor–acceptor photochemistry of N and Q with 1,4-
dicyanobenzene (DCB) in polar media. The reaction
products were compatible with the intermediacy of
radical cations, N and Q , nucleophilic capture from
the exo-face and rearrangements of and hydrogen
abstraction by the resulting free radicals.10 Our study
showed substantial disagreement with an earlier report
describing the donor–acceptor photochemistry of 1-
cyanonaphthalene (CNN) with N or Q in methanol.11
We observed much lower yields of methanol adducts, 1–
3, and isolated a different product type, containing both
methoxy and cyanophenyl substituents. The obvious
discrepancy led us to reinvestigate the donor–acceptor
photochemistry of CNN with N in alcoholic solvents.
Light-induced interactions of electron donor–acceptor
systems have been at the forefront of mechanistic
photochemistry for over two decades.1–3 The nature or
structure of the reaction intermediates has been the focus
of particular attention; depending on the energetics of
electron transfer between donor and acceptor, the
reaction may proceed via an excited-state complex
(exciplex)1,2 or a pair of radical ions.2,3 Typically, non-
polar media favor the formation of exciplexes, whereas
radical ion pairs are generated preferentially in polar
solvents. The involvement of exciplexes is typically
recognized by the observation of characteristic broad,
red-shifted (exciplex) emission or by highly stereospe-
cific cycloadditions. Radical ions, on the other hand, can
be observed and characterized by a variety of spectro-
scopic techniques (viz. ESR,4 CIDNP5 and ODMR)6;
alternatively, their involvement can be inferred from
characteristic reactions, such as rearrangements, nucleo-
philic capture or deprotonation.
‡
‡
Á
Á
The valence isomers norbornadiene (N) and quad-
ricyclane (Q) have been the target of considerable
*Correspondence to: H. D. Roth, Department of Chemistry, Rutgers
University, New Brunswick, New Jersey 08855-0939, USA. E-mail:
roth@rutchem.rutgers.edu.
Contract/grant sponsor: National Science Foundation; contract grant
number: CHE 9110487; contract grant number: CHE 9414271;
contract grant number: CHE 9107839.
1998 John Wiley & Sons, Ltd.
CCC 0894–3230/98/020101–08 $17.50

102
H. WENG AND H. D. ROTH
The product distribution with CNN is substantially
5 mm NMR tubes and preparative runs in water-cooled
30 mm i.d. tubes. The reactions were monitored by gas
chromatography on a GC–MS system (HP 5890 Series II
Plus GC with a HP 5972 mass-selective detector) on a 30
m  0.25 mm i.d.  0.25 mm film thickness HP-5 capil-
lary column (cross-linked methylsilicone on fused silica).
different from that obtained with DCB;10 CNN and N
were depleted and formed (at least) four [2 ‡ 2]-
cycloadducts, 4–7, in substantially higher yields than
obtained for the methanol adducts.12 The discrepancies
with earlier results11 are ascribed to advances in separa-
tion techniques during the intervening years.
Isolation of products. Products were isolated by pre-
parative GLC and column chromatography, preparative
GLC on a 6 ft column packed with 10% CP-5 on
Chromosorb W HP, liquid chromatography on columns
with i.d. ranging from 1 to 5 cm, packed with ca 15 cm of
TLC standard grade silica gel (Aldrich; without binder)
and eluted with solvent gradients, usually from light
petroleum ether (b.p. < 65°C) to mixtures with either
methylene chloride or ethyl acetate. Typically, several
passes were required to isolate the products.
Characterization of products. Structure assignments of
isolated products are based on MS and NMR data. Proton
NMR spectra were recorded on either a Varian XL-400 or
a Varian VXR-200 spectrometer. 13C and HETCOR
spectra were recorded on a Varian VXR-200 spectro-
meter operating at 50.3 MHz. The structural assignments
The different product types require the existence of
divergent mechanisms, featuring radical ions giving rise
to methanol adducts, and loose encounter ‘aggregates’
collapsing to the [2 ‡ 2]-cycloadducts. In order to
elucidate this intriguing mechanistic duality in more
detail, we further varied the reaction conditions. We used
solvents of different polarities, from acetonitrile to
benzene, in the presence and absence of methanol. In
order to obtain optimal mechanistic insight, we carried
out a complete analysis of all products formed in yields
>2%. The advent of routine analysis by GC–MS
techniques and separation by column chromatography
allowed us to perform a detailed examination of the entire
reaction mixture.
1
are based on 1D H, 2D COSY and 13C–1H HETCOR,
where appropriate. Extensive NOE difference spectra
were recorded to elucidate the structure and to probe
substituent stereochemistry and the spatial relationship
between the different functional groups.
RESULTS
Reactions in the presence of methanol
Reaction A. Irradiation of CNN in the presence of N in
methanol produced methanol adducts 1, 2 and 3 in yields
of 4, 3 and 3%, respectively.10 The sensitizer, CNN, was
consumed and two types of molecular adducts were
formed. Adducts containing N and CNN in a ratio of 1:1
were formed in ca 55% combined yield, (exo-[2 ‡ 2]-
adducts 4–7 50%; a meta-addition adduct 8, ca 6%).
Adducts composed of N, CNN and CH3OH in a ratio of
1:1:1 were formed in ca 25% combined yield. The gas
chromatogram in the region of 1:1:1 adducts is not well
resolved. The 1H NMR spectrum of the mixture of 1:1:1
adducts shows six distinct OCH3 singlets (ꢀ 3.2–3.5 ppm)
and an abnormal OCH3 signal at ꢀ 2.7 ppm in a ratio of
1:4:2:2:2:1:2; five products, 9a–11, were isolated and
identified.
EXPERIMENTAL
Materials and solvents. Norbornadiene (Aldrich; 99%)
was distilled after passing it through a short silica gel
column to remove the stabilizing inhibitor (0.05% BHT).
Quadricyclane (Aldrich; 99%) was used as received. 1-
Cyanonaphthalene (Aldrich; 98%) was purified by
column chromatography and recrystallized from n-
hexane. Acetonitrile and methylene chloride (Fischer,
ACS) were distilled from calcium hydride. Methanol
(Fischer, Spectranalyzed) was refluxed over ca 2 glꢀ1 of
sodium and distilled. The solvents so dried were stored
˚
over 4 A molecular sieve in brown bottles under an argon
atmosphere.
Reaction B. Irradiation in acetonitrile–methanol (3:1,
v/v) led to increased yields of methanol adducts (1, 3%;
2, 12%; 3, 12%) but decreased yields of the exo-[2 ‡ 2]-
cycloadducts (4–7, 36% combined yield). Again, the
1:1:1 adducts were formed in significant yields (ca 20%
combined yield). In addition, minor amounts of acetoni-
Photo-reactions. Solutions of 0.1 M donor and 0.1 M
acceptor were purged with argon for 15 min before
irradiation. They were irradiated in a Rayonet photo-
reactor with 16 RPR-3000 lamps, with analytical runs in
1998 John Wiley & Sons, Ltd.
JOURNAL OF PHYSICAL ORGANIC CHEMISTRY, VOL. 11, 101–108 (1998)

1-CYANONAPHTHALENE-NORBORNADIENE ENCOUNTER COMPLEXES
103
Table 1. Product distribution (%) of the donor± acceptor photoreactions between norbornadiene and 1-cyanonaphthalenea
Additional
1:1:1
[2 ‡ 2]-
CH3OH Adducts
N/CNN exo-[2 ‡ 2]-Adducts
meta-Adducts
Adducts Adductsb
Solvent
1
2
3
ꢁc
4
5
6
7
ꢁc
8
ꢁc
ꢁc
ꢁc
CH3OH
4
3
12
1
3
10
27
3
19
17
19
20
29
46
27
19
15
12
17
17
18
29
20
17
10
4
11
17
6
6
15
23
7
3
5
12
4
4
51
36
52
66
57
85
71
79
6
3
7
5
9
3
5
1
ꢁ10
ꢁ5
ꢁ25
ꢁ20
trace
—
trace
—
—
—
7
7
4
4
CH3CN/CH3OH
CH2Cl2/CH3OH
t-BuOH
3
12
1
1
ꢁ20
ꢁ20
ꢁ15
ꢁ5
0.5
1.2
—
—
—
0.1
0.1 <1
CH3CN/t-BuOH
CH3CN
CH2Cl2
7.6
—
—
—
4.4
—
—
—
13
—
—
—
9
20
ꢁ15
ꢁ5
—
—
10
11
C6H6
a
Normalized product distribution according to GC integration.
b Four adducts assigned as [2 ‡ 2]-adducts based on their MS patterns but not isolated.
c
Combined yields of product types of identical composition.
trile adducts and trace amounts of adducts containing N
and CH3OH in a ratio of 2:1 were observed.10 The meta-
adduct (3% yield) was again detected based on its
characteristic GC–MS behavior.
(highest detectable MS peak, m/z 243; base peak, m/z
177), possibly a norbornadiene-substituted cyanonaph-
thalene. The product distributions obtained under the
different reaction conditions are summarized in Table 1.
Reaction C. In methylene chloride–methanol (3:1), the
exo-[2 ‡ 2]-cycloadducts (ca 50% combined yield) were
prominent, whereas methanol adducts 1–3 (ca 1% each)
and 1:1:1 adducts were formed in very low yields. At
least five meta-adducts (ca 20% combined yield),
including 8, were also formed; their MS patterns are
essentially identical with that of 8. Also, at least four
adducts (ca 10% combined yield; M m/z 245, base peak
m/z 91) with MS fragmentation patterns similar to those
of adducts 4–7 were detected, but not isolated.
Reaction of quadricyclane
Reaction G. Irradiation of CNN and Q in methanol
causes rapid isomerization to N, as observed by previous
investigators.11 In addition, minor amounts of methanol
adducts 1–3 are formed in a ratio similar to those
obtained from the reaction of CNN with N. Prolonged
irradiation of this solution resulted in formation of
[2 ‡ 2]-cycloadducts and 1:1:1 adducts as in the CNN–N
reaction, apparently due to a secondary reaction between
CNN* and N.
‡
Á
Reactions in the absence of methanol
Reaction D. In acetonitrile, the exo-[2 ‡ 2]-adducts were
the predominant products (85% combined yield), parti-
cularly 4 (46%) and 5 (29%). Minor products include four
unidentified [2 ‡ 2]-adducts (ca 5% combined yield) and
at least one meta-adduct (ca 3%).
STRUCTURE ASSIGNMENTS
Because of the importance of the correct structure
assignments for the mechanistic conclusions, the spectral
features revealing key elements of the product structures
are briefly discussed below. In addition to H and 13C
1
Reaction E. In methylene chloride, the exo-[2 ‡ 2]-
adducts (4–7) were formed in high yields (27, 20, 15 and
9%, respectively), with decreased regio-preference but
increased yields of the unidentified [2 ‡ 2]-adducts (ca
11% combined yield) as well as the meta-adducts (ca
13% combined yield).
spectra, 2D COSY experiments provided significant
structural details. Extensive NOE difference spectra were
recorded to elucidate substituent stereochemistry and the
spatial relationship between different groups. A detailed
compilation of spectral data is available as supplemental
material.
Reaction F. In benzene, the four exo-[2 ‡ 2]-adducts
were formed in virtually identical yields (4–7, 19, 17, 23
and 20%, respectively); the unidentified [2 ‡ 2]-adducts
(ca 15% combined yield) and traces of meta-adducts (ca
5% combined yield) were also detected.
Cycloaddition products
We isolated five products resulting from cycloaddition
of N to CNN, in yields ranging from 19% (4) to 6%
(8). Four [2 ‡ 2]-cycloadducts (4–7) were described
earlier;12 the minor product (8) apparently resulted from
In all reactions, an apparent dehydrogenation product
of (a) CNN–N adduct(s) was formed in ca 3–5% yield
1998 John Wiley & Sons, Ltd.
JOURNAL OF PHYSICAL ORGANIC CHEMISTRY, VOL. 11, 101–108 (1998)

104
H. WENG AND H. D. ROTH
meta-addition. The mass spectra of all five adducts
times than the cyclobutane derivatives. Their molecular
‡
‡
show molecular ion peaks at
M
m/z 245
ions (M m/z 277, C19H19NO = CNN ‡ N ‡ CH3OH)
[C18H15N = C11H7N(CNN) ‡ C7H8(N)]. In contrast to
indicate that they contain the elements of N, CNN and
methanol in a ratio of 1:1:1. The 1H NMR spectrum of the
mixture shows six distinct OCH3 singlets (ꢀ 3.2–3.5 ppm)
in a ratio of 1:4:2:2:2:1. All six products must contain the
(unrearranged) norbornene function, because of promi-
nent MS peaks corresponding to the loss of cyclopenta-
the [2 ‡ 2]-adducts, which have prominent peaks at m/z
‡
153 (M ꢀ 92, C11H7N; loss of norbornadiene by retro-
‡
cycloaddition), 8 has a base peak of m/z 179 (M ꢀ 66)
corresponding to the loss of cyclopentadiene by retro-
Diels–Alder cleavage. The structures are revealed by the
1H NMR spectra.
‡
diene (m/z 211, M ꢀ 66, C14H13NO). In addition, the
All cycloaddition products show four features char-
acteristic for the 5,6-disubstituted norbornene unit.13
First, two resonances (d,d; ꢀ ca 5.8–6.2 ppm, J ꢂ 6,
3 Hz), sometimes overlapping, are typical for the olefinic
norbornene protons (H2',3'). The identification is con-
firmed by their correlation with the corresponding
bridgehead proton (H1' or H4') in the COSY spectrum.
The orientation of the bridgehead protons (ꢀ ca 3 ppm)
causes simultaneous weak coupling (J < 3 Hz) with
several adjacent protons, resulting in broadened signals.
The bridge protons (H7') appear as an AB system (ꢀ 1.5–
2.0 ppm; 2J ꢂ 9–12 Hz) with only small additional
splittings (J ꢃ 2 Hz). Finally, the resonances of H5' and
H6' show characteristic coupling patterns: J5n–6n ꢂ 8 Hz,
net composition requires that these products contain a
(cyano-)dihydronaphthalene function.
We were able to isolate and characterize five products,
9a–11, which were obtained in yields between 5% (9a)
and 2% (10). Similarly to the [2 ‡ 2]-adducts 4–7, their
NMR spectra show the general features characteristic for
a 5,6-disubstituted norbornene unit.12 The coupling
patterns of the resonances representing H5' and H6'
(J5n–6x = 3–6 Hz, J5n–4 < 1 Hz, J1–6x ꢂ 3 Hz) identify the
adducts as 6-endo-substituted 5-exo-methoxynorbor-
nenes.
Four of the adducts contain four aromatic and two
norbornene protons; therefore, the aromatic ring bearing
the CN function must be the site of addition. One pair of
adducts, 9a and 9b (OCH3 signals at 3.36 and 3.43 ppm,
respectively), contain two additional olefinic protons.
These adducts must be formed by addition at the ring
carbon bearing the cyano group. The similarity of the
spectra of 9a and 9b suggests that they differ in the
stereochemistry of the connection between the norbor-
nene and dihydronaphthalene units. The detailed stereo-
chemistry rests on NOE experiments.
J
5n–4 < 1 Hz, J1–6n < 1 Hz.
Product 8 has four aromatic and only two olefinic
protons, and the familiar norbornene features.11 The
presence of four aromatic protons identifies the CN-
substituted aromatic ring as the site of the addition; the
absence of olefinic protons (other than those of the
norbornene moiety) requires that the two reactants are
linked in a way that converts three unsaturated bonds to
saturated moieties. This change is typically observed for
the meta-cycloaddition of alkenes to benzenoid com-
pounds.14
The key to the structure of 8 lies in the resonances at
2.31 ppm (d, 6.6 Hz), representing a benzylic proton, and
an unresolved multiplet (3.5 ppm; 2H, correlated with the
2.31 ppm proton), which is identified as a cyclopropane
resonance. Although unusually high for cyclopropane
protons, this chemical shift has precedent in a related
meta-cycloadduct;15 in the case discussed here the adja-
cent cyano group causes a significant downfield shift. The
special nature of adduct 8 is also reflected in its MS
fragmentation pattern, which features a base peak at m/z
179 (M ꢀ 66), corresponding to loss of cyclopentadiene.
Another pair of adducts, 10a and 10b [OCH3 at 3.39
and 2.72 (!) ppm, respectively], contain one additional
olefinic proton, suggesting an olefinic bond between C1
and C2 and, accordingly, that the addition occurs in the
position para to the cyano group. The similarity between
the spectra of 10a and 10b suggests that the norbornene
and dihydronaphthalene moieties are linked in stereo-
chemically different fashion. The detailed stereochem-
istry was established by NOE experiments.
The fifth adduct, 11, (OCH3 at 3.35 ppm) has one
strongly deshielded (ꢀ 8.32), seven aromatic (ꢀ 7.8–8.0,
3H; ꢀ 7.5–7.6 ppm, 4H), and two olefinic protons (ꢀ
6.10 ppm). The presence of seven aromatic protons
shows that the naphthalene system is retained; the signal
at 8.32 ppm is characteristic of the a-1H of an aryl-
substituted imine.16 Therefore, this adduct must be
1:1:1 Adducts between N, CNN and methanol
Several adducts showed slightly longer GC retention
1998 John Wiley & Sons, Ltd.
JOURNAL OF PHYSICAL ORGANIC CHEMISTRY, VOL. 11, 101–108 (1998)

1-CYANONAPHTHALENE-NORBORNADIENE ENCOUNTER COMPLEXES
Excitation:
105
formed by addition of N to the C
ꢄ
N function. This type
1
A ! AÃ
ꢅ1†
of addition has precedence.17
Electron transfer:
‡
1Aà ‡ D ! Aꢀ_ ‡ D _
Nucleophilic capture:
ꢅ2†
ꢅ3†
‡
‡
D _ ‡ CH3OH ! CH3O ꢀ D_ ‡ H
Hydrogen abstraction:
CH3O ꢀ D_ ‡ CH3CN ! CH3O ꢀ D ꢀ H ‡ _CH2CN ꢅ4†
Coupling/substitution:
CH3O ꢀ D_ ‡ Aꢀ_ ! CH3O ꢀ D ꢀ Aꢀ
ꢅ5†
Scheme 1.
Á
E
(e.g. R = CH3O), resulting in a variety of pro-
ducts.10,12,19
DISCUSSION
The product distribution with CNN as sensitizer differs
greatly from the DCB-sensitized reaction10 and from the
results reported for the reaction with CNN in methanol.11
The energetics of key reaction steps rule out electron
The driving force for electron transfer [equation (2)]
1
1
from N to CNN* is different from that to DCB*. The
free energy of radical ion pair formation is given by the
excited state energy, E(0,0), the reduction and oxidation
potentials of the reagents, E(Aꢀ/A) and E(D/D‡), and a term
accounting for ion pairing which, in polar solvents, has a
value of ca 0.06 eV:20
1
transfer from N to CNN*. We confirmed the reported
conversion of Q to N as a major reaction, possibly in
competition with nucleophilic capture. This finding is in
striking contrast to the results observed for the photo-
reaction of Q with DCB, which clearly proceeds via
‡
ꢀ
ꢀÁG ˆ E
ꢀ E
‡ E
ꢀ e2="a ꢅ7†
ꢅ0;0†
ꢅD=D †
ꢅA =A†
‡
nucleophilic capture of Q by methanol.10 Similarly,
Á
TR-ESR results have demonstrated that the reaction of
With DCB as sensitizer, electron transfer is exergonic
Q
with methanol is much faster than its valence
(DCB, E0,0 = 4.29 eV, E(Aꢀ/A) = ꢀ1.60 V;21 N,
‡
Á
‡
18
Á
‡Á
isomerization to N . In view of these results, it is
questionable whether Q can be the major intermediate
in the quenching of CNN fluorescence by Q.
E
ꢅD=D
= 1.54 V;22 DG = ꢀ1.0 eV), allowing efficient
†
‡
Á
generation of radical ions. The CNN-sensitized reaction
is only marginally exergonic (E(0,0) = 3.75 eV,23
E(Aꢀ/A) = ꢀ1.98 V;21 DG ꢂ ꢀ0.2 eV); under these
conditions radical ion generation is inefficient. The
relatively low yields of the methanol adducts 1–3
observed in our hands is fully compatible with these
considerations. The significantly higher yields reported
earlier11 are hard to reconcile with the energetics of
The CNN-induced photochemistry of N reported here
is incompatible with the earlier results.11 The obvious
discrepancies are ascribed to advances in separation
techniques during the intervening years. The products
reported earlier were isolated by distillation and GC. It is
unlikely that adducts 4–11 would be isolated under these
conditions.
1
electron transfer from N to CNN*.
The general course of the reaction between DCB and N
was explained via photo-induced electron transfer
[Scheme 1, equation (2)], nucleophilic capture of the
resulting radical cation [equation (3)] and hydrogen
abstraction from acetonitrile [equation (4)] in competi-
tion with aromatic substitution [equation (5)].10 An
additional element of complexity lies in the fact that
the free radicals generated by nucleophilic capture
On the other hand, the reducing pꢀower of CNNꢀ is
Á
Á
greater by ca 0.4 eV than that of DCB ; it is sufficient to
Á
reduce the free radical, CH3—O—B , formed by
‡
Á
nucleophilic capture of N , and the rearranged free
Á
Á
radicals, CH3O—C and CH3O—E . The incorporation
of deuterium into the methanol adducts suggests the
protonation of anions, CH3O—Bꢀ, CH3O—Cꢀ and
CH3O—Eꢀ.10 The mechanism invoked for these inter-
‡
Á
Á
Á
undergo rapid skeletal rearrangements, B to C to
mediates implies the radical cation N
as a key
1998 John Wiley & Sons, Ltd.
JOURNAL OF PHYSICAL ORGANIC CHEMISTRY, VOL. 11, 101–108 (1998)

106
H. WENG AND H. D. ROTH
intermediate in their formation, even if formed with low
efficiency.
separation, which rapidly collapse to products. In polar
solvents, the reagents form more polar, more stable and
more selective aggregates.8,9 The adducts show little
One group of 1:1 adducts between CNN and N, viz. 8,
clearly belongs to a different structure type than the
[2 ‡ 2]-cycloadducts or the 1:1:1 adducts. The shortage
of olefinic resonances suggests the presence of an
additional (cyclopropane) ring; the assigned structure
identifies it as a meta-cycloadduct, a product type often
formed between an aromatic singlet state and alkenes,
and formulated via exciplex intermediates.24 Meta-
cycloadducts may be formed as major products from
aggregates with limited charge-transfer character; the
present system clearly has considerable charge-transfer
character, compatible with the formation of 8 and its
isomers as minor products. The predominant products
formed from CNN and N are [2 ‡ 2]-cycloaddition
products (4–7), suggesting exciplex formation as the
key mechanism.
1
preference for an orientation of N to CNN*. The lower
steric hindrance in anti-isomers suggests a lower barrier
for their formation; this is borne out to some extent by
product ratios, 4:5 and 6:7 (Table 1). However, the
preference for anti-combination is notable only in the
most polar solvent (acetonitrile; 4:5 ꢂ 1.6). The high
specificity of exo-attack on N is the most striking feature
of these reactions. One explanation for the absence of
endo-adducts lies in the formation of 1:1:1 adducts and,
possibly, in some unidentified [2 ‡ 2]-cycloadducts.
1:1:1 Adducts of cyanonaphthalene, methanol
and norbornadiene
Products formed by the combination of a cyano-aromatic
acceptor, an olefinic donor and an alcohol have been
observed in several systems, including CNN, 2,3-
dimethylbut-2-ene, methanol25 or 9-cyanophenanthrene,
2,3-dimethylbut-2-ene, methanol.2b,27,28 Several mech-
anistic pathways have been considered, including (a)
electron transfer–proton transfer–radical coupling,28 (b)
electron transfer– nucleophilic attack–coupling29 or (c)
electron transfer–coupling–nucleophilic attack;25 the
third mechanism is unique as it invokes a zwitterionic
species as the key intermediate.25 The three-dimensional
nature of N and the resulting stereochemistry of three
adducts, 9–11, provides a rigorous test for the proposed
mechanisms. The structures of the 1:1:1 adducts are
incompatible with all three mechanisms.
[2 ‡ 2]-Cycloaddition of CNN to N
The involvement of (an) exciplex(es) in the deactivation
of 1CNN* by N was considered for several reasons. The
photo-reactions of CNN with donor olefins (D) lead to
[2 ‡ 2]-cycloadducts25 and show exciplex emission;23,26
1
these results support (CNN–D)* exciplexes. The reac-
tion between 1CNN* and N shows no such emission; this
rules out exciplexes as well defined intermediates. Still,
the [2 ‡ 2]-adducts are compatible with short-lived
encounter complexes of various geometries. The struc-
tures of the adducts provide insight into the stereo-
chemistry of addition to N (endo or exo), the
regiochemistry of addition to CNN and the relative
orientation of the two fragments (syn or anti). The
addition occurs exclusively on the exo-face of N, on
either ring of the acceptor, with limited preference for
anti- vs syn-addition. In polar solvents, the reaction is
more selective, favoring anti-addition to the substituted
ring.12
1
In the system CNN*–N–methanol, proton transfer
‡
ꢀ
Á
Á
from N to CNN [mechanism (a)] can be eliminated
because deprotonation of the bridgehead is unfavorable.
Nucleophilic attack by methanol on N [mechanism (b)]
is known to generate the methoxy-substituted norborne-
nyl radical, B (R = OCH3), which undergoes a rapid
‡
Á
Á
Á
allylcarbinyl to cyclopropylcarbinyl rearrangement to C
1
(R = OCH3) and and E (R = OCH3).10 Finally, the
Some features of the adducts between CNN* and N
differ from those with other donors. The highly
Á
stereospecific exo-addition to N is reminiscent of the
stereospecific attack of methanol on N .10 Of course, the
‡
Á
stereochemistry of addition to the donor is not an issue
for the commonly used symmetrical substrates (2,3-
dimethylbut-2-ene, cyclopentene). Adducts 4–7 also
show an unusual regiochemistry of addition to CNN.
The donor typically adds at the site of the CN group,
whereas adducts 6 and 7 result from addition to the
unsubstituted ring. The relative yields of 6 and 7 depend
on solvent polarity; in benzene, 4–7 are formed
essentially randomly; the ratio (4 ‡ 5)/(6 ‡ 7) increases
to ca 2 in methanol or methylene chloride–methanol and
to ca 7.5 in acetonitrile (Table 1). This trend reflects the
selectivity–reactivity principle. In non-polar solvents,
1CNN* and N form aggregates with little charge
‡
adduct zwitterion, B [R = CNNꢀ; mechanism(c)] is
expected to undergo an allylcarbinyl to cyclopropylcar-
binyl rearrangement to C (R = CNNꢀ) followed by a
‡
‡
cyclopropylcarbinyl to allylcarbinyl rearrangement to E
(R = CNNꢀ). The five 1:1:1 adducts, 9–11, unambigu-
ously preclude any rearrangement of the norbornenyl
structure (B) to either the nortricyclyl (C) or the 7-
methoxynorbornenyl structure (E). All adducts have the
1998 John Wiley & Sons, Ltd.
JOURNAL OF PHYSICAL ORGANIC CHEMISTRY, VOL. 11, 101–108 (1998)

1-CYANONAPHTHALENE-NORBORNADIENE ENCOUNTER COMPLEXES
107
cation. Formally, such an aggregate is a contact radical
ion pair (CRIP),2 in contrast to the solvent-separated
radical ion pair (SSRIP) accessible by electron transfer to
1DCB*. If the free radicals formed by nucleophilic
capture of the contact radical ion pair add to the sensitizer
anion faster than they rearrange, the difference between
the reactions of 1DCB* and 1CNN* is simply that
between a CRIP and a SSRIP.
The variety of 1:1:1 adducts requires several ‘loose’
complexes with comparable free energies for a range of
geometries. A potential energy surface with broad
minima, rather than a single, well defined minimum,
would allow bonding of the ‘developing’ methoxynor-
Á
bornenyl radical (B , R = OCH3) not only at C-4 or C-2,
but also at the CN function of the CNN system. Coupling
in these positions would give rise to products 9, 10 and
11, respectively.
Scheme 2. Schematic representation of endo-CNN±N
encounter complexes and nucleophilic attack of methanol
on an endo-syn-complex
The charge-transfer induced coupling of an olefin to
the CN function of a cyanoaromatic has precedent in the
photo-reactions of benzonitrile with various olefins.30
The attack of nucleophilic solvents on ‘contact ion pairs’
has been invoked previously to explain structural features
of products resulting from the photoreactions of 2-
phenyl-1-pyrrolinium ion with (prop-2-enyl)-cyclopro-
pane or butadiene in methanol.31
same C7H8 framework; their MS fragmentation patterns
‡
show [Mꢀ66] fragment ions as base peaks, correspond-
ing to loss of cyclopentadiene via retro-Diels–Alder
cleavage. This finding is incompatible with all three
pathways considered above.
Substitution of an exciplex (or an encounter complex)
as the initial intermediate (in place of a radical cation) in
the previously postulated mechanisms does not resolve
the mechanistic problem. Formation of a single C—C
CONCLUSION
‡
bond in such a complex would generate B (R = CNNꢀ)
A series of [2 ‡ 2]-cycloadducts and 1:1:1 adducts
obtained in the photo-reaction of CNN with N in the
presence or absence of methanol are rationalized via
encounter complexes or exciplexes of different geome-
tries. The solvent polarity has a significant role in
determining the course of the reaction, ranging from
essentially random addition in benzene to somewhat
regioselective adduct formation in acetonitrile. The high
preference for exo-attack on N stands in contrast to a less
regiospecific attack on CNN and little preference for
relative orientation of the two reagents in the cycloaddi-
tion. The concept of endo- and exo-exciplexes and their
divergent reactivities poses interesting questions; a more
detailed investigation of these features is desirable, but
goes beyond the scope of the results presented here.
with structural consequences as discussed; proton
transfer is still unfavorable. Clearly, the adducts are
formed by a mechanism which is different in principle.
We suggest that the solution to the mechanistic
problem lies in the stereochemistry of the 1:1:1 adducts.
Products 9–11 have an endo-1-cyanodihydronaphthyl
moiety and an exo-methoxy group in common. The
stereochemistry of the approach of CNN to N is opposite
to that of the [2 ‡ 2]-adducts, whereas the methoxy
groups are introduced analogous to the attack of
‡
10
Á
methanol on N . In the light of these results, we
explain the formation of the 1:1:1 adducts via several
topologically different encounter complexes, with CNN
on the endo-face of N (Scheme 2). While exo-complexes
‘collapse’ to cyclobutane-type adducts (4–7), the endo-
complexes produce 1:1:1 adducts (9–11; Scheme 2), by
nucleophilic attack from the exo-face and formation of an
endo-C—C (or C—N) bond to the cyanoaromatic. This
reaction leads to zwitterionic intermediates which yield
the three-component adducts by deprotonation–protona-
tion. The C—C (C—N) bond formation need not be
concerted with nucleophilic capture; however, the timing
of these steps is important.
Acknowledgement
Financial support of this work by the National Science
Foundation through grants NSF CHE 9110487 and
9414271 and an equipment grant, NSF CHE 9107839,
is gratefully acknowledged.
‡
Á
The fact that methanol attacks the ‘aggregate’ and N
with the same stereochemistry suggests that the endo-
complexes have sufficient charge density on the exo-face
and that their nucleophilicity resembles that of the radical
REFERENCES
1. (a) M. Ottolenghi. Acc. Chem. Res. 6, 153–160 (1973); (b) A.
1998 John Wiley & Sons, Ltd.
JOURNAL OF PHYSICAL ORGANIC CHEMISTRY, VOL. 11, 101–108 (1998)

108
H. WENG AND H. D. ROTH
Ledwith. in The Exciplex, edited by M. Gordon and W. R. Ware,
12. H. Weng and H. D. Roth. Tetrahedron Lett. 37, 4895–4898 (1996).
13. A. P. Marchand. Stereochemical Applications of NMR Studies in
Rigid Bicyclic Systems, Chapt. 4. Verlag Chemie International
Deerfield Beach, FL, (1982).
pp. 209–245. Academic Press, New York, (1975).
2. (a) J. Mattay and M. Vondenhof. Top. Curr. Chem. 159, 219–255
(1991); (b) F. D. Lewis and R. J. DeVoe. Tetrahedron 38, 1069–
1077 (1982); (c) U. C. Yoon and P. S. Mariano. Acc. Chem. Res.
25, 233–240 (1992); (d) I. R. Gould, R. H. Young, L. J. Mueller
and S. Farid. J. Am. Chem. Soc. 116, 8176–8187 (1994).
3. (a) K. Yoshida. Electrooxidation in Organic Chemistry: The Role
of Cation Radicals as Synthetic Intermediates. Wiley, New York
(1984); (b) A. Lund and M. Shiotani (Eds). Radical Ionic Systems.
Kluwer, Dordrecht (1991); (c) H. D. Roth. Top. Curr. Chem. 163,
133–245 (1992); (d) T. Shida, E. Haselbach and T. Bally. Acc.
Chem. Res. 17, 180–186 (1984); (e) S. F. Nelsen. Acc. Chem. Res.
20, 269–275 (1987); (f) G. A. Mirafzal, J. Liu and N. Bauld. J. Am.
Chem. Soc. 115, 6072–6077 (1993).
14. (a) K. E. Wilzbach and L. Kaplan. J. Am. Chem. Soc. 88, 2066–
2067 (1966); (b) D. Bryce-Smith, A. Gilbert and B. H. Orger. J.
Chem. Soc., Chem. Commun. 512–513 (1966); (c) D. Bryce-Smith
and A. Gilbert. Tetrahedron 33, 2459–2489 (1977).
15. M. G. B. Drew, A. Gilbert, P. Heath, A. J. Mitchell and P. W.
Rodwell. J. Chem. Soc., Chem. Commun. 750–751 (1983).
16. E. Pretsch, T. Clerc, J. Seibl and W. Simon (Eds). Tables of
Spectral Data for Structure Determination of Organic Com-
pounds; 2nd ed., p. H180. Springer, Berlin (1989).
17. D. R. Arnold, K. A. McManus and X. Du. Can. J. Chem. 72, 415–
429 (1994).
4. (a) J. N. Aebischer, T. Bally, K. Roth, E. Haselbach, F. Gerson and
X.-Z. Qin. J. Am. Chem. Soc. 111, 7909–7914 (1989); (b) F.
Gerson, X.-Z. Qin, C. Ess and E. Kloster-Jensen. J. Am. Chem. Soc.
111, 6456–6457 (1989); (c) S. Dai, J. T. Wang and F. Williams. J.
Am. Chem. Soc. 112, 2835–2837 (1990); (d) S. Dai, J. T. Wang and
F. Williams. J. Am. Chem. Soc. 112, 2837–2839 (1990); (e) A.
Arnold, U. Burger, F. Gerson, E. Kloster-Jensen and S. P.
Schmidlin. J. Am. Chem. Soc. 115, 4271–4281 (1993).
18. K. Ishiguro, I. V. Khudyakov, P. F. McGarry, N. J. Turro and H. D.
Roth. J. Am. Chem. Soc. 116, 6933–6934 (1994).
19. (a) S. McIlroy, H. Weng and H. D. Roth. Tetrahedron Lett. 36,
7829–7832 (1995); (b) H. D. Roth, X.-M. Du, H. Weng, P. S.
Lakkaraju and C. J. Abelt. J. Am. Chem. Soc. 116, 7744–7752
(1994).
20. H. Knibbe, D. Rehm and A. Weller. Ber. Bunsenges. Phys. Chem.
72, 257–263 (1968).
21. S. L. Mattes and S. Farid. Org. Photochem. 6, 233–326 (1983).
22. P. G. Gassman, R. Yamaguchi and G. F. Koser. J. Org. Chem. 43,
4392–4393 (1978).
23. J. J. McCullough, R. C. Miller, D. Fung and W.-S. Wu. J. Am.
Chem. Soc. 97, 5942–5943 (1975).
24. J. J. McCullough. Chem. Rev. 87, 811–860 (1987); J. Mattay.
Angew. Chem., Int. Ed. Engl. 26, 825–845 (1987).
25. J. J. McCullough, R. C. Miller and W.-S. Wu. Can. J. Chem. 55,
2909–2915 (1977).
26. (a) G. N. Taylor. Chem. Phys. Lett. 10, 355 (1971); (b) W. R. Ware,
D. Watt and J. D. Holmes. J. Am. Chem. Soc. 96, 7853–7860
(1974).
27. (a) K. Mizuno, C. Pac and H. Sakurai. J. Am. Chem. Soc. 96, 2993–
2994 (1974); (b) C. Pac and H. Sakurai. Chem. Lett. 1067–1070
(1976).
5. H. D. Roth. Acc. Chem. Res. 20, 343–350 (1987).
6. (a) X.-Z. Qin and A. D. Trifunac. J. Am. Chem. Soc. 112, 4565–
4567 (1990); (b) X.-Z. Qin and A. D. Trifunac. J. Am. Chem. Soc.
113, 669–670 (1991).
7. (a) P. Bischof, J. A. Hashmall, E. Heilbronner and V. Hornung.
Helv. Chim. Acta 52, 1745–1749 (1969); (b) R. Hoffmann, E.
Heilbronner and R. Gleiter. J. Am. Chem. Soc. 92, 706–707 (1970);
(c) M. J. S. Dewar and J. S. Wasson. J. Am. Chem. Soc. 92, 3506–
3508 (1970); (d) R. Hoffmann. Acc. Chem. Res. 4, 1–9 (1971); (e)
E. Heilbronner and H. D. Martin. Helv. Chim. Acta 55, 1490–1502
(1972); (f) E. Heilbronner. Isr. J. Chem. 10, 143–156 (1972); (g) E.
Heilbronner and A. Schmelzer. Helv. Chim. Acta 58, 936–967
(1975); (h) E. Haselbach, T. Bally, Z. Lanyiova and P. Baertschi.
Helv. Chim. Acta 62, 583–592 (1979).
8. (a) H. Hogeveen and H. C. Volger. J. Am. Chem. Soc. 89, 2486–
2488 (1967); (b) P. G. Gassman, D. H. Aue and D. S. Patton. J. Am.
Chem. Soc. 90, 7271–7276 (1968); (c) K. B. Wiberg and H. A.
Connor. J. Am. Chem. Soc. 98, 5411–5412 (1976); (d) P. G.
Gassman and J. W. Hershberger. J. Org. Chem. 52, 1337–1339
(1987); (e) V. A. Bren, A. D. Dubonsov, V. I. Minkin and V. A.
Chernoivanov. Russ. Chem. Rev. 60, 451–469 (1991).
28. F. D. Lewis, R. J. DeVoe and D. B. MacBlane. J. Org. Chem. 47,
1392–1397 (1982).
29. (a) D. R. Arnold and M. S. Snow. Can. J. Chem. 66, 3012–3026
(1988); (b) D. R. Arnold and X. Du. J. Am. Chem. Soc. 111, 7666–
7667 (1989).
30. T. S. Cantrell. J. Org. Chem. 42, 4238–4245 (1977).
31. J. L. Stavinoha and P. S. Mariano. J. Am. Chem. Soc. 103, 3136–
3148 (1981).
9. S. Miki, Y. Asako and Z. Yoshida. Chem. Lett. 195–198 (1987).
10. H. Weng and H. D. Roth. J. Org. Chem. 60, 4136–4145 (1995).
11. P. G. Gassman and K. D. Olson. Tetrahedron Lett. 1, 19–22 (1983).
1998 John Wiley & Sons, Ltd.
JOURNAL OF PHYSICAL ORGANIC CHEMISTRY, VOL. 11, 101–108 (1998)