Photoinduced electron transfer reactions of aryl olefins. 2. Cis-trans isomerization and cycloadduct formation in anethole-fumaronitrile systems in polar solvents

Article abstract of DOI:10.1021/ja9511578

In acetonitrile, the photoreactions of cis-anethole, cA, or trans-anethole, tA, with fumarodinitrile, FN, lead to isomerization of both substrate and quencher and to mixed [2 + 2] cycloaddition. By NMR analysis and by NOE measurements it was shown that th

Full text of DOI:10.1021/ja9511578

1
40
J. Am. Chem. Soc. 1996, 118, 140
Photoinduced Electron Transfer Reactions of Aryl Olefins. 2.†
Cis-Trans Isomerization and Cycloadduct Formation in
Anethole-Fumaronitrile Systems in Polar Solvents
Martin Goez* and Gerd Eckert
Contribution from the Institut f u¨ r Physikalische und Theoretische Chemie, Technische UniVersit a¨ t
Braunschweig, Hans-Sommer-Strasse 10, D-38106 Braunschweig, FRG
X
ReceiVed April 10, 1995
Abstract: In acetonitrile, the photoreactions of cis-anethole, cA, or trans-anethole, tA, with fumarodinitrile, FN,
lead to isomerization of both substrate and quencher and to mixed [2 + 2] cycloaddition. By NMR analysis and by
NOE measurements it was shown that the same four stereoisomers of 1-anisyl-2-methyl-3,4-dicyanocyclobutane are
formed in equal yields regardless of whether the substrate is cA or tA. The configuration of the anethole-derived
moiety in these adducts is always trans, whereas all possible configurations of the cyano groups occur. From Stern-
Volmer experiments it was concluded that the quenching mechanism is electron transfer, not exciplex formation.
Electron-transfer quenching is also the pathway leading to the cycloadducts, as was established by photoinduced
electron-transfer sensitization. These photoreactions give rise to strong nuclear spin polarizations (CIDNP) in the
starting and isomerized forms of both substrate and quencher as well as in the cycloadducts. Radical pairs
•
+
•-
A FN (RP I) consisting of the radical cation of the anethole and the radical anion of fumarodinitrile were
identified as the predominant source of the polarizations. The starting materials are regenerated by back electron
3
3
transfer of singlet pairs; likewise, back electron transfer of triplet pairs RP I to give either triplet anethole, A, or
3
triplet fumarodinitrile, FN, occurs and constitutes the main pathway to isomerization of substrate and quencher.
The cycloadducts are also formed via RP I; a significant participation of free radicals in their generation was ruled
3
out. The stereochemistry of the products and the different ratios of polarization intensities of the isomerized olefin
3
and the cycloadducts can only be explained by the intermediacy of a triplet biradical BR. After intersystem crossing
1
to the singlet state, ring closure of BR competes with biradical scission. The latter process provides an additional
isomerization pathway, which differs from the pathway via triplet olefins by leading only to one-way cis-trans
isomerization of the substrate. 3BR is formed by geminate combination of triplet radical ion pairs RP I; an indirect
3
3
3
3
pathway via back electron transfer of RP I to give A or FN followed by attack to the other olefin, as has been
proposed for similar photocycloadditions, could be excluded unambiguously. Despite the different precursor
multiplicity, the mechanism of the [2 + 2] photocycloaddition of donor and acceptor olefins investigated in this
work is thus identical to the mechanism of the Paterno-B u¨ chi reaction between donor olefins and electron-deficient
carbonyl compounds which we recently reported. The CIDNP experiments at variable quencher concentration revealed
•
+
•-
the presence of an additional radical pair RP II, also containing A but an anion other than FN . It was shown that
RP II results from a biphotonic process. These findings were explained by two-photon ionization of the anethole
and formation of an oligomeric anion of acetonitrile.
The rapid development of organic photochemistry since the
importance still. Photosynthesis and the mechanism of vision
rely on the photophysical and photochemical properties of
molecules containing conjugated double bonds. As a final
example we mention the UV-induced [2 + 2] cycloaddition of
turn of the century is intimately linked with chemical transfor-
mations of olefins that are induced by visible or UV radiation,
1
the gamut of classical reactions ranging from the photodimer-
2
3
7
izations of cinnamic acid and of stilbene to the synthesis of
ascaridol on a technical scale by photooxygenation of R-ter-
pinene.4 Within the broad spectrum of today’s photochemical
standard methods, reactions of olefins again occupy a central
position, two outstanding representatives being the Paterno-
thymin, which may cause irreversible, and often canceroge-
8
neous, damage to DNA.
It is thus not surprising that olefin reactions have received
considerable attention, [2 + 2] cycloadditions in particular
belonging to the most intensively studied processes.9
,10
Only
5
5b,6
B u¨ chi reaction and photocycloadditions of enones to olefins.
in recent years has the intriguing fact been noted that the amount
of charge separation during mixed cycloadditions of donor and
acceptor olefins strongly influences the product distribution:
Yet, from a more general, and less application-oriented,
perspective natural processes involving olefins are of far greater
†
Preceding paper in this series: see ref 17.
Author to whom correspondence should be addressed.
Abstract published in AdVance ACS Abstracts, December 1, 1995.
(6) (a) Corey, E. J.; Bass, J. D.; LeMahieu, R.; Mitra, R. B. J. Am. Chem.
Soc. 1964, 86, 5570-5583. (b) deMayo, P. Acc. Chem. Res. 1971, 4, 41-
47. (c) Crimmins, M. T. Chem. ReV. 1988, 88, 1453-1473. (d) Schuster,
D. I.; Lem, G.; Kaprinidis, N. A. Chem. ReV. 1993, 93, 3-22.
(7) Blackburn, G. M.; Davies, R. J. H. J. Chem. Soc. (C) 1966, 2239-
2244.
*
X
(
(
(
(
1) Roth, H. D. Angew. Chem., Int. Ed. Engl. 1989, 28, 1193-1207.
2) Bertram, J.; K u¨ rsten, R. J. Prakt. Chem. 1895, 51, 316-325.
3) Ciamician, G.; Silber, P. Chem. Ber. 1902, 35, 4128-4131.
4) (a) Schenck, G. O.; Ziegler, K. Naturwissenschaften 1944, 32, 157-
(8) Stryer, L. Biochemistry; Freeman: New York, 1988.
(9) Lewis, F. D. In Photoinduced Electron Transfer. Part C. Organic
Substrates; Fox, M. A., Chanon, M., Eds.; Elsevier: New York, 1988; pp
1-69, and references therein.
1
57. (b) Schenck, G. O. Angew. Chem. 1952, 64, 12-23.
(
5) See, for instance: (a) Jones, G. In Organic Photochemistry, Padwa,
A., Ed.; Marcel Dekker: New York, 1981; Vol. 5, pp 1-122. (b) Demuth,
M.; Mikhail, G. Synthesis 1989, 145-162, and references therein.
(10) Mueller, F.; Mattay, J. Chem. ReV. 1993, 93, 99-117.
0
002-7863/96/1518-0140$12.00/0 © 1996 American Chemical Society

Photoinduced Electron Transfer Reactions of Aryl Olefins
J. Am. Chem. Soc., Vol. 118, No. 1, 1996 141
Often, quite different regioselectivities, stereoselectivities, peri-
selectivities, and adduct yields are observed when only partial
charge transfer occurs, i.e., when exciplexes are involved as
polar intermediates, than under conditions leading to full charge
separation, i.e., formation of radical ions.10 While the reactions
Table 1. Pertinent Photophysical and Photochemical Data of cis-
and trans-Anethole at Room Temperature
S
0
[M-1 s-1
[M-1 s-1]a
a
anethole
τ [ns]
k
q
]
k
M
Φ
iso
b,c
10
8
trans
cis
8.5c
3.0 × 10
3.2 × 10
0.12
0.06
10
7
6.1
4.3 × 10
8.0 × 10
9
,11
of the former category have been extensively investigated,
a
b
Reference 18b, in acetonitrile. This work, in acetonitrile. The
fluorescence decay was found to be monoexponential. c Reference 18b,
in hexane.
much less is known about the mechanisms of reactions falling
into the latter category.9 From sensitization and scavenging
studies as well as from energetic considerations it was concluded
that in polar solvents the [2 + 2] cycloadditions of 1,2,3-
triphenylcyclopropene with electron-deficient olefins such as
fumarodinitrile12 and the conceptually related [4 + 2] cyclo-
are comparable to the above-mentioned examples: full charge
separation is favored thermodynamically, and electron return
of the radical ion pairs to give triplet olefins is also feasible.
We present evidence that radical ion pairs are indeed formed
and that they undergo intersystem crossing and react to triplet
olefins. However, while the latter species provide the main
pathway for geometric isomerization of the starting materials,
cycloaddition does not proceed via this route but by geminate
combination of triplet pairs to give triplet biradicals in a direct
reaction. This is the same mechanism that we have previously
found for the Paterno-B u¨ chi reaction of these anetholes with
1
3
addition of a similar cyclopropene with dicyanoanthracene
proceed by an electron-transfer/triplet mechanism: A radical
ion pair is formed initially; this undergoes back electron transfer
to give an olefin triplet (the cyclopropene triplet in these
systems); attack by this species on the other olefin yields a triplet
biradical which finally leads to the cycloadducts. As the primary
excited species has singlet multiplicity in these reactions,
intersystem crossing occurring in the radical ion pairs is
therefore one of the key steps of this mechanism.
17
quinones in acetonitrile.
It is well known that in a magnetic field the rates of
intersystem crossing in radical pairs are modulated by the
hyperfine interactions with the nuclear spins, which leads to
Results
Photophysics, Quenching Experiments, and Thermody-
namics. The quencher fumarodinitrile, FN, does not display
any measurable UV/vis absorption above 250 nm in the solvent
used in this work, acetonitrile. When FN was added in high
concentration (0.1 M) to a dilute solution (5 × 10 M) of trans-
anethole, tA, a weak new absorption band was observed in the
range of 310-370 nm indicating formation of a ground-state
complex. However, even at 315 nm the absorbance of this
complex did not amount to more than 5% of that of tA; below
nonequilibrium populations of the nuclear spin states in the
singlet and triplet pairs.14 By ensuing chemical reactions, these
nuclear spin polarizations are ultimately transferred to diamag-
netic products, where they manifest themselves as anomalous
-3
NMR line intensities. This effect, chemically induced dynamic
nuclear polarization (CIDNP),14 thus appeared a convenient
means to study such cycloadditions. Other attractive features
of this method include the possibilities: first, to characterize
both the products and the intermediates, the former being
observed directly by high resolution NMR spectroscopy and
the latter indirectly, because the relative polarization intensities
of the different protons contain similar information as the EPR
spectrum of the radicals composing the pairs;15 second, to
determine the electron spin multiplicities of the species preced-
ing the pairs and of the successor species formed via the different
decay channels of the pairs, since the polarization phases depend
3
10 nm it was undetectable and totally obscured by the latter.
Selective excitation of the substrate tA is thus ensured in the
fluorescence studies of this section (λexc ) 303 nm) as well as
in the photo-CIDNP experiments described below (λexc ) 308
nm). A similar result is found with cis-anethole, cA, but in
this case the relative absorbance of the complex is noticeably
higher, about 13% of that of cA at 308 nm for the concentrations
given. However, under typical conditions of our CIDNP
measurements (c(cA) ) 0.02 M, c(FN) e 0.03 M) excitation
of the ground-state complex can also be neglected.
on both; third, to separate in-cage reactions and reactions of
escaping radicals by time-resolved experiments;16 and, finally,
to trace the pathways from the intermediates to the products,
because the polarizations are generated and detected at different
stages of the reaction.
In this work, we therefore apply CIDNP techniques to the
study of [2 + 2] photocycloadditions of donor olefins (cis- and
trans-anethole) with an acceptor olefin (fumarodinitrile) in polar
medium, mainly in acetonitrile. The energetics of these systems
In acetonitrile, both tA and cA possess unstructured fluores-
cence spectra with intensity maxima at 331 and 325 nm,
respectively. Upon addition of FN the fluorescence intensity
decreases, but the shape of the bands remains unchanged.
Formation of emitting secondary species is thus ruled out, and
relative fluorescence quantum yields can be obtained directly
by integrating the fluorescence spectra. Stern-Volmer plots
(
11) (a) Lewis, F. D. Acc. Chem. Res. 1979, 12, 152-158. (b) Yang,
were found to be linear within the range of quencher concentra-
N. C.; Yates, R. L.; Masnovi, J.; Shold, D. M.; Chiang, W. Pure Appl.
Chem. 1979, 51, 173-180. (c) Caldwell, R. A.; Creed, D. Acc. Chem.
Res. 1980, 13, 45-50. (d) Mattes, S. L.; Farid, S. Acc. Chem. Res. 1982,
S
0
tions used (c(FN) e 0.1 M). As the unquenched lifetimes τ
1
of the excited anetholes A are known (see Table 1), the rate
constants kq for quenching of tA and cA by FN can be
1
5, 80-86.
1
1
(
12) Wong, P. C.; Arnold, D. R. Can. J. Chem. 1979, 57, 1037-1049.
determined from these plots; they are also listed in the table.
(13) Brown-Wensley, K. A.; Mattes, S. L.; Farid, S. J. Am. Chem. Soc.
1
8
1
978, 100, 4162-4172.
14) (a) Chemically Induced Magnetic Polarization; Lepley, A. R., Closs,
It has long been known that photodimerization of tA or cA
leads to formation of (R,R,â,â)- and (R,R,R,R)-1,2-dianisyl-3,4-
dimethylcyclobutane, respectively. These self-quenching reac-
(
G. L., Eds.; Wiley: New York, 1973. (b) Richard, C.; Granger, P.
Chemically Induced Dynamic Nuclear and Electron PolarizationssCIDNP
and CIDEP; Springer: Berlin, 1974. (c) Kaptein, R. AdV. Free Rad. Chem.
1
8b
tions occur via intermediate excimers.
Owing to these
1
975, 5, 319-381. (d) Chemically Induced Magnetic Polarization; Muus,
S
processes, τ depends on the concentration of the substrate.
0
L. T., Atkins, P. W., McLauchlan, K. A., Pedersen, J. B., Eds.; D. Reidel:
Dordrecht, 1977. (e) Salikhov, K. M.; Molin, Yu. N.; Sagdeev, R. Z.;
Buchachenko, A. L. Spin Polarization and Magnetic Effects in Radical
Reactions; Elsevier: Amsterdam, 1984.
However, in the experiments of the present work the anethole
-
2
concentrations do not exceed 2 × 10 M. With the rate
(
15) (a) Roth, H. D. in ref 14d, pp 53-61. (b) Roth, H. D.; Manion, M.
L. J. Am. Chem. Soc. 1975, 97, 6886-6888.
16) Closs, G. L.; Miller, R. J. J. Am. Chem. Soc. 1979, 101, 1639-
641; 1981, 103, 3586-3588.
(17) Eckert, G.; Goez, M. J. Am. Chem. Soc. 1994, 116, 11999-12009.
(18) (a) Nozaki, H.; Otani, I.; Noyori, R.; Kawanisi, M. Tetrahedron
1968, 24, 2183-2192. (b) Lewis, F. D.; Kojima, M. J. Am. Chem. Soc.
1988, 110, 8660-8864.
(
1

1
42 J. Am. Chem. Soc., Vol. 118, No. 1, 1996
Goez and Eckert
Table 2. Relative Energies of Intermediates in the Systems
cA/FN, tA/FN, and CNA/tA
alone or by combining CIDNP detection with coherence transfer
or decoupling techniques, this was not possible in our systems
owing to severe overlap of signals. Hence we carried out the
photoreaction on a preparative scale, isolated the products from
the reaction mixture, and identified them separately. Apart from
dimers of the substrate, different cross coupling products were
found. Our main interest will be focussed on the latter in this
study.
g
E [kJ mol-1]
cA/FN
tA/FN
CNA/tA/FN
1
a
E( Q)
510
389
510
384
510
384b
299
261
260
260
1
a
E( M)
1
E( S)
•
+
•-
E(M /S )
•
+
•-
E(M /Q )
267
260
260
260
250
3
d
E( Q)
3
e
f
f
Besides traces of the already mentioned (R,R,â,â)-1,2-
dianisyl-3,4-dimethylcyclobutane stemming from photodimer-
E( M)
355
250
3
164b
E( S)
1
8b
ization via an exciplex
the only other dimer formed in
a
Determined from the turning points of the long-wave flanks in the
b
detectable yield in the system tA/FN was characterized as the
all-trans isomer (R,â,R,â)-1,2-dianisyl-3,4-dimethylcyclobutane,
fluorescence spectra. Van der Donckt, E.; Barthels, M. R.; Delestine,
c
A. J. J. Photochem. 1973, 1, 429-432. From the redox potentials in
+
21
acetonitrile vs SCE, positive values for the couples Φ (M /M) and
D1, which is known to result from cationic dimerization of tA.
-
21
negative values for the couples Φ (M/M ): cA, +1.47 V; tA, +1.39
Its NMR parameters are given in the Experimental Section. As
21
V; FN, -1.36 V (Arnold, D. R.; Wong, P. C. J. Am. Chem. Soc.
+
main products, however, four mixed adducts P1-P4 (M
)
1
979, 101, 1894-1895. Petrovitch, J. P.; Balzer, M. M.; Ort, M. R. J.
27 d
226) were isolated and found to be stereoisomers of 1-anisyl-
2-methyl-3,4-dicyanocyclobutane.
Electrochem. Soc. 1969, 116, 743-749.); CNA, -1.37 V.
Vertical
triplet energy (Mirbach, M. J.; Mirbach, M. F.; Saus, A. J. Photochem.
e
1
982, 18, 391-393). AM1 calculations gave an energy difference
between cis- and trans-anethole of 7 kJ mol in the ground state, and
of 12 kJ mol in the triplet state. Our estimate of the vertical triplet
energy given for cA is based on these results and on the experimental
By their chemical shifts and multiplet patterns, the resonances
of the methyl, the methoxy, and the aromatic protons could be
-1
-
1
2
3
identified trivially. Assignment of H was straightforward by
3
f
value for tA. Vertical triplet energy (Caldwell, R. A.; Ghali, N. I.;
decoupling the methyl protons at the four-membered ring. By
using NOE-difference experiments, we assigned the resonances
of the other protons and determined the stereochemistry. For
molecules of the size of our cycloadducts and in a solvent of
low viscosity (acetonitrile) the extreme narrowing limit is
realized at the NMR frequencies employed in this study. This
leads to positive NOE enhancement factors for protons that are
directly coupled by the dipolar interaction. Relaxation of the
CH3 protons predominantly occurs by dipolar relaxation among
themselves, so only negligible NOE effects result at the methyl
Chien, C.-K.; DeMarco, D.; Smith, L. J. Am. Chem. Soc. 1978, 100,
g
2
857-2863, ref 66). M denotes the substrate (the respective anethole),
Q the quencher FN, and S the auxiliary sensitizer (CNA) in the
experiments with PET sensitization.
constants kM for self-quenching given in Table 1 one arrives at
S
0
a reduction of τ by about 5% in the worst case. Correction
for this small effect was omitted.
Since ground-state complexes play no role in our reactions,
the fact that kq exceeds the rate of diffusion in acetonitrile tends
1
4
1
group when H to H are saturated. In contrast, the dipolar
coupling with the methyl protons provides an important
relaxation mechanism for the cyclobutane protons in their spatial
vicinity. Furthermore, with all four mixed adducts it turned
out to be difficult or even impossible to saturate all the
cyclobutane protons thoroughly enough and with adequate
selectivity, since their signals are broad multiplets and partly
possess very similar resonance frequencies. For these reasons,
we saturated the methyl protons and the ortho protons of the
anisyl group in each case, which was sufficient to fix the
stereochemistry, and besides only a few of the cyclobutane
protons for additional corroboration. The results are listed in
Table 3 together with the structures of P1-P4 that follow from
these experiments.
to rule out desactivation of A by formation of an exciplex
1
δ+
δ-
(
A ‚‚‚FN ), which is also consistent with the absence of
exciplex emission from our systems; instead, it points to
quenching by electron transfer occurring over distances larger
than the collision diameter.19 Photoinduced electron transfer
over distances up to 10.5 Å has also been observed in the
20
reaction of trans-stilbene with FN in acetonitrile. Further
corroboration of electron transfer from 1A to FN as the
predominant quenching mechanism is provided by the CIDNP
results presented below as well as by thermodynamic consid-
erations.
Pertinent thermodynamical data of possible reaction inter-
mediates are compiled in Table 2. Electron transfer quenching
1
0
-1
of A is seen to be strongly exergonic (-∆G > 120 kJ mol )
with both donors, so rapid quenching by this process is in
accordance with expectations. The energetics show that back
2
4
The positive NOEs between H
and H in the cycloadduct
1
3
P1 as well as between H and H in P3 point to a nonplanar
conformation of the cyclobutane skeleton, which moves these
protons nearer to each other. For all four adducts, force-field
computations predict a minimum energy for a cyclobutane ring
•
+
•-
electron transfer of the resulting radical ion pairs A FN is
not necessarily restricted to the singlet state: Formation of triplet
3
anethole A and FN in its ground state is thermodynamically
2
4
3
folded by approximately 35° along the C -C -axis such that
feasible, as is formation of A and FN. The data for the reaction
the anisyl substituent moves out radially. This is the same angle
with the auxiliary sensitizer 9-cyanoanthracene have also been
included in Table 2; these will be discussed in the section on
photoinduced electron transfer sensitization.
Reaction Products. Apart from geometric isomerization of
the substrate A as well as of the quencher, formation of
cycloadducts takes place in the anethole-fumarodinitrile sys-
tems upon irradiation. Knowledge of the structure of these
products is a prerequisite for the interpretation of the CIDNP
experiments described below. While such structural identifica-
tion can often be achieved on the basis of the CIDNP spectra
2
2
as in unsubstituted cyclobutane, which is slightly surprising
because one would expect the voluminous anisyl group to cause
higher Pitzer strain and consequently larger distortions. The
transannular distance between pseudoaxial protons of the folded
four-membered rings is calculated to be 2.65-2.75 Å. The
value obtained for vicinal protons occupying cis positions is
-
6
barely lower, 2.45 Å, so the d -dependence of dipolar
relaxation should only lead to a difference of NOE enhance-
ments in these two cases by a factor of about 2, in accordance
(
19) Hub, W.; Schneider, S.; D o¨ rr, F.; Oxman, J. D.; Lewis, F. D. J.
Am. Chem. Soc. 1984, 106, 701-708.
20) (a) Angel, S. A.; Peters, K. S. J. Phys. Chem. 1989, 93, 713-717.
b) Angel, S. A.; Peters, K. S. J. Phys. Chem. 1991, 95, 3606-3612. (c)
Peters, K. S.; Lee, J. J. Phys. Chem. 1992, 96, 8941-8945.
(21) Lewis, F. D.; Kojima, M. J. Am. Chem. Soc. 1988, 110, 8664-
8670.
(22) (a) Dows, D. A.; Rich, N. J. Chem. Phys. 1967, 47, 333-334. (b)
Stone, J. M. R.; Mills, I. M. Mol. Phys. 1970, 18, 631-682. (c) Miller, F.
A.; Capwell, R. J. Spectrochim. Acta A 1971, 27, 947-956.
(
(

Photoinduced Electron Transfer Reactions of Aryl Olefins
J. Am. Chem. Soc., Vol. 118, No. 1, 1996 143
Table 3. NOE Effects (at 250 MHz Except Where Noted) in the
Four Mixed Adducts P1-P4 and Resulting Stereochemistry
P4 can be rationalized by the interplay of four factors. Two of
these are due to the magnetic anisotropy of the cyano group.
First, a pseudoaxial cyano group induces a small downfield shift
of a transannular pseudoaxial proton (compare for instance H1
in P1 and P4). Second, for a vicinal proton in cis position to
a cyano group a very slight upfield shift is observed, as, for
e
2
example, H in P4 and in P1 shows. Third, pseudoaxial protons
have resonance frequencies that are a litte lower than those of
3
the corresponding pseudoequatorial protons (e.g., H in P3 and
P4), owing to the magnetic anisotropy of the C-C bonds of
the cyclobutane skeleton and the methyl group. Fourth, the ring
current of the anisyl group causes a noticeable diamagnetic shift
4
of a vicinal cis proton (see H in P2 and P1). All these trends
in the chemical shifts are in line with expectations. Despite
the smallness of the effects, their respective magnitudes are very
similar for all four compounds, allowing one to describe the
1
4
chemical shifts of H to H with a set of increments. Lastly, in
most cases the coupling constants extracted from the spectra
agree very well with those computed by the Karplus-Conroy
relation from the molecular geometries predicted by the force-
field calculations.
a
Measured at 400 MHz. b Saturation or detection at the resonance
c
frequency of the ortho protons of the anisyl substituent. Determination
not feasible owing to partial saturation caused by insufficient selectivity
of the irradiation. Superposition of the resonances of H and H did
not permit their separation in the NOE difference spectrum. However,
with the stereochemistry of P1, P2, and P4 known the question which
of the remaining cyclobutane configurations P3 possesses can already
23
d
2
3
NMR spectra of the reaction mixture show the yields of the
four cycloadducts P1-P4 to be practically equal. Interestingly,
with cA as the substrate instead of tA exactly the same mixed
1
4
be decided on the basis of the NOE effects at H and H . The
2
3
assignment of H and H in the table is the only one consistent with
the structure obtained. e Signal overlap did not permit exact determi-
nation of enhancement factors in all cases; hence only semiquantitative
values (++, strong increase; +, small increase; -, small decrease; °
position of irradiation) are given. An denotes the anisyl moiety.
24
adducts are formed. Hence, with both educts the configuration
of the anethole-derived moiety in the cycloadducts is always
trans. In contrast, with regard to the quencher-based moiety
there is obviously a random distribution of all possible
configurations.
Table 4. Structures and NMR Parameters (in Acetonitrile-d
Room Temperature) of the Four Cycloadducts P1-P4
3
at
Photoinduced Electron Transfer Sensitization. As ex-
plained above, our Stern-Volmer experiments pointed to
d
1
formation of radical ion pairs, not exciplexes, from A and FN
as the predominant quenching mechanism. To test whether or
not this pathway leads to the cycloadducts as well, we
synthesized these products also by PET (photoinduced electron
transfer) sensitization. In contrast to a “normal” sensitization
experiment where the excited sensitizer M* is quenched by Q,
2
5
with PET sensitization an auxiliary sensitizer S is excited, S*
is quenched by either M or Q to generate a primary radical ion
pair, and the latter is then intercepted by the third reactant to
give the same radical ion pair as upon direct sensitization but
as a secondary species. For suitable energetic and kinetic
parameters, neither electronically excited M nor electronically
excited Q are touched upon before formation of the secondary
radical ion pair. Mechanistic conclusions can then be drawn
from the fact that in consequence this pair must be produced
δ(
δ- 17
without the intervention of an exciplex M ‚‚‚Q
.
If
formation of the same products is observed in the direct and
the PET-sensitized experiments, such an exciplex is obviously
not a key intermediate for their generation.
For this purpose, 9-cyanoanthracene CNA was employed as
auxiliary sensitizer for the reaction of tA with FN. It can be
a
Center of the multiplet; without simulation. b Exact determination
2
3
impossible owing to the nearly identical chemical shifts of H and H .
Undetectable. Coupling constants calculated by the Karplus-Conroy
relation are given in parentheses. An denotes the anisyl moiety.
c
d
(
23) (a) Karplus, M. J. Chem. Phys. 1959, 30, 11-15. (b) Karplus, M.
with the experimental observations. Finally, for the product
P3 a decrease of the signal of H is observed upon saturation
J. Am. Chem. Soc. 1963, 85, 2870-2871. (c) Sternhell, S. Quart. ReV.
4
1
969, 23, 236-270.
of the methyl group. This can be explained by an indirect NOE
(24) We should like to emphasize that the product distribution observed
1
3
in these experiments does not unambiguously rule out that tA and cA react
to different products: With our reactants cycloaddition is accompanied by
geometric isomerization of the substrate to a large degree, so under
conditions of preparative photolysis the system moves far towards the
photostationary equilibrium between tA and cA regardless of whether the
starting anethole is cA or tA. However, the same product distribution is
also found in the CIDNP experiments described below, which are performed
at very low conversion of the substrate.
(25) (a) Mattes, S. L.; Farid, S. In Organic Photochemistry; Padwa, A.,
Eds.; 1983; Vol. 6, pp 233-326. (b) Chanon, M.; Eberson, L. In
Photoinduced Electron Transfer. Part A. Conceptual Basis; Fox, M. A.,
Chanon, M., Eds.; Elsevier: New York, 1988; pp 409-597. (c) Mattay, J.
Synthesis 1989, 233-252.
(
three-spin effect) via H as well as H .
From the spectra and with the NOE results, approximate
NMR parameters of the cycloadducts were derived, which
served as starting values for simulations of the J-coupled spin
systems consisting of the cyclobutane protons and the methyl
group. By iterative fitting of the simulated spectra to the
experimental spectra, the chemical shifts and coupling constants
displayed in Table 4 were obtained. These data are also
consistent with the structures given. For each cyclobutane
proton, the differences of the chemical shifts in the adducts P1-

1
44 J. Am. Chem. Soc., Vol. 118, No. 1, 1996
Goez and Eckert
dynamically as is formation of this radical pair; owing to the
almost identical reduction potentials of CNA and FN the driving
force of this process is nearly zero, and the rate constant is
estimated to be lower than the diffusion-controlled limit by about
2
8
one order of magnitude. Thus, this reaction should proceed
on a time scale of some 10 ns under the experimental conditions
(
c(FN) ) 0.1 M); still, as this is comparable to the life of the
1
4e
•+
•-
pair a substantial amount of tA FN is expected to result.
The same cycloadducts were found in this experiment with
1
PET sensitization as in the direct reaction of tA with FN, and
their yields were good enough to utilize this method for
preparative generation of P1-P4. These observations clearly
•+
•-
demonstrate that the route via the radical ion pairs tA FN is
at least a major, if not the predominant, pathway to the
cycloadducts. On the other hand, as our quenching studies
furnished evidence that electron transfer quenching of 1
A
prevails over quenching by exciplex formation, exciplexes
preceding the radical ion pairs may also be disregarded. In
1
δ+
δ-
combination, these findings show that (A ‚‚‚FN ) does not
play an important mechanistic role in the cycloadditions of A
and FN.
CIDNP Spectra. As shown in Figure 1, strong CIDNP
signals arise during the photoreactions of tA or cA with FN.
Nuclear spin polarizations are observed for the starting anethole
and its respective isomerization product, for the quencher FN
and its cis isomer maleodinitrile MN, and for the four cyclo-
adducts P1-P4.
Qualitative mechanistic information can be extracted from
these spectra with Kaptein’s rule for CIDNP net effects,30
Γ ) sgna × sgn∆g × µ × ꢀ
(1)
i
i
Figure 1. Background-free pseudo steady-state CIDNP spectra ob-
served in the photoreactions of trans-anethole tA (bottom trace) and
Equation 1 connects the polarization phase (Γi ) +1, absorption,
Γi ) -1, emission) of proton i in a particular reaction product
with magnetic parameters of the intermediate radical pairs (sgnai,
sign of the hyperfine coupling constant of the proton considered;
sgn∆g, sign of the difference of the g values of the two radicals
constituting the pair, where the radical counted first contains
the proton in question) and with the entry and exit channels of
the pairs (µ ) +1, pair formation from triplet precursors; µ )
cis-anethole cA (top trace) with fumarodinitrile FN in acetonitrile-d
3
-
3
at room temperature. Experimental parameters, c(tA) ) 3 × 10 M;
-
2
-3
c(cA) ) 2 × 10 M; c(FN) ) 8 × 10 M; T ) 274 K. Amplitudes
have been multiplied by a factor of 4 in the inset. The assignment of
the resonances of tA refers to the structural formula given at the top;
the same labels are used for the protons of cA. P denotes the
cycloadducts P1-P4, the signals of which coincide partly. For this
reason, only of few of the resonances of P are assigned in detail. These
are denoted as in Table 3. The insufficiently suppressed signal marked
-
1, from singlet precursors; ꢀ ) +1, product resulting from
singlet radical pairs; ꢀ ) -1, from triplet pairs). In addition,
3
1
[AN] is due to the solvent.
0
(
28) As ∆G for interception of tA•+CNA•- by FN is almost zero, the
excited at λexc g 385 nm, where the other reactants as well as
the ground-state complex between tA and FN do not absorb.
The relevant thermodynamical parameters of the system CNA/
tA/FN in acetonitrile have been given in Table 2. With these
rate constant for this process is approximately given by the geometric mean
of the rate constants kex for the self exchange in the systems FN/FN and
CNA/CNA (Kavarnos, G. J.; Turro, N. J. Chem. ReV. 1986, 86, 401-
49). CIDNP measurements show the former rate constant in acetonitrile
•-
•
-
4
9
-1 -1
to be as high as 1 × 10 M
s
even at 235 K (Eckert, G. Ph.D. Thesis,
1
data, it is seen that singlet-singlet energy transfer from CNA
Braunschweig, 1995); at room temperature the structurally very similar
tetracyanoethylene has29 kex ) 2.5 × 10 M
9
-1 -1
s
. An estimate for the
to tA or to FN can be ruled out as it would be much too
•
-
1
self-exchange rate constant of the couple CNA/CNA is provided by the
endergonic. Quenching of CNA by electron transfer from tA
9
-1 -1
value for 9,10-diaminoanthracene, kex ) 2.5 × 10 M
s
in acetonitrile
-
1
is exergonic by about 40 kJ mol , so it is assumed to proceed
at 293 K (Grampp, G.; Jaenicke, W. Ber. Bunsenges. Phys. Chem. 1984,
S
1
27
8
8, 325-334), since kex is comparable for systems of related structure, and
diffusion controlled. With τ of CNA being 17 ns, quantum
0
•
+
•-
anions seem to possess somewhat higher kex than cations (see the examples
yields for formation of the primary radical ion pair tA CNA
9
-1 -1
in ref 29). A rate constant of at least 2 × 10 M
s
thus seems reasonable
should be at least 80% at the concentration of tA used in our
for the interception process. However, even a much smaller value would
not invalidate our conclusions owing to the lack of other desactivation
experiment (0.025 M). Although thermodynamics indicate back
•
+
•-
•
+
•-
3
pathways of tA CNA besides interception by FN and back electron
transfer to give CNA.
electron transfer of the radical ion pairs tA CNA to give A
to be feasible, we found that in the absence of FN these pairs
3
(29) Grampp, G.; Harrer, W.; Jaenicke, W. J. Chem. Soc., Faraday Trans.
3
decay almost exclusively by back electron transfer to CNA,
1 1987, 83, 161-166.
(
(
30) Kaptein, R. J. Chem. Soc., Chem. Commun. 1971, 732-733.
31) According to the usual definition of ꢀ, this parameter is taken to
making the system CNA/tA very photostable. Presumably, this
behavior is due to the very low triplet energy of the auxiliary
sensitizer in comparison to the triplet energy of the substrate.
denote the mode of product formation (ꢀ ) +1, cage reaction; ꢀ ) -1,
reaction of escaping radicals). This interpretation is appealing to chemists
because it is directly linked to mechanistic concepts; very often it is also
permissible, since for the majority of radical pairs cage recombination can
take place in the singlet state only. In reactions involving radical ion pairs,
however, geminate processes of pairs of either multiplicity are frequently
possible, so the assignment of ꢀ (Roth, H. D.; Manion-Schilling, M. L. J.
Am. Chem. Soc. 1980, 102, 4303-4310) adopted here must be used.
•
+
•-
Interception of tA CNA by FN is not as favorable thermo-
(
26) For photoinduced electron transfer sensitization in a similar reaction
system see ref 13.
27) Lewis, F. D.; Bedell, A. M.; Dykstra, R. E.; Elbert, J. E.; Gould, I.
R.; Farid, S. J. Am. Chem. Soc. 1990, 112, 8055-8064.
(

Photoinduced Electron Transfer Reactions of Aryl Olefins
J. Am. Chem. Soc., Vol. 118, No. 1, 1996 145
for a given product the polarization intensity of a group of
equivalent protons i divided by their number, Pi, reflects the
magnitude of ai; if ∆g is large, there is even approximate
proportionality between Pi and ai.1 Hence, the polarization
patterns, i.e., the relative CIDNP intensities of the different
protons in a product, yield information almost equivalent to an
EPR spectrum of the intermediates with the additional benefit
that each hyperfine coupling constant can be related directly to
due to a side-reaction, whereas the products predominantly stem
from nonradical pathways.
By inspection of the two spectra of Figure 1, one can
immediately rule out that the observed isomerization of the
anetholes takes place before, or concomitantly with, CIDNP
generation, because in that case the same CIDNP signals would
be obtained as with a mixture of tA and cA, that is, the same
polarization phases would result for the starting as well as the
isomerized substrate. The unimportance of geometric isomer-
4e
1
5,32
a specific proton.
With regard to the starting anethole, one obtains from Figure
1
ization of A in our measurements also follows from the
1
that PHâ ≈ -PHγ . - POCH > PH3,5 > PH2,6 ≈ -PHR (for the
photophysical data and the results of our quenching studies:
With the values given in Table 1, one calculates a quantum
3
assignment see the figure). These intensity patterns are in good
17
1
agreement with the calculated spin density distributions in the
yield of geometric isomerization of 4% for tA and of 2% for
•
+
•+
1
radical cations tA or cA of the anetholes. Besides,
quantitatively the same patterns are found in the reactions of
tA and cA with quinones, for which it was shown that CIDNP
cA under the experimental conditions of Figure 1, taking into
account the temperature dependence of kq with the Smolu-
3
4
chowski equation and the known viscosities of acetonitrile,
while quantum yields for radical ion pair formation lie above
65% in these two cases. Likewise, the fact that no isomerization
of the radical ions occurs is not surprising as the configurational
stability of anethole radical cations has been verified both
•
+
•+
originates from radical ion pairs comprising tA and cA ,
respectively.
1
7
Together with the results of our quenching
studies, this furnishes proof that the observed polarizations are
•+
•-
•+
generated in radical ion pairs A FN , where A is the
radical cation of the respective anethole.
1
7
9,17,21
theoretically and experimentally.
It is well known that olefin isomerization accompanying
cycloaddition can also take place by bond rotation in a triplet
As can be seen in both spectra of Figure 1, the polarizations
of the isomerized anethole are mirror images of the polarizations
of the starting anethole. This is clear evidence that these two
species result from opposite exit channels of the same inter-
mediate pairs. The polarization patterns of all four cyclo-
adducts are equal to those of the isomerized substrate; in
addition, the phases of the corresponding protons are identical
in these two classes of products (compare, for instance, the
methyl groups in the isomerized anethole and in P1-P4, both
being the most strongly polarized signals of these compounds
and exhibiting enhanced absorption). Cycloadducts and isomer-
ized substrate are thus formed via the same exit channel of the
radical ion pairs.
The fact that the cycloadduct signals are absolutely identical
in the two spectra of Figure 1 convincingly demonstrates that
the reaction via radical ion pairs leads to the same adducts
regardless of the configuration of the starting anethole. From
an analysis of the superposition signal due to the methyl groups
in these four different products, it is further seen that the yields
of P1-P4 are practically equal. This simple evaluation is
permissible because protons are being considered that were
identical in the intermediates, and because the CIDNP spectra
presented are free from the effects of nuclear-spin relaxation
both in the products and the intermediates. The former is a
3
5
9,15,36
biradical or via the olefin triplet.
In a previous study of
1
7
3
quinone-sensitized anethole photoreactions, where A cannot
be formed for energetic reasons, we found the biradical pathway
to lead to efficient one-way cis-trans isomerization of the
starting olefin, paralleling the distribution of configurations of
the anethole moiety in the cycloadducts. In the present instance,
the configuration of the anethole part is trans in all four adducts,
so there is again one-way cis-trans isomerization of the
anethole fragment as far as cycloaddition is concerned, but
isomerization of the substrate proceeds in both directions, as is
evident from Figure 1, bottom trace. Hence, isomerization of
the substrate must be ascribed chiefly to the triplet state of the
anethole, the formation of which is thermodynamically feasible
in our systems (compare Table 2), while a participation of the
biradical pathway is not excluded by these qualitative arguments.
In both cases, however, the key intermediate between the radical
ion pair and the isomerized substrate is a triplet species, making
the parameter ꢀ of Kaptein’s rule negative. With the precursor
multiplicity being singlet, it follows from the absorptive
polarizations of the methyl protons (a > 0) in the isomerized
•
+
•+
anetholes that tA and cA possess larger g values than does
FN .
•
-
Geometric isomerization of the quencher FN manifests itself
consequence of the experimental technique (pseudo steady-state
photo-CIDNP measurements).33 The latter can be concluded
by the CIDNP signals of the vinyl protons of MN observed in
both photoredox systems. The absorptive polarizations of these
protons (a < 0) show that this process again occurs via the
triplet exit channel of the radical ion pairs. This cannot be
explained by free radical ions escaping from triplet pairs,
from the fact that the integral over all CIDNP signals of protons
having a common source (e.g., the methyl protons in the starting
and the isomerized anethole, and those in P1-P4) vanishes,
which was found to hold for all the CIDNP spectra of this work.
Compared to conventional studies of the product distribution,
these CIDNP determinations of relative yields possess the
advantage that they are performed at fairly low conversion of
the starting anethole to its geometric isomer (less than 5% in
our measurements). This is feasible, since the signal enhance-
ment by the CIDNP effect allows one to detect also products
formed in low amounts, hence at low turnover, while maintain-
ing the analytic potential of high-resolution NMR. On the other
hand, the very strong nuclear-spin polarizations and the results
of our fluorescence quenching and PET-sensitization studies,
which bear out formation of radical ion pairs as key intermedi-
ates, exclude the possibility that the CIDNP signals are merely
•
-
37
because FN is configurationally stable. Furthermore, from
the fact that isomerization of the anetholes via the biradical route
is obviously not predominant it also follows that this mechanism
is of secondary importance in the case of FN, since the same
intermediates would be involved in the isomerizations of both
the substrate and the quencher by that pathway. Hence,
isomerization of the quencher must also take place mainly by
(34) Grampp, G.; Jaenicke, W. Ber. Bunsenges. Phys. Chem. 1984, 88,
335-340.
(35) Caldwell, R. A.; Sovocool, G. W.; Gajewski, R. P. J. Am. Chem.
Soc. 1973, 95, 2549-2557.
(
36) Arnold, D. R. AdV. Photochem. 1968, 6, 301-423.
•-
(37) In time-resolved CIDNP studies of the self-exchange between FN
and FN we did not find any evidence for geometric isomerization of the
free radicals FN on a time scale of 10 µs (Eckert, G. Ph.D. Thesis,
Braunschweig, 1995).
•
-
(
(
32) Roth, H. D. Z. Phys. Chem. Neue Folge 1993, 180, 135-158.
33) Goez, M. Chem. Phys. Lett. 1992, 188, 451-456.

1
46 J. Am. Chem. Soc., Vol. 118, No. 1, 1996
Goez and Eckert
Figure 3. Ratios µ
P
/µiso of the polarizations of the cycloadducts and
Figure 2. Dependence of the polarization intensities of the respective
isomerized anethole and the cycloadducts on the concentration (given
at the traces) of the quencher fumarodinitrile. The spectra (range 1.0-
the isomerized substrate as a function of the fumarodinitrile concentra-
39
tion. Filled and open circles, starting anethole tA and cA, respectively.
-3
Experimental parameters: c(tA) ) 3.3 × 10 M, c(cA) ) 1.98 ×
2
.1 ppm) displaying the resonances of the methyl groups in educts and
-2
1
0
M, T ) 273 K.
products have been normalized to equal height of the emission signal
of the starting anethole.
concentrations of the quencher FN. From these spectra it is
obvious that in the reaction of tA with FN an increase of the
quencher concentration leads to a significant increase of these
polarizations in the photoadducts, while the signal that is due
to the isomerized substrate decreases by the same amount.
Although less pronounced, this effect is also observed with cA
as starting material.
the triplet route and population of 3FN, which is nearly
3
isoenergetic with A (Table 2), must provide an additional decay
3
•+
•-
pathway of the pairs A FN .
Finally, the protons of both the anethole- and the fumaro-
dinitrile-derived moieties in the cycloadducts P1-P4 are
polarized, as is seen from the signals of the cyclobutane protons
in the spectral range between 2.5 and 4.0 ppm. The ratio of
their CIDNP intensities agrees well with the ratio of the
hyperfine coupling constants of the respective protons in the
anethole cations and the quencher anions. This confirms that
Quantitative investigation of this concentration dependence
is complicated by partial isomerization of the anethole during
the measurements, which cannot be avoided. Despite the fact
that typically the degree of isomerization does not amount to
more than 5%, this leads to significant changes of relative
CIDNP signal strengths in the system cA/FN because at the
excitation wavelength the extinction coefficient of the trans
isomer accumulated during the experiment is about 10 times
higher than that of cA; quantitative analysis would thus be
impossible without appropriate correction. However, at low
consumption of the substrate, the composition of the sample
changes practically linearly with the number n of laser flashes
already absorbed, and so do the CIDNP intensities. For each
sample we therefore recorded several CIDNP spectra in suc-
cession, hence at different values of n, and then extrapolated
the polarizations to n ) 0. All CIDNP intensities evaluated in
the following were corrected in this way, but with tA as the
substrate this correction was invariably very small. In no case
did we observe significant deviations from linearity.
In Figure 3, the ratios µP/µiso of the polarizations of the
cycloadducts and the isomerized substrate have been plotted as
functions of the quencher concentration for both photoredox
systems. For given net polarization of a proton in a radical
pair, which only depends on the magnetic parameters and the
correlated life of the pair, and in the absence of nuclear spin
relaxation, the polarizations of the corresponding protons in the
products are directly proportional to the product yields. With
a simple kinetic scheme comprising two competing decay
reactions of a single radical pair, one leading to cycloaddition,
the other to isomerization, a constant value of µP/µiso would
thus be expected. This is indeed found at high quencher
concentrations, but the decrease of this ratio with decreasing
concentration of FN shows that the mechanism must be more
complicated.
•
+
•-
the intermediate pairs comprise both A and FN .
In a recently published investigation of the Paterno-B u¨ chi
reaction of anetholes with quinones in polar media,17 we found
that the cycloadducts were formed by geminate combination
of correlated radical ion pairs in triplet states to give triplet
biradicals, whereas back electron transfer of the singlet pairs
regained the educts; a significant participation of free radical
ions could be excluded. The results of the CIDNP experiments
of this section suggest that the same mechanism is realized in
the [2 + 2] photocycloaddition of the present work, with the
modification that in the anethole/fumarodinitrile systems back
electron transfer is also possible for triplet pairs and leads to
bidirectional geometric isomerization of donor and acceptor
2
1
olefin. The low yield of dimers such as D1, which is known
to result from a cationic route, and which can thus be formed
much more efficiently by attack of (long-lived) escaping radical
•
+
•+
cations tA to surplus substrate than by combination of tA
and tA during diffusive excursions of the radical ions (i.e.,
during the much shorter correlated life of the pairs), indicates
that free radicals do not play a significant role also in our
3
8
systems.
This is further corroborated by the absence of
relaxation loss, as well as by the time-resolved CIDNP measure-
ments and scavenging experiments described below.
Concentration Dependence of the CIDNP Intensities.
Figure 2 shows part of the CIDNP spectra of our systems
(resonances of the methyl protons in the starting and isomerized
anethole, and in the cycloadducts P1-P4) at two different
(38) As the spectra of Figure 1 show, no CIDNP signals can be detected
for the dimers. On the one hand, we attribute this to the low yields of
these products, and on the other hand, the primary species resulting from
combination of radical ion and substrate is still a radical ion. The nuclear
spin polarization generated in the radical ion pairs will therefore be largely
destroyed by nuclear spin relaxation during the life of the free paramagnetic
species. Absence of CIDNP in dimer formation has also been reported for
(39) The curves through the data points in Figure 3 as well as Figure 4
were drawn as best fits according to a theoretical model for the concentration
dependence of the CIDNP intensities, which is outside the scope of the
1
5a
65
other systems.
present work and will be published separately.

Photoinduced Electron Transfer Reactions of Aryl Olefins
J. Am. Chem. Soc., Vol. 118, No. 1, 1996 147
Figure 4. Polarizations µ∞ referred to unit quantum yield of radical
ion pair formation in the system tA/FN. Experimental parameters
as in Figure 3.
Figure 5. Dependence of the polarization ratio µ
/µiso on the substrate
P
3
9
concentration in the system tA/FN. Experimental parameters: c(FN)
-
2
) 1.5 × 10 M, T ) 273 K.
•
+
•-
Besides, and most strikingly, small but noticeable polariza-
tions of the starting and the isomerized anethole were also
observed in the absence of FN. The polarization patterns and
CIDNP phases of these two species were the same as in Figure
the free enthalpy of the radical ion pair tA tA lies some 60
kJ mol
-
1
1
below that of tA (compare Tables 1 and 2).
Photoinduced disproportionation of the substrate is thus ther-
modynamically feasible; also, as mentioned above, self-quench-
1
8b
1
. These findings indicate the presence of a parallel isomer-
ing of anetholes is a known reaction.
To test whether the
ization pathway via another radical ion pair besides
isomerization of the substrate occurring without the intervention
of FN relies on this process, we kept the quencher concentration
at a moderate value and varied the anethole concentration. As
Figure 5 shows, the polarization ratio µP/µiso is independent of
•
+
•-
A FN . As suggested by the CIDNP data, this second pair
also comprises the anethole radical cation but a radical anion
•
-
•-
other than FN ; the latter, X , possesses a lower g value than
•
+
•+ •-
does A .
Further information was obtained by evaluating the absolute
c(tA). This definitely rules out the pair tA tA as the
intermediate leading to the polarizations found in the absence
of FN. This is also consistent with expectations, because even
if such pairs were formed it seems very doubtful whether they
could lead to noticeable polarizations in the starting and the
polarizations rather than their ratio. In this instance, however,
an additional correction must be performed because CIDNP
signal strengths are directly proportional to the concentration
of the radical pairs, which in turn is a function of the quencher
concentration. Measured signals were therefore multiplied by
4
0
isomerized anethole.
The most trivial explanation of the polarizations found in the
samples without FN would be an impurity that acted either as
sensitizer or as quencher of the excited anetholes. We could
not detect the presence of an additional light-absorbing substance
in the UV/vis spectra of the substrates nor in the spectrum of
the acetonitrile employed; besides, this would lead to a
dependence of the polarization ratios on the substrate concentra-
tion, which is not found (Figure 5). Owing to the short lifetimes
of excited species, quenching by an impurity would demand
that this were present in quite high concentration; hence it could
only have been contained in the solvent used for the CIDNP
experiments, not in the starting anethole, the purity of which
was at least 99.5% and which was itself employed in millimolar
concentrations only. Since we could not observe it in the NMR
spectra, it would have to be a perdeuterated compound. We
could not find any evidence for the presence of such a species
in our solvent by optical spectroscopy; furthermore, we observed
identical CIDNP spectra of our systems with solvents purchased
from different manufacturers. On these grounds, we are inclined
to rule out the impurity hypothesis.
S
q 0
-1
1
+ [k τ c(FN)] to refer them to unit quantum yield ΦRP for
radical ion pair formation. Changes in the amount of light
absorbed by different samples were minimized by carrying out
all measurements of a series in immediate succession, in
identical NMR tubes, without changing the optical setup, and
∞
at the same laser energy. The polarizations µ obtained in this
way have been plotted in Figure 4 as a function of the quencher
concentration. The low fluctuations of the data establish the
feasibility of this procedure.
It is evident from the figure that the efficiency of cycloadduct
•+
•-
formation from the radical ion pairs A FN is independent
of the quencher concentration. The divergence of the polariza-
∞
∞
tions µ (cA) and µ (tA) for vanishing quencher concentration
is due to the correction of these CIDNP signals by the factor 1
S
q 0
-1
+
[k τ c(FN)] , which is solely applicable for that fraction of
•+
•-
them stemming from the pairs A FN . If that correction is
omitted, the absolute values of these polarizations start at a finite
intercept; they increase with increasing quencher concentration,
S
S
0
though initially slower than k τ c(FN)/[1 + k τ c(FN)], i.e.
The observed substrate isomerization without participation
of FN can thus only be explained by a reaction of the excited
anethole with the solvent to give radical ion pairs.
q
0
q
•+
•-
than the quantum yield of formation of A FN ; finally, they
approach a constant level. On the one hand, the fact that the
limiting polarizations are higher than the polarizations in the
absence of quencher confirms that isomerization does not only
occur in the as yet unidentified radical pair but also in
(40) Calculations show the spin density distributions of tA•+ and tA•-
to be similar. For the aromatic protons this follows also from the pairing
theorem. Corresponding protons in the radical cation and the anion of the
pairs tA•+tA•- would therefore bear opposite polarizations of comparable
•
+
•-
A FN . On the other hand, the observed concentration
dependence of the absolute polarizations of tA and cA points
to a competitive formation of both types of radical pairs.
Polarographic determination gave a value of -1.98 V vs SCE
magnitude. These would largely cancel in back electron transfer of singlet
•
+
•-
pairs, because tA and tA react to the same species, and also in back
electron transfer of triplet pairs, because in this process there is no preference
3
for formation of tA or tA from the cation or the anion. This argument
can be extended to isomerization via bond rotation in a triplet biradical
followed by biradical cleavage.
•
-
for the potential of the redox couple tA/tA in acetonitrile, so

1
48 J. Am. Chem. Soc., Vol. 118, No. 1, 1996
Goez and Eckert
P
Figure 6. Polarizations of the cycloadducts µ (filled circles) and of
Figure 7. Influence of temperature on the ratio µ
polarizations and polarizations of the isomerized anethole in the systems
P
/µiso of cycloadduct
the isomerized substrate µiso (open circles) in the system tA/FN as
functions of the light intensity Iexc. For clarity, µiso has been multiplied
tA/FN (filled circles) and cA/FN (open circles). Experimental param-
-3
by a factor of 2. Experimental parameters: c(tA) ) 3 × 10 M, c(FN)
8 × 10 M, T ) 251 K. The fit curves to µ and µiso are given by
eqs 2 and 3. The inset shows a plot of the ratio µiso/µ as a function
-3
-1
eters: c(tA) ) 4.2 × 10 M, c(FN) ) 1.5 × 10 M, and c(cA) ) 8
-
3
)
P
-3
-2
×
10 M, c(FN) ) 2.4 × 10 M, respectively.
P
of Iexc
.
systems tA/FN and cA/FN at large quencher concentration, i.e.,
in the plateau regime of Figure 3. It is seen that quite different
behavior is found in these two instances: in the former case,
an increase of the temperature leads to a noticeable decrease of
the polarization ratio, while in the latter case the influence of
the experimental temperature is negligible.
Influence of the Laser Intensity. Plots of the CIDNP signal
strengths for the cycloadducts and the isomerized substrate as
functions of the intensity Iexc of the excitation light are given
in Figure 6. As adduct formation was shown to occur via the
•
+
•-
radical ion pairs A FN , which are produced by electron
transfer quenching of 1A, it is natural to assume that the
polarizations µP are proportional to Iexc
CIDNP enhancement factors are functions of temperature
14d,e
through their dependence on the diffusion coefficient.
However, this is eliminated by considering ratios of polariza-
tions. The experimental conditions in these measurements were
such that the biphotonic process was suppressed so strongly as
to be undetectable. Hence, all polarizations exclusively stem
µ ) ø ‚I
(2)
P
1,P exc
where the constant ø1,P comprises among other things the CIDNP
enhancement factor for this radical pair. Equation 2 holds, as
can be seen in the figure.
Surprisingly, however, at low quencher concentrations µiso
grows faster than linearly with increasing Iexc. These data can
be fitted with a function composed of a term quadratic in Iexc
and a term linear in Iexc (compare Figure 6). Since the CIDNP
experiments at variable quencher concentration furnished evi-
•+
•-
from the radical ion pairs A FN , and the observed effects
must be attributed to processes occurring at later stages of the
reaction than these pairs.
Time-Resolved CIDNP Measurements and Scavenging
Experiments. As has already been established above, olefin
•+
•-
isomerization involving the pairs A FN chiefly occurs by
back electron transfer of triplet pairs in our systems, yielding
1
dence that A is desactivated by two competitive processes
3
3
one of the triplet species A or FN. However, with regard to
leading to two radical pairs and that both these pairs provide
pathways for isomerization of the substrate, the polarizations
µiso must indeed comprise a term linear in Iexc due to the pair
cycloaddition, which also takes place via the triplet exit channel
•+
•-
of A FN , a mechanistic question is left open: P1-P4 may
result from geminate combination of the triplet pairs yielding
triplet species (e.g., triplet biradicals), which then decay to the
species detected, or they may be formed by escape from the
cage to give free radicals, which ultimately react further to the
cycloadducts. In the photocycloadditions of anetholes with
quinones we found the first possibility to be realized;17 on the
other hand, formation of free radicals is the most common decay
pathway of triplet radical pairs. A differentiation between these
alternatives is possible because the former occurs on a nano-
second time scale, whereas the second is much slower. Hence,
the two can be distinguished either by time-resolved CIDNP
•
+
•-
A FN ; obviously, the term caused by the other pair depends
on the square of the light intensity
2
exc
µiso ) ø1,iso‚Iexc + ø2,iso‚I
(3)
The ratio µiso/µP is thus proportional to Iexc and possesses a
nonzero limiting value of ø1,iso/ø1,P for vanishing Iexc, as the inset
of Figure 6 shows. From the former observation, it immediately
follows that the additional radical pair is produced by a two-
quantum process, whereas the latter observation once more
confirms that isomerization takes place via both pairs.
At large quencher concentrations the linear dependence of
the product polarizations µP on the light intensity remains
unchanged whereas the curve for µiso is flattened out as to
become practically linear as well. This clearly reflects the fact
that formation of the two radical pairs occurs by competitive
reactions, a high concentration of FN favoring quenching, i.e.,
the monophotonic pathway and suppressing the biphotonic
process.
16
measurements (“flash CIDNP”) or by scavenging experiments.
In samples with very low concentrations of tA (c(tA) < 1 ×
-
3
10 M), we observed a time dependence of the signal due to
the starting anethole. The intensity of this emissive signal was
largest immediately after the flash and then decreased by some
30% within a few microseconds. This can be explained by
cancellation of the geminate polarizations in tA, which appear
instantaneously on the time scale of these measurements, by
•
+
the opposite polarizations in escaping tA that are gradually
transferred to tA by degenerate electron exchange with surplus
substrate, a phenomenon that is well known in CIDNP
Temperature Dependence. Figure 7 displays plots of the
temperature dependence of the polarization ratios µP/µiso in the

Photoinduced Electron Transfer Reactions of Aryl Olefins
J. Am. Chem. Soc., Vol. 118, No. 1, 1996 149
investigations of electron transfer reactions.41 In contrast, we
could not detect any change of the signals due to the isomerized
olefin and the cycloadducts. Owing to the low optical density
of these samples and the lower sensitivity of flash-CIDNP
measurements in comparison to pseudo steady-state CIDNP
experiments, all signals were very weak. While experimental
error was thus too large to permit quantitative evaluation, these
results may be taken to indicate that some amount of free
radicals is indeed formed. However, their polarizations evi-
dently do not appear in the cycloadducts nor in the isomerized
anethole but are predominantly transferred back to the starting
anethole by degenerate electron exchange; from the above-
mentioned absence of relaxation losses in the pseudo steady-
state experiments it is clear that a decay of these polarizations
by nuclear-spin relaxation in the free radicals or in subsequent
paramagnetic species is negligible.
both of these reactions necessarily being key steps in an electron-
transfer/triplet mechanism such as has been inferred1
2,13
for
systems similar to ours, we employed thioxanthone TX as triplet
3
3
sensitizer to prepare A and FN indirectly, i.e., without
intervention of radical ion pairs. Excitation of this sensitizer
can be performed at λexc g 390 nm where none of the other
compounds absorbs. TX possesses a triplet energy of 272 kJ
-
1 45
mol ; its reduction potential in acetonitrile vs SCE is -1.66
46
V. Though being only weakly exergonic, energy transfer from
3
TX is thus thermodynamically feasible for both anetholes as
well as in the case of FN (Table 2). In contrast, electron transfer
-
1
is seen to be endergonic by at least 20 kJ mol .
-
3
-2
A sample containing TX, 4 × 10 M of tA, and 1 × 10
M of FN in acetonitrile-d was illuminated, and the composition
3
of the solution was monitored by NMR. We found that the
photostationary equilibria between the geometric isomers of the
the substrate and those of the quencher were established fairly
Because of its low oxidation potential (+1.12 V in acetonitrile
vs SCE),42 1,2,4-trimethoxybenzene TMB has repeatedly been
3
3
rapidly, pointing to efficient formation of both A and FN.
However, adducts P1-P4 were nearly undetectable. This
9
,43
employed to scavenge radical cations.
As the reduction of
•
+
-1
3
free tA by TMB is exergonic by 26 kJ mol , this reaction
should be diffusion controlled. In the system tA/FN (c(tA) )
clearly shows that formation of a biradical from A + FN or
3
from FN + A must be much less efficient than decay of these
-
2
-3
2
× 10 M, c(FN) ) 8 × 10 M, T ) 279 K), absolute
triplet molecules to their ground states, unless this biradical
would almost entirely undergo cleavage after intersystem
crossing instead of ring closure.
polarizations of the cycloadducts were somewhat reduced upon
addition of TMB, by some 4, 14, 20, and 22% at scavenger
-
3
-3
-3
-3
concentrations of 2 × 10 , 4 × 10 , 6 × 10 , and 9 × 10
M, respectively. This decrease can be ascribed partly to
absorption of the excitation light by the sensitizer, the extinction
coefficient of TMB at 308 nm being about 10% of that of tA,
Discussion
Nature of the Additional Radical Pair. The findings
reported in the preceding sections show that irradiation of our
substrates in the absence of FN also leads to a radical pair, that
1
partly to quenching of tA by TMB, which should be exergonic
-
1
by more than 80 kJ mol , and partly to interception of the
•
+
one component of this pair is A and the other can only stem
from the solvent and that its formation relies on a biphotonic
process.
•+
•-
correlated radical pairs tA FN by the scavenger. However,
even at the lowest concentration of TMB employed the life of
free anethole cations is calculated to be reduced to less than 30
ns; this should lead to almost complete quenching of all escape
polarizations, yet only a very small effect on the polarizations
is found in that case. An involvement of free radicals in
cycloadduct formation is thus ruled out for the pathway giving
rise to CIDNP.
We interpret these experimental observations by two-photon
ionization of the anetholes. Photoionization of these substrates
by a biphotonic process is in accordance with expectations, since
the ionization potential of tA in the gas phase is 7.83 eV;47
despite the substantial reduction of the energy needed for
4
8
ionization that is often found in fluid phase, it is therefore
unlikely that absorption of a single photon of wavelength 308
nm (=4.03 eV) leads to emission of an electron. A recent laser
flash-photolysis investigation also gave evidence for two-photon
It is well known that various nucleophiles readily add to the
double bond of the radical cations of aryl olefins.9
,25a,44
Despite
the moderate reactivity of â-substituted derivatives of styrene
4
4
44
toward methanol, reactions involving free anethole cations
should be detectably suppressed if high concentrations of this
alcohol are employed. However, even in neat CD3OD the
CIDNP signals of the olefins and of P1-P4 persisted, and we
did not observe any polarizations indicating formation of
scavenging products. This further corroborates a pathway of
cycloadduct formation via geminate combination of triplet
radical pairs.
ionization of various other arylolefins.
The occurrence of nuclear spin polarizations in solutions of
triethylamine in acetonitrile, when a charge-transfer band around
300 nm was excited with light of high intensity, has been
reported, and these observations were explained by electron
4
9
transfer from the amine to the solvent. Likewise, the results
5
0
of theoretical studies as well as experimental investigations
5
1
by pulse radiolysis point to a sufficient stability of the anion
•
-
•-
Attempts to trap the intermediate triplet species with oxygen
were unsuccessful. Irradiation of oxygen-saturated solutions
led to no new NMR resonances, neither in the spectra taken
after illumination nor in the CIDNP spectra, which are more
sensitive than the former because the polarizations cause a large
signal enhancement. Furthermore, no significant influence of
oxygen on relative or absolute CIDNP signal intensities could
be detected.
ACN of acetonitrile or the dimeric species (ACN)2 to take
part in formation of a radical ion pair. Finally, upon photolysis
of 2,7-bis-(dimethylamino)-4,5,9,10-tetrahydropyrene in aceto-
nitrile a transient absorption with a half-life of 9 ns was detected,
which was attributed to an ion pair consisting of the radical
•- 52
cation of the pyrene derivative and (ACN)2
.
While definitive
identification of the anion contained in the radical pairs that
(
45) Dalton, J. C.; Montgomery, F. C. J. Am. Chem. Soc. 1974, 96, 6230-
232.
(46) Yates, S. F.; Schuster, G. B. J. Org. Chem. 1984, 49, 3349-3356.
47) Caldwell, R. A.; Smith, L. J. Am. Chem. Soc. 1974, 96, 2994-
2996.
(48) Lesclaux, R.; Joussot-Dubien, J. In Organic Molecular Photophysics;
Triplet-Sensitization Experiments. To obtain information
6
3
3
about photocycloadditions of A + FN or FN + A, either or
(
(
41) Closs, G. L.; Sitzmann, E. V. J. Am. Chem. Soc. 1981, 103, 3217-
3
219.
(
42) Zweig, A.; Hodgson, W. G.; Jura, W. H. J. Am. Chem. Soc. 1964,
Birks, J. B., Ed.; Wiley: New York, 1973; Vol. 1, p 457.
(49) Bargon, J. in ref 14d, pp 393-398.
(50) Jordan, K. D.; Wendoloski, J. J. Chem. Phys. 1977, 21, 145-154.
(51) Bell, I. P.; Rodgers, M. A. J. J. Chem. Soc., Faraday Trans. I 1977,
73, 315-326.
8
6, 4124-4129.
(
(
43) Eriksen, J.; Foote, C. S. J. Am. Chem. Soc. 1980, 102, 6083-6088.
44) Johnston, L. J.; Schepp, N. P. J. Am. Chem. Soc. 1993, 115, 6564-
6
571.

1
50 J. Am. Chem. Soc., Vol. 118, No. 1, 1996
Goez and Eckert
Chart 1
assume6d,54 that complete equilibration of conformations takes
place during the life of this open-chain compound. The fact
that in all cycloadducts the anisyl group occupies a trans position
with respect to the methyl substituent regardless of the config-
uration of the starting anethole, and the statistically distributed
configuration of the sterically undemanding cyano substituent
4
are formed in the absence of FN in our systems cannot be drawn
on the basis of our CIDNP experiments, and, besides, the actual
nature of these pairs is unimportant for the mechanistic
conclusions that follow, it thus seems very reasonable that the
electron removed from the substrate is taken up by the solvent
at C can be rationalized in this way. Analogous triplet
biradicals were shown to occur as intermediates in Paterno-
B u¨ chi photocycloadditions of anetholes to quinones, which also
17
proceed via radical ion pairs. In that case, equilibrium between
the different conformers is not always established owing to the
extremely short life of the Paterno-B u¨ chi biradicals involved,
and, consequently, the product distribution exhibits a temper-
ature dependence. However, the products formed at high
temperatures, where bond rotation is able to compete more
favorably with intersystem crossing, likewise possess trans
configurations of the anisyl and methyl groups independent of
the stereochemistry of the starting anethole.
•-
to give a monomeric or oligomeric anion (ACN)x . The g value
•
-
of this radical anion X is expected to be very similar to that
•
-
•+
of FN , i.e., lower than that of A , which is consistent with
the fact that the CIDNP phases of the starting and the isomerized
anethole are not influenced by the presence or absence of the
quencher FN.
Pathway to the Cycloadducts. Owing to the configurational
stability of the anethole radical cations, the radical ion pairs
occurring in the systems tA/FN and cA/FN are chemically
different species that cannot be interconverted during their life.
Yet, in both cases the same cycloadducts are found, and the
relative yields of their isomers P1-P4 are also identical. This
clearly shows that between the radical pairs and the adducts
there must be a common intermediate, in which the stereo-
chemical information of the olefin configuration is lost.
Even if a mechanism involving back electron transfer of
The different ratios of adduct polarizations to polarizations
of the isomerized substrate that are found in the systems tA/
FN and cA/FN further corroborate the intermediacy of such a
triplet biradical. After intersystem crossing of substituted
tetramethylene-1,4-biradicals, ring closure is frequently ac-
5
4,55
companied by homolytic scission of the 2,3-bond.
Thus,
in addition to bidirectional isomerization of the starting anethole
3
via A formed by back electron transfer of the radical ion pairs
there should be an isomerization route via the triplet biradicals,
•
+
•-
3
3
A FN to give either A or FN followed by cycloaddition in
the triplet state operated, this loss could not be ascribed to
isomerization at the stage of the triplet olefin, since intermediate
which, however, leads to one-way cis-trans isomerization only.
3
3
•+
•-
If the probabilities of formation of A from tA FN and
3
3
3
•+
•-
formation neither of A + FN nor of A + FN can account for
the product distribution, which is characterized by complete one-
way cis-trans isomerization of the anethole moiety; the first
alternative would lead to bidirectional isomerization of this
fragment and the second to retention of the configuration of
the starting anethole. Hence, neither of these two triplet species
qualifies as that key intermediate common to both our reaction
systems which is responsible for the stereochemistry of the
cycloadducts.
cA FN are similar, and if the same holds for the prob-
abilities of biradical formation from these two pairs, which both
seems reasonable to assume, the yield of isomerized substrate
is therefore expected to be higher for the system cA/FN than
in the case of tA/FN. This is indeed found, and manifests itself
impressively by the approximately three times higher limiting
value of the ratio µP/µiso at large quencher concentrations that
is observed with tA as substrate (compare Figure 3).
We should like to point out that neither the absence of CIDNP
in trapping products when the reactions are carried out in
oxygen-saturated solutions nor the unchanged relative CIDNP
intensities under these circumstances rule out the occurrence
of triplet biradicals. It is known that Heisenberg spin exchange
is an important, and perhaps even often a dominant, mechanism
In contrast, the observed product distribution is perfectly
consistent with formation of triplet biradicals and isomerization
at this stage by rotations around single bonds. In principle, the
3
3
two regioisomers BR and BR′ displayed in Chart 1 could both
explain the stereochemistry of the products. However, biradical
3
BR′, which lacks the benzylic stabilization of one radical center,
5
6
of oxygen quenching of biradicals. This pathway would not
lead to formation of new products bearing polarizations. Nor
would it destroy polarizations since CIDNP generation is
completed at the stage of the radical ion pairs; afterwards, the
existing polarizations are only transferred to subsequent inter-
mediates and finally to the end products. Significant effects
can be ruled out by thermodynamic arguments, as will be shown
below.
A methyl substituent will only exhibit a negligible interaction
with a vicinal cyano group. Furthermore, it is reasonable to
assume that the precursors to the biradical approach in such a
way that there is a small but nonzero angle between the
molecular planes, thereby minimizing any repulsive orbital
interactions involving the phenyl ring. The two diastereomers
3
could only result with this mechanism if the lifetime of BR
were reduced below about 1 ns such that equilibration of
conformations were no longer possible; then, the product
distribution and also the ratio µP/µiso would change. However,
3
of BR should therefore be formed in essentially equal amounts
regardless of whether the configuration of the starting olefin is
cis or trans. This is reflected by the fact that both with cA and
with tA the yields of P1 and P4 are identical within experimental
accuracy, as are those of P2 and P3.
5
7
as a lifetime of some 20 ns is estimated for the case that the
biradical life were only limited by oxygen quenching, this seems
unattainable under our experimental conditions.
For 1,4-biradicals of similar constitution triplet lifetimes
around 30 ns were determined. Thus, it is reasonable to
53
(54) Schuster, D. I.; Heibel, G. E.; Woning, J. Angew. Chem., Int. Ed.
Engl. 1991, 30, 1345-1347.
(52) (a) Hirata, Y.; Mataga, N.; Sakata, Y.; Misumi, S. J. Phys. Chem.
(55) (a) Lewis, F. D.; Hirsch, R. H. J. Am. Chem. Soc. 1976, 98, 5914-
5924. (b) Wagner, P. J. Acc. Chem. Res. 1971, 4, 168-177. (c) Barton,
D. H. R.; Charpiot, B.; Ingold, K. U.; Johnston, L. J.; Motherwell, W. B.;
Scaiano, J. C.; Stanforth, S. J. Am. Chem. Soc. 1985, 107, 3607-3611.
(56) Caldwell, R. A.; Creed, D. J. Am. Chem. Soc. 1978, 100, 2905-
2907.
1
982, 86, 1508-1511. (b) Hirata, Y.; Takimoto, M.; Mataga, N. Chem.
Phys. Lett. 1983, 97, 569-572. (c) Hirata, Y.; Mataga, N.; Sakata, Y.;
Misumi, S. J. Phys. Chem. 1983, 87, 1493-1498.
(53) Kaprinidis, N. A.; Lem, G.; Courtney, S. H.; Schuster, D. I. J. Am.
Chem. Soc. 1993, 115, 3324-3325.

Photoinduced Electron Transfer Reactions of Aryl Olefins
J. Am. Chem. Soc., Vol. 118, No. 1, 1996 151
3
3
The energy of BR must lie below that of A + FN or that
Scheme 1
3
of A + FN, because in all three instances the overall bond
order is the same, but BR possesses one additional σ-bond
3
instead of a π-bond and the unfavorable interaction of the
unpaired electrons is certainly lower in the 1,4-biradical than
in the triplet of the donor or the acceptor olefin. A numerical
5
8
59
value can be obtained by using Benson’s and Caldwell’s
60
0
group increments. By this procedure, we estimated ∆G of
BR relative to tA + FN to be +197 kJ mol , which is much
lower than the triplet energies, in accordance with the above
qualitative argument. On the other hand, formation of A as
well as of FN from triplet pairs is thermodynamically feasible
3
-1
3
3
in our systems (compare Table 2) and indeed takes place, as is
evidenced by the bidirectional isomerization of the substrates
and the isomerization of the quencher FN. Hence, both direct
3
•
+
•-
3
reaction of A FN to BR and indirect formation of this
triplet biradical from the radical ion pairs via the intermediacy
of A or FNsthe electron-transfer/triplet mechanism proposed
3
3
However, the fact that the yields of cycloadducts from the
radical ion pairs neither depend on the quencher concentration
(see Figure 4) nor on the concentration of the substrate (Figure
1
2,13
for photocycloadditions of other donor and acceptor olefins
6
1
or photodimerizations of olefins sare conceivable. On the
grounds of kinetic considerations, one might expect the indirect
routes to be preferred because geminate combination of a radical
ion pair to give a biradical requires cation and anion to be in
actual contact and presumably to possess a favorable orientation,
whereas back electron transfer of such a pair can occur at larger
5
) cannot be reconciled with biradical formation by attack of
3
3
A to FN or by attack of FN to A. Further corroboration is
3
obtained by the triplet sensitization studies, where both A and
3
FN are generated as is seen from the attainment of the
photostationary equilibria between the cis- and trans-forms of
both substrate and quencher, but where the cycloadduct yields
are negligible; as the biradical has been established as precursor
to P1 to P4 by the stereochemical arguments given above, this
6
2
separations and probably needs weak orbital overlap only.
(
57) The saturation concentration of oxygen in acetonitrile at room
-
3
temperature is 9 × 10 M (Murov, S. L.; Carmichael, I.; Hug, G. L.
3
observation must be ascribed to a very low yield of BR from
Handbook of Photochemistry; Marcel Dekker: New York, 1993). A rate
3
3
9
-1 -1
A + FN as well as from FN + A. Hence, these findings rule
constant of 6 × 10 M
s
was determined for O2 quenching of a Norrish
II 1,4-biradical (Small, R. D.; Scaiano, J. C. J. Phys. Chem. 1977, 81, 2126-
out a significant contribution of the indirect route, triplet radical
ion pair f olefin triplet f triplet biradical, to cycloaddition in
our systems; instead, the biradicals must be formed directly,
by geminate combination of radical ion pairs in triplet states.
This direct pathway is also consistent with the results of our
2
131). With these data, a first-order rate constant of (19 ns)- is thus
1
obtained.
(58) Benson, S. W.; Cruickshank, F. R.; Golden, D. M.; Haugen, G. R.;
O’Neal, H. E.; Rodgers, A. S.; Shaw, R.; Walsh, R. Chem. ReV. 1969, 69,
2
79-324.
(
59) Ni, Tuqiang; Caldwell, R. A.; Melton, L. A. J. Am. Chem. Soc.
17
studies of photocycloadditions of anetholes to quinones. In
1
989, 111, 457-464.
60) We use Benson’s nomenclature58 of groups in the following.
3
(
these reactions, neither A nor triplet quinones can be formed
Increments for nonradical groups are from ref 58, for radical groups from
ref 59. According to ref 59, olefin triplets may be regarded as 1,2-biradicals
to a very good approximation and described by the same group increments
for energetic reasons, yet efficient adduct formation takes place
by the direct route.
0
3
0
Finally, ∆G of the regioisomer BR′ can be calculated as
as 1,4-biradicals. For calculation of ∆H of the quencher-based moiety in
3
63
BR we therefore take half the triplet energy of FN, which corresponds to
above by using group increments. It is found that the free
•
replacing one group Cd(H)(CN) by C(H)(C)(CN). To take into account
3
-1
enthalpy of BR′ (+257 kJ mol relative to tA + FN) is higher
the combination with the anethole fragment, we subtract from this the
-
1
3
-
1
by 60 kJ mol than that of BR. This is of course a
manifestation of the difference in stabilization of a benzyl radical
and a secondary alkyl radical. Comparison with the data of
increment for C(CT)(C)2(H), +28 kJ mol , and add the increment for
-
1
-1
Cd(CT)(H), -7 kJ mol . The result is +95 kJ mol . Likewise, the triplet
energy of the anethole may be thought of as arising from converting Cd(CB)-
H) into C(CB)(C)(H), and Cd(H)(C) into C(C)2(H). The former change
also occurs in the transformation from tA to BR; furthermore, in going
•
•
(
3
3
Table 2 shows that formation of BR from the radical ion pairs
3
3
3
•
-1
•+
•-
from A to BR the group C(C)2(H) (increment +172 kJ mol ) must be
-1
tA FN is exergonic by some 60 kJ mol , whereas the
driving force for formation of BR′ is practically zero. By these
thermodynamic considerations, a participation of BR′ in our
reactions is seen to be very unlikely. BR is also kinetically
favored over BR′: Our AM1 computations of A gave the
highest positive charge density at the â carbon atom; it is thus
reasonable that combination with FN takes place at this
replaced by C(H)(C)3 (increment -8 kJ mol ). This puts ∆H0 of this
-
1
3
-
1
0
fragment at +70 kJ mol . The total value of ∆H of the biradical relative
-
1
3
to the value for the starting materials is therefore +165 kJ mol . The
0
3
predominant contribution to ∆S is the change of one olefinic carbon atom
3
in tA and one in FN into two aliphatic carbon atoms in BR. For the
3
•+
anethole part we must therefore subtract the increment for Cd(H)(C), +33
-
1
-1
-1
-1
J K mol , and add the increment for C(H)(C)3, -51 J K mol ; for
•
-
the quencher-based moiety we must subtract the increment for Cd(Ct)(H),
27 J K mol , and add the increment for C(Ct)(C)2(H), -47 J K mol
-1
-1
-1
-1
+
.
3
position, which leads to BR.
However, this overestimates the decrease in entropy occurring in combina-
tion to the biradical, since it does not take into account the conformational
flexibility introduced by the transformation of the two other olefinic carbon
atoms into two radical centers. To correct for this effect, we need the
Reaction Mechanism and Determination of Parameters.
All our experimental findings are explained by the reaction
mechanism given in Scheme 1. For simplicity, this Scheme
has been drawn for the system tA/FN only. An analogous
scheme results when the starting anethole is cA. The signs given
at the final products indicate the polarization phases of their
0
•
increment for ∆S of the group C(C)2(H), which we estimate to be +58 J
-
1
-1
-1
-1
K
mol by subtracting twice the increment for C(H)3(C), +116 J K
1 0 -1
-
mol , from the experimental value of ∆S for iso-C3H7, +290 J K mol
(
Tsang, W. J. Am. Chem. Soc. 1985, 107, 2872-2880). As the increment
-1 -1
for an olefinic group Cd(H)(C) is +33 J K mol , we thus get a total
0
-1
0
3
3
value of ∆S for formation of the biradical from tA + FN of -108 J K
(63) The difference in ∆G for BR and BR′ should be solely determined
0 3 3
-1
0
mol . Hence, at room temperature the entropy contribution to ∆G is about
by the difference in ∆H . In the transformation BR f BR′, the groups
•
-1 59
-1 58
2
5
0% of the enthalpy contribution.
61) Farid, S.; Brown, K. A. J. Chem. Soc., Chem. Commun. 1976, 564-
65.
62) Rehm, D.; Weller, A. Isr. J. Chem. 1970, 8, 259-271.
C(CB)(C)(H) (increment +116 kJ mol ) and C(C)3(H) (-8 kJ mol )
-1 58 • -1 59
(
change into C(CB)(C)2(H) (-4 kJ mol ) and C(C)2(H) (+172 kJ mol ),
0 3
0
3
respectively. ∆H of BR′ is thus more positive than ∆H of BR by 60
-
1
(
kJ mol .

1
52 J. Am. Chem. Soc., Vol. 118, No. 1, 1996
Goez and Eckert
methyl protons; these phases depend on the respective pathway
of product formation.
cycloadducts, whereas biradical cleavage (efficiency 1 - â)
produces tA only.
The excited substrate decays with its unquenched life τ0 to
the educt anethole and to a small degree also to the isomerized
anethole and to a dimer, all of which bear no polarizations and
have therefore been omitted from the scheme. 1tA is quenched
On the basis of this mechanism, the dependence of the
polarization ratio µP/µiso on quencher concentration (Figure 3)
and light intensity (Figure 6) can be rationalized with changes
in the yields of the intermediates RP I and RP II, which are
formed by competitive processes; to some degree, transforma-
•+
•-
by FN to give a pair tA FN (RP I) with rate constant kq and
•
-
tions of RP II into RP I by interception of X with FN must
also be taken into account. A quantitative description of the
chemical kinetics up to and including the radical pairs as well
as of the scavenging experiments will be given in a forthcoming
•+
•-
reacts to a radical pair tA X (RP II) in a biphotonic process
with rate constant ki. Both pairs originate in the singlet state.
In the radical ion pairs, intersystem crossing takes place.
Owing to the higher g value of the anethole radical cation
6
5
publication. The present work, apart from elucidation of the
general reaction mechanism, mainly focuses on geometric
isomerization and cycloaddition, i.e., processes occurring after
the stage of the radical ion pairs. Kinetic information about
these steps can be obtained from the CIDNP measurements at
•
-
•-
compared to that of X or FN and because the hyperfine
•
+
coupling constant of the methyl protons in A is positive,
intersystem crossing of these pairs is the faster the higher (i.e.,
more positive) the z projection of the particle spin of the methyl
group. In consequence, both in the case of RP I and in that of
RP II absorptive nuclear spin polarizations of the methyl protons
develop in the triplet pairs and emissive polarizations in the
singlet pairs.
1
high quencher concentrations, where A is essentially desacti-
vated by quenching with FN only, and where the role of RP II
is thus negligible.
With the starting anethole being tA, it is obvious from Scheme
1 that the ratio of adduct polarizations to polarizations of the
isomerized substrate in the limit of infinite quencher concentra-
As has been shown, escape from the cage plays no role with
respect to CIDNP in these systems except in time-resolved
measurements. By reverse, as far as steady-state polarizations
are concerned the singlet pairs of RP I and RP II can be thought
of as decaying exclusively by back electron transfer (compare
also note 64), thereby transferring their emissive polarizations
to the regained substrate. Back electron transfer of triplet pairs
lim
iso tA
tion, (µ /µ ) , is
P
µP lim
âγ
R(1 - γ)
)
(4)
(
)
µ
iso tA
3
to give A is also possible for RP I as well as RP II. Decay of
Setting up the reaction scheme for cA as starting anethole, we
this phantom triplet leads to cA and tA with efficiencies R and
lim
arrive at the corresponding quantity (µ /µ )
iso cA
3
P
1
- R, so the absorptive polarizations transferred to A from
both pairs are partitioned between the starting and the isomerized
lim
µP
âγ′
(1 - â)γ′ + (1 - R)(1 - γ′)
anethole.
)
(5)
(
)
µ
3
iso cA
•+
•-
In the same manner, back electron transfer of tA FN to
3
give FN + A occurs, as is evidenced by the polarizations of
where R and â by necessity are the same parameters as in the
system tA/FN, but where the efficiency γ′ of biradical formation
might differ from γ. However, it seems reasonable to assume
that γ and γ′ are identical, since the driving force for back
the isomerized quencher MN (see Figure 1). One could thus
3
3
hope to obtain the relative yields of A and FN from the CIDNP
spectra. While in principle this is indeed possible, in our
experiments the reproducibility of the polarization intensities
of FN and MN was too poor to permit such a quantitative
evaluation. We attribute this to protonation of these radical
anions by traces of water present in varying amount in the
samples. For clarity, this reaction has therefore been omitted
from Scheme 1. Its neglect does not influence the relative
polarizations of the anethole species and the anethole derived
moieties in the products.64
3
electron transfer of RP I to give FN plus the respective triplet
anethol is very similar in both cases (compare Table 2), the
same holds for the driving force of biradical formation, and an
influence of the configuration of the anethole cation on the rate
of biradical formation is not to be expected. The parameter R
is 0.59, as was determined from the photostationary state reached
66
in the benzophenone-sensitized isomerization of anethole; as
R depends only on the potential curves of anethole in the ground
state and in the triplet state, it should be constant within the
range of temperatures encountered in our experiments. With
this value, and on the assumption that γ and γ′ are equal, â and
γ can be obtained in a straightforward way from the observed
polarization ratios in the limit of high quencher concentration,
by using eqs 4 and 5. Error analysis shows that â and γ are
only weak functions of the polarization ratios, so the accuracy
(an estimated 1%) is not severely affected by possible small
deviations of the experimental values of µP/µiso from the limiting
values nor by fluctuations of the data.
Besides back electron transfer, geminate reaction to give a
3
triplet biradical BR provides an additional decay pathway of
3
•
+
•-
tA FN . We assign an efficiency γ to biradical formation,
and an efficiency 1 - γ to formation of the triplet anethole
from this pair, thereby referring both processes to unit efficiency
3
of formation of a species other than FN. Regardless of the
3
3
rate of the reaction RP I f FN + tA, the ratio of rate constants
3
3
3
3
for the processes RP I f BR and RP I f A + FN is thus
given by γ/(1 - γ). At the stage of the biradical, equilibration
of conformations occurs. Finally, after intersystem crossing to
1
The efficiency of biradical formation γ is found to be fairly
large, between 0.69 at 312 K and 0.73 at 241 K, so geminate
the singlet state, ring closure of BR (efficiency â) leads to the
3
3
3
64) If a fraction δ of tA•+FN•- reacts to FN + tA, this fraction of the
3
reaction of RP I to the biradical BR is faster than back electron
(
•
+
3
polarizations is transferred from tA in the triplet pairs to the starting
anethole, hence the opposite polarizations in this compound resulting from
the singlet exit channel are decreased by δ. Exactly the same fraction of
the polarizations from the triplet exit channel is missing in A and BR
and all their subsequent products. In evaluations of relative polarizations
stemming from the anethole radical cation only, this reaction is thus
unimportant. The same argument can be used to explain why omission of
escaping radical ions does not influence relative polarizations as long as
no relaxation losses occur.
transfer to give A by a factor of more than 2. Since only
relative rate constants can be obtained from these measurements,
it is not immediately clear whether back electron transfer is
rather slow or biradical formation unusually fast in these
systems. However, a comparatively low rate of back electron
3
3
(65) Eckert, G.; Goez, M. Manuscript in preparation.
(66) Caldwell, R. A.; Pac, C. Chem. Phys. Lett. 1979, 64, 303-306.

Photoinduced Electron Transfer Reactions of Aryl Olefins
J. Am. Chem. Soc., Vol. 118, No. 1, 1996 153
transfer is made plausible by the low driving force of this
it as established that in polar solvents basically the same
mechanism governs the [2 + 2] cycloadditions of photoexcited
donor olefins (anetholes) to acceptor olefins (fumarodinitrile)
studied in the present work and the Paterno-B u¨ chi reactions
of photoexcited electron-deficient carbonyl compounds (qui-
0
-1
process, -∆G of it being 10-12 kJ mol only. According
3
0
to the above calculations of the thermodynamics of BR, -∆G
for biradical formation is more than five times higher than this.
The fact that for spin-allowed electron transfer the critical
reaction distance is larger than for bond formation6 is thus
obviously outweighed by these thermodynamic aspects. In
contrast, singlet pairs preferentially decay by back electron
transfer, which is strongly exergonic in this case.
2
17
nones) with donor olefins (anetholes) investigated previously.
In both cases, the first chemical step is full charge transfer
to give radical ion pairs; we could unambiguously exclude
pathways bypassing these pairs, as for instance cycloadditions
via exciplexes, and could show that the pathways giving rise to
CIDNP are the main reaction routes. At the stage of the radical
ion pairs, intersystem crossing occurs. Singlet pairs decay by
back electron transfer; triplet pairs do likewise if this is
energetically feasible, e.g., in the anethole/fumarodinitrile
systems. The latter process leads to the triplet of one or both
reaction partners and thus provides a pathway to olefin isomer-
ization; for olefins such as those employed in this work, this
results in bidirectional isomerization.
The experimentally determined efficiencies â for ring closure
1
of BR lie between 0.41 and 0.42 in the temperature interval
considered. This agrees well with the values obtained in the
photoreactions of various cycloalkanones with cyanosubstituted
6
d,54
olefins,
.2-0.5.
The parameter γ is seen to decrease slightly when the
which were all reported to fall into the range of
0
temperature is raised. This appears reasonable, because higher
temperatures should favor formation of two molecules over
formation of a single species, i.e., favor back electron transfer
to triplet species over combination of two radical ions to give
a biradical. However, the same argument should also apply to
The triplet pairs possess an important additional decay
channel, namely geminate combination to give triplet biradicals.
Indirect formation of these key intermediates, via back electron
1
cleavage of BR, so one would expect a similar decrease of â
transfer of triplet pairs to give one of the reactants in the triplet
with increasing temperature, which is not found experimentally.
Instead, that parameter shows a very small rise with rising
temperature. From a plot of the logarithm of γ/(1 - γ) versus
12,13
state followed by attack to ground state molecules,
is
precluded by thermodynamics in the Paterno-B u¨ chi systems
17
studied. For the cycloadditions investigated in the present
1
/T one can conclude that the activation enthalpy of back
work, these indirect pathways would be thermodynamically
feasible, and olefin triplets are indeed formed in the anethole/
fumarodinitrile systems. Yet they do not lead to cycloaddition;
our experimental findings clearly show that only the direct route
to the biradical is taken.
3
electron transfer to give A must be marginally larger than the
activation enthalpy of biradical formation.
Finally, the noticeably different temperature dependence of
lim
the polarization ratios (µP/µiso) in the systems cA/FN and tA/
FN (compare Figure 7) is easily rationalized with the mechanism
of Scheme 1. For these limiting polarizations RP II plays a
negligible role. In both cases, cycloaddition relies on formation
Since the biradical is produced by combination of radical ions
bearing opposite charges, it is initially formed with a geometry
that is determined by maximum Coulomb stabilization of the
precursor complex, hence a sandwich-like structure of the
3
1
of BR and on ring closure of BR, both processes having
opposite temperature coefficients. The decrease in γ by some
1
7
transition state.
Whether or not the resulting biradical
5
% in the experimental temperature interval is partly cancelled
conformationswhich will usually be energetically unfavorable,
because a sandwich-type precursor with maximum area of
contact between the radical ions preferentially leads to anti
conformation of substituents in 1,4- and 2,3-positionssis
preserved is decided by the competition between thermally
acticated bond rotation in this open-chain compound and
temperature-independent intersystem crossing. If intersystem
crossing is much faster than bond rotation, as in Paterno-B u¨ chi
biradicals at low temperatures, the stereochemistry of the
cycloadducts reflects the initial stereochemistry of the biradical.
In that case, the reaction can lead to a single product; a different
stereoisomer is obtained when the configuration of the educt
olefin is changed. If the triplet life of the biradical is
comparatively long, as in the systems of the present work,
complete equilibration of conformations occurs, and several
products result, the stereochemistry of which does not depend
on the geometry of the starting olefin.
by the smaller increase in â (some 3% in that range), so the
product âγ varies only very little with T. In contrast, with regard
to isomerization the two systems behave differently. With tA
as starting olefin, isomerization solely occurs via back electron
3
transfer to give A, the relative contribution of this decay
pathway of RP I increasing with increasing temperature; to first
approximation, this leads to a temperature dependence of
lim
P iso tA
-1
(
µ /µ ) that is given by (1 - γ) . However, the system
cA/FN possesses an additional isomerization pathway, via the
biradical, so desactivation of RP I by either channel leads to a
species capable of isomerization. The temperature dependence
lim
iso cA
of (µ /µ ) is thus determined by the partitioning of RP I
P
between these two pathways, which to first order possess similar
isomerization efficiencies, 1 - R and 1 - â being within 30%
of each other. The remaining imbalance of these two channels
is only a second-order effect. Indeed, by rewriting eq 5 and
setting γ′ equal to γ one arrives at
Finally, ring closure of the singlet biradical is accompanied
by biradical cleavage. In the reactions of anetholes with both
fumarodinitrile and with quinones, this provides an important
pathway to unidirectional cis-trans isomerization of the
substrate, which should be important in the cycloadditions of
other aryl-substituted olefins as well.
lim
µP
âγ
)
(6)
(
)
µ
(1 - R) + (R - â)γ
iso cA
The numerator of this equation is almost temperature-inde-
pendent, as explained above, and the constant first term in the
denominator is much larger than the temperature-dependent
second term.
It seems reasonable to assume that the mechanism of ionic
photocycloaddition found for the Paterno-B u¨ chi reactions of
17
quinones with anetholes and for the [2 + 2] cycloadditions
of the present work applies also to related reactions of donor
and acceptor olefins in polar media. The following two features
of this mechanism may be of interest for synthetic applications.
First, if such ionic cycloadditions proceed via short-lived
Conclusions
Based on the results of our photophysical and photochemical
experiments, especially the CIDNP measurements, we regard

1
54 J. Am. Chem. Soc., Vol. 118, No. 1, 1996
Goez and Eckert
biradicals and are carried out at low temperatures, they may
lead to adducts possessing a stereochemistry that is decided by
the Coulomb interactions in the transition state leading from
the radical ion pair to the biradical. This may be exploited to
obtain stereoisomers that are different from those resulting from
nonionic cycloadditions of the same substrates and presumably
are the sterically more demanding isomers in many cases.
Second, the basic reaction mechanism is independent of the
multiplicity of the excited species because intersystem crossing
to the triplet state can also occur at the stage of the radical ion
pairs. Hence, reactants can be employed that carry neither
carbonyl functions nor other substituents enhancing the ef-
ficiency of intersystem crossing in the primary excited mol-
ecules. This should broaden considerably the range of com-
pounds useable for these cycloadditions and could thus increase
the synthetic potential of these reactions.
Sample Preparation. Acetonitrile-d
3
(ICB, 99.5% D) was dried
over 3 Å molecular sieve prior to use. Except for the measurements
with varying concentrations of tA, sensitizer concentrations of the
CIDNP experiments were chosen such as to give an absorbance of the
samples at the excitation wavelength of approximately 1. Directly after
preparation, the samples were deoxygenated by bubbling purified
nitrogen through the solutions, and then they were immediately sealed.
Photophysical, Stern-Volmer, and Potential Measurements. The
1
lifetime of tA in acetonitrile was determined with a time-correlated
single-photon-counting apparatus that is described elsewhere.69 Fluo-
rescence measurements were carried out with a Perkin Elmer MPF-44
fluorescence spectrometer. The reduction potential of tA in acetonitrile
was determined with a Metrohm 506 polarograph (supporting electro-
lyte, 0.1 M tetrabutylammonium hexafluorophosphate) against fer-
rocene/ferrocinium.
NMR and CIDNP Measurements. The NOE experiments of P3
were performed with a Bruker AM-400 NMR spectrometer. For all
1
other H-NMR and CIDNP measurements, a Bruker WM-250 console
was used. For data acquisition and experimental control the apparatus
was interfaced to an 80486-based multitasking workstation equipped
with a Keithley AD-converter and connected to a home-made program-
mable pulse generator. Illumination was done by an excimer laser,
Experimental Section
Chemical Substances. trans-Anethole, tA, was doubly distilled in
vacuum, fumarodinitrile, FN, 9-cyanoanthracene, CNA, and thio-
-
6
xanthone were purified by repeated sublimation at 10 bar, 1,2,4-
trimethoxybenzene (97%) was used as received, and cis-anethole, cA,
was prepared by bromination of the trans-isomer followed by dehy-
which was triggered by the pulse generator. A special probe allowed
side-on illumination of the samples.7
0,71
Typically, 10% of the laser
energy was absorbed in the samples, as determined actinometrically.
The pulse sequences employed for the pseudo steady-state CIDNP
6
7
drohalogenation and finally partial hydrogenation with a poisoned
6
8
33
nickel catalyst under atmospheric pressure. Spinning band column
distillation gave a product containing about 5% of the trans-isomer.
As the extinction coefficients of tA at the excitation wavelength is about
a factor of 10 higher than that of cA, further purification was essential.
This was done by GC (carbowax, column length 3 m, 180 °C, carrier
measurements have been described previously. This technique leads
to strong suppression of background signals, and the CIDNP intensities
are practically undisturbed by nuclear-spin relaxation in the diamagnetic
products.
Calculations. Force-field calculations were performed with the
72
H
2
, solvent pentane), yielding a product with a purity of 99.6% (GC).
program PC MODEL (version 4.0) on a personal computer, and
AM173 calculations with MOPAC 5.0 package were run on an IBM
3090 mainframe computer.
For the synthesis of the cycloadducts P1-P4 by photoinduced
electron transfer sensitization, 180 mg of tA, 380 mg of FN, and 15
mg of CNA were solved in 50 mL of acetonitrile. The solution was
irradiated under nitrogen at λexc > 385 nm for several hours in a quartz
vessel. After removal of the solvent in vacuum, the residue was
Acknowledgment. Financial support by the Deutsche For-
schungsgemeinschaft is gratefully acknowledged. We are
indebted to the following members of the Technische Universit a¨ t
Braunschweig, the respective institute given in parentheses: to
J. K u¨ ster (Physical Chemistry) for determining the singlet
lifetime of tA in acetonitrile; to A. Plagens (Organic Chemistry)
for his help in the GC purification of cA; and to Mrs. P. Schulz
fractionated by chromatography (ca. 30 cm SiO
ethylacetate, 1:1), the eluate being monitored by NMR. The early
fluorescent) fractions consisted of tA and the homodimer (R,â,R,â)-
2
, cyclohexane/
(
1
1
,2-dianisyl-3,4-dimethylcyclobutane, D1. The H-NMR parameters
3
4
(
2
CD
3
CN) of the latter are δ 1.131 (CH
3
, m), δ 1.170 (H , H , m), δ
, s), δ 6.822 (Ph, m), δ 7.138 (Ph,
1
2
.776 (H , H , m), δ 3.726 (OCH
3
(Organic Chemistry) for carrying out the 400 MHz NOE
m). The following two fractions contained P1 plus P2 and P3 plus
P4, respectively. By chromatography (SiO , CHCl ) of these fractions,
measurements on compound P3.
2
3
all four stereoisomers were finally obtained in sufficient purity and
with yields of a few milligrams each. Their spectral parameters are
given in Table 4.
JA9511578
(69) Dreeskamp, H.; Salthammer, T.; L a¨ ufer, A. J. Lumin. 1989, 44,
161-165.
(
70) L a¨ ufer, M.; Dreeskamp, H. J. Magn. Reson. 1984, 60, 357-365.
(
67) Fahey, R. C.; Schneider, H. J. J. Am. Chem. Soc. 1968, 90, 4429-
434.
68) Davies, D. E.; Gilchrist, T. L.; Roberts, T. G. J. Chem. Soc., Perkin
Trans. I 1983, 1275-1281.
(71) Goez, M. Chem. Phys. 1990, 147, 143-154.
(72) Serena Software, Bloomington, 1990.
4
(
(73) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. J.
Am. Chem. Soc. 1985, 107, 3902-3909.